Datasets:

gabrielaltay

/

pmcoa

like 0

==== Front Acad Radiol Acad Radiol Academic Radiology 1076-6332 1878-4046 The Association of University Radiologists. Published by Elsevier Inc. S1076-6332(22)00632-8 10.1016/j.acra.2022.11.027 Original Investigation Stacking Ensemble and ECA-EfficientNetV2 Convolutional Neural Networks on Classification of Multiple Chest Diseases Including COVID-19 Huang Mei-Ling PhD. in Industrial Engineering & Management ⁎ Liao Yu-Chieh Master in Industrial Engineering & Management Department of Industrial Engineering & Management, National Chin-Yi University of Technology, 57, Sec. 2, Zhongshan Rd., Taiping Dist., Taichung, 411030, Taiwan ⁎ Address correspondence to: M-L.H. 25 11 2022 25 11 2022 6 9 2022 15 11 2022 20 11 2022 © 2022 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved. 2022 The Association of University Radiologists Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Rationale and Objectives Early detection and treatment of COVID-19 patients is crucial. Convolutional neural networks have been proven to accurately extract features in medical images, which accelerates time required for testing and increases the effectiveness of COVID-19 diagnosis. This study proposes two classification models for multiple chest diseases including COVID-19. Materials and Methods The first is Stacking-ensemble model, which stacks six pretrained models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M. The second model is self-designed model ECA-EfficientNetV2 based on ECA-Net and EfficientNetV2. Ten-fold cross validation was performed for each model on chest X-ray and CT images. One more dataset, COVID-CT dataset, was tested to verify the performance of the proposed Stacking-ensemble and ECA-EfficientNetV2 models. Results The best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.21%, Precision of 99.23%, Recall of 99.25%, F1-score of 99.20%, and (area under the curve) AUC of 99.51% on chest X-ray dataset; the best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.81%, Precision of 99.80%, Recall of 99.80%, F1-score of 99.81%, and AUC of 99.87% on chest CT dataset. The differences for five metrics between Stacking-ensemble and ECA-EfficientNetV2 models are not significant. Conclusion Ensemble model achieves better performance than single pretrained models. Compared to the SOTA, Stacking-ensemble and ECA-EfficientNetV2 models proposed in this study demonstrate promising performance on classification of multiple chest diseases including COVID-19. KEY WORDS COVID-19 Convolutional neural network Ensemble learning Stacking ==== Body pmcINTRODUCTION COVID-19 causes disease in humans and vertebrates, and is a zoonotic infectious disease. Some confirmed patients may have severe pneumonia and respiratory failure (1). The most common tools for detecting lung infections are Chest X-ray and Computed tomography (CT). Chest X-ray is a very common noninvasive radiological test that has been widely used to screen for a variety of lung diseases such as COVID-19, pneumonia, pulmonary effusion, lung cancer, and emphysema (2). In clinical practice, chest X-ray images are often interpreted by radiologists, which is time-consuming and prone to errors in subjective assessments. A CT image is composed of a certain number of pixels with different grayscales arranged in a matrix. If the number of pixels is larger, the pixel value is smaller, and the image will be clearer. Although CT images provide very fine details, but it has more radiation than a chest X-ray image, and the equipment is relatively expensive (3). In recent years, due to the rise of artificial intelligence, researchers have applied deep learning to detect COVID-19 by using chest X-ray images and CT images. Compared to traditional machine learning, deep learning can automatically extract features of images to reduce processing time (4). At present, there have been many researches applying convolutional neural network (CNN) on image recognition. This method has been proved to be a powerful image recognition technology and has been widely used for COVID-19 detection. Common CNNs include GoogLeNet, ResNet, Xception, DenseNet, MobileNet, and EfficientNet, etc. For example, Ieracitano et al. proposed a fuzzy logic-based convolutional neural network (CovNNet) and the classification accuracy is 81.00% in 2022 by using a total of 155 chest X-ray images (5). Khan et al. proposed two convolutional neural networks named DHL and DBHL in 2021, using a total of 6448 chest X-ray images. The results showed that the accuracy of the binary classification of COVID-19 and normal was 98.53% (6). Loey et al. proposed a Bayesian-optimized convolutional neural network in 2022 using a total of 10,848 chest X-ray images. The results showed that the three-category accuracy of COVID-19, general pneumonia and normal was 96.00% (7). Hu et al. proposed a two-stage detection method in 2021 using a total of 11,092 chest X-ray images (8). The first stage is to train a CNN as a feature extractor, and the second stage uses extreme learning machines (ELMs) for real-time detection. In addition, the Chimp optimization algorithm is used to improve the results and increase the reliability of the network, and finally it is compared to the general CNN, Genetic algorithm optimized ELM, Cuckoo search optimized ELM, and Whale optimization algorithm optimized ELM. The results show that the proposed method has a binary classification accuracy of 99.11% for COVID-19 and non-COVID-19. Thaseen et al. applied ensemble learning by combining ResNet, FitNet, IRCNN, MobileNet, and EfficientNet in 2022, using a total of 13,808 chest X-ray images (9). The results showed that the three-category accuracy of COVID-19, general pneumonia and normal was 99.00%. A hybrid convolutional neural network (HCNN) combining CNN and RNN on classification of three-category accuracy of COVID-19, general pneumonia and normal was proposed with the classification accuracy of 98.20% (10). Musallam et al. proposed a convolutional neural network called DeepChest in 2022, using a total of 7512 chest X-ray images. The results showed that the three-category accuracy of COVID-19, general pneumonia and normal was 96.56% (11). A convolutional neural network named DenResCov-19 composing of DenseNet121 and ResNet50 networks was proposed, and DenResCov-19 was combined with existing ResNet50, DenseNet121, VGG16, and InceptionV3 networks, using a total of 6469 chest X-ray images. The results showed that the four-category area under the curve (AUC)-ROC of COVID-19, general pneumonia, tuberculosis and normal was 99.60% (12). The literature on COVID-19 detection using CT images such as Garg et al. used a series of models such as EfficientNet, DenseNet, VGG, and ResNet with a total of 20 convolutional neural networks in 2022, using a total of 4173 CT images. The results show that the binary classification accuracy of EfficientNet-B5 in detecting COVID-19 and non-COVID-19 is 98.45% (13). Lahsaini et al. used transfer learning on VGG16, VGG19, Xception, InceptionV2, ResNet, DenseNet121, and DenseNet201 in 2021, combined with GradCam, using a total of 4986 CT images. The results show that in the binary classification of COVID-19 and non-COVID-19, DenseNet201+ GradCam achieves the best accuracy rate of 98.80% (14). Rahimzadeh et al. used ResNet50V2 as the backbone network and compared the model with ResNet50V2 and Xception after adding a feature pyramid network (FPN), using a total of 63,849 CT images. The results show that ResNet50V2+ FPN has an accuracy of 98.49% in COVID-19 and normal binary classification (15). Qi et al. first used five models of UNet, LinkNet, R2UNet, Attention UNet, and UNet++ to segment CT images, and then used pretrained DenseNet121, InceptionV3 and ResNet50 for classification, using a total of Over 10,000 CT images. The results show that in the binary classification of COVID-19 and CAO, LinkNet performs best in lung segmentation with a Dice coefficient of 0.9830, while DenseNet121 with capsule network has a prediction accuracy of 97.10% (16). Abdel-Basset et al. proposed a two-stage detection method in 2021. The first stage is to use the proposed GR-UNet to segment the area of lung infection, and then transfer learning is used as feature extraction; the second stage is to use the proposed GR-UNet to segment the lung infected area. The stage is to propose an infection prediction module that uses the infected location to make decisions about classification, using a total of 9,593 CT images. The results showed that the binary classification accuracy of COVID-19 and CAP was 96.80% (17). Ye et al. proposed a convolutional neural network named CIFD-Net in 2022, which can effectively handle the multi-region displacement problem through a new robust supervised learning, using a total of 45,167 CT images picture. The results showed that the binary classification accuracy of COVID-19 and non-COVID-19 was 91.19% (18). Balaha et al. used transfer learning models including ResNet50, ResNet101, VGG16, VGG19, Xception, MobileNetV1, MobileNetV2, DenseNet121, and DenseNet169 in 2021, and added the Harris Hawks optimization to optimize hyper-parameters, and finally use fast classification stage and compact stacking stage to stack the best models into one, using a total of 15,535 CT images. The results show that in the binary classification of COVID-19 and non-COVID-19, the weighted sum method (WSM) is used to obtain an accuracy of 99.33% (19). Qi et al. proposed a detection method named DR-MIL in 2021, which first treats a 3D CT image of a patient as a bag and selects ten CT slices as initial instances. The deep features were then extracted from the pretrained ResNet50 by fine-tuning and treated as a Deep Represented Instance Score. The bag with DRIS was input to K-Nearest Neighbor (KNN) to generate the final prediction, using a total of 241 patients CT images. The results showed that the binary classification accuracy of COVID-19 and CAP was 95.00% (20). Some scholars used both chest X-ray images and CT images to detect COVID-19. Related studies such as Ahamed et al. fine-tune the pretrained ResNet50V2 in 2021, using a total of 4593 chest X-ray images and 3000 CT images. The results showed that the four-category accuracy for detecting COVID-19, viral pneumonia, bacterial pneumonia, and normal in chest X-ray images was 96.45%; the three-category accuracy for COVID-19, bacterial pneumonia and normal was 97.24%; COVID-19 and normal binary classification accuracy was 98.95%. The three-category accuracy for detecting COVID-19, CAP, and normal in CT images was 99.01%; the two-category accuracy for COVID-19 and normal was 99.99% (21). Kumari et al. used four convolutional neural networks, InceptionV3, VGG16, Xception and ResNet50 in 2021, using a total of 2000 chest X-ray images and 2000 CT images. The results show that the binary classification accuracy of VGG16 in detecting COVID-19 and non-COVID-19 in chest X-ray images was 98.00%; the binary classification accuracy of Xception in detecting COVID-19 and non-COVID-19 in CT images was 83.00% (22). Ahsan et al. fine-tuned eight convolutional neural networks including VGG16, VGG19, ResNet15V2, InceptionResNetV2, ResNet50, DenseNet201, MobilenetV2, and NasNetMobile in 2020, using a total of 400 chest X-ray images and 400 CT images. The results show that the binary classification accuracy of NasNetMobile in detecting COVID-19 and non-COVID-19 in chest X-ray images was 100.00%; the binary classification accuracy of NasNetMobile in detecting COVID-19 and non-COVID-19 in CT images was 95.20% (23). Jia et al. proposed a dynamic CNN modification method in 2021 and applied it to a fine-tuned ResNet, and finally compared the results with VGG16, InceptionV3, ResNet18, DenseNet121, MobileNetV3, and SqueezeNet, using a total of 7,592 chests X-ray images and 104,009 CT images. The results showed that the five-category accuracy of detecting COVID-19, viral pneumonia, bacterial pneumonia, tuberculosis and normal in chest X-ray images was 99.60%; in CT images, detecting COVID-19, non-COVID-19 and normal three-category accuracy was 99.30% (24). Kassania et al. use eight models such as DenseNet, ResNet, MobileNet, InceptionV3, Xception, InceptionResNetV2, VGG and NASNet to extract features, and then extracted features were input to decision tree, random forest, XGBoost, AdaBoost, Bagging, and LightGBM. A total of 137 chest X-ray images and 137 CT images were used. The results show that DenseNet121+ Bagging combines chest X-ray images and CT images to detect COVID-19 and normal with a binary classification accuracy of 99.00% (25). Gour and Jain proposed an integrated stacked CNN in 2022. After fine-tuning VGG19 and Xception and generating five sub-models, all sub-models were stacked using a softmax classifier with a total of 3040 images Chest X-ray images and 4,645 CT images. The results showed that the three-category accuracy of detecting COVID-19, general pneumonia, and normal in chest X-ray images was 97.27%; and the two-category accuracy of detecting COVID-19 and normal in CT images was 98.30% (26). Kamil et al. used fine-tuned VGG19 in 2021, using a total of 977 chest X-ray images and 23 CT images. The results showed a 99.00% accuracy of the binary classification between COVID-19 and normal (27). Saygılı applied Bag of Tree, K-ELM, KNN, and SVM with a total of 1125 chest X-ray images and 3228 CT images. The results showed that SVM has an accuracy of 85.96% in detecting COVID-19, general pneumonia and normal in chest X-ray images; K-ELM is accurate in detecting COVID-19 and non-COVID-19 in CT images. The accuracy was 98.88% (28). Existing pretrained models are designed for general natural images and fine-tuned for classified image types, which is not specifically designed for COVID-19 detection. The general natural images are large and simple, while the images of COVID-19 have specific patterns and textures that differ significantly from natural images. Based on previous studies, we can find that the use of ensemble learning and the author's self-proposed convolutional neural network has good performance in detecting COVID-19 in chest X-ray images and CT images. Ensemble learning mainly combines multiple existing convolutional neural networks, which can not only reduce the probability of misjudgment by a single model, but also improve the classification accuracy in less time. Ensemble learning solves the current need to detect COVID-19 without designing a new model to get good detection performance in the most time-saving way. Challenging the possibility of improving accuracy on multiple chest disease diagnosis, this study proposes two classification models for multiple chest diseases including COVID-19. First, we obtained chest X-ray images and CT images from multiple public databases. Next, we select six pretraining models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M in the EfficientNetV2 series. The reason for choosing the previous models is the training speed and argument efficiency of EfficientNetV2 is better than some previous networks (29). To the best of our knowledge, this is the first study ensembles the series of EfficientNetV2 models for COVID-19 detection. In addition, this study proposes a self-designed model ECA-EfficientNetV2 based on ECA-Net and EfficientNetV2. MATERIALS AND METHODS Figure 1 represents the architecture of this study.• Step 1 Data extraction: The chest X-ray and CT images are collected. • Step 2 Image preprocessing: The size of all selected images is equalized and saved in PNG. The datasets are split into training, validation, and test subsets. • Step 3 Pretrained models: Six EfficientNetV2 models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, and EfficientNetV2-M are used. • Step 4 Proposed Ensemble-stacking model: The previous six EfficientNetV2 models are stacked. • Step 5 Proposed model: ECA-EfficientNetV2 model. • Step 6 Performance evaluation: Accuracy, precision, recall, F1-score, and AUC are recorded for each model. • Step 7 Comparison with SOTA: The results of models are compared to the related studies. Figure 1 The architecture of this study. (Color version of figure is available online.) Figure 1 Data Extraction Chest X-ray images and CT images are collected from five Kaggle datasets described as follows. The first Kaggle dataset is COVID-19 Radiography Dataset (30), which contains 3616 COVID-19, 1345 viral pneumonia, 6012 lung opacity and 10,192 normal for a total of 21,165 chest X-ray images (PNG). The second Kaggle dataset is Chest X-Ray Images (Pneumonia) (31), collected from pediatric patients aged 1 to 5 from Guangzhou Women and Children's Medical Center with 4273 pneumonia and 1583 normal chest X-ray images (JPEG). The third Kaggle dataset is Tuberculosis Chest X-ray Dataset (32), which contains 3500 tuberculosis and 3500 normal chest X-ray images (PNG). Large COVID-19 CT scan slice dataset (33), containing 7593 COVID-19, 2618 chest abdominal pelvis and 6893 normal CT images for a total of 17,104 CT images (PNG), is the forth Kaggle dataset used in this study. The fifth Kaggle dataset is COVID-19 and Normal and PneumoniaCT_Images (34), containing 2035 COVID-19, 3309 pneumonia and 2119 normal CT images for a total of 7463 CT images (PNG). The COVID-19 X-ray images was formed based on the previous datasets 1–3. The pneumonia images in the second dataset were split into groups of viral pneumonia and bacterial pneumonia. The total COVID-19 X-ray images contain five groups, which are 3616 COVID-19, 2780 viral pneumonia, 2838 bacterial pneumonia, 3500 Tuberculosis and 15,275 Normal for a total of 28,009 chest X-ray images (PNG). The COVID-19 CT images was formed based on the previous datasets 4 and 5, which contains four groups including 9628 COVID-19, 2618 Chest Abdomen Pelvis (CAP), 3309 pneumonia, and 9012 Normal for a total number of 24567 chest CT images. Image Preprocessing To equalize the number of images for each group, 1200 images are randomly selected for each group for both X-ray and CT datasets in this study. The number of images in training, validation and test subsets are 900, 100, and 200, respectively for X-ray dataset; The number of images in training, validation and test subsets are 4500, 500, and 400, respectively for CT dataset. Ten-fold cross validation was performed in this study. Tables 1 and 2 display the number of images for each group for COVID-19 X-ray dataset (Dataset X-ray) and COVID-19 CT dataset (Dataset CT) used in the study.Table 1 COVID-19 X-ray Dataset (Dataset X-ray) Table 1Group Training Validation Test COVID-19 900 100 200 Bacterial pneumonia 900 100 200 Tuberculosis 900 100 200 Viral pneumonia 900 100 200 Normal 900 100 200 Total 4500 500 1000 Table 2 COVID-19 CT Dataset (Dataset CT) Table 2Group Training Validation Test COVID-19 900 100 200 CAP 900 100 200 Pneumonia 900 100 200 Normal 900 100 200 Total 3600 400 800 The original images collected from the previous datasets differ in size and format. As the model requirement, all the images are preprocessed to in PNG with size of 224  ×  224. Figure 2, Figure 3 display the examples of images after preprocessed for chest X-ray (Dataset X-ray) and CT (Dataset CT) images, respectively.Figure 2 Chest X-ray images after preprocessing. (Color version of figure is available online.) Figure 2 Figure 3 CT images after preprocessing. (Color version of figure is available online.) Figure 3 Convolutional Neural Network Convolutional Neural Network (CNN) is a feedforward neural network, mainly composed of multiple convolution layers, pooling layers and fully connected layers. Compared to other neural networks, convolutional neural networks have better performances on image or speech recognitions. The goal of training a convolutional neural network is to find the most appropriate weights in the process of multiple forward and reverse iterations. Transfer learning is having previously trained model on a larger database that we can directly apply the architecture and weights of the pretrained model to various studies to speed the efficiency of the training model. The Application module in Keras currently provides about 40 pretrained models, all of which are trained on the ImageNet dataset. We choose the current newer and efficient EfficientNetV2, which includes six series of models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M. The reason for not choosing EfficientNetV2-L is that the model has a large number of parameters and cannot be executed in our existing hardware resources. EfficientNetV2 There are some problems with the previous EfficientNet series models, such as (1) when the size of the image is large, the training speed for EfficientNet-B3 to EfficientNet-B7 is very slow. (2) Training is slow when Depthwise convolutions is used. (3) It is not the best choice to enlarge each layer with the same magnification. Constantly increasing image size, which also leads to large memory consumption, which in turn affects training speed. Therefore, Tan and Le (35) proposed a new convolutional neural network EfficientNetV2 in 2021, which uses a nonuniform scaling strategy and can increase the number of deeper layers and has faster training speed and better parameter efficiency as compared to the previous models. In addition, Tan and Le (35) also used EfficientNet-B4 to do some experiments, and found that replacing the MBConv module with the Fused-MBConv module in the early stage can greatly improve the training speed. However, if each layer is replaced with the Fused-MBConv module, the number of parameters and FLOPs will be significantly increased, and the training speed will be relatively reduced. Therefore, Neural Architecture Search (NAS) was used to find the best combination of MBConv and Fused-MBConv. Figure 4 is the architecture of EfficientNetV2.Figure 4 Architecture of EfficientNetV2. (Color version of figure is available online.) Figure 4 ECA-Net Channel attention has greatly improved the performance of convolutional neural networks, and most scholars are currently working on developing more complex attention modules. The most representative method such as Hu et al. (36) used Squeeze-and-excitation (SE) module and proposed SE-Net in 2017. SE-Net first uses a global average pooling layer for each channel, and then uses two nonlinear fully connected layers and a sigma function to generate the weights for each channel. Although the SE module is widely used in some researches on the current channel attention module, it has been proved that dimensionality reduction will affect both the prediction performance for channel attention, and the efficiency of obtaining the weights between all channels. Wang et al. (37) uses a lightweight and efficient channel attention (ECA) module, which only adds a small number of parameters, but can achieve significant performance gains. The ECA module does not use dimensionality reduction and operates through a one-dimensional convolution of size k. According to the experimental results of ECA-Net on ImageNet-1K and MS COCO, ECA-Net has lower model complexity than the state-of-the-art methods. In addition, ECA-Net has better efficiency no matter in image classification, image segmentation or object detection. The Proposed Stacking-Ensemble Model Ensemble learning is a type of supervised learning that has been widely used in the fields of statistics, machine learning and deep learning. Compared to a single learning algorithm, the purpose of ensemble learning is to combine multiple algorithms or models to form a model with better predictive performance. Using an ensemble approach will yield better results when there are significant model-to-model differences (38,39) reviews on ensemble deep leaning. Details of ensemble classifiers with improved overfitting have been investigated by several studies, for instance (40,41). In recent years, due to the continuous improvement of computing power of computers, large-scale integrated models can be trained within a reasonable time, and have been applied on medical image recognition, face recognition, emotion recognition and financial decision-making, etc. Successful applications of ensemble classifiers could be seen in (42, 43, 44, 45). The main methods of ensemble learning can be divided into three categories: bagging, boosting and stacking. The main function of stacking is to combine multiple algorithms to make predictions, that is, the result integration of the voting method or the average method. Ensemble stacking is a powerful CNN ensemble method thanks to its unique meta-learning algorithm. Meta-learning algorithms work by taking the probabilities of the input from each sub-model and determining which model performs best in extracting features. The learning algorithms directly extend the learning of each sub-model and combine the best predictions from each model. If each model is unique, then each model learns differently. Stacking achieves better results than either trained model and is used to evaluate the error rate of bagging. This study chooses stacking as our approach for ensemble learning. The stacking-ensemble model developed in this study selects six pretrained models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M in the EfficientNetV2 series. Figure 5 presents the architecture of the proposed stacking-ensemble model in this study.Figure 5 Stacking-ensemble model. (Color version of figure is available online.) Figure 5 The Proposed ECA-EfficientNetV2 Model In addition to the proposed stacking-ensemble model, challenging the possibility of improving accuracy on multiple chest disease diagnosis, this study proposes a self-designed model ECA-EfficientNetV2 based on ECA-Net and EfficientNetV2. Motivation of choosing the previous two models as basis are described as follows:1. The advantages of EfficientNetV2 includes: (1) The network is better than some previous networks in terms of training speed and number of parameters; (2) It proposes improved incremental learning methods that dynamically adjust regularization (e.g., dropout, data augmentation, and mix-up) according to the size of the training images; (3) Progressive learning is used to perform well on pretrained ImageNet, CIFAR, cars, and flowers datasets. 2. The major advantages of ECA-Net are: (1) ECA is a lightweight attention module that only contains k parameters (k ≤ 9), which can be used to improve the performance of large convolutional neural networks; (2) The ECA module uses a dimensionality-free local cross-channel interaction method that adaptively selects suitable adjacent channels to compute attention; (3) The features extracted by the SE module between different channels are relatively similar, while the features extracted by the ECA module from different channels are different, which indicates that the features extracted by the ECA module are better for classification. Based on the previous two points, the proposed ECA-EfficientNetV2 are designed and introduced as follows:1. ECA-EfficientNetV2 uses the dilated convolution module as the first two layers. The advantage is that while the kernel size is increased, the parameters or calculation amount of the original model can be maintained. Each module contains two dilated convolutional layers and an activation function SELU, where the dilation rate of the first layer is set to two, and the dilation rate of the second layer is set to three. 2. To reduce both the number of parameters and the computation complexity, ECA-EfficientNetV2 replaces SE module with ECA module in MBConv and Fused-MBConv convolution modules, and renames as MBEConv and Fused-MBEConv. 3. Zero-padding is added to the convolutional layers in both the MBEConv and Fused-MBEConv modules to prevent the input image be affected by the kernel size. In addition, the original activation function SELU is changed to ReLU to overcome the problem of vanishing gradient. 4. After MBEConv and Fused-MBEConv, two general convolution modules are added. The internal parameters include Zero-padding, Stride, and SELU. In addition, we add a batch normalization after each convolutional layer, which makes training easier and more stable, and improves the performance of the neural network. 5. Use the global average pooling layer to improve the problem of a large number of parameters that occurs in the fully connected layer. After that, add a dropout layer before the classification layer to generate multiple results by continuously updating the weights, and finally remove the outliers to avoid the problem of overfitting. The number of parameters for ECA-EfficientNetV2 is 5,706,965, which is much less than the number of parameters (117,753,253) used in EfficientNetV2-L. Figure 6 presents the architecture of the proposed Fused-MBEConv layer and MBEConv layers, and Figure 7 shows the architecture of the proposed ECA-EfficientNetV2.Figure 6 The architecture of the proposed Fused-MBEConv layer and MBEConv layers. (H, W, C denotes network height, width and channel respectively.) (Color version of figure is available online.) Figure 6 Figure 7 Architecture of ECA-EfficientNetV2. (Color version of figure is available online.) Figure 7 Model Evaluation Measures The performance indices include accuracy, precision, recall, F1-Score and the AUC. The mathematical formulas of these indicators are shown in Eqs. (1–5), respectively. Accuracy is the ratio of the number of correctly classified samples to the total number of samples; Precision is the ratio of the number of true positives to the total number of elements labelled to the positive class; Recall presents the number of true positives divided by the total number of true positives and false negatives; F1-Score is a measure of precision and recall. The higher the F1-score, the better the classification performance of the model; AUC denotes the measure at distinguishing between the positive and negative classes. Higher the AUC, better the model.(1) Accuracyc=(TPc+TNc)(TPc+FPc+FNc+TNc) (2) Precisionc=TPc(TPc+FPc) (3) Recallc=TPc(TPc+FNc) (4) F1−score=2n∑c=1nRecallc×PrecisioncRecallc+Precisionc (5) AUC=∑cnrankn−TPc×(1+TPc)2TPc×TNc where c is the number of classes; TP (true positive) represents the number of positive categories that are correctly classified as positive, FP (false positive) represents the number of negative categories that are incorrectly classified as positive, TN (true negative) refers to the number of negative categories that are correctly classified as negative, and FN (false negative) refers to the number of positive categories that are incorrectly classified as negative. RESULTS The equipment used in the experiment is an Intel(R) Core(TM) i9-10900F 2.81 GHz CPU, NVIDIA GeForce RTX 3070 8G GPU. The whole experiment process is performed using Python 3.8 [Python Software Foundation, Fredericksburg, Virginia, USA], which contains Keras 2.6 and Tensorflow 2.6. Ten-fold cross validation is used to evaluate the performance of each model under batch size of 16; epochs count of 30; optimizer of Adam, learning rate of 1e-5 and dropout of 0.4. Performance Results on Dataset X-ray On dataset 1, Table 3, Table 4, Table 5, Table 6 report the results of Accuracies, Precisions, Recalls, F1-Scores, and AUCs in the test sets, and the average and standard deviation for EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models. The best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.21%, Precision of 99.23%, Recall of 99.25%, F1-score of 99.20%, and AUC of 99.51%. Figure 8, Figure 9 show the examples of training accuracy and loss, confusion matrix and ROC curve for each model, respectively.Table 3 Performance of EfficientNetV2-B0 and EfficientNetV2-B1 Models Table 3Fold EfficientNetV2-B0 EfficientNetV2-B1 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 82.10% 82.33% 82.14% 82.13% 88.81% 83.10% 83.23% 83.14% 83.11% 89.44% 2 85.20% 84.99% 85.24% 85.06% 90.75% 85.90% 85.81% 85.93% 85.85% 91.19% 3 88.50% 88.62% 88.53% 88.57% 92.81% 86.70% 86.69% 86.72% 86.59% 91.69% 4 85.50% 85.73% 85.51% 85.49% 90.94% 82.30% 82.26% 82.34% 82.25% 88.94% 5 85.90% 85.93% 85.91% 85.96% 91.19% 84.30% 84.47% 84.32% 84.34% 90.19% 6 84.40% 84.46% 84.42% 84.43% 90.25% 85.20% 85.15% 85.23% 85.16% 90.75% 7 85.60% 85.62% 85.64% 85.56% 91.00% 86.50% 86.40% 86.54% 86.42% 91.56% 8 84.40% 84.59% 84.46% 84.34% 90.25% 85.82% 85.94% 85.81% 85.84% 91.13% 9 84.50% 84.80% 84.51% 84.54% 90.31% 88.90% 89.01% 88.92% 88.93% 93.06% 10 86.20% 86.42% 86.23% 86.22% 91.38% 87.60% 87.79% 87.64% 87.65% 92.25% Average 85.23± 1.63% 85.35± 1.61% 85.26± 1.63% 85.23± 1.65% 90.77± 1.02% 85.63± 2.00% 85.68± 2.02% 85.66± 2.00% 85.61± 2.01% 91.02± 1.25% Table 4 Performance of EfficientNetV2-B2 and EfficientNetV2-B3 Models Table 4Fold EfficientNetV2-B2 EfficientNetV2-B3 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 79.90% 79.85% 79.91% 79.76% 87.44% 85.40% 85.72% 85.43% 85.37% 90.88% 2 84.40% 84.63% 84.42% 84.30% 90.25% 86.30% 86.19% 86.32% 86.22% 91.44% 3 86.60% 87.08% 86.61% 86.43% 91.69% 87.00% 87.27% 87.42% 86.97% 91.88% 4 81.70% 81.71% 81.74% 81.63% 88.56% 84.20% 84.06% 84.23% 84.11% 90.12% 5 83.70% 83.80% 83.72% 83.74% 89.81% 85.70% 85.84% 85.71% 85.69% 91.06% 6 80.60% 80.98% 80.64% 80.59% 87.88% 84.60% 84.94% 84.62% 84.61% 90.38% 7 83.20% 83.14% 83.23% 83.15% 89.50% 86.40% 86.67% 86.42% 86.38% 91.50% 8 87.30% 87.36% 87.34% 87.27% 92.06% 85.20% 85.49% 85.21% 85.27% 90.75% 9 80.60% 80.89% 80.64% 80.65% 87.88% 87.10% 87.22% 87.14% 87.12% 91.94% 10 85.10% 85.31% 85.14% 85.15% 90.69% 85.90% 86.26% 85.94% 85.92% 91.19% Average 83.31± 2.59% 83.48± 2.63% 83.34± 2.59% 83.27± 2.58% 89.58± 1.62% 85.78± 0.96% 85.97± 0.99% 85.84± 1.02% 85.77± 0.96% 91.11± 0.60% Table 5 Performance of EfficientNetV2-S and EfficientNetV2-M Models Table 5Fold EfficientNetV2-S EfficientNetV2-M Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 86.90% 86.82% 86.94% 86.81% 91.81% 87.30% 87.29% 87.34% 87.21% 92.06% 2 88.30% 88.31% 88.34% 88.27% 92.69% 88.70% 88.74% 88.71% 88.68% 92.64% 3 87.70% 88.01% 87.92% 87.95% 92.44% 89.00% 89.04% 89.27% 88.97% 93.13% 4 87.00% 87.37% 87.42% 87.03% 91.88% 87.70% 87.65% 87.72% 87.67% 92.31% 5 87.50% 87.55% 87.52% 87.48% 92.19% 89.40% 89.43% 89.41% 89.44% 93.38% 6 87.00% 86.93% 87.12% 86.91% 91.88% 89.70% 89.71% 89.73% 89.65% 93.56% 7 87.10% 87.13% 87.21% 87.05% 91.94% 89.00% 88.90% 89.29% 88.94% 93.13% 8 88.00% 88.03% 88.28% 87.98% 92.50% 89.70% 89.72% 89.74% 89.69% 93.56% 9 87.10% 87.13% 87.14% 87.11% 91.94% 86.70% 86.98% 86.74% 86.76% 91.69% 10 87.50% 87.69% 87.53% 87.56% 92.19% 89.30% 89.33% 89.31% 89.34% 93.31% Average 87.41± 0.47% 87.50± 0.51% 87.54± 0.49% 87.42± 0.51% 92.15± 0.31% 88.65± 1.05% 88.68± 1.01% 88.73± 1.07% 88.64± 1.05% 92.88± 0.66% Table 6 Performance of Stacking-Ensemble and ECA-EfficientNetV2 Models Table 6Fold Proposed Stacking-Ensemble Model Proposed ECA-EfficientNetV2 Accuracy Precision Recall F1-score AUC Accuracy Precision Recall F1-Score AUC 1 98.40% 98.42% 98.47% 98.39% 99.00% 98.80% 98.84% 98.81% 98.79% 99.25% 2 98.70% 98.71% 98.74% 98.72% 99.19% 99.60% 99.64% 99.61% 99.59% 99.75% 3 98.90% 98.91% 98.93% 98.94% 99.31% 98.80% 98.82% 98.84% 98.78% 99.25% 4 98.50% 95.54% 98.51% 98.53% 99.06% 99.50% 99.51% 99.54% 99.49% 99.69% 5 99.00% 99.01% 99.04% 98.99% 99.38% 99.10% 99.12% 99.11% 99.09% 99.44% 6 99.00% 99.02% 99.05% 98.99% 99.38% 99.40% 99.41% 99.44% 99.37% 99.62% 7 99.00% 99.05% 99.03% 99.01% 99.38% 99.50% 99.48% 99.54% 99.53% 99.69% 8 98.30% 98.35% 98.31% 98.29% 98.94% 99.00% 99.04% 99.07% 98.99% 99.38% 9 98.60% 98.64% 98.67% 98.62% 99.13% 99.00% 99.02% 99.07% 98.97% 99.38% 10 98.70% 98.73% 98.71% 98.75% 99.19% 99.40% 99.41% 99.44% 99.38% 99.62% Average 98.71± 0.26% 98.44± 1.05% 98.75± 0.26% 98.72± 0.26% 99.20± 0.16% 99.21± 0.30% 99.23± 0.30% 99.25± 0.30% 99.20± 0.31% 99.51± 0.19% Figure 8 Examples of training accuracy and loss from each model. (Color version of figure is available online.) Figure 8 Figure 9 Examples of confusion matrix and ROC curve from each model. (Color version of figure is available online.) Figure 9 Performance Results on Dataset CT On dataset 2, Table 7, Table 8, Table 9, Table 10 report the results of Accuracies, Precisions, Recalls, F1-Scores, and AUCs in the test sets, and the average and standard deviation for EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models. The best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.81%, Precision of 99.80%, Recall of 99.80%, F1-score of 99.81%, and AUC of 99.87%. Figure 10, Figure 11 show the examples of training accuracy and loss, confusion matrix and ROC curve for each model, respectively.Table 7 Performance of EfficientNetV2-B0 and EfficientNetV2-B1 Models Table 7Fold EfficientNetV2-B0 EfficientNetV2-B1 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 95.13% 95.22% 95.15% 95.10% 96.75% 93.63% 93.65% 93.62% 93.59% 95.75% 2 93.87% 93.88% 93.85% 93.86% 95.92% 96.25% 96.41% 96.25% 96.24% 97.50% 3 95.63% 95.60% 95.62% 95.59% 97.08% 95.13% 95.32% 95.15% 95.07% 96.75% 4 95.13% 95.09% 95.15% 95.08% 96.75% 93.25% 93.20% 93.25% 93.17% 95.50% 5 95.63% 95.78% 95.62% 95.60% 97.08% 93.00% 93.03% 93.17% 92.88% 95.33% 6 94.25% 94.29% 94.27% 94.23% 96.17% 93.25% 93.37% 93.27% 93.28% 95.50% 7 93.25% 93.43% 93.28% 93.13% 95.50% 94.13% 94.35% 94.17% 94.04% 96.08% 8 94.37% 94.39% 94.35% 94.36% 96.25% 90.87% 90.85% 90.88% 90.77% 93.92% 9 93.50% 93.51% 93.54% 93.47% 95.67% 95.13% 95.11% 95.15% 95.17% 96.75% 10 94.63% 94.60% 94.56% 94.61% 96.42% 91.37% 91.27% 91.75% 91.27% 94.25% Average 94.54± 0.84% 94.58± 0.83% 94.54± 0.83% 94.50± 0.85% 96.36± 0.56% 93.60± 1.67% 93.66± 1.74% 93.67± 1.61% 93.55± 1.70% 95.73± 1.11% Table 8 Performance of EfficientNetV2-B2 and EfficientNetV2-B3 Models Table 8Fold EfficientNetV2-B2 EfficientNetV2-B3 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 94.87% 98.88% 94.85% 94.83% 96.58% 94.37% 94.54% 94.38% 94.36% 96.25% 2 97.00% 96.99% 97.41% 96.87% 98.00% 95.63% 95.65% 95.62% 95.61% 97.08% 3 94.87% 94.92% 94.88% 94.84% 96.58% 94.87% 95.02% 94.85% 94.83% 96.58% 4 95.00% 94.97% 95.02% 94.94% 96.67% 95.13% 95.10% 95.12% 95.08% 96.75% 5 95.00% 94.99% 95.17% 94.97% 96.67% 96.37% 96.39% 96.38% 96.36% 97.58% 6 92.87% 92.95% 92.88% 92.85% 95.25% 95.75% 95.74% 95.75% 95.73% 97.17% 7 92.13% 92.19% 92.15% 92.03% 94.75% 95.00% 95.21% 95.14% 94.97% 96.67% 8 94.63% 94.58% 94.65% 94.59% 96.42% 95.50% 95.54% 95.51% 95.48% 97.00% 9 95.25% 95.24% 95.19% 95.24% 96.83% 94.37% 94.36% 94.35% 94.76% 96.25% 10 92.50% 92.45% 92.47% 92.38% 95.00% 94.00% 93.98% 94.37% 93.97% 96.00% Average 94.41± 1.48% 94.82± 2.04% 94.47± 1.57% 94.35± 1.49% 96.28± 0.99% 95.10± 0.73% 95.15± 0.72% 95.15± 0.68% 95.12± 0.70% 96.73± 0.49% Table 9 Performance of EfficientNetV2-S and EfficientNetV2-M Models Table 9Fold EfficientNetV2-S EfficientNetV2-M Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 96.37% 96.36% 96.35% 96.63% 97.58% 95.25% 95.22% 95.27% 95.22% 96.83% 2 96.63% 96.71% 96.62% 96.65% 97.75% 96.13% 96.14% 96.15% 96.10% 97.42% 3 97.38% 97.40% 97.37% 97.39% 98.25% 95.50% 95.54% 95.51% 95.47% 97.00% 4 96.63% 96.62% 96.65% 96.33% 97.75% 95.13% 95.12% 95.15% 95.10% 96.75% 5 96.50% 96.47% 96.54% 96.48% 97.67% 96.75% 96.76% 96.73% 96.78% 97.83% 6 95.75% 95.74% 95.77% 95.78% 97.17% 96.00% 96.05% 96.04% 95.99% 97.33% 7 96.50% 96.58% 96.54% 96.51% 97.67% 95.25% 95.28% 95.26% 95.24% 96.83% 8 96.88% 96.86% 96.87% 96.89% 97.92% 96.50% 96.51% 95.54% 96.49% 97.67% 9 95.63% 95.67% 95.65% 95.62% 97.08% 95.37% 95.33% 95.35% 95.34% 96.92% 10 95.13% 95.09% 95.12% 95.10% 96.75% 94.75% 94.70% 94.77% 94.72% 96.50% Average 96.34± 0.66% 96.35± 0.67% 96.35± 0.66% 96.34± 0.67% 97.56± 0.44% 95.66± 0.65% 95.67± 0.67% 95.58± 0.57% 95.65± 0.66% 97.11± 0.43% Table 10 Performance of Stacking-ensemble and ECA-EfficientNetV2 Models Table 10Fold Stacking-ensemble model ECA-EfficientNetV2 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 99.00% 99.01% 99.24% 99.89% 99.33% 99.38% 99.34% 99.35% 99.37% 99.58% 2 98.62% 98.67% 98.63% 98.54% 99.08% 99.88% 99.87% 99.85% 99.89% 99.92% 3 99.38% 99.46% 99.74% 99.25% 99.58% 99.75% 99.72% 99.71% 99.77% 99.83% 4 98.75% 98.74% 98.81% 98.93% 99.17% 99.25% 99.27% 99.24% 99.28% 99.50% 5 98.75% 98.78% 98.87% 98.79% 99.17% 99.88% 99.85% 99.87% 99.84% 99.92% 6 99.88% 99.87% 99.85% 99.89% 99.92% 99.94% 99.92% 99.96% 99.95% 99.98% 7 98.25% 98.33% 98.41% 98.24% 98.83% 100.00% 100.00% 100.00% 100.00% 100.00% 8 98.75% 98.77% 98.75% 98.72% 99.17% 100.00% 100.00% 100.00% 100.00% 100.00% 9 98.88% 98.87% 98.85% 98.86% 99.25% 100.00% 100.00% 100.00% 100.00% 100.00% 10 98.13% 98.16% 98.15% 98.12% 98.75% 100.00% 100.00% 100.00% 100.00% 100.00% Average 98.84± 0.51% 98.87± 0.50% 98.93± 0.54% 98.92± 0.61% 99.23± 0.34% 99.81± 0.27% 99.80± 0.27% 99.80± 0.28% 99.81± 0.27% 99.87± 0.18% Figure 10 Examples of training accuracy and loss from each model. (Color version of figure is available online.) Figure 10 Figure 11 Examples of confusion matrix and ROC curve from each model. (Color version of figure is available online.) Figure 11 DISCUSSION Dataset X-ray The performances of 5 metrics among EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models for dataset X-ray were compared in Table 11 . Two accuracies are above 98%, which come from Stacking-ensemble and ECA-EfficientNetV2 models, while the accuracies from six pretrained models are inferior. In fact, the highest two performance for five metrics from Stacking-ensemble and ECA-EfficientNetV2 models are significantly different from six pretrained models (p-value < 0.01). Although the best performance, the accuracy (99.21%), precision (99.23%), recall (99.25%), F1-score (99.20%), and AUC (99.51%), comes from the proposed ECA-EfficientNetV2 model, the differences for five metrics between Stacking-ensemble and ECA-EfficientNetV2 models are not significant. Comparing the standard deviations of five metrics from Stacking-ensemble and ECA-EfficientNetV2 models, those from the latter are relatively small, which stands the stability of the model.Table 11 Performance Comparison on Dataset X-ray Table 11Models Dataset X-ray Accuracy (%) Precision (%) Recall (%) F1-Score (%) AUC (%) EfficientNetV2-B0 85.23(±1.63) 85.35(±1.61) 85.26(±1.63) 85.23(±1.65) 90.77(±1.02) EfficientNetV2-B1 85.63(±2.00) 85.68(±2.02) 85.66(±2.00) 85.61(±2.01) 91.02(±1.25) EfficientNetV2-B2 83.31(±2.59) 83.48(±2.63) 83.34(±2.59) 83.27(±2.58) 89.58(±1.62) \ EfficientNetV2-B3 85.78(±0.96) 85.97(±0.99) 85.84(±1.02) 85.77(±0.96) 91.11(±0.60) EfficientNetV2-S 87.41(±0.47) 87.50(±0.51) 87.54(±0.49) 87.42(±0.51) 92.15(±0.31) EfficientNetV2-M 88.65(±1.05) 88.68(±1.01) 88.73(±1.07) 88.64(±1.05) 92.88(±0.66) Stacking-ensemble 98.71(±0.26) 98.44(±1.05) 98.75(±0.26) 98.72(±0.26) 99.20(±0.16) ECA-EfficientNetV2 99.21(±0.30) 99.23(±0.30) 99.25(±0.30) 99.20(±0.31) 99.51(±0.19) p-value 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ ⁎⁎ p < 0.01 Figure 12 shows examples of confusion matrix from EfficientNetV2-B0, EfficientNetV2-M, Stacking-ensemble, and ECA-EfficientNetV2 models. In EfficientNetV2-B0, 58 Bacterial Pneumonia are misclassified to Tuberculosis, and 42 Tuberculosis are misclassified to Bacterial Pneumonia. In EfficientNetV2-M, 40 Bacterial Pneumonia are misclassified to Tuberculosis, and 51 Tuberculosis are misclassified to Bacterial Pneumonia. Obviously, Bacterial Pneumonia and Tuberculosis are misclassified to each other from EfficientNetV2-B0 and EfficientNetV2-M models, while the wrong phenomenon is much improved in Stacking-ensemble and ECA-EfficientNetV2 models.Figure 12 Confusion matrix of Dataset X-ray. (Color version of figure is available online.) Figure 12 Dataset CT The performances of 5 metrics among EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models for dataset CT were compared in Table 12 . The accuracies from the previous eight models are greater than 93%; the highest two are from Stacking-ensemble and ECA-EfficientNetV2 models. Actually, the same results could be found in precision, recall, F1-score and AUC. The five metrics from Stacking-ensemble and ECA-EfficientNetV2 models are statistically significant from the six EfficientNetV2 models (p-value < 0.01). The best performance, the accuracy (99.81%), precision (99.80%), recall (99.80%), F1-score (99.81%), and AUC (99.87%), comes from the proposed ECA-EfficientNetV2, which are not significantly different from the Stacking-ensemble model. Same as we found in Dataset X-ray, the standard deviations of five metrics from ECA-EfficientNetV2 model are relatively small, which stands the stability of the model.Table 12 Performance Comparison on Dataset CT Table 12Models Dataset CT Accuracy (%) Precision (%) Recall (%) F1-Score (%) AUC (%) EfficientNetV2-B0 94.54(±0.84) 94.58(±0.83) 94.54(±0.83) 94.50(±0.85) 96.36(±0.56) EfficientNetV2-B1 93.60(±1.67) 93.66(±1.74) 93.67(±1.61) 93.55(±1.70) 95.73(±1.11) EfficientNetV2-B2 94.41(±1.48) 94.82(±2.04) 94.47(±1.57) 94.35(±1.49) 96.28(±0.99) EfficientNetV2-B3 95.10(±0.73) 95.15(±0.72) 95.15(±0.68) 95.12(±0.70) 96.73(±0.49) EfficientNetV2-S 96.34(±0.66) 96.35(±0.67) 96.35(±0.66) 96.34(±0.67) 97.56(±0.44) EfficientNetV2-M 95.66(±0.65) 95.67(±0.67) 95.58(±0.57) 95.65(±0.66) 97.11(±0.43) Stacking-ensemble 98.84(±0.51) 98.87(±0.50) 98.93(±0.54) 98.92(±0.61) 99.23(±0.34) ECA-EfficientNetV2 99.81(±0.27) 99.80(±0.27) 99.80(±0.28) 99.81(±0.27) 99.87(±0.18) p-value 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ ⁎⁎ p < 0.01 Figure 13 shows examples of confusion matrix from EfficientNetV2-B0, EfficientNetV2-M, Stacking-ensemble, and ECA-EfficientNetV2 models for Dataset CT. In EfficientNetV2-B0, 23 COVID-19 are misclassified to Normal, and 8 Normal are misclassified to COVID-19. In EfficientNetV2-M, 18 COVID-19 are misclassified to Normal, and 12 Normal are misclassified to COVID-19. In Stacking-ensemble, 7 COVID-19 are misclassified to Normal, and 1 Normal are misclassified to COVID-19. In ECA-EfficientNetV2, 2 COVID-19 are misclassified to Normal, and 0 Normal are misclassified to COVID-19. Obviously, COVID-19 and Normal are misclassified to each other from EfficientNetV2-B0 and EfficientNetV2-M models, while Stacking-ensemble and ECA-EfficientNetV2 models greatly improved the wrong phenomenon.Figure 13 Confusion matrix of Dataset CT. (Color version of figure is available online.) Figure 13 Comparison With SOTA The results of the proposed Stacking-ensemble and ECA-EfficientNetV2 models are compared to the related studies using stacking ensemble models for COVID-19 diagnosis in Table 13 . Most of the related studies worked on classification of 2 to 3 groups, while this study focuses on classification of five groups in X-ray and classification of 4 groups in CT. Most of the accuracies are greater than 90% by using stacking models, which represents the favorable performance of stacking method.Table 13 Comparison of the Proposed Models With SOTA Table 13No. Study(s) Dataset Architecture Class Accuracy 1 Ieracitano et al. (2022) (5) 155 X-ray images. CovNNet 2 81.00% 2 Khan et al. (2021) (6) 6448 X-ray images. DHL DBHL 2 98.53% 3 Loey et al. (2022) (7) 10,848 X-ray images. CNN+ Bayesian 3 96.00% 4 Hu et al. (2021) (8) 10,848 X-ray images. CNN+ ELMs+ ChOA 2 99.11% 5 Thaseen et al. (2022) (9) 13,808 X-ray images. Ensemble CNN model 3 99.00% 6 Kumar et al. (2021) (10) 6000 X-ray images. HCNN 3 98.20% 7 Musallam et al. (2022) (11) 7512 X-ray images. DeepChest 3 96.56% 8 Mamalakis et al. (2021) (12) 6000 X-ray images. DenResCov-19 4 99.60% 9 Garg et al. (2022) (13) 4173 CT images. EfficientNet-B5 2 98.45% 10 Lahsaini et al. (2021) (14) 4986 CT images. DenseNet201+GradCam 2 98.80% 11 Rahimzadeh et al. (2021) (15) 63,849 CT images. ResNet50V2+ FPN 2 98.49% 12 Qi et al. (2022) (16) 10,000 CT images. UNet+ DenseNet121 2 97.10% 13 Abdel-Basset et al. (2021) (17) 9593 CT images. GR-UNet+ CNN 2 96.80% 14 Ye et al. (2022) (18) 45,167 CT images. CIFD-Net 2 91.19% 15 Balaha et al. (2021) (19) 15,535 CT images. CNN+HHO+FCS+CSS+WSM 2 99.33% 16 Ahamed et al. (2021) (21) 4593 X-ray images. ResNet50V2(fine-tuning) 4 96.45% 3000 CT images. 3 99.01% 17 Kumari et al. (2021) (22) 2000 X-ray images. VGG16 2 98.00% 2000 CT images. Xception 2 83.00% 18 Ahsan et al. (2020) (23) 400 X-ray images. NasNetMobile 2 100.00% 400 CT images. 2 95.20% 19 Jia et al. (2021) (24) 7592 X-ray images. ResNet(fine-tuning) 5 99.60% 104,009 CT images. 3 99.30% 20 Kassania et al. (2021) (25) 137 X-ray images. DenseNet121+ Bagging 2 99.00% 137 CT images. 21 Gour and Jain (2022) (26) 3040 X-ray images. Ensemble CNN 3 97.27% 4645 CT images. 2 98.30% 22 Kamil et al. (2021) (27) 977 X-ray images. VGG19(fine-tuning) 2 99.00% 23 CT images. 23 Saygılı (2021) (28) 1125 X-ray images. SVM 3 85.96% 3228 CT images. K-ELM 2 98.88% 24 Stacking-ensemble 6000 X-ray images. Stacking-ensemble 5 98.71% 4800 CT images. 4 98.84% 25 ECA-EfficientNetV2 6000 X-ray images. ECA-EfficientNetV2 5 99.21% 4800 CT images. 4 99.81% Although the accuracy from our Stacking-ensemble model is slightly inferior than (8,9,12,19,21,23,24,25,27,28). As we have mentioned before, most of those related studies worked on classification of two groups, while this study is working on the classification of four groups in X-ray and five groups in CT. Even in this situation, the accuracies of our proposed Stacking-ensemble model are pretty close to the previous studies. In addition, the performance of the proposed ECA-EfficientNetV2 model dominates most of the related studies. The accuracies for X-ray and CT from ECA-EfficientNetV2 model are close to 100%, which illustrates the great classification capability for multiple groups on chest diseases. To verify the performance of the two proposed Stacking-ensemble and ECA-EfficientNetV2 models, one more open public dataset COVID-CT (https://github.com/UCSD-AI4H/COVID-CT) was tested and the performance metrics were displayed in Table 14 . There are 746 CT images including two groups of COVID-19 and Non-COVID-19 in dataset COVID-CT. Except the accuracy of 93.33% from Shaik and Cherukuri (53), the accuracies from other related studies are lower than 90%. The accuracies are 94.86% and 95.29% from our proposed Stacking-ensemble and ECA-EfficientNetV2 models, respectively, which are higher than all of the related studies compared in Table 14. Especially, the number of parameter in ECA-EfficientNetV2 model is 5,706,965, which is much less than those in (48,50,51).Table 14 Comparisons on COVID-CT Dataset Table 14Dataset Study(s) Accuracy Precision Recall F1-Score Parameter COVID-CT Mishra et al. (2020) (46) 88.30% — — 86.70% — Saqib et al. (2020) (47) 80.30% 78.20% 85.70% 81.80% — He et al. (2020) (48) 86.00% — — 85.00% 14,149,480 Mobiny et al. (2020) (49) — 84.00% — — — Polsinelli et al. (2020) (50) 85.00% 85.00% 87.00% 86.00% 12,600,000 Yang et al. (2020) (51) — — 89.10% — 25,600,000 Cruz (2021) (52) 86.00% — 89.00% 85.00% — Shaik and Cherukuri (2021) (53) 93.33% 93.17% 93.54% 93.29% — Stacking-ensemble 94.86% 94.79% 94.83% 94.84% — ECA-EfficientNetV2 95.29% 95.15% 95.24% 95.27% 5706,965 CONCLUSION This study applies six pretrained MobileNetV2 models on COVID-19 diagnosis, and stacks ensemble the previous six models to achieve better classification results. The self-designed Stacking-ensemble and ECA-MobileNetV2 models were proposed to classify multiple chest diseases including COVID-19 in this study. Classification was executed on five groups using X-ray images and four groups using CT images. The experimental results of two proposed models were compared to six pretrained MobileNetV2 models. The proposed Stacking-ensemble achieves accuracies of 98.71% and 98.84% on X-ray and CT datasets, respectively. With stability, the proposed ECA-MobileNetV2 achieves the highest accuracies of 99.21% and 99.81% on X-ray and CT datasets, respectively. The proposed Stacking-ensemble model is the first study ensembles the series of EfficientNetV2 models on multiple chest diseases including COVID-19 detection, which reduce prediction variance to the training data and improves overall classification performance when compared to any individual EfficientNetV2 model. As compared to the proposed ECA-MobileNetV2 model, the stacking model is computationally expensive; it takes around 1.5 hours in this experiment while the execution time is only 17 minutes in ECA-MobileNetV2 model. The major contribution of the study is the proposed ECA-MobileNetV2 model combines the advantages of EfficientNetV2 and ECA-Net to demonstrate superior classification performance with least variance and training time. The results of this study imply that the architecture of the proposed ECA-MobileNetV2 model can be used to assist radiologists in the X-ray and CT scans on multiple chest diseases including COVID-19. The 6,000 X-ray and 4800 CT images used in this study were collected from five Kaggle datasets. More chest images are encouraged to achieve more robust classification performance. Since the related studies compared in the study were searched from Scopus and were limited, more comprehensive review is suggested in the future research. ==== Refs REFERENCES 1 C. Pneumonia, “COVID-19,” vol. 2019, 2019. 2 Liu H. Wang L. Nan Y. SDFN: Segmentation-based deep fusion network for thoracic disease classification in chest X-ray images Computerized Med Imaging and Graphics 75 2019 66 73 10.1016/j.compmedimag.2019.05.005 3 Lin L. Qin L. Xu Z. Artificial intelligence distinguishes COVID-19 from community acquired pneumonia on Chest CT Appl Intel 2019 2020 1 5 [Online]. Available: http://arxiv.org/abs/2003.13865 (accessed February 01, 2022) 4 Sharma N. Sharma R. Jindal N. Machine learning and deep learning applications-a vision Global Transitions Proceedings 2021 0 8 10.1016/j.gltp.2021.01.004 5 Ieracitano C. Mammone N. Versaci M. A fuzzy-enhanced deep learning approach for early detection of Covid-19 pneumonia from portable chest X-ray images Neurocomputing 481 2022 202 215 10.1016/j.neucom.2022.01.055 35079203 6 Khan S.H. Sohail A. Khan A. COVID-19 detection in chest X-ray images using deep boosted hybrid learning Comp Biol Med 137 2021 104816 10.1016/j.compbiomed.2021.104816 7 Loey M. El-Sappagh S. Mirjalili S. Bayesian-based optimized deep learning model to detect COVID-19 patients using chest X-ray image data Comp Biol Med 142 2022 105213 10.1016/j.compbiomed.2022.105213 8 Hu T. Khishe M. Mohammadi M. Realtime COVID-19 diagnosis from X-Ray images using deep CNN and extreme learning machines stabilized by chimp optimization algorithm Biomed Signal Proces Contrl 68 2021 102764 10.1016/j.bspc.2021.102764 9 Sakthivel R. Sumaiya Thaseen I. Vanithac M. An efficient hardware architecture based on an ensemble of deep learning models for COVID-19 prediction Sustainable Cities and Soc Feb. 2022 103713 10.1016/j.scs.2022.103713 10 Kumar M. Shakya D. Kurup V. COVID-19 prediction through X-ray images using transfer learning-based hybrid deep learning approach Materials Today: Proceedings Dec. 2021 10.1016/j.matpr.2021.12.123 11 Musallam A.S. Sherif A.S. Hussein M.K. Efficient framework for detecting COVID-19 and pneumonia from chest X-Ray using deep convolutional network Egyptian Info J 2021 2022 10.1016/j.eij.2022.01.002 12 Mamalakis M. Swift A.J. Vorselaars B. DenResCov-19: A deep transfer learning network for robust automatic classification of COVID-19, pneumonia, and tuberculosis from X-rays Computerized Med Imaging and Graphics 94 2021 102008 10.1016/j.compmedimag.2021.102008 13 Garg A. Salehi S. la Rocca M. Efficient and visualizable convolutional neural networks for COVID-19 classification using Chest CT Expert Systems with Appl 195 2022 116540 10.1016/j.eswa.2022.116540 14 LAHSAINI I. EL HABIB DAHO M. CHIKH M.A. Deep transfer learning based classification model for covid-19 using chest CT-scans Pattern Recog Letters 152 2021 122 128 10.1016/j.patrec.2021.08.035 15 Rahimzadeh M. Attar A. Sakhaei S.M. A fully automated deep learning-based network for detecting COVID-19 from a new and large lung CT scan dataset Biomed Signal Proces Contrl 68 2021 10.1016/j.bspc.2021.102588 16 Qi Q.Q. Qi S.L. Wu Y.N. Fully automatic pipeline of convolutional neural networks and capsule networks to distinguish COVID-19 from community-acquired pneumonia via CT images Comp in Biol Med 141 2022 10.1016/j.compbiomed.2021.105182 17 Abdel-Basset M. Hawash H. Moustafa N. Two-stage deep learning framework for discrimination between COVID-19 and community-acquired pneumonia from chest CT scans Pattern Recog Letters 152 2021 311 319 10.1016/j.patrec.2021.10.027 18 Ye Q.H. Gqu Y. Ding W.P. Robust weakly supervised learning for COVID-19 recognition using multi-center CT images Appl Soft Computing 116 2022 108291 10.1016/j.asoc.2021.108291 19 Balaha H.M. El-Gendy E.M. Saafan M.M. CovH2SD: a COVID-19 detection approach based on Harris Hawks optimization and stacked deep learning Expert Systems with Appl 186 2021 10.1016/j.eswa.2021.115805 20 Qi S.L. Xu C.W. Li C. DR-MIL: deep represented multiple instance learning distinguishes COVID-19 from community-acquired pneumonia in CT images Comp Methods and Programs in Biomed 211 2021 10.1016/j.cmpb.2021.106406 21 Ahamed K.U. Islam M. Uddin A. A deep learning approach using effective preprocessing techniques to detect COVID-19 from chest CT-scan and X-ray images Comp Biol Med 2021 105014 10.1016/j.compbiomed.2021.105014 22 Kumari S. Ranjith E. Gujjar A. Comparative analysis of deep learning models for COVID-19 detection Global Trans Proc 2 2 2021 559 565 10.1016/j.gltp.2021.08.030 23 M. M. Ahsan, K. D. Gupta, M. M. Islam, et al., “Study of different deep learning approach with explainable AI for screening patients with COVID-19 symptoms: using CT scan and chest X-ray image dataset,” 2020, doi:10.3390/make2040027. 24 Jia G. Lam H.K. Xu Y. Classification of COVID-19 chest X-Ray and CT images using a type of dynamic CNN modification method Computers in Biol and Med 134 April 2021 104425 10.1016/j.compbiomed.2021.104425 25 Kassania S.H. Kassanib P.H. Wesolowskic M.J. Automatic detection of coronavirus disease (COVID-19) in X-ray And CT images: a machine learning based approach Biocybernetics and Biomed Eng 41 3 2021 867 879 10.1016/j.bbe.2021.05.013 26 Gour M. Jain S. Automated COVID-19 detection from X-ray and CT images with stacked ensemble convolutional neural network Biocybernetics and Biomed Eng 42 1 2022 27 41 10.1016/j.bbe.2021.12.001 27 Kamil M.Y. A deep learning framework to detect Covid-19 disease via chest X-ray and CT scan images Int J Electrical and Comp Eng 11 1 2021 844 850 10.11591/ijece.v11i1.pp844-850 28 Saygılı A. A new approach for computer-aided detection of coronavirus (COVID-19) from CT and X-ray images using machine learning methods Appl Soft Computing 105 2021 107323 10.1016/j.asoc.2021.107323 29 M. Tan and Q. v. Le, “EfficientNetV2: smaller models and faster training,” 2021, [Online]. Available: http://arxiv.org/abs/2104.00298. (accessed December 01, 2021) 30 “COVID-19 Radiography Database | Kaggle.” https://www.kaggle.com/tawsifurrahman/covid19-radiography-database (accessed Nov. 04, 2021). 31 “Chest X-Ray Images (Pneumonia) | Kaggle.” https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia (accessed Nov. 04, 2021). 32 “Tuberculosis (TB) chest x-ray database | Kaggle.” https://www.kaggle.com/datasets/tawsifurrahman/tuberculosis-tb-chest-xray-dataset (accessed May 06, 2022). 33 “Large COVID-19 CT scan slice dataset | Kaggle.” https://www.kaggle.com/maedemaftouni/large-covid19-ct-slice-dataset (accessed Nov. 09, 2021). 34 “COVID-19&Normal&Pneumonia_CT_Images | Kaggle.” https://www.kaggle.com/datasets/anaselmasry/covid19normalpneumonia-ct-images (accessed May 08, 2022). 35 M. Tan and Q. V. Le, “EfficientNetV2: smaller models and faster training,” 2021. 36 J. Hu, L. Shen, S. Albanie, et al., “Squeeze-and-excitation networks,” 2017, [Online]. Available: http://arxiv.org/abs/1709.01507 37 Q. Wang, B. Wu, P. Zhu, et al., “ECA-Net: efficient channel attention for deep convolutional neural networks,” 2019, [Online]. Available: http://arxiv.org/abs/1910.03151(accessed February 01, 2022). 38 Opitz D. Maclin R. Popular ensemble methods: an empirical study J Artificial Intelligence Res 11 1999 169 198 10.1613/jair.614 39 M. A. Ganaie and M Hu. “Ensemble deep learning: A review,” arXiv preprint arXiv:2104.02395 (2021). 40 Sollich P. Krogh A. Learning with ensembles: how over-fitting can be useful Proceedings of the 8th International Conference on Neural Information Processing Systems 1995 190 196 Pages 41 Pourtaheri Z.K. Zahiri S.H. Ensemble classifiers with improved overfitting 1st Conference on Swarm Intelligence and Evolutionary Computation 2016 10.1109/CSIEC.2016.7482130 42 Rajaraman S. Kim I. Antani S.K. Detection and visualization of abnormality in chest radiographs using modality-specific convolutional neural network ensembles PeerJ 8 2020 e8693 10.7717/peerj.8693 32211231 43 Devnath L. Fan Z. Luo S. Detection and visualisation of pneumoconiosis using an ensemble of multi-dimensional deep features learned from Chest X-rays Int J Environ Res and Public Health 19 18 2022 11193 10.3390/ijerph191811193 36141457 44 Rajaraman S. Cemir S. Xue Z A novel stacked model ensemble for improved TB detection in chest radiographs Chapter 1, Medical Imaging 2019 CRC Press Boca Raton 45 Devnath L. Luo S. Summons P. Deep ensemble learning for the automatic detection of pneumoconiosis in coal worker's Chest X-ray radiography J Clin Med 11 18 2022 5342 10.3390/jcm11185342 36142989 46 Mishra A.K. Das S.K. Roy P. Identifying COVID19 from Chest CT images: a deep convolutional neural networks based approach J Healthcare Eng 2020 2020 10.1155/2020/8843664 47 M. Saqib, S. Anwar, A. Anwar, et al., “COVID19 detection from radiographs: is deep learning able to handle the crisis?,” no. June, pp. 1–14, 2020, [Online]. Available: www.preprints.org 48 He X. Yang X.Y. Zhang S.H. Sample-efficient deep learning for COVID-19 diagnosis based on CT scans IEEE Transactions on Med Imaging XX Xx 2020 10.1101/2020.04.13.20063941 49 A. Mobiny et al., “Radiologist-level COVID-19 detection using CT Scans with detail-oriented capsule networks,” 2020, [Online]. Available: http://arxiv.org/abs/2004.07407 50 Polsinelli M. Cinque L. Placidi G. A light CNN for detecting COVID-19 from CT scans of the chest Pattern Recog Letters 140 2020 95 100 10.1016/j.patrec.2020.10.001 51 X. Yang, X. He, J. Zhao, et al., “COVID-CT-Dataset: A CT scan dataset about COVID-19,” 2020, [Online]. Available: http://arxiv.org/abs/2003.13865. (accessed December 05, 2021). 52 Hernández Santa Cruz J.F. An ensemble approach for multi-stage transfer learning models for COVID-19 detection from chest CT scans Intelligence-Based Med 5 February 2021 100027 10.1016/j.ibmed.2021.100027 53 Shaik N.S. Cherukuri T.K. Transfer learning based novel ensemble classifier for COVID-19 detection from chest CT-scans Computers in Biol Med 141 June 2021 2021 105127 10.1016/j.compbiomed.2021.105127

0

PMC9748720

NO-CC CODE

2022-12-15 23:22:46

no

Acad Radiol. 2022 Nov 25; doi: 10.1016/j.acra.2022.11.027

utf-8

Acad Radiol

10.1016/j.acra.2022.11.027

oa_other

==== Front Acad Radiol Acad Radiol Academic Radiology 1076-6332 1878-4046 The Association of University Radiologists. Published by Elsevier Inc. S1076-6332(22)00632-8 10.1016/j.acra.2022.11.027 Original Investigation Stacking Ensemble and ECA-EfficientNetV2 Convolutional Neural Networks on Classification of Multiple Chest Diseases Including COVID-19 Huang Mei-Ling PhD. in Industrial Engineering & Management ⁎ Liao Yu-Chieh Master in Industrial Engineering & Management Department of Industrial Engineering & Management, National Chin-Yi University of Technology, 57, Sec. 2, Zhongshan Rd., Taiping Dist., Taichung, 411030, Taiwan ⁎ Address correspondence to: M-L.H. 25 11 2022 25 11 2022 6 9 2022 15 11 2022 20 11 2022 © 2022 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved. 2022 The Association of University Radiologists Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Rationale and Objectives Early detection and treatment of COVID-19 patients is crucial. Convolutional neural networks have been proven to accurately extract features in medical images, which accelerates time required for testing and increases the effectiveness of COVID-19 diagnosis. This study proposes two classification models for multiple chest diseases including COVID-19. Materials and Methods The first is Stacking-ensemble model, which stacks six pretrained models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M. The second model is self-designed model ECA-EfficientNetV2 based on ECA-Net and EfficientNetV2. Ten-fold cross validation was performed for each model on chest X-ray and CT images. One more dataset, COVID-CT dataset, was tested to verify the performance of the proposed Stacking-ensemble and ECA-EfficientNetV2 models. Results The best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.21%, Precision of 99.23%, Recall of 99.25%, F1-score of 99.20%, and (area under the curve) AUC of 99.51% on chest X-ray dataset; the best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.81%, Precision of 99.80%, Recall of 99.80%, F1-score of 99.81%, and AUC of 99.87% on chest CT dataset. The differences for five metrics between Stacking-ensemble and ECA-EfficientNetV2 models are not significant. Conclusion Ensemble model achieves better performance than single pretrained models. Compared to the SOTA, Stacking-ensemble and ECA-EfficientNetV2 models proposed in this study demonstrate promising performance on classification of multiple chest diseases including COVID-19. KEY WORDS COVID-19 Convolutional neural network Ensemble learning Stacking ==== Body pmcINTRODUCTION COVID-19 causes disease in humans and vertebrates, and is a zoonotic infectious disease. Some confirmed patients may have severe pneumonia and respiratory failure (1). The most common tools for detecting lung infections are Chest X-ray and Computed tomography (CT). Chest X-ray is a very common noninvasive radiological test that has been widely used to screen for a variety of lung diseases such as COVID-19, pneumonia, pulmonary effusion, lung cancer, and emphysema (2). In clinical practice, chest X-ray images are often interpreted by radiologists, which is time-consuming and prone to errors in subjective assessments. A CT image is composed of a certain number of pixels with different grayscales arranged in a matrix. If the number of pixels is larger, the pixel value is smaller, and the image will be clearer. Although CT images provide very fine details, but it has more radiation than a chest X-ray image, and the equipment is relatively expensive (3). In recent years, due to the rise of artificial intelligence, researchers have applied deep learning to detect COVID-19 by using chest X-ray images and CT images. Compared to traditional machine learning, deep learning can automatically extract features of images to reduce processing time (4). At present, there have been many researches applying convolutional neural network (CNN) on image recognition. This method has been proved to be a powerful image recognition technology and has been widely used for COVID-19 detection. Common CNNs include GoogLeNet, ResNet, Xception, DenseNet, MobileNet, and EfficientNet, etc. For example, Ieracitano et al. proposed a fuzzy logic-based convolutional neural network (CovNNet) and the classification accuracy is 81.00% in 2022 by using a total of 155 chest X-ray images (5). Khan et al. proposed two convolutional neural networks named DHL and DBHL in 2021, using a total of 6448 chest X-ray images. The results showed that the accuracy of the binary classification of COVID-19 and normal was 98.53% (6). Loey et al. proposed a Bayesian-optimized convolutional neural network in 2022 using a total of 10,848 chest X-ray images. The results showed that the three-category accuracy of COVID-19, general pneumonia and normal was 96.00% (7). Hu et al. proposed a two-stage detection method in 2021 using a total of 11,092 chest X-ray images (8). The first stage is to train a CNN as a feature extractor, and the second stage uses extreme learning machines (ELMs) for real-time detection. In addition, the Chimp optimization algorithm is used to improve the results and increase the reliability of the network, and finally it is compared to the general CNN, Genetic algorithm optimized ELM, Cuckoo search optimized ELM, and Whale optimization algorithm optimized ELM. The results show that the proposed method has a binary classification accuracy of 99.11% for COVID-19 and non-COVID-19. Thaseen et al. applied ensemble learning by combining ResNet, FitNet, IRCNN, MobileNet, and EfficientNet in 2022, using a total of 13,808 chest X-ray images (9). The results showed that the three-category accuracy of COVID-19, general pneumonia and normal was 99.00%. A hybrid convolutional neural network (HCNN) combining CNN and RNN on classification of three-category accuracy of COVID-19, general pneumonia and normal was proposed with the classification accuracy of 98.20% (10). Musallam et al. proposed a convolutional neural network called DeepChest in 2022, using a total of 7512 chest X-ray images. The results showed that the three-category accuracy of COVID-19, general pneumonia and normal was 96.56% (11). A convolutional neural network named DenResCov-19 composing of DenseNet121 and ResNet50 networks was proposed, and DenResCov-19 was combined with existing ResNet50, DenseNet121, VGG16, and InceptionV3 networks, using a total of 6469 chest X-ray images. The results showed that the four-category area under the curve (AUC)-ROC of COVID-19, general pneumonia, tuberculosis and normal was 99.60% (12). The literature on COVID-19 detection using CT images such as Garg et al. used a series of models such as EfficientNet, DenseNet, VGG, and ResNet with a total of 20 convolutional neural networks in 2022, using a total of 4173 CT images. The results show that the binary classification accuracy of EfficientNet-B5 in detecting COVID-19 and non-COVID-19 is 98.45% (13). Lahsaini et al. used transfer learning on VGG16, VGG19, Xception, InceptionV2, ResNet, DenseNet121, and DenseNet201 in 2021, combined with GradCam, using a total of 4986 CT images. The results show that in the binary classification of COVID-19 and non-COVID-19, DenseNet201+ GradCam achieves the best accuracy rate of 98.80% (14). Rahimzadeh et al. used ResNet50V2 as the backbone network and compared the model with ResNet50V2 and Xception after adding a feature pyramid network (FPN), using a total of 63,849 CT images. The results show that ResNet50V2+ FPN has an accuracy of 98.49% in COVID-19 and normal binary classification (15). Qi et al. first used five models of UNet, LinkNet, R2UNet, Attention UNet, and UNet++ to segment CT images, and then used pretrained DenseNet121, InceptionV3 and ResNet50 for classification, using a total of Over 10,000 CT images. The results show that in the binary classification of COVID-19 and CAO, LinkNet performs best in lung segmentation with a Dice coefficient of 0.9830, while DenseNet121 with capsule network has a prediction accuracy of 97.10% (16). Abdel-Basset et al. proposed a two-stage detection method in 2021. The first stage is to use the proposed GR-UNet to segment the area of lung infection, and then transfer learning is used as feature extraction; the second stage is to use the proposed GR-UNet to segment the lung infected area. The stage is to propose an infection prediction module that uses the infected location to make decisions about classification, using a total of 9,593 CT images. The results showed that the binary classification accuracy of COVID-19 and CAP was 96.80% (17). Ye et al. proposed a convolutional neural network named CIFD-Net in 2022, which can effectively handle the multi-region displacement problem through a new robust supervised learning, using a total of 45,167 CT images picture. The results showed that the binary classification accuracy of COVID-19 and non-COVID-19 was 91.19% (18). Balaha et al. used transfer learning models including ResNet50, ResNet101, VGG16, VGG19, Xception, MobileNetV1, MobileNetV2, DenseNet121, and DenseNet169 in 2021, and added the Harris Hawks optimization to optimize hyper-parameters, and finally use fast classification stage and compact stacking stage to stack the best models into one, using a total of 15,535 CT images. The results show that in the binary classification of COVID-19 and non-COVID-19, the weighted sum method (WSM) is used to obtain an accuracy of 99.33% (19). Qi et al. proposed a detection method named DR-MIL in 2021, which first treats a 3D CT image of a patient as a bag and selects ten CT slices as initial instances. The deep features were then extracted from the pretrained ResNet50 by fine-tuning and treated as a Deep Represented Instance Score. The bag with DRIS was input to K-Nearest Neighbor (KNN) to generate the final prediction, using a total of 241 patients CT images. The results showed that the binary classification accuracy of COVID-19 and CAP was 95.00% (20). Some scholars used both chest X-ray images and CT images to detect COVID-19. Related studies such as Ahamed et al. fine-tune the pretrained ResNet50V2 in 2021, using a total of 4593 chest X-ray images and 3000 CT images. The results showed that the four-category accuracy for detecting COVID-19, viral pneumonia, bacterial pneumonia, and normal in chest X-ray images was 96.45%; the three-category accuracy for COVID-19, bacterial pneumonia and normal was 97.24%; COVID-19 and normal binary classification accuracy was 98.95%. The three-category accuracy for detecting COVID-19, CAP, and normal in CT images was 99.01%; the two-category accuracy for COVID-19 and normal was 99.99% (21). Kumari et al. used four convolutional neural networks, InceptionV3, VGG16, Xception and ResNet50 in 2021, using a total of 2000 chest X-ray images and 2000 CT images. The results show that the binary classification accuracy of VGG16 in detecting COVID-19 and non-COVID-19 in chest X-ray images was 98.00%; the binary classification accuracy of Xception in detecting COVID-19 and non-COVID-19 in CT images was 83.00% (22). Ahsan et al. fine-tuned eight convolutional neural networks including VGG16, VGG19, ResNet15V2, InceptionResNetV2, ResNet50, DenseNet201, MobilenetV2, and NasNetMobile in 2020, using a total of 400 chest X-ray images and 400 CT images. The results show that the binary classification accuracy of NasNetMobile in detecting COVID-19 and non-COVID-19 in chest X-ray images was 100.00%; the binary classification accuracy of NasNetMobile in detecting COVID-19 and non-COVID-19 in CT images was 95.20% (23). Jia et al. proposed a dynamic CNN modification method in 2021 and applied it to a fine-tuned ResNet, and finally compared the results with VGG16, InceptionV3, ResNet18, DenseNet121, MobileNetV3, and SqueezeNet, using a total of 7,592 chests X-ray images and 104,009 CT images. The results showed that the five-category accuracy of detecting COVID-19, viral pneumonia, bacterial pneumonia, tuberculosis and normal in chest X-ray images was 99.60%; in CT images, detecting COVID-19, non-COVID-19 and normal three-category accuracy was 99.30% (24). Kassania et al. use eight models such as DenseNet, ResNet, MobileNet, InceptionV3, Xception, InceptionResNetV2, VGG and NASNet to extract features, and then extracted features were input to decision tree, random forest, XGBoost, AdaBoost, Bagging, and LightGBM. A total of 137 chest X-ray images and 137 CT images were used. The results show that DenseNet121+ Bagging combines chest X-ray images and CT images to detect COVID-19 and normal with a binary classification accuracy of 99.00% (25). Gour and Jain proposed an integrated stacked CNN in 2022. After fine-tuning VGG19 and Xception and generating five sub-models, all sub-models were stacked using a softmax classifier with a total of 3040 images Chest X-ray images and 4,645 CT images. The results showed that the three-category accuracy of detecting COVID-19, general pneumonia, and normal in chest X-ray images was 97.27%; and the two-category accuracy of detecting COVID-19 and normal in CT images was 98.30% (26). Kamil et al. used fine-tuned VGG19 in 2021, using a total of 977 chest X-ray images and 23 CT images. The results showed a 99.00% accuracy of the binary classification between COVID-19 and normal (27). Saygılı applied Bag of Tree, K-ELM, KNN, and SVM with a total of 1125 chest X-ray images and 3228 CT images. The results showed that SVM has an accuracy of 85.96% in detecting COVID-19, general pneumonia and normal in chest X-ray images; K-ELM is accurate in detecting COVID-19 and non-COVID-19 in CT images. The accuracy was 98.88% (28). Existing pretrained models are designed for general natural images and fine-tuned for classified image types, which is not specifically designed for COVID-19 detection. The general natural images are large and simple, while the images of COVID-19 have specific patterns and textures that differ significantly from natural images. Based on previous studies, we can find that the use of ensemble learning and the author's self-proposed convolutional neural network has good performance in detecting COVID-19 in chest X-ray images and CT images. Ensemble learning mainly combines multiple existing convolutional neural networks, which can not only reduce the probability of misjudgment by a single model, but also improve the classification accuracy in less time. Ensemble learning solves the current need to detect COVID-19 without designing a new model to get good detection performance in the most time-saving way. Challenging the possibility of improving accuracy on multiple chest disease diagnosis, this study proposes two classification models for multiple chest diseases including COVID-19. First, we obtained chest X-ray images and CT images from multiple public databases. Next, we select six pretraining models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M in the EfficientNetV2 series. The reason for choosing the previous models is the training speed and argument efficiency of EfficientNetV2 is better than some previous networks (29). To the best of our knowledge, this is the first study ensembles the series of EfficientNetV2 models for COVID-19 detection. In addition, this study proposes a self-designed model ECA-EfficientNetV2 based on ECA-Net and EfficientNetV2. MATERIALS AND METHODS Figure 1 represents the architecture of this study.• Step 1 Data extraction: The chest X-ray and CT images are collected. • Step 2 Image preprocessing: The size of all selected images is equalized and saved in PNG. The datasets are split into training, validation, and test subsets. • Step 3 Pretrained models: Six EfficientNetV2 models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, and EfficientNetV2-M are used. • Step 4 Proposed Ensemble-stacking model: The previous six EfficientNetV2 models are stacked. • Step 5 Proposed model: ECA-EfficientNetV2 model. • Step 6 Performance evaluation: Accuracy, precision, recall, F1-score, and AUC are recorded for each model. • Step 7 Comparison with SOTA: The results of models are compared to the related studies. Figure 1 The architecture of this study. (Color version of figure is available online.) Figure 1 Data Extraction Chest X-ray images and CT images are collected from five Kaggle datasets described as follows. The first Kaggle dataset is COVID-19 Radiography Dataset (30), which contains 3616 COVID-19, 1345 viral pneumonia, 6012 lung opacity and 10,192 normal for a total of 21,165 chest X-ray images (PNG). The second Kaggle dataset is Chest X-Ray Images (Pneumonia) (31), collected from pediatric patients aged 1 to 5 from Guangzhou Women and Children's Medical Center with 4273 pneumonia and 1583 normal chest X-ray images (JPEG). The third Kaggle dataset is Tuberculosis Chest X-ray Dataset (32), which contains 3500 tuberculosis and 3500 normal chest X-ray images (PNG). Large COVID-19 CT scan slice dataset (33), containing 7593 COVID-19, 2618 chest abdominal pelvis and 6893 normal CT images for a total of 17,104 CT images (PNG), is the forth Kaggle dataset used in this study. The fifth Kaggle dataset is COVID-19 and Normal and PneumoniaCT_Images (34), containing 2035 COVID-19, 3309 pneumonia and 2119 normal CT images for a total of 7463 CT images (PNG). The COVID-19 X-ray images was formed based on the previous datasets 1–3. The pneumonia images in the second dataset were split into groups of viral pneumonia and bacterial pneumonia. The total COVID-19 X-ray images contain five groups, which are 3616 COVID-19, 2780 viral pneumonia, 2838 bacterial pneumonia, 3500 Tuberculosis and 15,275 Normal for a total of 28,009 chest X-ray images (PNG). The COVID-19 CT images was formed based on the previous datasets 4 and 5, which contains four groups including 9628 COVID-19, 2618 Chest Abdomen Pelvis (CAP), 3309 pneumonia, and 9012 Normal for a total number of 24567 chest CT images. Image Preprocessing To equalize the number of images for each group, 1200 images are randomly selected for each group for both X-ray and CT datasets in this study. The number of images in training, validation and test subsets are 900, 100, and 200, respectively for X-ray dataset; The number of images in training, validation and test subsets are 4500, 500, and 400, respectively for CT dataset. Ten-fold cross validation was performed in this study. Tables 1 and 2 display the number of images for each group for COVID-19 X-ray dataset (Dataset X-ray) and COVID-19 CT dataset (Dataset CT) used in the study.Table 1 COVID-19 X-ray Dataset (Dataset X-ray) Table 1Group Training Validation Test COVID-19 900 100 200 Bacterial pneumonia 900 100 200 Tuberculosis 900 100 200 Viral pneumonia 900 100 200 Normal 900 100 200 Total 4500 500 1000 Table 2 COVID-19 CT Dataset (Dataset CT) Table 2Group Training Validation Test COVID-19 900 100 200 CAP 900 100 200 Pneumonia 900 100 200 Normal 900 100 200 Total 3600 400 800 The original images collected from the previous datasets differ in size and format. As the model requirement, all the images are preprocessed to in PNG with size of 224  ×  224. Figure 2, Figure 3 display the examples of images after preprocessed for chest X-ray (Dataset X-ray) and CT (Dataset CT) images, respectively.Figure 2 Chest X-ray images after preprocessing. (Color version of figure is available online.) Figure 2 Figure 3 CT images after preprocessing. (Color version of figure is available online.) Figure 3 Convolutional Neural Network Convolutional Neural Network (CNN) is a feedforward neural network, mainly composed of multiple convolution layers, pooling layers and fully connected layers. Compared to other neural networks, convolutional neural networks have better performances on image or speech recognitions. The goal of training a convolutional neural network is to find the most appropriate weights in the process of multiple forward and reverse iterations. Transfer learning is having previously trained model on a larger database that we can directly apply the architecture and weights of the pretrained model to various studies to speed the efficiency of the training model. The Application module in Keras currently provides about 40 pretrained models, all of which are trained on the ImageNet dataset. We choose the current newer and efficient EfficientNetV2, which includes six series of models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M. The reason for not choosing EfficientNetV2-L is that the model has a large number of parameters and cannot be executed in our existing hardware resources. EfficientNetV2 There are some problems with the previous EfficientNet series models, such as (1) when the size of the image is large, the training speed for EfficientNet-B3 to EfficientNet-B7 is very slow. (2) Training is slow when Depthwise convolutions is used. (3) It is not the best choice to enlarge each layer with the same magnification. Constantly increasing image size, which also leads to large memory consumption, which in turn affects training speed. Therefore, Tan and Le (35) proposed a new convolutional neural network EfficientNetV2 in 2021, which uses a nonuniform scaling strategy and can increase the number of deeper layers and has faster training speed and better parameter efficiency as compared to the previous models. In addition, Tan and Le (35) also used EfficientNet-B4 to do some experiments, and found that replacing the MBConv module with the Fused-MBConv module in the early stage can greatly improve the training speed. However, if each layer is replaced with the Fused-MBConv module, the number of parameters and FLOPs will be significantly increased, and the training speed will be relatively reduced. Therefore, Neural Architecture Search (NAS) was used to find the best combination of MBConv and Fused-MBConv. Figure 4 is the architecture of EfficientNetV2.Figure 4 Architecture of EfficientNetV2. (Color version of figure is available online.) Figure 4 ECA-Net Channel attention has greatly improved the performance of convolutional neural networks, and most scholars are currently working on developing more complex attention modules. The most representative method such as Hu et al. (36) used Squeeze-and-excitation (SE) module and proposed SE-Net in 2017. SE-Net first uses a global average pooling layer for each channel, and then uses two nonlinear fully connected layers and a sigma function to generate the weights for each channel. Although the SE module is widely used in some researches on the current channel attention module, it has been proved that dimensionality reduction will affect both the prediction performance for channel attention, and the efficiency of obtaining the weights between all channels. Wang et al. (37) uses a lightweight and efficient channel attention (ECA) module, which only adds a small number of parameters, but can achieve significant performance gains. The ECA module does not use dimensionality reduction and operates through a one-dimensional convolution of size k. According to the experimental results of ECA-Net on ImageNet-1K and MS COCO, ECA-Net has lower model complexity than the state-of-the-art methods. In addition, ECA-Net has better efficiency no matter in image classification, image segmentation or object detection. The Proposed Stacking-Ensemble Model Ensemble learning is a type of supervised learning that has been widely used in the fields of statistics, machine learning and deep learning. Compared to a single learning algorithm, the purpose of ensemble learning is to combine multiple algorithms or models to form a model with better predictive performance. Using an ensemble approach will yield better results when there are significant model-to-model differences (38,39) reviews on ensemble deep leaning. Details of ensemble classifiers with improved overfitting have been investigated by several studies, for instance (40,41). In recent years, due to the continuous improvement of computing power of computers, large-scale integrated models can be trained within a reasonable time, and have been applied on medical image recognition, face recognition, emotion recognition and financial decision-making, etc. Successful applications of ensemble classifiers could be seen in (42, 43, 44, 45). The main methods of ensemble learning can be divided into three categories: bagging, boosting and stacking. The main function of stacking is to combine multiple algorithms to make predictions, that is, the result integration of the voting method or the average method. Ensemble stacking is a powerful CNN ensemble method thanks to its unique meta-learning algorithm. Meta-learning algorithms work by taking the probabilities of the input from each sub-model and determining which model performs best in extracting features. The learning algorithms directly extend the learning of each sub-model and combine the best predictions from each model. If each model is unique, then each model learns differently. Stacking achieves better results than either trained model and is used to evaluate the error rate of bagging. This study chooses stacking as our approach for ensemble learning. The stacking-ensemble model developed in this study selects six pretrained models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M in the EfficientNetV2 series. Figure 5 presents the architecture of the proposed stacking-ensemble model in this study.Figure 5 Stacking-ensemble model. (Color version of figure is available online.) Figure 5 The Proposed ECA-EfficientNetV2 Model In addition to the proposed stacking-ensemble model, challenging the possibility of improving accuracy on multiple chest disease diagnosis, this study proposes a self-designed model ECA-EfficientNetV2 based on ECA-Net and EfficientNetV2. Motivation of choosing the previous two models as basis are described as follows:1. The advantages of EfficientNetV2 includes: (1) The network is better than some previous networks in terms of training speed and number of parameters; (2) It proposes improved incremental learning methods that dynamically adjust regularization (e.g., dropout, data augmentation, and mix-up) according to the size of the training images; (3) Progressive learning is used to perform well on pretrained ImageNet, CIFAR, cars, and flowers datasets. 2. The major advantages of ECA-Net are: (1) ECA is a lightweight attention module that only contains k parameters (k ≤ 9), which can be used to improve the performance of large convolutional neural networks; (2) The ECA module uses a dimensionality-free local cross-channel interaction method that adaptively selects suitable adjacent channels to compute attention; (3) The features extracted by the SE module between different channels are relatively similar, while the features extracted by the ECA module from different channels are different, which indicates that the features extracted by the ECA module are better for classification. Based on the previous two points, the proposed ECA-EfficientNetV2 are designed and introduced as follows:1. ECA-EfficientNetV2 uses the dilated convolution module as the first two layers. The advantage is that while the kernel size is increased, the parameters or calculation amount of the original model can be maintained. Each module contains two dilated convolutional layers and an activation function SELU, where the dilation rate of the first layer is set to two, and the dilation rate of the second layer is set to three. 2. To reduce both the number of parameters and the computation complexity, ECA-EfficientNetV2 replaces SE module with ECA module in MBConv and Fused-MBConv convolution modules, and renames as MBEConv and Fused-MBEConv. 3. Zero-padding is added to the convolutional layers in both the MBEConv and Fused-MBEConv modules to prevent the input image be affected by the kernel size. In addition, the original activation function SELU is changed to ReLU to overcome the problem of vanishing gradient. 4. After MBEConv and Fused-MBEConv, two general convolution modules are added. The internal parameters include Zero-padding, Stride, and SELU. In addition, we add a batch normalization after each convolutional layer, which makes training easier and more stable, and improves the performance of the neural network. 5. Use the global average pooling layer to improve the problem of a large number of parameters that occurs in the fully connected layer. After that, add a dropout layer before the classification layer to generate multiple results by continuously updating the weights, and finally remove the outliers to avoid the problem of overfitting. The number of parameters for ECA-EfficientNetV2 is 5,706,965, which is much less than the number of parameters (117,753,253) used in EfficientNetV2-L. Figure 6 presents the architecture of the proposed Fused-MBEConv layer and MBEConv layers, and Figure 7 shows the architecture of the proposed ECA-EfficientNetV2.Figure 6 The architecture of the proposed Fused-MBEConv layer and MBEConv layers. (H, W, C denotes network height, width and channel respectively.) (Color version of figure is available online.) Figure 6 Figure 7 Architecture of ECA-EfficientNetV2. (Color version of figure is available online.) Figure 7 Model Evaluation Measures The performance indices include accuracy, precision, recall, F1-Score and the AUC. The mathematical formulas of these indicators are shown in Eqs. (1–5), respectively. Accuracy is the ratio of the number of correctly classified samples to the total number of samples; Precision is the ratio of the number of true positives to the total number of elements labelled to the positive class; Recall presents the number of true positives divided by the total number of true positives and false negatives; F1-Score is a measure of precision and recall. The higher the F1-score, the better the classification performance of the model; AUC denotes the measure at distinguishing between the positive and negative classes. Higher the AUC, better the model.(1) Accuracyc=(TPc+TNc)(TPc+FPc+FNc+TNc) (2) Precisionc=TPc(TPc+FPc) (3) Recallc=TPc(TPc+FNc) (4) F1−score=2n∑c=1nRecallc×PrecisioncRecallc+Precisionc (5) AUC=∑cnrankn−TPc×(1+TPc)2TPc×TNc where c is the number of classes; TP (true positive) represents the number of positive categories that are correctly classified as positive, FP (false positive) represents the number of negative categories that are incorrectly classified as positive, TN (true negative) refers to the number of negative categories that are correctly classified as negative, and FN (false negative) refers to the number of positive categories that are incorrectly classified as negative. RESULTS The equipment used in the experiment is an Intel(R) Core(TM) i9-10900F 2.81 GHz CPU, NVIDIA GeForce RTX 3070 8G GPU. The whole experiment process is performed using Python 3.8 [Python Software Foundation, Fredericksburg, Virginia, USA], which contains Keras 2.6 and Tensorflow 2.6. Ten-fold cross validation is used to evaluate the performance of each model under batch size of 16; epochs count of 30; optimizer of Adam, learning rate of 1e-5 and dropout of 0.4. Performance Results on Dataset X-ray On dataset 1, Table 3, Table 4, Table 5, Table 6 report the results of Accuracies, Precisions, Recalls, F1-Scores, and AUCs in the test sets, and the average and standard deviation for EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models. The best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.21%, Precision of 99.23%, Recall of 99.25%, F1-score of 99.20%, and AUC of 99.51%. Figure 8, Figure 9 show the examples of training accuracy and loss, confusion matrix and ROC curve for each model, respectively.Table 3 Performance of EfficientNetV2-B0 and EfficientNetV2-B1 Models Table 3Fold EfficientNetV2-B0 EfficientNetV2-B1 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 82.10% 82.33% 82.14% 82.13% 88.81% 83.10% 83.23% 83.14% 83.11% 89.44% 2 85.20% 84.99% 85.24% 85.06% 90.75% 85.90% 85.81% 85.93% 85.85% 91.19% 3 88.50% 88.62% 88.53% 88.57% 92.81% 86.70% 86.69% 86.72% 86.59% 91.69% 4 85.50% 85.73% 85.51% 85.49% 90.94% 82.30% 82.26% 82.34% 82.25% 88.94% 5 85.90% 85.93% 85.91% 85.96% 91.19% 84.30% 84.47% 84.32% 84.34% 90.19% 6 84.40% 84.46% 84.42% 84.43% 90.25% 85.20% 85.15% 85.23% 85.16% 90.75% 7 85.60% 85.62% 85.64% 85.56% 91.00% 86.50% 86.40% 86.54% 86.42% 91.56% 8 84.40% 84.59% 84.46% 84.34% 90.25% 85.82% 85.94% 85.81% 85.84% 91.13% 9 84.50% 84.80% 84.51% 84.54% 90.31% 88.90% 89.01% 88.92% 88.93% 93.06% 10 86.20% 86.42% 86.23% 86.22% 91.38% 87.60% 87.79% 87.64% 87.65% 92.25% Average 85.23± 1.63% 85.35± 1.61% 85.26± 1.63% 85.23± 1.65% 90.77± 1.02% 85.63± 2.00% 85.68± 2.02% 85.66± 2.00% 85.61± 2.01% 91.02± 1.25% Table 4 Performance of EfficientNetV2-B2 and EfficientNetV2-B3 Models Table 4Fold EfficientNetV2-B2 EfficientNetV2-B3 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 79.90% 79.85% 79.91% 79.76% 87.44% 85.40% 85.72% 85.43% 85.37% 90.88% 2 84.40% 84.63% 84.42% 84.30% 90.25% 86.30% 86.19% 86.32% 86.22% 91.44% 3 86.60% 87.08% 86.61% 86.43% 91.69% 87.00% 87.27% 87.42% 86.97% 91.88% 4 81.70% 81.71% 81.74% 81.63% 88.56% 84.20% 84.06% 84.23% 84.11% 90.12% 5 83.70% 83.80% 83.72% 83.74% 89.81% 85.70% 85.84% 85.71% 85.69% 91.06% 6 80.60% 80.98% 80.64% 80.59% 87.88% 84.60% 84.94% 84.62% 84.61% 90.38% 7 83.20% 83.14% 83.23% 83.15% 89.50% 86.40% 86.67% 86.42% 86.38% 91.50% 8 87.30% 87.36% 87.34% 87.27% 92.06% 85.20% 85.49% 85.21% 85.27% 90.75% 9 80.60% 80.89% 80.64% 80.65% 87.88% 87.10% 87.22% 87.14% 87.12% 91.94% 10 85.10% 85.31% 85.14% 85.15% 90.69% 85.90% 86.26% 85.94% 85.92% 91.19% Average 83.31± 2.59% 83.48± 2.63% 83.34± 2.59% 83.27± 2.58% 89.58± 1.62% 85.78± 0.96% 85.97± 0.99% 85.84± 1.02% 85.77± 0.96% 91.11± 0.60% Table 5 Performance of EfficientNetV2-S and EfficientNetV2-M Models Table 5Fold EfficientNetV2-S EfficientNetV2-M Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 86.90% 86.82% 86.94% 86.81% 91.81% 87.30% 87.29% 87.34% 87.21% 92.06% 2 88.30% 88.31% 88.34% 88.27% 92.69% 88.70% 88.74% 88.71% 88.68% 92.64% 3 87.70% 88.01% 87.92% 87.95% 92.44% 89.00% 89.04% 89.27% 88.97% 93.13% 4 87.00% 87.37% 87.42% 87.03% 91.88% 87.70% 87.65% 87.72% 87.67% 92.31% 5 87.50% 87.55% 87.52% 87.48% 92.19% 89.40% 89.43% 89.41% 89.44% 93.38% 6 87.00% 86.93% 87.12% 86.91% 91.88% 89.70% 89.71% 89.73% 89.65% 93.56% 7 87.10% 87.13% 87.21% 87.05% 91.94% 89.00% 88.90% 89.29% 88.94% 93.13% 8 88.00% 88.03% 88.28% 87.98% 92.50% 89.70% 89.72% 89.74% 89.69% 93.56% 9 87.10% 87.13% 87.14% 87.11% 91.94% 86.70% 86.98% 86.74% 86.76% 91.69% 10 87.50% 87.69% 87.53% 87.56% 92.19% 89.30% 89.33% 89.31% 89.34% 93.31% Average 87.41± 0.47% 87.50± 0.51% 87.54± 0.49% 87.42± 0.51% 92.15± 0.31% 88.65± 1.05% 88.68± 1.01% 88.73± 1.07% 88.64± 1.05% 92.88± 0.66% Table 6 Performance of Stacking-Ensemble and ECA-EfficientNetV2 Models Table 6Fold Proposed Stacking-Ensemble Model Proposed ECA-EfficientNetV2 Accuracy Precision Recall F1-score AUC Accuracy Precision Recall F1-Score AUC 1 98.40% 98.42% 98.47% 98.39% 99.00% 98.80% 98.84% 98.81% 98.79% 99.25% 2 98.70% 98.71% 98.74% 98.72% 99.19% 99.60% 99.64% 99.61% 99.59% 99.75% 3 98.90% 98.91% 98.93% 98.94% 99.31% 98.80% 98.82% 98.84% 98.78% 99.25% 4 98.50% 95.54% 98.51% 98.53% 99.06% 99.50% 99.51% 99.54% 99.49% 99.69% 5 99.00% 99.01% 99.04% 98.99% 99.38% 99.10% 99.12% 99.11% 99.09% 99.44% 6 99.00% 99.02% 99.05% 98.99% 99.38% 99.40% 99.41% 99.44% 99.37% 99.62% 7 99.00% 99.05% 99.03% 99.01% 99.38% 99.50% 99.48% 99.54% 99.53% 99.69% 8 98.30% 98.35% 98.31% 98.29% 98.94% 99.00% 99.04% 99.07% 98.99% 99.38% 9 98.60% 98.64% 98.67% 98.62% 99.13% 99.00% 99.02% 99.07% 98.97% 99.38% 10 98.70% 98.73% 98.71% 98.75% 99.19% 99.40% 99.41% 99.44% 99.38% 99.62% Average 98.71± 0.26% 98.44± 1.05% 98.75± 0.26% 98.72± 0.26% 99.20± 0.16% 99.21± 0.30% 99.23± 0.30% 99.25± 0.30% 99.20± 0.31% 99.51± 0.19% Figure 8 Examples of training accuracy and loss from each model. (Color version of figure is available online.) Figure 8 Figure 9 Examples of confusion matrix and ROC curve from each model. (Color version of figure is available online.) Figure 9 Performance Results on Dataset CT On dataset 2, Table 7, Table 8, Table 9, Table 10 report the results of Accuracies, Precisions, Recalls, F1-Scores, and AUCs in the test sets, and the average and standard deviation for EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models. The best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.81%, Precision of 99.80%, Recall of 99.80%, F1-score of 99.81%, and AUC of 99.87%. Figure 10, Figure 11 show the examples of training accuracy and loss, confusion matrix and ROC curve for each model, respectively.Table 7 Performance of EfficientNetV2-B0 and EfficientNetV2-B1 Models Table 7Fold EfficientNetV2-B0 EfficientNetV2-B1 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 95.13% 95.22% 95.15% 95.10% 96.75% 93.63% 93.65% 93.62% 93.59% 95.75% 2 93.87% 93.88% 93.85% 93.86% 95.92% 96.25% 96.41% 96.25% 96.24% 97.50% 3 95.63% 95.60% 95.62% 95.59% 97.08% 95.13% 95.32% 95.15% 95.07% 96.75% 4 95.13% 95.09% 95.15% 95.08% 96.75% 93.25% 93.20% 93.25% 93.17% 95.50% 5 95.63% 95.78% 95.62% 95.60% 97.08% 93.00% 93.03% 93.17% 92.88% 95.33% 6 94.25% 94.29% 94.27% 94.23% 96.17% 93.25% 93.37% 93.27% 93.28% 95.50% 7 93.25% 93.43% 93.28% 93.13% 95.50% 94.13% 94.35% 94.17% 94.04% 96.08% 8 94.37% 94.39% 94.35% 94.36% 96.25% 90.87% 90.85% 90.88% 90.77% 93.92% 9 93.50% 93.51% 93.54% 93.47% 95.67% 95.13% 95.11% 95.15% 95.17% 96.75% 10 94.63% 94.60% 94.56% 94.61% 96.42% 91.37% 91.27% 91.75% 91.27% 94.25% Average 94.54± 0.84% 94.58± 0.83% 94.54± 0.83% 94.50± 0.85% 96.36± 0.56% 93.60± 1.67% 93.66± 1.74% 93.67± 1.61% 93.55± 1.70% 95.73± 1.11% Table 8 Performance of EfficientNetV2-B2 and EfficientNetV2-B3 Models Table 8Fold EfficientNetV2-B2 EfficientNetV2-B3 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 94.87% 98.88% 94.85% 94.83% 96.58% 94.37% 94.54% 94.38% 94.36% 96.25% 2 97.00% 96.99% 97.41% 96.87% 98.00% 95.63% 95.65% 95.62% 95.61% 97.08% 3 94.87% 94.92% 94.88% 94.84% 96.58% 94.87% 95.02% 94.85% 94.83% 96.58% 4 95.00% 94.97% 95.02% 94.94% 96.67% 95.13% 95.10% 95.12% 95.08% 96.75% 5 95.00% 94.99% 95.17% 94.97% 96.67% 96.37% 96.39% 96.38% 96.36% 97.58% 6 92.87% 92.95% 92.88% 92.85% 95.25% 95.75% 95.74% 95.75% 95.73% 97.17% 7 92.13% 92.19% 92.15% 92.03% 94.75% 95.00% 95.21% 95.14% 94.97% 96.67% 8 94.63% 94.58% 94.65% 94.59% 96.42% 95.50% 95.54% 95.51% 95.48% 97.00% 9 95.25% 95.24% 95.19% 95.24% 96.83% 94.37% 94.36% 94.35% 94.76% 96.25% 10 92.50% 92.45% 92.47% 92.38% 95.00% 94.00% 93.98% 94.37% 93.97% 96.00% Average 94.41± 1.48% 94.82± 2.04% 94.47± 1.57% 94.35± 1.49% 96.28± 0.99% 95.10± 0.73% 95.15± 0.72% 95.15± 0.68% 95.12± 0.70% 96.73± 0.49% Table 9 Performance of EfficientNetV2-S and EfficientNetV2-M Models Table 9Fold EfficientNetV2-S EfficientNetV2-M Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 96.37% 96.36% 96.35% 96.63% 97.58% 95.25% 95.22% 95.27% 95.22% 96.83% 2 96.63% 96.71% 96.62% 96.65% 97.75% 96.13% 96.14% 96.15% 96.10% 97.42% 3 97.38% 97.40% 97.37% 97.39% 98.25% 95.50% 95.54% 95.51% 95.47% 97.00% 4 96.63% 96.62% 96.65% 96.33% 97.75% 95.13% 95.12% 95.15% 95.10% 96.75% 5 96.50% 96.47% 96.54% 96.48% 97.67% 96.75% 96.76% 96.73% 96.78% 97.83% 6 95.75% 95.74% 95.77% 95.78% 97.17% 96.00% 96.05% 96.04% 95.99% 97.33% 7 96.50% 96.58% 96.54% 96.51% 97.67% 95.25% 95.28% 95.26% 95.24% 96.83% 8 96.88% 96.86% 96.87% 96.89% 97.92% 96.50% 96.51% 95.54% 96.49% 97.67% 9 95.63% 95.67% 95.65% 95.62% 97.08% 95.37% 95.33% 95.35% 95.34% 96.92% 10 95.13% 95.09% 95.12% 95.10% 96.75% 94.75% 94.70% 94.77% 94.72% 96.50% Average 96.34± 0.66% 96.35± 0.67% 96.35± 0.66% 96.34± 0.67% 97.56± 0.44% 95.66± 0.65% 95.67± 0.67% 95.58± 0.57% 95.65± 0.66% 97.11± 0.43% Table 10 Performance of Stacking-ensemble and ECA-EfficientNetV2 Models Table 10Fold Stacking-ensemble model ECA-EfficientNetV2 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 99.00% 99.01% 99.24% 99.89% 99.33% 99.38% 99.34% 99.35% 99.37% 99.58% 2 98.62% 98.67% 98.63% 98.54% 99.08% 99.88% 99.87% 99.85% 99.89% 99.92% 3 99.38% 99.46% 99.74% 99.25% 99.58% 99.75% 99.72% 99.71% 99.77% 99.83% 4 98.75% 98.74% 98.81% 98.93% 99.17% 99.25% 99.27% 99.24% 99.28% 99.50% 5 98.75% 98.78% 98.87% 98.79% 99.17% 99.88% 99.85% 99.87% 99.84% 99.92% 6 99.88% 99.87% 99.85% 99.89% 99.92% 99.94% 99.92% 99.96% 99.95% 99.98% 7 98.25% 98.33% 98.41% 98.24% 98.83% 100.00% 100.00% 100.00% 100.00% 100.00% 8 98.75% 98.77% 98.75% 98.72% 99.17% 100.00% 100.00% 100.00% 100.00% 100.00% 9 98.88% 98.87% 98.85% 98.86% 99.25% 100.00% 100.00% 100.00% 100.00% 100.00% 10 98.13% 98.16% 98.15% 98.12% 98.75% 100.00% 100.00% 100.00% 100.00% 100.00% Average 98.84± 0.51% 98.87± 0.50% 98.93± 0.54% 98.92± 0.61% 99.23± 0.34% 99.81± 0.27% 99.80± 0.27% 99.80± 0.28% 99.81± 0.27% 99.87± 0.18% Figure 10 Examples of training accuracy and loss from each model. (Color version of figure is available online.) Figure 10 Figure 11 Examples of confusion matrix and ROC curve from each model. (Color version of figure is available online.) Figure 11 DISCUSSION Dataset X-ray The performances of 5 metrics among EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models for dataset X-ray were compared in Table 11 . Two accuracies are above 98%, which come from Stacking-ensemble and ECA-EfficientNetV2 models, while the accuracies from six pretrained models are inferior. In fact, the highest two performance for five metrics from Stacking-ensemble and ECA-EfficientNetV2 models are significantly different from six pretrained models (p-value < 0.01). Although the best performance, the accuracy (99.21%), precision (99.23%), recall (99.25%), F1-score (99.20%), and AUC (99.51%), comes from the proposed ECA-EfficientNetV2 model, the differences for five metrics between Stacking-ensemble and ECA-EfficientNetV2 models are not significant. Comparing the standard deviations of five metrics from Stacking-ensemble and ECA-EfficientNetV2 models, those from the latter are relatively small, which stands the stability of the model.Table 11 Performance Comparison on Dataset X-ray Table 11Models Dataset X-ray Accuracy (%) Precision (%) Recall (%) F1-Score (%) AUC (%) EfficientNetV2-B0 85.23(±1.63) 85.35(±1.61) 85.26(±1.63) 85.23(±1.65) 90.77(±1.02) EfficientNetV2-B1 85.63(±2.00) 85.68(±2.02) 85.66(±2.00) 85.61(±2.01) 91.02(±1.25) EfficientNetV2-B2 83.31(±2.59) 83.48(±2.63) 83.34(±2.59) 83.27(±2.58) 89.58(±1.62) \ EfficientNetV2-B3 85.78(±0.96) 85.97(±0.99) 85.84(±1.02) 85.77(±0.96) 91.11(±0.60) EfficientNetV2-S 87.41(±0.47) 87.50(±0.51) 87.54(±0.49) 87.42(±0.51) 92.15(±0.31) EfficientNetV2-M 88.65(±1.05) 88.68(±1.01) 88.73(±1.07) 88.64(±1.05) 92.88(±0.66) Stacking-ensemble 98.71(±0.26) 98.44(±1.05) 98.75(±0.26) 98.72(±0.26) 99.20(±0.16) ECA-EfficientNetV2 99.21(±0.30) 99.23(±0.30) 99.25(±0.30) 99.20(±0.31) 99.51(±0.19) p-value 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ ⁎⁎ p < 0.01 Figure 12 shows examples of confusion matrix from EfficientNetV2-B0, EfficientNetV2-M, Stacking-ensemble, and ECA-EfficientNetV2 models. In EfficientNetV2-B0, 58 Bacterial Pneumonia are misclassified to Tuberculosis, and 42 Tuberculosis are misclassified to Bacterial Pneumonia. In EfficientNetV2-M, 40 Bacterial Pneumonia are misclassified to Tuberculosis, and 51 Tuberculosis are misclassified to Bacterial Pneumonia. Obviously, Bacterial Pneumonia and Tuberculosis are misclassified to each other from EfficientNetV2-B0 and EfficientNetV2-M models, while the wrong phenomenon is much improved in Stacking-ensemble and ECA-EfficientNetV2 models.Figure 12 Confusion matrix of Dataset X-ray. (Color version of figure is available online.) Figure 12 Dataset CT The performances of 5 metrics among EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models for dataset CT were compared in Table 12 . The accuracies from the previous eight models are greater than 93%; the highest two are from Stacking-ensemble and ECA-EfficientNetV2 models. Actually, the same results could be found in precision, recall, F1-score and AUC. The five metrics from Stacking-ensemble and ECA-EfficientNetV2 models are statistically significant from the six EfficientNetV2 models (p-value < 0.01). The best performance, the accuracy (99.81%), precision (99.80%), recall (99.80%), F1-score (99.81%), and AUC (99.87%), comes from the proposed ECA-EfficientNetV2, which are not significantly different from the Stacking-ensemble model. Same as we found in Dataset X-ray, the standard deviations of five metrics from ECA-EfficientNetV2 model are relatively small, which stands the stability of the model.Table 12 Performance Comparison on Dataset CT Table 12Models Dataset CT Accuracy (%) Precision (%) Recall (%) F1-Score (%) AUC (%) EfficientNetV2-B0 94.54(±0.84) 94.58(±0.83) 94.54(±0.83) 94.50(±0.85) 96.36(±0.56) EfficientNetV2-B1 93.60(±1.67) 93.66(±1.74) 93.67(±1.61) 93.55(±1.70) 95.73(±1.11) EfficientNetV2-B2 94.41(±1.48) 94.82(±2.04) 94.47(±1.57) 94.35(±1.49) 96.28(±0.99) EfficientNetV2-B3 95.10(±0.73) 95.15(±0.72) 95.15(±0.68) 95.12(±0.70) 96.73(±0.49) EfficientNetV2-S 96.34(±0.66) 96.35(±0.67) 96.35(±0.66) 96.34(±0.67) 97.56(±0.44) EfficientNetV2-M 95.66(±0.65) 95.67(±0.67) 95.58(±0.57) 95.65(±0.66) 97.11(±0.43) Stacking-ensemble 98.84(±0.51) 98.87(±0.50) 98.93(±0.54) 98.92(±0.61) 99.23(±0.34) ECA-EfficientNetV2 99.81(±0.27) 99.80(±0.27) 99.80(±0.28) 99.81(±0.27) 99.87(±0.18) p-value 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ ⁎⁎ p < 0.01 Figure 13 shows examples of confusion matrix from EfficientNetV2-B0, EfficientNetV2-M, Stacking-ensemble, and ECA-EfficientNetV2 models for Dataset CT. In EfficientNetV2-B0, 23 COVID-19 are misclassified to Normal, and 8 Normal are misclassified to COVID-19. In EfficientNetV2-M, 18 COVID-19 are misclassified to Normal, and 12 Normal are misclassified to COVID-19. In Stacking-ensemble, 7 COVID-19 are misclassified to Normal, and 1 Normal are misclassified to COVID-19. In ECA-EfficientNetV2, 2 COVID-19 are misclassified to Normal, and 0 Normal are misclassified to COVID-19. Obviously, COVID-19 and Normal are misclassified to each other from EfficientNetV2-B0 and EfficientNetV2-M models, while Stacking-ensemble and ECA-EfficientNetV2 models greatly improved the wrong phenomenon.Figure 13 Confusion matrix of Dataset CT. (Color version of figure is available online.) Figure 13 Comparison With SOTA The results of the proposed Stacking-ensemble and ECA-EfficientNetV2 models are compared to the related studies using stacking ensemble models for COVID-19 diagnosis in Table 13 . Most of the related studies worked on classification of 2 to 3 groups, while this study focuses on classification of five groups in X-ray and classification of 4 groups in CT. Most of the accuracies are greater than 90% by using stacking models, which represents the favorable performance of stacking method.Table 13 Comparison of the Proposed Models With SOTA Table 13No. Study(s) Dataset Architecture Class Accuracy 1 Ieracitano et al. (2022) (5) 155 X-ray images. CovNNet 2 81.00% 2 Khan et al. (2021) (6) 6448 X-ray images. DHL DBHL 2 98.53% 3 Loey et al. (2022) (7) 10,848 X-ray images. CNN+ Bayesian 3 96.00% 4 Hu et al. (2021) (8) 10,848 X-ray images. CNN+ ELMs+ ChOA 2 99.11% 5 Thaseen et al. (2022) (9) 13,808 X-ray images. Ensemble CNN model 3 99.00% 6 Kumar et al. (2021) (10) 6000 X-ray images. HCNN 3 98.20% 7 Musallam et al. (2022) (11) 7512 X-ray images. DeepChest 3 96.56% 8 Mamalakis et al. (2021) (12) 6000 X-ray images. DenResCov-19 4 99.60% 9 Garg et al. (2022) (13) 4173 CT images. EfficientNet-B5 2 98.45% 10 Lahsaini et al. (2021) (14) 4986 CT images. DenseNet201+GradCam 2 98.80% 11 Rahimzadeh et al. (2021) (15) 63,849 CT images. ResNet50V2+ FPN 2 98.49% 12 Qi et al. (2022) (16) 10,000 CT images. UNet+ DenseNet121 2 97.10% 13 Abdel-Basset et al. (2021) (17) 9593 CT images. GR-UNet+ CNN 2 96.80% 14 Ye et al. (2022) (18) 45,167 CT images. CIFD-Net 2 91.19% 15 Balaha et al. (2021) (19) 15,535 CT images. CNN+HHO+FCS+CSS+WSM 2 99.33% 16 Ahamed et al. (2021) (21) 4593 X-ray images. ResNet50V2(fine-tuning) 4 96.45% 3000 CT images. 3 99.01% 17 Kumari et al. (2021) (22) 2000 X-ray images. VGG16 2 98.00% 2000 CT images. Xception 2 83.00% 18 Ahsan et al. (2020) (23) 400 X-ray images. NasNetMobile 2 100.00% 400 CT images. 2 95.20% 19 Jia et al. (2021) (24) 7592 X-ray images. ResNet(fine-tuning) 5 99.60% 104,009 CT images. 3 99.30% 20 Kassania et al. (2021) (25) 137 X-ray images. DenseNet121+ Bagging 2 99.00% 137 CT images. 21 Gour and Jain (2022) (26) 3040 X-ray images. Ensemble CNN 3 97.27% 4645 CT images. 2 98.30% 22 Kamil et al. (2021) (27) 977 X-ray images. VGG19(fine-tuning) 2 99.00% 23 CT images. 23 Saygılı (2021) (28) 1125 X-ray images. SVM 3 85.96% 3228 CT images. K-ELM 2 98.88% 24 Stacking-ensemble 6000 X-ray images. Stacking-ensemble 5 98.71% 4800 CT images. 4 98.84% 25 ECA-EfficientNetV2 6000 X-ray images. ECA-EfficientNetV2 5 99.21% 4800 CT images. 4 99.81% Although the accuracy from our Stacking-ensemble model is slightly inferior than (8,9,12,19,21,23,24,25,27,28). As we have mentioned before, most of those related studies worked on classification of two groups, while this study is working on the classification of four groups in X-ray and five groups in CT. Even in this situation, the accuracies of our proposed Stacking-ensemble model are pretty close to the previous studies. In addition, the performance of the proposed ECA-EfficientNetV2 model dominates most of the related studies. The accuracies for X-ray and CT from ECA-EfficientNetV2 model are close to 100%, which illustrates the great classification capability for multiple groups on chest diseases. To verify the performance of the two proposed Stacking-ensemble and ECA-EfficientNetV2 models, one more open public dataset COVID-CT (https://github.com/UCSD-AI4H/COVID-CT) was tested and the performance metrics were displayed in Table 14 . There are 746 CT images including two groups of COVID-19 and Non-COVID-19 in dataset COVID-CT. Except the accuracy of 93.33% from Shaik and Cherukuri (53), the accuracies from other related studies are lower than 90%. The accuracies are 94.86% and 95.29% from our proposed Stacking-ensemble and ECA-EfficientNetV2 models, respectively, which are higher than all of the related studies compared in Table 14. Especially, the number of parameter in ECA-EfficientNetV2 model is 5,706,965, which is much less than those in (48,50,51).Table 14 Comparisons on COVID-CT Dataset Table 14Dataset Study(s) Accuracy Precision Recall F1-Score Parameter COVID-CT Mishra et al. (2020) (46) 88.30% — — 86.70% — Saqib et al. (2020) (47) 80.30% 78.20% 85.70% 81.80% — He et al. (2020) (48) 86.00% — — 85.00% 14,149,480 Mobiny et al. (2020) (49) — 84.00% — — — Polsinelli et al. (2020) (50) 85.00% 85.00% 87.00% 86.00% 12,600,000 Yang et al. (2020) (51) — — 89.10% — 25,600,000 Cruz (2021) (52) 86.00% — 89.00% 85.00% — Shaik and Cherukuri (2021) (53) 93.33% 93.17% 93.54% 93.29% — Stacking-ensemble 94.86% 94.79% 94.83% 94.84% — ECA-EfficientNetV2 95.29% 95.15% 95.24% 95.27% 5706,965 CONCLUSION This study applies six pretrained MobileNetV2 models on COVID-19 diagnosis, and stacks ensemble the previous six models to achieve better classification results. The self-designed Stacking-ensemble and ECA-MobileNetV2 models were proposed to classify multiple chest diseases including COVID-19 in this study. Classification was executed on five groups using X-ray images and four groups using CT images. The experimental results of two proposed models were compared to six pretrained MobileNetV2 models. The proposed Stacking-ensemble achieves accuracies of 98.71% and 98.84% on X-ray and CT datasets, respectively. With stability, the proposed ECA-MobileNetV2 achieves the highest accuracies of 99.21% and 99.81% on X-ray and CT datasets, respectively. The proposed Stacking-ensemble model is the first study ensembles the series of EfficientNetV2 models on multiple chest diseases including COVID-19 detection, which reduce prediction variance to the training data and improves overall classification performance when compared to any individual EfficientNetV2 model. As compared to the proposed ECA-MobileNetV2 model, the stacking model is computationally expensive; it takes around 1.5 hours in this experiment while the execution time is only 17 minutes in ECA-MobileNetV2 model. The major contribution of the study is the proposed ECA-MobileNetV2 model combines the advantages of EfficientNetV2 and ECA-Net to demonstrate superior classification performance with least variance and training time. The results of this study imply that the architecture of the proposed ECA-MobileNetV2 model can be used to assist radiologists in the X-ray and CT scans on multiple chest diseases including COVID-19. The 6,000 X-ray and 4800 CT images used in this study were collected from five Kaggle datasets. More chest images are encouraged to achieve more robust classification performance. Since the related studies compared in the study were searched from Scopus and were limited, more comprehensive review is suggested in the future research. ==== Refs REFERENCES 1 C. Pneumonia, “COVID-19,” vol. 2019, 2019. 2 Liu H. Wang L. Nan Y. SDFN: Segmentation-based deep fusion network for thoracic disease classification in chest X-ray images Computerized Med Imaging and Graphics 75 2019 66 73 10.1016/j.compmedimag.2019.05.005 3 Lin L. Qin L. Xu Z. Artificial intelligence distinguishes COVID-19 from community acquired pneumonia on Chest CT Appl Intel 2019 2020 1 5 [Online]. Available: http://arxiv.org/abs/2003.13865 (accessed February 01, 2022) 4 Sharma N. Sharma R. Jindal N. Machine learning and deep learning applications-a vision Global Transitions Proceedings 2021 0 8 10.1016/j.gltp.2021.01.004 5 Ieracitano C. Mammone N. Versaci M. A fuzzy-enhanced deep learning approach for early detection of Covid-19 pneumonia from portable chest X-ray images Neurocomputing 481 2022 202 215 10.1016/j.neucom.2022.01.055 35079203 6 Khan S.H. Sohail A. Khan A. COVID-19 detection in chest X-ray images using deep boosted hybrid learning Comp Biol Med 137 2021 104816 10.1016/j.compbiomed.2021.104816 7 Loey M. El-Sappagh S. Mirjalili S. Bayesian-based optimized deep learning model to detect COVID-19 patients using chest X-ray image data Comp Biol Med 142 2022 105213 10.1016/j.compbiomed.2022.105213 8 Hu T. Khishe M. Mohammadi M. Realtime COVID-19 diagnosis from X-Ray images using deep CNN and extreme learning machines stabilized by chimp optimization algorithm Biomed Signal Proces Contrl 68 2021 102764 10.1016/j.bspc.2021.102764 9 Sakthivel R. Sumaiya Thaseen I. Vanithac M. An efficient hardware architecture based on an ensemble of deep learning models for COVID-19 prediction Sustainable Cities and Soc Feb. 2022 103713 10.1016/j.scs.2022.103713 10 Kumar M. Shakya D. Kurup V. COVID-19 prediction through X-ray images using transfer learning-based hybrid deep learning approach Materials Today: Proceedings Dec. 2021 10.1016/j.matpr.2021.12.123 11 Musallam A.S. Sherif A.S. Hussein M.K. Efficient framework for detecting COVID-19 and pneumonia from chest X-Ray using deep convolutional network Egyptian Info J 2021 2022 10.1016/j.eij.2022.01.002 12 Mamalakis M. Swift A.J. Vorselaars B. DenResCov-19: A deep transfer learning network for robust automatic classification of COVID-19, pneumonia, and tuberculosis from X-rays Computerized Med Imaging and Graphics 94 2021 102008 10.1016/j.compmedimag.2021.102008 13 Garg A. Salehi S. la Rocca M. Efficient and visualizable convolutional neural networks for COVID-19 classification using Chest CT Expert Systems with Appl 195 2022 116540 10.1016/j.eswa.2022.116540 14 LAHSAINI I. EL HABIB DAHO M. CHIKH M.A. Deep transfer learning based classification model for covid-19 using chest CT-scans Pattern Recog Letters 152 2021 122 128 10.1016/j.patrec.2021.08.035 15 Rahimzadeh M. Attar A. Sakhaei S.M. A fully automated deep learning-based network for detecting COVID-19 from a new and large lung CT scan dataset Biomed Signal Proces Contrl 68 2021 10.1016/j.bspc.2021.102588 16 Qi Q.Q. Qi S.L. Wu Y.N. Fully automatic pipeline of convolutional neural networks and capsule networks to distinguish COVID-19 from community-acquired pneumonia via CT images Comp in Biol Med 141 2022 10.1016/j.compbiomed.2021.105182 17 Abdel-Basset M. Hawash H. Moustafa N. Two-stage deep learning framework for discrimination between COVID-19 and community-acquired pneumonia from chest CT scans Pattern Recog Letters 152 2021 311 319 10.1016/j.patrec.2021.10.027 18 Ye Q.H. Gqu Y. Ding W.P. Robust weakly supervised learning for COVID-19 recognition using multi-center CT images Appl Soft Computing 116 2022 108291 10.1016/j.asoc.2021.108291 19 Balaha H.M. El-Gendy E.M. Saafan M.M. CovH2SD: a COVID-19 detection approach based on Harris Hawks optimization and stacked deep learning Expert Systems with Appl 186 2021 10.1016/j.eswa.2021.115805 20 Qi S.L. Xu C.W. Li C. DR-MIL: deep represented multiple instance learning distinguishes COVID-19 from community-acquired pneumonia in CT images Comp Methods and Programs in Biomed 211 2021 10.1016/j.cmpb.2021.106406 21 Ahamed K.U. Islam M. Uddin A. A deep learning approach using effective preprocessing techniques to detect COVID-19 from chest CT-scan and X-ray images Comp Biol Med 2021 105014 10.1016/j.compbiomed.2021.105014 22 Kumari S. Ranjith E. Gujjar A. Comparative analysis of deep learning models for COVID-19 detection Global Trans Proc 2 2 2021 559 565 10.1016/j.gltp.2021.08.030 23 M. M. Ahsan, K. D. Gupta, M. M. Islam, et al., “Study of different deep learning approach with explainable AI for screening patients with COVID-19 symptoms: using CT scan and chest X-ray image dataset,” 2020, doi:10.3390/make2040027. 24 Jia G. Lam H.K. Xu Y. Classification of COVID-19 chest X-Ray and CT images using a type of dynamic CNN modification method Computers in Biol and Med 134 April 2021 104425 10.1016/j.compbiomed.2021.104425 25 Kassania S.H. Kassanib P.H. Wesolowskic M.J. Automatic detection of coronavirus disease (COVID-19) in X-ray And CT images: a machine learning based approach Biocybernetics and Biomed Eng 41 3 2021 867 879 10.1016/j.bbe.2021.05.013 26 Gour M. Jain S. Automated COVID-19 detection from X-ray and CT images with stacked ensemble convolutional neural network Biocybernetics and Biomed Eng 42 1 2022 27 41 10.1016/j.bbe.2021.12.001 27 Kamil M.Y. A deep learning framework to detect Covid-19 disease via chest X-ray and CT scan images Int J Electrical and Comp Eng 11 1 2021 844 850 10.11591/ijece.v11i1.pp844-850 28 Saygılı A. A new approach for computer-aided detection of coronavirus (COVID-19) from CT and X-ray images using machine learning methods Appl Soft Computing 105 2021 107323 10.1016/j.asoc.2021.107323 29 M. Tan and Q. v. Le, “EfficientNetV2: smaller models and faster training,” 2021, [Online]. Available: http://arxiv.org/abs/2104.00298. (accessed December 01, 2021) 30 “COVID-19 Radiography Database | Kaggle.” https://www.kaggle.com/tawsifurrahman/covid19-radiography-database (accessed Nov. 04, 2021). 31 “Chest X-Ray Images (Pneumonia) | Kaggle.” https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia (accessed Nov. 04, 2021). 32 “Tuberculosis (TB) chest x-ray database | Kaggle.” https://www.kaggle.com/datasets/tawsifurrahman/tuberculosis-tb-chest-xray-dataset (accessed May 06, 2022). 33 “Large COVID-19 CT scan slice dataset | Kaggle.” https://www.kaggle.com/maedemaftouni/large-covid19-ct-slice-dataset (accessed Nov. 09, 2021). 34 “COVID-19&Normal&Pneumonia_CT_Images | Kaggle.” https://www.kaggle.com/datasets/anaselmasry/covid19normalpneumonia-ct-images (accessed May 08, 2022). 35 M. Tan and Q. V. Le, “EfficientNetV2: smaller models and faster training,” 2021. 36 J. Hu, L. Shen, S. Albanie, et al., “Squeeze-and-excitation networks,” 2017, [Online]. Available: http://arxiv.org/abs/1709.01507 37 Q. Wang, B. Wu, P. Zhu, et al., “ECA-Net: efficient channel attention for deep convolutional neural networks,” 2019, [Online]. Available: http://arxiv.org/abs/1910.03151(accessed February 01, 2022). 38 Opitz D. Maclin R. Popular ensemble methods: an empirical study J Artificial Intelligence Res 11 1999 169 198 10.1613/jair.614 39 M. A. Ganaie and M Hu. “Ensemble deep learning: A review,” arXiv preprint arXiv:2104.02395 (2021). 40 Sollich P. Krogh A. Learning with ensembles: how over-fitting can be useful Proceedings of the 8th International Conference on Neural Information Processing Systems 1995 190 196 Pages 41 Pourtaheri Z.K. Zahiri S.H. Ensemble classifiers with improved overfitting 1st Conference on Swarm Intelligence and Evolutionary Computation 2016 10.1109/CSIEC.2016.7482130 42 Rajaraman S. Kim I. Antani S.K. Detection and visualization of abnormality in chest radiographs using modality-specific convolutional neural network ensembles PeerJ 8 2020 e8693 10.7717/peerj.8693 32211231 43 Devnath L. Fan Z. Luo S. Detection and visualisation of pneumoconiosis using an ensemble of multi-dimensional deep features learned from Chest X-rays Int J Environ Res and Public Health 19 18 2022 11193 10.3390/ijerph191811193 36141457 44 Rajaraman S. Cemir S. Xue Z A novel stacked model ensemble for improved TB detection in chest radiographs Chapter 1, Medical Imaging 2019 CRC Press Boca Raton 45 Devnath L. Luo S. Summons P. Deep ensemble learning for the automatic detection of pneumoconiosis in coal worker's Chest X-ray radiography J Clin Med 11 18 2022 5342 10.3390/jcm11185342 36142989 46 Mishra A.K. Das S.K. Roy P. Identifying COVID19 from Chest CT images: a deep convolutional neural networks based approach J Healthcare Eng 2020 2020 10.1155/2020/8843664 47 M. Saqib, S. Anwar, A. Anwar, et al., “COVID19 detection from radiographs: is deep learning able to handle the crisis?,” no. June, pp. 1–14, 2020, [Online]. Available: www.preprints.org 48 He X. Yang X.Y. Zhang S.H. Sample-efficient deep learning for COVID-19 diagnosis based on CT scans IEEE Transactions on Med Imaging XX Xx 2020 10.1101/2020.04.13.20063941 49 A. Mobiny et al., “Radiologist-level COVID-19 detection using CT Scans with detail-oriented capsule networks,” 2020, [Online]. Available: http://arxiv.org/abs/2004.07407 50 Polsinelli M. Cinque L. Placidi G. A light CNN for detecting COVID-19 from CT scans of the chest Pattern Recog Letters 140 2020 95 100 10.1016/j.patrec.2020.10.001 51 X. Yang, X. He, J. Zhao, et al., “COVID-CT-Dataset: A CT scan dataset about COVID-19,” 2020, [Online]. Available: http://arxiv.org/abs/2003.13865. (accessed December 05, 2021). 52 Hernández Santa Cruz J.F. An ensemble approach for multi-stage transfer learning models for COVID-19 detection from chest CT scans Intelligence-Based Med 5 February 2021 100027 10.1016/j.ibmed.2021.100027 53 Shaik N.S. Cherukuri T.K. Transfer learning based novel ensemble classifier for COVID-19 detection from chest CT-scans Computers in Biol Med 141 June 2021 2021 105127 10.1016/j.compbiomed.2021.105127

0

PMC9748725

NO-CC CODE

2022-12-15 23:22:46

no

Infect Dis Health. 2022 Nov 14; 27:S11

latin-1

Infect Dis Health

10.1016/j.idh.2022.09.038

oa_other

==== Front Acad Radiol Acad Radiol Academic Radiology 1076-6332 1878-4046 The Association of University Radiologists. Published by Elsevier Inc. S1076-6332(22)00632-8 10.1016/j.acra.2022.11.027 Original Investigation Stacking Ensemble and ECA-EfficientNetV2 Convolutional Neural Networks on Classification of Multiple Chest Diseases Including COVID-19 Huang Mei-Ling PhD. in Industrial Engineering & Management ⁎ Liao Yu-Chieh Master in Industrial Engineering & Management Department of Industrial Engineering & Management, National Chin-Yi University of Technology, 57, Sec. 2, Zhongshan Rd., Taiping Dist., Taichung, 411030, Taiwan ⁎ Address correspondence to: M-L.H. 25 11 2022 25 11 2022 6 9 2022 15 11 2022 20 11 2022 © 2022 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved. 2022 The Association of University Radiologists Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Rationale and Objectives Early detection and treatment of COVID-19 patients is crucial. Convolutional neural networks have been proven to accurately extract features in medical images, which accelerates time required for testing and increases the effectiveness of COVID-19 diagnosis. This study proposes two classification models for multiple chest diseases including COVID-19. Materials and Methods The first is Stacking-ensemble model, which stacks six pretrained models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M. The second model is self-designed model ECA-EfficientNetV2 based on ECA-Net and EfficientNetV2. Ten-fold cross validation was performed for each model on chest X-ray and CT images. One more dataset, COVID-CT dataset, was tested to verify the performance of the proposed Stacking-ensemble and ECA-EfficientNetV2 models. Results The best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.21%, Precision of 99.23%, Recall of 99.25%, F1-score of 99.20%, and (area under the curve) AUC of 99.51% on chest X-ray dataset; the best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.81%, Precision of 99.80%, Recall of 99.80%, F1-score of 99.81%, and AUC of 99.87% on chest CT dataset. The differences for five metrics between Stacking-ensemble and ECA-EfficientNetV2 models are not significant. Conclusion Ensemble model achieves better performance than single pretrained models. Compared to the SOTA, Stacking-ensemble and ECA-EfficientNetV2 models proposed in this study demonstrate promising performance on classification of multiple chest diseases including COVID-19. KEY WORDS COVID-19 Convolutional neural network Ensemble learning Stacking ==== Body pmcINTRODUCTION COVID-19 causes disease in humans and vertebrates, and is a zoonotic infectious disease. Some confirmed patients may have severe pneumonia and respiratory failure (1). The most common tools for detecting lung infections are Chest X-ray and Computed tomography (CT). Chest X-ray is a very common noninvasive radiological test that has been widely used to screen for a variety of lung diseases such as COVID-19, pneumonia, pulmonary effusion, lung cancer, and emphysema (2). In clinical practice, chest X-ray images are often interpreted by radiologists, which is time-consuming and prone to errors in subjective assessments. A CT image is composed of a certain number of pixels with different grayscales arranged in a matrix. If the number of pixels is larger, the pixel value is smaller, and the image will be clearer. Although CT images provide very fine details, but it has more radiation than a chest X-ray image, and the equipment is relatively expensive (3). In recent years, due to the rise of artificial intelligence, researchers have applied deep learning to detect COVID-19 by using chest X-ray images and CT images. Compared to traditional machine learning, deep learning can automatically extract features of images to reduce processing time (4). At present, there have been many researches applying convolutional neural network (CNN) on image recognition. This method has been proved to be a powerful image recognition technology and has been widely used for COVID-19 detection. Common CNNs include GoogLeNet, ResNet, Xception, DenseNet, MobileNet, and EfficientNet, etc. For example, Ieracitano et al. proposed a fuzzy logic-based convolutional neural network (CovNNet) and the classification accuracy is 81.00% in 2022 by using a total of 155 chest X-ray images (5). Khan et al. proposed two convolutional neural networks named DHL and DBHL in 2021, using a total of 6448 chest X-ray images. The results showed that the accuracy of the binary classification of COVID-19 and normal was 98.53% (6). Loey et al. proposed a Bayesian-optimized convolutional neural network in 2022 using a total of 10,848 chest X-ray images. The results showed that the three-category accuracy of COVID-19, general pneumonia and normal was 96.00% (7). Hu et al. proposed a two-stage detection method in 2021 using a total of 11,092 chest X-ray images (8). The first stage is to train a CNN as a feature extractor, and the second stage uses extreme learning machines (ELMs) for real-time detection. In addition, the Chimp optimization algorithm is used to improve the results and increase the reliability of the network, and finally it is compared to the general CNN, Genetic algorithm optimized ELM, Cuckoo search optimized ELM, and Whale optimization algorithm optimized ELM. The results show that the proposed method has a binary classification accuracy of 99.11% for COVID-19 and non-COVID-19. Thaseen et al. applied ensemble learning by combining ResNet, FitNet, IRCNN, MobileNet, and EfficientNet in 2022, using a total of 13,808 chest X-ray images (9). The results showed that the three-category accuracy of COVID-19, general pneumonia and normal was 99.00%. A hybrid convolutional neural network (HCNN) combining CNN and RNN on classification of three-category accuracy of COVID-19, general pneumonia and normal was proposed with the classification accuracy of 98.20% (10). Musallam et al. proposed a convolutional neural network called DeepChest in 2022, using a total of 7512 chest X-ray images. The results showed that the three-category accuracy of COVID-19, general pneumonia and normal was 96.56% (11). A convolutional neural network named DenResCov-19 composing of DenseNet121 and ResNet50 networks was proposed, and DenResCov-19 was combined with existing ResNet50, DenseNet121, VGG16, and InceptionV3 networks, using a total of 6469 chest X-ray images. The results showed that the four-category area under the curve (AUC)-ROC of COVID-19, general pneumonia, tuberculosis and normal was 99.60% (12). The literature on COVID-19 detection using CT images such as Garg et al. used a series of models such as EfficientNet, DenseNet, VGG, and ResNet with a total of 20 convolutional neural networks in 2022, using a total of 4173 CT images. The results show that the binary classification accuracy of EfficientNet-B5 in detecting COVID-19 and non-COVID-19 is 98.45% (13). Lahsaini et al. used transfer learning on VGG16, VGG19, Xception, InceptionV2, ResNet, DenseNet121, and DenseNet201 in 2021, combined with GradCam, using a total of 4986 CT images. The results show that in the binary classification of COVID-19 and non-COVID-19, DenseNet201+ GradCam achieves the best accuracy rate of 98.80% (14). Rahimzadeh et al. used ResNet50V2 as the backbone network and compared the model with ResNet50V2 and Xception after adding a feature pyramid network (FPN), using a total of 63,849 CT images. The results show that ResNet50V2+ FPN has an accuracy of 98.49% in COVID-19 and normal binary classification (15). Qi et al. first used five models of UNet, LinkNet, R2UNet, Attention UNet, and UNet++ to segment CT images, and then used pretrained DenseNet121, InceptionV3 and ResNet50 for classification, using a total of Over 10,000 CT images. The results show that in the binary classification of COVID-19 and CAO, LinkNet performs best in lung segmentation with a Dice coefficient of 0.9830, while DenseNet121 with capsule network has a prediction accuracy of 97.10% (16). Abdel-Basset et al. proposed a two-stage detection method in 2021. The first stage is to use the proposed GR-UNet to segment the area of lung infection, and then transfer learning is used as feature extraction; the second stage is to use the proposed GR-UNet to segment the lung infected area. The stage is to propose an infection prediction module that uses the infected location to make decisions about classification, using a total of 9,593 CT images. The results showed that the binary classification accuracy of COVID-19 and CAP was 96.80% (17). Ye et al. proposed a convolutional neural network named CIFD-Net in 2022, which can effectively handle the multi-region displacement problem through a new robust supervised learning, using a total of 45,167 CT images picture. The results showed that the binary classification accuracy of COVID-19 and non-COVID-19 was 91.19% (18). Balaha et al. used transfer learning models including ResNet50, ResNet101, VGG16, VGG19, Xception, MobileNetV1, MobileNetV2, DenseNet121, and DenseNet169 in 2021, and added the Harris Hawks optimization to optimize hyper-parameters, and finally use fast classification stage and compact stacking stage to stack the best models into one, using a total of 15,535 CT images. The results show that in the binary classification of COVID-19 and non-COVID-19, the weighted sum method (WSM) is used to obtain an accuracy of 99.33% (19). Qi et al. proposed a detection method named DR-MIL in 2021, which first treats a 3D CT image of a patient as a bag and selects ten CT slices as initial instances. The deep features were then extracted from the pretrained ResNet50 by fine-tuning and treated as a Deep Represented Instance Score. The bag with DRIS was input to K-Nearest Neighbor (KNN) to generate the final prediction, using a total of 241 patients CT images. The results showed that the binary classification accuracy of COVID-19 and CAP was 95.00% (20). Some scholars used both chest X-ray images and CT images to detect COVID-19. Related studies such as Ahamed et al. fine-tune the pretrained ResNet50V2 in 2021, using a total of 4593 chest X-ray images and 3000 CT images. The results showed that the four-category accuracy for detecting COVID-19, viral pneumonia, bacterial pneumonia, and normal in chest X-ray images was 96.45%; the three-category accuracy for COVID-19, bacterial pneumonia and normal was 97.24%; COVID-19 and normal binary classification accuracy was 98.95%. The three-category accuracy for detecting COVID-19, CAP, and normal in CT images was 99.01%; the two-category accuracy for COVID-19 and normal was 99.99% (21). Kumari et al. used four convolutional neural networks, InceptionV3, VGG16, Xception and ResNet50 in 2021, using a total of 2000 chest X-ray images and 2000 CT images. The results show that the binary classification accuracy of VGG16 in detecting COVID-19 and non-COVID-19 in chest X-ray images was 98.00%; the binary classification accuracy of Xception in detecting COVID-19 and non-COVID-19 in CT images was 83.00% (22). Ahsan et al. fine-tuned eight convolutional neural networks including VGG16, VGG19, ResNet15V2, InceptionResNetV2, ResNet50, DenseNet201, MobilenetV2, and NasNetMobile in 2020, using a total of 400 chest X-ray images and 400 CT images. The results show that the binary classification accuracy of NasNetMobile in detecting COVID-19 and non-COVID-19 in chest X-ray images was 100.00%; the binary classification accuracy of NasNetMobile in detecting COVID-19 and non-COVID-19 in CT images was 95.20% (23). Jia et al. proposed a dynamic CNN modification method in 2021 and applied it to a fine-tuned ResNet, and finally compared the results with VGG16, InceptionV3, ResNet18, DenseNet121, MobileNetV3, and SqueezeNet, using a total of 7,592 chests X-ray images and 104,009 CT images. The results showed that the five-category accuracy of detecting COVID-19, viral pneumonia, bacterial pneumonia, tuberculosis and normal in chest X-ray images was 99.60%; in CT images, detecting COVID-19, non-COVID-19 and normal three-category accuracy was 99.30% (24). Kassania et al. use eight models such as DenseNet, ResNet, MobileNet, InceptionV3, Xception, InceptionResNetV2, VGG and NASNet to extract features, and then extracted features were input to decision tree, random forest, XGBoost, AdaBoost, Bagging, and LightGBM. A total of 137 chest X-ray images and 137 CT images were used. The results show that DenseNet121+ Bagging combines chest X-ray images and CT images to detect COVID-19 and normal with a binary classification accuracy of 99.00% (25). Gour and Jain proposed an integrated stacked CNN in 2022. After fine-tuning VGG19 and Xception and generating five sub-models, all sub-models were stacked using a softmax classifier with a total of 3040 images Chest X-ray images and 4,645 CT images. The results showed that the three-category accuracy of detecting COVID-19, general pneumonia, and normal in chest X-ray images was 97.27%; and the two-category accuracy of detecting COVID-19 and normal in CT images was 98.30% (26). Kamil et al. used fine-tuned VGG19 in 2021, using a total of 977 chest X-ray images and 23 CT images. The results showed a 99.00% accuracy of the binary classification between COVID-19 and normal (27). Saygılı applied Bag of Tree, K-ELM, KNN, and SVM with a total of 1125 chest X-ray images and 3228 CT images. The results showed that SVM has an accuracy of 85.96% in detecting COVID-19, general pneumonia and normal in chest X-ray images; K-ELM is accurate in detecting COVID-19 and non-COVID-19 in CT images. The accuracy was 98.88% (28). Existing pretrained models are designed for general natural images and fine-tuned for classified image types, which is not specifically designed for COVID-19 detection. The general natural images are large and simple, while the images of COVID-19 have specific patterns and textures that differ significantly from natural images. Based on previous studies, we can find that the use of ensemble learning and the author's self-proposed convolutional neural network has good performance in detecting COVID-19 in chest X-ray images and CT images. Ensemble learning mainly combines multiple existing convolutional neural networks, which can not only reduce the probability of misjudgment by a single model, but also improve the classification accuracy in less time. Ensemble learning solves the current need to detect COVID-19 without designing a new model to get good detection performance in the most time-saving way. Challenging the possibility of improving accuracy on multiple chest disease diagnosis, this study proposes two classification models for multiple chest diseases including COVID-19. First, we obtained chest X-ray images and CT images from multiple public databases. Next, we select six pretraining models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M in the EfficientNetV2 series. The reason for choosing the previous models is the training speed and argument efficiency of EfficientNetV2 is better than some previous networks (29). To the best of our knowledge, this is the first study ensembles the series of EfficientNetV2 models for COVID-19 detection. In addition, this study proposes a self-designed model ECA-EfficientNetV2 based on ECA-Net and EfficientNetV2. MATERIALS AND METHODS Figure 1 represents the architecture of this study.• Step 1 Data extraction: The chest X-ray and CT images are collected. • Step 2 Image preprocessing: The size of all selected images is equalized and saved in PNG. The datasets are split into training, validation, and test subsets. • Step 3 Pretrained models: Six EfficientNetV2 models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, and EfficientNetV2-M are used. • Step 4 Proposed Ensemble-stacking model: The previous six EfficientNetV2 models are stacked. • Step 5 Proposed model: ECA-EfficientNetV2 model. • Step 6 Performance evaluation: Accuracy, precision, recall, F1-score, and AUC are recorded for each model. • Step 7 Comparison with SOTA: The results of models are compared to the related studies. Figure 1 The architecture of this study. (Color version of figure is available online.) Figure 1 Data Extraction Chest X-ray images and CT images are collected from five Kaggle datasets described as follows. The first Kaggle dataset is COVID-19 Radiography Dataset (30), which contains 3616 COVID-19, 1345 viral pneumonia, 6012 lung opacity and 10,192 normal for a total of 21,165 chest X-ray images (PNG). The second Kaggle dataset is Chest X-Ray Images (Pneumonia) (31), collected from pediatric patients aged 1 to 5 from Guangzhou Women and Children's Medical Center with 4273 pneumonia and 1583 normal chest X-ray images (JPEG). The third Kaggle dataset is Tuberculosis Chest X-ray Dataset (32), which contains 3500 tuberculosis and 3500 normal chest X-ray images (PNG). Large COVID-19 CT scan slice dataset (33), containing 7593 COVID-19, 2618 chest abdominal pelvis and 6893 normal CT images for a total of 17,104 CT images (PNG), is the forth Kaggle dataset used in this study. The fifth Kaggle dataset is COVID-19 and Normal and PneumoniaCT_Images (34), containing 2035 COVID-19, 3309 pneumonia and 2119 normal CT images for a total of 7463 CT images (PNG). The COVID-19 X-ray images was formed based on the previous datasets 1–3. The pneumonia images in the second dataset were split into groups of viral pneumonia and bacterial pneumonia. The total COVID-19 X-ray images contain five groups, which are 3616 COVID-19, 2780 viral pneumonia, 2838 bacterial pneumonia, 3500 Tuberculosis and 15,275 Normal for a total of 28,009 chest X-ray images (PNG). The COVID-19 CT images was formed based on the previous datasets 4 and 5, which contains four groups including 9628 COVID-19, 2618 Chest Abdomen Pelvis (CAP), 3309 pneumonia, and 9012 Normal for a total number of 24567 chest CT images. Image Preprocessing To equalize the number of images for each group, 1200 images are randomly selected for each group for both X-ray and CT datasets in this study. The number of images in training, validation and test subsets are 900, 100, and 200, respectively for X-ray dataset; The number of images in training, validation and test subsets are 4500, 500, and 400, respectively for CT dataset. Ten-fold cross validation was performed in this study. Tables 1 and 2 display the number of images for each group for COVID-19 X-ray dataset (Dataset X-ray) and COVID-19 CT dataset (Dataset CT) used in the study.Table 1 COVID-19 X-ray Dataset (Dataset X-ray) Table 1Group Training Validation Test COVID-19 900 100 200 Bacterial pneumonia 900 100 200 Tuberculosis 900 100 200 Viral pneumonia 900 100 200 Normal 900 100 200 Total 4500 500 1000 Table 2 COVID-19 CT Dataset (Dataset CT) Table 2Group Training Validation Test COVID-19 900 100 200 CAP 900 100 200 Pneumonia 900 100 200 Normal 900 100 200 Total 3600 400 800 The original images collected from the previous datasets differ in size and format. As the model requirement, all the images are preprocessed to in PNG with size of 224  ×  224. Figure 2, Figure 3 display the examples of images after preprocessed for chest X-ray (Dataset X-ray) and CT (Dataset CT) images, respectively.Figure 2 Chest X-ray images after preprocessing. (Color version of figure is available online.) Figure 2 Figure 3 CT images after preprocessing. (Color version of figure is available online.) Figure 3 Convolutional Neural Network Convolutional Neural Network (CNN) is a feedforward neural network, mainly composed of multiple convolution layers, pooling layers and fully connected layers. Compared to other neural networks, convolutional neural networks have better performances on image or speech recognitions. The goal of training a convolutional neural network is to find the most appropriate weights in the process of multiple forward and reverse iterations. Transfer learning is having previously trained model on a larger database that we can directly apply the architecture and weights of the pretrained model to various studies to speed the efficiency of the training model. The Application module in Keras currently provides about 40 pretrained models, all of which are trained on the ImageNet dataset. We choose the current newer and efficient EfficientNetV2, which includes six series of models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M. The reason for not choosing EfficientNetV2-L is that the model has a large number of parameters and cannot be executed in our existing hardware resources. EfficientNetV2 There are some problems with the previous EfficientNet series models, such as (1) when the size of the image is large, the training speed for EfficientNet-B3 to EfficientNet-B7 is very slow. (2) Training is slow when Depthwise convolutions is used. (3) It is not the best choice to enlarge each layer with the same magnification. Constantly increasing image size, which also leads to large memory consumption, which in turn affects training speed. Therefore, Tan and Le (35) proposed a new convolutional neural network EfficientNetV2 in 2021, which uses a nonuniform scaling strategy and can increase the number of deeper layers and has faster training speed and better parameter efficiency as compared to the previous models. In addition, Tan and Le (35) also used EfficientNet-B4 to do some experiments, and found that replacing the MBConv module with the Fused-MBConv module in the early stage can greatly improve the training speed. However, if each layer is replaced with the Fused-MBConv module, the number of parameters and FLOPs will be significantly increased, and the training speed will be relatively reduced. Therefore, Neural Architecture Search (NAS) was used to find the best combination of MBConv and Fused-MBConv. Figure 4 is the architecture of EfficientNetV2.Figure 4 Architecture of EfficientNetV2. (Color version of figure is available online.) Figure 4 ECA-Net Channel attention has greatly improved the performance of convolutional neural networks, and most scholars are currently working on developing more complex attention modules. The most representative method such as Hu et al. (36) used Squeeze-and-excitation (SE) module and proposed SE-Net in 2017. SE-Net first uses a global average pooling layer for each channel, and then uses two nonlinear fully connected layers and a sigma function to generate the weights for each channel. Although the SE module is widely used in some researches on the current channel attention module, it has been proved that dimensionality reduction will affect both the prediction performance for channel attention, and the efficiency of obtaining the weights between all channels. Wang et al. (37) uses a lightweight and efficient channel attention (ECA) module, which only adds a small number of parameters, but can achieve significant performance gains. The ECA module does not use dimensionality reduction and operates through a one-dimensional convolution of size k. According to the experimental results of ECA-Net on ImageNet-1K and MS COCO, ECA-Net has lower model complexity than the state-of-the-art methods. In addition, ECA-Net has better efficiency no matter in image classification, image segmentation or object detection. The Proposed Stacking-Ensemble Model Ensemble learning is a type of supervised learning that has been widely used in the fields of statistics, machine learning and deep learning. Compared to a single learning algorithm, the purpose of ensemble learning is to combine multiple algorithms or models to form a model with better predictive performance. Using an ensemble approach will yield better results when there are significant model-to-model differences (38,39) reviews on ensemble deep leaning. Details of ensemble classifiers with improved overfitting have been investigated by several studies, for instance (40,41). In recent years, due to the continuous improvement of computing power of computers, large-scale integrated models can be trained within a reasonable time, and have been applied on medical image recognition, face recognition, emotion recognition and financial decision-making, etc. Successful applications of ensemble classifiers could be seen in (42, 43, 44, 45). The main methods of ensemble learning can be divided into three categories: bagging, boosting and stacking. The main function of stacking is to combine multiple algorithms to make predictions, that is, the result integration of the voting method or the average method. Ensemble stacking is a powerful CNN ensemble method thanks to its unique meta-learning algorithm. Meta-learning algorithms work by taking the probabilities of the input from each sub-model and determining which model performs best in extracting features. The learning algorithms directly extend the learning of each sub-model and combine the best predictions from each model. If each model is unique, then each model learns differently. Stacking achieves better results than either trained model and is used to evaluate the error rate of bagging. This study chooses stacking as our approach for ensemble learning. The stacking-ensemble model developed in this study selects six pretrained models including EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S and EfficientNetV2-M in the EfficientNetV2 series. Figure 5 presents the architecture of the proposed stacking-ensemble model in this study.Figure 5 Stacking-ensemble model. (Color version of figure is available online.) Figure 5 The Proposed ECA-EfficientNetV2 Model In addition to the proposed stacking-ensemble model, challenging the possibility of improving accuracy on multiple chest disease diagnosis, this study proposes a self-designed model ECA-EfficientNetV2 based on ECA-Net and EfficientNetV2. Motivation of choosing the previous two models as basis are described as follows:1. The advantages of EfficientNetV2 includes: (1) The network is better than some previous networks in terms of training speed and number of parameters; (2) It proposes improved incremental learning methods that dynamically adjust regularization (e.g., dropout, data augmentation, and mix-up) according to the size of the training images; (3) Progressive learning is used to perform well on pretrained ImageNet, CIFAR, cars, and flowers datasets. 2. The major advantages of ECA-Net are: (1) ECA is a lightweight attention module that only contains k parameters (k ≤ 9), which can be used to improve the performance of large convolutional neural networks; (2) The ECA module uses a dimensionality-free local cross-channel interaction method that adaptively selects suitable adjacent channels to compute attention; (3) The features extracted by the SE module between different channels are relatively similar, while the features extracted by the ECA module from different channels are different, which indicates that the features extracted by the ECA module are better for classification. Based on the previous two points, the proposed ECA-EfficientNetV2 are designed and introduced as follows:1. ECA-EfficientNetV2 uses the dilated convolution module as the first two layers. The advantage is that while the kernel size is increased, the parameters or calculation amount of the original model can be maintained. Each module contains two dilated convolutional layers and an activation function SELU, where the dilation rate of the first layer is set to two, and the dilation rate of the second layer is set to three. 2. To reduce both the number of parameters and the computation complexity, ECA-EfficientNetV2 replaces SE module with ECA module in MBConv and Fused-MBConv convolution modules, and renames as MBEConv and Fused-MBEConv. 3. Zero-padding is added to the convolutional layers in both the MBEConv and Fused-MBEConv modules to prevent the input image be affected by the kernel size. In addition, the original activation function SELU is changed to ReLU to overcome the problem of vanishing gradient. 4. After MBEConv and Fused-MBEConv, two general convolution modules are added. The internal parameters include Zero-padding, Stride, and SELU. In addition, we add a batch normalization after each convolutional layer, which makes training easier and more stable, and improves the performance of the neural network. 5. Use the global average pooling layer to improve the problem of a large number of parameters that occurs in the fully connected layer. After that, add a dropout layer before the classification layer to generate multiple results by continuously updating the weights, and finally remove the outliers to avoid the problem of overfitting. The number of parameters for ECA-EfficientNetV2 is 5,706,965, which is much less than the number of parameters (117,753,253) used in EfficientNetV2-L. Figure 6 presents the architecture of the proposed Fused-MBEConv layer and MBEConv layers, and Figure 7 shows the architecture of the proposed ECA-EfficientNetV2.Figure 6 The architecture of the proposed Fused-MBEConv layer and MBEConv layers. (H, W, C denotes network height, width and channel respectively.) (Color version of figure is available online.) Figure 6 Figure 7 Architecture of ECA-EfficientNetV2. (Color version of figure is available online.) Figure 7 Model Evaluation Measures The performance indices include accuracy, precision, recall, F1-Score and the AUC. The mathematical formulas of these indicators are shown in Eqs. (1–5), respectively. Accuracy is the ratio of the number of correctly classified samples to the total number of samples; Precision is the ratio of the number of true positives to the total number of elements labelled to the positive class; Recall presents the number of true positives divided by the total number of true positives and false negatives; F1-Score is a measure of precision and recall. The higher the F1-score, the better the classification performance of the model; AUC denotes the measure at distinguishing between the positive and negative classes. Higher the AUC, better the model.(1) Accuracyc=(TPc+TNc)(TPc+FPc+FNc+TNc) (2) Precisionc=TPc(TPc+FPc) (3) Recallc=TPc(TPc+FNc) (4) F1−score=2n∑c=1nRecallc×PrecisioncRecallc+Precisionc (5) AUC=∑cnrankn−TPc×(1+TPc)2TPc×TNc where c is the number of classes; TP (true positive) represents the number of positive categories that are correctly classified as positive, FP (false positive) represents the number of negative categories that are incorrectly classified as positive, TN (true negative) refers to the number of negative categories that are correctly classified as negative, and FN (false negative) refers to the number of positive categories that are incorrectly classified as negative. RESULTS The equipment used in the experiment is an Intel(R) Core(TM) i9-10900F 2.81 GHz CPU, NVIDIA GeForce RTX 3070 8G GPU. The whole experiment process is performed using Python 3.8 [Python Software Foundation, Fredericksburg, Virginia, USA], which contains Keras 2.6 and Tensorflow 2.6. Ten-fold cross validation is used to evaluate the performance of each model under batch size of 16; epochs count of 30; optimizer of Adam, learning rate of 1e-5 and dropout of 0.4. Performance Results on Dataset X-ray On dataset 1, Table 3, Table 4, Table 5, Table 6 report the results of Accuracies, Precisions, Recalls, F1-Scores, and AUCs in the test sets, and the average and standard deviation for EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models. The best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.21%, Precision of 99.23%, Recall of 99.25%, F1-score of 99.20%, and AUC of 99.51%. Figure 8, Figure 9 show the examples of training accuracy and loss, confusion matrix and ROC curve for each model, respectively.Table 3 Performance of EfficientNetV2-B0 and EfficientNetV2-B1 Models Table 3Fold EfficientNetV2-B0 EfficientNetV2-B1 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 82.10% 82.33% 82.14% 82.13% 88.81% 83.10% 83.23% 83.14% 83.11% 89.44% 2 85.20% 84.99% 85.24% 85.06% 90.75% 85.90% 85.81% 85.93% 85.85% 91.19% 3 88.50% 88.62% 88.53% 88.57% 92.81% 86.70% 86.69% 86.72% 86.59% 91.69% 4 85.50% 85.73% 85.51% 85.49% 90.94% 82.30% 82.26% 82.34% 82.25% 88.94% 5 85.90% 85.93% 85.91% 85.96% 91.19% 84.30% 84.47% 84.32% 84.34% 90.19% 6 84.40% 84.46% 84.42% 84.43% 90.25% 85.20% 85.15% 85.23% 85.16% 90.75% 7 85.60% 85.62% 85.64% 85.56% 91.00% 86.50% 86.40% 86.54% 86.42% 91.56% 8 84.40% 84.59% 84.46% 84.34% 90.25% 85.82% 85.94% 85.81% 85.84% 91.13% 9 84.50% 84.80% 84.51% 84.54% 90.31% 88.90% 89.01% 88.92% 88.93% 93.06% 10 86.20% 86.42% 86.23% 86.22% 91.38% 87.60% 87.79% 87.64% 87.65% 92.25% Average 85.23± 1.63% 85.35± 1.61% 85.26± 1.63% 85.23± 1.65% 90.77± 1.02% 85.63± 2.00% 85.68± 2.02% 85.66± 2.00% 85.61± 2.01% 91.02± 1.25% Table 4 Performance of EfficientNetV2-B2 and EfficientNetV2-B3 Models Table 4Fold EfficientNetV2-B2 EfficientNetV2-B3 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 79.90% 79.85% 79.91% 79.76% 87.44% 85.40% 85.72% 85.43% 85.37% 90.88% 2 84.40% 84.63% 84.42% 84.30% 90.25% 86.30% 86.19% 86.32% 86.22% 91.44% 3 86.60% 87.08% 86.61% 86.43% 91.69% 87.00% 87.27% 87.42% 86.97% 91.88% 4 81.70% 81.71% 81.74% 81.63% 88.56% 84.20% 84.06% 84.23% 84.11% 90.12% 5 83.70% 83.80% 83.72% 83.74% 89.81% 85.70% 85.84% 85.71% 85.69% 91.06% 6 80.60% 80.98% 80.64% 80.59% 87.88% 84.60% 84.94% 84.62% 84.61% 90.38% 7 83.20% 83.14% 83.23% 83.15% 89.50% 86.40% 86.67% 86.42% 86.38% 91.50% 8 87.30% 87.36% 87.34% 87.27% 92.06% 85.20% 85.49% 85.21% 85.27% 90.75% 9 80.60% 80.89% 80.64% 80.65% 87.88% 87.10% 87.22% 87.14% 87.12% 91.94% 10 85.10% 85.31% 85.14% 85.15% 90.69% 85.90% 86.26% 85.94% 85.92% 91.19% Average 83.31± 2.59% 83.48± 2.63% 83.34± 2.59% 83.27± 2.58% 89.58± 1.62% 85.78± 0.96% 85.97± 0.99% 85.84± 1.02% 85.77± 0.96% 91.11± 0.60% Table 5 Performance of EfficientNetV2-S and EfficientNetV2-M Models Table 5Fold EfficientNetV2-S EfficientNetV2-M Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 86.90% 86.82% 86.94% 86.81% 91.81% 87.30% 87.29% 87.34% 87.21% 92.06% 2 88.30% 88.31% 88.34% 88.27% 92.69% 88.70% 88.74% 88.71% 88.68% 92.64% 3 87.70% 88.01% 87.92% 87.95% 92.44% 89.00% 89.04% 89.27% 88.97% 93.13% 4 87.00% 87.37% 87.42% 87.03% 91.88% 87.70% 87.65% 87.72% 87.67% 92.31% 5 87.50% 87.55% 87.52% 87.48% 92.19% 89.40% 89.43% 89.41% 89.44% 93.38% 6 87.00% 86.93% 87.12% 86.91% 91.88% 89.70% 89.71% 89.73% 89.65% 93.56% 7 87.10% 87.13% 87.21% 87.05% 91.94% 89.00% 88.90% 89.29% 88.94% 93.13% 8 88.00% 88.03% 88.28% 87.98% 92.50% 89.70% 89.72% 89.74% 89.69% 93.56% 9 87.10% 87.13% 87.14% 87.11% 91.94% 86.70% 86.98% 86.74% 86.76% 91.69% 10 87.50% 87.69% 87.53% 87.56% 92.19% 89.30% 89.33% 89.31% 89.34% 93.31% Average 87.41± 0.47% 87.50± 0.51% 87.54± 0.49% 87.42± 0.51% 92.15± 0.31% 88.65± 1.05% 88.68± 1.01% 88.73± 1.07% 88.64± 1.05% 92.88± 0.66% Table 6 Performance of Stacking-Ensemble and ECA-EfficientNetV2 Models Table 6Fold Proposed Stacking-Ensemble Model Proposed ECA-EfficientNetV2 Accuracy Precision Recall F1-score AUC Accuracy Precision Recall F1-Score AUC 1 98.40% 98.42% 98.47% 98.39% 99.00% 98.80% 98.84% 98.81% 98.79% 99.25% 2 98.70% 98.71% 98.74% 98.72% 99.19% 99.60% 99.64% 99.61% 99.59% 99.75% 3 98.90% 98.91% 98.93% 98.94% 99.31% 98.80% 98.82% 98.84% 98.78% 99.25% 4 98.50% 95.54% 98.51% 98.53% 99.06% 99.50% 99.51% 99.54% 99.49% 99.69% 5 99.00% 99.01% 99.04% 98.99% 99.38% 99.10% 99.12% 99.11% 99.09% 99.44% 6 99.00% 99.02% 99.05% 98.99% 99.38% 99.40% 99.41% 99.44% 99.37% 99.62% 7 99.00% 99.05% 99.03% 99.01% 99.38% 99.50% 99.48% 99.54% 99.53% 99.69% 8 98.30% 98.35% 98.31% 98.29% 98.94% 99.00% 99.04% 99.07% 98.99% 99.38% 9 98.60% 98.64% 98.67% 98.62% 99.13% 99.00% 99.02% 99.07% 98.97% 99.38% 10 98.70% 98.73% 98.71% 98.75% 99.19% 99.40% 99.41% 99.44% 99.38% 99.62% Average 98.71± 0.26% 98.44± 1.05% 98.75± 0.26% 98.72± 0.26% 99.20± 0.16% 99.21± 0.30% 99.23± 0.30% 99.25± 0.30% 99.20± 0.31% 99.51± 0.19% Figure 8 Examples of training accuracy and loss from each model. (Color version of figure is available online.) Figure 8 Figure 9 Examples of confusion matrix and ROC curve from each model. (Color version of figure is available online.) Figure 9 Performance Results on Dataset CT On dataset 2, Table 7, Table 8, Table 9, Table 10 report the results of Accuracies, Precisions, Recalls, F1-Scores, and AUCs in the test sets, and the average and standard deviation for EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models. The best performance comes from the proposed ECA-EfficientNetV2 model with the highest Accuracy of 99.81%, Precision of 99.80%, Recall of 99.80%, F1-score of 99.81%, and AUC of 99.87%. Figure 10, Figure 11 show the examples of training accuracy and loss, confusion matrix and ROC curve for each model, respectively.Table 7 Performance of EfficientNetV2-B0 and EfficientNetV2-B1 Models Table 7Fold EfficientNetV2-B0 EfficientNetV2-B1 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 95.13% 95.22% 95.15% 95.10% 96.75% 93.63% 93.65% 93.62% 93.59% 95.75% 2 93.87% 93.88% 93.85% 93.86% 95.92% 96.25% 96.41% 96.25% 96.24% 97.50% 3 95.63% 95.60% 95.62% 95.59% 97.08% 95.13% 95.32% 95.15% 95.07% 96.75% 4 95.13% 95.09% 95.15% 95.08% 96.75% 93.25% 93.20% 93.25% 93.17% 95.50% 5 95.63% 95.78% 95.62% 95.60% 97.08% 93.00% 93.03% 93.17% 92.88% 95.33% 6 94.25% 94.29% 94.27% 94.23% 96.17% 93.25% 93.37% 93.27% 93.28% 95.50% 7 93.25% 93.43% 93.28% 93.13% 95.50% 94.13% 94.35% 94.17% 94.04% 96.08% 8 94.37% 94.39% 94.35% 94.36% 96.25% 90.87% 90.85% 90.88% 90.77% 93.92% 9 93.50% 93.51% 93.54% 93.47% 95.67% 95.13% 95.11% 95.15% 95.17% 96.75% 10 94.63% 94.60% 94.56% 94.61% 96.42% 91.37% 91.27% 91.75% 91.27% 94.25% Average 94.54± 0.84% 94.58± 0.83% 94.54± 0.83% 94.50± 0.85% 96.36± 0.56% 93.60± 1.67% 93.66± 1.74% 93.67± 1.61% 93.55± 1.70% 95.73± 1.11% Table 8 Performance of EfficientNetV2-B2 and EfficientNetV2-B3 Models Table 8Fold EfficientNetV2-B2 EfficientNetV2-B3 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 94.87% 98.88% 94.85% 94.83% 96.58% 94.37% 94.54% 94.38% 94.36% 96.25% 2 97.00% 96.99% 97.41% 96.87% 98.00% 95.63% 95.65% 95.62% 95.61% 97.08% 3 94.87% 94.92% 94.88% 94.84% 96.58% 94.87% 95.02% 94.85% 94.83% 96.58% 4 95.00% 94.97% 95.02% 94.94% 96.67% 95.13% 95.10% 95.12% 95.08% 96.75% 5 95.00% 94.99% 95.17% 94.97% 96.67% 96.37% 96.39% 96.38% 96.36% 97.58% 6 92.87% 92.95% 92.88% 92.85% 95.25% 95.75% 95.74% 95.75% 95.73% 97.17% 7 92.13% 92.19% 92.15% 92.03% 94.75% 95.00% 95.21% 95.14% 94.97% 96.67% 8 94.63% 94.58% 94.65% 94.59% 96.42% 95.50% 95.54% 95.51% 95.48% 97.00% 9 95.25% 95.24% 95.19% 95.24% 96.83% 94.37% 94.36% 94.35% 94.76% 96.25% 10 92.50% 92.45% 92.47% 92.38% 95.00% 94.00% 93.98% 94.37% 93.97% 96.00% Average 94.41± 1.48% 94.82± 2.04% 94.47± 1.57% 94.35± 1.49% 96.28± 0.99% 95.10± 0.73% 95.15± 0.72% 95.15± 0.68% 95.12± 0.70% 96.73± 0.49% Table 9 Performance of EfficientNetV2-S and EfficientNetV2-M Models Table 9Fold EfficientNetV2-S EfficientNetV2-M Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 96.37% 96.36% 96.35% 96.63% 97.58% 95.25% 95.22% 95.27% 95.22% 96.83% 2 96.63% 96.71% 96.62% 96.65% 97.75% 96.13% 96.14% 96.15% 96.10% 97.42% 3 97.38% 97.40% 97.37% 97.39% 98.25% 95.50% 95.54% 95.51% 95.47% 97.00% 4 96.63% 96.62% 96.65% 96.33% 97.75% 95.13% 95.12% 95.15% 95.10% 96.75% 5 96.50% 96.47% 96.54% 96.48% 97.67% 96.75% 96.76% 96.73% 96.78% 97.83% 6 95.75% 95.74% 95.77% 95.78% 97.17% 96.00% 96.05% 96.04% 95.99% 97.33% 7 96.50% 96.58% 96.54% 96.51% 97.67% 95.25% 95.28% 95.26% 95.24% 96.83% 8 96.88% 96.86% 96.87% 96.89% 97.92% 96.50% 96.51% 95.54% 96.49% 97.67% 9 95.63% 95.67% 95.65% 95.62% 97.08% 95.37% 95.33% 95.35% 95.34% 96.92% 10 95.13% 95.09% 95.12% 95.10% 96.75% 94.75% 94.70% 94.77% 94.72% 96.50% Average 96.34± 0.66% 96.35± 0.67% 96.35± 0.66% 96.34± 0.67% 97.56± 0.44% 95.66± 0.65% 95.67± 0.67% 95.58± 0.57% 95.65± 0.66% 97.11± 0.43% Table 10 Performance of Stacking-ensemble and ECA-EfficientNetV2 Models Table 10Fold Stacking-ensemble model ECA-EfficientNetV2 Accuracy Precision Recall F1-Score AUC Accuracy Precision Recall F1-Score AUC 1 99.00% 99.01% 99.24% 99.89% 99.33% 99.38% 99.34% 99.35% 99.37% 99.58% 2 98.62% 98.67% 98.63% 98.54% 99.08% 99.88% 99.87% 99.85% 99.89% 99.92% 3 99.38% 99.46% 99.74% 99.25% 99.58% 99.75% 99.72% 99.71% 99.77% 99.83% 4 98.75% 98.74% 98.81% 98.93% 99.17% 99.25% 99.27% 99.24% 99.28% 99.50% 5 98.75% 98.78% 98.87% 98.79% 99.17% 99.88% 99.85% 99.87% 99.84% 99.92% 6 99.88% 99.87% 99.85% 99.89% 99.92% 99.94% 99.92% 99.96% 99.95% 99.98% 7 98.25% 98.33% 98.41% 98.24% 98.83% 100.00% 100.00% 100.00% 100.00% 100.00% 8 98.75% 98.77% 98.75% 98.72% 99.17% 100.00% 100.00% 100.00% 100.00% 100.00% 9 98.88% 98.87% 98.85% 98.86% 99.25% 100.00% 100.00% 100.00% 100.00% 100.00% 10 98.13% 98.16% 98.15% 98.12% 98.75% 100.00% 100.00% 100.00% 100.00% 100.00% Average 98.84± 0.51% 98.87± 0.50% 98.93± 0.54% 98.92± 0.61% 99.23± 0.34% 99.81± 0.27% 99.80± 0.27% 99.80± 0.28% 99.81± 0.27% 99.87± 0.18% Figure 10 Examples of training accuracy and loss from each model. (Color version of figure is available online.) Figure 10 Figure 11 Examples of confusion matrix and ROC curve from each model. (Color version of figure is available online.) Figure 11 DISCUSSION Dataset X-ray The performances of 5 metrics among EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models for dataset X-ray were compared in Table 11 . Two accuracies are above 98%, which come from Stacking-ensemble and ECA-EfficientNetV2 models, while the accuracies from six pretrained models are inferior. In fact, the highest two performance for five metrics from Stacking-ensemble and ECA-EfficientNetV2 models are significantly different from six pretrained models (p-value < 0.01). Although the best performance, the accuracy (99.21%), precision (99.23%), recall (99.25%), F1-score (99.20%), and AUC (99.51%), comes from the proposed ECA-EfficientNetV2 model, the differences for five metrics between Stacking-ensemble and ECA-EfficientNetV2 models are not significant. Comparing the standard deviations of five metrics from Stacking-ensemble and ECA-EfficientNetV2 models, those from the latter are relatively small, which stands the stability of the model.Table 11 Performance Comparison on Dataset X-ray Table 11Models Dataset X-ray Accuracy (%) Precision (%) Recall (%) F1-Score (%) AUC (%) EfficientNetV2-B0 85.23(±1.63) 85.35(±1.61) 85.26(±1.63) 85.23(±1.65) 90.77(±1.02) EfficientNetV2-B1 85.63(±2.00) 85.68(±2.02) 85.66(±2.00) 85.61(±2.01) 91.02(±1.25) EfficientNetV2-B2 83.31(±2.59) 83.48(±2.63) 83.34(±2.59) 83.27(±2.58) 89.58(±1.62) \ EfficientNetV2-B3 85.78(±0.96) 85.97(±0.99) 85.84(±1.02) 85.77(±0.96) 91.11(±0.60) EfficientNetV2-S 87.41(±0.47) 87.50(±0.51) 87.54(±0.49) 87.42(±0.51) 92.15(±0.31) EfficientNetV2-M 88.65(±1.05) 88.68(±1.01) 88.73(±1.07) 88.64(±1.05) 92.88(±0.66) Stacking-ensemble 98.71(±0.26) 98.44(±1.05) 98.75(±0.26) 98.72(±0.26) 99.20(±0.16) ECA-EfficientNetV2 99.21(±0.30) 99.23(±0.30) 99.25(±0.30) 99.20(±0.31) 99.51(±0.19) p-value 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ ⁎⁎ p < 0.01 Figure 12 shows examples of confusion matrix from EfficientNetV2-B0, EfficientNetV2-M, Stacking-ensemble, and ECA-EfficientNetV2 models. In EfficientNetV2-B0, 58 Bacterial Pneumonia are misclassified to Tuberculosis, and 42 Tuberculosis are misclassified to Bacterial Pneumonia. In EfficientNetV2-M, 40 Bacterial Pneumonia are misclassified to Tuberculosis, and 51 Tuberculosis are misclassified to Bacterial Pneumonia. Obviously, Bacterial Pneumonia and Tuberculosis are misclassified to each other from EfficientNetV2-B0 and EfficientNetV2-M models, while the wrong phenomenon is much improved in Stacking-ensemble and ECA-EfficientNetV2 models.Figure 12 Confusion matrix of Dataset X-ray. (Color version of figure is available online.) Figure 12 Dataset CT The performances of 5 metrics among EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, EfficientNetV2-B3, EfficientNetV2-S, EfficientNetV2-M, the proposed stacking-ensemble model, and the proposed ECA-EfficientNetV2 models for dataset CT were compared in Table 12 . The accuracies from the previous eight models are greater than 93%; the highest two are from Stacking-ensemble and ECA-EfficientNetV2 models. Actually, the same results could be found in precision, recall, F1-score and AUC. The five metrics from Stacking-ensemble and ECA-EfficientNetV2 models are statistically significant from the six EfficientNetV2 models (p-value < 0.01). The best performance, the accuracy (99.81%), precision (99.80%), recall (99.80%), F1-score (99.81%), and AUC (99.87%), comes from the proposed ECA-EfficientNetV2, which are not significantly different from the Stacking-ensemble model. Same as we found in Dataset X-ray, the standard deviations of five metrics from ECA-EfficientNetV2 model are relatively small, which stands the stability of the model.Table 12 Performance Comparison on Dataset CT Table 12Models Dataset CT Accuracy (%) Precision (%) Recall (%) F1-Score (%) AUC (%) EfficientNetV2-B0 94.54(±0.84) 94.58(±0.83) 94.54(±0.83) 94.50(±0.85) 96.36(±0.56) EfficientNetV2-B1 93.60(±1.67) 93.66(±1.74) 93.67(±1.61) 93.55(±1.70) 95.73(±1.11) EfficientNetV2-B2 94.41(±1.48) 94.82(±2.04) 94.47(±1.57) 94.35(±1.49) 96.28(±0.99) EfficientNetV2-B3 95.10(±0.73) 95.15(±0.72) 95.15(±0.68) 95.12(±0.70) 96.73(±0.49) EfficientNetV2-S 96.34(±0.66) 96.35(±0.67) 96.35(±0.66) 96.34(±0.67) 97.56(±0.44) EfficientNetV2-M 95.66(±0.65) 95.67(±0.67) 95.58(±0.57) 95.65(±0.66) 97.11(±0.43) Stacking-ensemble 98.84(±0.51) 98.87(±0.50) 98.93(±0.54) 98.92(±0.61) 99.23(±0.34) ECA-EfficientNetV2 99.81(±0.27) 99.80(±0.27) 99.80(±0.28) 99.81(±0.27) 99.87(±0.18) p-value 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ 0.00⁎⁎ ⁎⁎ p < 0.01 Figure 13 shows examples of confusion matrix from EfficientNetV2-B0, EfficientNetV2-M, Stacking-ensemble, and ECA-EfficientNetV2 models for Dataset CT. In EfficientNetV2-B0, 23 COVID-19 are misclassified to Normal, and 8 Normal are misclassified to COVID-19. In EfficientNetV2-M, 18 COVID-19 are misclassified to Normal, and 12 Normal are misclassified to COVID-19. In Stacking-ensemble, 7 COVID-19 are misclassified to Normal, and 1 Normal are misclassified to COVID-19. In ECA-EfficientNetV2, 2 COVID-19 are misclassified to Normal, and 0 Normal are misclassified to COVID-19. Obviously, COVID-19 and Normal are misclassified to each other from EfficientNetV2-B0 and EfficientNetV2-M models, while Stacking-ensemble and ECA-EfficientNetV2 models greatly improved the wrong phenomenon.Figure 13 Confusion matrix of Dataset CT. (Color version of figure is available online.) Figure 13 Comparison With SOTA The results of the proposed Stacking-ensemble and ECA-EfficientNetV2 models are compared to the related studies using stacking ensemble models for COVID-19 diagnosis in Table 13 . Most of the related studies worked on classification of 2 to 3 groups, while this study focuses on classification of five groups in X-ray and classification of 4 groups in CT. Most of the accuracies are greater than 90% by using stacking models, which represents the favorable performance of stacking method.Table 13 Comparison of the Proposed Models With SOTA Table 13No. Study(s) Dataset Architecture Class Accuracy 1 Ieracitano et al. (2022) (5) 155 X-ray images. CovNNet 2 81.00% 2 Khan et al. (2021) (6) 6448 X-ray images. DHL DBHL 2 98.53% 3 Loey et al. (2022) (7) 10,848 X-ray images. CNN+ Bayesian 3 96.00% 4 Hu et al. (2021) (8) 10,848 X-ray images. CNN+ ELMs+ ChOA 2 99.11% 5 Thaseen et al. (2022) (9) 13,808 X-ray images. Ensemble CNN model 3 99.00% 6 Kumar et al. (2021) (10) 6000 X-ray images. HCNN 3 98.20% 7 Musallam et al. (2022) (11) 7512 X-ray images. DeepChest 3 96.56% 8 Mamalakis et al. (2021) (12) 6000 X-ray images. DenResCov-19 4 99.60% 9 Garg et al. (2022) (13) 4173 CT images. EfficientNet-B5 2 98.45% 10 Lahsaini et al. (2021) (14) 4986 CT images. DenseNet201+GradCam 2 98.80% 11 Rahimzadeh et al. (2021) (15) 63,849 CT images. ResNet50V2+ FPN 2 98.49% 12 Qi et al. (2022) (16) 10,000 CT images. UNet+ DenseNet121 2 97.10% 13 Abdel-Basset et al. (2021) (17) 9593 CT images. GR-UNet+ CNN 2 96.80% 14 Ye et al. (2022) (18) 45,167 CT images. CIFD-Net 2 91.19% 15 Balaha et al. (2021) (19) 15,535 CT images. CNN+HHO+FCS+CSS+WSM 2 99.33% 16 Ahamed et al. (2021) (21) 4593 X-ray images. ResNet50V2(fine-tuning) 4 96.45% 3000 CT images. 3 99.01% 17 Kumari et al. (2021) (22) 2000 X-ray images. VGG16 2 98.00% 2000 CT images. Xception 2 83.00% 18 Ahsan et al. (2020) (23) 400 X-ray images. NasNetMobile 2 100.00% 400 CT images. 2 95.20% 19 Jia et al. (2021) (24) 7592 X-ray images. ResNet(fine-tuning) 5 99.60% 104,009 CT images. 3 99.30% 20 Kassania et al. (2021) (25) 137 X-ray images. DenseNet121+ Bagging 2 99.00% 137 CT images. 21 Gour and Jain (2022) (26) 3040 X-ray images. Ensemble CNN 3 97.27% 4645 CT images. 2 98.30% 22 Kamil et al. (2021) (27) 977 X-ray images. VGG19(fine-tuning) 2 99.00% 23 CT images. 23 Saygılı (2021) (28) 1125 X-ray images. SVM 3 85.96% 3228 CT images. K-ELM 2 98.88% 24 Stacking-ensemble 6000 X-ray images. Stacking-ensemble 5 98.71% 4800 CT images. 4 98.84% 25 ECA-EfficientNetV2 6000 X-ray images. ECA-EfficientNetV2 5 99.21% 4800 CT images. 4 99.81% Although the accuracy from our Stacking-ensemble model is slightly inferior than (8,9,12,19,21,23,24,25,27,28). As we have mentioned before, most of those related studies worked on classification of two groups, while this study is working on the classification of four groups in X-ray and five groups in CT. Even in this situation, the accuracies of our proposed Stacking-ensemble model are pretty close to the previous studies. In addition, the performance of the proposed ECA-EfficientNetV2 model dominates most of the related studies. The accuracies for X-ray and CT from ECA-EfficientNetV2 model are close to 100%, which illustrates the great classification capability for multiple groups on chest diseases. To verify the performance of the two proposed Stacking-ensemble and ECA-EfficientNetV2 models, one more open public dataset COVID-CT (https://github.com/UCSD-AI4H/COVID-CT) was tested and the performance metrics were displayed in Table 14 . There are 746 CT images including two groups of COVID-19 and Non-COVID-19 in dataset COVID-CT. Except the accuracy of 93.33% from Shaik and Cherukuri (53), the accuracies from other related studies are lower than 90%. The accuracies are 94.86% and 95.29% from our proposed Stacking-ensemble and ECA-EfficientNetV2 models, respectively, which are higher than all of the related studies compared in Table 14. Especially, the number of parameter in ECA-EfficientNetV2 model is 5,706,965, which is much less than those in (48,50,51).Table 14 Comparisons on COVID-CT Dataset Table 14Dataset Study(s) Accuracy Precision Recall F1-Score Parameter COVID-CT Mishra et al. (2020) (46) 88.30% — — 86.70% — Saqib et al. (2020) (47) 80.30% 78.20% 85.70% 81.80% — He et al. (2020) (48) 86.00% — — 85.00% 14,149,480 Mobiny et al. (2020) (49) — 84.00% — — — Polsinelli et al. (2020) (50) 85.00% 85.00% 87.00% 86.00% 12,600,000 Yang et al. (2020) (51) — — 89.10% — 25,600,000 Cruz (2021) (52) 86.00% — 89.00% 85.00% — Shaik and Cherukuri (2021) (53) 93.33% 93.17% 93.54% 93.29% — Stacking-ensemble 94.86% 94.79% 94.83% 94.84% — ECA-EfficientNetV2 95.29% 95.15% 95.24% 95.27% 5706,965 CONCLUSION This study applies six pretrained MobileNetV2 models on COVID-19 diagnosis, and stacks ensemble the previous six models to achieve better classification results. The self-designed Stacking-ensemble and ECA-MobileNetV2 models were proposed to classify multiple chest diseases including COVID-19 in this study. Classification was executed on five groups using X-ray images and four groups using CT images. The experimental results of two proposed models were compared to six pretrained MobileNetV2 models. The proposed Stacking-ensemble achieves accuracies of 98.71% and 98.84% on X-ray and CT datasets, respectively. With stability, the proposed ECA-MobileNetV2 achieves the highest accuracies of 99.21% and 99.81% on X-ray and CT datasets, respectively. The proposed Stacking-ensemble model is the first study ensembles the series of EfficientNetV2 models on multiple chest diseases including COVID-19 detection, which reduce prediction variance to the training data and improves overall classification performance when compared to any individual EfficientNetV2 model. As compared to the proposed ECA-MobileNetV2 model, the stacking model is computationally expensive; it takes around 1.5 hours in this experiment while the execution time is only 17 minutes in ECA-MobileNetV2 model. The major contribution of the study is the proposed ECA-MobileNetV2 model combines the advantages of EfficientNetV2 and ECA-Net to demonstrate superior classification performance with least variance and training time. The results of this study imply that the architecture of the proposed ECA-MobileNetV2 model can be used to assist radiologists in the X-ray and CT scans on multiple chest diseases including COVID-19. The 6,000 X-ray and 4800 CT images used in this study were collected from five Kaggle datasets. More chest images are encouraged to achieve more robust classification performance. Since the related studies compared in the study were searched from Scopus and were limited, more comprehensive review is suggested in the future research. ==== Refs REFERENCES 1 C. Pneumonia, “COVID-19,” vol. 2019, 2019. 2 Liu H. Wang L. Nan Y. SDFN: Segmentation-based deep fusion network for thoracic disease classification in chest X-ray images Computerized Med Imaging and Graphics 75 2019 66 73 10.1016/j.compmedimag.2019.05.005 3 Lin L. Qin L. Xu Z. Artificial intelligence distinguishes COVID-19 from community acquired pneumonia on Chest CT Appl Intel 2019 2020 1 5 [Online]. Available: http://arxiv.org/abs/2003.13865 (accessed February 01, 2022) 4 Sharma N. Sharma R. Jindal N. Machine learning and deep learning applications-a vision Global Transitions Proceedings 2021 0 8 10.1016/j.gltp.2021.01.004 5 Ieracitano C. Mammone N. Versaci M. A fuzzy-enhanced deep learning approach for early detection of Covid-19 pneumonia from portable chest X-ray images Neurocomputing 481 2022 202 215 10.1016/j.neucom.2022.01.055 35079203 6 Khan S.H. Sohail A. Khan A. COVID-19 detection in chest X-ray images using deep boosted hybrid learning Comp Biol Med 137 2021 104816 10.1016/j.compbiomed.2021.104816 7 Loey M. El-Sappagh S. Mirjalili S. Bayesian-based optimized deep learning model to detect COVID-19 patients using chest X-ray image data Comp Biol Med 142 2022 105213 10.1016/j.compbiomed.2022.105213 8 Hu T. Khishe M. Mohammadi M. Realtime COVID-19 diagnosis from X-Ray images using deep CNN and extreme learning machines stabilized by chimp optimization algorithm Biomed Signal Proces Contrl 68 2021 102764 10.1016/j.bspc.2021.102764 9 Sakthivel R. Sumaiya Thaseen I. Vanithac M. An efficient hardware architecture based on an ensemble of deep learning models for COVID-19 prediction Sustainable Cities and Soc Feb. 2022 103713 10.1016/j.scs.2022.103713 10 Kumar M. Shakya D. Kurup V. COVID-19 prediction through X-ray images using transfer learning-based hybrid deep learning approach Materials Today: Proceedings Dec. 2021 10.1016/j.matpr.2021.12.123 11 Musallam A.S. Sherif A.S. Hussein M.K. Efficient framework for detecting COVID-19 and pneumonia from chest X-Ray using deep convolutional network Egyptian Info J 2021 2022 10.1016/j.eij.2022.01.002 12 Mamalakis M. Swift A.J. Vorselaars B. DenResCov-19: A deep transfer learning network for robust automatic classification of COVID-19, pneumonia, and tuberculosis from X-rays Computerized Med Imaging and Graphics 94 2021 102008 10.1016/j.compmedimag.2021.102008 13 Garg A. Salehi S. la Rocca M. Efficient and visualizable convolutional neural networks for COVID-19 classification using Chest CT Expert Systems with Appl 195 2022 116540 10.1016/j.eswa.2022.116540 14 LAHSAINI I. EL HABIB DAHO M. CHIKH M.A. Deep transfer learning based classification model for covid-19 using chest CT-scans Pattern Recog Letters 152 2021 122 128 10.1016/j.patrec.2021.08.035 15 Rahimzadeh M. Attar A. Sakhaei S.M. A fully automated deep learning-based network for detecting COVID-19 from a new and large lung CT scan dataset Biomed Signal Proces Contrl 68 2021 10.1016/j.bspc.2021.102588 16 Qi Q.Q. Qi S.L. Wu Y.N. Fully automatic pipeline of convolutional neural networks and capsule networks to distinguish COVID-19 from community-acquired pneumonia via CT images Comp in Biol Med 141 2022 10.1016/j.compbiomed.2021.105182 17 Abdel-Basset M. Hawash H. Moustafa N. Two-stage deep learning framework for discrimination between COVID-19 and community-acquired pneumonia from chest CT scans Pattern Recog Letters 152 2021 311 319 10.1016/j.patrec.2021.10.027 18 Ye Q.H. Gqu Y. Ding W.P. Robust weakly supervised learning for COVID-19 recognition using multi-center CT images Appl Soft Computing 116 2022 108291 10.1016/j.asoc.2021.108291 19 Balaha H.M. El-Gendy E.M. Saafan M.M. CovH2SD: a COVID-19 detection approach based on Harris Hawks optimization and stacked deep learning Expert Systems with Appl 186 2021 10.1016/j.eswa.2021.115805 20 Qi S.L. Xu C.W. Li C. DR-MIL: deep represented multiple instance learning distinguishes COVID-19 from community-acquired pneumonia in CT images Comp Methods and Programs in Biomed 211 2021 10.1016/j.cmpb.2021.106406 21 Ahamed K.U. Islam M. Uddin A. A deep learning approach using effective preprocessing techniques to detect COVID-19 from chest CT-scan and X-ray images Comp Biol Med 2021 105014 10.1016/j.compbiomed.2021.105014 22 Kumari S. Ranjith E. Gujjar A. Comparative analysis of deep learning models for COVID-19 detection Global Trans Proc 2 2 2021 559 565 10.1016/j.gltp.2021.08.030 23 M. M. Ahsan, K. D. Gupta, M. M. Islam, et al., “Study of different deep learning approach with explainable AI for screening patients with COVID-19 symptoms: using CT scan and chest X-ray image dataset,” 2020, doi:10.3390/make2040027. 24 Jia G. Lam H.K. Xu Y. Classification of COVID-19 chest X-Ray and CT images using a type of dynamic CNN modification method Computers in Biol and Med 134 April 2021 104425 10.1016/j.compbiomed.2021.104425 25 Kassania S.H. Kassanib P.H. Wesolowskic M.J. Automatic detection of coronavirus disease (COVID-19) in X-ray And CT images: a machine learning based approach Biocybernetics and Biomed Eng 41 3 2021 867 879 10.1016/j.bbe.2021.05.013 26 Gour M. Jain S. Automated COVID-19 detection from X-ray and CT images with stacked ensemble convolutional neural network Biocybernetics and Biomed Eng 42 1 2022 27 41 10.1016/j.bbe.2021.12.001 27 Kamil M.Y. A deep learning framework to detect Covid-19 disease via chest X-ray and CT scan images Int J Electrical and Comp Eng 11 1 2021 844 850 10.11591/ijece.v11i1.pp844-850 28 Saygılı A. A new approach for computer-aided detection of coronavirus (COVID-19) from CT and X-ray images using machine learning methods Appl Soft Computing 105 2021 107323 10.1016/j.asoc.2021.107323 29 M. Tan and Q. v. Le, “EfficientNetV2: smaller models and faster training,” 2021, [Online]. Available: http://arxiv.org/abs/2104.00298. (accessed December 01, 2021) 30 “COVID-19 Radiography Database | Kaggle.” https://www.kaggle.com/tawsifurrahman/covid19-radiography-database (accessed Nov. 04, 2021). 31 “Chest X-Ray Images (Pneumonia) | Kaggle.” https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia (accessed Nov. 04, 2021). 32 “Tuberculosis (TB) chest x-ray database | Kaggle.” https://www.kaggle.com/datasets/tawsifurrahman/tuberculosis-tb-chest-xray-dataset (accessed May 06, 2022). 33 “Large COVID-19 CT scan slice dataset | Kaggle.” https://www.kaggle.com/maedemaftouni/large-covid19-ct-slice-dataset (accessed Nov. 09, 2021). 34 “COVID-19&Normal&Pneumonia_CT_Images | Kaggle.” https://www.kaggle.com/datasets/anaselmasry/covid19normalpneumonia-ct-images (accessed May 08, 2022). 35 M. Tan and Q. V. Le, “EfficientNetV2: smaller models and faster training,” 2021. 36 J. Hu, L. Shen, S. Albanie, et al., “Squeeze-and-excitation networks,” 2017, [Online]. Available: http://arxiv.org/abs/1709.01507 37 Q. Wang, B. Wu, P. Zhu, et al., “ECA-Net: efficient channel attention for deep convolutional neural networks,” 2019, [Online]. Available: http://arxiv.org/abs/1910.03151(accessed February 01, 2022). 38 Opitz D. Maclin R. Popular ensemble methods: an empirical study J Artificial Intelligence Res 11 1999 169 198 10.1613/jair.614 39 M. A. Ganaie and M Hu. “Ensemble deep learning: A review,” arXiv preprint arXiv:2104.02395 (2021). 40 Sollich P. Krogh A. Learning with ensembles: how over-fitting can be useful Proceedings of the 8th International Conference on Neural Information Processing Systems 1995 190 196 Pages 41 Pourtaheri Z.K. Zahiri S.H. Ensemble classifiers with improved overfitting 1st Conference on Swarm Intelligence and Evolutionary Computation 2016 10.1109/CSIEC.2016.7482130 42 Rajaraman S. Kim I. Antani S.K. Detection and visualization of abnormality in chest radiographs using modality-specific convolutional neural network ensembles PeerJ 8 2020 e8693 10.7717/peerj.8693 32211231 43 Devnath L. Fan Z. Luo S. Detection and visualisation of pneumoconiosis using an ensemble of multi-dimensional deep features learned from Chest X-rays Int J Environ Res and Public Health 19 18 2022 11193 10.3390/ijerph191811193 36141457 44 Rajaraman S. Cemir S. Xue Z A novel stacked model ensemble for improved TB detection in chest radiographs Chapter 1, Medical Imaging 2019 CRC Press Boca Raton 45 Devnath L. Luo S. Summons P. Deep ensemble learning for the automatic detection of pneumoconiosis in coal worker's Chest X-ray radiography J Clin Med 11 18 2022 5342 10.3390/jcm11185342 36142989 46 Mishra A.K. Das S.K. Roy P. Identifying COVID19 from Chest CT images: a deep convolutional neural networks based approach J Healthcare Eng 2020 2020 10.1155/2020/8843664 47 M. Saqib, S. Anwar, A. Anwar, et al., “COVID19 detection from radiographs: is deep learning able to handle the crisis?,” no. June, pp. 1–14, 2020, [Online]. Available: www.preprints.org 48 He X. Yang X.Y. Zhang S.H. Sample-efficient deep learning for COVID-19 diagnosis based on CT scans IEEE Transactions on Med Imaging XX Xx 2020 10.1101/2020.04.13.20063941 49 A. Mobiny et al., “Radiologist-level COVID-19 detection using CT Scans with detail-oriented capsule networks,” 2020, [Online]. Available: http://arxiv.org/abs/2004.07407 50 Polsinelli M. Cinque L. Placidi G. A light CNN for detecting COVID-19 from CT scans of the chest Pattern Recog Letters 140 2020 95 100 10.1016/j.patrec.2020.10.001 51 X. Yang, X. He, J. Zhao, et al., “COVID-CT-Dataset: A CT scan dataset about COVID-19,” 2020, [Online]. Available: http://arxiv.org/abs/2003.13865. (accessed December 05, 2021). 52 Hernández Santa Cruz J.F. An ensemble approach for multi-stage transfer learning models for COVID-19 detection from chest CT scans Intelligence-Based Med 5 February 2021 100027 10.1016/j.ibmed.2021.100027 53 Shaik N.S. Cherukuri T.K. Transfer learning based novel ensemble classifier for COVID-19 detection from chest CT-scans Computers in Biol Med 141 June 2021 2021 105127 10.1016/j.compbiomed.2021.105127

0

PMC9748726

NO-CC CODE

2022-12-15 23:22:46

no

Infect Dis Health. 2022 Nov 14; 27:S14-S15

latin-1

Infect Dis Health

10.1016/j.idh.2022.09.051

oa_other

==== Front Disabil Health J Disabil Health J Disability and Health Journal 1936-6574 1876-7583 Elsevier S1936-6574(22)00170-4 10.1016/S1936-6574(22)00170-4 101412 Article Table of Contents 14 12 2022 1 2023 14 12 2022 16 1 101412101412 2020 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc

0

PMC9748728

NO-CC CODE

2022-12-15 23:22:46

no

Disabil Health J. 2023 Jan 14; 16(1):101412

utf-8

Disabil Health J

10.1016/S1936-6574(22)00170-4

oa_other

==== Front Disabil Health J Disabil Health J Disability and Health Journal 1936-6574 1876-7583 Elsevier S1936-6574(22)00170-4 10.1016/S1936-6574(22)00170-4 101412 Article Table of Contents 14 12 2022 1 2023 14 12 2022 16 1 101412101412 2020 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc

0

PMC9748729

NO-CC CODE

2022-12-15 23:22:46

no

Infect Dis Health. 2022 Nov 14; 27:S6

latin-1

Infect Dis Health

10.1016/j.idh.2022.09.023

oa_other

==== Front Disabil Health J Disabil Health J Disability and Health Journal 1936-6574 1876-7583 Elsevier S1936-6574(22)00170-4 10.1016/S1936-6574(22)00170-4 101412 Article Table of Contents 14 12 2022 1 2023 14 12 2022 16 1 101412101412 2020 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc

0

PMC9748730

NO-CC CODE

2022-12-15 23:22:46

no

Infect Dis Health. 2022 Nov 14; 27:S13

latin-1

Infect Dis Health

10.1016/j.idh.2022.09.046

oa_other

==== Front Disabil Health J Disabil Health J Disability and Health Journal 1936-6574 1876-7583 Elsevier S1936-6574(22)00170-4 10.1016/S1936-6574(22)00170-4 101412 Article Table of Contents 14 12 2022 1 2023 14 12 2022 16 1 101412101412 2020 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc

0

PMC9748731

NO-CC CODE

2022-12-15 23:22:46

no

Infect Dis Health. 2022 Nov 14; 27:S5

latin-1

Infect Dis Health

10.1016/j.idh.2022.09.019

oa_other

==== Front Atmospheric and Oceanic Science Letters 1674-2834 1674-2834 The Authors. Publishing Services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. S1674-2834(20)30015-5 10.1016/j.aosl.2020.100015 100015 Article Global air quality change during the COVID-19 pandemic: Regionally different ozone pollution responses COVID-19 疫情期间全球空气质量变化:臭氧响应的区域间差异 Tang Rong ab Huang Xin ab Zhou Derong ab Wang Haikun a Xu Jiawei ab Ding Aijun ab⁎ a School of Atmospheric Sciences, Nanjing University, Nanjing, China b Collaborative Innovation Center of Climate Change, Nanjing, Jiangsu Province, China ⁎ Corresponding author. 14 12 2020 7 2021 14 12 2020 14 4 100015100015 14 9 2020 21 10 2020 5 11 2020 © 2020 The Authors 2020 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The explosive spread of the 2019 novel coronavirus (COVID-19) provides a unique chance to rethink the relationship between human activity and air pollution. Though related studies have revealed substantial reductions in primary emissions, obvious differences do exist in the responses of secondary pollutants, like ozone (O3) pollution. However, the regional disparities of O3 responses and their causes have still not been fully investigated. To better elucidate the interrelationship between anthropogenic emissions, chemical production, and meteorological conditions, O3 responses caused by lockdowns over different regions were comprehensively explored at a global scale. Observational signals of air-quality change were derived from multi-year surface measurements and satellite retrievals. With similar substantial drops in nitrogen dioxide (NO2), ozone shows rising signals in most areas of both East Asia and Europe, even up to ∼14 ppb, while a non-negligible declining signal exists in North America, by about 2–4 ppb. Furthermore, the drivers behind the different O3 responses are discussed based on meteorological analysis and O3 sensitivity diagnosis. On the one hand, O3 responses to NO2 declines can be affected by the primary dependence on its precursors. On the other hand, it is also highly dependent on meteorological factors, especially temperature. Our study further highlights the great importance of taking into consideration both the regional disparities and synergistic effects of precursor reductions and meteorological influence for scientific mitigation of O3 pollution. 摘要 疫情期间全球各地一次排放大幅削减, 而臭氧等二次污染的响应则存在着区域间差异.结合地面和卫星观测发现, 同在氮氧化物大幅下降的情况下,臭氧在东亚和欧洲呈现出可达14ppb的上升信号, 而北美则下降为主 (约2–4ppb) .我们结合气象分析和臭氧敏感性进一步讨论了臭氧响应差异性的原因, 一方面受臭氧与前体物间关系的影响;另一方面来自于气象, 尤其是温度.研究明晰了人为排放,化学和气象三者的内在关联, 强调了在臭氧控制过程中考虑前体物削减和气象条件协同的重要性. Graphical abstract Image, graphical abstract Keywords Air quality COVID-19 lockdown Ozone Nitrogen dioxide Meteorology 关键词: 空气质量 COVID-19 臭氧 二氧化氮 气象条件 ==== Body pmc1 Introduction Tropospheric ozone (O3), one of the most important air pollutants, has drawn increasing concern owing to exacerbated O3 pollution and increased knowledge about its climatic impacts and health hazards (Atkinson et al., 2016). It can create serious pollution events of great concern, but is also an invisible killer (Haagen-Smit, 1952). Robust evidence exists of higher mortality in association with exposure to O3 (Jerrett et al., 2009). In addition, tropospheric O3 is also the third largest greenhouse gas, contributing about 3%–7% of the greenhouse effect, and has a substantial impact on climate change (Chang et al., 2009). Moreover, O3 exposure also causes oxidation damage to crops, resulting in huge economic and ecological losses and posing great threats to food security (Feng et al., 2015). As a typical secondary pollutant, surface O3 is mainly generated by volatile organic compounds (VOCs) and nitrogen oxides (NOx) through a series of photochemical reactions (Haagen-Smit, 1952; Wang et al., 2010). Its formation is driven by precursor emissions, chemical conversion, and weather with complicated mechanisms (Ding et al., 2008; Ding et al., 2013; Xue et al., 2014; Wang et al., 2017; Xu et al., 2018; Huang et al., 2020). The short-term daily changes in O3 are mainly affected by meteorological factors, while the long-term changes in O3 are affected by both climate and emissions (Gao et al., 2013; Zhu et al., 2015; Wang et al., 2019; Chen et al., 2020). Moreover, the nonlinear relationship between O3 production and its precursors further makes it challenging to understand the nature of O3 pollution and formulate specific control measures under different conditions (Sillman, 1999). With the explosive spread of the 2019 novel coronavirus (COVID-19), the epidemic rapidly deteriorated into a global pandemic and kept escalating. Governments around the world imposed unconventional containments on economic activities and even restrictions on population mobility (lockdown), one after another (Cohen and Kupferschmidt, 2020). A near-complete standstill of social and economic activities occupied many countries, leading to large reductions in emissions, especially those from traffic and the manufacturing sector (Huang et al., 2020; Venter et al., 2020). However, despite such unprecedented decreases in primary pollution (Menut et al., 2020), obvious regional disparities still exist in global air-quality changes, especially with respect to O3 anomalies (Sicard et al., 2020). In China, large decreases in NOx emissions from transportation increased O3 and nighttime NO3 radical formation, leading to the enhancement of atmospheric oxidizing capacity, which even made the secondary pollution offset the role of emissions reduction (Huang et al., 2020). On the contrary, O3 also shows declines together with nitrogen dioxide (NO2) in some regions like Canada and California (Adams, 2020; Connerton et al., 2020). Regionally different O3 pollution responses to NO2 declines during the global pandemic may exist, owing to differences in emission reduction measures, natural conditions, and nonlinear O3 chemistry. In this study, we integrated multi-year ground-based measurements of major pollutants, meteorological analysis, and satellite-retrieved atmospheric composition columns to explore the NO2 and O3 pollution responses during the COVID-19 lockdown. Furthermore, regional disparities in O3 responses and the possible contributions from meteorological factors and emission changes are discussed, combining statistical analysis and O3 sensitivity diagnosis (Duncan et al., 2010; Souri et al., 2020). The interplay between emissions, atmospheric chemistry and meteorological conditions is assessed, with a focus on O3. 2 Data and methods 2.1 Ground-based observations and satellite retrievals To analyze global air-quality changes during the COVID-19 lockdown, observational signals were extracted from surface air-quality observations and satellite-retrieved column amounts for three major regions, i.e., East Asia, Europe, and North America. The ground-based air-quality observations were obtained from the World Air Quality Index platform (https://aqicn.org/data-platform/covid19/; last access: 31 July 2020) and originated from government-grade sources, mainly environmental protection agencies worldwide. The median of daily measurements at several stations was adopted for each major city. Only data with more than 70% valid temporal coverage were collected for further analysis. Moreover, outliers in the data with z-scores exceeding an absolute value of 3 (within 3 standard deviations from the mean) were removed for quality control (Cousineau and Chartier, 2010). Besides, the average of observations during the same period from 2015 to 2019 was inferred as the climatic normal level, to analyze from a second dimension apart from the comparison between the pre-COVID and COVID-lock periods. To supplement surface observations, we also employed satellite data to illustrate the spatial pattern and temporal variation of air pollution. The Tropospheric Monitoring Instrument (TROPOMI) onboard the Copernicus Sentinel-5 Precursor satellite provides retrievals of tropospheric NO2 and formaldehyde (HCHO) column amounts. The Level-2 daily products, with an overpass time of around 1330 LST and a spatial resolution of 3.5 km latitude × 5.5 km longitude, were accessed (https://s5phub.copernicus.eu/dhus/#/home; last access: 31 July 2020). Measurements with a quality assurance value below 0.5 were omitted to reduce uncertainties resulting from processing errors, anomalously high signals, and sun glints. For NO2, the trajectory product was then regridded to a 0.25° × 0.25° spatial grid. 2.2 Meteorological analyses The temperature at a height of 2 m, relative humidity, and solar radiation reaching the surface were taken from the NCEP FNL dataset (spatial resolution: 1° × 1°), which is updated every 6 h. The FNL product has integrated abundant observation and satellite retrievals, and has been widely used in research on weather and climate (Huang et al., 2016; Tang et al., 2020). 2.3 Ozone chemistry sensitivity diagnosis The nonlinear chemical processes involved in O3 production have necessitated using proxy indicators to convey information about the primary dependence of O3 production on VOCs or NOx (Sillman, 1999; Souri et al., 2020). The space-based tropospheric column ratio of HCHO (a proxy for VOCs) to NO2 has been widely used in previous studies as an indicator to qualitatively separate NOx-limited and NOx-saturated O3 formation regimes (Duncan et al., 2010; Choi et al., 2012; Jin and Holloway, 2015). Following these studies, data points with HCHO/NO2 ratios below 1 are considered as an NOx-saturated regime, those with HCHO/NO2 ratios above 2 are considered as an NOx-limited regime, while those between 1 and 2 would fall into a transition regime. Indeed, Jin et al. (2017) also revealed that the column HCHO/NO2 ratios marking the boundary between the NOx-saturated and transitional regimes are 0.9 from 2006 to 2012 for all three regions, and the thresholds are higher in the cold season than the warm season. Since the COVID-lock period defined in our research is mainly in the cold season, the adopted critical values in this study are relatively reasonable for qualitative differentiation. The TROPOMI NO2 and HCHO products were also regridded to a 1° × 1° spatial grid, to calculate the ratio of HCHO/NO2 and then diagnose O3 sensitivity regimes. 3 Results 3.1 Obvious NO2 reduction caused by COVID-19 in the Northern Hemisphere To obtain an overview of air-quality changes caused by the COVID-19 pandemic in the Northern Hemisphere, we compared the monthly average ground-based observed NO2 concentration from January to March in 2020 with the climatic state, which is defined as the average of the same period in 2015–19 (Fig. S1). As shown, with the continuous expansion of the global pandemic, sharp decreases in NO2 concentrations occurred in East Asia, Europe, and North America, successively. As the earliest epicenter of the pandemic, China was the first country to shut down commercial activities, restrict travel, and announce home quarantine, immediately after the Chinese Spring Festival in late January (Wang et al., 2020), reflecting the dramatic negative NO2 anomalies in January and February (Fig. S1(a)). Then, European countries successively entered region-wide states of emergency and carried out measures with the ensuing outbreak in Europe, and NO2 concentration started to decrease by ∼10 ppb compared with the climatic state in February (Fig. S1(b)). As for North America, a large-scale stay-at-home order was finally implemented in late March. Prior to this, the NO2 concentration had already showed scattered signals of decline. Moreover, time series of daily average observations in 2020 and normal levels in the same period are compared, for the three regions, respectively (Fig. 1 ). For East Asia, the region-averaged NO2 concentration shows a sharp decline of ∼46% (∼8 ppb), remarkably coincident with the start of lockdown, and remained at a low level for nearly two months after the outbreak of growth in newly confirmed COVID-19 cases. Prior studies have also revealed a significant drop of up to ∼40% in the average NO2 column over all Chinese cities relative to the same period in 2019 (Bauwens et al., 2020). Unlike NO2, O3 sometimes shows an opposite rising signal, especially when NO2 drops sharply (Zhao et al., 2020). As for Europe and North America, the downward trend of NO2 and the upward trend of O3 can on the one hand be partly attributed to seasonal changes, in terms of radiation and temperature (Fiore et al., 2002), while on the other hand the average ground-level NO2 during the pandemic is ∼25% and ∼17% lower, respectively, than climatic normal levels even before the large-scale sudden emergence of the COVID-19 pandemic (Fig. 1(b-c)). Moreover, the global population-weighted concentration of ground-level NO2 was indicated to have declined by ∼60% (Venter et al., 2020).Fig. 1 Changes in air pollution indicated by comparison between observations in 2020 and normal years: (a) time series of the daily average of observations in 2020 and normal levels in the same period in East Asia; (b, c) as in (a) but for Europe and North America, respectively. The average of observations from 2015 to 2019 is considered as the normal level. (d–f) Changes in satellite-retrieved tropospheric NO2 column amounts between the COVID-lock and pre-COVID periods in 2020 for the three regions. Fig 1 Ozone in both East Asia and Europe during the pandemic is obviously higher than the climatic state, which is clearly opposite to the case for NO2. Still, O3 in North America shows an opposite signal (Fig. 1(c)): a decline from the climatic normal level in sync with NO2 since March when some local governments started to call on people to segregate at home. Therefore, the pre-COVID and COVID-lock periods are selected for the three regions respectively, according to the approximate time node of the stay-at-home order and the changes in newly confirmed COVID-19 cases (Fig. 1). In addition, declines by almost −18 ppb in satellite-retrieved tropospheric NO2 column concentrations between the COVID-lock and pre-COVID period in 2020 further proves the sharp reductions in air pollutant emissions caused by lockdown (Fig. 1(d–f)). 3.2 Ozone response to emissions reduction and its regional disparity To make a clearer comparison, regional statistics are counted based on ground-based observations for the four time periods, i.e. pre-COVID and COVID-lock in 2020 and 2015–19 respectively. In this case (Fig. 2 ), the average differences in 2015–19 between the same periods of pre-COVID and COVID-lock are considered as seasonal changes, while the differences between 2020 and the climatic state during the pre-COVID period are inferred as changes caused by meteorological factors and anthropogenic emission declines during the pandemic (Venter et al., 2020).Fig. 2 Ground-based station observations of NO2 and O3 in East Asia, Europe, and North America. The figure compares the average concentration level before the COVID-19 pandemic (2020-Pre), during the lockdown period (2020-Lock), and the five-year climatological level for 2015–19 in the same period. Panels (a) and (b) are for NO2 and O3, respectively. The dashed lines on 2020-Lock mark the expected value, defined as Norm-lock plus the difference between 2020-Pre and Norm-Pre. Fig 2 As illustrated in Fig. 2(a), regional average NO2 concentrations reach an abnormally low value during the COVID-lock period in 2020 for all three regions, declining from climatic normal levels by ∼7 ppb in East Asia, ∼5 ppb in Europe, and ∼3 ppb in North America. The values are much lower than the expected ones defined in Fig. 2, i.e., the magnitudes of decline are much larger than the difference between 2020 and the climatic state, indicating the existence of contributing factors other than interannual changes in emissions, especially in East Asia and Europe. With a declining NO2, the regional average O3 in Europe shows a significant increase (∼5 ppb) from the climatic level, and ∼4 ppb from the expected value, while there is a weak increase in East Asia and a moderate decline in North America. Apart from increased O3 caused by seasonal changes (Fiore et al., 2002), the significantly increased O3 in Europe and the moderately decreased value in North America are mainly caused by changes in anthropogenic emissions or meteorological conditions. Prior studies have also indicated that O3 concentrations have increased differently in urban areas throughout western Europe during lockdown (Collivignarelli et al., 2020; Menut et al., 2020), while declines or little change are apparent in American cities like New York and Los Angeles (Connerton et al., 2020). To elucidate the relationship between changes in NO2 and O3, ground-based observational changes are further analyzed for cities over the three regions, respectively (Fig. S2). Points of NO2 and O3 changes distribute evenly around zero before the pandemic in all three regions, while changes in NO2 centralize in the negative area and can even be up to −14 ppb during the COVID-lock period. As for O3 in East Asia and Europe, changes are more concentrated in the positive area than the negative area during the COVID-Lock period, showing an obvious upward signal, except for a few points in the negative area. In addition, linear fitting results reveal a significant inverse correlation between changes in NO2 and O3 in Europe and East Asia, which is even more significant during the COVID-lock period than the pre-COVID period (Fig. 2(a-b)). By contrast, the scatterplot for North America indicates moderate NO2 declines and an even more positive association between NO2 and O3 changes. It is further confirmed that significant disparities in the O3 response exist among these regions, though similarly with decreased anthropogenic emissions and NO2. Combined with the spatial distribution of NO2 and O3 observational changes during the COVID-19 lockdown (Fig. 3 ), signals of decreased NO2 are densely distributed in eastern China and central Europe, but scattered in western and northeastern America. As shown in Fig. 3(e), the O3 rising signals over Europe are highly consistent in space with the NO2 decreasing signals, which is almost perfectly in accordance with the negative correlation between NO2 and O3 (Sicard et al., 2020), while coincidences are shown in southern China. Therefore, we attempt to provide a comprehensive explanation from the perspective of meteorological drivers and chemical factors like O3 sensitivity.Fig. 3 Spatial distribution of changes in NO2 and O3 during the COVID-19 lockdown: (a, b) ground-based observational changes in NO2 and O3 for cities in East Asia during the COVID-19 lockdown in 2020 relative to the same period in 2019; (c, d) and (e, f) as in (a, b) but for Europe and North America, respectively. Fig 3 3.3 Meteorological and chemical drivers behind the different O3 responses Tropospheric O3 is primarily built by photochemical reactions under solar radiation, covering complex nonlinear photochemistry with multiple precursors including NOx, carbon monoxide, VOCs, and methane (Sillman, 1999). So, the primary dependence of O3 production on its precursors, i.e., O3 sensitivity, is an important factor for the O3 responses in the face of NOx reduction. Here, the ratio of HCHO, a proxy to VOCs, to NO2 columns has been applied to diagnose O3 sensitivity regimes (Duncan et al., 2010; Choi et al., 2012) (see Section 2.3 for details). The spatial distribution of the ratio of satellite-retrieved HCHO to NO2 during the COVID-19 lockdown in 2020 is shown in Fig. 4 . In East Asia, large areas with low HCHO/NO2 values less than 1 are considered as an NOx-saturated regime (Souri et al., 2020) in eastern and southern China, especially in dense urban areas. Most cities in China would fall into the NOx-saturated regime owing to high NOx emissions, which means the O3 is expected to increase with decreased NOx (Wang et al., 2019). The increased O3 in eastern China can be explained to a certain extent; nevertheless, the observed O3 anomalies in southern coastal areas still indicate the involvement of other factors like meteorological influences. As for Europe, the low ratio values over the northern cities show diagnoses of NOx-saturated regimes, fitting well with the observed O3 anomaly, together with the high values over the Iberian Peninsula. As for North America, except for metropolitan areas, most areas are inclined toward NOx-limited regimes, with ratios even above 5, consistent with the transition from an NOx-saturated regime to an NOx-limited regime in North America, as revealed in prior studies (Choi et al., 2012; He et al., 2020). In addition, since the lockdown in North America almost fell in late spring, increased highly active VOC emissions from natural sources and the accelerated oxidation rate by the rising temperature would make O3 sensitivity tend toward a transitional regime or even an NOx-limited regime (Jin and Holloway, 2015; Xu et al., 2021).Fig. 4 Ratio of satellite-retrieved HCHO to NO2 during the COVID-19 lockdown: (a) spatial distribution of the ratio of HCHO to NO2 in East Asia during the COVID-19 lockdown; (b, c) as in (a) but for Europe and North America, respectively. Fig 4 Besides, surface O3 production is also strongly affected by meteorological factors like solar radiation fluxes, air temperature, and humidity (Ding et al., 2008; Pope et al., 2016; Lu et al., 2019). Fractional changes in meteorological conditions between the COVID-lock period in 2020 and the same period in 2015–19 are shown in Fig. 5 . Air temperature is one of the most important drivers of tropospheric O3 production, which can enhance the rate of photochemical production and promote its accumulation (Swackhamer 1991). During the COVID-lock period, temperatures were around 2 °C warmer than the climatic state in eastern China and almost throughout Europe, contributing to O3 formation (Fig. 5(a-b)). However, the effect of rising temperature can be partly offset by declines in solar radiation reaching the surface and increased relative humidity, which is mainly reflected in southern China and southern Europe (Fig. 5(d–i)), consistent with the barely changed O3 in the Pearl River Delta and Iberian Peninsula regions (Fig. 3). For North America, due to substantially decreased temperatures and increased relative humidity, decreased O3 shows in the southwestern coastal cities like California (Connerton et al., 2020). Though no significant changes in solar radiation exist, large increases of humidity by ∼40% in the eastern coastal region of America reflects the sporadic signals of decreased O3 (Fig. 5(f-i)). It was found that although O3 responses to NO2 declines can be partly affected by chemical sensitivity, it is mainly dominated by meteorological factors, especially temperature and radiation. The drivers behind the O3 responses further emphasize the importance of taking into consideration both the synergistic effects of precursor reductions and meteorological influences for refined O3 pollution control over different regions.Fig. 5 Fractional changes in meteorological conditions between the COVID-19 lockdown period in 2020 and the same period in 2015–19: (a–c) spatial distribution of changes (unit:°C) in temperature at 2 m during the COVID-19 lockdown in East Asia, Europe and North America, respectively; (d–f) and (g–i) as in (a–c) but for the fractional changes (unit:%) in downward shortwave radiation at the surface and relative humidity, respectively. Note that the crosses mark the area with significant changes (t-test, α = 0.05). Fig 5 4 Conclusion The outbreak of COVID-19 raised a question about the relationship between anthropogenic emissions and air pollution, which has aroused heated discussion. Though research on air-quality changes caused by the lockdowns in different areas shows similar substantial reductions in primary emissions, obvious differences exist in the responses of secondary pollutants like O3. To better elucidate the reasons behind the regional differences in O3 responses and the interplay between emissions, atmospheric chemistry, and meteorological conditions, global air-quality changes caused by COVID-19 lockdowns and regional specific O3 responses to sharp NO2 declines were explored. Observational signals of air-quality change were extracted from multi-year ground-based measurements of major pollutants and satellite-retrieved atmospheric composition columns. Ozone shows rising signals in most areas of both East Asia and Europe, while a non-negligible declining signal exists in North America, in the face of NO2 reductions over the three regions, indicating significant differences in relations between NO2 and O3 changes. Furthermore, meteorological and atmospheric chemical drivers behind the different O3 responses were discussed based on analysis data and proxy indicators (HCHO/NO2) for O3 sensitivity diagnosis. Ozone responses to NO2 declines can be affected by the primary dependence on its precursors to a certain extent, and the O3 response in Europe fits particularly well with the O3 sensitivity regimes. Meanwhile, meteorological factors are a rather important driver of the O3 responses, especially air temperature and solar radiation. Apart from weakened titration effects caused by NO declines, increased O3 in East Asia and Europe can be largely dominated by the climatologically warmer temperatures during the lockdowns in 2020. However, the contribution of rising temperature can be partly offset by a decline in solar radiation reaching the surface and an increase in relative humidity in southern China and southern Europe. For North America, declines in temperature and substantial increases in humidity can be important contributors to the decreased O3 over the western coasts. This study investigated the impact of meteorological conditions and chemical sensitivity under emission changes, which further emphasizes the great importance of taking into consideration the regional disparities and synergistic effects of precursor reductions and meteorological influences for scientific mitigation of O3 pollution. Appendix Supplementary materials Image, application 1 Funding This research was supported by the 10.13039/501100001809 National Natural Science Foundation of China [grant numbers 91744311 and 41922038] and the International Cooperation project of Jiangsu Provincial Science and Technology Agency [grant number BZ2017066]. Acknowledgments The air-quality observations provided by the World Air Quality Index Project Team are archived at https://aqicn.org/data-platform/covid19/. The global meteorological analysis data are available at https://rda.ucar.edu/datasets/ds083.2/, and the Sentiel-5P satellite retrievals can be obtained from https://s5phub.copernicus.eu/dhus/#/home. Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.aosl.2020.100015. ==== Refs References Adams M.D. Air pollution in Ontario, Canada during the COVID-19 state of emergency Sci. Total Environ. 742 2020 140516 10.1016/j.scitotenv.2020.140516 Atkinson R.W. Butland B.K. Dimitroulopoulou C. Heal M.R. Stedman J.R. Carslaw N. Jarvis D. Long-term exposure to ambient ozone and mortality: a quantitative systematic review and meta-analysis of evidence from cohort studies BMJ Open 6 2 2016 e009493 10.1136/bmjopen-2015-009493 Bauwens M. Compernolle S. Stavrakou T. Müller J.-F Gent J. Eskes H. Levelt P.F. Impact of coronavirus outbreak on NO 2 pollution assessed using TROPOMI and OMI observations Geophys. Res. Lett. 47 11 2020 10.1029/2020 gl087978 Chang W.Y. Liao H. Wang H.J. Climate responses to direct radiative forcing of anthropogenic aerosols, tropospheric ozone, and long-lived greenhouse gases in eastern China over 1951–2000 Adv. Atmos. Sci. 26 4 2009 748 762 10.1007/s00376-009-9032-4 Chen Z. Li R. Chen D. Zhuang Y. Gao B. Yang L. Li M. Understanding the causal influence of major meteorological factors on ground ozone concentrations across China J. Clean. Prod. 242 2020 118498 10.1016/j.jclepro.2019.118498 Choi Y. Kim H. Tong D. Lee P. Summertime weekly cycles of observed and modeled NOx and O3 concentrations as a function of satellite-derived ozone production sensitivity and land use types over the Continental United States Atmos. Chem. Phys. 12 14 2012 6291 6307 10.5194/acp-12-6291-2012 Cohen J. Kupferschmidt K. Strategies shift as coronavirus pandemic looms Science 367 6481 2020 962 963 10.1126/science.367.6481.962 32108093 Collivignarelli M.C. Abbà A. Bertanza G. Pedrazzani R. Ricciardi P. Miino M.C. Lockdown for CoViD-2019 in Milan: what are the effects on air quality? Sci. Total Environ. 732 2020 139280 10.1016/j.scitotenv.2020.139280 Connerton P. Assunção J.V.De Miranda R.M.De Slovic A.D. Pérez-Martínez P.J. Ribeiro H. Air quality during COVID-19 in four megacities: lessons and challenges for public health Int. J. Environ. Res. Public Health 17 14 2020 5067 10.3390/ijerph17145067 32674410 Cousineau D. Chartier S. Outliers detection and treatment: a review Int. J. Psychol. Res. Medellin 3 1 2010 58 67 10.21500/20112084.844 Ding A.J. Fu C.B. Yang X.Q. Sun J.N. Zheng L.F. Xie Y.N. Herrmann E. Ozone and fine particle in the western Yangtze River Delta: an overview of 1yr data at the SORPES station Atmos. Chem. Phys. 13 11 2013 5813 5830 10.5194/acp-13-5813-2013 Ding A.J. Wang T. Thouret V. Cammas J.P. N´ed´elec P. Tropospheric ozone climatology over Beijing: analysis of aircraft data from the MOZAIC program Atmos. Chem. Phys. 8 2008 1 13 10.5194/acp-8-1-2008 Duncan B.N. Yoshida Y. Olson J.R. Sillman S. Martin R.V. Lamsal L. Hu Y. Application of OMI observations to a space-based indicator of NOx and VOC controls on surface ozone formation Atmos. Environ. 44 18 2010 2213 2223 10.1016/j.atmosenv.2010.03.010 Feng Z. Hu E. Wang X. Jiang L. Liu X. Ground-level O3 pollution and its impacts on food crops in China: a review Environ. Pollut. 199 2015 42 48 10.1016/j.envpol.2015.01.016 25618365 Fiore, A.M., Jacob, D.J., Bey, I., Yantosca, R.M., Field, B.D., Fusco. A.C., 2002. Background ozone over the United States in summer: origin, trend, and contribution to pollution episodes. J. Geophys. Res.: Atmos., 107(D15).107 (D15). doi:10.1029/2001jd000982. Gao Y. Fu J.S. Drake J.B. Lamarque J.F. Liu Y. The impact of emission and climate change on ozone in the United States under representative concentration pathways (RCPs) Atmos. Chem. Phys. 13 18 2013 9607 9621 10.5194/acp-13-9607-2013 34135946 Haagen-Smit A.J. Chemistry and physiology of Los Angeles smog Ind. Eng. Chem. 44 6 1952 1342 1346 10.1021/ie50510a045 He H. Liang X.-Z. Sun C. Tao Z. Tong D.Q. The long-term trend and production sensitivity change in the US ozone pollution from observations and model simulations Atmos. Chem. Phys. 20 5 2020 3191 3208 10.5194/acp-20-3191-2020 Huang X. Ding A.J. Liu L.X. Liu Q. Ding K. Niu X.R. Nie W. Effects of aerosol–radiation interaction on precipitation during biomass-burning season in East China Atmos. Chem. Phys. 16 15 2016 10063 10082 10.5194/acp-16-10063-2016 Huang X. Ding A.J. Gao J. Zheng B. Zhou D.R. Qi X.M. Tang R. Enhanced secondary pollution offset reduction of primary emissions during COVID-19 lockdown in China Natl. Sci. Rev. 2020 10.1093/nsr/nwaa137 Jerrett M. Burnett R.T. Arden Pope C. Ito K. Thurston G. Krewski D. Shi Y. Calle E. Thun M. Long-term ozone exposure and mortality N. Engl. J. Med. 360 11 2009 1085 1095 10.1056/nejmoa0803894 19279340 Jin X. Fiore A.M. Murray L.T. Valin L.C. Lamsal L.N. Duncan B. Folkert Boersma K. Evaluating a space-based indicator of surface ozone-NOx-VOC sensitivity over midlatitude source regions and application to decadal trends J. Geophys. Res. Atmos. 122 19 2017 439 461 10.1002/2017jd026720 Jin X. Holloway T. Spatial and temporal variability of ozone sensitivity over China observed from the ozone monitoring instrument J. Geophys. Res. Atmos. 120 14 2015 7229 7246 10.1002/2015jd023250 Lu X. Zhang L. Shen L. Meteorology and climate influences on tropospheric ozone: a review of natural sources, chemistry, and transport patterns Curr. Pollut. Rep. 5 4 2019 238 260 10.1007/s40726-019-00118-3 Menut L. Bessagnet B. Siour G. Mailler S. Pennel R. Cholakian A. Impact of lockdown measures to combat Covid-19 on air quality over western Europe Sci. Total Environ. 741 2020 140426 10.1016/j.scitotenv.2020.140426 Pope R. Butt E. Chipperfield M. Doherty R. Fenech S. Schmidt A. Arnold S. Savage N. The impact of synoptic weather on UK surface ozone and implications for premature mortality Environ. Res. Lett. 11 2016 124004 10.1088/1748-9326/11/12/124004 Sicard P. Marco A.De Agathokleous E. Feng Z. Xu X. Paoletti E. Rodriguez J.J.D. Calatayud V. Amplified ozone pollution in cities during the COVID-19 lockdown Sci. Total Environ. 735 2020 139542 10.1016/j.scitotenv.2020.139542 Sillman S. The relation between ozone, NOx and hydrocarbons in urban and polluted rural environments Atmos. Environ. 33 12 1999 1821 1845 10.1016/s1352-2310(98)00345-8 Souri A.H. Nowlan C.R. Wolfe G.M. Lamsal L.N. Miller C.E.C. Abad G.G. Janz S.J. Revisiting the effectiveness of HCHO/NO2 ratios for inferring ozone sensitivity to its precursors using high resolution airborne remote sensing observations in a high ozone episode during the KORUS-AQ campaign Atmos. Environ. 224 2020 117341 10.1016/j.atmosenv.2020.117341 Swackhamer D.L. Rethinking the ozone problem in urban and regional air pollution J. Aerosol Sci. 24 7 1991 977 978 10.1016/0021-8502(93)90076-L Tang R. Huang X. Zhou D. Ding A. Biomass-burning-induced surface darkening and its impact on regional meteorology in eastern China Atmos. Chem. Phys. 20 10 2020 6177 6191 10.5194/acp-20-6177-2020 Venter Z.S. Aunan K. Chowdhury S. Lelieveld J. COVID-19 lockdowns cause global air pollution declines Proc. Natl. Acad. Sci. 2020 10.1073/pnas.2006853117 202006853 Wang C. Horby P.W. Hayden F.G. Gao G.F. A novel coronavirus outbreak of global health concern Lancet 395 10223 2020 470 473 10.1016/s0140-6736(20)30185-9 31986257 Wang N. Lyu X. Deng X. Huang X. Jiang F. Ding A. Aggravating O3 pollution due to NOx emission control in eastern China Sci. Total Environ. 677 2019 732 744 10.1016/j.scitotenv.2019.04.388 31075619 Wang T. Nie W. Gao J. Xue L.K. Gao X.M. Wang X.F. Qiu J. Air quality during the 2008 Beijing Olympics: secondary pollutants and regional impact Atmos. Chem. Phys. 10 16 2010 7603 7615 10.5194/acp-10-7603-2010 Wang T. Xue L. Brimblecombe P. Lam Y.F. Li L. Zhang L. Ozone pollution in China: a review of concentrations, meteorological influences, chemical precursors, and effects Sci. Total Environ. 575 2017 1582 1596 10.1016/j.scitotenv.2016.10.081 27789078 Xu J. Huang X. Wang N. Li Y. Ding A. Understanding ozone pollution in the Yangtze River Delta of eastern China from the perspective of diurnal cycles Sci. Total Environ. 2020 752 10.1016/j.scitotenv.2020.141928 Xu Z. Huang X. Nie W. Shen Y. Zheng L. Xie Y. Wang T. Impact of biomass burning and vertical mixing of residual-layer aged plumes on ozone in the Yangtze River Delta, China: a tethered-balloon measurement and modeling study of a multiday ozone episode J. Geophys. Res. Atmos. 123 20 2018 786 803 10.1029/2018jd028994 Xue L.K. Wang T. Gao J. Ding A.J. Zhou X.H. Blake D.R. Wang X.F. Ground-level ozone in four Chinese cities: precursors, regional transport and heterogeneous processes Atmos. Chem. Phys. 14 23 2014 13175 13188 10.5194/acp-14-13175-2014 Zhao Y. Zhang K. Xu X. Shen H. Zhu X. Zhang Y. Hu Y. Shen G. Substantial changes in nitrogen dioxide and ozone after excluding meteorological impacts during the COVID-19 outbreak in mainland China Environ. Sci. Technol. Lett. 7 6 2020 402 408 10.1021/acs.estlett.0c00304 Zhu B. Kang H. Zhu T. Su J. Hou X. Gao J. Impact of Shanghai urban land surface forcing on downstream city ozone chemistry J. Geophys. Res. Atmos. 120 9 2015 4340 4351 10.1002/2014jd022859

0

PMC9748732

NO-CC CODE

2022-12-15 23:22:46

no

2021 Jul 14; 14(4):100015

utf-8

oa_other

==== Front Atmospheric and Oceanic Science Letters 1674-2834 1674-2834 The Authors. Publishing Services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. S1674-2834(21)00032-5 10.1016/j.aosl.2021.100060 100060 Article Changes in air pollutants during the COVID-19 lockdown in Beijing: Insights from a machine-learning technique and implications for future control policy Hu Jiabao ab Pan Yuepeng bc⁎ He Yuexin bc Chi Xiyuan d Zhang Qianqian e Song Tao b Shen Weishou a⁎ a Collaborative Innovation Centre of Atmospheric Environment and Equipment Technology, Jiangsu Key Laboratory of Atmospheric Environment Monitoring and Pollution Control, School of Environmental Science andEngineering, Nanjing University of Information Science & Technology, Nanjing, China b State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry (LAPC), Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China c University of Chinese Academy of Sciences, Beijing, China d National Meteorological Center, China Meteorological Administration, Beijing, China e National Satellite Meteorological Center, China Meteorological Administration, Beijing, China ⁎ Corresponding authors. 30 4 2021 7 2021 30 4 2021 14 4 100060100060 28 1 2021 26 3 2021 23 4 2021 © 2021 The Authors. Publishing Services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The COVID-19 lockdowns led to abrupt reductions in human-related emissions worldwide and had an unintended impact on air quality improvement. However, quantifying this impact is difficult as meteorological conditions may mask the real effect of changes in emissions on the observed concentrations of pollutants. Based on the air quality and meteorological data at 35 sites in Beijing from 2015 to 2020, a machine learning technique was applied to decouple the impacts of meteorology and emissions on the concentrations of air pollutants. The results showed that the real (“deweathered”) concentrations of air pollutants (expect for O3) dropped significantly due to lockdown measures. Compared with the scenario without lockdowns (predicted concentrations), the observed values of PM2.5, PM10, SO2, NO2, and CO during lockdowns decreased by 39.4%, 50.1%, 51.8%, 43.1%, and 35.1%, respectively. In addition, a significant decline for NO2 and CO was found at the background sites (51% and 37.8%) rather than the traffic sites (37.1% and 35.5%), which is different from the common belief. While the primary emissions reduced during the lockdown period, episodic haze events still occurred due to unfavorable meteorological conditions. Thus, developing an optimized strategy to tackle air pollution in Beijing is essential in the future. 摘要 基于2015–2020年北京35个环境空气站和20个气象站观测资料, 应用机器学习方法 (随机森林算法) 分离了气象条件和源排放对大气污染物浓度的影响. 结果发现, 为应对疫情采取的隔离措施使北京2020年春节期间大气污染物浓度降低了35.1%–51.8%; 其中, 背景站氮氧化物和一氧化碳浓度的降幅最大, 超过了以往报道较多的交通站点. 同时, 2020年春节期间的气象条件不利于污染物扩散, 导致多次霾污染事件发生.为进一步改善北京空气质量, 未来需要优化减排策略. Graphical abstract Image, graphical abstract Keywords Random forest model Air pollutants Meteorological normalization COVID-19 Emission control strategy 关键词: 机器学习  大气污染  去气象化  COVID-19  减排策略  ==== Body pmc1 Introduction The coronavirus disease (COVID-19) broke out abruptly in December 2019 and spread rapidly across the world. In response to the COVID-19 crisis, most governments around the world introduced restrictions on behavior or lockdown measures. As a consequence, there has been a sharp slowdown in global economic growth and human-activity-related pollutant emissions. During the lockdown period, for instance, emissions of CO2 declined by 6.9%, 12.1%, and 9.5% in China, Europe, and the U.S., respectively (Liu et al., 2020). In addition, the concentrations of air pollutants also decreased greatly in most cities in the world, especially NO2, with a reduction by 20%, 43.5%, 50%, 51%, 52.7%, and 54.3% in Baghdad (Hashim et al., 2021), Rome (Kumari and Toshniwal, 2020), Barcelona (Baldasano, 2020), New York (Zangari et al., 2020), Delhi (Mahato et al., 2020), and São Paulo (Nakada and Urban, 2020), respectively. However, these ratios were mostly obtained via simple statistical analysis that compared concentrations of air pollutants before and after the lockdowns, or during the lockdowns, with the same periods in previous years (Shi et al., 2021). This common statistical approach comes with a major caveat that meteorological conditions may have masked the real effect of changes in emissions on the concentrations of target air pollutants (Zhang et al., 2020; Shi et al., 2021). Thus, there are difficulties in using such a method to explain the observed haze events during the lockdowns in some cities (Huang et al., 2020). While atmospheric chemistry and transport models can decouple the effect of emission changes from meteorology (Vu et al., 2019), this method still suffers from the lack of a timely emissions inventory to reflect the real-world changes (Wang et al., 2020). As an alternative approach, machine learning offers a reliable way to quantify changes in air quality due to emissions and meteorological factors (Shi et al., 2021). Indeed, a random forest (RF) method was successfully used to assess the short-term changes in selected cities in China due to the COVID-19 lockdowns (Wang et al., 2020). However, this study did not decouple the impacts of meteorology on the observed concentrations. Since the measured concentrations are determined by both emissions and meteorology, it is essential to decouple the effects of meteorology to understand the link between emissions and interventions (Zhang et al., 2020). Here, we employed a novel machine learning technique based on an RF algorithm to evaluate the impacts of the COVID-19 lockdowns in early 2020 on air quality in the megacity of Beijing. The selected city covers a range of air pollution levels, from highly to less polluted regimes. The 35 monitoring sites were divided into urban, suburban, background, and transport (road traffic) sites to better understand the impacts of various emissions on air quality changes. This study aims to (1) document the concentrations of air pollutants in early 2020 and the same periods in the past five years, (2) decouple the effects of meteorology from short-term emission changes on air quality, and (3) predict the concentrations of air pollutants without lockdown measures. Our findings can be used to elucidate the changes in air pollutants at different site types during the pandemic and provide guidance for policymakers in designing mitigation strategies towards improving air quality in future. 2 Data and methods 2.1 Data sources The major air pollutants of particulate matter (PM2.5 and PM10), sulfur dioxide (SO2), nitrogen dioxide (NO2), ozone (O3), and carbon monoxide (CO) at 35 stations in Beijing were used in this study (Fig. 1 , Table S1). The data were downloaded from the Beijing Municipal Ecological and Environmental Monitoring Center (http://www.bjmemc.com.cn/). In this study, hourly mean concentrations were used for the air pollutants, except O3, for which the maximum daily 8-h average concentrations were calculated for each period.Fig. 1 Monitoring sites for air quality and meteorology in Beijing. Fig 1 The hourly meteorological parameters including temperature, relative humidity, wind speed and wind direction, and atmospheric pressure, were obtained from the National Meteorological Information Center (http://data.cma.cn). In total, there are 20 meteorological stations in Beijing (Fig. 1, Table S2). Note that the sites for monitoring the meteorology were not always collocated with those for air quality observations. To address this mismatch, the nearest meteorological site for each air quality site was assigned based on the principle of proximity. The distance between the air quality and meteorological sites was calculated based on the Geosphere R package (Table S3), the corresponding matched results of which are shown in Table S1. 2.2 Definition of the study period In this study, we divided the whole period into four stages as follows: P1 (1–19 January 2020) was the early stage before the outbreak of COVID-19; P2 (24–30 January 2020) was the Spring Festival when the First-Level Public Health Emergency Response was started; P3 (31 January to 9 February 2020) was the prolonged holiday, which coincided with strictest lockdown measures; and P4 (10 February to 30 April 2020) saw closed-off management in the communities implemented and industries gradually restored. 2.3 RF models RF is a machine learning algorithm that uses multiple trees to train and predict samples (Breiman, 2001). Two basic scenarios were performed following previous study (He et al., 2021). First, meteorological normalization was used to exclude the impact of meteorological conditions on the concentrations of air pollutants (Grange et al., 2018). And second, a prediction experiment was used to predict concentrations of air pollutants under the scenario without lockdown (Wang et al., 2020). The configurations of these scenarios were as follows: 2.3.1 Meteorological normalization experiment We established 35 models (RF1) based on each air quality site to achieve meteorological normalization. In RF1, hourly concentrations of air pollutants were the dependent variables; and the meteorological parameters, time predictors, and regional transport parameters served as the independent predictors (Table S4). RF1 was trained on datasets from 1 January to 30 April 2015–20. The training set accounted for 70% of the data, with the remaining 30% used as the testing set. To remove the short-term meteorological impacts, meteorological datasets were randomly selected and replaced 1000 times from 2015 to 2020, rather than early 2020, to generate a new meteorological dataset. Specifically, meteorological parameters at a selected hour of a particular day in the 2020 meteorological data were replaced randomly with meteorological data from 2015 to 2020 within 2 weeks before and after the selected data point (Vu et al., 2019; Zhang et al., 2020). For example, the new input meteorological data at 0600 LST 15 January 2020 were randomly selected from the observed data at 0600 LST on any date from 1 to 29 January of any year in 2015–20. The above 1000 pieces of data were fed into the RF1 model to predict the concentrations of air pollutants. The 1000 predicted concentrations were then aggregated using the arithmetic mean to calculate the final meteorological normalized concentration (referred to as the “deweathered” concentration). Relative changes (R dew) between the observed concentrations (C observed) and deweathered concentrations (C deweathered) for each pollutant were calculated using the following equation:(1) Rdew=Cobserved−CdeweatheredCdeweathered×100%. 2.3.2 Prediction experiment We further established 24 (4 site types × 6 pollutants) models (RF2) to predict hourly concentrations of air pollutants without lockdown. In RF2, we used the same dependent variables and independent predictors as in RF1. The training set accounted for 70% of the data, and the remaining 30% was the testing set. The RF2 model was trained on datasets from 1 January to 30 April 2015–19 (note that the dataset in 2020 was not used in this stage). Then, the RF2 model was used to predict the concentrations of air pollutants during the COVID-19 outbreak from 1 January to 30 April 2020. Relative changes (R pre) between the observed concentrations (C observed) and predicted concentrations (C predicted) for each pollutant were calculated using the following equation:(2) Rpre=Cobserved−CpredictedCpredicted×100%. In general, the RF models performed well for most sites. A detailed model evaluation, including the coefficients of determination (R 2), the fraction of predictions within a factor of 2 (FAC2), mean bias, normalized mean bias, root-mean-square error (RMSE), and Pearson correlation coefficient (PCC), is displayed in Tables S5 and S6. In brief, the model performance indicated a low bias, low RMSE, and high PCC. For RF1, R 2 values ranged from 0.77 to 0.86 and FAC2 was greater than 0.76. For RF2, R 2 values ranged from 0.57 to 0.87 and FAC2 was greater than 0.71. 3 Results 3.1 Changes in observed concentrations of air pollutants We first analyzed the annual variation of air pollutants from 2015 to 2020, selecting the period from 1 January to 30 April as a whole (Fig. 2 ). With the exception of O3, the mean concentrations of PM2.5, PM10, SO2, NO2, and CO during January–April gradually decreased from 2015 to 2020, with a reduction of 9.2%, 10.4%, 16.4%, 8.8%, and 9.8% per year, respectively. This declining trend is indicative of the remarkable achievements with respect to the implementation of clean air actions in recent years (Wang et al., 2019; Huang et al., 2020).Fig. 2 Interannual variations of air pollutants from 1 January to 30 April 2015–20. Fig 2 In addition, we calculated the overall mean concentrations of each pollutant for the period 2015–19 and compared them with those in 2020 (Fig. 3 ). As shown, the concentrations of most air pollutants (except O3) were relatively low in 2020, particularly those of SO2 and NO2. In summary, the mean concentrations of PM2.5, PM10, SO2, NO2, and CO in 2020 were 31.5%, 41.0%, 69.2%, 36.8%, and 34.6% lower than the overall mean concentrations of 2015–19, respectively. The substantial decrease in air pollutants in 2020 was likely due to the reduction in anthropogenic emissions after the outbreak of COVID-19 (Pei et al., 2020), although some pollution episodes still occurred unexpectedly (discussed below).Fig. 3 Temporal variations of air pollutants in Beijing. The light gray area indicates 2015–19 variations in concentrations, with their average given by the thick gray line. The color line in each subfigure shows the hourly concentration of each pollutant in 2020. The color shadowed areas represent the different periods of P1 (white, before lockdown), P2 (pink, Spring Festival), P3 (blue, during lockdown), and P4 (yellow, after lockdown). Fig 3 While a decline in the annual trend was found, two severe haze episodes were also seen during 24–28 January and 6–13 February in 2020, with higher PM2.5, PM10, and CO concentrations than those of 2015–19. These two episodes coincided with unfavorable meteorological conditions, e.g., high relative humidity, low wind speed, and lowered mixed-layer height (Le et al., 2020). One implication here is that the meteorological conditions may have masked the real changes in air pollutants, leading to the unexpected increase in observed concentrations. In the next section, to better understand the real changes in concentrations for each pollutant, we try to decouple the meteorological impacts. 3.2 Changes in deweathered concentrations of air pollutants Meteorological normalization based on RF1 provides us with an opportunity to quantify the concentrations of air pollutants without the impact of meteorological conditions. Fig. 4 shows the deweathered concentrations against observations of each air pollutant at the four site types in Beijing, with detailed statistics listed in Table S7.Fig. 4 Observed and “deweathered” concentrations of each air pollutant at four site types in different periods in Beijing. Numbers on bars indicate relative changes between the observed and deweathered concentrations. Fig 4 As shown in Fig. 4 and Fig. S1, the deweathered concentrations of air pollutants show a similar temporal pattern with the observations at different site types. Except for the P2 period, there is a minor difference between the deweathered and observed concentrations, with an R dew of only a few percent or higher. This result indicates that the meteorological conditions in 2020 did not differ significantly to those during 2015–19. During the period of P2, however, the deweathered concentrations of air pollutants were significantly lower than their observed values, indicating unfavorable meteorological conditions for pollutant dispersion (Zhang et al., 2020). In other words, these unfavorable meteorological conditions in P2 caused an increase by 92.0%–126.4%, 59.6%–87.0%, 10.7%–25.3%, 15.2%–32.4%, and 42.2%–65.7% for the concentrations of PM2.5, PM10, SO2, NO2, and CO, respectively (Fig. 4). Clearly, particulate concentrations had the most significant increase, with the R dew changing from 92% at the background sites to 126.4% at the urban sites. After meteorological normalization, the deweathered concentrations of air pollutants, to some extent, reflected the real changes in emissions of the target air pollutants. As shown in Fig. 4, the deweathered concentrations of air pollutants (except O3) decreased significantly from P2 to P3 at all site types. Compared with the P2 period, the deweathered concentrations of PM2.5, PM10, SO2, NO2, and CO in P3 decreased by 15.8%–22.7%, 25.1%–53.6%, 27.6%–37.7%, 9.5%–15.6%, and 11.1%–12.5%, suggesting the large reduction was due to lockdown measures rather than meteorology. 3.3 Changes in predicted concentrations of air pollutants After decoupling the effects of meteorology, we confirmed that the concentrations of most air pollutants decreased because of the lockdown measures during P3, while the impact level of lockdown on air quality remains unclear. Thus, here we applied RF2 to predict the concentrations of air pollutants assuming no lockdown measures. The predicted results are shown in Fig. 5 and Table S8.Fig. 5 Observed and predicted concentrations of air pollutants at four site types in different periods in Beijing. Numbers on bars indicate relative changes between the observed and predicted concentrations. Fig 5 With the exception of PM2.5 (in P2) and O3, the predicted concentrations of other pollutants were higher than the observations at most sites in P2–P4 (Fig. S2 and Fig. S3). The findings indicate that the decrease in primary pollutants can be attributed to the lockdown measures in Beijing (Li et al., 2020; Wang et al., 2020). Similar changes in air pollutants during the COVID-19 pandemic have been found at other cities worldwide (Table S9). Note that during the P1 period, there is no remarkable difference between the observed and predicted concentrations of any pollutant. This result may indicate that the emissions in P1 were not markedly different between 2020 and 2015–19. After the series of lockdown measures adopted in P3, however, a significant R pre of 36.5%–42.3%, 48.4%–52.5%, 48.6%–54.4%, 37.1%–51.0%, and 33.5%–37.8% was found for PM2.5, PM10, SO2, NO2, and CO, respectively. Similar decreases in air pollutants in P3 have been reported in Hebei Province, China, and the whole of Korea (Jiang et al., 2021; Ju et al., 2021). With the resumption of “normal life” and the reopening of industries in P4, on the other hand, the R pre shows a decreasing tendency compared to P3, indicating the increased emissions with the resumption of economic activities. Since there are 35 air quality monitoring sites, we further compared the R pre of each pollutant at different site types. The R pre of PM2.5 and SO2 in P3 at the transport sites were 42.3% and 54.4%, respectively, which was higher than those at other site types. Notably, PM10 had the most significant relative change at the urban sites, due to the limitations imposed on the transportation and construction industries (Zhang et al., 2017; Cheriyan et al., 2020). Note that previous studies highlighted the reduction of NO2 concentrations at transport sites in response to lockdown measures such as in Shanghai (from 58.4 to 39.4 µg m−3, decrease of 32.5%) (Wu et al., 2021), Rome (50.8–22.4 µg m−3, 55.9%), Nice (46.9–14.6 µg m−3, 68.9%) (Sicard et al., 2020), and Madrid (42.7–15.9 µg m−3, 62.7%) (Baldasano, 2020). However, there are few reports on background sites. In our study, the most significant change for NO2 was found at the background sites (31.4–15.4 µg m−3, 51%) rather than the transport sites (55.3–34.8 µg m−3, 37.1%). In addition, a slightly larger decline in CO at the background sites (37.8%) than transport sites (35.5%) was also found in this study. These findings indicated that the reduced traffic emissions in urban areas may have a wide impact on regional air quality during the pandemic (Sicard et al., 2020). 3.4 Implications for future control policy The COVID-19 pandemic provided a natural experiment to evaluate the impacts of human activities on air pollutants. We found that most air pollutants decreased significantly—by 39.4%, 50.1%, 51.8%, 43.1%, and 35.1% for PM2.5, PM10, SO2, NO2, and CO, respectively—during the lockdown period, due to substantial reductions in anthropogenic emissions. The largest reduction was found at transportation sties for PM2.5 and SO2, at background sites for CO and NO2, and at urban sites for PM10. These reduction ratios may represent the upper limit of emission control for Beijing at the present economic and technological level. Although the air quality has improved gradually in recent years and pollution levels reached a minimum in 2020 for most air pollutants, several severe haze events still occurred around the Spring Festival, with higher PM2.5, PM10 and CO concentrations than those of 2015–19. These unexpected pollution episodes coincided with unfavorable meteorological conditions rather than enhanced emissions. In other words, the unfavorable meteorological conditions may have offset the decreases in concentrations of air pollutants caused by the reduction in emissions. Thus, stricter mitigation strategies in reducing emissions under stagnant weathers are still needed in the future. Different from the decline trends of the other air pollutants, the concentration of O3 were more stable or increased in recent years. In addition, the observed concentrations of O3 during P2 and P3 were 41.9%–84.3% and 11.3%–38.6% higher than its predicted values. This pattern was also different from the other air pollutants that a large reduction was found during lockdown periods. Some studies argue that the imbalanced reductions of NOx versus volatile organic compounds (VOCs) lead to higher VOCs-NOx ratio, and hence enhanced the unintended increase of O3 (Sicard et al., 2020; Zhang et al., 2021). Besides, the reduction in particulate matter was also suggested to increase O3 concentrations via the heterogeneous chemical processes (Li et al., 2017). Thus, future control policies will require a coordinated and balanced approach toward O3 and PM2.5, taking into full account both primary emissions and secondary processes. Appendix Supplementary materials Image, application 1 Disclosure Statement No potential conflict of interest was reported by the authors. Acknowledgment This work was supported by the 10.13039/501100001809 National Natural Science Foundation of China (Grant number 42077204) and the National Key Research and Development Project (Grant number 2017YFC0210103), with data support provided by the National Earth System Science Data Center, National Science & Technology Infrastructure of China (http://www.geodata.cn). Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.aosl.2021.100060. ==== Refs References Baldasano J.M. COVID-19 lockdown effects on air quality by NO2 in the cities of Barcelona and Madrid (Spain) Sci. Total Environ. 741 2020 140353 10.1016/j.scitotenv.2020.140353 Breiman L. Random forests Mach. Learn. 45 2001 5 32 10.1023/A:1010933404324 Cheriyan D. Hyun K.Y. Jaegoo H. Choi J.-H. Assessing the distributional characteristics of PM10, PM2.5, and PM1 exposure profile produced and propagated from a construction activity J. Clean. Prod. 276 2020 124335 10.1016/j.jclepro.2020.124335 Grange S.K. Carslaw D.C. Lewis A.C. Boleti E. Hueglin C. Random forest meteorological normalisation models for Swiss PM10 trend analysis Atmos. Chem. Phys. 18 9 2018 6223 6239 10.5194/acp-18-6223-2018 Hashim B.M. Al-Naseri S.K. Al-Maliki A. Al-Ansari N. Impact of COVID-19 lockdown on NO2, O3, PM2.5 and PM10 concentrations and assessing air quality changes in Baghdad Iraq. Sci. Total Environ. 754 2021 141978 10.1016/j.scitotenv.2020.141978 He Y. Pan Y. Gu M. Sun Q. Zhang Q. Zhang R. Wang Y. Changes of ammonia concentrations in wintertime on the North China Plain from 2018 to 2020 Atmos. Res. 253 2021 105490 10.1016/j.atmosres.2021.105490 Huang X. Ding A. Gao J. Zheng B. Zhou D. Qi X. Tang R. Enhanced secondary pollution offset reduction of primary emissions during COVID-19 lockdown in China Natl. Sci. Rev. 8 2 2020 nwaa137 10.1093/nsr/nwaa137 34676092 Jiang S. Zhao C. Fan H. Toward understanding the variation of air quality based on a comprehensive analysis in Hebei province under the influence of COVID-19 lockdown Atmos 12 2 2021 267 10.3390/atmos12020267 Ju M.J. Oh J. Choi Y.-H. Changes in air pollution levels after COVID-19 outbreak in Korea Sci. Total Environ. 750 2021 141521 10.1016/j.scitotenv.2020.141521 Kumari P. Toshniwal D. Impact of lockdown on air quality over major cities across the globe during COVID-19 pandemic Urban Clim 34 2020 100719 10.1016/j.uclim.2020.100719 Le T. Wang Y. Liu L. Yang J. Yung Y.L. Li J. Seinfeld J.H. Unexpected air pollution with marked emission reductions during the COVID-19 outbreak in China Science 369 6504 2020 702 706 10.1126/science.abb7431 32554754 Li L. Li Q. Huang L. Wang Q. Zhu A. Xu J. Liu Z. Air quality changes during the COVID-19 lockdown over the Yangtze River Delta Region: An insight into the impact of human activity pattern changes on air pollution variation Sci. Total Environ. 732 2020 139282 10.1016/j.scitotenv.2020.139282 Li M. Wang T. Xie M. Zhuang B. Li S. Han Y. Chen P. Impacts of aerosol-radiation feedback on local air quality during a severe haze episode in Nanjing megacity, eastern China Tellus B: Chem. Phys. Meteorol. 69 1 2017 1339548 10.1080/16000889.2017.1339548 Liu Z. Ciais P. Deng Z. Lei R. Davis S.J. Feng S. Zheng B. Near-real-time monitoring of global CO2 emissions reveals the effects of the COVID-19 pandemic Nat. Commun. 11 1 2020 5172 10.1038/s41467-020-18922-7 33057164 Mahato S. Pal S. Ghosh K.G. Effect of lockdown amid COVID-19 pandemic on air quality of the megacity Delhi India. Sci. Total Environ. 730 2020 139086 10.1016/j.scitotenv.2020.139086 Nakada L.Y.K. Urban R.C. COVID-19 pandemic: Impacts on the air quality during the partial lockdown in Sao Paulo state Braz. Sci. Total Environ. 730 2020 139087 10.1016/j.scitotenv.2020.139087 Pei Z. Han G. Ma X. Su H. Gong W. Response of major air pollutants to COVID-19 lockdowns in China Sci. Total Environ. 743 2020 140879 10.1016/j.scitotenv.2020.140879 Shi Z. Song C. Liu B. Lu G. Xu J. Van Vu T. Elliott R.J.R. Abrupt but smaller than expected changes in surface air quality attributable to COVID-19 lockdowns Sci. Adv. 7 3 2021 eabd6696 10.1126/sciadv.abd6696 33523881 Sicard P. De Marco A. Agathokleous E. Feng Z. Xu X. Paoletti E. Rodriguez J.J.D. Calatayud V. Amplified ozone pollution in cities during the COVID-19 lockdown Sci. Total Environ. 735 2020 139542 10.1016/j.scitotenv.2020.139542 Vu T.V. Shi Z. Cheng J. Zhang Q. He K. Wang S. Harrison R.M. Assessing the impact of clean air action on air quality trends in Beijing using a machine learning technique Atmos. Chem. Phys. 19 17 2019 11303 11314 10.5194/acp-19-11303-2019 Wang Y. Li W. Gao W. Liu Z. Tian S. Shen R. Ji D. Trends in particulate matter and its chemical compositions in China from 2013–2017 Sci.China Earth Sci 62 12 2019 1857 1871 10.1007/s11430-018-9373-1 Wang Y. Wen Y. Wang Y. Zhang S. Zhang K.M. Zheng H. Xing J. Wu Y. Hao J. Four-month changes in air quality during and after the COVID-19 Lockdown in six megacities in China Environ. Sci. Technol. Lett. 7 11 2020 802 808 10.1021/acs.estlett.0c00605 Wu C.L. Wang H.W. Cai W.J. He H.D. Ni A.N. Peng Z.R. Impact of the COVID-19 lockdown on roadside traffic-related air pollution in Shanghai China. Build. Environ. 194 2021 107718 10.1016/j.buildenv.2021.107718 Zangari S. Hill D.T. Charette A.T. Mirowsky J.E. Air quality changes in New York City during the COVID-19 pandemic Sci. Total Environ. 742 2020 140496 10.1016/j.scitotenv.2020.140496 Zhang Q. Pan Y. He Y. Walters W.W. Ni Q. Liu X. Xu G. Shao J. Jiang C. Substantial nitrogen oxides emission reduction from China due to COVID-19 and its impact on surface ozone and aerosol pollution Sci. Total Environ. 753 2021 142238 10.1016/j.scitotenv.2020.142238 Zhang Y. Vu T.V. Sun J. He J. Shen X. Lin W. Zhang X. Significant changes in chemistry of fine particles in wintertime Beijing from 2007 to 2017: Impact of clean air actions Environ. Sci. Technol. 54 3 2020 1344 1352 10.1021/acs.est.9b04678 31766848 Zhang Z. Hu M. Ren J. Zhang Z. Christakos G. Wang J. Probabilistic assessment of high concentrations of particulate matter (PM10) in Beijing China. Atmos. Pollut. Res. 8 6 2017 1143 1150 10.1016/j.apr.2017.04.006

0

PMC9748733

NO-CC CODE

2022-12-15 23:22:46

no

2021 Jul 30; 14(4):100060

utf-8

oa_other

==== Front Chest Chest Chest 0012-3692 1931-3543 American College of Chest Physicians. Published by Elsevier Inc. S0012-3692(20)34929-1 10.1016/j.chest.2020.10.034 Editorial Physician Leadership New Responsibilities Require New Skills Kelley Mark A. MD ∗ Pulmonary-Critical Care Division, Massachusetts General Hospital, Boston, MA ∗ CORRESPONDENCE TO: Mark A. Kelley, MD 4 3 2021 3 2021 4 3 2021 159 3 902903 © 2020 American College of Chest Physicians. Published by Elsevier Inc. All rights reserved. 2020 American College of Chest Physicians Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcFOR RELATED ARTICLE, SEE PAGE 1147 In the coronavirus disease 2019 (COVID-19) pandemic, health professionals have saved thousands of patients lives. Pulmonary and critical care physicians have been particularly outstanding in confronting this crisis. That outcome is not a surprise. ICU physicians are experienced in leading teams that treat critically ill patients. Some of these leaders have advanced to become senior executives in their institutions. That trend happened over the last 50 years, as health care became one of the largest sectors of the US economy. During that time, the industry learned that experienced physician leaders are effective because they understand both medical science and patient care. The philosophy of “patient first” in quality and safety has led hospitals to place more physicians into leadership roles.1 , 2 That change has led to better quality in many institutions.3 , 4 Historically, the education of physician leaders was informal at best. Although some leaders pursued advanced degrees, many jumped into a new leadership role and hoped they could learn on the job. That era of “do one, teach one” is over, in both clinical medicine and medical leadership. Each requires its own structured educational program.5 , 6 In this month’s publication, CHEST begins a four-part series on physician leadership. The content, although focused on chest physicians, is useful for any health professional interested in leadership. The author, Dr James Stoller, is a well-known chest physician and educator, with extensive experience developing physician leaders at Cleveland Clinic. Health care is a challenging blend of social services, business, and rapidly changing medical discovery. The first essay in the series entitled, “Leadership Essentials,”7 goes right to the point. No matter how well versed in medicine, most physicians have not been trained to lead people and manage organizations. Consequently, clinicians can find themselves unprepared for a significant leadership role. The good news is that leadership skills needed for such an assignment can be taught and mastered. As Dr Stoller explains, leadership is more sophisticated than one might imagine. For example, using “situational leadership,” leaders can solve different problems by using different leadership styles, such as telling, selling, participating, or delegating. This approach resembles that of a quarterback who uses different plays during a game. There is general agreement that leadership requires mastery in two areas. The first is “threshold competencies” (ie, content not taught in medical training). These topics include operations, finances, policy, quality, and negotiation. After acquiring such knowledge, the new leader focuses on learning the “people skills” of leadership. This CHEST series describes the core of these skills in three subsequent essays: “Managing Change”; “Emotional Intelligence”; and “Building Teams.” These important “social competencies” are best mastered during ongoing leadership experience. Because of the demand for more physician leaders, there are many ways to receive this education. Advanced degrees (eg, MPH, MHA, MBA) can be attractive but may focus more on professional content than leadership skills. Leadership taught in nondegree programs may concentrate more on “people” skills but also may include business topics. Some physician organizations offer a blend of both options. For example, the Association for Physician leaders and the American College of Physicians have organized leadership programs than can also include a master’s degree. Like medicine itself, leadership can be learned from sources with different strengths in teaching the core curriculum.6 Some institutions, such as the Mayo Clinic and the Cleveland Clinic, have developed their own internal leadership programs.5 The goal is to continue to educate and support leaders as their experience and responsibilities expand. Over time, this creates a systemwide culture in which all leaders share common skills in supporting their teams and each other. This customized model of leadership education is becoming more popular, especially in other large health organizations such as MGH Brigham, Penn Medicine, and the Henry Ford Health System. Many physicians are currently in leadership positions—whether in their practice, hospital, or health system. These leadership roles range in complexity—from managing a small clinic to serving as the hospital CEO. Regardless of the assignment, success depends on teams and those who lead them. In this modern era of medicine, leadership training at some level is important for every physician leader. It is up to the individual to determine how, when, and where to pursue leadership education. This CHEST leadership series is an excellent introduction for beginners and may offer some new insights for those with experience. In many ways, leadership education resembles clinical training. The process begins with new knowledge, followed by learning new interpersonal skills. In medicine, physicians learn clinical science and the skills to interact with their patients. In leadership, physicians learn management and how to inspire their teams. In both scenarios, many lessons are learned in jobs that demand continuous education. The physician leader stands at the crossroads of health care delivery, clinical medicine, and its “business” infrastructure: operations, economics, human resources, and informatics. Skillfully aligning these forces to help patients is a noble task. To create a patient-friendly system that provides excellent, reliable, and safe care is the most important goal in medicine—and the most difficult. We need more physician leaders who are well prepared and eager to meet that challenge. FINANCIAL/NONFINANCIAL DISCLOSURES: None declared. ==== Refs References 1 Berwick D.M. Nolan T.W. Physicians as leaders in improving health care Ann Intern Med 128 4 1998 289 292 9471932 2 Lee T.H. Turning doctors into leaders Harv Bus Rev 88 4 2010 50 58 3 Tasi M.C. Keswani A. Bozic K.J. Does physician leadership affect hospital quality, operational efficiency, and financial performance? Health Care Manage Rev 44 3 2019 256 262 28700509 4 Goodall A. Physician-leaders and hospital performance: Is there an association? Soc Sci Med 73 4 2011 535 539 21802184 5 Stoller J.K. Developing physician leaders: a perspective on rationale, current experience, and needs Chest 154 1 2018 16 20 30044730 6 Sonnino R.E. Health care leadership development and training: progress and pitfalls J Healthc Leadersh 8 2016 19 29 29355187 7 Stoller J.K. Leadership essentials for CHEST Medicine Professionals: models, attributes, and styles Chest 159 3 2021 1147 1154 32956716

33678271

PMC9748787

NO-CC CODE

2022-12-15 23:22:46

no

Chest. 2021 Mar 4; 159(3):902-903

utf-8

Chest

10.1016/j.chest.2020.10.034

oa_other

==== Front Chest Chest Chest 0012-3692 1931-3543 American College of Chest Physicians. Published by Elsevier Inc. S0012-3692(20)34910-2 10.1016/j.chest.2020.10.018 General Interest Commentary and Announcement Never Let a Good Crisis Go to Waste Metersky Mark L. MD a∗ Aliberti Stefano MD b Feldman Charles MBBCh, PhD c Luna Carlos M. MD, PhD d Shindo Yuichiro MD, PhD e Sotgiu Giovanni PhD, MD f Waterer Grant MBBS, PhD gh a Division of Pulmonary, Critical Care and Sleep Medicine, University of Connecticut School of Medicine, Farmington, CT b Department of Pathophysiology and Transplantation, and Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico, Internal Medicine Department, Respiratory Unit and Cystic Fibrosis Adult Center, University of Milan, Milan, Italy c Division of Pulmonology, Department of Internal Medicine, Charlotte Maxeke Johannesburg Academic Hospital, Faculty of Health Sciences, University of the Witwatersrand and Department of Internal Medicine, University of the Witwatersrand Medical School, Johannesburg, South Africa d Facultad de Medicina, Universidad de Buenos Aires, Buenos Aires, Argentina e Department of Respiratory Medicine, Nagoya University Graduate School of Medicine, Nagoya, Japan f Clinical Epidemiology and Medical Statistics Unit, Department of Medical, Surgical, Experimental Sciences, University of Sassari and Clinical Epidemiology and Medical Statistics Unit, Department of Clinical and Experimental Medicine, University of Sassari, Sassari, Italy g University of Western Australia, Perth, Australia h Northwestern University, Chicago, IL ∗ CORRESPONDENCE TO: Mark L. Metersky, MD 4 3 2021 3 2021 4 3 2021 159 3 917919 © 2020 American College of Chest Physicians. Published by Elsevier Inc. All rights reserved. 2020 American College of Chest Physicians Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Key Words community-acquired pneumonia COVID-19 crisis hand sanitizer hand washing mask wearing pandemic physical distancing respiratory infection respiratory virus Abbreviations CAP, community-acquired pneumonia COVID-19, coronavirus disease 2019 ==== Body pmcA scourge of respiratory infections kills millions of people across the world every year. The coronavirus disease 2019 (COVID-19) pandemic has caused tremendous suffering and mortality, both directly and indirectly as a result of the resulting economic slowdown. However, the “usual” highly incident respiratory viruses, including rhinovirus, enterovirus, coronavirus, respiratory syncytial virus, influenza and parainfluenza viruses, and adenovirus, are also responsible for a huge toll of morbidity, mortality, and lost economic productivity. Human rhinovirus may be the most common cause of community-acquired pneumonia (CAP) requiring hospital admission1 and respiratory viruses overall are responsible for more than half of all CAP in which an etiologic agent can be determined,1 and one third of severe pneumonias.2 Given the 13% 30-day mortality among the 1.6 million CAP hospitalizations per year in the United States,3 viral pneumonia causes nearly 70,000 deaths per year. Viral infections also cause approximately 40% of COPD exacerbations4 and are thus responsible for approximately 400,000 of the one million annual COPD hospitalizations,5 of which more than 8%6 (32,000) are fatal. Influenza, even without pneumonia, often results in exacerbations of underlying comorbidities, explaining the Centers for Disease Control and Prevention estimate of approximately 40,000 average annual influenza deaths. Less appreciated is the fact that “uncomplicated” viral respiratory infections can activate thrombotic pathways, leading to stroke, myocardial infarction, exacerbation of heart failure, and other cardiovascular events, adding to the morbidity and mortality associated with viral respiratory infections. Based on these data, a conservative estimate is that respiratory viruses cause at least 150,000 deaths per year in the United States. The incidence of CAP7 and the prevalence of COPD8 in the rest of the world are similar to that seen in the United States, suggesting that the global mortality associated with highly incident respiratory viruses is over 3 million per year. The additional impact on antibiotic resistance due to inappropriate antibiotic treatment as well as appropriate treatment of secondary bacterial infections is likely substantial. Of course, the most common clinical syndrome associated with these viruses is the “common cold,” generally perceived as a mild illness with little morbidity. Furthermore, the impact is seen throughout the year and is the “usual state.” Likely for these reasons, the burden of morbidity and mortality associated with the highly incident respiratory viruses does not receive adequate attention from medical professionals, public health officials, and the public. Although much work is ongoing to develop compounds to treat or prevent many of these infections, there has been little success, and, other than influenza vaccine, whose effectiveness can greatly vary, vaccines are not available. Knowledge and experience gained during the current COVID-19 pandemic provide an opportunity to decrease the morbidity and mortality associated with highly incident respiratory viruses: their airborne human-to-human transmission shows the same pattern as COVID-19, highlighting the importance of physical distancing and mask wearing to decrease the forward spread. This is evidenced by the very low incidence of influenza, influenza-like illness, and respiratory syncytial virus reported from numerous locations around the world this year, almost surely explained by physical distancing and mask wearing. Numerous studies demonstrate correlation of COVID-19 rates with physical distancing measures, and other studies have demonstrated the effectiveness of mask use in preventing respiratory virus transmission. A year ago, it would have been unthinkable to call for patients with just a “cold” to use hand sanitizer, to isolate, or to wear a mask while in public. However, COVID-19 has made mask wearing acceptable in most countries and has been widely adopted. Similarly, there has been widespread acceptance of the necessity of isolating people with proven severe acute respiratory syndrome coronavirus 2 infection or quarantining those with documented exposure. It is not a stretch to believe that the public could be convinced of the importance of commonsense efforts to reduce respiratory viral transmission with appropriate education and advocacy. In fact, in some Asian countries (eg, Japan and China), mask wearing when suffering from a respiratory infection has been considered a common courtesy for decades. It has been said that one should “never let a good crisis go to waste.” The COVID-19 pandemic provides the opportunity to change the perception of respiratory viral infections as benign, self-limited illnesses that do not require efforts to prevent their transmission. With an appropriate public health campaign, knowledge, attitudes, and ultimately, human behaviors can be changed. Of course, it would be folly to suggest that businesses should be shut down and schools should be kept closed to control highly incident viral respiratory infections. Nor is it likely to be accepted that anyone with a cold should be prevented from going to school or work, or that anyone exposed to such a person should be quarantined. However, our experience with COVID-19 strongly suggests that less drastic interventions targeting people with active respiratory infection, including physical distancing (not isolation), mask wearing (eg, indoors and on public transportation), and perhaps more frequent use of hand sanitizers and hand washing would decrease viral transmission. Accordingly, we have the following proposals for consideration by the clinical and public health communities:1. Public health, respiratory, and infectious disease experts should collaborate on research examining the effectiveness of interventions such as mask-wearing, physical distancing, and hand washing/hand sanitizers in preventing transmission of respiratory viruses in real-life settings. Subsequent research will need to examine the potential acceptance of the various interventions found to have benefit. Both effectiveness and acceptability will likely vary among different countries and socioeconomic and cultural settings. What works or is accepted in urban high-resource areas might not work or be acceptable or affordable in rural or low-resource areas. Research will need to address the expected variability and should find suitable solutions. 2. Funding agencies should recognize the tremendous societal benefit of such research, despite it not being “mechanistic” or “cutting edge.” Specific funding lines should be created to foster these areas of investigation. 3. Once the optimum interventions are established, media campaigns will be needed to educate the public on their benefits and, similar to anti-smoking campaigns, modify attitudes and behaviors. These could include advocacy, employing survivors of severe respiratory infections. The initial campaigns themselves will need to be examined critically to assess impact and allow modification as needed. The response to COVID-19 has demonstrated that simple interventions can slow community spread of viral respiratory disease and that large populations can be motivated to rapidly change behavior when convinced of the need. It should be possible to achieve the same goals with respect to highly incident respiratory viruses, and in doing so, reduce on a global level their associated morbidity, mortality, and lost economic productivity. FINANCIAL/NONFINANCIAL DISCLOSURES: None declared. ==== Refs References 1 Jain S. Self W.H. Wunderink R.G. Community-acquired pneumonia requiring hospitalization among U.S. adults N Engl J Med 373 5 2015 415 427 26172429 2 Choi S.H. Hong S.B. Ko G.B. Viral infection in patients with severe pneumonia requiring intensive care unit admission Am J Respir Crit Care Med 186 4 2012 325 332 22700859 3 Ramirez J.A. Wiemken T.L. Peyrani P. Adults hospitalized with pneumonia in the United States: incidence, epidemiology, and mortality Clin Infect Dis 65 11 2017 1806 1812 29020164 4 Wedzicha J.A. Role of viruses in exacerbations of chronic obstructive pulmonary disease Proc Am Thorac Soc 1 2 2004 115 120 16113423 5 Jacobs D.M. Noyes K. Zhao J. Early hospital readmissions after an acute exacerbation of chronic obstructive pulmonary disease in the Nationwide Readmissions Database Ann Am Thorac Soc 15 7 2018 837 845 29611719 6 Hospital Compare. U.S. Centers for Medicare & Medicaid Services https://www.medicare.gov/hospitalcompare/search.html 7 Shi T. Denouel A. Tietjen A.K. Global and regional burden of hospital admissions for pneumonia in older adults: a systematic review and meta-analysis J Infect Dis 222 Suppl 7 2020 S570 S576 30849172 8 Ruvuna L. Sood A. Epidemiology of chronic obstructive pulmonary disease Clin Chest Med 41 3 2020 315 327 32800187

33678278

PMC9748788

NO-CC CODE

2022-12-15 23:22:47

no

Chest. 2021 Mar 4; 159(3):917-919

utf-8

Chest

10.1016/j.chest.2020.10.018

oa_other

==== Front Chest Chest Chest 0012-3692 1931-3543 American College of Chest Physicians S0012-3692(20)35502-1 10.1016/j.chest.2020.12.019 Humanities: Vantage Categorized Priority Systems A New Tool for Fairly Allocating Scarce Medical Resources in the Face of Profound Social Inequities Sönmez Tayfun PhD a Pathak Parag A. PhD b Ünver M. Utku PhD a Persad Govind JD, PhD c Truog Robert D. MD d White Douglas B. MD e∗ a Department of Economics, Boston College, Chestnut Hill, MA b Department of Economics, Massachusetts Institute of Technology, Cambridge, MA c University of Denver Sturm College of Law, Denver, CO d Center for Bioethics, Harvard Medical School, and the Department of Anesthesiology, Critical Care, and Pain Medicine, Boston Children’s Hospital, Boston, MA e Department of Critical Care Medicine, Program on Ethics and Decision Making in Critical Illness, University of Pittsburgh School of Medicine, Pittsburgh, PA ∗ CORRESPONDENCE TO: Douglas B. White, MD 26 12 2020 3 2021 26 12 2020 159 3 12941299 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcThe coronavirus disease 2019 (COVID-19) pandemic has motivated medical ethicists and several task forces to revisit or issue new guidelines on allocating scarce medical resources.1, 2, 3 Such guidelines are relevant for the allocation of scarce therapeutics and vaccines and for allocation of ICU beds, ventilators, and other life-sustaining treatments or potentially scarce interventions. Principles underlying these guidelines, like saving the most lives, mitigating disparities, reciprocity to those who assume additional risk (eg, essential workers and clinical trial participants), and equal access4 , 5 may compete with one another. We propose the use of a “categorized priority system” (also known as a “reserve system”) as an improvement over existing allocation methods, particularly because it may be able to achieve disparity mitigation better than other methods. Early in the pandemic, several states adopted a single-principle priority point system (PPS) to allocate scarce ventilators, based on patients’ chances of survival to hospital discharge. In contrast, White6 developed a multi-principle PPS to accommodate multiple ethical principles. The framework is designed to promote population health outcomes by giving priority to those most likely to survive the hospitalization and to survive in the near term after hospital discharge. It is designed to promote equity by (1) giving heightened priority to essential personnel, which includes a racially and ethnically diverse group of health-care workers7 and individuals who play a critical role in the public health response,8 and (2) giving some priority to patients who have had the least chance to live through life’s stages. Some version of a PPS is the leading mechanism for rationing ventilators in the United States. Several states, including New York and Minnesota, use a PPS as part of guidelines that predate the COVID-19 pandemic. During the current COVID-19 pandemic, at least 25 states used a PPS as part of enacted or proposed guidelines for ventilator allocation. Although a PPS is a valuable allocation mechanism because it can incorporate multiple values, it may not be able to achieve the most ethically- compelling balance between certain values. A PPS assigns individual attributes to a numeric scale and does not handle ethical values that are not commensurable. A reliance on individual attributes also means that a PPS struggles to incorporate group-based policies, such as the mitigation of population-level health disparities, which is a pressing issue in the COVID-19 pandemic given the disproportionate death rates among Black, Hispanic/Latino, and indigenous communities. Existing debates over prioritizing essential workers illustrate a central problem with the use of a PPS as the allocation mechanism. Several rationing guidelines endorse prioritizing essential personnel, but others hesitate out of concern that essential personnel may take all of the scarce resources under a PPS.9, 10, 11 These difficulties can be overcome by moving beyond a PPS to a categorized priority system (CPS). How a CPS Can Enable More Fair Allocation A CPS divides resources into multiple categories, enabling the use of different criteria for allocation of resources within each category.12 Category-specific criteria can reflect the balance of ethical values guiding allocation of units in the category. A CPS does not need to use uniform criteria across all categories, making it more flexible than a PPS. CPS, which commonly are more known as reserve systems, have been used in practice. In medicine, they were adopted for allocation of deceased donor kidneys in 2014.13 They are used in school choice systems in Boston, Chicago, and New York City, for the assignment of marathon slots in Boston and New York City, for the allocation of H-1B visas in the United States, and for affirmative action policies worldwide.12 , 14, 15, 16 CPS for Essential Personnel: A Case Study We use a stylized example of giving heightened priority to essential personnel for the scarce therapeutic remdesivir to illustrate how a CPS can enable a wider, and potentially fairer, set of allocation options and avoid requiring decision-makers to choose between two extreme essential worker policies (Fig 1 ).Figure 1 A-D, Three different allocation mechanisms: A, Remdesivir and patients; B, No priority for essential personnel; C, Absolute priority for essential personnel; D, 80/20 categorized priority system. Suppose there are 100 courses of remdesivir available for a population of 200 patients. Eighty patients are identified as essential personnel, and 120 patients are from the general community. A task force wishes to allocate remdesivir based on clinical criteria and would also like to give some priority for essential personnel. Suppose that both patient types have an identical distribution of clinical scores. Figure 1A shows the total supply of remdesivir and the two patient types, ordered by their clinical score. Our first scenario (Fig 1B) is a PPS based only on clinical criteria. Because we stipulate that the distribution of clinical scores is identical between the two groups, the allocation corresponds to the proportion of the two patient types in the population. Essential personnel receive (80/200) × 100 = 40 courses of remdesivir, and general-community patients receive (120/200) × 100 = 60. Our second scenario (Fig 1C) is a PPS based on giving absolute priority to essential personnel and then allocating based on clinical criteria among essential personnel and also among the general community population. In this scenario, all 80 essential personnel receive treatment because their total number is less than the total number of courses available. After that, the 20 general community patients with highest priority based on clinical criteria receive courses of remdesivir. Our third scenario (Fig 1D) is a CPS with two categories: an 80-course open category and a 20-course category for which essential personnel receive first priority. Units in the open category are allocated solely based on clinical criteria, whereas essential personnel status is considered prior to clinical criteria in the essential personnel category. Under this CPS, the first 80 courses of remdesivir are assigned by clinical score, just as all were in the first scenario. Therefore, essential personnel receive (80/200) × 80 = 32 open-category courses, and general community members receive (120/200) × 80 = 48 open-category courses. Next, essential personnel receive all 20 courses of remdesivir in the second category. In total, essential personnel receive 32 + 20 = 52 courses, and general community patients receive 48 courses. In addition to specifying categories and specifying criteria for allocation in each category, a CPS permits two other types of policy choices. One is the quantity of remdesivir assigned to each category. Figure 2 shows how the distribution of remdesivir changes as the essential personnel category increases in size from 0 to 100, with the 20 specified in our example highlighted. As the essential personnel category grows, the quantity assigned to essential personnel likewise increases from 40 to 80 courses. Figure 2 shows that a CPS can accommodate allocations that range from no essential personnel priority (used for ventilators in Minnesota and New York) to absolute essential personnel priority (used for ventilators in Michigan). A CPS can enable allocation guidelines to implement a compromise rather than being confined to extreme options.Figure 2 Allocation under a categorized priority system. MI = Michigan; MN = Minnesota; NY = New York. The last policy choice is the order in which categories are processed. In our example, remdesivir in the open category is assigned first. However, the two categories could be processed in the opposite order. Research shows that processing the essential personnel category last provides them with the maximum benefit.12 , 14 , 16 Using a CPS in Current Debates A CPS can also expand the set of solutions to other key debates. Most prominent recently has been the debate over whether to incorporate disparity reduction into allocation policies.17 Compared with alternatives like random assignment,18 a CPS with categories for disadvantaged groups can address these concerns more effectively and can serve antiracist goals while complying with restrictions in some countries against the consideration of race at an individual level.19 As a parallel, after the Parents Involved decision barred school districts in the United States from considering individual students’ race, many school systems addressed educational disparities by implementing a CPS based on socioeconomic criteria: for instance, by categorizing some seats in selective enrollment high schools as preferentially accessible to qualified students from socioeconomically disadvantaged areas.16 Using a socioeconomic index such as the Area Deprivation Index or Social Vulnerability Index to allocate treatments,20, 21, 22 a CPS could define a disadvantaged-group category and then use clinical criteria within it, both aiming to save more lives and to ensure that disadvantaged groups are not excluded from access. Meanwhile, many disability rights advocates fear disabilities will impact prognostication inappropriately, and some go further to reject the ethical relevance of probability or length of survival, preferring random or first-come, first-served assignment.23 , 24 In this view, equal access trumps other ethical considerations. A CPS could include a category for disabled individuals (as defined by the triage committee), with different prioritization rules for this category. This would permit patients with disabilities to achieve their desired within-group allocation without affecting the allocation criteria used for others. Similarly, a small open category with random prioritization among all patients who can benefit would provide everyone some chance of obtaining the scarce resource at issue. Another debate involves adults and pediatric patients, between whom the metrics used to measure death risk are not readily comparable. Nevertheless, some guidelines that use a PPS assign point scores for both groups using the same index.25 A CPS allows separate criteria for children and adults. Operationalizing a CPS A CPS enables allocation guidelines to use four policy levers: (1) the number and specification of categories, (2) the size of each category, (3) the rules for prioritization within each category, and (4) the order in which categories are processed (Fig 3 ). Although each of these levers require more detailed specification, we view this as a strength, because a CPS allows policy choices to be identified separately while a PPS necessitates the potentially more obscure translation of multiple ethical values into a single scale. There is a clear association between a category and the desired balance of ethical values for its beneficiaries. Moreover, the size of the category clearly represents the extent to which the beneficiaries of the category receive greater access. Finally, a CPS can be adjusted easily and be responsive to emerging data.Figure 3 Three main ingredients of a categorized priority system. As with a PPS, decisions about these policy levers should come from community engagement exercises where citizens deliberatively examine ethical trade-offs. Community engagement seems most important for the identification of the categories and their sizes. For instance, community engagement might endorse a Good Samaritan category, which gives priority to participants in vaccine clinical trials or donors of blood, plasma, or even a kidney, based on reciprocity.1 , 12 A smaller number of categories may help ensure that a CPS is transparent and practical. Prioritization within each category, in contrast, might be based on medical factors. For example, if the community wishes to save more lives within a category, the priority rule for that category should be based on medical factors associated with survival. Allocation experts should be enlisted to ensure that the processing order of the categories promotes the community’s values, because the categories processed after the general population category will receive more resources than if they were processed before the general population category.12 For example, the importance of recognizing essential personnel’s contribution and ensuring their continued availability may warrant processing their category after the general category. In contrast, categories that aim to prevent exclusion from access should be processed before the general category. For instance, equal access may support a category for disadvantaged populations. Once this group obtains a certain share by processing their category before the general category, this minimum guarantee share can be sufficient, even if they cannot get extra units through later categories. Conclusion Although policymakers and clinicians have debated extensively about which ethical values should guide allocation of scarce medical resources during the pandemic, there has been comparatively little discussion of which allocation mechanism will best realize the ethical values selected. We believe an ideal allocation mechanism should permit a wide range of options for balancing different ethical values, rather than requiring a strict ordering of principles. A CPS improves on a PPS by allowing greater flexibility to balance ethical principles and ensure that allocation outcomes reflect ethical values. Although confronting scarcity in life-and-death situations is a dire and hopefully rare possibility, allocation guidelines must balance a variety of ethical values. The limitations that a PPS faces in balancing ethical principles risk upsetting the social contract between different community members. When revising or modifying guidelines during or after the COVID-19 pandemic, a CPS should be part of the arsenal. FINANCIAL/NONFINANCIAL DISCLOSURES: The authors have reported to CHEST the following: G. P. made public statements related to the allocation of scarce treatments as part of disseminating prior work in this area; was funded by a Greenwall Foundation grant (Faculty Scholars Program) outside the submitted work; personal fees from the American Society of Clinical Oncology and World Health Organization outside the submitted work. D. W. was funded by an NIH grant (NHLBI K24 HL148314) while the study was being conducted; personal fees from the American Thoracic Society, personal fees from UpToDate outside of the submitted work. None declared (T. S., M. Ü., P. P., and R. T.). ==== Refs References 1 Emanuel E.J. Persad G. Upshur R. Fair allocation of scarce medical resources in the time of Covid-19 N Engl J Med 382 21 2020 2049 2055 32202722 2 Truog R.D. Mitchell C. Daley G.Q. The toughest triage: allocating ventilators in a pandemic N Engl J Med 382 21 2020 1973 1975 32202721 3 White D.B. Lo B. A framework for rationing ventilators and critical care beds during the COVID-19 pandemic JAMA 323 18 2020 1773 1774 32219367 4 Emanuel E.J. Wertheimer A. Public health Who should get influenza vaccine when not all can? Science 312 5775 2006 854 855 16690847 5 Persad G. Wertheimer A. Emanuel E.J. Principles for allocation of scarce medical interventions Lancet 373 9661 2009 423 431 19186274 6 White D.B. A model hospital policy for allocating scarce critical care resources. Updated April 15, 2020 https://ccm.pitt.edu/sites/default/files/UnivPittsburgh_ModelHospitalResourcePolicy_2020_04_15.pdf 7 Artiga S. Rae M. Pham O. Hamel L. Muñana C. COVID-19 risks and impacts among health care workers by race/ethnicity. KFF. 2020 https://www.kff.org/racial-equity-and-health-policy/issue-brief/covid-19-risks-impacts-health-care-workers-race-ethnicity/ 8 Rogers T.N. Rogers C.R. VanSant-Webb E. Gu L.Y. Yan B. Qeadan F. Racial disparities in COVID-19 mortality among essential workers in the United States [published online ahead of print August 5, 2020] World Med Health Policy 2020 10.1002/wmh3.358 9 Michigan Guidelines for Ethical Allocation of Scarce Medical Resources and Services During Public Health Emergencies in Michigan. 2012 https://asprtracie.hhs.gov/technical-resources/resource/6396/guidelines-for-ethical-allocation-of-scarce-medical-resources-and-services-during-public-health-emncies-in-michigan 10 Vawter D.E. Garrett J.E. Gervais K.G. For the good of us all: ethically rationing health resources in Minnesota in a severe influenza pandemic 2010 Minnesota Pandemic Ethics Project, Minnesota Department of Health https://www.health.state.mn.us/communities/ep/surge/crisis/ethics.pdf 11 Zucker H.A. Adler K.P. Berens D.P. Ventilator allocation guidelines New York State Task Force on Life and the Law, New York Department of Health. November 2015 https://www.health.ny.gov/regulations/task_force/reports_publications/docs/ventilator_guidelines.pdf 12 Pathak P.A. Sönmez T. Ünver M.U. Yenmez M.B. Fair allocation of vaccines, ventilators and antiviral treatments: leaving no ethical value behind in health care rationing 2020 arXiv:2008.00374 [econ.TH] 13 UNOS Revised national kidney transplant allocation system is now in effect December 4 2014 https://unos.org/news/revised-national-kidney-transplant-allocation-system-match-kidneys-effectively-improve-access-hard-match-patients/ 14 Dur U. Kominers S.D. Pathak P.A. Sönmez T. Reserve design: unintended consequences and the demise of Boston’s walk zones Journal of Political Economy 126 6 2018 2457 2479 15 Hafalir I. Yenmez M.B. Yildirim M. Effective affirmative action in school choice Theoretical Economics 8 2 2013 325 363 16 Dur U. Pathak P.A. Sönmez T. Explicit vs statistical targeting in affirmative action: theory and evidence from Chicago’s exam schools J Econ Theory 187 2020 104996 17 Twohey M. Who gets a vaccine first? U.S. considers race in coronavirus plans, NY Times. July 16, 2020 https://www.nytimes.com/2020/07/09/us/coronavirus-vaccine.html 18 Jones C.P. Coronavirus disease discriminates. Our health care doesn't have to. Newsweek. April 7, 2020 https://www.newsweek.com/2020/04/24/coronavirus-disease-discriminates-our-health-care-doesnt-have-opinion-1496405.html 19 Persad G. Allocating medicine fairly in an unfair pandemic. University of Illinois Law Review. In press. 20 Kind A.J.H. Buckingham W.R. Making neighborhood-disadvantage metrics accessible: the Neighborhood Atlas N Engl J Med 378 26 2018 2456 2458 29949490 21 Schmidt H. Vaccine rationing and the urgency of social justice in the COVID-19 response Hastings Center Report 50 3 2020 46 49 22 Dasgupta S. Bowen V.B. Leidner A. Association between social vulnerability and a county’s risk for becoming a COVID-19 hotspot: United States, June 1-July 25, 2020 MMWR Morb Mortal Wkly Rep 69 42 2020 1535 1541 33090977 23 DREDF Preventing discrimination in the treatment of COVID-19 patients: The Illegality of Medical Rationing on the Basis of Disability, Disability Rights Education & Defense Fund. March 25, 2020 https://dredf.org/wp-content/uploads/2020/03/DREDF-Policy-Statement-on-COVID-19-and-Medical-Rationing-3-25-2020.pdf 24 Applying HHS’s guidance for states and health care providers on avoiding disability-based discrimination in treatment rationing. April 3, 2020 http://thearc.org/wp-content/uploads/2020/04/Guidance-to-States-Hospitals_FINAL.pdf 25 Crisis standards of care, planning guidance for the COVID-19 Pandemic, Commonwealth of Massachusetts, Department of Public Health. April 7, 2020 https://www.documentcloud.org/documents/6843353-Revised-Crisis-Standards-of-Care-Planning-Guidance.html

33373597

PMC9748801

NO-CC CODE

2022-12-15 23:22:47

no

Chest. 2021 Mar 26; 159(3):1294-1299

utf-8

Chest

10.1016/j.chest.2020.12.019

oa_other

==== Front Contraception Contraception Contraception 0010-7824 1879-0518 Published by Elsevier Inc. S0010-7824(21)00317-6 10.1016/j.contraception.2021.07.083 Article POSTER ABSTRACTS P65 RAPID INNOVATION AND IMPLEMENTATION OF TELEMEDICINE FOR CONTRACEPTION: PROVIDERS’ PERSPECTIVES Hasselbacher L Wong Z Boulware A Song B Delay R Young D Whitaker A Stulberg D Ci3, University of Chicago, Chicago, IL, US 6 9 2021 10 2021 6 9 2021 104 4 467467 Copyright © 2021 Published by Elsevier Inc. 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcObjectives: Many family planning providers faced disruptions to service delivery from the COVID-19 pandemic and needed to quickly implement telemedicine for contraceptive care. This study describes how Illinois providers responded to the uptake of telemedicine. Methods: We interviewed clinicians (n=20) from non-Planned Parenthood clinics (July – September 2020) and clinicians and staff (n=17) from Planned Parenthood of Illinois sites (December 2020 – March 2021) across the state. Interviews were conducted by phone or video, audio-recorded, transcribed, and coded in Dedoose. Analysis revealed themes focusing on telemedicine's effects on patient care and access to comprehensive contraceptive services, and on steps needed to improve telemedicine and sustain its benefits post-pandemic. Results: Interviewees expressed mostly positive attitudes towards telemedicine, highlighting its utility for counseling and prescription contraceptive methods, with additional benefits for rural patients and patients facing transportation, childcare, or other barriers to in-person visits. Challenges included rapid telemedicine rollout, patient barriers to accessing technology platforms, and reduced access to long-acting reversible contraception (LARC). Providers implemented changes to mitigate barriers, such as prioritizing same-day LARC insertion and removals, and eliminating required post-LARC follow-up visits. Providers noted virtual visits enhanced privacy for some and compromised privacy for others. All participants observed that continuation of telemedicine contraceptive services would depend on equitable reimbursement for telehealth services. Conclusions: Telemedicine contraception services can enhance access for patients who face barriers to care beyond those created by the pandemic. However, reimbursement parity for providers, patient-centered flexibility regarding telemedicine options, and measures to ensure access to all methods are important for long-term success.

0

PMC9748835

NO-CC CODE

2022-12-15 23:22:47

no

Contraception. 2021 Oct 6; 104(4):467

utf-8

Contraception

10.1016/j.contraception.2021.07.083

oa_other

==== Front Contraception Contraception Contraception 0010-7824 1879-0518 Published by Elsevier Inc. S0010-7824(21)00313-9 10.1016/j.contraception.2021.07.079 Article POSTER ABSTRACTS P61 YOUNG PEOPLE'S ACCESS TO CONTRACEPTIVE SERVICES THROUGH TELEMEDICINE: INEQUITIES BY FOOD AND HOUSING INSECURITY Yarger J Hopkins K Elmes S Rossetto I Melena S De La White K Harper CC Department of Obstetrics, Gynecology, and Reproductive Sciences, University of California, San Francisco, CA, US 6 9 2021 10 2021 6 9 2021 104 4 466466 Copyright © 2021 Published by Elsevier Inc. 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcObjectives: To examine disparities in access to telemedicine visits for contraception during the COVID-19 pandemic by young people's experiences of basic needs insecurity. Methods: We collected data from May 2020 to March 2021 from people at risk of pregnancy aged 18–28 in an ongoing study of community college students in California and Texas (n=1,352). Multivariate logistic regression analyses, adjusted for clustering by site, were conducted to examine differences in access to contraceptive services through telemedicine by food and housing insecurity, controlling for age, race/ethnicity, health insurance, and other key sociodemographic characteristics. Results: Only 9% of participants received their birth control method through a phone or video visit. One quarter (24%) reported it would be difficult to have a telemedicine visit for birth control. Perceived barriers to telemedicine included lacking privacy at home (42%), not knowing how to do a telemedicine visit (25%), lacking a device or Internet connection (23%), clinics not offering telemedicine (16%), and insurance not covering telemedicine (13%). Half (51%) stated they needed to get their method in person, while 36% would not feel comfortable using telemedicine, and 78% preferred in-person visits. Participants experiencing food insecurity (adjustedOR [aOR], 2.14; 95% confidence interval [CI], 1.59–2.88) and housing insecurity (aOR, 1.66; 95% CI, 1.16–2.38) were significantly more likely to report that they would have difficulty using telemedicine for birth control. Conclusions: Efforts are needed to remove barriers to telemedicine, particularly for young people facing basic needs insecurity, and to ensure that safe, high-quality in-person contraceptive services also remain accessible.

0

PMC9748836

NO-CC CODE

2022-12-15 23:22:47

no

Contraception. 2021 Oct 6; 104(4):466

utf-8

Contraception

10.1016/j.contraception.2021.07.079

oa_other

==== Front Chem Biol Interact Chem Biol Interact Chemico-Biological Interactions 0009-2797 1872-7786 Elsevier B.V. S0009-2797(21)00064-8 10.1016/j.cbi.2021.109428 109428 Research Paper Low risk of the TMPRSS2 inhibitor camostat mesylate and its metabolite GBPA to act as perpetrators of drug-drug interactions Weiss Johanna ∗ Bajraktari-Sylejmani Gzona Haefeli Walter Emil Department of Clinical Pharmacology and Pharmacoepidemiology, University of Heidelberg, Im Neuenheimer Feld 410, 69120, Heidelberg, Germany ∗ Corresponding author. University Hospital Heidelberg, Department of Clinical Pharmacology and Pharmacoepidemiology, Im Neuenheimer Feld 410, 69120, Heidelberg, Germany. 27 2 2021 1 4 2021 27 2 2021 338 109428109428 18 1 2021 23 2 2021 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Camostat mesylate, a potent inhibitor of the human transmembrane protease, serine 2 (TMPRSS2), is currently under investigation for its effectiveness in COVID-19 patients. For its safe application, the risks of camostat mesylate to induce pharmacokinetic drug-drug interactions with co-administered drugs should be known. We therefore tested in vitro the potential inhibition of important efflux (P-glycoprotein (P-gp, ABCB1), breast cancer resistance protein (BCRP, ABCG2)), and uptake transporters (organic anion transporting polypeptides OATP1B1, OATP1B3, OATP2B1) by camostat mesylate and its active metabolite 4-(4-guanidinobenzoyloxy)phenylacetic acid (GBPA). Transporter inhibition was evaluated using fluorescent probe substrates in transporter over-expressing cell lines and compared to the respective parental cell lines. Moreover, possible mRNA induction of pharmacokinetically relevant genes regulated by the nuclear pregnane X receptor (PXR) and aryl hydrocarbon receptor (AhR) was analysed in LS180 cells by quantitative real-time PCR. The results of our study for the first time demonstrated that camostat mesylate and GBPA do not relevantly inhibit P-gp, BCRP, OATP1B1 or OATP1B3. Only OATP2B1 was profoundly inhibited by GBPA with an IC50 of 11 μM. Induction experiments in LS180 cells excluded induction of PXR-regulated genes such as cytochrome P450 3A4 (CYP3A4) and ABCB1 and AhR-regulated genes such as CYP1A1 and CYP1A2 by camostat mesylate and GBPA. Together with the summary of product characteristics of camostat mesylate indicating no inhibition of CYP1A2, 2C9, 2C19, 2D6, and 3A4 in vitro, our data suggest a low potential of camostat mesylate to act as a perpetrator in pharmacokinetic drug-drug interactions. Only inhibition of OATP2B1 by GBPA warrants further investigation. Keywords Camostat mesylate GBPA Drug-drug interactions Drug transporters SARS-CoV-2 COVID-19 Abbreviations AhR, aryl hydrocarbon receptor BCRP, breast cancer resistance protein calcein-AM, calcein acetoxymethylester COVID-19, corona virus disease 2019 CYP, cytochrome P450 DBF, 4′,5′-dibromofluorescein FTC, fumitremorgin C GBA, 4-guanidinobenzoic acid GBPA, 4-(4-guanidinobenzoyloxy) phenylacetic acid GU, glucuronidase β OATP, organic anion transporting polypeptide P-gp, P-glycoprotein PXR, pregnane X receptor SARS-CoV-2, SARS-coronavirus 2 SPC, summary of product characteristics TMPRSS2, transmembrane protease, serine 2 ==== Body pmc1 Introduction Camostat mesylate, a potent inhibitor of the human transmembrane protease, serine 2 (TMPRSS2), is approved in Japan for the treatment of chronic pancreatitis and postoperative reflux esophagitis. Because SARS-coronavirus 2 (SARS-CoV-2) needs the TMPRSS2 for its spike protein priming, this protease is essential for the cellular entry of this virus [1] and recent evidence indicates that camostat mesylate effectively inhibits SARS-CoV-2 entry into lung cells [1]. Therefore, several clinical studies are ongoing testing its effectiveness in corona virus disease 2019 (COVID-19) patients. In humans and animals, camostat mesylate is rapidly metabolised by carboxyesterases to its metabolite, 4-(4-guanidinobenzoyloxy) phenylacetic acid (GBPA, FOY-251), which has a terminal half-life of 0.75–1.4 h in humans and is further metabolised to 4-guanidinobenzoic acid (GBA) [[2], [3], [4], [5]] (Fig. 1 ). Whereas GBA has a low inhibition potency for TMPRSS2, GBPA is pharmacologically active but with an about 10-fold lower potency than the parent compound camostat.Fig. 1 Metabolism of camostat mesylate. Chemical structures were plotted with ChemDraw Professional, version 20.0.0.41. Fig. 1 Efficacy and safety of drugs can be influenced by drug–drug interactions. In pharmacokinetic interactions, the perpetrator drug leads to changes in absorption, metabolism, distribution, or excretion of the victim drug causing toxic side effects or non-response. Most pharmacokinetic drug-drug interactions can be attributed to inhibition or induction of drug-metabolising enzymes and drug transporters involved in the absorption, distribution, and clearance of drugs. Up to now, nearly no published information exists whether camostat acts as a perpetrator in drug-drug interactions. The summary of product characteristics (SPC) of FOIPAN® only states that camostat and GBPA do not inhibit cytochrome P450 (CYP) 1A2, 2C9, 2C19, 2D6, and 3A4 in vitro [6] excluding one of the important mechanisms of drug-drug interactions. However, apart from CYPs and other metabolising enzymes, drug uptake or efflux transporters can also have substantial impact on the pharmacokinetics of drugs. Among them are efflux transporters such as P-glycoprotein (P-gp, ABCB1) and breast cancer resistance protein (BCRP, ABCG2) and uptake transporters such as the organic anion transporting polypeptides (OATP) 1B1, 1B3, and 2B1 [[7], [8], [9], [10], [11]]. Thus far, no data are published on possible inhibiting effects of camostat or GBPA on drug transporters. Moreover, there is no information on possible inducing effects of these compounds on enzymes or transporters, which may lead to decreased exposure of co-administered drugs and thus to non-response. We therefore evaluated the inhibitory potential of camostat and GBPA on selected pharmacokinetically important drug transporters as well as their inducing potential on exemplary genes regulated by the pregnane X receptor (PXR) and the aryl-hydrocarbon receptor (AhR): CYP1A1, CYP1A2, CYP3A4, ABCB1, and ABCG2. 2 Material and methods 2.1 Materials Cell culture medium M199, foetal calf serum (FCS), supplements, Hank's buffered salt solution (HBSS), phosphate buffered saline (PBS), GenElute™ Mammalian Total RNA Miniprep Kit, Cytotoxicity Detection Kit (LDH), 4′,5′-dibromofluorescein (DBF), tetracycline, rifampicin, naringin, omeprazole, and camostat mesylate were obtained from Sigma-Aldrich (Taufkirchen, Germany). DMEM was purchased from PAN Biotech (Aidenbach, Germany). GBPA mesylate was purchased from Santa Cruz Biotechnology (Dallas, TX, USA). Dimethyl sulfoxide (DMSO), Triton X-100, G418, and crystal violet were from AppliChem (Darmstadt, Germany). Vincristine was purchased from Biotrend (Cologne, Germany). Fumitremorgin C (FTC) was obtained from Merck Millipore (Darmstadt, Germany). Calcein acetoxymethylester (calcein-AM) was obtained from Invitrogen (Karlsruhe, Germany), pheophorbide A from Frontier Scientific Europe (Carnforth, UK), LY335979 (zosuquidar) from Toronto Research Chemicals (Toronto, Canada), and 8-fluorescein-cAMP from BIOLOG Life Science Institute (Bremen, Germany). The RevertAid™ H Minus First Strand cDNA Synthesis Kit and the Absolute QPCR SYBR Green Mix were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Primers were synthesised by Eurofins MWG Operon (Ebersberg, Germany). 2.2 Stock solutions Stock solution of camostat mesylate was prepared in aqua bidest (50 mM) and of GBPA mesylate in DMSO (10 mM). Stock solutions of all other compounds were prepared in DMSO. All stock solutions were stored in aliquots at −20 °C. 2.3 Cytotoxicity assay Because cytotoxic effects can influence transporter inhibition assays, we tested for possible cytotoxic effects of camostat mesylate and GBPA in all cell lines using the Cytotoxicity Detection Kit according to the manufacturer's instructions. Neither camostat nor GBPA revealed any cytotoxic effects up to 100 μM. Moreover, in the flow cytometry assays, no shift of the cell populations in the forward versus side scatter occurred also confirming the absence of any short-term cytotoxic effects. 2.4 P-gp inhibition assay (calcein assay) For assessing whether camostat and GBPA inhibit P-gp, a calcein assay was performed as validated and described in detail previously [12]. As a cell system, the L-MDR1 cell line over-expressing human P-gp [13] (kindly provided by A. H. Schinkel (The Netherlands Cancer Institute, Amsterdam, The Netherlands)) and the corresponding parental cell line LLC-PK1 were used. The two cell lines were cultured under standard cell culture conditions with medium M199 supplemented with 10% FCS, 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin sulphate. To maintain P-gp expression, the medium used for L-MDR1 cells was amended with 0.64 μM vincristine. One day before the calcein assay, both cell lines were fed with vincristine-free culture medium. Each concentration of camostat mesylate and GBPA mesylate (0.05–100 μM) was tested in octuplet and 1 μM LY335979 was used as a positive control. The experiment was performed in quadruplicate. 2.5 BCRP inhibition assay (flow cytometric pheophorbide A efflux assay) For testing BCRP inhibition, the human BCRP-overexpressing cell line MDCKII-BCRP [14] (kindly provided by A. H. Schinkel (The Netherlands Cancer Institute, Amsterdam, The Netherlands)) and the corresponding parental cell line MDCKII were used. Cells were cultured under standard cell culture conditions in DMEM containing 10% FCS, 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin sulphate. The flow cytometric BCRP inhibition assay was conducted as validated and described previously [15] using the fluorescent pheophorbide A as a specific BCRP substrate. In each sample, 30,000 cells were counted. Living cells were gated in the forward versus side scatter. The inhibitory effect of test compounds was quantified by calculating the ratio between the median fluorescence with and without inhibitor and normalised to the effects in the parental cell line. For camostat and GBPA 0.5–100 μM were tested and as a positive control, 10 μM FTC was used. The experiment was performed in triplicate. 2.6 OATP1B1 and OATP1B3 inhibition assay (flow cytometric 8-fluorescein-cAMP uptake assay) For measuring OATP1B1 and OATP1B3 inhibition, the human embryonic kidney cell line HEK293 stably transfected with OATP1B1 (HEK-OATP1B1), OATP1B3 (HEK-OATP1B3), or the empty control vector (HEK293-VC G418) (kindly provided by D. Keppler (German Cancer Research Centre, Heidelberg, Germany)) were used [16,17]. Cells were cultured under standard cell culture conditions with DMEM supplemented with 10% FCS, 2 mM glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin sulphate, and 800 μg/ml G418. Inhibition of OATP1B1 and OATP1B3 was analysed by measuring the uptake of the fluorescent substrate 8-fluorescein-cAMP as described previously [18]. In each sample, 30,000 cells were counted. Cell debris was eliminated by gating the viable cells in the forward versus side scatter. For determination of the inhibitor effects, the ratio between the median fluorescence of intracellular 8-fluorescein-cAMP with and without inhibitor was calculated and normalised to the mock transfected control cell line. For camostat and GBPA 0.5–100 μM were tested and as a positive control 20 μM rifampicin was used. The experiments were conducted in triplicate. 2.7 OATP2B1 inhibition assay (flow cytometric DBF uptake assay) HEK293 cells overexpressing OATP2B1 in presence of tetracycline [19] (Grosser et al., 2015) (kindly supplied by G. Grosser and J. Geyer (University of Gieβen, Germany)) were used for measuring OATP2B1 inhibition as described previously [20]. Cells were cultured under standard cell culture conditions with DMEM/Ham's F12 medium supplemented with 10% FCS, 4 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin sulphate. To generate OATP2B1 overexpression, cells were treated for 72 h with 1 μg/ml tetracycline before the assay. In each sample, 30,000 cells were counted. Cell debris was eliminated by gating the viable cells in the forward versus side scatter. For determination of the inhibitor effects, the ratio between the median fluorescence of intracellular DBF with and without inhibitor was calculated and normalised to the control cell line. For camostat and GBPA 0.5–100 μM were tested and as a positive control 1 mM naringin was used. The experiments were conducted in triplicate. 2.8 Induction assay For testing possible inducing effects of camostat and GBPA, the human colon adenocarcinoma cell line LS180 was used. LS180 cells are a well-established model for investigating induction mediated by PXR and AhR [[21], [22], [23], [24], [25], [26], [27]]. Cells were cultured under standard cell culture conditions in DMEM supplemented with 10% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin sulphate, 0.1 mM non-essential amino acids, and 2 mM glutamine. Antiproliferative effects can influence the results of induction assays. Thus, growth inhibition by camostat mesylate was investigated in LS180 before starting the induction assay. Proliferation was quantified by crystal violet staining and the assays were conducted as described previously [28]. Camostat mesylate was tested from 0.01 to 100 μM with each concentration tested in octuplet and the experiment was performed in quadruplicate. Camostat mesylate did not show any antiproliferative effects up to the maximum concentration tested (100 μM), which was therefore also the maximum concentration tested in the induction assay. In FCS-containing medium, similar to the in vivo situation, camostat mesylate is rapidly metabolised to GBPA and further to GBA with a half-live of about 140 min [29]. Thus, we did not test GBPA separately in our induction assay, because during our 4-day incubation, the metabolites were present anyway. To test induction, LS180 cells were treated for four days with camostat mesylate (5, 10, 50, 100 μM), omeprazole (150 μM, positive control for AhR-driven genes), rifampicin (20 μM, positive control for PXR-driven genes), or with medium only as a negative control. All media were adjusted to 0.1% DMSO. Immediately after harvesting the cells, RNA was extracted using the GenElute™ Mammalian Total RNA Miniprep Kit, purity and concentration measured photometrically and stored at −80 °C until processing. The experiment was conducted in quintuplicate. RNA was reverse transcribed to cDNA with the RevertAid™ H Minus First Strand cDNA Synthesis Kit according to the manufacturer's instructions. mRNA expression was quantified by real-time RT-PCR with the LightCycler® 480 (Roche Applied Science, Mannheim, Germany) as described previously [30]. Primer sequences were also published previously: ABCB1 and ABCG2 [30], CYP1A1 [31], CYP1A2 [32], CYP3A4 [33], and glucuronidase β (GU) as housekeeping gene [34]. PCR amplification was carried out in 20 μl reaction volume containing 5 μl 1:10 diluted cDNA, 1x Absolute QPCR SYBR Green Mix, and 0.15–0.5 μM primers. The most suitable housekeeping gene for normalisations was identified using geNorm (version 3.4, Center for Medical Genetics, Ghent, Belgium), which determines the most stable reference gene from a set of tested genes in a given cDNA sample panel [35]. Among the 7 reference genes tested, GU proved to be the most stable in LS180 cells under our experimental conditions. Data were evaluated via calibrator-normalised relative quantification with efficiency correction using the LightCycler® 480 software version 1.5.1.62 (Roche Applied Science). Results were expressed as the target/reference ratio divided by the target/reference ratio of the calibrator. The results are therefore corrected for variance caused by detection and sample inhomogeneities. All samples were amplified in technical duplicate and the mean was used for further calculation. 2.9 Statistical analysis Non-linear regression curves were calculated with GraphPad Prism version 9.0.0 (GraphPad Software Inc., La Jolla, CA, USA) using the four-parameter fit (sigmoidal dose-response curves with variable slope). Differences between mRNA expressions were tested using ANOVA with Dunnett's post hoc test using InStat version 3.06 (GraphPad Software Inc., La Jolla, CA, USA). A p-value < 0.05 was considered significant. 3 Results 3.1 Inhibition of drug efflux transporters P-gp and BCRP by camostat and GBPA Camostat neither inhibited P-gp nor BCRP up to 100 μM (data not shown). In contrast, its metabolite GBPA slightly increased intracellular calcein fluorescence at concentrations above 10 μM in P-gp-overexpressing L-MDR1 cells, but not in the parental cell line LLC-PK1 indicating very weak P-gp inhibition (Fig. 2 ). Similarly, GBPA had a small inhibitory effect on BCRP at concentrations above 10 μM. However, compared to the respective positive controls (LY335979 and FTC) the effect was very small (Fig. 2, Fig. 3 ).Fig. 2 Concentration-dependent effect of GBPA on intracellular calcein fluorescence in L-MDR 1 indicating P-gp inhibition. The curve depicts the results of 3 experiments with each concentration tested in octuplet and data are expressed as mean ± S.E.M. Due to the small errors, the error bars are not visible. Fig. 2 Fig. 3 Concentration-dependent effect of GBPA on the BCRP activity. For determination of the inhibitor effects, the ratio between the median fluorescence of intracellular pheophorbide A with and without inhibitor was calculated in BCRP-overexpressing cells and normalised to the control cell line. Each data point depicts the results of 3 experiments with 30,000 cells each and is expressed as mean ± S.E.M. Due to the small errors, the error bars are not visible. Fig. 3 3.2 Inhibition of OATPs by camostat and GBPA Camostat revealed only minor effects on OATP activities: it did not inhibit OATP1B3 and OATP2B1 (Fig. 4, Fig. 5 ) and only slightly inhibited OATP1B1 at concentrations ≥ 50 μM (Fig. 4A). GBPA also had no relevant effects on OATP1B3 activity (Fig. 4B), but inhibited OATP1B1 about 50% at 100 μM and potently inhibited OATP2B1 with an IC50 of 11.1 ± 3.3 μM (Fig. 5).Fig. 4 Concentration-dependent effect of camostat mesylate and GBPA on the activity of OATP1B1 and OATP1B3. For determination of the inhibitor effects, the ratio between the median fluorescence of intracellular 8-fluorescein-cAMP with and without inhibitor was calculated in OATP-overexpressing cells and normalised to the control cell line. Each data point depicts the results of 3–4 experiments with 30,000 cells each and is expressed as mean ± S.E.M. Fig. 4 Fig. 5 Concentration-dependent effect of camostat mesylate and GBPA on the activity of OATP2B1. For determination of the inhibitor effects, the ratio between the median fluorescence of intracellular DBF with and without inhibitor was calculated in OATP2B1-overexpressing cells and normalised to the control cell line. Each data point depicts the results of 3 experiments with 30,000 cells each and is expressed as mean ± S.E.M. Fig. 5 3.3 Induction of PXR- and AhR-regulated genes by camostat and GBPA In contrast to substantial induction by the respective positive controls (rifampicin, omeprazole), incubation of LS180 cells for 4 days with camostat mesylate and thus also GBPA had no significant effect on the mRNA expression of CYP1A1, CYP1A2, CYP3A4, ABCB1, and ABCG2 clearly excluding the induction of genes regulated by PXR and/or AhR (Fig. 6 ).Fig. 6 Effects of camostat mesylate on mRNA expression in LS180 cells compared to untreated medium control after 4 days of incubation. Rifampicin (20 μM) served as a positive control for PXR-driven genes (CYP3A4, ABCB1, ABCG2) and omeprazole (150 μM) served as a positive control for AhR-driven genes (ABCG2, CYP1A1, CYP1A2). Expression data were normalised to the housekeeping gene GU. Data are expressed as mRNA changes ±SEM for n = 5 biological replicates. Data were analysed using ANOVA with Dunnett's post hoc test compared to the medium control. **p < 0.01. Fig. 6 4 Discussion The pandemic caused by SARS-CoV-2 has triggered a feverish search for effective pharmacological treatment of infected patients. The development and approval of new drugs is a very time-consuming process, which can be abbreviated by repurposing already licensed drugs. One of those approved drugs proposed as possible treatment option for COVID-19 patients is camostat mesylate. This serine protease inhibitor, marketed in Japan as FOIPAN®, has already been clinically used for over two decades to treat pancreatitis and postoperative reflux esophagitis due to its potent inactivation of trypsin preventing auto-digestion [36]. Camostat mesylate also potently inhibits the transmembrane serine protease TMPRSS2, on which SARS-CoV-2 critically depends for cellular entry and subsequent virus spread in the host [1,29]. It is therefore currently tested for its efficacy in COVID-19 patients. Beyond its efficacy, possible drug-drug interactions provoked by camostat are also of concern, because they could harm the particularly vulnerable COVID-19 patients. So far, the possible perpetrator characteristics of camostat mesylate in pharmacokinetic drug-drug interactions have not been investigated in detail. Our study aimed to close this knowledge gap. In humans and other mammals, after oral intake camostat mesylate itself does not reach systemic circulation in measurable concentrations, because it is rapidly metabolised during intestinal absorption to GBPA and GBA [[2], [3], [4], [5]]. Thus, although GBPA is about 10-fold less potent than the parent compound, inhibition of TMPRSS2 in humans and thus its possible anti-SARS-CoV-2 activity has to be attributed to this metabolite [29]. For the same reason, systemic drug-drug interactions can only be provoked by GBPA, which reaches maximum plasma concentrations of about 87 ng/ml (= 0.3 μM) after oral intake of 200 mg camostat mesylate [6]. In contrast, intestinal drug-drug interactions might also be provoked by both, the short-lived camostat itself and by its metabolite GBPA, which can reach concentrations in the intestine up to 1600 μM after oral intake of 200 mg according to the formula published by Zhang and co-workers [37]. Conclusive earlier data demonstrated that in FCS-containing medium camostat mesylate is rapidly metabolised to GBPA with a half-live of about 140 min [29]. Thus, we assume that in our induction assays, in which LS180 cells are incubated with camostat mesylate in FCS-containing medium, we also tested the influence of GBPA on the mRNA expression of several drug metabolising genes and drug transporters. Our data clearly demonstrate that camostat mesylate/GBPA did not induce CYP3A4 and ABCB1 (PXR-regulated), ABCG2 (regulated by PXR and AhR), or CYP1A1 and CYP1A2 (AhR-regulated) genes. We therefore conclude that camostat will not act as a perpetrator in drug-drug interactions based on the induction of drug metabolising enzymes or drug transporters regulated by PXR or AhR. In our inhibition assays, we tested both compounds individually, because these were short-term experiments and some of the buffers used did not contain FCS excluding the degradation of camostat mesylate. Our results demonstrated that neither camostat mesylate nor its metabolite GBPA are inhibitors of the efflux transporters P-gp and BCRP. Whereas camostat mesylate had no relevant effects on the uptake transporters OATP1B1, 1B3, and 2B1, GBPA exerted different effects on these OATPs: No pronounced inhibition of OATP1B3, weak inhibition of OATP1B1, and potent inhibition of OATP2B1 (IC50 about 11 μM). Although the small effect on OATP1B1 is most likely not relevant in vivo, because systemic concentrations of GBPA are too low for inhibition of this liver-specific uptake transporter, the potency for intestinal OATP2B1 inhibition is clearly high enough to be of clinical relevance. Inhibition of this uptake transporter represents a typical feature of several citrus flavonoids [[38], [39], [40]] and can lead to decreased bioavailability of respective substrates as already postulated e.g. for aliskiren [41,42], celiprolol [43], and rosuvastatin [44]. Whether therapy with camostat mesylate substantially influences the pharmacokinetics of OATP2B1 substrates should be investigated in clinical studies. 5 Conclusions In conclusion, the capability of the potential COVID-19 therapeutic drug camostat mesylate to act as a perpetrator in pharmacokinetic drug-drug interactions appears to be low. The SPC of camostat mesylate (FOIPAN®) states that camostat mesylate and GBPA do not inhibit CYP1A2, 2C9, 2C19, 2D6, and 3A4 in vitro. Our data for the first time demonstrate that these compounds also do not relevantly inhibit P-gp, BCRP, OATP1B1, and OATP1B3 and do not induce drug transporters or drug metabolising enzymes regulated by PXR or AhR in vitro. Only inhibition of OATP2B1 by GBPA might be of clinical relevance and should be further investigated. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgement GBS was supported in part by the Physician Scientist program of the Faculty of Medicine of Heidelberg University, Germany. The authors thank Corina Mueller, Stephanie Rosenzweig, and Jutta Kocher for excellent technical assistance, Alfred Schinkel for providing the cell lines L-MDR1 and MDCKII-BCRP, Dietrich Keppler for providing the cell lines HEK-OATP1B1/1B3 and Gary Grosser and Joachim Geyer for providing the cell line HEK-OATP2B1. ==== Refs References 1 Hoffmann M. Kleine-Weber H. Schroeder S. Krüger N. Herrler T. Erichsen S. Schiergens T.S. Herrler G. Wu N.H. Nitsche A. Müller M.A. Drosten C. Pöhlmann S. SARS-CoV-2 cell entry depends on ACE2 and TMPRSS2 and is blocked by a clinically proven protease inhibitor Cell 181 2020 271 280 10.1016/j.cell.2020.02.052 e8 32142651 2 Beckh K. Goke B. Muller R. Arnold R. Elimination of the low-molecular weight proteinase inhibitor camostate (FOY 305) and its degradation products by the rat liver Res. Exp. Med. 187 1987 401 406 10.1007/BF01852177 3 Beckh K. Weidenbach H. Weidenbach F. Muller R. Adler G. Hepatic and pancreatic metabolism and biliary excretion of the protease inhibitor camostat mesilate Int. J. Pancreatol. 10 1991 197 205 1787334 4 Ohki S. Nishiyama H. Ozeki K. Ito H. Hirata F. Studies on absorption, distribution, metabolism and excretion of [14C] FOY-305 Gendai-Iryo 12 1980 71 82 5 Midgley I. Hood A.J. Proctor P. Chasseaud L.F. Irons S.R. Cheng K.N. Brindley C.J. Bonn R. Metabolic fate of 14C-camostat mesylate in man, rat and dog after intravenous administration Xenobiotica 24 1994 79 92 10.3109/00498259409043223 8165824 6 Ono Pharmaceutical Co., Ltd.: SPC Foipan: www.shijiebiaopin.net/upload/product/201272318373223.PDF. 7 Nakanishi T. Tamai I. Interaction of drug or food with drug transporters in intestine and liver Curr. Drug Metabol. 16 2015 753 764 10.2174/138920021609151201113537 8 König J. Müller F. Fromm M.F. Transporters and drug-drug interactions: important determinants of drug disposition and effects Pharmacol. Rev. 65 2013 944 966 10.1124/pr.113.007518 23686349 9 Liu C.X. Yi X.L. Fan H.R. Wu W.D. Zhang X. Xiao X.F. He X. Effects of drug transporters on pharmacological responses and safety Curr. Drug Metabol. 16 2015 732 752 10.2174/138920021609151201112629 10 Shitara Y. Clinical importance of OATP1B1 and OATP1B3 in drug-drug interactions Drug Metabol. Pharmacokinet. 26 2011 220 227 10.2133/dmpk.DMPK-10-RV-094 11 Gessner A. König J. Fromm M.F. Clinical aspects of transporter-mediated drug-drug interactions Clin. Pharmacol. Ther. 105 2019 1386 1394 10.1002/cpt.13 30648735 12 Weiss J. Dormann S.M. Martin-Facklam M. Kerpen C.J. Ketabi-Kiyanvash N. Haefeli W.E. Inhibition of P-glycoprotein by newer antidepressants J. Pharmacol. Exp. Therapeut. 305 2003 197 204 10.1124/jpet.102.046532 13 Schinkel A.H. Wagenaar E. Mol C.A. van Deemter L. P-glycoprotein in the blood brain barrier of mice influences the brain penetration and pharmacological activity of many drugs J. Clin. Invest. 97 1996 2517e24 10.1172/JCI118699 8647944 14 Pavek P. Merino G. Wagenaar E. Bolscher E. Novotna M. Jonker J.W. Schinkel A.H. Human breast cancer resistance protein: interactions with steroid drugs, hormones, the dietary carcinogen 2-amino-1-methyl-6-phenylimidazo(4,5-b)pyridine, and transport of cimetidine J. Pharmacol. Exp. Therapeut. 312 2005 144 152 10.1124/jpet.104.073916 15 Weiss J. Rose J. Storch C.H. Ketabi-Kiyanvash N. Sauer A. Haefeli W.E. Efferth T. Modulation of human BCRP (ABCG2) activity by anti-HIV drugs J. Antimicrob. Chemother. 59 2007 238 245 10.1093/jac/dkl474 17202245 16 König J J. Cui Y. Nies A.T. Keppler D. Localization and genomic organization of a new hepatocellular organic anion transporting polypeptide J. Biol. Chem. 275 2000 23161 23168 10.1074/jbc.M001448200 10779507 17 König J. Cui Y. Nies A.T. Keppler D. A novel human organic anion transporting polypeptide localized to the basolateral hepatocyte membrane Am. J. Physiol. Gastrointest. Liver Physiol. 278 2000 G156 G164 10.1152/ajpgi.2000.278.1.G156 10644574 18 Weiss J. Theile D. Spalwisz A. Burhenne J. Riedel K.-D. Haefeli W.E. Influence of sildenafil and tadalafil on the enzyme- and transporter-inducing effects of bosentan and ambrisentan in LS180 cells Biochem. Pharmacol. 85 2013 265 273 10.1016/j.bcp.2012.11.020 23219525 19 Grosser G. Döring B. Uegele B. Geyer J. Kulling S.E. Soukup S.T. Transport of the soy isoflavone diadzein and its conjugative metabolites by the carriers SOAT, NTCP, OAT4, and OATP2B1 Arch. Toxicol. 89 2015 2253 2263 10.1007/s00204-014-1379-3 25319728 20 Bajraktari G. Weiss J. The aglycone diosmetin has the higher perpetrator drug-drug interaction potential compared to the parent flavone diosmin Funct. Foods 67 2020 103842 10.1016/j.jff.2020.103842 21 Gupta A. Mugundu G.M. Desai P.B. Thummel K.E. Unadkat J.D. Intestinal human colon adenocarcinoma cell line LS180 is an excellent model to study pregnane X receptor, but not constitutive androstane receptor, mediated CYP3A4 and multidrug resistance transporter 1 induction: studies with anti-human immunodeficiency virus protease inhibitors Drug Metab. Dispos. 36 2008 1172 1180 10.1124/dmd.107.018689 18332086 22 Harmsen S. Koster A.S. Beijnen J.H. Schellens J.H. Meijerman I. Comparison of two immortalized human cell lines to study nuclear receptor-mediated CYP3A4 induction Drug Metab. Dispos. 36 2008 1166 1171 10.1124/dmd.107.017335 18347084 23 Harper P.A. Prokipcak R.D. Bush L.E. Golas C.L. Okey A.B. Detection and characterization of the Ah receptor for 2,3,7,8-tetrachlorodibenzo-p-dioxin in the human colon adenocarcinoma cell line LS180 Arch. Biochem. Biophys. 290 1991 27 36 10.1016/0003-9861(91)90587-9 1654865 24 Weiss J. Herzog M. Haefeli W.E. Differential modulation of the expression of important drug metabolising enzymes and transporters by endothelin-1 receptor antagonists ambrisentan and bosentan in vitro Eur. J. Pharmacol. 660 2011 298 304 10.1016/j.ejphar.2011.04.003 21501604 25 Brandin H. Viitanen E. Myrberg O. Arvidsson A.K. Effects of herbal medicinal products and food supplements on induction of CYP1A2, CYP3A4 and MDR1 in the human colon carcinoma cell line LS180 Phytother Res. 21 2007 239 244 10.1002/ptr.2057 17163579 26 Yamasaki D. Nakamura T. Okamura N. Kokudai M. Effects of acid and lactone forms of 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors on the induction of MDR1 expression and function in LS180 cells Eur. J. Pharmaceut. Sci. 37 2009 126 132 10.1016/j.ejps.2009.01.009 27 Li W. Harper P.A. Tang B.K. Okey A.B. Regulation of cytochrome P450 enzymes by aryl hydrocarbon receptor in human cells: CYP1A2 expression in the LS180 colon carcinoma cell line after treatment with 2,3,7,8-tetrachlorodibenzo-p-dioxin or 3-methylcholanthrene Biochem. Pharmacol. 56 1998 599 612 10.1016/s0006-2952(98)00208-1 9783729 28 Peters T. Lindenmaier H. Haefeli W.E. Weiss J. Interaction of the mitotic kinesin Eg5 inhibitor monastrol with P-glycoprotein Naunyn-Schmiedeberg’s Arch. Pharmacol. 372 2006 291 299 10.1007/s00210-005-0022-5 16365780 29 Hoffmann M. Hofmann-Winkler H. Smith J.C. Krüger N. Sørensen L.K. Søgaard O.S. Hasselstrøm J.B. Winkler M. Hempel T. Raich L. Olsson S. Yamazoe T. Yamatsuta K. Mizuno H. Ludwig S. Noé F. Sheltzer J.M. Kjolby M. Pöhlmann S. Camostat Mesylate Inhibits SARS-CoV-2 Activation by TMPRSS2-related Proteases and its Metabolite GBPA Exerts Antiviral Activity 2020 Aug 5 2020 10.1101/2020.08.05.237651 bioRxiv. 08.05.237651. Preprint 30 Albermann N. Schmitz-Winnenthal F.H. Z'graggen K. Volk C. Hoffmann M.M. Haefeli W.E. Weiss J. Expression of the drug transporters MDR1/ABCB1, MRP1/ABCC1, MRP2/ABCC2, BCRP/ABCG2, and PXR in peripheral blood mononuclear cells and their relationship with the expression in intestine and liver Biochem. Pharmacol. 70 2005 949 958 10.1016/j.bcp.2005.06.018 16054595 31 Dvorak Z. Vrzal R. Henklova P. Jancova P. Anzenbacherova E. Maurel P. Svecova L. Pavek P. Ehrmann J. Havlik R. Bednar P. Lemr K. Ulrichova J. JNK inhibitor SP600125 is a partial agonist of human aryl hydrocarbon receptor and induces CYP1A1 and CYP1A2 genes in primary human hepatocytes Biochem. Pharmacol. 5 2008 580 588 10.1016/j.bcp.2007.09.013 32 Ayed-Boussema I. Pascussi J.M. Maurel P. Bacha H. Hassen W. Zearalenone activates pregnane X receptor; constitutive androstane receptor and aryl hydrocarbon receptor and corresponding phase I target genes mRNA in primary cultures of human hepatocytes Environ. Toxicol. Pharmacol. 31 2011 79 87 10.1016/j.etap.2010.09.008 21787672 33 Cerveny L. Svecova L. Anzenbacherova E. Vrzal R. Staud F. Dvorak Z. Ulrichova J. Anzenbacher P. Pavek P. Valproic acid induces CYP3A4 and MDR1 gene expression by activation of constitutive androstane receptor and pregnane X receptor pathways Drug Metab. Dispos. 35 2007 1032 1041 10.1124/dmd.106.014456 17392393 34 Zisowsky J. Koegel S. Leyers S. Devarakonda K. Kassack M.U. Osmak M. Jaehde U. Biochem. Pharmacol. 73 2007 298 307 10.1016/j.bcp.2006.10.003 17097621 35 Vandesompele J. De Preter K. Pattyn F. Poppe B. Van Roy B.N. De Paepe A. Speleman F. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes Genome Biol. 3 2002 10.1186/gb-2002-3-7-research0034 RESEARCH0034 36 Breining P. Frølund A.L. Højen J.F. Gunst J.D. Staerke N.B. Saedder E. Cases-Thomas M. Little P. Nielsen L.P. Søgaard O.S. Kjolby M. Camostat mesylate against SARS-CoV-2 and COVID-19-Rationale, dosing and safety Basic Clin. Pharmacol. Toxicol. 2020 Nov 11 10.1111/bcpt.13533 (Online ahead of print) 37 Zhang L. Zhang Y.D. Strong J.M. Reynolds K.S. Huang S.M. A regulatory viewpoint on transporter-based drug interactions Xenobiotica 38 2008 709 724 10.1080/00498250802017715 18668428 38 McFeely S.J. Wu L. Ritchie T.K. Unadkat J. Organic anion transporting polypeptide 2B1 - more than a glass-full of drug interactions Pharmacol. Ther. 196 2019 204 215 10.1016/j.pharmthera.2018.12.009 30557631 39 Johnson E. Won C.S. Köck K. Paine M.F. Prioritizing pharmacokinetic drug interaction precipitants in natural products: application to OATP inhibitors in grapefruit juice Biopharm Drug Dispos. 3 2017 251 259 10.1002/bdd.2061 40 Bajraktari-Sylejmani G. Weiss J. Potential Risk of Food-Drug Interactions: citrus polymethoxyflavones and flavanones as inhibitors of the organic anion transporting polypeptides (OATP) 1B1, 1B3, and 2B1 Eur. J. Drug Metab. Pharmacokinet. 45 2020 809 815 10.1007/s13318-020-00634-4 32661908 41 Tapaninen T. Neuvonen P.J. Niemi M. Grapefruit juice greatly reduces the plasma concentrations of the OATP2B1 and CYP3A4 substrate aliskiren Clin. Pharmacol. Ther. 88 2010 339 342 10.1038/clpt.2010.101 20664534 42 Tapaninen T. Neuvonen P.J. Niemi M M. Orange and apple juice greatly reduce the plasma concentrations of the OATP2B1 substrate aliskiren Br. J. Clin. Pharmacol. 71 2017 718 726 10.1111/j.1365-2125.2010.03898.x 43 Lilja J.J. Juntti-Patinen L. Neuvonen P.J. Orange juice substantially reduces the bioavailability of t he beta-adrenergic-blocking agent celiprolol Clin. Pharmacol. Ther. 75 2004 184 190 10.1016/j.clpt.2003.11.002 15001969 44 Johnson M. Patel D. Matheny C. Ho M. Chen L. Ellens H. Inhibition of intestinal OATP2B1 by the calcium receptor antagonist ronacaleret results in a significant drug-drug interaction by causing a 2-fold decrease in exposure of rosuvastatin Drug Metab. Dispos. 45 2017 27 34 10.1124/dmd.116.072397 27737931

33647240

PMC9748837

NO-CC CODE

2022-12-15 23:22:47

no

Chem Biol Interact. 2021 Apr 1; 338:109428

utf-8

Chem Biol Interact

10.1016/j.cbi.2021.109428

oa_other

==== Front Cognition Cognition Cognition 0010-0277 1873-7838 Elsevier B.V. S0010-0277(21)00113-X 10.1016/j.cognition.2021.104694 104694 Article The possible effects of the COVID-19 pandemic on the contents and organization of autobiographical memory: A Transition-Theory perspective Brown Norman R. ⁎ University of Alberta, Canada ⁎ Corresponding author at: Department of Psychology, University of Alberta, Edmonton, AB T6G 2E9, Canada. 31 3 2021 7 2021 31 3 2021 212 104694104694 23 6 2020 17 3 2021 18 3 2021 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The COVID-19 Pandemic is unique in its near universal scope and in the way that it has changed our lives. These facts suggest that it might also be unique in its effects on memory. A framework outlined in this article, Transition Theory, is used to explicate the mnemonically relevant ways in which the onset of the Pandemic differs from other personal and collective transitions and how the Pandemic Period might differ from other personally-defined and historically-defined autobiographical periods. Transition Theory also provides the basis for several predictions. Specifically, it predicts (a) a COVID bump (an increase in availability of event memories at the outset of the Pandemic) followed by (b) a lockdown dip (a decrease in availability of event memories from lockdown periods compared to other stable periods). It also predicts that (c) people may consider the Pandemic an important chapter in their life stories, but only when there is little continuity between their pre-Pandemic and post-Pandemic lives. Time will tell whether these predictions pan out. However, it is not too soon to highlight those aspects of the COVID-19 Pandemic that are likely to shape our personal and collective memories of this very unusual historical period. Keywords COVID-19 Pandemic Autobiographical memory Transition theory Stability ratio Continuity ratio Stability profile COVID bump Lockdown dip Event components H-DAP ==== Body pmc There is the world B.C. — Before Corona — and the world A.C. — After Corona. Thomas Friedman, New York Times, March 17, 2020 The evidence of calamity is overwhelmingly of absence… Geoff Dyer, The New Yorker, April 13, 2020 At this point (December 2020), it is clear, at least intuitively clear, that the COVID-19 virus has created a “new normal”, a Pandemic Period. What is less clear is how well events from this period will be remembered, whether it will stand out as a major chapter in our lives, and whether its effect on memory will differ across regions, between groups or from one individual to another. It is too soon to address some of these issues empirically. However, given what we now know about autobiographical memory, it is possible to speculate about them in an informed manner. To this end, I first outline a framework, Transition Theory (Brown, 2016; Brown, Hansen, Lee, Vanderveen, & Conrad, 2012; Brown, Schweickart, & Svob, 2016), that explains how transitional events affect the contents and organization of autobiographical memory and then use this framework to discuss the ways in which the Pandemic appears to differ from other historically significant public events and to consider how the Pandemic is likely to affect our memories. 1 Transition Theory: an overview Transition Theory provides an account of autobiographical memory that takes the Event Component (EC), as its basic unit. ECs are those concrete, identifiable elements of our lives that can be captured in a verbal description of an event (Barsalou, 1988; Linton, 1986; Morton, Hammersley, & Bekerian, 1985; Norman & Bobrow, 1979). Familiar people, locations, activities and objects are all considered ECs. Transition Theory assumes that: (a) ECs are represented in memory as individual units, (b) each memorable experience is represented as a bound set of ECs and indexed by them (Barsalou, 1988; Brewin, Dalgleish, & Joseph, 1996; Conway, 2009; Morton et al., 1985; Norman & Bobrow, 1979; Rubin, Boals, & Berntsen, 2008; Shimamura, 2014), and (c) experiences also function as Hebbian learning trials, creating and strengthening associations between co-occurring and contiguous ECs (Hebb, 1949; McClelland, McNaughton, & O'Reilly, 1995; Munakata & Pfaffly, 2004; Nelson & Shiffrin, 2013; Smith & DeCoster, 2000). During stable periods, we spend much of our time engaged in mundane and repetitive activities (e.g., commuting, shopping, preparing meals, socializing; Kahneman, Krueger, Schkade, Schwarz, & Stone, 2004; White & Dolan, 2009). As a result, Hebbian learning processes give rise to robust, richly interconnected networks, which link frequently encountered ECs to one another. These EC networks provide the mnemonic foundation for lifetime periods, where a lifetime period is defined as a high-level memory structure that integrates period-specific knowledge (e.g., ECs) and subsumes thematically related event memories (Conway, 2005; Kubovy, 2020; Linton, 1986; Neisser, 1986; Thomsen, 2015). On this view, transitions mark the beginnings and ends of lifetime periods by causing the synchronized termination of one set of familiar ECs and/or by creating conditions that establish encounters with a new set of ECs. Over time and with repeated exposure, some of these new elements become (highly) familiar and hence come to define a new post-transitional period. The focus on ECs enables us to think about autobiographical memory in a quantitative manner. In particular, it is useful to consider two theoretical values: The Stability Ratio (Eq. (1)) and the Continuity Ratio (Eqs. (2a) and (2b)).(1) StabilityRatioT=#FamiliarECT#NovelECT+#FamiliarECT Underlying the Stability Ratio (Eq. 1) are three observations. First, at any point in time, some ECs are familiar (e.g., one's spouse, one's bicycle) and others are novel (e.g., a new colleague, a just-opened café); second, the mix of familiar and novel ECs fluctuates overtime; and third, there are times when this mix is dominated by familiar ECs and others when it is not. Obviously, EC familiarity is continuous. However for simplicity's sake, it is assumed that familiar ECs are those that have been encountered, say, 50 times or more and novel ECs are those that have been encountered, say, l0 times or less. Also, to keep things simple, each ratio is computed over some standard unit of time (e.g., a week, a month). The Stability Ratio at TimeT approaches 1.0 during stable lifetime periods, periods during which an individual spends most of his or her time following a well-established routine and hence encounters many familiar ECs and few novel ones. In contrast, the ratio approaches 0.0 immediately following major transitions (e.g., relocation to a foreign country), when everything seems new – when novel ECs are much more common than familiar ones.(2a) Continuity Ratio=1−#AbandonedECpre+#NewECpost#AbandonedECpre+#RetainedECpre+#NewECpost (2b) Continuity Ratio=#RetainedECpre#ECpre∪post Where:ECpre⋃post = ECpre ⋃ ECpost [PRE ⋃ POST – the set of all ECs from the Pre- and Post-transition periods] Retained ECpre = ECpre ⋂ ECpost [PRE ⋂ POST: the set of ECs present during both the pre- and post-transition periods] Abandoned ECpre = ECpre – Retained ECpre [PRE ~ POST: the set of ECs present ONLY during the pre-transition period] New ECpost = ECpost – Retained ECpre [POST~ PRE: the set of ECs present ONLY during the post-transition period] Periodpre < Periodpost The Continuity Ratio, a simplified version of Tversky's ratio model (Tversky, 1977, p. 333), reflects the degree to which life following some identified transition (i.e., Periodpost) is similar to life during some prior period (Periodpre). It does so by assessing the degree to which the ECs present during one period are also present during the other. The intuition here is that two periods will be experienced as very similar (a) when the two share many of the same ECs and (b) when neither includes many ECs that are unique to one period but not the other. When these conditions are not met, the two periods will be experienced as very different. More specifically, the Continuity Ratio recognizes that some familiar ECs from Periodpre will be retained across a transition (Retained ECpre), that others will not (Abandoned ECpre), and that the post-transitional period should eventually give rise to new familiar ECs (New ECpost). This ratio captures two aspects of our understanding of transitions (Brown et al., 2012). First, transitions are graded; hence, the ratio is large when a post-transitional period (Periodpost) is very similar to a pre-transitional period (Periodpre) i.e., when #Retained ECpre is relatively large and the #Abandoned ECpre and/or #New ECpost are relatively small. Conversely, this ratio is small when the two periods are very different, i.e., when #Retained ECpre is relatively small and #Abandoned ECpre and/or #New ECpost are relatively large. Second, transitions can affect lives through a process of addition (yielding a larger #New ECpost), through the process of subtraction (yielding a smaller #Retained ECpre, and a larger #Abandoned ECpre) or both. Note, although the ratio, as presented, is defined over adjacent periods, the Continuity Ratio can be computed between any two points in time. In principle, an individual's Stability Ratio can be plotted over time to produce a Stability Profile. Fig. 1 provides an example. This profile fragment begins with a period of stability (e.g., a post-doctoral position in the US), followed by a major transition that involves both a change of country and a new role (e.g., a professorship in Canada), which in turn leads to a new stable period. In this figure, this second stable period is punctuated twice, first by a vacation in Mexico, and then by the birth of a child. The changes in the shape (from square to circle) and color (red, dark blue, light blue) of the symbols represent different sets of familiar ECs and hence give an indication of the continuity between periods. This figure illustrates the fact that there is little continuity between the first and second stable periods (resulting in a relatively small Continuity Ratio), and that the birth of a child served as a transition but not a radical one (resulting in a relative large Continuity Ratio). The Mexican vacation provides an example of a brief interlude, defined as a span of time dominated by novel ECs but which has no lasting effect on the set of familiar ECs; when the vacation is over, things return to normal; here, the Continuity Ratio computed across the vacation would be at ceiling.Fig. 1 An example of a stability profile presenting changes in the Stability Ratio over time. Note: See text for explanation. Fig. 1 Broadly speaking, the temporal distribution of memorable personal events and the Stability Profile should be reverse images of one another. On the one hand, steep declines in the Stability Profile and long stretches of instability should give rise to memory bumps of one sort or another (e.g., the reminiscence bump, the upheaval bump, the immigration bump; Brown et al., 2016; Koppel & Berntsen, 2015; Shi & Brown, 2016). On the other, relatively few memories should be available during stable periods, i.e., stretches of time during which the Stability Ratio is high and remains steady. These points are developed further below. 2 The COVID-19 pandemic: implications for autobiographical memory 2.1 How the COVID transition differs from other transitions Important transitions typically bring to an abrupt end one way of life and usher in another (Brown, 2016; Gu, Tse, & Brown, 2017; Shi & Brown, 2016; Uzer & Brown, 2015). In other words, they produce: (a) a sharp decline and gradual increase in the Stability Ratio and (b) a small Continuity Ratio. This general pattern recognizes that life altering transitions typically involve an adjustment phase, during which people necessarily encounter many novel ECs (i.e., #Novel EC > > # Familiar EC). With time, however, some novel ECs become familiar, resulting in a large-scale replacement of ECs across the transition (i.e., #Abandoned ECPre ≈ #New ECPost; see Fig. 2 , Panel A). Consider, for example, relocation from one city to another. A transition of this sort would involve, at minimum, the replacement of one dwelling for another, one set of neighbors for and another, and one set of shops for another.Fig. 2 Venn diagrams representing a transition-by-replacement (Panel A) and a transition-by-omission (Panel B). Note. In each panel, the horizontally-lined region represents the set of Abandoned Event Components (ECs), the region marked by the square grids represents the set of Retained ECs, and the vertically-lined regions represent the set of New ECs. Fig. 2 One of the unusual features of the COVID Pandemic is that it appears to have changed our lives by narrowing them. Simply put, at least during lockdown – and by definition – the new normal differs from the old, largely in what we cannot do, where we cannot go, and whom we cannot meet. As a result, we are experiencing a transition-by-omission (i.e., #Abandoned ECPre > > #New ECPost; see Fig. 2, Panel B). A recent study by Heanoy, Shi, and Brown (2020) provides empirical support for the notion that the onset of the COVID Pandemic was experienced as a transition-by-omission. The study itself is based on a large (n = 1215) North American, web-based survey conducted in late March 2020, some two weeks after the World Health Organization declared COVID-19 a global pandemic. Of direct relevance were responses to a set of questions drawn from the Transitional Impact Scale (TIS, Svob, Brown, Reddon, Uzer, & Lee, 2014). These questions provided information about the degree of change and the type of change people experienced as a result of the COVID Pandemic. Two findings are of particular interest. First, the Cronbach's Alpha for this scale is usually in the 0.80 to 0.90 range (Svob et al., 2014; Uzer, 2020) indicating that major transitions (e.g., relocation) typically change many aspects of a person's life in a synchronized manner. In contrast, the Pandemic produced a Cronbach's Alpha of 0.60; this implies that some elements were changing and others were not. The second finding clarifies this point. Specifically, it turned out that only two of five items received moderately high scores; these concerned the activities people engaged in (mean = 3.95 on a 1 to 5 scale) and the locations where they spent their time (mean = 3.33). The other items, which concerned people, possessions and general material circumstances, produced ratings at or below the midpoint of the scale. One way to interpret these findings is to recognize that, at the outset of the Pandemic, many areas went into lockdown and many schools and businesses were moved online. As a result, people were not going where they had previously gone and were not doing what they had previously done. They were, however, still living in the same location where they had been before the Pandemic and spending time with a subset of people with whom they were familiar (e.g., family members). Thus, these data indicate that the fabric of daily life did change for many people, but not in a wholesale manner. From a Transition-Theory perspective, then, for many of us1 , the onset of the Pandemic can be characterized by a relatively shallow decline and then a moderately steep increase in the Stability Profile (i.e., a relatively brief and shallow adjustment period). The decline was caused by the introduction of a modest number of novel ECs (e.g., online video meetings, masks, hand sanitizer) in combination with a substantial decrease in the number of familiar ECs (e.g., in-person classes, restaurant meals, social gatherings). The Stability Ratio was expected to increase rapidly because the life under lockdown tends to be highly repetitive. As a result, newly encountered ECs should lose their novelty rapidly. As for the Continuity Ratio, it should be less than 1.0 – yes, the Pandemic has been disruptive – but considerably greater than the Continuity Ratio produced by a major life-changing event (e.g., relocation to a foreign country). This moderate degree of continuity reflects the fact that, for many of us, the Pandemic has introduced a relatively small number of new ECpost; it has caused us to abandon, at least temporally, a fairly large number of ECpre, but has not resulted in the large-scale replacement of one set of ECs for another. 2.2 Predictions: A COVID bump and a lockdown dip This situation has implications for both the contents and organization of autobiographical memory. First, consider how the Pandemic might affect the temporal distribution of memorable personal events. Prior research has demonstrated that memorable personal events tend to “pile up” around impactful transitions (Enz, Pillemer, & Johnson, 2016; Kurbat, Shevell, & Rips, 1998; Pillemer, Goldsmith, Panther, & White, 1988; Shi & Brown, 2016; Thomsen and Berntsen, 2005) and occur more often during unstable periods (e.g., wars and natural disasters) than during stable ones (Bohn & Habermas, 2016; Brown et al., 2016; Gu et al., 2017; Zebian & Brown, 2014). This happens because people in transition (or living in unstable times) have left their old routines and are taking part in many one-off events and having many first-time experiences (Robinson, 1992; Rubin, Rahhal, & Poon, 1998). We know that the episodic distinctiveness conferred by an unusual experience predicts memorability (Hunt, 2006; Hunt & Rawson, 2011; Wagenaar, 1986). Thus, it makes sense that more events are remembered, per unit time, from the transitional/adjustment periods that separate stable periods than from the stable periods themselves. Given that we have all had a number of COVID-related first-time experiences (e.g., online teaching, home-schooling, shopping trips involving cleaning routines, etc.), it seems likely that we will observe a COVID bump. In addition to predicting a COVID bump, there are reasons for predicting a lockdown dip – a decrease in availability of specific event memories for a lockdown period, compared to other stable periods (Uzer & Brown, 2015). First, during lockdowns, people are restricted in their activities and movements, and as a result, they have fewer opportunities to do “interesting” things. Hence, relative to stable periods, individuals living under lockdown conditions are less likely to have memorable experiences. Second, even activities introduced by or required as a result of the Pandemic (e.g., online meetings) might not be well remembered as individual events. This is because these activities are likely to be frequently repeated and as a result, rapidly schematized (Barsalou, 1988; Conrad, Brown, & Cashman, 1998; Linton, 1982; Neisser, 1981, Neisser, 1986; Rumelhart, Smolensky, McClelland, & Hinton, 1986; Schank, 1999). One way to test these predictions is to collect two sets of freely recalled autobiographical memories, one from 2020 and the other from a subsequent “normal” year. More concretely, consider a study that requires participants, in this case first-year undergraduates in their second term of university, to recall a dozen memorable events from the previous calendar year. This method is known to produce a calendar effect (Kurbat et al., 1998; Pillemer et al., 1988; Pillemer, Rhinehart, & White, 1986; Robinson, 1986), which is the tendency to recall more events from the beginning and the end of the academic year than from other times2 . Using this design, the expectation is that data collected from the “normal” year should display the typical calendar effect, producing pronounced peaks for June and September. The temporal distribution of event memories from 2020 is predicted to look very different, though the exact locations of the COVID bump and the lockdown dip should differ depending on when the Pandemic hit and for how long a population was locked down. More concretely, consider the situation in Edmonton, Alberta, Canada (the author's residence). Here, schools went online and non-essential businesses closed in mid-March (approximately two weeks after the World Health Organization declared the COVID-19 outbreak to be a global Pandemic) and remained closed until mid-June. Thus, compared to a typical distribution of event memories, the 2020 distribution should display an increase in the frequency of memories from March 2020, the COVID bump, and a decrease in memories from April and May, the lockdown dip. It also seems possible that the June and September peaks – the defining characteristics of the calendar effect – will be attenuated in 2020, compared to those observed following a “normal” year. This prediction recognizes that the Pandemic caused the cancellation of many distinctive, graduation-related activities and that incoming first-year students were denied a normal (and hence memorable) “freshman experience” in September 2020 because all university classes were conducted online and because incoming students did not have the opportunity to relocate to campus. This general method – a method that elicits personal event memories from 2020 – can be extended to non-university samples drawn from different regions. The expectation is that the location of the COVID bump should reflect the arrival of the Pandemic in a particular region and that dips should occur during different months in different regions depending on when those regions were locked down. Finally, it should be noted that for some people (e.g., frontline workers, people who have lost their livelihood), the Pandemic has been a highly emotional and eventful time. These people are likely to produce a relatively large COVID bump in this event-recall task and should also produce evidence indicating that they represent the Pandemic as a historically-defined autobiographical period (see below). 2.3 The pandemic period – H-DAP or extended interlude? Public events sometimes define lifetime periods (Bohn & Habermas, 2016: Brown, 2016; Brown & Lee, 2010; Brown et al., 2009; Brown et al., 2012; Gu et al., 2017; Zebian & Brown, 2014). These historically-defined autobiographical periods (H-DAPs) are uncommon, occurring only when a population has undergone a major collective transition, a transition that has imposed rapid, fundamental, and enduring changes in the lives of all people affected by it. Typically, then, and in contrast to the COVID Pandemic (see above), H-DAP formation involves (a) a sharp decrease in the stability Ratio brought on by some identifiable public event or process. This is followed by (b) an extended span of instability – the H-DAP itself – which in turn leads to (c) the gradual establishment of a personally-defined post-transitional period, a period that is often very different from (i.e., discontinuous with) either the earlier personally-defined autobiographical period or the more recently experienced H-DAP. Where H-DAPs do exist, they are considered important chapters in a person's life story (Gu et al., 2017), are often conveyed from one generation to next (Gu, Tse, & Brown, 2020; Svob & Brown, 2012) and frequently serve as temporal landmarks (Bohn & Habermas, 2016; Brown et al., 2009, Brown et al., 2016; Zebian & Brown, 2014). At present, several billion people are in (or have recently been in) lockdown, self-isolation, or quarantine; all these people are living (or have been living) lives that are to some degree different from the ones they lived before the Pandemic. At least in a numerical sense, the arrival of the Novel Coronavirus has brought about the largest collective transition humanity has experienced. Does it follow that all these people will inevitably represent the Pandemic as an important historically-defined autobiographical period? In other words, in the future, will we frequently mention the Pandemic as a temporal landmark when dating unrelated events (Brown et al., 2016; Friedman, 1993; Shum, 1998)? Will we list it as an important event in our lives and include it as a major chapter in our life stories (Habermas & Bluck, 2000; McAdams, 2001; Thomsen, 2015)? In order to predict how the COVID Pandemic might affect the organization of autobiographical memory, it is necessary to consider the degree to which it has produced a fundamental and enduring change in people's lives3 . As noted above, it appears that we have arrived at a new normal, a state of affairs that is, at least, moderately stable and somewhat different from life pre-COVID. It is also clear that despite its near universal scope, the immediate impact of the Pandemic has differed greatly from place to place (e.g., Italy with 116.5 COVID deaths per 100,000 people vs Indonesia with 7.6 COVID deaths per 100,000; Johns Hopkins Coronavirus Resource Center, 2020, December 24), from individual to individual (e.g., a bankrupted business owner vs a professor working from home), and from time to time (life during and after lockdown). What is less clear is the degree to which the next new normal – life after the Pandemic – will differ from the pre-COVID period. This is an important issue because only some transitional events are considered important and only some identifiable periods become chapters in the life story. In general, in retrospect, transitions are considered important and personal periods treated as life-story chapters only when they have shaped the trajectory of a person's life (Gu et al., 2017, Gu et al., 2020; Habermas & Bluck, 2000; McAdams, 2001; Thomsen et al., 2015, Thomsen and Berntsen, 2008). Thus, if there is a high degree of continuity between the two periods, the Pandemic is likely to have the status of an extended interlude 4 . In other words, a person whose post-Pandemic life is much the same as his/her pre-Pandemic life may well remember the Pandemic as a special time, but should not, in retrospect, rate it as being particularly important nor consider it to be a major chapter in the life story. It will be possible to test these ideas directly by combining the TIS (Svob et al., 2014) with standard methods used to extract life stories. Again, the TIS can be used to identify those aspects of a person's life that have changed because of the COVID Pandemic. In a research setting, life stories are elicited by asking participants to provide an open-ended narrative of their lives (McAdams, 2001; Thomsen, 2009), by asking them to divide their lives into chapters (Thomsen, Pillemer, & Ivcevic, 2011) or by asking for a listing of most-important life events (Glück & Bluck, 2007). In each case, it will be necessary to collect the life-story events first and then to assess the transitional impact of the Pandemic using the TIS. It is reasonable to expect that some people will mention the Pandemic in their life stories and others will not. If so, for reasons developed above, (a) TIS scores should be high when people include the Pandemic in their life stories and (b) people who include the Pandemic in their life stories will provide higher TIS ratings than those who do not. (Gu et al., 2017; Shi & Brown, 2016). This approach can be extended in a quasi-experimental manner to assess predictable between-group differences in the Pandemic's mnemonic impact. For example, one could compare people who have lost their homes and their jobs to the Pandemic with those who have not. Lastly, as mentioned above, the COVID-19 Pandemic can be seen as the most extensive collective transition in history. Yet Transition Theory treats collective transitions and personal transitions in the same way. Thus the theory predicts that the Pandemic, like any other transition, will come to organize autobiographical memory only if it has changed a person's life in fundamental ways and that this should be true regardless of the Pandemic's scope, its cultural prominence or its historical importance (Brown, 2016; Brown et al., 2012; Gu et al., 2017). It will be interesting, going forward, to determine whether these social aspects of the Pandemic will also affect its mnemonic impact. Given its scope and scale, it is possible that “life during the Pandemic” will be a lasting topic of conversation and that, by convention, people, even those unscathed by it, will claim the Pandemic as an important chapter in their lives (Cohn, Mehl, & Pennebaker, 2004; Hirst & Echterhoff, 2012). If so, the collective and pervasive nature of the Pandemic may override or at least augment the basic memory processes that are thought to give rise to H-DAPs. This in turn might weaken or eliminate the predicted relation between the Pandemic's long-term transitional impact and its inclusion in the life story. This possibility can be tested by focusing on people whose lives were not radically altered the Pandemic. If social factors play an important role in determining the inclusion of a COVID chapter in the life story, then, other things being equal (i.e., holding the transitional impact constant), these chapters should be common in hard-hit areas and uncommon in areas that were relatively unaffected by the Pandemic. 3 General discussion In many ways, the COVID Pandemic is unique; it is unique in its near universal scope; it is unique in the way that it has changed our lives; and it is likely to be unique in the way that we remember it. Transition Theory provides a language to describe at least some of this uniqueness. From a Transition-Theory perspective, the onset of the COVID-19 Pandemic can be characterized as a transition-by-omission, one that has produced a relatively shallow decline and then a steep increase in the Stability Profile leading to a moderately stable Pandemic Period. Transition Theory has also provided the basis for predicting how the Pandemic might affect autobiographical memory. Specifically, it seems likely that that the Pandemic will produce a COVID bump and, depending on location, one or more lockdown dips. It is also likely that Pandemic will create an identifiable period in autobiographical memory. Whether this period will play a central role in a person's life story will depend on its long-term impact. The prediction here is that people will consider the Pandemic to be an important chapter in their life stories only if there is little continuity between their pre-Pandemic and post-Pandemic lives. Otherwise, the Pandemic will be treated as an extended interlude. Two final points. The first concerns the quantitative approach outlined above. In this article, the Stability Ratio, the Stability Profile, and the Continuity Ratio were introduced to highlight the special properties of the COVID Pandemic. However, these notions are grounded in what appear to be fundamental aspects of experience, aspects that have, to some extent, been underappreciated by autobiographical memory researchers, but surely play an important role in determining what we remember about our lives and how we organize what we do remember. Specifically, (a) some time spans are dominated by predictable and repetitive activities, others are filled with novel experiences and new challenges – this aspect is captured by the Stability Ratio; (b) some transitions change many facets of our lives and cause a sharp break between the past and present, others alter one or two facets, but are not particularly disruptive – differences in transitional impact are reflected in the Stability Profile and the Continuity Ratio; finally, on a larger scale, (c) some lives are tumultuous and chaotic; others placid and orderly – these different ways of being yield markedly different Stability Profiles. In brief, these tools – the Stability Ratio, the Continuity Ratio, and the Stability Profile –make it possible to describe the texture of life at a given point in time, to assess the similarity of periods across time, and to plot the contours of a life over time. We know that change and stability, also distinctiveness and repetition, the concepts that underlie the current approach, affect the organization and content of autobiographical memory (Brown, 2016; McAdams, 2001; Thomsen, 2015); we also know that these factors affect people's mental health and physical well-being (Hobson & Delunas, 2001; Holmes & Rahe, 1967; Lundberg, Theorell, & Lind, 1975; Tennant, 2002; Wheaton, 1990). Thus, an approach like the present one that quantifies these aspects of a person's life is likely to have a wide range of research applications. Admittedly, moving from theory to application will be a challenge – this is the second point. The challenge lies in the requirement to catalogue and classify ECs accurately, at various points in time. Fortunately, methods are available to help with this task. These include traditional approaches like social-network elicitation (McCallister & Fischer, 1978), diary keeping (Thompson, Skowronski, Larsen, & Betz, 1996), and experience sampling (Larson & Csikszentmihalyi, 2014). In addition, there exist computer technologies that capture people's activities, locations and interactions as they happen (e.g., Greaves et al., 2015; Hodges et al., 2006). There are also techniques for extracting information about people's lives from their email, scheduling software and social media (Gemmell, Bell, & Lueder, 2006; Vianna, Yong, Xia, Marian, & Nguyen, 2014). The existence of this suite of methods indicates that it is feasible to create individualized EC catalogues and to update them over time, and hence that it is realistic to believe we can acquire the data required to compute Stability and Continuity Ratios and to construct Stability Profiles. 4 Conclusion There is no doubt that the Pandemic will be seen as an extremely important historical event, and there is no doubt that it is permanently altering the lives of a vast number of people and temporarily altering the lives of many more. And it seems equally certain that the Pandemic will leave its mark on autobiographical memory. What is less certain is how the Pandemic will be remembered. This article provides a theoretical characterization of the COVID-19 Pandemic and develops a set of predictions based on this characterization. Although it is too soon to assess the accuracy of all of these predictions, it would seem that there is much to gain by considering the memory issues associated with this world-changing historical event as it unfolds rather than in retrospect. On the one hand, the unusual nature of the Pandemic has provided the impetus for extending Transition Theory itself (cf. Brown, 2016); on the other, predictions derived from this theory call for a longitudinal approach, one grounded in an accurate and nuanced understanding of the Pandemic experience. Thus, it is clear that the time to begin studying the long-term effects of the COVID Pandemic is now. Acknowledgements This research was supported by the author's NSERC Discovery Grant RES0038944. I'd like to thank Jill Hollis, Peter Lee, John Reddon, Eva Rubinova and Connie Svob for their comments on an earlier version of this article. 1 The focus here is on people who have retained their jobs or continued with their studies during the Pandemic, but have been forced to work or study from home. Other scenarios are considered below. 2 From a Transition-Theory perspective, the calendar effect is understood as a specific example of the more general tendency for transitions (e.g., high school graduation, starting university) to give rise to memorable personal events (see above). 3 Even very important historical events (e.g., the 9/11 Attacks, the collapse of the Soviet Union) fail to induce H-DAPs if they do not bring about “fundamental and enduring” changes in the lives of people in the effected populations (Brown et al., 2009; Brown & Lee, 2010; Nourkova & Brown, 2015). 4 A semester-long study leave in a foreign country exemplifies this concept. At first one experiences all the novelty and challenges of a major transition. However, with time, life in this new location becomes routine. In this case, the continuity between the leave period and the pre- and post-leave periods is low whereas the continuity between the pre- and post-leave periods is very high. The intuition here is that though the leave period may well be remembered fondly, it should generally not be included in the list of important life chapters because it failed to produce important and enduring changes in the fabric of daily life. ==== Refs References Barsalou L.W. The content and organization of autobiographical memories Neisser U. Winograd E. Remembering reconsidered: Ecological and traditional approaches to the study of memory 1988 Cambridge University Press Cambridge, UK 193 243 Bohn A. Habermas T. Living in history and living by the cultural life script: How older Germans date their autobiographical memories Memory 24 2016 482 495 10.1080/09658211.2015.1019890 25768233 Brewin C.R. Dalgleish T. Joseph S. A dual representation theory of posttraumatic stress disorder Psychological Review 103 1996 670 686 10.1037/0033-295x.103.4.670 8888651 Brown N.R. Transition theory: A minimalist perspective on the organization of autobiographical memory Journal of Applied Research in Memory and Cognition 5 2016 128 134 Brown N.R. Hansen T.G.B. Lee P.J. Vanderveen S.A. Conrad F.G. Historically-defined autobiographical periods: Their origins and implications Berntsen D. Rubin D. Understanding autobiographical memory: Theories and approaches 2012 Cambridge University Press Cambridge, UK 160 180 Brown N.R. Lee P.J. Public events and the organization of autobiographical memory: An overview of the living-in-history project Behavioral Sciences of Terrorism and Political Aggression 2 2010 133 149 Brown N.R. Lee P.J. Krslak M. Conrad F.G. Hansen T. Havelka J. Reddon J. Living in history: How war, terrorism, and natural disaster affect the organization of autobiographical memory Psychological Science 20 2009 399 405 19298262 Brown N.R. Schweickart O. Svob C. The effect of collective transitions on organization and contents of autobiographical memory: A transition-theory perspective American Journal of Psychology 129 2016 259 282 29558591 Cohn M.A. Mehl M.R. Pennebaker J.W. Linguistic markers of psychological change surrounding September 11, 2001 Psychological Science 15 2004 687 693 10.1111/j.0956-7976.2004.00741.x 15447640 Conrad F. Brown N.R. Cashman E. Strategies for estimating behavioral frequency in survey interviews Memory 6 1998 339 366 9829096 Conway M.A. Memory and the self Journal of Memory and Language 53 2005 594 628 Conway M.A. Episodic memories Neuropsychologia 47 2009 2305 2313 19524094 Enz K.F. Pillemer D.B. Johnson K.M. The relocation bump: Memories of middle adulthood are organized around residential moves Journal of Experimental Psychology: General 145 2016 935 940 27454038 Friedman W.J. Memory for the time of past events Psychological Bulletin 113 1993 44 66 Gemmell J. Bell G. Lueder R. MyLifeBits: A personal database for everything Communications of the ACM 49 1 2006 88 95 10.1145/1107458.1107460 Glück J. Bluck S. Looking back across the life span: A life story account of the reminiscence bump Memory & Cognition 35 2007 1928 1939 18265609 Greaves S. Ellison A. Ellison R. Rance D. Standen C. Rissel C. Crane M. A web-based diary and companion smartphone app for travel/activity surveys Transportation Research Procedia 11 2015 297 310 10.1016/j.trpro.2015.12.026 Gu X. Tse C. Brown N.R. Factors that modulate the intergenerational transmission of autobiographical memory from older to younger generations Memory 28 2020 204 215 31888407 Gu X. Tse C.-S. Brown N.R. The effects of collective and personal transitions on the organization and contents of autobiographical memory in older Chinese adults Memory & Cognition 45 2017 1335 1349 28710601 Habermas T. Bluck S. Getting a life: The emergence of the life story in adolescence Psychological Bulletin 126 2000 748 769 10.1037/0033-2909.126.5.748 10989622 Heanoy E.Z. Shi L. Brown N.R. Assessing the transitional impact and mental health consequences of the COVID-19 pandemic onset Frontiers of Psychology 11 2020 607976 Hebb D.O. The organization of behavior 1949 Wiley New York, NY Hirst W. Echterhoff G. Remembering in conversations: The social sharing and reshaping of memories Annual Review of Psychology 63 2012 55 79 Hobson C.J. Delunas L. National norms and life-event frequencies for the revised social readjustment rating scale International Journal of Stress Management 8 2001 299 314 Hodges S. SenseCam: A retrospective memory aid Dourish P. Friday A. UbiComp 2006. Lecture Notes in Computer Science 4206 2006 Springer Berlin, Heidelberg 10.1007/11853565_11 Holmes T.H. Rahe R.H. The social readjustment rating scale Journal of Psychosomatic Research 1967 213 218 6059863 Hunt R.R. The concept of distinctiveness in memory research Hunt R.R. Worthen J.B. Distinctiveness and memory 2006 Oxford Oxford, UK 3 26 Hunt R.R. Rawson K.A. Knowledge affords distinctive processing in memory Journal of Memory and Language 65 2011 390 405 Johns Hopkins Coronavirus Resource Center Mortality analysis https://coronavirus.jhu.edu/data/mortality 2020, December 24 Kahneman D. Krueger A.B. Schkade D.A. Schwarz N. Stone A.A. A survey method for characterizing daily life experience: The day reconstruction method Science 306 2004 1776 1780 15576620 Koppel J. Berntsen D. The peaks of life: The differential temporal locations of the reminiscence bump across disparate cueing methods Journal of Applied Research in Memory and Cognition 4 2015 66 80 10.1016/j.jarmac.2014.11.004 Kubovy M. Lives as collections of strands: An essay in descriptive psychology Perspectives on Psychological Science 15 2020 497 515 31935336 Kurbat M.A. Shevell S.K. Rips L.J. A year’s memories: The calendar effect in autobiographical recall Memory & Cognition 26 1998 532 552 9610123 Larson R. Csikszentmihalyi M. The experience sampling method Flow and the foundations of positive psychology: the collected works of mihaly csikszentmihalyi 2014 Springer Netherlands 21 34 10.1007/978-94-017-9088-8_2 Linton M. Transformations of memory in everyday life Neisser U. Memory observed - remembering in natural contexts 1982 W.H. Freeman and Company NY, US 77 91 Linton M. Ways of searching the contents of memory Rubin D.C. Autobiographical memory 1986 Cambridge University Press New York, NY 50 67 Lundberg U. Theorell T. Lind E. Life changes and myocardial infarction: Individual differences in life change scaling Journal of Psychosomatic Research 19 1975 1975 27 32 1142312 McAdams D.P. The psychology of life stories Review of General Psychology 5 2001 100 122 10.1037/1089-2680.5.2.100 McCallister L. Fischer C.S. A procedure for surveying personal networks Sociological Methods & Research 7 1978 131 148 McClelland J.L. McNaughton B.L. O'Reilly R.C. Why there are complementary learning systems in the hippocampus and neocortex: Insights from the successes and failures of connectionist models of learning and memory Psychological Review 102 1995 419 457 7624455 Morton J. Hammersley R.H. Bekerian D.A. Headed records: A model for memory and its failures Cognition 20 1985 1 23 4017518 Munakata Y. Pfaffly J. Hebbian learning and development Developmental Science 2 2004 141 148 Neisser U. John Dean’s memory: A case study Cognition 9 1981 102 115 Neisser U. Nested structure in autobiographical memory Rubin D.C. Autobiographical memory 1986 Cambridge University Press Cambridge, UK 71 81 10.1017/CBO9780511558313.009 Nelson A.B. Shiffrin R.M. The co-evolution of knowledge and event memory Psychological Review 120 2013 356 394 23458086 Norman D.A. Bobrow D.C. Descriptions: An intermediate stage in memory retrieval Cognitive Psychology 11 1979 107 123 Nourkova V.V. Brown N.R. Assessing the impact of “the collapse” on the organization and content of autobiographical memory in the former Soviet Union Journal of Social Issues 71 2015 324 337 Pillemer D.B. Goldsmith L.R. Panther A.T. White S.H. Very long-term memories of the first year in college Journal of Experimental Psychology. Learning, Memory, and Cognition 14 1988 709 715 Pillemer D.B. Rhinehart E.D. White S.H. Memories of life transitions: The first year in college Human Learning: Journal of Practical Research & Applications 5 1986 109 123 Robinson J.A. Temporal reference systems and autobiographical memory Rubin D. Autobiographical memory 1986 Cambridge University Press New York, NY 159 188 Robinson J.A. First experience memories: Contexts and functions in personal memories Conway M.A. Rubin D.C. Spinnler H. Wagenaar W.A. Theoretical perspectives on autobiographical memory 1992 Kluwer Academic Dordrecht, NL 223 239 Rubin D.C. Boals A. Berntsen D. Memory in posttraumatic stress disorder: Properties of voluntary and involuntary, traumatic and nontraumatic autobiographical memories in people with and without posttraumatic stress disorder symptoms Journal of Experimental Psychology: General 137 2008 591 614 10.1037/a0013165 18999355 Rubin D.C. Rahhal T.A. Poon L.W. Things learned in early adulthood are remembered best Memory & Cognition 26 1998 3 19 10.3758/bf03211366 9519693 Rumelhart D.E. Smolensky P. McClelland J.L. Hinton G. Schemata and sequential thought processes in PDP models McClelland J.L. Rumelhart D.E. the PDP Research Group Parallel distributed processing: Explorations in the microstructures of cognition 1986 MIT Cambridge, MA 3 57 Schank R.C. Dynamic memory revisited 1999 Cambridge University Press Cambridge, UK Shi L. Brown N.R. The effect of immigration on the contents and organization of autobiographical memory: A transition-theory perspective Journal of Applied Research on Memory and Cognition 5 2016 135 142 Shimamura A.P. Remembering the past neural substrates underlying episodic encoding and retrieval Current Directions in Psychological Science 234 2014 257 263 Shum M.S. The role of temporal landmarks in autobiographical memory processes Psychological Bulletin 124 1998 423 442 10.1037//0033-2909.124.3.423 9849113 Smith E.R. DeCoster J. Dual-process models in social and cognitive psychology: Conceptual integration and links to underlying memory systems Personality and Social Psychology Review 4 2000 108 131 Svob C. Brown N.R. Intergenerational transmission of the reminiscence bump and biographical conflict knowledge Psychological Science 23 2012 1404 1409 23012267 Svob C. Brown N.R. Reddon J.R. Uzer T. Lee P.J. The transitional impact scale: Assessing the material and psychological impact of life transitions Behavior Research Methods 46 2014 448 455 23943584 Tennant C. Life events, stress and depression: A review of recent findings Australian and New Zealand Journal of Psychiatry 36 2002 174 182 Thompson C.P. Skowronski J.J. Larsen S.F. Betz A. Autobiographical memory: Remembering what and remembering when 1996 Lawrence Erlbaum Associates Inc Thomsen D.K. There is more to life stories than memories Memory 17 2009 445 457 19241217 Thomsen D.K. Autobiographical periods: A review and central components of a theory Review of General Psychology 19 2015 294 310 Thomsen D.K. Berntsen D. The end point effect in autobiographical memory: More than a calendar is needed Memory 13 2005 846 861 10.1080/0965821044400044 16298892 Thomsen D.K. Berntsen D. The cultural life script and life story chapters contribute to the reminiscence bump Memory 16 2008 420 436 18432486 Thomsen D.K. Jensen T. Holm T. Olesen M.H. Schnieber A. Tønnesvang J. A 3.5 year diary study: Remembering and life story importance are predicted by different event characteristics Consciousness and Cognition 36 2015 180 195 10.1016/j.concog.2015.06.011 26164104 Thomsen D.K. Pillemer D.B. Ivcevic Z. Life story chapters, specific memories and the reminiscence bump Memory 19 2011 267 279 21500087 Tversky A. Features of similarity Psychological Review 84 1977 327 352 10.1037/0033-295X.84.4.327 Uzer T. Validity and reliability testing of the transitional impact scale Stress and Health 36 2020 478 486 10.1002/smi.2944 32281242 Uzer T. Brown N.R. Disruptive individual experiences create lifetime periods: A study of autobiographical memory in persons with spinal cord injury Applied Cognitive Psychology 29 2015 768 774 Vianna D. Yong A. Xia C. Marian A. Nguyen T. A tool for personal data extraction 2014 IEEE 30th International Conference on Data Engineering Workshops 2014 80 93 Chicago, IL 10.1109/ICDEW.2014.6818307 Wagenaar W.A. My memory: A study of autobiographical memory over six years Cognitive Psychology 18 1986 225 252 Wheaton B. Life transitions, role histories, and mental health American Sociological Review 55 1990 209 223 White M.P. Dolan P. Accounting for the richness of daily activities Psychological Science 20 2009 1000 1008 19549079 Zebian S. Brown N.R. Living in history in Lebanon: The influence of chronic social upheaval on the organization of autobiographical memory Memory 22 2014 194 211 23473445

33798951

PMC9748850

NO-CC CODE

2022-12-15 23:22:47

no

Cognition. 2021 Jul 31; 212:104694

utf-8

Cognition

10.1016/j.cognition.2021.104694

oa_other

==== Front ISA Trans ISA Trans ISA Transactions 0019-0578 1879-2022 ISA. Published by Elsevier Ltd. S0019-0578(22)00637-1 10.1016/j.isatra.2022.12.006 Article The effect mitigation measures for COVID-19 by a fractional-order SEIHRDP model with individuals migration Lu Zhenzhen a Chen YangQuan b Yu Yongguang a⁎ Ren Guojian a Xu Conghui a Ma Weiyuan c Meng Xiangyun a a Department of Mathematics, Beijing Jiaotong University, Beijing, 100044, PR China b Mechatronics, Embedded Systems and Automation Lab, University of California, Merced, CA 95343, USA c School of Mathematics and Computer Science, Northwest Minzu University, Lanzhou, 730000, PR China ⁎ Corresponding author. 14 12 2022 14 12 2022 5 8 2020 22 11 2022 10 12 2022 © 2022 ISA. Published by Elsevier Ltd. All rights reserved. 2022 ISA Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. In this paper, the generalized SEIHRDP (susceptible-exposed-infective-hospitalized-recovered-death-insusceptible) fractional-order epidemic model is established with individual migration. Firstly, the global properties of the proposed system are studied. Particularly, the sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. Secondly, according to the real data in India and Brazil, it can all be concluded that the bilinear incidence rate has a better description of COVID-19 transmission. Meanwhile, multi-peak situation is considered in China, and it is shown that the proposed system can better predict the next peak. Finally, taking individual migration between Los Angeles and New York as an example, the spread of COVID-19 between cities can be effectively controlled by limiting individual movement, enhancing nucleic acid testing and reducing individual contact. Keywords Individual migration Fractional-order epidemic model Peak prediction Sensitivity ==== Body pmc1 Introduction End of 2019 saw the outbreak of the dangerous infectious disease COVID-19, which is brought on by a novel coronavirus. Global public health has been significantly impacted by the 13,837,395 diagnosed cases and the 590,702 death cases as of July 17, 2020 [1]. The quick rise in infection cases suggests that COVID-19 has a much greater ability to spread than MERS-CoV and SARS coronaviruses [2], [3]. A direction for taking the right steps can be provided by a deeper comprehension and insight of the epidemic tendencies. Numerous nations have implemented a variety of mitigating strategies to prevent the spread of COVID-19 since 23 January, 2020. These strategies include home isolation, herd immunity, limiting individual migration, and others. Lack of information on the dynamic mechanism relating to the severity of COVID-19 at the this early stage makes it extremely difficult to limit the spread of COVID-19. However, using a mathematical model, strategies can be measured to serve as a benchmark for determining whether mitigation strategies are adequate. During the modeling process, it is very important to describe how infectious diseases are transmitted between susceptible and infected individuals. Numerous studies have shown that the incidence rate, which measures the infection capacity of a single infected person per unit time, is a crucial tool for describing this process, where susceptible individuals come into contact with infected individuals and are then infected with such a predetermined probability [4], [5], [6], [7]. Meanwhile, Korobeinikov et al. [8] indicated that the stability of the endemic equilibrium point is closely related to the concave of the incidence rate with respect to the infected individuals. Therefore, many researchers established the epidemic model of COVID-19 under various incidence rates  [9], [10], [11], for example, Peng et al. [9] constructed the SEIR (E-exposed) epidemic model and they discovered that COVID-19’s first appearance might be traced to the end of December 2019. A SEIQRD (Q-diagnosed) model was taken into consideration by Xu et al. [10], which has some basic guiding relevance for predicting COVID-19. Besides, individual migration has a crucial effect on the evolution of infectious diseases. With the convenience between cities, individuals move more and more frequent and new infectious diseases develop more rapidly regionally and globally [12]. Numerous deterministic models with multiple patches have been presented in attempt to better understand how individual migration affects the spread of infectious illnesses [13], [14], [15]. Contrary to what was initially reported [16], COVID-19 is in fact spreading from person to person through continuous interpersonal contact [2]. Lu et al. [17] considered a fractional-order SEIHRD (H-hospitalized) model with inter-city networks and they found that COVID-19 could be reduced in low-risk areas, but increased in high-risk areas by restricting communication between cities. Meanwhile, cross-infection among cities are considered, while there is not consider for self-migration [17]. Therefore, it is of great practical significance to include individual migration in different cities or different countries with the modeling COVID-19. Furthermore, the migration of susceptible individuals, exposed individuals, infected individuals are studied in this paper. It is worth noting that the time which patients waits for treatment follows the power law distribution [18], which prompts the use of the Caputo fractional-order derivative [19]. Angstmann et al. [20] discovered how fractional operators naturally appear in their model if the recovery time is a power law distribution after building a SIR epidemic model. Meanwhile, this offers a chronic disease epidemic model in which long-term infected people have little chance of recovering. Based on this statement, several authors have stated that the fractional-order model plays an important role in the process of disease transmission. Khan et al. [21] recounted how individuals, bats, unidentified hosts, and the source of the illness interacted, and considered how crucial the fractional-order system was in preventing the spread of the infection. To predict the spread of COVID-19, Chen et al. [22] developed a fractional-order epidemic model. Amjad et al. [23] built a fractional-order COVID-19 model and calculated the consequences of several mitigation and prevention strategies. Motivated by the above discussion, a fractional-order SEIHRDP epidemic model with individuals movement is established in this paper to study COVID-19. Meanwhile, the number of hospitalizations is the same as confirmed isolation in China, and but in other countries, these two are not equal, which the number of confirmed case is greater than that of hospitalized case. So in order to give a more generalized model, the purpose of this paper is to describe hospitalized individuals in response to the spread of COVID-19. The infectiousness of the incubation time is also taken into consideration, as inspired by [24]. Then, the proposed system’s dynamic behaviors are investigated in order to show the existence and uniqueness of the nonnegative solution, the global asymptotic stability of the disease-free equilibrium, and the uniform persistence, all of which have theoretical implications for future COVID-19 intervention and prevention. Meanwhile, the basic reproduction number with and without individual migration are compared, and it is found that adding individual migration can effectively describe the spread of COVID-19. Furthermore, the sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. Meanwhile, considering India and Brazil, results suggest that the bilinear incidence rate may be more fitted than the saturation incidence rate for stimulating the spread of COVID-19. When individuals movement is not considered, it can be found the proposed fractional-order model can better predict than the integer-order for multi peaks of COVID-19 in China. Meanwhile, when individuals movement is considered, the epidemic in the United States is analyzed and some mitigation measures are carried out to control the development of COVID-19. An implication of the achieved results is the possibility that the United States peaked on 24 November, 2020 (integer-order system) and 1 January, 2021 (fractional-order system), however, the number of infections shows an downward trend after 17 July, 2020 as enhancing nucleic acid detection and reducing the contact rate. Meanwhile, considering measures to limit migration between New York and Los Angeles, and enhance nucleic acid detection and reduce exposure rates, it is evident that there is an immediate increase in confirmed cases before a drop. Based on the above analysis, a generalized fractional-order SEIHRDP epidemic model with individual migration is considered. The main contributions of this study are as follows: • A fractional epidemic model with self-migration is considered, in which the infectivity of exposed individuals and hospitalized individuals are also taken into account. • The global properties of the proposed model are investigated, including the existence and uniqueness of global positive solutions, the local and global stability of disease-free equilibrium points, the persistence of disease transmission. • The sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. • Based on real data, the impact of the incidence rate on modeling COVID-19 is studied in India and Brazil. • Multiple peaks of COVID-19 transmission in China are analyzed by the proposed system. • Individual movement in the spread of COVID-19 in the United States is investigated and the peak are analyzed based on mitigation measures, such as enhanced nucleic acid testing, reduced of individual exposure, and control of individual movement. The rest of this paper is organized as follows. The SEIHRDP fractional-order model with individual movement is developed for COVID-19 in Section 2 and provides some preliminaries. Then dynamic properties of the proposed system are examined in Section 3. The theoretical results are shown using numerical simulations in Section 4. Finally, Section 5 provides the conclusions. 2 System description and preliminaries Fractional-order operator have been determined to have a wide range of uses in the modeling of many dynamic processes, including those in engineering, biology, medicine, and others [25], [26], [27], [28]. In this part, some necessary preliminaries are introduced before the fractional-order epidemic model is presented. 2.1 Preliminaries Definition 2.1 [29] The Caputo fractional-order operator is defined by t0CDtαgt=dαgtdtα=1Γn−α∫0tg(n)(s)t−sα−n+1ds,(n−1<α<n), where g(n)(s) is the nth derivative of g(s) with respect to s. Remark 2.1 If α=n, one has t0CDtαgt=g(n)(t). Lemma 2.1 [30] The Caputo nonlinear system is considered as follows: 0CDtαx(t)=g(x),(α∈(0,1]), with the initial condition x0 . If all eigenvalues of J|x=x∗=∂g∂x|x=x∗ satisfy |arg(λ)|>απ2 , the equilibrium points x∗ are locally asymptotically stable. Lemma 2.2 [31] Suppose X⊂R and the continuous operator T(t):X→X satisfies (1) T(t) is point dissipative in X and compact for t≥0 . (2) there is a finite sequence M={M1,M2,…,Mk} of compact and isolated invariant sets such that (i) Mi∩Mj=0̸ for any i,j=1,2,…,k and i≠j ; (ii) Ω(∂X0)≜∪x∈∂X0ω(x)⊂∪i=1kMi ; (iii) in the case of ∂X0 , no a cycle is formed by any subset of M ; (iv) Ws(Mi)∩X0=0̸ for each i=1,2,…,k . Then T(t) is uniformly persistent in X . 2.2 Graph theory In this paper, a weighted graph ζ=(ϑ,ω,A) will be considered to model the spread of infectious diseases between cities, where ϑ={ϑ1,ϑ2,…,ϑn} denotes the node set and ϑi represents the ith city; ω⊆ϑ×ϑ is the edge set, and if there is individual movement between any two cities, it means that there is a edge between this two cities; matrixes M=[mij]1≤i,j≤n, N=[nij]1≤i,j≤n, P=[pij]1≤i,j≤n and Q=[qij]1≤i,j≤n represent the weighted adjacency matrix of susceptible, exposed, infected and recovered individual, respectively; mij, nij, pij and qij denote the migrate rate of susceptible, exposed, infected and recovered individual from city j to city i with aij≥0 (i≠j) and aii=0 (a=m,n,porq), respectively. Furthermore, based on the directivity of individual migration, the directed graph ζ is studied in this paper. 2.3 System description Starting from 23 January, 2020, the Chinese government has adopted a series of mitigation measures to effectively suppress the spread of COVID-19, such as implementing strict home isolation, restricting various traffic, strengthening nucleic acid testing, establishing shelter hospitals and so on. Meanwhile, other countries around the world have adopted different measures from China, such as social distancing and herd community strategy by British, protecting sensitive compartment from infection by Italy, transferring of critically ill patients with military aircraft by France, etc. Therefore, it is important to establish a generalized model of individual migration to simultaneously quantify the impact of interruption of policies on virus transmission. Moreover, Tang et al. [32] proposed that the exposed individual is infectious of COVID-19. Therefore, a fractional-order SEIHRDP epidemic model with individual migration is considered as follows: (1) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),0CDtαEk=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),0CDtαIk=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),0CDtαHk=δkIk−(λk+κk)Hk,0CDtαRk=λkHk+∑j=1n(qkjRj−qjkRk),0CDtαDk=κkHk,0CDtαPk=ρkSk. with the initial condition (2) Sk(0)=Sk0>0,Ek(0)=Ek0≥0,Ik(0)=Ik0≥0,Hk(0)=Hk0≥0, Rk(0)=Rk0≥0,Dk(0)=Dk0≥0,Pk(0)=Pk0≥0. The specific explanation of system (1) are as follows: • Susceptible Sk: the number of susceptible class within city k at time t. • Exposed Ek: the number of exposed class within city k at time t (neither any clinical symptoms nor high infectivity). • Infectious Ik: the number of infected class within city k at time t (with overt symptoms). • Hospitalized Hk: the number of hospitalized class within city k at time t. • Recovered Rk: the number of recovered class within city k at time t. • Dead Dk: the number of dead class within city k at time t. • Insusceptible Pk: the number of susceptible class who are not exposed to the external community within city k at time t. Meanwhile, the process of disease transmission are as follows: • The susceptible individual Sk contacts with Ek and Ik, and then is infected by β1kSkfk(Ik)+β2kSkgk(Ek), where βik (i=1,2) are the transmission coefficient, fk(Ik) and gk(Ek) are generalized incidence rates. • The parameter Λk is the inflow rate; λk, ϵk, δk and κk represent the recovery, incubation, diagnosis, mortality rate. • The susceptible, exposed, infective and recovered individuals in city j move to city k with probability mkj, nkj, pkj and qkj, respectively. The terms ∑j=1n(mkjSj−mjkSk), ∑j=1n(nkjEj−njkEk), ∑j=1n(pkjIj−pjkIk) and ∑j=1n(qkjRj−qjkRk) represent the movement of Sk, Ek Ik and Rk individual, where ∑j=1nakjWj represents the individuals moving into k city from other cities j (k≠j) and ∑j=1najkWj represents the individuals leaving city k (W=S, E, I, R, respectively, a=m, n, p, q, respectively). • The movement of insuspectible and hospitalized individuals is not considered in this paper. Furthermore, Λk, βik (i=1,2), ρk and ϵk are positive constants; functions δk(t), λk(t), κk(t), mkj(t), nkj(t), pkj(t) and qkj(t) satisfy |δk(t)|≤M1k, |λk(t)|≤M2k, |κk(t)|≤M3k, |mkj(t)|≤M4k, |nkj(t)|≤M5k, |pkj(t)|≤M6k and |qkj(t)|≤M7k for all t≥0 and k,j=1,2,…,n, where M1k, M2k, M3k, M4k, M5k, M6k and M7k are positive constants. The transmission diagram of the generalized SEIHRDP model (1) is shown in Fig. 1. Before presenting the major findings, the following generalized incidence rate hypothesis is put forth: (H):(i)gk(Ek)andfk(Ik)satisfythelocalLipschitzconditionand gk(0)=0,fk(0)=0fork=1,2,…,n;(ii)fk(Ik)isstrictlymonotoneincreasingonIk∈[0,∞)and gk(Ek)isstrictlymonotoneincreasingon Ek∈[0,∞)forallk=1,2,…,n;(iii)fk(Ik)≤akIkforallIk≥0, whereak=fk′(0)forallk=1,2,…,n;(iv)gk(Ek)≤bkEkforallEk≥0, wherebk=gk′(0)forallk=1,2,…,n. Fig. 1 The schematic diagram of SEIHRDP epidemic model with individual migration (i,j=1,2,…,n). Remark 2.2 It should be noted that many current models can be viewed as a special type of system (1) with the hypothesis (H), such as gk(Ek)=bkEk, gk(Ek)=bkEk1+vkEk, fk(Ik)=akIk, fk(Ik)=akIk1+ukIk and others [33] with nonnegative constants ak, bk, uk and vk. Remark 2.3 Compared with [34], the individual movement in this paper can be described as follows: (1) the self-migration of individuals is described in system (1), which is caused by a self-chemotactic-like forcing [35]. However, the cross-infection among cities is considered which is a travel infectious [34]. (2) system (1) describes not only the migration of infected individuals, but also the movement of exposed and recovered individuals. (3) M=[mij]1≤i,j≤n, N=[nij]1≤i,j≤n, P=[pij]1≤i,j≤n and Q=[qij]1≤i,j≤n are not irreducible in this paper, but irreducible in [34]. Then the influence of network structure on disease transmission can be discussed in this paper, such as fully connected network, ring network and centralized network. However, [34] only consider fully connected network. (4) the total population of each city is changed (without considering the decrease in population due to death) in system (1). But in [34], the total population of each city remains constant. 3 System analysis This study explores system (1)’s dynamic analysis. As can be seen, the death class Dk and the insusceptible class Pk have no effect on the susceptible class Sk, exposed class Ek, infected class Ik, hospitalized class Hk, or recovered class Rk of systems (1). Accordingly, the following system is discussed in the next section: (3) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),0CDtαEk=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),0CDtαIk=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),0CDtαHk=δkIk−(λk+κk)Hk,0CDtαRk=λkHk+∑j=1n(qkjRj−qjkRk), with the initial condition (4) Sk(0)=Sk0>0,Ek(0)=Ek0≥0,Ik(0)=Ik0≥0, Hk(0)=Hk0≥0,Rk(0)=Rk0≥0,(k=1,2,…,n). 3.1 Existence and uniqueness of the positive solution The existence, uniqueness, and boundedness of the nonnegative solution for system (3) should be taken into account prior to the numerical process. Therefore, this subsection will be discussed these properties for system (3). Theorem 3.1 For any nonnegative initial condition (4) , there are a unique solution for system (3) and the region Ω={(S1,E1,I1,H1,R1,…,Sn,En,In,Hn,Rn): 0<Si≤Λ¯ρ,0≤Ei≤Λ¯ρ,0≤Ii≤Λ¯ρ, 0≤Hi≤Λ¯ρ,0≤Ri≤Λ¯ρ,i=1,2,…,n} is positively invariant for system (3) , where Λ¯=∑j=1nΛj and ρ=min{ρ1,ρ2,…,ρn} . Proof Let consider the following function: f1k=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),f2k=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),f3k=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),f4k=δkIk−(λk+κk)Hk,f5k=λkHk+∑j=1n(qkjRj−qjkRk). It is obvious that Fk=(f1k,f2k,f3k,f4k,f5k) satisfies the local Lipschitz condition about (Sk,Ek,Ik,Hk,Rk), then system (3) has a unique solution. Next, the nonnegative solution will be analyzed. Consider the following auxiliary system: 0CDtαS_k=−β1kS_kfk(I_k)−β2kS_kgk(E_k)−ρkS_k+∑j=1n(mkjS_j−mjkS_k),0CDtαE_k=β1kS_kfk(I_k)+β2kS_kgk(I_k)−ϵkE_k+∑j=1n(nkjE_j−njkE_k),0CDtαI_k=ϵkE_k−δkI_k+∑j=1n(pkjI_j−pjkI_k),0CDtαH_k=δkI_k−(λk+κk)H_k,0CDtαR_k=λkH_k+∑j=1n(qkjR_j−qjkR_k),Sk_(0)=Ek_(0)=Ik_(0)=Hk_(0)=Rk_(0)=0. Through the comparison theorem, it is not difficult to find that the following auxiliary system has a unique solution (0,0,0,0,0). then the following equation holds: (Sk,Ek,Ik,Hk,Rk)>(0,0,0,0,0). Next, adding all equations gives 0CDtαN≤Λ¯−ρN where N=∑j=1n(Sj+Ej+Ij+Hj+Rj+Dj), Λ¯=∑j=1nΛj and ρ=min{ρ1,ρ2,…,ρn}. Then N(t)≤(N(0)−Λ¯ρ)Eα(−ρtα)+Λ¯ρ. Therefore, the region Ω is positively invariant for system (3).  □ 3.2 Local stability The exploration of the existence and local stability of the disease-free equilibrium point is the focus of this section. Theorem 3.2 There are a unique disease-free equilibrium point E0=(S1∗,0,0,0,0,…,Sn∗,0,0,0,0) for system (3) where S∗=(S1∗,…,Sn∗) , S∗=A−1Λ , Λ=(Λ1,…,Λn) and A=ρ1+∑j≠1nmj1−m12⋯−m1n−m21ρ2+∑j≠2nmj2⋯−m2n⋮⋮⋮⋮−mn1−mn2⋯ρ2+∑j≠nnmjn. Proof Obviously, E0 satisfies the following equation: Λk−ρkSk∗+∑j=1n(mkjSj∗−mjkSk∗)=0, then the above equation can be written as the following matrix form: AS∗=Λ. It can be found that the matrix A is strictly diagonally dominant, and then it follows from [36] that one has A−1≥0. So according to [37], there exists a unique solution S∗=A−1Λ. Therefore, there exists a unique disease-free equilibrium point E0 of system (3). □ The predicted number of secondary cases that a typical infectious individual should create in a community that is totally susceptible is known as the basic reproduction number R0. According to Watmough et al. [38], it can be determined that an infectious disease can commonly infect the community if one diseased individual can typically infect more than one susceptible individual when R0≥1. On the other hand, if R0<1, each infected individual produces less than one new infection, and the infectious diseases can not grow. Thus, it is very important to describe the relationship between the basic reproduction number and the spread of infectious diseases. Here, the basic reproduction number R0 is stated as follows. Theorem 3.3 Under hypothesis H, the basic reproduction number R0 is R0=ρ(F11V11−1−F12V11−1V21V22−1), where matrixes F11=diag(β21b1S1,…,β2nbnSn) , F12=diag(β11a1S1,…,β1nanSn) , V21=diag(−ϵ1,…,−ϵn) , V11=ϵ1+∑j≠1nn1j−n12⋯−n1n−n21ϵ2+∑j≠2nn2j⋯−n2n⋮⋮⋮⋮−nn1−nn2⋯ϵn+∑j≠nnnnj, and V22=δ1+∑j≠1np1j−p12⋯−p1n−p21δ2+∑j≠2np2j⋯−p2n⋮⋮⋮⋮−pn1−pn2⋯δn+∑j≠nnpnj. Proof Let consider the following matrixes: F0=β11S1f1(I1)+β21S1gk(E1)⋮β1nSnfn(In)+β2nSngn(En)0⋮00⋮0andV0=ϵ1E1−∑j=1n(n1jEj−nj1E1)⋮ϵnEn−∑j=nn(nnjEj−njnEn)−ϵ1E1+δ1I1−∑j=1n(p1jIj−pj1I1)⋮−ϵnEn+δnIn−∑j=nn(pnjIj−pjnIn)−δ1I1+(λ1+κ1)H1⋮−δ1In+(λn+κn)Hn. Let u=(E1,…,En,I1,…,In,H1,…,Hn), then take the derivative of F0 and V0 for u at the disease-free equilibrium point E0, respectively, we can see as follows: F=F11F120000000andV=V1100V21V2200V32V33, where F11=diag(β21b1S1∗,…,β2nbnSn∗), F12=diag(β11a1S1∗,…,β1nanSn∗), V21=diag(−ϵ1,…,−ϵn), V33=diag((λ1+κ1),…,(λn+κn)), V32=diag (−δ1,…,−δn), V11=ϵ1+∑j≠1nn1j−n12⋯−n1n−n21ϵ2+∑j≠2nn2j⋯−n2n⋮⋮⋮⋮−nn1−nn2⋯ϵn+∑j≠nnnnj, and V22=δ1+∑j≠1np1j−p12⋯−p1n−p21δ2+∑j≠2np2j⋯−p2n⋮⋮⋮⋮−pn1−pn2⋯δn+∑j≠nnpnj. Then according to [39], the basic reproduction number is as follows: R0=ρ(FV−1)=ρ(F11V11−1−F12V11−1V21V22−1), where ρ(F11V11−1−F12V11−1V21V22−1) is the spectral radius of the matrix (F11V11−1−F12V11−1V21V22−1). □ Remark 3.1 According to [40], the epidemic size ςk=Sk(0)−Sk∗ of city k is defined as the number of individuals affected by the infectious disease, where Sk(0) is initial condition and Sk∗ is the disease-free equilibrium point of susceptible individuals within city k. Remark 3.2 When individual migration is not taken into consideration, it can be calculated from [10] that the basic reproduction number R0uk of city k is R0uk=Sk∗(β2kbkϵk+β1kakδk). Remark 3.3 When individual migration is taken into consideration, R0k of city k is R0k=Sk∗ϵk+∑j⁄=knnkj(β2kbk+β1kakϵkδk+∑j≠knpkj). Remark 3.4 It is easy to see that R0k are not dependent on λk, κk and mkj. Like [41], the other Λk, β1k, β2k, ρk, ϵk, nkj, pkj and δk are calculated as follows: AΛk=ΛkR0k∂R0k∂Λk=1,Aρk=ρkR0k∂R0k∂ρk=−1, Aβ1k=β1kR0k∂R0k∂β1k=β1kakϵkδk+∑j≠knpkjβ2kbk+β1kakϵkδk+∑j≠knpkj,Aβ2k=β2kR0k∂R0k∂β2k=β2kbkβ2kbk+β1kakϵkδk+∑j≠knpkj, Aδk=δkR0k∂R0k∂δk=−1(δk+∑j≠knpkj2)(β2kbk+β1kakϵkδk+∑j≠knpkj),Aϵk=ϵkR0k∂R0k∂ϵk=−1ϵk+∑j≠knnkj(β2kbk+ϵk+∑j≠knnkjδk+∑j≠knpkj), Ankj=nkjR0k∂R0k∂nkj=−1,Apkj=pkjR0k∂R0k∂pkj=−1(δk+∑j≠knpkj2)(β2kbk+β1kakϵkδk+∑j≠knpkj), where AΛk, Aρk, Aβ1k, Aβ2k, Aδk, Aϵk, Ankj and Apkj represent the normalized sensitivity on Λk, ρk, β1k, β2k, δk, ϵk, nkj and pkj, respectively. Through the above calculation found that the increase on Λk, β1k and β2k leads to the increase on R0k, but the increase on ρk, δk, ϵk, nkj and pkj leads to the decrease on R0k. In addition, the movement of susceptible individuals has no impact of R0k, but the movement of exposed and infected individuals is negatively correlated with R0k, and the movement of exposed individuals is more likely to influence the spread of the infectious disease with |Ankj|>|Apkj|. Theorem 3.4 Under hypothesis H, system (3) is locally asymptotically stable at the disease-free equilibrium point E0 if |arg(sF−V)|>απ2 . Proof The following Jacobian matrix at the disease-free equilibrium point E0 is considered: JE0=J11∗00F−V00∗J33, where matrixs J11=−ρ1−∑j=1nmj1m12⋯m1nm21−ρ2−∑j=1nmj2⋯m2n⋮⋮⋮⋮mn1mn2⋯−ρn−∑j=1nmjn, J33=−∑j=1nqj1q12⋯q1nq21−∑j=1nqj2⋯q2n⋮⋮⋮⋮qn1qn2⋯−∑j=1nqjn, F and V see Theorem 3.3. Then if all eigenvalues of the Jacobian matrix JE0 satisfy |arg(si)|>απ2, E0 is locally asymptotically stable and unstable if for some eigenvalues si, |arg(si)|≤απ2. Obviously, J11 and J33 are a nonsingular M-matrix, so J11 and J33 has all eigenvalues with negative real parts according to [42]. Consequently the local stability of E0 depends only on eigenvalues of F−V. Thus, if all eigenvalues of F−V satisfy |arg(sF−V)|>απ2, system (3) is locally asymptotically stable. □ Remark 3.5 If all the eigenvalues of F−V are negative, that is |arg(sF−V)|=π>απ2, system (3) is locally asymptotically stable. Meanwhile, it is obvious that |arg(sF−V)|=π⇔sF−V<0⇔ρ(FV−1)<1⇔R0<1. It can be yielded that if R0<1, the disease-free equilibrium point E0 is locally asymptotically stable of system (3) . 3.3 Global asymptotic stability of the disease-free equilibrium In this subsection, the global asymptotic stability of the disease-free equilibrium point E0 is discussed firstly. Furthermore, the uniform persistence of system (3) is also considered. Theorem 3.5 Under hypothesis (H) and |arg(sF−V)|>απ2 , the disease-free equilibrium point E0 is globally asymptotically stable of system (3) . Proof We use a method similar to the one used in [43]. Firstly, the boundedness of the susceptible class will be analyzed. According to Theorem 3.1 and hypothesis (H), we know Sk, Ek and Ik (k=1,2,…,n) are nonnegative, thus one has (5) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk +∑j=1n(mkjSj−mjkSk)≤Λk−ρkSk+∑j=1n(mkjSj−mjkSk). Let S=(S1,…,Sn), S∗=(S1∗,…,Sn∗), Λ=(Λ1,…,Λn) and A=ρ1+∑j≠1nmj1−m12⋯−m1n−m21ρ2+∑j≠2nmj2⋯−m2n⋮⋮⋮⋮−mn1−mn2⋯ρ2+∑j≠nnmjn, then Eq. (5) can be written in the following matrix: 0CDtαS≤Λ−AS=AS∗−AS, so it is easy to see that the conclusion holds as follows: S(t)≤(S0−S∗)Eα(−Atα)+S∗. Obviously, one has Sk≤Sk∗. Next, the global stability of Ek, Ik and Hk will be discussed. Based on hypothesis (H), one has fk(Ik)≤akIk and gk(Ek)≤bkEk. Then the following auxiliary system is considered: (6) 0CDtαEk¯=β1kSk∗akIk¯+β2kSk∗bkEk¯−ϵkEk¯+∑j=1n(nkjEj¯−njkEk¯),0CDtαIk¯=ϵkEk¯−δkIk¯+∑j=1n(pkjI¯j−pjkIk¯),0CDtαHk¯=δkIk¯−(λk+κk)Hk¯. It is easy to see that (7) 0CDtαW=(F−V)W, where W=(E¯,I¯,H¯), E¯=(E1¯,…,En¯), I¯=(I1¯,…,In¯), H¯=(H¯1,…,H¯n), F and V see Theorem 3.3. Thus, if |arg(sF−V)|>απ2, the above linear system (7) is locally asymptotically stable as well as globally asymptotically stable, that is limt→∞E¯k=limt→∞I¯k=limt→∞H¯k=0. According to the comparison theory and the nonnegative solution of Ek, Ik and Hk, one has limt→∞Ek=limt→∞Ik=limt→∞Hk=0. Based on the above analysis, when t→∞, one has 0CDtαS=AS∗−AS, then one has S(t)→S∗(t→∞). So E0 is globally asymptotically stable if |arg(sF−V)|>απ2. □ Remark 3.6 Similar to Theorem 3.3, it can be concluded that if R0<1, system (3) is globally asymptotically stable at the disease-free equilibrium point E0. Furthermore, the uniform persistence for system (3) is discussed in the following theorem. Theorem 3.6 Under hypothesis (H) and R0>1 , system (3) is uniformly persist, implying there exists a positive constant δ such that lim inft→+∞Sk≥δ,lim inft→+∞Ek≥δ,lim inft→+∞Ik≥δ,lim inft→+∞Hk≥δ,lim inft→+∞Rk≥δ,1≤k≤n. Proof Let consider the following space: X=X1×X2×⋯×Xn,X0=X10×X20×⋯×Xn0,∂X=∂X1×∂X2×⋯×∂Xn, where X0 represents the interior of X, ∂X denotes the boundary of X and Xk={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek≥0,Ik≥0,Hk≥0,Rk≥0}, Xk0={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek>0,Ik>0,Hk>0,Rk>0}, ∂Xk={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek=0,Ik=0,Hk=0,Rk=0}. Meanwhile, let W(t)=(S1,E1,I1,H1,R1,…,Sn,En,In,Hn,Rn) be the solution of system (3) with initial value W(0)=W0∈X, then W(t)∈X according to Theorem 3.1. For any t≥0, a continuous map F(t):X→X is defined as follows: F(t)W0=W(t). In the following, the uniformly persistent of the map F will be analyzed based on Lemma 2.3. When t=0, one has F(0)W0=W(0), this is F(0)=I where I is the identity matrix. Meanwhile, it can be deduced that the following equation holds: F(t+s)W0=W(t+s)=F(t)W(s)=F(t)F(s)W0, implying F(0)=I. Additionly, it is easy to see that F(t) is C0-semigroup on X, point dissipative and compact in X. Furthermore, consider the following system: 0CDtαSk=Λk−ρkSk+∑j=1n(mkjSj−mjkSk). According to Theorem 3.3, Sk∗ is asymptotically stable, which finds that E0 in ∂X is a global attractor of F(t). Let M={M1}, where M1={E0}. Because of ak=fk′(0) and bkgk′(0), for all ϵ, there exists ϵ¯ that f(ϵ)>(ak−ϵ)ϵ¯andg(ϵ)>(bk−ϵ)ϵ¯. Let the stable set Ws(E0) of a compact invariant set E0 defined by Ws(E0)={Y0∈X:ω(Y0)≠0̸,ω(Y0)∈E0}, where ω(Y0) is ω-limit set through Y0. System (3) has a solution (Sk,Ek,Ik,Hk,Rk) when Ws(E0)∩X0≠0̸, implying Sk→0, Ek→0, Ik→0, Hk→0, Rk→0 (k=1,2,…,n) as t→∞. So there exists a constant τ>0 such that Sk>Sk∗−ϵ, Ek>ϵ, Ik>ϵ, Hk>ϵ and Rk>ϵ for t≥τ. Then according to the monotonicity of fk(Ik) and gk(Ek), one has f(Ik)>f(ϵ)>(ak−ϵ)ϵ¯andg(Ek)>g(ϵ)>(bk−ϵ)ϵ¯. So the following auxiliary system is considered: (8) 0CDtαEk_=β1kSk∗(ak−ϵ)ϵ¯Ik_+β2kSk∗(bk−ϵ)ϵ¯Ek_−ϵkEk_+∑j=1n(nkjEj_−njkEk_),0CDtαIk_=ϵkEk_−δkIk_+∑j=1n(pkjIj_−pjkIk_),0CDtαHk_=δkIk_−(λk+κk)Hk_. It is easy to see from system (8) that (9) 0CDtαW=(F−V)(ϵ¯,ϵ)W, where W=(E_,I_,H_), E_=(E_1,…,E_n), I_=(I_1,…,I_n) and H_=(H_1,…,H_n). Consider the basic reproduction number R0>1, then one has ρ(F11V11−1−F12V11−1V21V22−1)(ϵ¯,ϵ)>1, which results in a contradiction with Ek(t)→0 (t→∞). Hence one has Ws(E0)∩X0=0̸, implying it is uniformly persistent at the operator T(t), so system (3) is uniformly persistent if R0>1.  □ The existence of a positive equilibrium point is implied by the system (3)’s ultimate boundedenss and uniform persistence. As a result, we can derive the following theorem. Theorem 3.7 Under hypothesis (H) and R0>1 , there is at least one endemic equilibrium E∗=(S1∗,E1∗,I1∗,H1∗,R1∗,…,Sn∗,En∗,In∗,Hn∗,Rn∗) of system (3) satisfying Λk−β1kSk∗fk(Ik∗)−β2kSk∗gk(Ek∗)−ρkSk∗+∑j=1n(mkjSj∗−mjkSk∗)=0,β1kSk∗fk(Ik∗)+β2kSk∗gk(Ek∗)−ϵkEk∗+∑j=1n(nkjEj∗−njkEk∗)=0,ϵkEk∗−δkIk∗+∑j=1n(pkjIj∗−pjkIk∗)=0,δkIk∗−(λk−κk)Hk∗=0,λkHk∗+∑j=1n(qkjRj∗−qjkRk∗)=0. 4 Numerical simulation From the previous description, it is clear that E0 is globally asymptotically stable when R0<1 and conversely, system (3) is persistent, which can offer theoretical evidence for further COVID-19 prediction and control. Meanwhile, in order to analyze COVID-19 in different cities, this section is divided into two parts: no restrictions on individual migration and restrictions on individual migration. Furthermore, consider the corresponding integer-order model as follows: (10) dSkdt=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),dEkdt=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),dIkdt=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),dHkdt=δkIk−(λk+κk)Hk,dRkdt=λkHk+∑j=1n(qkjRj−qjkRk). 4.1 Data source The Johns Hopkins University Center for System Science and Engineering provided the real data for this study [1]. Data on accumulated and confirmed cases, recovered cases, and COVID-19 death cases were shared by the Johns Hopkins University on January 23, 2020. Assuming that the confirmed individuals mut be hospitalized, one has Hospitalized=Confirmed−Recovered−Death. Hence, we can get the real data of H(t), D(t) and R(t) for different city from 23 January to 17 July, 2020. 4.2 The generalized incidence rate As we know, Korobeinikov et al. [8] indicated that the stability of the endemic equilibrium point for infectious diseases is closely related to the concave of the incidence rate with respect to the infected individuals. Therefore, it is of practical significance to understand the role of different incidence rates in COVID-19. In this section, according to hypothesis (H), the bilinear incidence rate and the saturation incidence rate are discussed as follows: fk(Ik)=Ik,gk(Ek)=Ek. fk(Ik)=Ik1+ukIk,gk(Ek)=Ek1+vkEk. Meanwhile, as the public learns about COVID-19, the recovered rate and the disease-related mortality are time-varying rather than constant. Similar to [11], the best recovered rate λk and the best disease-related mortality κk are selected from the following equation: (11) κk=p1ep2(t−p3)+e−p2(t−p3),p1e(p2(t−p3))2,p1+e(p2(t+p3)),andλk=q11+e−q2(t−q3),q1+e−q2(t+q3), where qi and qi (i=1,2,3) are parameters for κk and λk, respectively. According to the real data reported by [2], the spread of COVID-19 in India and Brazil began on 30 January and 26 February, 2020, as the beginning of the outbreak of India and Brazil in this paper, respectively. According to Matlab function lsqcurvefit [11], the parameter identification results with system (3) and system (10) are depicted in Table 1, Table 2, respectively. Meanwhile, based on Table 1, Table 2, the five days forecast of India and Brazil are shown in Tables 3, 4, Figs. 2, 3, 4, 5, which the solid lines represent simulation results and circles represent real data. The results in Table 1, Table 2 show that the fractional-order system (3) can accurately forecast the real data in the upcoming week, with the real data of currently confirmed cases falling between 95% and 105% of the projected values. Table 1 Parameter identification of India. India Integer (Bilinear) Fractional (Bilinear) Integer (Saturation) Fractional (Saturation) Λ 0.3 0.6245 0.465 0.3 β1 0.3869 1.206 1.156 2.358 β2 0.5133 0.3079 0.3414 0.8 ϵ 0.0023 0.0555 0.0014 0.0692 ρ 0.0264 0.03 0.0188 0.0094 λ p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) κ b1e−q2(t−q3)2 q1e−q2(t−q3)2 q1e−q2(t−q3)2 q1e−q2(t−q3)2 Table 2 Parameter identification of Brazil. Brazil Integer (Bilinear) Fractional (Bilinear) Integer (Saturation) Fractional (Saturation) Λ 0.3 0.5 0.8802 0.5189 β1 1.839 1.082 0.2378 4.804 β2 0.3733 0.928 0.4397 0.3 ϵ 0.0303 0.7905 0.1076 0.9609 ρ 0.0211 0.0214 0.0216 0.0431 δ 0.99 0.2434 0.3975 5.609×10−5 λ p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) κ q1eq2(t−q3) q1eq2(t−q3) q1eq2(t−q3) q1eq2(t−q3) Table 3 Estimate the number of confirmed cases within five days in India (×105). India 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 3.735 3.906 4.027 4.113 4.263 Integer (bilinear incidence rate) 3.454 3.499 3.52 3.556 3.583 Fractional (bilinear incidence rate) 3.781 3.869 3.936 4.002 4.079 Integer (saturation incidence rate) 3.426 3.476 3.515 3.553 3.578 Fractional (saturation incidence rate) 3.645 3.696 3.741 3.792 3.815 Table 4 Estimate the number of confirmed cases within five days in Brazil (×105). Brazil 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 5.487 5.598 5.242 5.228 5.528 Integer (bilinear incidence rate) 5.864 5.847 5.828 5.806 5.781 Fractional (bilinear incidence rate) 5.502 5.49 5.475 5.458 5.438 Integer (saturation incidence rate) 5.897 5.882 5.864 5.853 5.821 Fractional (saturation incidence rate) 5.857 5.844 5.819 5.793 5.769 Fig. 2 The number of cases in India (Integer-order with the bilinear incidence rate (left), Fractional-order with the bilinear incidence rate (right)). Fig. 3 The number of cases in India (Integer-order with the saturation incidence rate (left), Fractional-order with the saturation incidence rate (right)). Fig. 4 The number of cases in Brazil (Integer-order with the bilinear incidence rate (left), Fractional-order with the bilinear incidence rate (right)). Fig. 5 The number of cases in Brazil (Integer-order with the saturation incidence rate (left), Fractional-order with the saturation incidence rate (right)). 4.3 Restrict individual migration When individual movement is not considered, the parameters satisfy mkj=nkj=pkj=qkj=0. Meanwhile, according to Section 4.2, the bilinear incidence rate is considered in this section. Then based on system (1), the following auxiliary system is considered: (12) 0CDtαS=Λ−β1SIβ2SE−ρS,0CDtαE=β1SI+β2SEϵE,0CDtαI=ϵE−δI,0CDtαH=δI−(λ+κ)H,0CDtαR=λH,0CDtαD=κH. 4.3.1 Sensitivity analysis of parameters in R0uk When individual movement is not taken into consideration in this section, Partial Rank Correlation Coefficients (PRCC) value and Latin hypercube sampling (LHS) [44], which are one of the Monte Carlo (MC) sampling methods established by Mckay in 1979 [45], can be used to account for the sensitivity of the parameter to the basic reproduction number. LHS has the advantage of using fewer iterations than other random sampling techniques and avoiding the clustering phenomenon of sampling [45]. In order to determine which aspects of a certain intervention have the greatest impact on how quickly a new infection spreads, it can be seen from Remark 3.2 that the parameters of system (12) all affect the basic reproduction number to varying degrees, thereby affecting the spread of the infectious disease. We perform LHS on the parameters that appear in R0uk. PRCC are calculated, and a total of 1000 simulations per LHS run are carried out. A uniform distribution is chose as the prior distribution when performing parameter sampling. The parameters Λ, ρ, ϵ, β1, β2 and δ of system (12) are set as input variables, and the basic reproduction number R0uk as the output. The specific process is as follows: (1) There are six parameters that affect the change of R0uk, which are Λ, ρ, ϵ, β1, β2 and δ. Through LSH, [0,1] is divided into 1000 simulations, and 6 × 1000 parameters are generated through random selection on each interval by a uniform distribution. (2) Calculate the basic reproduction number R0uk for each parameter. (3) PRCC is calculated by Matlab function partialcorr. (4) The PRCC’s influence on the basic reproduction number R0uk can increase with increasing PRCC absolute value. However, it is believed that the parameter is not significant if the p value is greater than 0.05. Table 5 lists the PRCC values of the six parameters associated with R0uk and Fig. 6 shows the histogram of PRCC value. From Table 5 and Fig. 6, the following conclusion holds: (1) the parameters Λ, β1 and β2 have a positive influence on R0uk, but ρ, ϵ and δ have a negative influence, which is consistent with Remark 3.4; (2) the positive impact of birth rate Λ is the most obvious with PRCC(Λ)=0.5868; (3) the positive impact of the transmission rate β2 for the exposed population is more obvious than that of the infected population with PRCC(β2)>PRCC(β1). That is, the greater the transmission coefficient of the exposed population, the greater the value of the basic reproduction number R0uk, and then the greater the number of people infected with COVID-19. Therefore, it is more critical to limit exposed individual. However, because exposed individual doed not show any symptoms, identifying them is very difficult, which is a key reason for the spread of COVID-19; (4) the diagnosis rate δ has more greater negative impact on R0uk. That is to say, enhancing nucleic acid detection can effectively reduce R0ui, thereby reducing the number of infected people; (5) from the p-value, it can be found that the p-values of all parameters are less than 0.05, so they all have a significant impact on the basic reproduction number R0uk. Therefore, based on the above analysis, it can be obtained that controlling the influx of foreign population and enhancing nucleic acid detection are the most effective measures to control COVID-19. Meanwhile, home isolation can also control COVID-19. Therefore, this evidence confirms the effectiveness of Chinese government’s interruption policies, such as home isolation, prohibition of the inflow of foreign population, and enhancing nucleic acid detection, which may provide a good reference for the other countries. Table 5 The PRCC values and p-value of the parameters with respect to R0uk. Input PRCC values p-value Λ 0.5868 0 ρ −0.5363 0 ϵ −0.1368 0 β2 0.1035 3.5×10−6 β1 0.0847 1.448×10−4 δ −0.4362 0 Fig. 6 The sensitivity analysis of R0uk. 4.3.2 China’s second outbreak From the analysis in Section 4.3.1, it can be found that enhancing the diagnosis rate and controlling the inflow of foreign population can effectively control the spread of the epidemic. For China, individual migration has been strictly restricted at the beginning of COVID-19. Therefore, the impact of enhanced diagnosis rate will be only considered in this section. Due to the increase in public awareness and the development of detection technology, the time from onset to diagnosis is gradually shortened. Additionally, despite the use of the nucleic acid test method, the number of confirmed cases climbed significantly and peaked in early February 2020 as a result of the use of the CT diagnosis method. As a result, it is assumed that starting on 12 February, 2020, China’s diagnosis rate can reach and remain at its highest level. However, the third COVID-19 wave has been occurring in Beijing since the end of June 2020. Beijing has said that starting on 17 June, 2020, nucleic acid could be more readily detected. As a result, a new distribution, rather than the max level dated June 17, now governs the diagnostic rate. Similar to [46], the following piecewise function are described the diagnosed period of two and three peaks: (13) 1δk=(1δ0−1δe)e−w1t+1δe,t<t1,1δe,t≥t1,and 1δk=(1δ0−1δe)e−w1t+1δe,t<t1,1δe,t1≤t≤t2,(1δe−1δf)e−w2(t−t2)+1δf,t>t2, where δ0, δe (δe>δ0), w1, w2 and δf are similar to [46], t1 is 13 February, 2020, t2 is 17 June, 2020. Meanwhile, similar to [11], the best recovered rate λk and the best disease-related mortality κk are selected from Eq. (11). Then system (12) and system (10) are solved by predictor–correctors scheme and least squares method [11] by the real data from 23 January to 17 July, which 17 March, 2020 is considered as the beginning of the emergency in Heilongjiang, Shanghai and Guangdong, and 17 June, 2020 are considered as the beginning of the emergency in Beijing, respectively. From Fig. 7, Fig. 8, the fractional-order system (12) is found to fit the real data more accurately than the integer-order system (10) does. and COVID-19 in Beijing, Shanghai reaches its highest peak in a short time but there may be fourth wave peak, however, Heilongjiang and Guangdong are only two peaks and the third wave of epidemic peaks will not occur in a short time (current policies remain unchanged). Therefore, under the condition of restricting the migration of individuals, the fractional system (12) can better simulate the multi-peak problem of COVID-19, and the strengthening of nucleic acid detection can predict the new wave in advance, which provides a theoretical basis for the control of the epidemic. Fig. 7 The number of cases in Beijing and Shanghai. Fig. 8 The number of cases in Guangdong and Heilongjiang. 4.4 Individual migration As of 17 July, 2020, the United States has a total of 3,647,715 confirmed cases, 139266 deaths and 1,107,204 recovery cases. It is urgent to formulate reasonable and effective mitigation measures. Thus in this section, based on the sensitivity analysis of parameter to R0k, the effect mitigation measures are provided to control the development of COVID-19 in US. 4.4.1 Sensitivity analysis of parameters in R0k Similar to Section 4.3.1, consider two cities to examine the sensitivity of parameters to R0k (k=1,2). Then when n=2, system (3) can be simplified as follows: (14) 0CDtαS1=Λ1−β11S1I1−β21S1E1−ρ1S1+(m12S2−m21S1),0CDtαE1=β11S1I1+β21S1E1−ϵ1E1+(n12E2−n21E2),0CDtαI1=ϵ1E1−δ1I1+(p12I2−p21I1),0CDtαH1=δ1I1−(λ1+κ1)H1,0CDtαR1=λ1H1+(q12R2−q21R1),0CDtαS2=Λ2−β12S2I2−β22S2E2−ρ2S2+(m21S1−m12S2),0CDtαE2=β12S1I2+β22S2E2−ϵ1E2+(n21E1−n12E1),0CDtαI2=ϵ2E2−δ2I2+(p21I1−p12I2),0CDtαH2=δ2I2−(λ2+κ2)H2,0CDtαR2=λ2H2+(q21R1−q12R2). It can be found from Remark 3.4 that there exists 16 parameter of the basic reproduction number R0k (k=1,2), and then the 16 parameters are set as input variables, and R0k as the output. Similar to Section 4.3.1, Table 6 lists the PRCC values and Fig. 9 shows the histogram of PRCC value. According to Table 6 and Fig. 9, it can be found that the following conclusion holds: (1) the movement of susceptible individuals mkj (k,j=1,2) does not affect R0k; (2) the sensitivity of the parameter to R0k (k=1,2) is same as that of Section 4.2.1, except for n12, n21, p12 and p21; (3) considering the basic reproduction number R01 of city 1, the p-value of p21 is large than 0.05, which means that infected individuals migrating from city 1 have a significant impact on COVID-19 in city 1. But exposed and infected individuals migrating to city 1 have an impact on the spread of COVID-19 in city 1, and the impact of the inflow of exposed individuals is more significant because of |PRCC(n12)|>|PRCC(p12)|; (4) contrary to the situation in city 1, the p-value of and p12 is large than 0.05, which means that infected individuals migrating from city 2 have a significant impact on the spread of disease in city 2. But exposed and infected individuals migrating from city 2 have an impact on the spread of COVID-19 in city 2, and the impact of the inflow of exposed individuals is more significant because of |PRCC(n21)|>|PRCC(p21)|. Therefore, in order to alleviate the situation in severe areas of COVID-19, migration of exposed individuals must be strictly controlled. Table 6 The PRCC values and p-value of the parameters with respect to R01 (left) and R02 (right). Input PRCC values p-value Λ1 0.6513 0 ρ1 −0.5294 0 ϵ1 −0.0348 0.1194 β21 0.1983 0 β11 0.0564 0.0116 δ1 −0.1427 0 n12 −0.3584 0 n21 −0.0286 0.2005 p12 −0.1732 0 p21 −0.0062 0.7806 Input PRCC values p-value Λ2 0.6496 0 ρ2 −0.5223 0 ϵ2 −0.0451 0.0435 β22 0.1896 0 β12 0.052 0.0153 δ2 −0.1691 0 n12 −0.0217 0.3318 n21 −0.2732 0 p12 −0.0048 0.8316 p21 −0.1355 0 Fig. 9 The sensitivity analysis of R01 (left) and R02 (right). 4.4.2 US outbreak In this subsection, the overall spread of COVID-19 in the US is considered first. Then system (10) and system (12) are solved by least squares method [11]. However, beginning 17 May, 2020, the number of confirmed individuals in the US had significantly increased. Emergency situations may have changed government regulations and people’s attitudes, which led to an increase in the number of sick people. Therefore, it is assumed that the emergency starts on 17 May, and the outbreak’s spread in the US is then examined in two stages as follows: (1) 23 January-17 May, 2020; (2) 17 May-17 July, 2020. Therefore, parameter identification is provided in Table 7 based on actual data from 23 January to 17 July 2020. From Fig. 10 and Table 8, it is clear that the fractional-order system (12) is capable of accurately forecasting the confirmed case for the upcoming week. In the meantime, Table 8 shows that, regardless of whether in the first stage or the second stage, the parameter findings of the fractional-order system and integer-order fitting are totally different. Based on Remark 3.2, R0uk=49.84 is very high. From the analysis in Section 4.2.1, it can be found that enhancing the diagnosis rate, reducing contact with infected people and controlling the inflow of foreign population can effectively control the spread of COVID-19. However, the United States is not currently doing anything to limit the influx of foreign population, so it is only considering enhancing nucleic acid testing and reducing contact with infected people to control COVID-19. Like [46], the diagnosed period 1δk of US are as follows: (15) 1δ(t)=1δet≤t3,(1δ0−1δe)e−w(t−t3)+1δet>t3. The meaning of each symbol is similar to that in Section 4.2.3. t3 is 17 July, 2020, which mean increasing the diagnosis rate δ(t) from 17 July, 2020. At the same time, the contact rate βi (i=1,2) is limited by the number of hospitalizations like [46] as follows: (16) βi(t)=βi,logH(t)≤1,βilogH(t),logH(t)>1. It can be seen from Fig. 10 that increasing the diagnosis rate δ(t) and controlling the infection rate βi (i=1,2) can effectively contain COVID-19. Therefore, enhanced nucleic acid testing and limited contact with infected individuals are important to control COVID-19.Table 7 Parameter identification of US. US Integer (first stage) Fractional (first stage) Integer (second stage) Fractional (second stage) Λ 0.3 0.0935 0.3 0.4593 β1 1.056 1.217 4.999 2.641 β2 0.2969 0.4882 1.648×10−7 3.724×10−7 ϵ 0.1791 0.2876 0.0087 0.0077 ρ 0.0281 0.0497 0.0358 0.0272 δ 0.1243 0.2434 0.3975 0.205 λ p1+e−p2(t+p3) p1+e−p2(t+p3) p1+e−p2(t+p3) p1+e−p2(t+p3) κ q1eq2(t−q3)+e−q2(t−q3) q1eq2(t−q3)+e−q2(t−q3) q1+e−q2(t+q3) q1+e−q2(t+q3) Table 8 Estimate the number of confirmed cases within five days in US (×106). Date Real data Fractional Integer 18 July 2.449 2.503 2.397 19 July 2.502 2.541 2.425 20 July 2.534 2.578 2.454 21 July 2.575 2.616 2.482 22 July 2.617 2.655 2.511 Fig. 10 The number of cases in US (without control (left), with control (right)). 4.4.3 US with individual migration This subsection considers the impact of individual migration on COVID-19. We need to preprocess the data to remove data that are less than 0.5% of the current maximum number of confirmed cases. Therefore, the real data after 3 April are selected to identify the parameters of system (14). Similar to the analysis of Section 4.4.2, we consider 17 May, 2020 as the beginning of the emergency, and the COVID-19 spread in New York and Los Angeles into two phases: (1) 3 April-17 May, 2020; (2) 17 May-17 July, 2020. Meanwhile, the recovered data of New York and Los Angeles have not been collected by [1], and then we take hospitalized+recovered individuals as a whole to conduct parameter identification and short-term prediction according to [11]. It can be found from Table 9, Table 10 and Fig. 11 that system (14) can better predict COVID-19. Meanwhile, it can be seen from Fig. 11 that the COVID-19 in New York has been peaked but not in Los Angles. From the analysis of Section 4.4.1, we know that controlling the infection rate, improving the diagnosis rate and controlling the movement of exposed individuals have a significant effect on the control of COVID-19 in US. Therefore, similar to Section 4.4.2, the diagnosis rate δk and the migration rate nkj are utilized as follows: (17) 1δk(t)=1δe,t≤t3,(1δ0−1δe)e−w(t−t3)+1δe,t>t3,and nkj=nkj,log(Hk)<1,nkjlog(Hk),log(Hk)≥1. Meanwhile, the infection rate controlled by the number of hospitalizations is Eq. (16). From Fig. 12, we can seen the fractional-order system (14) with control (Eqs. (16), (17)) in Los Angles can be control quickly but not in New York, which is still an open question and will be discussed later.Table 9 Estimate the number of confirmed cases within five days in New York (×105). New York 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 1.977 1.98 1.983 1.987 1.99 Integer 1.978 1.979 1.98 1.981 1.981 Fractional 1.985 1.986 1.987 1.988 1.989 Table 10 Estimate the number of confirmed cases within five days in Los Angles (×105). Los Angles 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 1.491 1.518 1.549 1.579 1.609 Integer 1.414 1.444 1.464 1.482 1.514 Fractional 1.494 1.517 1.547 1.579 1.608 Fig. 11 The number of cases in New York and Los Angeles with individual movement. Fig. 12 The number of cases in New York and Los Angeles with control Eqs. (16), (17). 5 Conclusion Based on individual migration, a fractional-order SEIHRDP model is proposed with the generalized incidence rate. Meanwhile, some results and effective mitigation measures is suggested to control COVID-19 as follows: (1) The local and global asymptotic stability of the disease-free and endemic equilibrium points are investigated based on the basic reproduction number R0. (2) Based on the real data, it is found that the bilinear incidence rate has a better description of COVID-19 transmission than the saturation incidence rate. Therefore, the bilinear incidence rate is applied in modelling COVID-19. Meanwhile, this is the first time that looked at the impact of the incidence rate in the spread of COVID-19 using real data. (3) By applying the value of PRCC, the sensitivity of the parameters to the basic reproduction number R0k and R0uk are obtained, which is consistent with Remark 3.4. Through the PRCC value, the diagnosis rate, the migration rate and the movement of the infected population are most sensitive to control COVID-19. (4) Multiple peaks have been analyzed for COVID-19 and using four cities in China to show that the fractional-order system (1) works well. Moreover, by increasing the diagnosis rate, it can be found that the third wave of epidemic in Beijing has reached its peak, but the arrival of the next wave of COVID-19 is not ruled out. (5) Analyzing the situation in the United States, it can be seen that system (12) has better predictability than system (10). Meanwhile, by reducing the infection rate and increasing the diagnosis rate, the peak of the epidemic in the US can be accelerated. (6) Results show that the fractional-order system can accurately forecast the real data in the upcoming week when taking into account individual migration between two cities. By limiting the movement of exposed individuals, raising the diagnosis rate, and lowering the infection rate, Los Angeles’ peaks can appear and then decline immediately. Furthermore, this study makes several contributions to predict multi-peak of COVID-19 in China and suggestions on controlling epidemic in the US by changing certain parameters. Nevertheless, this research raises some issues that require more investigation, including how medical and other factors affect the spread of infectious diseases, how to properly administer vaccines, how network topology affects disease transmission and so on. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgment This work is supported by the Natural Science Foundation of Beijing Municipality [grant numbers Z180005], the National Nature Science Foundation of China [grant numbers 61772063] and the Fundamental Research Funds of the Central Universities [grant numbers 2020JBM074]. ==== Refs References 1 The Johns Hopkins University Center for System Science and Engineering, Data of accumulated and newly confirmed cases, recovered case and death case of COVID-19, URL https://github.com/CSSEGISandData/COVID-19. 2 Chan J. Yuan S. Kok K. To K. Chu H. Yang J. A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: a study of a family cluster Lancet 395 10223 2020 514 523 31986261 3 Huang C. Wang Y. Li X. Ren L. Zhao J. Hu Y. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China Lancet 395 10223 2020 497 506 31986264 4 Anderson R. Anderson B. May R. Infectious diseases of humans: Dynamics and control 1992 Oxford University Press 5 Bailey N. The mathematical theory of infectious diseases and its applications 1975 Charles Griffin Company Ltd 6 Upadhyay R. Pal A. Kumari S. Roy P. Dynamics of an SEIR epidemic model with nonlinear incidence and treatment rates Nonlinear Dynam 96 4 2019 2351 2368 7 Zou X. Wang K. Optimal harvesting for a stochastic regime-switching logistic diffusion system with jumps Nonlinear Anal Hybrid 13 2014 32 44 8 Korobeinikov A. Maini P. Nonlinear incidence and stability of infectious disease models Math Med Biol 22 2 2005 113 128 15778334 9 Peng L. Yang W. Zhang D. Zhuge C. Hong L. Epidemic analysis of COVID-19 in China by dynamical modeling 2020 10.1101/2020.02.16.20023465 arXiv preprint arXiv:2002.06563 10 Xu C. Yu Y. Chen Y. Lu Z. Forecast analysis of the epidemics trend of COVID-19 in the USA by a generalized fractional-order SEIR model Nonlinear Dynam 101 3 2020 1621 12634 11 Cheynet E. Generalized SEIR epidemic model (fitting and compution) 2020 GitHub URL http://www.github.com/ECheynet/SEIR 12 Gao D.Z. Cosner C. Cantrell R.S. Beier J.C. Ruan S.G. Modeling the spatial spread of Rift Valley fever in Egypt Bull Math Biol 75 3 2013 523 542 23377629 13 Phaijoo G.R. Gurung D.B. Mathematical study of dengue disease transmission in multi-patch environment Appl Math 7 14 2016 1521 1533 14 Eisenberg C.M. Shuai Z.S. Tien J.H. den Driessche P.V. A cholera model in a patchy environment with water and human movement Math Biosci 246 1 2013 105 112 23958383 15 Julien A. Davis J.R. David H. Richard J. Miller J.M. den Driessche P.V. A multi-species epidemic model with spatial dynamics Math Med Biol 2 2005 129 142 15778332 16 Cheng Z.K.J. Shan J. 2019 Novel coronavirus: where we are and what we know Infection 49 197 2021 17 Lu Z. Yu Y. Chen Y.Q. Ren G. Xu C. Wang S. A fractional-order SEIHDR model for COVID-19 with inter-city networked coupling effects Nonlinear Dynam 101 3 2020 1717 1730 18 Smethurst D. Williams H. Are hospital waiting lists self-regulating? Nature 410 6829 2001 652 653 11287943 19 Meerschaert M. Sikorskii A. Stochastic models for fractional calculus 2011 De Gruyter 20 Angstmann C. Henry B. Mcgann A. A fractional-order infectivity SIR model Physica A 2016 86 93 21 Khan M.A. Atangana A. Modeling the dynamics of novel coronavirus (2019-nCov) with fractional derivative Alex Eng J 59 4 2020 2379 2389 22 Chen Y. Cheng J. Jiang X. Xu X. The reconstruction and prediction algorithm of the fractional TDD for the local outbreak of COVID-19 2020 arXiv preprint arXiv:2002.10302 23 Amjad S.S. Najiroddin S.I. Kottakkaran N.S. A mathematical model of COVID-19 using fractional derivative: Outbreak in India with dynamics of transmission and control Adv Differ Equ 373 2020 1 19 32226454 24 Tang Z. Li X.B. Li H.Q. Prediction of new coronavirus infection based on a modified SEIR model MedRxiv 2020 10.1101/2020.03.03.20030858 25 Zhou P. Ma J. Tang J. Clarify the physical process for fractional dynamical systems Nonlinear Dynam 2020 1 12 26 Li H. Zhang L. Hu C. Jiang Y. Teng Z. Dynamical analysis of a fractional-order predator-prey model incorporating a prey refuge J Appl Math Comput 54 1–2 2017 435 449 27 Singh J. Kumar D. Hammouch Z. Atangana A. A fractional epidemiological model for computer viruses pertaining to a new fractional derivative Appl Math Comput 316 2018 504 515 28 Sierociuk D. Skovranek T. Macias M. Podlubny I. Petras I. Dzielinski A. Diffusion process modeling by using fractional-order models Appl Math Comput 257 2015 2 11 29 Podlubny I. Fractional differential equations 1999 Academic Press 30 Li Y. Chen Y. Podlubny I. Mittag–Leffler stability of fractional order nonlinear dynamic systems Automatica 45 8 2009 1965 1969 31 Tang Q. Teng Z. Jiang H. Global behaviors for a class of multi-group SIRS epidemic models with nonlinear incidence rate Taiwan J Math 19 5 2015 1509 1532 32 Tang Z. Li X. Li H. Prediction of new coronavirus infection based on a modified SEIR model Cold Spring Harbor Lab 2020 1 13 33 Lin M. Huang J. Ruan S. Yu P. Bifurcation analysis of an SIRS epidemic model with a generalized nonmonotone and saturated incidence rate J Differ Equ 267 3 2019 1859 1898 32226129 34 Muroya Y. Enatsu Y. Kuniya T. Global stability for a multi-group SIRS epidemic model with varying population sizes Nonlinear Anal Real 14 3 2013 1693 1704 35 Angstmann C.N. Donnelly I.C. Henry B.I. Langlands T. Continuous-time random walks on networks with vertex- and time-dependent forcing Phys Rev E 88 2 2013 022811 36 Li M. Graef J.R. Wang L.C. Karsai J. Global dynamics of a SEIR model with varying total population size Math Biosci 160 2 1999 191 213 10472754 37 Kheiri H. Jafari M. Stability analysis of a fractional order model for the HIV/AIDS epidemic in a patchy environment J Comput Appl Math 2018 323 339 38 van den Driessche P. Watmough J. Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmission Math Biosci 180 1–2 2002 29 48 12387915 39 Diekmann O. Heesterbeek J. Metz J. On the definition and the computation of the basic reproduction ratio R0 in models for infectious diseases in heterogeneous populations J Math Biol 28 4 1990 365 382 2117040 40 van den Driessche P. Watmough J. Heterogeneous social interactions and the COVID-19 lockdown outcome in a multi-group SEIR model Math Model Nat Phenom 15 36 2020 41 Khajanchi S. Bera S. Roy T.K. Mathematical analysis of the global dynamics of a HTLV-I infection model, considering the role of cytotoxic T-lymphocytes Math Comput Simulat 180 2021 42 Berman A. Plemmons R. Nonnegative matrices in the mathematical sciences 1979 Academic Press 43 Ruan S.G. Wang W.D. Levin S.A. The effect of global travel on the spread of SARS Math Biosci Eng 3 1 2012 205 218 44 Zhang K. Ji Y.P. Pan Q.W. Wei Y.M. Liu H. Sensitivity analysis and optimal treatment control for a mathematical model of Human Papillomavirus infection AIMS Math 5 5 2020 2646 2670 45 Huo H.F. Feng L.X. Global stability for an HIV/AIDS epidemic model with different latent stages and treatment Appl Math Model 37 3 2013 1480 1489 46 Tang B. Xia F. Tang S. Bragazzi N. Li Q. Sun X. The effectiveness of quarantine and isolation determine the trend of the COVID-19 epidemic in the final phase of the current outbreak in China Int J Infect Dis 96 2020 636 647 32689711

0

PMC9748852

NO-CC CODE

2022-12-15 23:22:47

no

ISA Trans. 2022 Dec 14; doi: 10.1016/j.isatra.2022.12.006

utf-8

ISA Trans

10.1016/j.isatra.2022.12.006

oa_other

==== Front ISA Trans ISA Trans ISA Transactions 0019-0578 1879-2022 ISA. Published by Elsevier Ltd. S0019-0578(22)00637-1 10.1016/j.isatra.2022.12.006 Article The effect mitigation measures for COVID-19 by a fractional-order SEIHRDP model with individuals migration Lu Zhenzhen a Chen YangQuan b Yu Yongguang a⁎ Ren Guojian a Xu Conghui a Ma Weiyuan c Meng Xiangyun a a Department of Mathematics, Beijing Jiaotong University, Beijing, 100044, PR China b Mechatronics, Embedded Systems and Automation Lab, University of California, Merced, CA 95343, USA c School of Mathematics and Computer Science, Northwest Minzu University, Lanzhou, 730000, PR China ⁎ Corresponding author. 14 12 2022 14 12 2022 5 8 2020 22 11 2022 10 12 2022 © 2022 ISA. Published by Elsevier Ltd. All rights reserved. 2022 ISA Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. In this paper, the generalized SEIHRDP (susceptible-exposed-infective-hospitalized-recovered-death-insusceptible) fractional-order epidemic model is established with individual migration. Firstly, the global properties of the proposed system are studied. Particularly, the sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. Secondly, according to the real data in India and Brazil, it can all be concluded that the bilinear incidence rate has a better description of COVID-19 transmission. Meanwhile, multi-peak situation is considered in China, and it is shown that the proposed system can better predict the next peak. Finally, taking individual migration between Los Angeles and New York as an example, the spread of COVID-19 between cities can be effectively controlled by limiting individual movement, enhancing nucleic acid testing and reducing individual contact. Keywords Individual migration Fractional-order epidemic model Peak prediction Sensitivity ==== Body pmc1 Introduction End of 2019 saw the outbreak of the dangerous infectious disease COVID-19, which is brought on by a novel coronavirus. Global public health has been significantly impacted by the 13,837,395 diagnosed cases and the 590,702 death cases as of July 17, 2020 [1]. The quick rise in infection cases suggests that COVID-19 has a much greater ability to spread than MERS-CoV and SARS coronaviruses [2], [3]. A direction for taking the right steps can be provided by a deeper comprehension and insight of the epidemic tendencies. Numerous nations have implemented a variety of mitigating strategies to prevent the spread of COVID-19 since 23 January, 2020. These strategies include home isolation, herd immunity, limiting individual migration, and others. Lack of information on the dynamic mechanism relating to the severity of COVID-19 at the this early stage makes it extremely difficult to limit the spread of COVID-19. However, using a mathematical model, strategies can be measured to serve as a benchmark for determining whether mitigation strategies are adequate. During the modeling process, it is very important to describe how infectious diseases are transmitted between susceptible and infected individuals. Numerous studies have shown that the incidence rate, which measures the infection capacity of a single infected person per unit time, is a crucial tool for describing this process, where susceptible individuals come into contact with infected individuals and are then infected with such a predetermined probability [4], [5], [6], [7]. Meanwhile, Korobeinikov et al. [8] indicated that the stability of the endemic equilibrium point is closely related to the concave of the incidence rate with respect to the infected individuals. Therefore, many researchers established the epidemic model of COVID-19 under various incidence rates  [9], [10], [11], for example, Peng et al. [9] constructed the SEIR (E-exposed) epidemic model and they discovered that COVID-19’s first appearance might be traced to the end of December 2019. A SEIQRD (Q-diagnosed) model was taken into consideration by Xu et al. [10], which has some basic guiding relevance for predicting COVID-19. Besides, individual migration has a crucial effect on the evolution of infectious diseases. With the convenience between cities, individuals move more and more frequent and new infectious diseases develop more rapidly regionally and globally [12]. Numerous deterministic models with multiple patches have been presented in attempt to better understand how individual migration affects the spread of infectious illnesses [13], [14], [15]. Contrary to what was initially reported [16], COVID-19 is in fact spreading from person to person through continuous interpersonal contact [2]. Lu et al. [17] considered a fractional-order SEIHRD (H-hospitalized) model with inter-city networks and they found that COVID-19 could be reduced in low-risk areas, but increased in high-risk areas by restricting communication between cities. Meanwhile, cross-infection among cities are considered, while there is not consider for self-migration [17]. Therefore, it is of great practical significance to include individual migration in different cities or different countries with the modeling COVID-19. Furthermore, the migration of susceptible individuals, exposed individuals, infected individuals are studied in this paper. It is worth noting that the time which patients waits for treatment follows the power law distribution [18], which prompts the use of the Caputo fractional-order derivative [19]. Angstmann et al. [20] discovered how fractional operators naturally appear in their model if the recovery time is a power law distribution after building a SIR epidemic model. Meanwhile, this offers a chronic disease epidemic model in which long-term infected people have little chance of recovering. Based on this statement, several authors have stated that the fractional-order model plays an important role in the process of disease transmission. Khan et al. [21] recounted how individuals, bats, unidentified hosts, and the source of the illness interacted, and considered how crucial the fractional-order system was in preventing the spread of the infection. To predict the spread of COVID-19, Chen et al. [22] developed a fractional-order epidemic model. Amjad et al. [23] built a fractional-order COVID-19 model and calculated the consequences of several mitigation and prevention strategies. Motivated by the above discussion, a fractional-order SEIHRDP epidemic model with individuals movement is established in this paper to study COVID-19. Meanwhile, the number of hospitalizations is the same as confirmed isolation in China, and but in other countries, these two are not equal, which the number of confirmed case is greater than that of hospitalized case. So in order to give a more generalized model, the purpose of this paper is to describe hospitalized individuals in response to the spread of COVID-19. The infectiousness of the incubation time is also taken into consideration, as inspired by [24]. Then, the proposed system’s dynamic behaviors are investigated in order to show the existence and uniqueness of the nonnegative solution, the global asymptotic stability of the disease-free equilibrium, and the uniform persistence, all of which have theoretical implications for future COVID-19 intervention and prevention. Meanwhile, the basic reproduction number with and without individual migration are compared, and it is found that adding individual migration can effectively describe the spread of COVID-19. Furthermore, the sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. Meanwhile, considering India and Brazil, results suggest that the bilinear incidence rate may be more fitted than the saturation incidence rate for stimulating the spread of COVID-19. When individuals movement is not considered, it can be found the proposed fractional-order model can better predict than the integer-order for multi peaks of COVID-19 in China. Meanwhile, when individuals movement is considered, the epidemic in the United States is analyzed and some mitigation measures are carried out to control the development of COVID-19. An implication of the achieved results is the possibility that the United States peaked on 24 November, 2020 (integer-order system) and 1 January, 2021 (fractional-order system), however, the number of infections shows an downward trend after 17 July, 2020 as enhancing nucleic acid detection and reducing the contact rate. Meanwhile, considering measures to limit migration between New York and Los Angeles, and enhance nucleic acid detection and reduce exposure rates, it is evident that there is an immediate increase in confirmed cases before a drop. Based on the above analysis, a generalized fractional-order SEIHRDP epidemic model with individual migration is considered. The main contributions of this study are as follows: • A fractional epidemic model with self-migration is considered, in which the infectivity of exposed individuals and hospitalized individuals are also taken into account. • The global properties of the proposed model are investigated, including the existence and uniqueness of global positive solutions, the local and global stability of disease-free equilibrium points, the persistence of disease transmission. • The sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. • Based on real data, the impact of the incidence rate on modeling COVID-19 is studied in India and Brazil. • Multiple peaks of COVID-19 transmission in China are analyzed by the proposed system. • Individual movement in the spread of COVID-19 in the United States is investigated and the peak are analyzed based on mitigation measures, such as enhanced nucleic acid testing, reduced of individual exposure, and control of individual movement. The rest of this paper is organized as follows. The SEIHRDP fractional-order model with individual movement is developed for COVID-19 in Section 2 and provides some preliminaries. Then dynamic properties of the proposed system are examined in Section 3. The theoretical results are shown using numerical simulations in Section 4. Finally, Section 5 provides the conclusions. 2 System description and preliminaries Fractional-order operator have been determined to have a wide range of uses in the modeling of many dynamic processes, including those in engineering, biology, medicine, and others [25], [26], [27], [28]. In this part, some necessary preliminaries are introduced before the fractional-order epidemic model is presented. 2.1 Preliminaries Definition 2.1 [29] The Caputo fractional-order operator is defined by t0CDtαgt=dαgtdtα=1Γn−α∫0tg(n)(s)t−sα−n+1ds,(n−1<α<n), where g(n)(s) is the nth derivative of g(s) with respect to s. Remark 2.1 If α=n, one has t0CDtαgt=g(n)(t). Lemma 2.1 [30] The Caputo nonlinear system is considered as follows: 0CDtαx(t)=g(x),(α∈(0,1]), with the initial condition x0 . If all eigenvalues of J|x=x∗=∂g∂x|x=x∗ satisfy |arg(λ)|>απ2 , the equilibrium points x∗ are locally asymptotically stable. Lemma 2.2 [31] Suppose X⊂R and the continuous operator T(t):X→X satisfies (1) T(t) is point dissipative in X and compact for t≥0 . (2) there is a finite sequence M={M1,M2,…,Mk} of compact and isolated invariant sets such that (i) Mi∩Mj=0̸ for any i,j=1,2,…,k and i≠j ; (ii) Ω(∂X0)≜∪x∈∂X0ω(x)⊂∪i=1kMi ; (iii) in the case of ∂X0 , no a cycle is formed by any subset of M ; (iv) Ws(Mi)∩X0=0̸ for each i=1,2,…,k . Then T(t) is uniformly persistent in X . 2.2 Graph theory In this paper, a weighted graph ζ=(ϑ,ω,A) will be considered to model the spread of infectious diseases between cities, where ϑ={ϑ1,ϑ2,…,ϑn} denotes the node set and ϑi represents the ith city; ω⊆ϑ×ϑ is the edge set, and if there is individual movement between any two cities, it means that there is a edge between this two cities; matrixes M=[mij]1≤i,j≤n, N=[nij]1≤i,j≤n, P=[pij]1≤i,j≤n and Q=[qij]1≤i,j≤n represent the weighted adjacency matrix of susceptible, exposed, infected and recovered individual, respectively; mij, nij, pij and qij denote the migrate rate of susceptible, exposed, infected and recovered individual from city j to city i with aij≥0 (i≠j) and aii=0 (a=m,n,porq), respectively. Furthermore, based on the directivity of individual migration, the directed graph ζ is studied in this paper. 2.3 System description Starting from 23 January, 2020, the Chinese government has adopted a series of mitigation measures to effectively suppress the spread of COVID-19, such as implementing strict home isolation, restricting various traffic, strengthening nucleic acid testing, establishing shelter hospitals and so on. Meanwhile, other countries around the world have adopted different measures from China, such as social distancing and herd community strategy by British, protecting sensitive compartment from infection by Italy, transferring of critically ill patients with military aircraft by France, etc. Therefore, it is important to establish a generalized model of individual migration to simultaneously quantify the impact of interruption of policies on virus transmission. Moreover, Tang et al. [32] proposed that the exposed individual is infectious of COVID-19. Therefore, a fractional-order SEIHRDP epidemic model with individual migration is considered as follows: (1) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),0CDtαEk=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),0CDtαIk=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),0CDtαHk=δkIk−(λk+κk)Hk,0CDtαRk=λkHk+∑j=1n(qkjRj−qjkRk),0CDtαDk=κkHk,0CDtαPk=ρkSk. with the initial condition (2) Sk(0)=Sk0>0,Ek(0)=Ek0≥0,Ik(0)=Ik0≥0,Hk(0)=Hk0≥0, Rk(0)=Rk0≥0,Dk(0)=Dk0≥0,Pk(0)=Pk0≥0. The specific explanation of system (1) are as follows: • Susceptible Sk: the number of susceptible class within city k at time t. • Exposed Ek: the number of exposed class within city k at time t (neither any clinical symptoms nor high infectivity). • Infectious Ik: the number of infected class within city k at time t (with overt symptoms). • Hospitalized Hk: the number of hospitalized class within city k at time t. • Recovered Rk: the number of recovered class within city k at time t. • Dead Dk: the number of dead class within city k at time t. • Insusceptible Pk: the number of susceptible class who are not exposed to the external community within city k at time t. Meanwhile, the process of disease transmission are as follows: • The susceptible individual Sk contacts with Ek and Ik, and then is infected by β1kSkfk(Ik)+β2kSkgk(Ek), where βik (i=1,2) are the transmission coefficient, fk(Ik) and gk(Ek) are generalized incidence rates. • The parameter Λk is the inflow rate; λk, ϵk, δk and κk represent the recovery, incubation, diagnosis, mortality rate. • The susceptible, exposed, infective and recovered individuals in city j move to city k with probability mkj, nkj, pkj and qkj, respectively. The terms ∑j=1n(mkjSj−mjkSk), ∑j=1n(nkjEj−njkEk), ∑j=1n(pkjIj−pjkIk) and ∑j=1n(qkjRj−qjkRk) represent the movement of Sk, Ek Ik and Rk individual, where ∑j=1nakjWj represents the individuals moving into k city from other cities j (k≠j) and ∑j=1najkWj represents the individuals leaving city k (W=S, E, I, R, respectively, a=m, n, p, q, respectively). • The movement of insuspectible and hospitalized individuals is not considered in this paper. Furthermore, Λk, βik (i=1,2), ρk and ϵk are positive constants; functions δk(t), λk(t), κk(t), mkj(t), nkj(t), pkj(t) and qkj(t) satisfy |δk(t)|≤M1k, |λk(t)|≤M2k, |κk(t)|≤M3k, |mkj(t)|≤M4k, |nkj(t)|≤M5k, |pkj(t)|≤M6k and |qkj(t)|≤M7k for all t≥0 and k,j=1,2,…,n, where M1k, M2k, M3k, M4k, M5k, M6k and M7k are positive constants. The transmission diagram of the generalized SEIHRDP model (1) is shown in Fig. 1. Before presenting the major findings, the following generalized incidence rate hypothesis is put forth: (H):(i)gk(Ek)andfk(Ik)satisfythelocalLipschitzconditionand gk(0)=0,fk(0)=0fork=1,2,…,n;(ii)fk(Ik)isstrictlymonotoneincreasingonIk∈[0,∞)and gk(Ek)isstrictlymonotoneincreasingon Ek∈[0,∞)forallk=1,2,…,n;(iii)fk(Ik)≤akIkforallIk≥0, whereak=fk′(0)forallk=1,2,…,n;(iv)gk(Ek)≤bkEkforallEk≥0, wherebk=gk′(0)forallk=1,2,…,n. Fig. 1 The schematic diagram of SEIHRDP epidemic model with individual migration (i,j=1,2,…,n). Remark 2.2 It should be noted that many current models can be viewed as a special type of system (1) with the hypothesis (H), such as gk(Ek)=bkEk, gk(Ek)=bkEk1+vkEk, fk(Ik)=akIk, fk(Ik)=akIk1+ukIk and others [33] with nonnegative constants ak, bk, uk and vk. Remark 2.3 Compared with [34], the individual movement in this paper can be described as follows: (1) the self-migration of individuals is described in system (1), which is caused by a self-chemotactic-like forcing [35]. However, the cross-infection among cities is considered which is a travel infectious [34]. (2) system (1) describes not only the migration of infected individuals, but also the movement of exposed and recovered individuals. (3) M=[mij]1≤i,j≤n, N=[nij]1≤i,j≤n, P=[pij]1≤i,j≤n and Q=[qij]1≤i,j≤n are not irreducible in this paper, but irreducible in [34]. Then the influence of network structure on disease transmission can be discussed in this paper, such as fully connected network, ring network and centralized network. However, [34] only consider fully connected network. (4) the total population of each city is changed (without considering the decrease in population due to death) in system (1). But in [34], the total population of each city remains constant. 3 System analysis This study explores system (1)’s dynamic analysis. As can be seen, the death class Dk and the insusceptible class Pk have no effect on the susceptible class Sk, exposed class Ek, infected class Ik, hospitalized class Hk, or recovered class Rk of systems (1). Accordingly, the following system is discussed in the next section: (3) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),0CDtαEk=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),0CDtαIk=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),0CDtαHk=δkIk−(λk+κk)Hk,0CDtαRk=λkHk+∑j=1n(qkjRj−qjkRk), with the initial condition (4) Sk(0)=Sk0>0,Ek(0)=Ek0≥0,Ik(0)=Ik0≥0, Hk(0)=Hk0≥0,Rk(0)=Rk0≥0,(k=1,2,…,n). 3.1 Existence and uniqueness of the positive solution The existence, uniqueness, and boundedness of the nonnegative solution for system (3) should be taken into account prior to the numerical process. Therefore, this subsection will be discussed these properties for system (3). Theorem 3.1 For any nonnegative initial condition (4) , there are a unique solution for system (3) and the region Ω={(S1,E1,I1,H1,R1,…,Sn,En,In,Hn,Rn): 0<Si≤Λ¯ρ,0≤Ei≤Λ¯ρ,0≤Ii≤Λ¯ρ, 0≤Hi≤Λ¯ρ,0≤Ri≤Λ¯ρ,i=1,2,…,n} is positively invariant for system (3) , where Λ¯=∑j=1nΛj and ρ=min{ρ1,ρ2,…,ρn} . Proof Let consider the following function: f1k=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),f2k=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),f3k=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),f4k=δkIk−(λk+κk)Hk,f5k=λkHk+∑j=1n(qkjRj−qjkRk). It is obvious that Fk=(f1k,f2k,f3k,f4k,f5k) satisfies the local Lipschitz condition about (Sk,Ek,Ik,Hk,Rk), then system (3) has a unique solution. Next, the nonnegative solution will be analyzed. Consider the following auxiliary system: 0CDtαS_k=−β1kS_kfk(I_k)−β2kS_kgk(E_k)−ρkS_k+∑j=1n(mkjS_j−mjkS_k),0CDtαE_k=β1kS_kfk(I_k)+β2kS_kgk(I_k)−ϵkE_k+∑j=1n(nkjE_j−njkE_k),0CDtαI_k=ϵkE_k−δkI_k+∑j=1n(pkjI_j−pjkI_k),0CDtαH_k=δkI_k−(λk+κk)H_k,0CDtαR_k=λkH_k+∑j=1n(qkjR_j−qjkR_k),Sk_(0)=Ek_(0)=Ik_(0)=Hk_(0)=Rk_(0)=0. Through the comparison theorem, it is not difficult to find that the following auxiliary system has a unique solution (0,0,0,0,0). then the following equation holds: (Sk,Ek,Ik,Hk,Rk)>(0,0,0,0,0). Next, adding all equations gives 0CDtαN≤Λ¯−ρN where N=∑j=1n(Sj+Ej+Ij+Hj+Rj+Dj), Λ¯=∑j=1nΛj and ρ=min{ρ1,ρ2,…,ρn}. Then N(t)≤(N(0)−Λ¯ρ)Eα(−ρtα)+Λ¯ρ. Therefore, the region Ω is positively invariant for system (3).  □ 3.2 Local stability The exploration of the existence and local stability of the disease-free equilibrium point is the focus of this section. Theorem 3.2 There are a unique disease-free equilibrium point E0=(S1∗,0,0,0,0,…,Sn∗,0,0,0,0) for system (3) where S∗=(S1∗,…,Sn∗) , S∗=A−1Λ , Λ=(Λ1,…,Λn) and A=ρ1+∑j≠1nmj1−m12⋯−m1n−m21ρ2+∑j≠2nmj2⋯−m2n⋮⋮⋮⋮−mn1−mn2⋯ρ2+∑j≠nnmjn. Proof Obviously, E0 satisfies the following equation: Λk−ρkSk∗+∑j=1n(mkjSj∗−mjkSk∗)=0, then the above equation can be written as the following matrix form: AS∗=Λ. It can be found that the matrix A is strictly diagonally dominant, and then it follows from [36] that one has A−1≥0. So according to [37], there exists a unique solution S∗=A−1Λ. Therefore, there exists a unique disease-free equilibrium point E0 of system (3). □ The predicted number of secondary cases that a typical infectious individual should create in a community that is totally susceptible is known as the basic reproduction number R0. According to Watmough et al. [38], it can be determined that an infectious disease can commonly infect the community if one diseased individual can typically infect more than one susceptible individual when R0≥1. On the other hand, if R0<1, each infected individual produces less than one new infection, and the infectious diseases can not grow. Thus, it is very important to describe the relationship between the basic reproduction number and the spread of infectious diseases. Here, the basic reproduction number R0 is stated as follows. Theorem 3.3 Under hypothesis H, the basic reproduction number R0 is R0=ρ(F11V11−1−F12V11−1V21V22−1), where matrixes F11=diag(β21b1S1,…,β2nbnSn) , F12=diag(β11a1S1,…,β1nanSn) , V21=diag(−ϵ1,…,−ϵn) , V11=ϵ1+∑j≠1nn1j−n12⋯−n1n−n21ϵ2+∑j≠2nn2j⋯−n2n⋮⋮⋮⋮−nn1−nn2⋯ϵn+∑j≠nnnnj, and V22=δ1+∑j≠1np1j−p12⋯−p1n−p21δ2+∑j≠2np2j⋯−p2n⋮⋮⋮⋮−pn1−pn2⋯δn+∑j≠nnpnj. Proof Let consider the following matrixes: F0=β11S1f1(I1)+β21S1gk(E1)⋮β1nSnfn(In)+β2nSngn(En)0⋮00⋮0andV0=ϵ1E1−∑j=1n(n1jEj−nj1E1)⋮ϵnEn−∑j=nn(nnjEj−njnEn)−ϵ1E1+δ1I1−∑j=1n(p1jIj−pj1I1)⋮−ϵnEn+δnIn−∑j=nn(pnjIj−pjnIn)−δ1I1+(λ1+κ1)H1⋮−δ1In+(λn+κn)Hn. Let u=(E1,…,En,I1,…,In,H1,…,Hn), then take the derivative of F0 and V0 for u at the disease-free equilibrium point E0, respectively, we can see as follows: F=F11F120000000andV=V1100V21V2200V32V33, where F11=diag(β21b1S1∗,…,β2nbnSn∗), F12=diag(β11a1S1∗,…,β1nanSn∗), V21=diag(−ϵ1,…,−ϵn), V33=diag((λ1+κ1),…,(λn+κn)), V32=diag (−δ1,…,−δn), V11=ϵ1+∑j≠1nn1j−n12⋯−n1n−n21ϵ2+∑j≠2nn2j⋯−n2n⋮⋮⋮⋮−nn1−nn2⋯ϵn+∑j≠nnnnj, and V22=δ1+∑j≠1np1j−p12⋯−p1n−p21δ2+∑j≠2np2j⋯−p2n⋮⋮⋮⋮−pn1−pn2⋯δn+∑j≠nnpnj. Then according to [39], the basic reproduction number is as follows: R0=ρ(FV−1)=ρ(F11V11−1−F12V11−1V21V22−1), where ρ(F11V11−1−F12V11−1V21V22−1) is the spectral radius of the matrix (F11V11−1−F12V11−1V21V22−1). □ Remark 3.1 According to [40], the epidemic size ςk=Sk(0)−Sk∗ of city k is defined as the number of individuals affected by the infectious disease, where Sk(0) is initial condition and Sk∗ is the disease-free equilibrium point of susceptible individuals within city k. Remark 3.2 When individual migration is not taken into consideration, it can be calculated from [10] that the basic reproduction number R0uk of city k is R0uk=Sk∗(β2kbkϵk+β1kakδk). Remark 3.3 When individual migration is taken into consideration, R0k of city k is R0k=Sk∗ϵk+∑j⁄=knnkj(β2kbk+β1kakϵkδk+∑j≠knpkj). Remark 3.4 It is easy to see that R0k are not dependent on λk, κk and mkj. Like [41], the other Λk, β1k, β2k, ρk, ϵk, nkj, pkj and δk are calculated as follows: AΛk=ΛkR0k∂R0k∂Λk=1,Aρk=ρkR0k∂R0k∂ρk=−1, Aβ1k=β1kR0k∂R0k∂β1k=β1kakϵkδk+∑j≠knpkjβ2kbk+β1kakϵkδk+∑j≠knpkj,Aβ2k=β2kR0k∂R0k∂β2k=β2kbkβ2kbk+β1kakϵkδk+∑j≠knpkj, Aδk=δkR0k∂R0k∂δk=−1(δk+∑j≠knpkj2)(β2kbk+β1kakϵkδk+∑j≠knpkj),Aϵk=ϵkR0k∂R0k∂ϵk=−1ϵk+∑j≠knnkj(β2kbk+ϵk+∑j≠knnkjδk+∑j≠knpkj), Ankj=nkjR0k∂R0k∂nkj=−1,Apkj=pkjR0k∂R0k∂pkj=−1(δk+∑j≠knpkj2)(β2kbk+β1kakϵkδk+∑j≠knpkj), where AΛk, Aρk, Aβ1k, Aβ2k, Aδk, Aϵk, Ankj and Apkj represent the normalized sensitivity on Λk, ρk, β1k, β2k, δk, ϵk, nkj and pkj, respectively. Through the above calculation found that the increase on Λk, β1k and β2k leads to the increase on R0k, but the increase on ρk, δk, ϵk, nkj and pkj leads to the decrease on R0k. In addition, the movement of susceptible individuals has no impact of R0k, but the movement of exposed and infected individuals is negatively correlated with R0k, and the movement of exposed individuals is more likely to influence the spread of the infectious disease with |Ankj|>|Apkj|. Theorem 3.4 Under hypothesis H, system (3) is locally asymptotically stable at the disease-free equilibrium point E0 if |arg(sF−V)|>απ2 . Proof The following Jacobian matrix at the disease-free equilibrium point E0 is considered: JE0=J11∗00F−V00∗J33, where matrixs J11=−ρ1−∑j=1nmj1m12⋯m1nm21−ρ2−∑j=1nmj2⋯m2n⋮⋮⋮⋮mn1mn2⋯−ρn−∑j=1nmjn, J33=−∑j=1nqj1q12⋯q1nq21−∑j=1nqj2⋯q2n⋮⋮⋮⋮qn1qn2⋯−∑j=1nqjn, F and V see Theorem 3.3. Then if all eigenvalues of the Jacobian matrix JE0 satisfy |arg(si)|>απ2, E0 is locally asymptotically stable and unstable if for some eigenvalues si, |arg(si)|≤απ2. Obviously, J11 and J33 are a nonsingular M-matrix, so J11 and J33 has all eigenvalues with negative real parts according to [42]. Consequently the local stability of E0 depends only on eigenvalues of F−V. Thus, if all eigenvalues of F−V satisfy |arg(sF−V)|>απ2, system (3) is locally asymptotically stable. □ Remark 3.5 If all the eigenvalues of F−V are negative, that is |arg(sF−V)|=π>απ2, system (3) is locally asymptotically stable. Meanwhile, it is obvious that |arg(sF−V)|=π⇔sF−V<0⇔ρ(FV−1)<1⇔R0<1. It can be yielded that if R0<1, the disease-free equilibrium point E0 is locally asymptotically stable of system (3) . 3.3 Global asymptotic stability of the disease-free equilibrium In this subsection, the global asymptotic stability of the disease-free equilibrium point E0 is discussed firstly. Furthermore, the uniform persistence of system (3) is also considered. Theorem 3.5 Under hypothesis (H) and |arg(sF−V)|>απ2 , the disease-free equilibrium point E0 is globally asymptotically stable of system (3) . Proof We use a method similar to the one used in [43]. Firstly, the boundedness of the susceptible class will be analyzed. According to Theorem 3.1 and hypothesis (H), we know Sk, Ek and Ik (k=1,2,…,n) are nonnegative, thus one has (5) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk +∑j=1n(mkjSj−mjkSk)≤Λk−ρkSk+∑j=1n(mkjSj−mjkSk). Let S=(S1,…,Sn), S∗=(S1∗,…,Sn∗), Λ=(Λ1,…,Λn) and A=ρ1+∑j≠1nmj1−m12⋯−m1n−m21ρ2+∑j≠2nmj2⋯−m2n⋮⋮⋮⋮−mn1−mn2⋯ρ2+∑j≠nnmjn, then Eq. (5) can be written in the following matrix: 0CDtαS≤Λ−AS=AS∗−AS, so it is easy to see that the conclusion holds as follows: S(t)≤(S0−S∗)Eα(−Atα)+S∗. Obviously, one has Sk≤Sk∗. Next, the global stability of Ek, Ik and Hk will be discussed. Based on hypothesis (H), one has fk(Ik)≤akIk and gk(Ek)≤bkEk. Then the following auxiliary system is considered: (6) 0CDtαEk¯=β1kSk∗akIk¯+β2kSk∗bkEk¯−ϵkEk¯+∑j=1n(nkjEj¯−njkEk¯),0CDtαIk¯=ϵkEk¯−δkIk¯+∑j=1n(pkjI¯j−pjkIk¯),0CDtαHk¯=δkIk¯−(λk+κk)Hk¯. It is easy to see that (7) 0CDtαW=(F−V)W, where W=(E¯,I¯,H¯), E¯=(E1¯,…,En¯), I¯=(I1¯,…,In¯), H¯=(H¯1,…,H¯n), F and V see Theorem 3.3. Thus, if |arg(sF−V)|>απ2, the above linear system (7) is locally asymptotically stable as well as globally asymptotically stable, that is limt→∞E¯k=limt→∞I¯k=limt→∞H¯k=0. According to the comparison theory and the nonnegative solution of Ek, Ik and Hk, one has limt→∞Ek=limt→∞Ik=limt→∞Hk=0. Based on the above analysis, when t→∞, one has 0CDtαS=AS∗−AS, then one has S(t)→S∗(t→∞). So E0 is globally asymptotically stable if |arg(sF−V)|>απ2. □ Remark 3.6 Similar to Theorem 3.3, it can be concluded that if R0<1, system (3) is globally asymptotically stable at the disease-free equilibrium point E0. Furthermore, the uniform persistence for system (3) is discussed in the following theorem. Theorem 3.6 Under hypothesis (H) and R0>1 , system (3) is uniformly persist, implying there exists a positive constant δ such that lim inft→+∞Sk≥δ,lim inft→+∞Ek≥δ,lim inft→+∞Ik≥δ,lim inft→+∞Hk≥δ,lim inft→+∞Rk≥δ,1≤k≤n. Proof Let consider the following space: X=X1×X2×⋯×Xn,X0=X10×X20×⋯×Xn0,∂X=∂X1×∂X2×⋯×∂Xn, where X0 represents the interior of X, ∂X denotes the boundary of X and Xk={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek≥0,Ik≥0,Hk≥0,Rk≥0}, Xk0={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek>0,Ik>0,Hk>0,Rk>0}, ∂Xk={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek=0,Ik=0,Hk=0,Rk=0}. Meanwhile, let W(t)=(S1,E1,I1,H1,R1,…,Sn,En,In,Hn,Rn) be the solution of system (3) with initial value W(0)=W0∈X, then W(t)∈X according to Theorem 3.1. For any t≥0, a continuous map F(t):X→X is defined as follows: F(t)W0=W(t). In the following, the uniformly persistent of the map F will be analyzed based on Lemma 2.3. When t=0, one has F(0)W0=W(0), this is F(0)=I where I is the identity matrix. Meanwhile, it can be deduced that the following equation holds: F(t+s)W0=W(t+s)=F(t)W(s)=F(t)F(s)W0, implying F(0)=I. Additionly, it is easy to see that F(t) is C0-semigroup on X, point dissipative and compact in X. Furthermore, consider the following system: 0CDtαSk=Λk−ρkSk+∑j=1n(mkjSj−mjkSk). According to Theorem 3.3, Sk∗ is asymptotically stable, which finds that E0 in ∂X is a global attractor of F(t). Let M={M1}, where M1={E0}. Because of ak=fk′(0) and bkgk′(0), for all ϵ, there exists ϵ¯ that f(ϵ)>(ak−ϵ)ϵ¯andg(ϵ)>(bk−ϵ)ϵ¯. Let the stable set Ws(E0) of a compact invariant set E0 defined by Ws(E0)={Y0∈X:ω(Y0)≠0̸,ω(Y0)∈E0}, where ω(Y0) is ω-limit set through Y0. System (3) has a solution (Sk,Ek,Ik,Hk,Rk) when Ws(E0)∩X0≠0̸, implying Sk→0, Ek→0, Ik→0, Hk→0, Rk→0 (k=1,2,…,n) as t→∞. So there exists a constant τ>0 such that Sk>Sk∗−ϵ, Ek>ϵ, Ik>ϵ, Hk>ϵ and Rk>ϵ for t≥τ. Then according to the monotonicity of fk(Ik) and gk(Ek), one has f(Ik)>f(ϵ)>(ak−ϵ)ϵ¯andg(Ek)>g(ϵ)>(bk−ϵ)ϵ¯. So the following auxiliary system is considered: (8) 0CDtαEk_=β1kSk∗(ak−ϵ)ϵ¯Ik_+β2kSk∗(bk−ϵ)ϵ¯Ek_−ϵkEk_+∑j=1n(nkjEj_−njkEk_),0CDtαIk_=ϵkEk_−δkIk_+∑j=1n(pkjIj_−pjkIk_),0CDtαHk_=δkIk_−(λk+κk)Hk_. It is easy to see from system (8) that (9) 0CDtαW=(F−V)(ϵ¯,ϵ)W, where W=(E_,I_,H_), E_=(E_1,…,E_n), I_=(I_1,…,I_n) and H_=(H_1,…,H_n). Consider the basic reproduction number R0>1, then one has ρ(F11V11−1−F12V11−1V21V22−1)(ϵ¯,ϵ)>1, which results in a contradiction with Ek(t)→0 (t→∞). Hence one has Ws(E0)∩X0=0̸, implying it is uniformly persistent at the operator T(t), so system (3) is uniformly persistent if R0>1.  □ The existence of a positive equilibrium point is implied by the system (3)’s ultimate boundedenss and uniform persistence. As a result, we can derive the following theorem. Theorem 3.7 Under hypothesis (H) and R0>1 , there is at least one endemic equilibrium E∗=(S1∗,E1∗,I1∗,H1∗,R1∗,…,Sn∗,En∗,In∗,Hn∗,Rn∗) of system (3) satisfying Λk−β1kSk∗fk(Ik∗)−β2kSk∗gk(Ek∗)−ρkSk∗+∑j=1n(mkjSj∗−mjkSk∗)=0,β1kSk∗fk(Ik∗)+β2kSk∗gk(Ek∗)−ϵkEk∗+∑j=1n(nkjEj∗−njkEk∗)=0,ϵkEk∗−δkIk∗+∑j=1n(pkjIj∗−pjkIk∗)=0,δkIk∗−(λk−κk)Hk∗=0,λkHk∗+∑j=1n(qkjRj∗−qjkRk∗)=0. 4 Numerical simulation From the previous description, it is clear that E0 is globally asymptotically stable when R0<1 and conversely, system (3) is persistent, which can offer theoretical evidence for further COVID-19 prediction and control. Meanwhile, in order to analyze COVID-19 in different cities, this section is divided into two parts: no restrictions on individual migration and restrictions on individual migration. Furthermore, consider the corresponding integer-order model as follows: (10) dSkdt=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),dEkdt=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),dIkdt=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),dHkdt=δkIk−(λk+κk)Hk,dRkdt=λkHk+∑j=1n(qkjRj−qjkRk). 4.1 Data source The Johns Hopkins University Center for System Science and Engineering provided the real data for this study [1]. Data on accumulated and confirmed cases, recovered cases, and COVID-19 death cases were shared by the Johns Hopkins University on January 23, 2020. Assuming that the confirmed individuals mut be hospitalized, one has Hospitalized=Confirmed−Recovered−Death. Hence, we can get the real data of H(t), D(t) and R(t) for different city from 23 January to 17 July, 2020. 4.2 The generalized incidence rate As we know, Korobeinikov et al. [8] indicated that the stability of the endemic equilibrium point for infectious diseases is closely related to the concave of the incidence rate with respect to the infected individuals. Therefore, it is of practical significance to understand the role of different incidence rates in COVID-19. In this section, according to hypothesis (H), the bilinear incidence rate and the saturation incidence rate are discussed as follows: fk(Ik)=Ik,gk(Ek)=Ek. fk(Ik)=Ik1+ukIk,gk(Ek)=Ek1+vkEk. Meanwhile, as the public learns about COVID-19, the recovered rate and the disease-related mortality are time-varying rather than constant. Similar to [11], the best recovered rate λk and the best disease-related mortality κk are selected from the following equation: (11) κk=p1ep2(t−p3)+e−p2(t−p3),p1e(p2(t−p3))2,p1+e(p2(t+p3)),andλk=q11+e−q2(t−q3),q1+e−q2(t+q3), where qi and qi (i=1,2,3) are parameters for κk and λk, respectively. According to the real data reported by [2], the spread of COVID-19 in India and Brazil began on 30 January and 26 February, 2020, as the beginning of the outbreak of India and Brazil in this paper, respectively. According to Matlab function lsqcurvefit [11], the parameter identification results with system (3) and system (10) are depicted in Table 1, Table 2, respectively. Meanwhile, based on Table 1, Table 2, the five days forecast of India and Brazil are shown in Tables 3, 4, Figs. 2, 3, 4, 5, which the solid lines represent simulation results and circles represent real data. The results in Table 1, Table 2 show that the fractional-order system (3) can accurately forecast the real data in the upcoming week, with the real data of currently confirmed cases falling between 95% and 105% of the projected values. Table 1 Parameter identification of India. India Integer (Bilinear) Fractional (Bilinear) Integer (Saturation) Fractional (Saturation) Λ 0.3 0.6245 0.465 0.3 β1 0.3869 1.206 1.156 2.358 β2 0.5133 0.3079 0.3414 0.8 ϵ 0.0023 0.0555 0.0014 0.0692 ρ 0.0264 0.03 0.0188 0.0094 λ p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) κ b1e−q2(t−q3)2 q1e−q2(t−q3)2 q1e−q2(t−q3)2 q1e−q2(t−q3)2 Table 2 Parameter identification of Brazil. Brazil Integer (Bilinear) Fractional (Bilinear) Integer (Saturation) Fractional (Saturation) Λ 0.3 0.5 0.8802 0.5189 β1 1.839 1.082 0.2378 4.804 β2 0.3733 0.928 0.4397 0.3 ϵ 0.0303 0.7905 0.1076 0.9609 ρ 0.0211 0.0214 0.0216 0.0431 δ 0.99 0.2434 0.3975 5.609×10−5 λ p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) κ q1eq2(t−q3) q1eq2(t−q3) q1eq2(t−q3) q1eq2(t−q3) Table 3 Estimate the number of confirmed cases within five days in India (×105). India 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 3.735 3.906 4.027 4.113 4.263 Integer (bilinear incidence rate) 3.454 3.499 3.52 3.556 3.583 Fractional (bilinear incidence rate) 3.781 3.869 3.936 4.002 4.079 Integer (saturation incidence rate) 3.426 3.476 3.515 3.553 3.578 Fractional (saturation incidence rate) 3.645 3.696 3.741 3.792 3.815 Table 4 Estimate the number of confirmed cases within five days in Brazil (×105). Brazil 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 5.487 5.598 5.242 5.228 5.528 Integer (bilinear incidence rate) 5.864 5.847 5.828 5.806 5.781 Fractional (bilinear incidence rate) 5.502 5.49 5.475 5.458 5.438 Integer (saturation incidence rate) 5.897 5.882 5.864 5.853 5.821 Fractional (saturation incidence rate) 5.857 5.844 5.819 5.793 5.769 Fig. 2 The number of cases in India (Integer-order with the bilinear incidence rate (left), Fractional-order with the bilinear incidence rate (right)). Fig. 3 The number of cases in India (Integer-order with the saturation incidence rate (left), Fractional-order with the saturation incidence rate (right)). Fig. 4 The number of cases in Brazil (Integer-order with the bilinear incidence rate (left), Fractional-order with the bilinear incidence rate (right)). Fig. 5 The number of cases in Brazil (Integer-order with the saturation incidence rate (left), Fractional-order with the saturation incidence rate (right)). 4.3 Restrict individual migration When individual movement is not considered, the parameters satisfy mkj=nkj=pkj=qkj=0. Meanwhile, according to Section 4.2, the bilinear incidence rate is considered in this section. Then based on system (1), the following auxiliary system is considered: (12) 0CDtαS=Λ−β1SIβ2SE−ρS,0CDtαE=β1SI+β2SEϵE,0CDtαI=ϵE−δI,0CDtαH=δI−(λ+κ)H,0CDtαR=λH,0CDtαD=κH. 4.3.1 Sensitivity analysis of parameters in R0uk When individual movement is not taken into consideration in this section, Partial Rank Correlation Coefficients (PRCC) value and Latin hypercube sampling (LHS) [44], which are one of the Monte Carlo (MC) sampling methods established by Mckay in 1979 [45], can be used to account for the sensitivity of the parameter to the basic reproduction number. LHS has the advantage of using fewer iterations than other random sampling techniques and avoiding the clustering phenomenon of sampling [45]. In order to determine which aspects of a certain intervention have the greatest impact on how quickly a new infection spreads, it can be seen from Remark 3.2 that the parameters of system (12) all affect the basic reproduction number to varying degrees, thereby affecting the spread of the infectious disease. We perform LHS on the parameters that appear in R0uk. PRCC are calculated, and a total of 1000 simulations per LHS run are carried out. A uniform distribution is chose as the prior distribution when performing parameter sampling. The parameters Λ, ρ, ϵ, β1, β2 and δ of system (12) are set as input variables, and the basic reproduction number R0uk as the output. The specific process is as follows: (1) There are six parameters that affect the change of R0uk, which are Λ, ρ, ϵ, β1, β2 and δ. Through LSH, [0,1] is divided into 1000 simulations, and 6 × 1000 parameters are generated through random selection on each interval by a uniform distribution. (2) Calculate the basic reproduction number R0uk for each parameter. (3) PRCC is calculated by Matlab function partialcorr. (4) The PRCC’s influence on the basic reproduction number R0uk can increase with increasing PRCC absolute value. However, it is believed that the parameter is not significant if the p value is greater than 0.05. Table 5 lists the PRCC values of the six parameters associated with R0uk and Fig. 6 shows the histogram of PRCC value. From Table 5 and Fig. 6, the following conclusion holds: (1) the parameters Λ, β1 and β2 have a positive influence on R0uk, but ρ, ϵ and δ have a negative influence, which is consistent with Remark 3.4; (2) the positive impact of birth rate Λ is the most obvious with PRCC(Λ)=0.5868; (3) the positive impact of the transmission rate β2 for the exposed population is more obvious than that of the infected population with PRCC(β2)>PRCC(β1). That is, the greater the transmission coefficient of the exposed population, the greater the value of the basic reproduction number R0uk, and then the greater the number of people infected with COVID-19. Therefore, it is more critical to limit exposed individual. However, because exposed individual doed not show any symptoms, identifying them is very difficult, which is a key reason for the spread of COVID-19; (4) the diagnosis rate δ has more greater negative impact on R0uk. That is to say, enhancing nucleic acid detection can effectively reduce R0ui, thereby reducing the number of infected people; (5) from the p-value, it can be found that the p-values of all parameters are less than 0.05, so they all have a significant impact on the basic reproduction number R0uk. Therefore, based on the above analysis, it can be obtained that controlling the influx of foreign population and enhancing nucleic acid detection are the most effective measures to control COVID-19. Meanwhile, home isolation can also control COVID-19. Therefore, this evidence confirms the effectiveness of Chinese government’s interruption policies, such as home isolation, prohibition of the inflow of foreign population, and enhancing nucleic acid detection, which may provide a good reference for the other countries. Table 5 The PRCC values and p-value of the parameters with respect to R0uk. Input PRCC values p-value Λ 0.5868 0 ρ −0.5363 0 ϵ −0.1368 0 β2 0.1035 3.5×10−6 β1 0.0847 1.448×10−4 δ −0.4362 0 Fig. 6 The sensitivity analysis of R0uk. 4.3.2 China’s second outbreak From the analysis in Section 4.3.1, it can be found that enhancing the diagnosis rate and controlling the inflow of foreign population can effectively control the spread of the epidemic. For China, individual migration has been strictly restricted at the beginning of COVID-19. Therefore, the impact of enhanced diagnosis rate will be only considered in this section. Due to the increase in public awareness and the development of detection technology, the time from onset to diagnosis is gradually shortened. Additionally, despite the use of the nucleic acid test method, the number of confirmed cases climbed significantly and peaked in early February 2020 as a result of the use of the CT diagnosis method. As a result, it is assumed that starting on 12 February, 2020, China’s diagnosis rate can reach and remain at its highest level. However, the third COVID-19 wave has been occurring in Beijing since the end of June 2020. Beijing has said that starting on 17 June, 2020, nucleic acid could be more readily detected. As a result, a new distribution, rather than the max level dated June 17, now governs the diagnostic rate. Similar to [46], the following piecewise function are described the diagnosed period of two and three peaks: (13) 1δk=(1δ0−1δe)e−w1t+1δe,t<t1,1δe,t≥t1,and 1δk=(1δ0−1δe)e−w1t+1δe,t<t1,1δe,t1≤t≤t2,(1δe−1δf)e−w2(t−t2)+1δf,t>t2, where δ0, δe (δe>δ0), w1, w2 and δf are similar to [46], t1 is 13 February, 2020, t2 is 17 June, 2020. Meanwhile, similar to [11], the best recovered rate λk and the best disease-related mortality κk are selected from Eq. (11). Then system (12) and system (10) are solved by predictor–correctors scheme and least squares method [11] by the real data from 23 January to 17 July, which 17 March, 2020 is considered as the beginning of the emergency in Heilongjiang, Shanghai and Guangdong, and 17 June, 2020 are considered as the beginning of the emergency in Beijing, respectively. From Fig. 7, Fig. 8, the fractional-order system (12) is found to fit the real data more accurately than the integer-order system (10) does. and COVID-19 in Beijing, Shanghai reaches its highest peak in a short time but there may be fourth wave peak, however, Heilongjiang and Guangdong are only two peaks and the third wave of epidemic peaks will not occur in a short time (current policies remain unchanged). Therefore, under the condition of restricting the migration of individuals, the fractional system (12) can better simulate the multi-peak problem of COVID-19, and the strengthening of nucleic acid detection can predict the new wave in advance, which provides a theoretical basis for the control of the epidemic. Fig. 7 The number of cases in Beijing and Shanghai. Fig. 8 The number of cases in Guangdong and Heilongjiang. 4.4 Individual migration As of 17 July, 2020, the United States has a total of 3,647,715 confirmed cases, 139266 deaths and 1,107,204 recovery cases. It is urgent to formulate reasonable and effective mitigation measures. Thus in this section, based on the sensitivity analysis of parameter to R0k, the effect mitigation measures are provided to control the development of COVID-19 in US. 4.4.1 Sensitivity analysis of parameters in R0k Similar to Section 4.3.1, consider two cities to examine the sensitivity of parameters to R0k (k=1,2). Then when n=2, system (3) can be simplified as follows: (14) 0CDtαS1=Λ1−β11S1I1−β21S1E1−ρ1S1+(m12S2−m21S1),0CDtαE1=β11S1I1+β21S1E1−ϵ1E1+(n12E2−n21E2),0CDtαI1=ϵ1E1−δ1I1+(p12I2−p21I1),0CDtαH1=δ1I1−(λ1+κ1)H1,0CDtαR1=λ1H1+(q12R2−q21R1),0CDtαS2=Λ2−β12S2I2−β22S2E2−ρ2S2+(m21S1−m12S2),0CDtαE2=β12S1I2+β22S2E2−ϵ1E2+(n21E1−n12E1),0CDtαI2=ϵ2E2−δ2I2+(p21I1−p12I2),0CDtαH2=δ2I2−(λ2+κ2)H2,0CDtαR2=λ2H2+(q21R1−q12R2). It can be found from Remark 3.4 that there exists 16 parameter of the basic reproduction number R0k (k=1,2), and then the 16 parameters are set as input variables, and R0k as the output. Similar to Section 4.3.1, Table 6 lists the PRCC values and Fig. 9 shows the histogram of PRCC value. According to Table 6 and Fig. 9, it can be found that the following conclusion holds: (1) the movement of susceptible individuals mkj (k,j=1,2) does not affect R0k; (2) the sensitivity of the parameter to R0k (k=1,2) is same as that of Section 4.2.1, except for n12, n21, p12 and p21; (3) considering the basic reproduction number R01 of city 1, the p-value of p21 is large than 0.05, which means that infected individuals migrating from city 1 have a significant impact on COVID-19 in city 1. But exposed and infected individuals migrating to city 1 have an impact on the spread of COVID-19 in city 1, and the impact of the inflow of exposed individuals is more significant because of |PRCC(n12)|>|PRCC(p12)|; (4) contrary to the situation in city 1, the p-value of and p12 is large than 0.05, which means that infected individuals migrating from city 2 have a significant impact on the spread of disease in city 2. But exposed and infected individuals migrating from city 2 have an impact on the spread of COVID-19 in city 2, and the impact of the inflow of exposed individuals is more significant because of |PRCC(n21)|>|PRCC(p21)|. Therefore, in order to alleviate the situation in severe areas of COVID-19, migration of exposed individuals must be strictly controlled. Table 6 The PRCC values and p-value of the parameters with respect to R01 (left) and R02 (right). Input PRCC values p-value Λ1 0.6513 0 ρ1 −0.5294 0 ϵ1 −0.0348 0.1194 β21 0.1983 0 β11 0.0564 0.0116 δ1 −0.1427 0 n12 −0.3584 0 n21 −0.0286 0.2005 p12 −0.1732 0 p21 −0.0062 0.7806 Input PRCC values p-value Λ2 0.6496 0 ρ2 −0.5223 0 ϵ2 −0.0451 0.0435 β22 0.1896 0 β12 0.052 0.0153 δ2 −0.1691 0 n12 −0.0217 0.3318 n21 −0.2732 0 p12 −0.0048 0.8316 p21 −0.1355 0 Fig. 9 The sensitivity analysis of R01 (left) and R02 (right). 4.4.2 US outbreak In this subsection, the overall spread of COVID-19 in the US is considered first. Then system (10) and system (12) are solved by least squares method [11]. However, beginning 17 May, 2020, the number of confirmed individuals in the US had significantly increased. Emergency situations may have changed government regulations and people’s attitudes, which led to an increase in the number of sick people. Therefore, it is assumed that the emergency starts on 17 May, and the outbreak’s spread in the US is then examined in two stages as follows: (1) 23 January-17 May, 2020; (2) 17 May-17 July, 2020. Therefore, parameter identification is provided in Table 7 based on actual data from 23 January to 17 July 2020. From Fig. 10 and Table 8, it is clear that the fractional-order system (12) is capable of accurately forecasting the confirmed case for the upcoming week. In the meantime, Table 8 shows that, regardless of whether in the first stage or the second stage, the parameter findings of the fractional-order system and integer-order fitting are totally different. Based on Remark 3.2, R0uk=49.84 is very high. From the analysis in Section 4.2.1, it can be found that enhancing the diagnosis rate, reducing contact with infected people and controlling the inflow of foreign population can effectively control the spread of COVID-19. However, the United States is not currently doing anything to limit the influx of foreign population, so it is only considering enhancing nucleic acid testing and reducing contact with infected people to control COVID-19. Like [46], the diagnosed period 1δk of US are as follows: (15) 1δ(t)=1δet≤t3,(1δ0−1δe)e−w(t−t3)+1δet>t3. The meaning of each symbol is similar to that in Section 4.2.3. t3 is 17 July, 2020, which mean increasing the diagnosis rate δ(t) from 17 July, 2020. At the same time, the contact rate βi (i=1,2) is limited by the number of hospitalizations like [46] as follows: (16) βi(t)=βi,logH(t)≤1,βilogH(t),logH(t)>1. It can be seen from Fig. 10 that increasing the diagnosis rate δ(t) and controlling the infection rate βi (i=1,2) can effectively contain COVID-19. Therefore, enhanced nucleic acid testing and limited contact with infected individuals are important to control COVID-19.Table 7 Parameter identification of US. US Integer (first stage) Fractional (first stage) Integer (second stage) Fractional (second stage) Λ 0.3 0.0935 0.3 0.4593 β1 1.056 1.217 4.999 2.641 β2 0.2969 0.4882 1.648×10−7 3.724×10−7 ϵ 0.1791 0.2876 0.0087 0.0077 ρ 0.0281 0.0497 0.0358 0.0272 δ 0.1243 0.2434 0.3975 0.205 λ p1+e−p2(t+p3) p1+e−p2(t+p3) p1+e−p2(t+p3) p1+e−p2(t+p3) κ q1eq2(t−q3)+e−q2(t−q3) q1eq2(t−q3)+e−q2(t−q3) q1+e−q2(t+q3) q1+e−q2(t+q3) Table 8 Estimate the number of confirmed cases within five days in US (×106). Date Real data Fractional Integer 18 July 2.449 2.503 2.397 19 July 2.502 2.541 2.425 20 July 2.534 2.578 2.454 21 July 2.575 2.616 2.482 22 July 2.617 2.655 2.511 Fig. 10 The number of cases in US (without control (left), with control (right)). 4.4.3 US with individual migration This subsection considers the impact of individual migration on COVID-19. We need to preprocess the data to remove data that are less than 0.5% of the current maximum number of confirmed cases. Therefore, the real data after 3 April are selected to identify the parameters of system (14). Similar to the analysis of Section 4.4.2, we consider 17 May, 2020 as the beginning of the emergency, and the COVID-19 spread in New York and Los Angeles into two phases: (1) 3 April-17 May, 2020; (2) 17 May-17 July, 2020. Meanwhile, the recovered data of New York and Los Angeles have not been collected by [1], and then we take hospitalized+recovered individuals as a whole to conduct parameter identification and short-term prediction according to [11]. It can be found from Table 9, Table 10 and Fig. 11 that system (14) can better predict COVID-19. Meanwhile, it can be seen from Fig. 11 that the COVID-19 in New York has been peaked but not in Los Angles. From the analysis of Section 4.4.1, we know that controlling the infection rate, improving the diagnosis rate and controlling the movement of exposed individuals have a significant effect on the control of COVID-19 in US. Therefore, similar to Section 4.4.2, the diagnosis rate δk and the migration rate nkj are utilized as follows: (17) 1δk(t)=1δe,t≤t3,(1δ0−1δe)e−w(t−t3)+1δe,t>t3,and nkj=nkj,log(Hk)<1,nkjlog(Hk),log(Hk)≥1. Meanwhile, the infection rate controlled by the number of hospitalizations is Eq. (16). From Fig. 12, we can seen the fractional-order system (14) with control (Eqs. (16), (17)) in Los Angles can be control quickly but not in New York, which is still an open question and will be discussed later.Table 9 Estimate the number of confirmed cases within five days in New York (×105). New York 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 1.977 1.98 1.983 1.987 1.99 Integer 1.978 1.979 1.98 1.981 1.981 Fractional 1.985 1.986 1.987 1.988 1.989 Table 10 Estimate the number of confirmed cases within five days in Los Angles (×105). Los Angles 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 1.491 1.518 1.549 1.579 1.609 Integer 1.414 1.444 1.464 1.482 1.514 Fractional 1.494 1.517 1.547 1.579 1.608 Fig. 11 The number of cases in New York and Los Angeles with individual movement. Fig. 12 The number of cases in New York and Los Angeles with control Eqs. (16), (17). 5 Conclusion Based on individual migration, a fractional-order SEIHRDP model is proposed with the generalized incidence rate. Meanwhile, some results and effective mitigation measures is suggested to control COVID-19 as follows: (1) The local and global asymptotic stability of the disease-free and endemic equilibrium points are investigated based on the basic reproduction number R0. (2) Based on the real data, it is found that the bilinear incidence rate has a better description of COVID-19 transmission than the saturation incidence rate. Therefore, the bilinear incidence rate is applied in modelling COVID-19. Meanwhile, this is the first time that looked at the impact of the incidence rate in the spread of COVID-19 using real data. (3) By applying the value of PRCC, the sensitivity of the parameters to the basic reproduction number R0k and R0uk are obtained, which is consistent with Remark 3.4. Through the PRCC value, the diagnosis rate, the migration rate and the movement of the infected population are most sensitive to control COVID-19. (4) Multiple peaks have been analyzed for COVID-19 and using four cities in China to show that the fractional-order system (1) works well. Moreover, by increasing the diagnosis rate, it can be found that the third wave of epidemic in Beijing has reached its peak, but the arrival of the next wave of COVID-19 is not ruled out. (5) Analyzing the situation in the United States, it can be seen that system (12) has better predictability than system (10). Meanwhile, by reducing the infection rate and increasing the diagnosis rate, the peak of the epidemic in the US can be accelerated. (6) Results show that the fractional-order system can accurately forecast the real data in the upcoming week when taking into account individual migration between two cities. By limiting the movement of exposed individuals, raising the diagnosis rate, and lowering the infection rate, Los Angeles’ peaks can appear and then decline immediately. Furthermore, this study makes several contributions to predict multi-peak of COVID-19 in China and suggestions on controlling epidemic in the US by changing certain parameters. Nevertheless, this research raises some issues that require more investigation, including how medical and other factors affect the spread of infectious diseases, how to properly administer vaccines, how network topology affects disease transmission and so on. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgment This work is supported by the Natural Science Foundation of Beijing Municipality [grant numbers Z180005], the National Nature Science Foundation of China [grant numbers 61772063] and the Fundamental Research Funds of the Central Universities [grant numbers 2020JBM074]. ==== Refs References 1 The Johns Hopkins University Center for System Science and Engineering, Data of accumulated and newly confirmed cases, recovered case and death case of COVID-19, URL https://github.com/CSSEGISandData/COVID-19. 2 Chan J. Yuan S. Kok K. To K. Chu H. Yang J. A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: a study of a family cluster Lancet 395 10223 2020 514 523 31986261 3 Huang C. Wang Y. Li X. Ren L. Zhao J. Hu Y. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China Lancet 395 10223 2020 497 506 31986264 4 Anderson R. Anderson B. May R. Infectious diseases of humans: Dynamics and control 1992 Oxford University Press 5 Bailey N. The mathematical theory of infectious diseases and its applications 1975 Charles Griffin Company Ltd 6 Upadhyay R. Pal A. Kumari S. Roy P. Dynamics of an SEIR epidemic model with nonlinear incidence and treatment rates Nonlinear Dynam 96 4 2019 2351 2368 7 Zou X. Wang K. Optimal harvesting for a stochastic regime-switching logistic diffusion system with jumps Nonlinear Anal Hybrid 13 2014 32 44 8 Korobeinikov A. Maini P. Nonlinear incidence and stability of infectious disease models Math Med Biol 22 2 2005 113 128 15778334 9 Peng L. Yang W. Zhang D. Zhuge C. Hong L. Epidemic analysis of COVID-19 in China by dynamical modeling 2020 10.1101/2020.02.16.20023465 arXiv preprint arXiv:2002.06563 10 Xu C. Yu Y. Chen Y. Lu Z. Forecast analysis of the epidemics trend of COVID-19 in the USA by a generalized fractional-order SEIR model Nonlinear Dynam 101 3 2020 1621 12634 11 Cheynet E. Generalized SEIR epidemic model (fitting and compution) 2020 GitHub URL http://www.github.com/ECheynet/SEIR 12 Gao D.Z. Cosner C. Cantrell R.S. Beier J.C. Ruan S.G. Modeling the spatial spread of Rift Valley fever in Egypt Bull Math Biol 75 3 2013 523 542 23377629 13 Phaijoo G.R. Gurung D.B. Mathematical study of dengue disease transmission in multi-patch environment Appl Math 7 14 2016 1521 1533 14 Eisenberg C.M. Shuai Z.S. Tien J.H. den Driessche P.V. A cholera model in a patchy environment with water and human movement Math Biosci 246 1 2013 105 112 23958383 15 Julien A. Davis J.R. David H. Richard J. Miller J.M. den Driessche P.V. A multi-species epidemic model with spatial dynamics Math Med Biol 2 2005 129 142 15778332 16 Cheng Z.K.J. Shan J. 2019 Novel coronavirus: where we are and what we know Infection 49 197 2021 17 Lu Z. Yu Y. Chen Y.Q. Ren G. Xu C. Wang S. A fractional-order SEIHDR model for COVID-19 with inter-city networked coupling effects Nonlinear Dynam 101 3 2020 1717 1730 18 Smethurst D. Williams H. Are hospital waiting lists self-regulating? Nature 410 6829 2001 652 653 11287943 19 Meerschaert M. Sikorskii A. Stochastic models for fractional calculus 2011 De Gruyter 20 Angstmann C. Henry B. Mcgann A. A fractional-order infectivity SIR model Physica A 2016 86 93 21 Khan M.A. Atangana A. Modeling the dynamics of novel coronavirus (2019-nCov) with fractional derivative Alex Eng J 59 4 2020 2379 2389 22 Chen Y. Cheng J. Jiang X. Xu X. The reconstruction and prediction algorithm of the fractional TDD for the local outbreak of COVID-19 2020 arXiv preprint arXiv:2002.10302 23 Amjad S.S. Najiroddin S.I. Kottakkaran N.S. A mathematical model of COVID-19 using fractional derivative: Outbreak in India with dynamics of transmission and control Adv Differ Equ 373 2020 1 19 32226454 24 Tang Z. Li X.B. Li H.Q. Prediction of new coronavirus infection based on a modified SEIR model MedRxiv 2020 10.1101/2020.03.03.20030858 25 Zhou P. Ma J. Tang J. Clarify the physical process for fractional dynamical systems Nonlinear Dynam 2020 1 12 26 Li H. Zhang L. Hu C. Jiang Y. Teng Z. Dynamical analysis of a fractional-order predator-prey model incorporating a prey refuge J Appl Math Comput 54 1–2 2017 435 449 27 Singh J. Kumar D. Hammouch Z. Atangana A. A fractional epidemiological model for computer viruses pertaining to a new fractional derivative Appl Math Comput 316 2018 504 515 28 Sierociuk D. Skovranek T. Macias M. Podlubny I. Petras I. Dzielinski A. Diffusion process modeling by using fractional-order models Appl Math Comput 257 2015 2 11 29 Podlubny I. Fractional differential equations 1999 Academic Press 30 Li Y. Chen Y. Podlubny I. Mittag–Leffler stability of fractional order nonlinear dynamic systems Automatica 45 8 2009 1965 1969 31 Tang Q. Teng Z. Jiang H. Global behaviors for a class of multi-group SIRS epidemic models with nonlinear incidence rate Taiwan J Math 19 5 2015 1509 1532 32 Tang Z. Li X. Li H. Prediction of new coronavirus infection based on a modified SEIR model Cold Spring Harbor Lab 2020 1 13 33 Lin M. Huang J. Ruan S. Yu P. Bifurcation analysis of an SIRS epidemic model with a generalized nonmonotone and saturated incidence rate J Differ Equ 267 3 2019 1859 1898 32226129 34 Muroya Y. Enatsu Y. Kuniya T. Global stability for a multi-group SIRS epidemic model with varying population sizes Nonlinear Anal Real 14 3 2013 1693 1704 35 Angstmann C.N. Donnelly I.C. Henry B.I. Langlands T. Continuous-time random walks on networks with vertex- and time-dependent forcing Phys Rev E 88 2 2013 022811 36 Li M. Graef J.R. Wang L.C. Karsai J. Global dynamics of a SEIR model with varying total population size Math Biosci 160 2 1999 191 213 10472754 37 Kheiri H. Jafari M. Stability analysis of a fractional order model for the HIV/AIDS epidemic in a patchy environment J Comput Appl Math 2018 323 339 38 van den Driessche P. Watmough J. Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmission Math Biosci 180 1–2 2002 29 48 12387915 39 Diekmann O. Heesterbeek J. Metz J. On the definition and the computation of the basic reproduction ratio R0 in models for infectious diseases in heterogeneous populations J Math Biol 28 4 1990 365 382 2117040 40 van den Driessche P. Watmough J. Heterogeneous social interactions and the COVID-19 lockdown outcome in a multi-group SEIR model Math Model Nat Phenom 15 36 2020 41 Khajanchi S. Bera S. Roy T.K. Mathematical analysis of the global dynamics of a HTLV-I infection model, considering the role of cytotoxic T-lymphocytes Math Comput Simulat 180 2021 42 Berman A. Plemmons R. Nonnegative matrices in the mathematical sciences 1979 Academic Press 43 Ruan S.G. Wang W.D. Levin S.A. The effect of global travel on the spread of SARS Math Biosci Eng 3 1 2012 205 218 44 Zhang K. Ji Y.P. Pan Q.W. Wei Y.M. Liu H. Sensitivity analysis and optimal treatment control for a mathematical model of Human Papillomavirus infection AIMS Math 5 5 2020 2646 2670 45 Huo H.F. Feng L.X. Global stability for an HIV/AIDS epidemic model with different latent stages and treatment Appl Math Model 37 3 2013 1480 1489 46 Tang B. Xia F. Tang S. Bragazzi N. Li Q. Sun X. The effectiveness of quarantine and isolation determine the trend of the COVID-19 epidemic in the final phase of the current outbreak in China Int J Infect Dis 96 2020 636 647 32689711

0

PMC9748869

NO-CC CODE

2022-12-15 23:22:47

no

TVZ verpleegkd prakt wet. 2022 Dec 14; 132(6):22

latin-1

oa_other

==== Front ISA Trans ISA Trans ISA Transactions 0019-0578 1879-2022 ISA. Published by Elsevier Ltd. S0019-0578(22)00637-1 10.1016/j.isatra.2022.12.006 Article The effect mitigation measures for COVID-19 by a fractional-order SEIHRDP model with individuals migration Lu Zhenzhen a Chen YangQuan b Yu Yongguang a⁎ Ren Guojian a Xu Conghui a Ma Weiyuan c Meng Xiangyun a a Department of Mathematics, Beijing Jiaotong University, Beijing, 100044, PR China b Mechatronics, Embedded Systems and Automation Lab, University of California, Merced, CA 95343, USA c School of Mathematics and Computer Science, Northwest Minzu University, Lanzhou, 730000, PR China ⁎ Corresponding author. 14 12 2022 14 12 2022 5 8 2020 22 11 2022 10 12 2022 © 2022 ISA. Published by Elsevier Ltd. All rights reserved. 2022 ISA Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. In this paper, the generalized SEIHRDP (susceptible-exposed-infective-hospitalized-recovered-death-insusceptible) fractional-order epidemic model is established with individual migration. Firstly, the global properties of the proposed system are studied. Particularly, the sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. Secondly, according to the real data in India and Brazil, it can all be concluded that the bilinear incidence rate has a better description of COVID-19 transmission. Meanwhile, multi-peak situation is considered in China, and it is shown that the proposed system can better predict the next peak. Finally, taking individual migration between Los Angeles and New York as an example, the spread of COVID-19 between cities can be effectively controlled by limiting individual movement, enhancing nucleic acid testing and reducing individual contact. Keywords Individual migration Fractional-order epidemic model Peak prediction Sensitivity ==== Body pmc1 Introduction End of 2019 saw the outbreak of the dangerous infectious disease COVID-19, which is brought on by a novel coronavirus. Global public health has been significantly impacted by the 13,837,395 diagnosed cases and the 590,702 death cases as of July 17, 2020 [1]. The quick rise in infection cases suggests that COVID-19 has a much greater ability to spread than MERS-CoV and SARS coronaviruses [2], [3]. A direction for taking the right steps can be provided by a deeper comprehension and insight of the epidemic tendencies. Numerous nations have implemented a variety of mitigating strategies to prevent the spread of COVID-19 since 23 January, 2020. These strategies include home isolation, herd immunity, limiting individual migration, and others. Lack of information on the dynamic mechanism relating to the severity of COVID-19 at the this early stage makes it extremely difficult to limit the spread of COVID-19. However, using a mathematical model, strategies can be measured to serve as a benchmark for determining whether mitigation strategies are adequate. During the modeling process, it is very important to describe how infectious diseases are transmitted between susceptible and infected individuals. Numerous studies have shown that the incidence rate, which measures the infection capacity of a single infected person per unit time, is a crucial tool for describing this process, where susceptible individuals come into contact with infected individuals and are then infected with such a predetermined probability [4], [5], [6], [7]. Meanwhile, Korobeinikov et al. [8] indicated that the stability of the endemic equilibrium point is closely related to the concave of the incidence rate with respect to the infected individuals. Therefore, many researchers established the epidemic model of COVID-19 under various incidence rates  [9], [10], [11], for example, Peng et al. [9] constructed the SEIR (E-exposed) epidemic model and they discovered that COVID-19’s first appearance might be traced to the end of December 2019. A SEIQRD (Q-diagnosed) model was taken into consideration by Xu et al. [10], which has some basic guiding relevance for predicting COVID-19. Besides, individual migration has a crucial effect on the evolution of infectious diseases. With the convenience between cities, individuals move more and more frequent and new infectious diseases develop more rapidly regionally and globally [12]. Numerous deterministic models with multiple patches have been presented in attempt to better understand how individual migration affects the spread of infectious illnesses [13], [14], [15]. Contrary to what was initially reported [16], COVID-19 is in fact spreading from person to person through continuous interpersonal contact [2]. Lu et al. [17] considered a fractional-order SEIHRD (H-hospitalized) model with inter-city networks and they found that COVID-19 could be reduced in low-risk areas, but increased in high-risk areas by restricting communication between cities. Meanwhile, cross-infection among cities are considered, while there is not consider for self-migration [17]. Therefore, it is of great practical significance to include individual migration in different cities or different countries with the modeling COVID-19. Furthermore, the migration of susceptible individuals, exposed individuals, infected individuals are studied in this paper. It is worth noting that the time which patients waits for treatment follows the power law distribution [18], which prompts the use of the Caputo fractional-order derivative [19]. Angstmann et al. [20] discovered how fractional operators naturally appear in their model if the recovery time is a power law distribution after building a SIR epidemic model. Meanwhile, this offers a chronic disease epidemic model in which long-term infected people have little chance of recovering. Based on this statement, several authors have stated that the fractional-order model plays an important role in the process of disease transmission. Khan et al. [21] recounted how individuals, bats, unidentified hosts, and the source of the illness interacted, and considered how crucial the fractional-order system was in preventing the spread of the infection. To predict the spread of COVID-19, Chen et al. [22] developed a fractional-order epidemic model. Amjad et al. [23] built a fractional-order COVID-19 model and calculated the consequences of several mitigation and prevention strategies. Motivated by the above discussion, a fractional-order SEIHRDP epidemic model with individuals movement is established in this paper to study COVID-19. Meanwhile, the number of hospitalizations is the same as confirmed isolation in China, and but in other countries, these two are not equal, which the number of confirmed case is greater than that of hospitalized case. So in order to give a more generalized model, the purpose of this paper is to describe hospitalized individuals in response to the spread of COVID-19. The infectiousness of the incubation time is also taken into consideration, as inspired by [24]. Then, the proposed system’s dynamic behaviors are investigated in order to show the existence and uniqueness of the nonnegative solution, the global asymptotic stability of the disease-free equilibrium, and the uniform persistence, all of which have theoretical implications for future COVID-19 intervention and prevention. Meanwhile, the basic reproduction number with and without individual migration are compared, and it is found that adding individual migration can effectively describe the spread of COVID-19. Furthermore, the sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. Meanwhile, considering India and Brazil, results suggest that the bilinear incidence rate may be more fitted than the saturation incidence rate for stimulating the spread of COVID-19. When individuals movement is not considered, it can be found the proposed fractional-order model can better predict than the integer-order for multi peaks of COVID-19 in China. Meanwhile, when individuals movement is considered, the epidemic in the United States is analyzed and some mitigation measures are carried out to control the development of COVID-19. An implication of the achieved results is the possibility that the United States peaked on 24 November, 2020 (integer-order system) and 1 January, 2021 (fractional-order system), however, the number of infections shows an downward trend after 17 July, 2020 as enhancing nucleic acid detection and reducing the contact rate. Meanwhile, considering measures to limit migration between New York and Los Angeles, and enhance nucleic acid detection and reduce exposure rates, it is evident that there is an immediate increase in confirmed cases before a drop. Based on the above analysis, a generalized fractional-order SEIHRDP epidemic model with individual migration is considered. The main contributions of this study are as follows: • A fractional epidemic model with self-migration is considered, in which the infectivity of exposed individuals and hospitalized individuals are also taken into account. • The global properties of the proposed model are investigated, including the existence and uniqueness of global positive solutions, the local and global stability of disease-free equilibrium points, the persistence of disease transmission. • The sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. • Based on real data, the impact of the incidence rate on modeling COVID-19 is studied in India and Brazil. • Multiple peaks of COVID-19 transmission in China are analyzed by the proposed system. • Individual movement in the spread of COVID-19 in the United States is investigated and the peak are analyzed based on mitigation measures, such as enhanced nucleic acid testing, reduced of individual exposure, and control of individual movement. The rest of this paper is organized as follows. The SEIHRDP fractional-order model with individual movement is developed for COVID-19 in Section 2 and provides some preliminaries. Then dynamic properties of the proposed system are examined in Section 3. The theoretical results are shown using numerical simulations in Section 4. Finally, Section 5 provides the conclusions. 2 System description and preliminaries Fractional-order operator have been determined to have a wide range of uses in the modeling of many dynamic processes, including those in engineering, biology, medicine, and others [25], [26], [27], [28]. In this part, some necessary preliminaries are introduced before the fractional-order epidemic model is presented. 2.1 Preliminaries Definition 2.1 [29] The Caputo fractional-order operator is defined by t0CDtαgt=dαgtdtα=1Γn−α∫0tg(n)(s)t−sα−n+1ds,(n−1<α<n), where g(n)(s) is the nth derivative of g(s) with respect to s. Remark 2.1 If α=n, one has t0CDtαgt=g(n)(t). Lemma 2.1 [30] The Caputo nonlinear system is considered as follows: 0CDtαx(t)=g(x),(α∈(0,1]), with the initial condition x0 . If all eigenvalues of J|x=x∗=∂g∂x|x=x∗ satisfy |arg(λ)|>απ2 , the equilibrium points x∗ are locally asymptotically stable. Lemma 2.2 [31] Suppose X⊂R and the continuous operator T(t):X→X satisfies (1) T(t) is point dissipative in X and compact for t≥0 . (2) there is a finite sequence M={M1,M2,…,Mk} of compact and isolated invariant sets such that (i) Mi∩Mj=0̸ for any i,j=1,2,…,k and i≠j ; (ii) Ω(∂X0)≜∪x∈∂X0ω(x)⊂∪i=1kMi ; (iii) in the case of ∂X0 , no a cycle is formed by any subset of M ; (iv) Ws(Mi)∩X0=0̸ for each i=1,2,…,k . Then T(t) is uniformly persistent in X . 2.2 Graph theory In this paper, a weighted graph ζ=(ϑ,ω,A) will be considered to model the spread of infectious diseases between cities, where ϑ={ϑ1,ϑ2,…,ϑn} denotes the node set and ϑi represents the ith city; ω⊆ϑ×ϑ is the edge set, and if there is individual movement between any two cities, it means that there is a edge between this two cities; matrixes M=[mij]1≤i,j≤n, N=[nij]1≤i,j≤n, P=[pij]1≤i,j≤n and Q=[qij]1≤i,j≤n represent the weighted adjacency matrix of susceptible, exposed, infected and recovered individual, respectively; mij, nij, pij and qij denote the migrate rate of susceptible, exposed, infected and recovered individual from city j to city i with aij≥0 (i≠j) and aii=0 (a=m,n,porq), respectively. Furthermore, based on the directivity of individual migration, the directed graph ζ is studied in this paper. 2.3 System description Starting from 23 January, 2020, the Chinese government has adopted a series of mitigation measures to effectively suppress the spread of COVID-19, such as implementing strict home isolation, restricting various traffic, strengthening nucleic acid testing, establishing shelter hospitals and so on. Meanwhile, other countries around the world have adopted different measures from China, such as social distancing and herd community strategy by British, protecting sensitive compartment from infection by Italy, transferring of critically ill patients with military aircraft by France, etc. Therefore, it is important to establish a generalized model of individual migration to simultaneously quantify the impact of interruption of policies on virus transmission. Moreover, Tang et al. [32] proposed that the exposed individual is infectious of COVID-19. Therefore, a fractional-order SEIHRDP epidemic model with individual migration is considered as follows: (1) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),0CDtαEk=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),0CDtαIk=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),0CDtαHk=δkIk−(λk+κk)Hk,0CDtαRk=λkHk+∑j=1n(qkjRj−qjkRk),0CDtαDk=κkHk,0CDtαPk=ρkSk. with the initial condition (2) Sk(0)=Sk0>0,Ek(0)=Ek0≥0,Ik(0)=Ik0≥0,Hk(0)=Hk0≥0, Rk(0)=Rk0≥0,Dk(0)=Dk0≥0,Pk(0)=Pk0≥0. The specific explanation of system (1) are as follows: • Susceptible Sk: the number of susceptible class within city k at time t. • Exposed Ek: the number of exposed class within city k at time t (neither any clinical symptoms nor high infectivity). • Infectious Ik: the number of infected class within city k at time t (with overt symptoms). • Hospitalized Hk: the number of hospitalized class within city k at time t. • Recovered Rk: the number of recovered class within city k at time t. • Dead Dk: the number of dead class within city k at time t. • Insusceptible Pk: the number of susceptible class who are not exposed to the external community within city k at time t. Meanwhile, the process of disease transmission are as follows: • The susceptible individual Sk contacts with Ek and Ik, and then is infected by β1kSkfk(Ik)+β2kSkgk(Ek), where βik (i=1,2) are the transmission coefficient, fk(Ik) and gk(Ek) are generalized incidence rates. • The parameter Λk is the inflow rate; λk, ϵk, δk and κk represent the recovery, incubation, diagnosis, mortality rate. • The susceptible, exposed, infective and recovered individuals in city j move to city k with probability mkj, nkj, pkj and qkj, respectively. The terms ∑j=1n(mkjSj−mjkSk), ∑j=1n(nkjEj−njkEk), ∑j=1n(pkjIj−pjkIk) and ∑j=1n(qkjRj−qjkRk) represent the movement of Sk, Ek Ik and Rk individual, where ∑j=1nakjWj represents the individuals moving into k city from other cities j (k≠j) and ∑j=1najkWj represents the individuals leaving city k (W=S, E, I, R, respectively, a=m, n, p, q, respectively). • The movement of insuspectible and hospitalized individuals is not considered in this paper. Furthermore, Λk, βik (i=1,2), ρk and ϵk are positive constants; functions δk(t), λk(t), κk(t), mkj(t), nkj(t), pkj(t) and qkj(t) satisfy |δk(t)|≤M1k, |λk(t)|≤M2k, |κk(t)|≤M3k, |mkj(t)|≤M4k, |nkj(t)|≤M5k, |pkj(t)|≤M6k and |qkj(t)|≤M7k for all t≥0 and k,j=1,2,…,n, where M1k, M2k, M3k, M4k, M5k, M6k and M7k are positive constants. The transmission diagram of the generalized SEIHRDP model (1) is shown in Fig. 1. Before presenting the major findings, the following generalized incidence rate hypothesis is put forth: (H):(i)gk(Ek)andfk(Ik)satisfythelocalLipschitzconditionand gk(0)=0,fk(0)=0fork=1,2,…,n;(ii)fk(Ik)isstrictlymonotoneincreasingonIk∈[0,∞)and gk(Ek)isstrictlymonotoneincreasingon Ek∈[0,∞)forallk=1,2,…,n;(iii)fk(Ik)≤akIkforallIk≥0, whereak=fk′(0)forallk=1,2,…,n;(iv)gk(Ek)≤bkEkforallEk≥0, wherebk=gk′(0)forallk=1,2,…,n. Fig. 1 The schematic diagram of SEIHRDP epidemic model with individual migration (i,j=1,2,…,n). Remark 2.2 It should be noted that many current models can be viewed as a special type of system (1) with the hypothesis (H), such as gk(Ek)=bkEk, gk(Ek)=bkEk1+vkEk, fk(Ik)=akIk, fk(Ik)=akIk1+ukIk and others [33] with nonnegative constants ak, bk, uk and vk. Remark 2.3 Compared with [34], the individual movement in this paper can be described as follows: (1) the self-migration of individuals is described in system (1), which is caused by a self-chemotactic-like forcing [35]. However, the cross-infection among cities is considered which is a travel infectious [34]. (2) system (1) describes not only the migration of infected individuals, but also the movement of exposed and recovered individuals. (3) M=[mij]1≤i,j≤n, N=[nij]1≤i,j≤n, P=[pij]1≤i,j≤n and Q=[qij]1≤i,j≤n are not irreducible in this paper, but irreducible in [34]. Then the influence of network structure on disease transmission can be discussed in this paper, such as fully connected network, ring network and centralized network. However, [34] only consider fully connected network. (4) the total population of each city is changed (without considering the decrease in population due to death) in system (1). But in [34], the total population of each city remains constant. 3 System analysis This study explores system (1)’s dynamic analysis. As can be seen, the death class Dk and the insusceptible class Pk have no effect on the susceptible class Sk, exposed class Ek, infected class Ik, hospitalized class Hk, or recovered class Rk of systems (1). Accordingly, the following system is discussed in the next section: (3) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),0CDtαEk=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),0CDtαIk=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),0CDtαHk=δkIk−(λk+κk)Hk,0CDtαRk=λkHk+∑j=1n(qkjRj−qjkRk), with the initial condition (4) Sk(0)=Sk0>0,Ek(0)=Ek0≥0,Ik(0)=Ik0≥0, Hk(0)=Hk0≥0,Rk(0)=Rk0≥0,(k=1,2,…,n). 3.1 Existence and uniqueness of the positive solution The existence, uniqueness, and boundedness of the nonnegative solution for system (3) should be taken into account prior to the numerical process. Therefore, this subsection will be discussed these properties for system (3). Theorem 3.1 For any nonnegative initial condition (4) , there are a unique solution for system (3) and the region Ω={(S1,E1,I1,H1,R1,…,Sn,En,In,Hn,Rn): 0<Si≤Λ¯ρ,0≤Ei≤Λ¯ρ,0≤Ii≤Λ¯ρ, 0≤Hi≤Λ¯ρ,0≤Ri≤Λ¯ρ,i=1,2,…,n} is positively invariant for system (3) , where Λ¯=∑j=1nΛj and ρ=min{ρ1,ρ2,…,ρn} . Proof Let consider the following function: f1k=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),f2k=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),f3k=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),f4k=δkIk−(λk+κk)Hk,f5k=λkHk+∑j=1n(qkjRj−qjkRk). It is obvious that Fk=(f1k,f2k,f3k,f4k,f5k) satisfies the local Lipschitz condition about (Sk,Ek,Ik,Hk,Rk), then system (3) has a unique solution. Next, the nonnegative solution will be analyzed. Consider the following auxiliary system: 0CDtαS_k=−β1kS_kfk(I_k)−β2kS_kgk(E_k)−ρkS_k+∑j=1n(mkjS_j−mjkS_k),0CDtαE_k=β1kS_kfk(I_k)+β2kS_kgk(I_k)−ϵkE_k+∑j=1n(nkjE_j−njkE_k),0CDtαI_k=ϵkE_k−δkI_k+∑j=1n(pkjI_j−pjkI_k),0CDtαH_k=δkI_k−(λk+κk)H_k,0CDtαR_k=λkH_k+∑j=1n(qkjR_j−qjkR_k),Sk_(0)=Ek_(0)=Ik_(0)=Hk_(0)=Rk_(0)=0. Through the comparison theorem, it is not difficult to find that the following auxiliary system has a unique solution (0,0,0,0,0). then the following equation holds: (Sk,Ek,Ik,Hk,Rk)>(0,0,0,0,0). Next, adding all equations gives 0CDtαN≤Λ¯−ρN where N=∑j=1n(Sj+Ej+Ij+Hj+Rj+Dj), Λ¯=∑j=1nΛj and ρ=min{ρ1,ρ2,…,ρn}. Then N(t)≤(N(0)−Λ¯ρ)Eα(−ρtα)+Λ¯ρ. Therefore, the region Ω is positively invariant for system (3).  □ 3.2 Local stability The exploration of the existence and local stability of the disease-free equilibrium point is the focus of this section. Theorem 3.2 There are a unique disease-free equilibrium point E0=(S1∗,0,0,0,0,…,Sn∗,0,0,0,0) for system (3) where S∗=(S1∗,…,Sn∗) , S∗=A−1Λ , Λ=(Λ1,…,Λn) and A=ρ1+∑j≠1nmj1−m12⋯−m1n−m21ρ2+∑j≠2nmj2⋯−m2n⋮⋮⋮⋮−mn1−mn2⋯ρ2+∑j≠nnmjn. Proof Obviously, E0 satisfies the following equation: Λk−ρkSk∗+∑j=1n(mkjSj∗−mjkSk∗)=0, then the above equation can be written as the following matrix form: AS∗=Λ. It can be found that the matrix A is strictly diagonally dominant, and then it follows from [36] that one has A−1≥0. So according to [37], there exists a unique solution S∗=A−1Λ. Therefore, there exists a unique disease-free equilibrium point E0 of system (3). □ The predicted number of secondary cases that a typical infectious individual should create in a community that is totally susceptible is known as the basic reproduction number R0. According to Watmough et al. [38], it can be determined that an infectious disease can commonly infect the community if one diseased individual can typically infect more than one susceptible individual when R0≥1. On the other hand, if R0<1, each infected individual produces less than one new infection, and the infectious diseases can not grow. Thus, it is very important to describe the relationship between the basic reproduction number and the spread of infectious diseases. Here, the basic reproduction number R0 is stated as follows. Theorem 3.3 Under hypothesis H, the basic reproduction number R0 is R0=ρ(F11V11−1−F12V11−1V21V22−1), where matrixes F11=diag(β21b1S1,…,β2nbnSn) , F12=diag(β11a1S1,…,β1nanSn) , V21=diag(−ϵ1,…,−ϵn) , V11=ϵ1+∑j≠1nn1j−n12⋯−n1n−n21ϵ2+∑j≠2nn2j⋯−n2n⋮⋮⋮⋮−nn1−nn2⋯ϵn+∑j≠nnnnj, and V22=δ1+∑j≠1np1j−p12⋯−p1n−p21δ2+∑j≠2np2j⋯−p2n⋮⋮⋮⋮−pn1−pn2⋯δn+∑j≠nnpnj. Proof Let consider the following matrixes: F0=β11S1f1(I1)+β21S1gk(E1)⋮β1nSnfn(In)+β2nSngn(En)0⋮00⋮0andV0=ϵ1E1−∑j=1n(n1jEj−nj1E1)⋮ϵnEn−∑j=nn(nnjEj−njnEn)−ϵ1E1+δ1I1−∑j=1n(p1jIj−pj1I1)⋮−ϵnEn+δnIn−∑j=nn(pnjIj−pjnIn)−δ1I1+(λ1+κ1)H1⋮−δ1In+(λn+κn)Hn. Let u=(E1,…,En,I1,…,In,H1,…,Hn), then take the derivative of F0 and V0 for u at the disease-free equilibrium point E0, respectively, we can see as follows: F=F11F120000000andV=V1100V21V2200V32V33, where F11=diag(β21b1S1∗,…,β2nbnSn∗), F12=diag(β11a1S1∗,…,β1nanSn∗), V21=diag(−ϵ1,…,−ϵn), V33=diag((λ1+κ1),…,(λn+κn)), V32=diag (−δ1,…,−δn), V11=ϵ1+∑j≠1nn1j−n12⋯−n1n−n21ϵ2+∑j≠2nn2j⋯−n2n⋮⋮⋮⋮−nn1−nn2⋯ϵn+∑j≠nnnnj, and V22=δ1+∑j≠1np1j−p12⋯−p1n−p21δ2+∑j≠2np2j⋯−p2n⋮⋮⋮⋮−pn1−pn2⋯δn+∑j≠nnpnj. Then according to [39], the basic reproduction number is as follows: R0=ρ(FV−1)=ρ(F11V11−1−F12V11−1V21V22−1), where ρ(F11V11−1−F12V11−1V21V22−1) is the spectral radius of the matrix (F11V11−1−F12V11−1V21V22−1). □ Remark 3.1 According to [40], the epidemic size ςk=Sk(0)−Sk∗ of city k is defined as the number of individuals affected by the infectious disease, where Sk(0) is initial condition and Sk∗ is the disease-free equilibrium point of susceptible individuals within city k. Remark 3.2 When individual migration is not taken into consideration, it can be calculated from [10] that the basic reproduction number R0uk of city k is R0uk=Sk∗(β2kbkϵk+β1kakδk). Remark 3.3 When individual migration is taken into consideration, R0k of city k is R0k=Sk∗ϵk+∑j⁄=knnkj(β2kbk+β1kakϵkδk+∑j≠knpkj). Remark 3.4 It is easy to see that R0k are not dependent on λk, κk and mkj. Like [41], the other Λk, β1k, β2k, ρk, ϵk, nkj, pkj and δk are calculated as follows: AΛk=ΛkR0k∂R0k∂Λk=1,Aρk=ρkR0k∂R0k∂ρk=−1, Aβ1k=β1kR0k∂R0k∂β1k=β1kakϵkδk+∑j≠knpkjβ2kbk+β1kakϵkδk+∑j≠knpkj,Aβ2k=β2kR0k∂R0k∂β2k=β2kbkβ2kbk+β1kakϵkδk+∑j≠knpkj, Aδk=δkR0k∂R0k∂δk=−1(δk+∑j≠knpkj2)(β2kbk+β1kakϵkδk+∑j≠knpkj),Aϵk=ϵkR0k∂R0k∂ϵk=−1ϵk+∑j≠knnkj(β2kbk+ϵk+∑j≠knnkjδk+∑j≠knpkj), Ankj=nkjR0k∂R0k∂nkj=−1,Apkj=pkjR0k∂R0k∂pkj=−1(δk+∑j≠knpkj2)(β2kbk+β1kakϵkδk+∑j≠knpkj), where AΛk, Aρk, Aβ1k, Aβ2k, Aδk, Aϵk, Ankj and Apkj represent the normalized sensitivity on Λk, ρk, β1k, β2k, δk, ϵk, nkj and pkj, respectively. Through the above calculation found that the increase on Λk, β1k and β2k leads to the increase on R0k, but the increase on ρk, δk, ϵk, nkj and pkj leads to the decrease on R0k. In addition, the movement of susceptible individuals has no impact of R0k, but the movement of exposed and infected individuals is negatively correlated with R0k, and the movement of exposed individuals is more likely to influence the spread of the infectious disease with |Ankj|>|Apkj|. Theorem 3.4 Under hypothesis H, system (3) is locally asymptotically stable at the disease-free equilibrium point E0 if |arg(sF−V)|>απ2 . Proof The following Jacobian matrix at the disease-free equilibrium point E0 is considered: JE0=J11∗00F−V00∗J33, where matrixs J11=−ρ1−∑j=1nmj1m12⋯m1nm21−ρ2−∑j=1nmj2⋯m2n⋮⋮⋮⋮mn1mn2⋯−ρn−∑j=1nmjn, J33=−∑j=1nqj1q12⋯q1nq21−∑j=1nqj2⋯q2n⋮⋮⋮⋮qn1qn2⋯−∑j=1nqjn, F and V see Theorem 3.3. Then if all eigenvalues of the Jacobian matrix JE0 satisfy |arg(si)|>απ2, E0 is locally asymptotically stable and unstable if for some eigenvalues si, |arg(si)|≤απ2. Obviously, J11 and J33 are a nonsingular M-matrix, so J11 and J33 has all eigenvalues with negative real parts according to [42]. Consequently the local stability of E0 depends only on eigenvalues of F−V. Thus, if all eigenvalues of F−V satisfy |arg(sF−V)|>απ2, system (3) is locally asymptotically stable. □ Remark 3.5 If all the eigenvalues of F−V are negative, that is |arg(sF−V)|=π>απ2, system (3) is locally asymptotically stable. Meanwhile, it is obvious that |arg(sF−V)|=π⇔sF−V<0⇔ρ(FV−1)<1⇔R0<1. It can be yielded that if R0<1, the disease-free equilibrium point E0 is locally asymptotically stable of system (3) . 3.3 Global asymptotic stability of the disease-free equilibrium In this subsection, the global asymptotic stability of the disease-free equilibrium point E0 is discussed firstly. Furthermore, the uniform persistence of system (3) is also considered. Theorem 3.5 Under hypothesis (H) and |arg(sF−V)|>απ2 , the disease-free equilibrium point E0 is globally asymptotically stable of system (3) . Proof We use a method similar to the one used in [43]. Firstly, the boundedness of the susceptible class will be analyzed. According to Theorem 3.1 and hypothesis (H), we know Sk, Ek and Ik (k=1,2,…,n) are nonnegative, thus one has (5) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk +∑j=1n(mkjSj−mjkSk)≤Λk−ρkSk+∑j=1n(mkjSj−mjkSk). Let S=(S1,…,Sn), S∗=(S1∗,…,Sn∗), Λ=(Λ1,…,Λn) and A=ρ1+∑j≠1nmj1−m12⋯−m1n−m21ρ2+∑j≠2nmj2⋯−m2n⋮⋮⋮⋮−mn1−mn2⋯ρ2+∑j≠nnmjn, then Eq. (5) can be written in the following matrix: 0CDtαS≤Λ−AS=AS∗−AS, so it is easy to see that the conclusion holds as follows: S(t)≤(S0−S∗)Eα(−Atα)+S∗. Obviously, one has Sk≤Sk∗. Next, the global stability of Ek, Ik and Hk will be discussed. Based on hypothesis (H), one has fk(Ik)≤akIk and gk(Ek)≤bkEk. Then the following auxiliary system is considered: (6) 0CDtαEk¯=β1kSk∗akIk¯+β2kSk∗bkEk¯−ϵkEk¯+∑j=1n(nkjEj¯−njkEk¯),0CDtαIk¯=ϵkEk¯−δkIk¯+∑j=1n(pkjI¯j−pjkIk¯),0CDtαHk¯=δkIk¯−(λk+κk)Hk¯. It is easy to see that (7) 0CDtαW=(F−V)W, where W=(E¯,I¯,H¯), E¯=(E1¯,…,En¯), I¯=(I1¯,…,In¯), H¯=(H¯1,…,H¯n), F and V see Theorem 3.3. Thus, if |arg(sF−V)|>απ2, the above linear system (7) is locally asymptotically stable as well as globally asymptotically stable, that is limt→∞E¯k=limt→∞I¯k=limt→∞H¯k=0. According to the comparison theory and the nonnegative solution of Ek, Ik and Hk, one has limt→∞Ek=limt→∞Ik=limt→∞Hk=0. Based on the above analysis, when t→∞, one has 0CDtαS=AS∗−AS, then one has S(t)→S∗(t→∞). So E0 is globally asymptotically stable if |arg(sF−V)|>απ2. □ Remark 3.6 Similar to Theorem 3.3, it can be concluded that if R0<1, system (3) is globally asymptotically stable at the disease-free equilibrium point E0. Furthermore, the uniform persistence for system (3) is discussed in the following theorem. Theorem 3.6 Under hypothesis (H) and R0>1 , system (3) is uniformly persist, implying there exists a positive constant δ such that lim inft→+∞Sk≥δ,lim inft→+∞Ek≥δ,lim inft→+∞Ik≥δ,lim inft→+∞Hk≥δ,lim inft→+∞Rk≥δ,1≤k≤n. Proof Let consider the following space: X=X1×X2×⋯×Xn,X0=X10×X20×⋯×Xn0,∂X=∂X1×∂X2×⋯×∂Xn, where X0 represents the interior of X, ∂X denotes the boundary of X and Xk={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek≥0,Ik≥0,Hk≥0,Rk≥0}, Xk0={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek>0,Ik>0,Hk>0,Rk>0}, ∂Xk={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek=0,Ik=0,Hk=0,Rk=0}. Meanwhile, let W(t)=(S1,E1,I1,H1,R1,…,Sn,En,In,Hn,Rn) be the solution of system (3) with initial value W(0)=W0∈X, then W(t)∈X according to Theorem 3.1. For any t≥0, a continuous map F(t):X→X is defined as follows: F(t)W0=W(t). In the following, the uniformly persistent of the map F will be analyzed based on Lemma 2.3. When t=0, one has F(0)W0=W(0), this is F(0)=I where I is the identity matrix. Meanwhile, it can be deduced that the following equation holds: F(t+s)W0=W(t+s)=F(t)W(s)=F(t)F(s)W0, implying F(0)=I. Additionly, it is easy to see that F(t) is C0-semigroup on X, point dissipative and compact in X. Furthermore, consider the following system: 0CDtαSk=Λk−ρkSk+∑j=1n(mkjSj−mjkSk). According to Theorem 3.3, Sk∗ is asymptotically stable, which finds that E0 in ∂X is a global attractor of F(t). Let M={M1}, where M1={E0}. Because of ak=fk′(0) and bkgk′(0), for all ϵ, there exists ϵ¯ that f(ϵ)>(ak−ϵ)ϵ¯andg(ϵ)>(bk−ϵ)ϵ¯. Let the stable set Ws(E0) of a compact invariant set E0 defined by Ws(E0)={Y0∈X:ω(Y0)≠0̸,ω(Y0)∈E0}, where ω(Y0) is ω-limit set through Y0. System (3) has a solution (Sk,Ek,Ik,Hk,Rk) when Ws(E0)∩X0≠0̸, implying Sk→0, Ek→0, Ik→0, Hk→0, Rk→0 (k=1,2,…,n) as t→∞. So there exists a constant τ>0 such that Sk>Sk∗−ϵ, Ek>ϵ, Ik>ϵ, Hk>ϵ and Rk>ϵ for t≥τ. Then according to the monotonicity of fk(Ik) and gk(Ek), one has f(Ik)>f(ϵ)>(ak−ϵ)ϵ¯andg(Ek)>g(ϵ)>(bk−ϵ)ϵ¯. So the following auxiliary system is considered: (8) 0CDtαEk_=β1kSk∗(ak−ϵ)ϵ¯Ik_+β2kSk∗(bk−ϵ)ϵ¯Ek_−ϵkEk_+∑j=1n(nkjEj_−njkEk_),0CDtαIk_=ϵkEk_−δkIk_+∑j=1n(pkjIj_−pjkIk_),0CDtαHk_=δkIk_−(λk+κk)Hk_. It is easy to see from system (8) that (9) 0CDtαW=(F−V)(ϵ¯,ϵ)W, where W=(E_,I_,H_), E_=(E_1,…,E_n), I_=(I_1,…,I_n) and H_=(H_1,…,H_n). Consider the basic reproduction number R0>1, then one has ρ(F11V11−1−F12V11−1V21V22−1)(ϵ¯,ϵ)>1, which results in a contradiction with Ek(t)→0 (t→∞). Hence one has Ws(E0)∩X0=0̸, implying it is uniformly persistent at the operator T(t), so system (3) is uniformly persistent if R0>1.  □ The existence of a positive equilibrium point is implied by the system (3)’s ultimate boundedenss and uniform persistence. As a result, we can derive the following theorem. Theorem 3.7 Under hypothesis (H) and R0>1 , there is at least one endemic equilibrium E∗=(S1∗,E1∗,I1∗,H1∗,R1∗,…,Sn∗,En∗,In∗,Hn∗,Rn∗) of system (3) satisfying Λk−β1kSk∗fk(Ik∗)−β2kSk∗gk(Ek∗)−ρkSk∗+∑j=1n(mkjSj∗−mjkSk∗)=0,β1kSk∗fk(Ik∗)+β2kSk∗gk(Ek∗)−ϵkEk∗+∑j=1n(nkjEj∗−njkEk∗)=0,ϵkEk∗−δkIk∗+∑j=1n(pkjIj∗−pjkIk∗)=0,δkIk∗−(λk−κk)Hk∗=0,λkHk∗+∑j=1n(qkjRj∗−qjkRk∗)=0. 4 Numerical simulation From the previous description, it is clear that E0 is globally asymptotically stable when R0<1 and conversely, system (3) is persistent, which can offer theoretical evidence for further COVID-19 prediction and control. Meanwhile, in order to analyze COVID-19 in different cities, this section is divided into two parts: no restrictions on individual migration and restrictions on individual migration. Furthermore, consider the corresponding integer-order model as follows: (10) dSkdt=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),dEkdt=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),dIkdt=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),dHkdt=δkIk−(λk+κk)Hk,dRkdt=λkHk+∑j=1n(qkjRj−qjkRk). 4.1 Data source The Johns Hopkins University Center for System Science and Engineering provided the real data for this study [1]. Data on accumulated and confirmed cases, recovered cases, and COVID-19 death cases were shared by the Johns Hopkins University on January 23, 2020. Assuming that the confirmed individuals mut be hospitalized, one has Hospitalized=Confirmed−Recovered−Death. Hence, we can get the real data of H(t), D(t) and R(t) for different city from 23 January to 17 July, 2020. 4.2 The generalized incidence rate As we know, Korobeinikov et al. [8] indicated that the stability of the endemic equilibrium point for infectious diseases is closely related to the concave of the incidence rate with respect to the infected individuals. Therefore, it is of practical significance to understand the role of different incidence rates in COVID-19. In this section, according to hypothesis (H), the bilinear incidence rate and the saturation incidence rate are discussed as follows: fk(Ik)=Ik,gk(Ek)=Ek. fk(Ik)=Ik1+ukIk,gk(Ek)=Ek1+vkEk. Meanwhile, as the public learns about COVID-19, the recovered rate and the disease-related mortality are time-varying rather than constant. Similar to [11], the best recovered rate λk and the best disease-related mortality κk are selected from the following equation: (11) κk=p1ep2(t−p3)+e−p2(t−p3),p1e(p2(t−p3))2,p1+e(p2(t+p3)),andλk=q11+e−q2(t−q3),q1+e−q2(t+q3), where qi and qi (i=1,2,3) are parameters for κk and λk, respectively. According to the real data reported by [2], the spread of COVID-19 in India and Brazil began on 30 January and 26 February, 2020, as the beginning of the outbreak of India and Brazil in this paper, respectively. According to Matlab function lsqcurvefit [11], the parameter identification results with system (3) and system (10) are depicted in Table 1, Table 2, respectively. Meanwhile, based on Table 1, Table 2, the five days forecast of India and Brazil are shown in Tables 3, 4, Figs. 2, 3, 4, 5, which the solid lines represent simulation results and circles represent real data. The results in Table 1, Table 2 show that the fractional-order system (3) can accurately forecast the real data in the upcoming week, with the real data of currently confirmed cases falling between 95% and 105% of the projected values. Table 1 Parameter identification of India. India Integer (Bilinear) Fractional (Bilinear) Integer (Saturation) Fractional (Saturation) Λ 0.3 0.6245 0.465 0.3 β1 0.3869 1.206 1.156 2.358 β2 0.5133 0.3079 0.3414 0.8 ϵ 0.0023 0.0555 0.0014 0.0692 ρ 0.0264 0.03 0.0188 0.0094 λ p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) κ b1e−q2(t−q3)2 q1e−q2(t−q3)2 q1e−q2(t−q3)2 q1e−q2(t−q3)2 Table 2 Parameter identification of Brazil. Brazil Integer (Bilinear) Fractional (Bilinear) Integer (Saturation) Fractional (Saturation) Λ 0.3 0.5 0.8802 0.5189 β1 1.839 1.082 0.2378 4.804 β2 0.3733 0.928 0.4397 0.3 ϵ 0.0303 0.7905 0.1076 0.9609 ρ 0.0211 0.0214 0.0216 0.0431 δ 0.99 0.2434 0.3975 5.609×10−5 λ p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) κ q1eq2(t−q3) q1eq2(t−q3) q1eq2(t−q3) q1eq2(t−q3) Table 3 Estimate the number of confirmed cases within five days in India (×105). India 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 3.735 3.906 4.027 4.113 4.263 Integer (bilinear incidence rate) 3.454 3.499 3.52 3.556 3.583 Fractional (bilinear incidence rate) 3.781 3.869 3.936 4.002 4.079 Integer (saturation incidence rate) 3.426 3.476 3.515 3.553 3.578 Fractional (saturation incidence rate) 3.645 3.696 3.741 3.792 3.815 Table 4 Estimate the number of confirmed cases within five days in Brazil (×105). Brazil 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 5.487 5.598 5.242 5.228 5.528 Integer (bilinear incidence rate) 5.864 5.847 5.828 5.806 5.781 Fractional (bilinear incidence rate) 5.502 5.49 5.475 5.458 5.438 Integer (saturation incidence rate) 5.897 5.882 5.864 5.853 5.821 Fractional (saturation incidence rate) 5.857 5.844 5.819 5.793 5.769 Fig. 2 The number of cases in India (Integer-order with the bilinear incidence rate (left), Fractional-order with the bilinear incidence rate (right)). Fig. 3 The number of cases in India (Integer-order with the saturation incidence rate (left), Fractional-order with the saturation incidence rate (right)). Fig. 4 The number of cases in Brazil (Integer-order with the bilinear incidence rate (left), Fractional-order with the bilinear incidence rate (right)). Fig. 5 The number of cases in Brazil (Integer-order with the saturation incidence rate (left), Fractional-order with the saturation incidence rate (right)). 4.3 Restrict individual migration When individual movement is not considered, the parameters satisfy mkj=nkj=pkj=qkj=0. Meanwhile, according to Section 4.2, the bilinear incidence rate is considered in this section. Then based on system (1), the following auxiliary system is considered: (12) 0CDtαS=Λ−β1SIβ2SE−ρS,0CDtαE=β1SI+β2SEϵE,0CDtαI=ϵE−δI,0CDtαH=δI−(λ+κ)H,0CDtαR=λH,0CDtαD=κH. 4.3.1 Sensitivity analysis of parameters in R0uk When individual movement is not taken into consideration in this section, Partial Rank Correlation Coefficients (PRCC) value and Latin hypercube sampling (LHS) [44], which are one of the Monte Carlo (MC) sampling methods established by Mckay in 1979 [45], can be used to account for the sensitivity of the parameter to the basic reproduction number. LHS has the advantage of using fewer iterations than other random sampling techniques and avoiding the clustering phenomenon of sampling [45]. In order to determine which aspects of a certain intervention have the greatest impact on how quickly a new infection spreads, it can be seen from Remark 3.2 that the parameters of system (12) all affect the basic reproduction number to varying degrees, thereby affecting the spread of the infectious disease. We perform LHS on the parameters that appear in R0uk. PRCC are calculated, and a total of 1000 simulations per LHS run are carried out. A uniform distribution is chose as the prior distribution when performing parameter sampling. The parameters Λ, ρ, ϵ, β1, β2 and δ of system (12) are set as input variables, and the basic reproduction number R0uk as the output. The specific process is as follows: (1) There are six parameters that affect the change of R0uk, which are Λ, ρ, ϵ, β1, β2 and δ. Through LSH, [0,1] is divided into 1000 simulations, and 6 × 1000 parameters are generated through random selection on each interval by a uniform distribution. (2) Calculate the basic reproduction number R0uk for each parameter. (3) PRCC is calculated by Matlab function partialcorr. (4) The PRCC’s influence on the basic reproduction number R0uk can increase with increasing PRCC absolute value. However, it is believed that the parameter is not significant if the p value is greater than 0.05. Table 5 lists the PRCC values of the six parameters associated with R0uk and Fig. 6 shows the histogram of PRCC value. From Table 5 and Fig. 6, the following conclusion holds: (1) the parameters Λ, β1 and β2 have a positive influence on R0uk, but ρ, ϵ and δ have a negative influence, which is consistent with Remark 3.4; (2) the positive impact of birth rate Λ is the most obvious with PRCC(Λ)=0.5868; (3) the positive impact of the transmission rate β2 for the exposed population is more obvious than that of the infected population with PRCC(β2)>PRCC(β1). That is, the greater the transmission coefficient of the exposed population, the greater the value of the basic reproduction number R0uk, and then the greater the number of people infected with COVID-19. Therefore, it is more critical to limit exposed individual. However, because exposed individual doed not show any symptoms, identifying them is very difficult, which is a key reason for the spread of COVID-19; (4) the diagnosis rate δ has more greater negative impact on R0uk. That is to say, enhancing nucleic acid detection can effectively reduce R0ui, thereby reducing the number of infected people; (5) from the p-value, it can be found that the p-values of all parameters are less than 0.05, so they all have a significant impact on the basic reproduction number R0uk. Therefore, based on the above analysis, it can be obtained that controlling the influx of foreign population and enhancing nucleic acid detection are the most effective measures to control COVID-19. Meanwhile, home isolation can also control COVID-19. Therefore, this evidence confirms the effectiveness of Chinese government’s interruption policies, such as home isolation, prohibition of the inflow of foreign population, and enhancing nucleic acid detection, which may provide a good reference for the other countries. Table 5 The PRCC values and p-value of the parameters with respect to R0uk. Input PRCC values p-value Λ 0.5868 0 ρ −0.5363 0 ϵ −0.1368 0 β2 0.1035 3.5×10−6 β1 0.0847 1.448×10−4 δ −0.4362 0 Fig. 6 The sensitivity analysis of R0uk. 4.3.2 China’s second outbreak From the analysis in Section 4.3.1, it can be found that enhancing the diagnosis rate and controlling the inflow of foreign population can effectively control the spread of the epidemic. For China, individual migration has been strictly restricted at the beginning of COVID-19. Therefore, the impact of enhanced diagnosis rate will be only considered in this section. Due to the increase in public awareness and the development of detection technology, the time from onset to diagnosis is gradually shortened. Additionally, despite the use of the nucleic acid test method, the number of confirmed cases climbed significantly and peaked in early February 2020 as a result of the use of the CT diagnosis method. As a result, it is assumed that starting on 12 February, 2020, China’s diagnosis rate can reach and remain at its highest level. However, the third COVID-19 wave has been occurring in Beijing since the end of June 2020. Beijing has said that starting on 17 June, 2020, nucleic acid could be more readily detected. As a result, a new distribution, rather than the max level dated June 17, now governs the diagnostic rate. Similar to [46], the following piecewise function are described the diagnosed period of two and three peaks: (13) 1δk=(1δ0−1δe)e−w1t+1δe,t<t1,1δe,t≥t1,and 1δk=(1δ0−1δe)e−w1t+1δe,t<t1,1δe,t1≤t≤t2,(1δe−1δf)e−w2(t−t2)+1δf,t>t2, where δ0, δe (δe>δ0), w1, w2 and δf are similar to [46], t1 is 13 February, 2020, t2 is 17 June, 2020. Meanwhile, similar to [11], the best recovered rate λk and the best disease-related mortality κk are selected from Eq. (11). Then system (12) and system (10) are solved by predictor–correctors scheme and least squares method [11] by the real data from 23 January to 17 July, which 17 March, 2020 is considered as the beginning of the emergency in Heilongjiang, Shanghai and Guangdong, and 17 June, 2020 are considered as the beginning of the emergency in Beijing, respectively. From Fig. 7, Fig. 8, the fractional-order system (12) is found to fit the real data more accurately than the integer-order system (10) does. and COVID-19 in Beijing, Shanghai reaches its highest peak in a short time but there may be fourth wave peak, however, Heilongjiang and Guangdong are only two peaks and the third wave of epidemic peaks will not occur in a short time (current policies remain unchanged). Therefore, under the condition of restricting the migration of individuals, the fractional system (12) can better simulate the multi-peak problem of COVID-19, and the strengthening of nucleic acid detection can predict the new wave in advance, which provides a theoretical basis for the control of the epidemic. Fig. 7 The number of cases in Beijing and Shanghai. Fig. 8 The number of cases in Guangdong and Heilongjiang. 4.4 Individual migration As of 17 July, 2020, the United States has a total of 3,647,715 confirmed cases, 139266 deaths and 1,107,204 recovery cases. It is urgent to formulate reasonable and effective mitigation measures. Thus in this section, based on the sensitivity analysis of parameter to R0k, the effect mitigation measures are provided to control the development of COVID-19 in US. 4.4.1 Sensitivity analysis of parameters in R0k Similar to Section 4.3.1, consider two cities to examine the sensitivity of parameters to R0k (k=1,2). Then when n=2, system (3) can be simplified as follows: (14) 0CDtαS1=Λ1−β11S1I1−β21S1E1−ρ1S1+(m12S2−m21S1),0CDtαE1=β11S1I1+β21S1E1−ϵ1E1+(n12E2−n21E2),0CDtαI1=ϵ1E1−δ1I1+(p12I2−p21I1),0CDtαH1=δ1I1−(λ1+κ1)H1,0CDtαR1=λ1H1+(q12R2−q21R1),0CDtαS2=Λ2−β12S2I2−β22S2E2−ρ2S2+(m21S1−m12S2),0CDtαE2=β12S1I2+β22S2E2−ϵ1E2+(n21E1−n12E1),0CDtαI2=ϵ2E2−δ2I2+(p21I1−p12I2),0CDtαH2=δ2I2−(λ2+κ2)H2,0CDtαR2=λ2H2+(q21R1−q12R2). It can be found from Remark 3.4 that there exists 16 parameter of the basic reproduction number R0k (k=1,2), and then the 16 parameters are set as input variables, and R0k as the output. Similar to Section 4.3.1, Table 6 lists the PRCC values and Fig. 9 shows the histogram of PRCC value. According to Table 6 and Fig. 9, it can be found that the following conclusion holds: (1) the movement of susceptible individuals mkj (k,j=1,2) does not affect R0k; (2) the sensitivity of the parameter to R0k (k=1,2) is same as that of Section 4.2.1, except for n12, n21, p12 and p21; (3) considering the basic reproduction number R01 of city 1, the p-value of p21 is large than 0.05, which means that infected individuals migrating from city 1 have a significant impact on COVID-19 in city 1. But exposed and infected individuals migrating to city 1 have an impact on the spread of COVID-19 in city 1, and the impact of the inflow of exposed individuals is more significant because of |PRCC(n12)|>|PRCC(p12)|; (4) contrary to the situation in city 1, the p-value of and p12 is large than 0.05, which means that infected individuals migrating from city 2 have a significant impact on the spread of disease in city 2. But exposed and infected individuals migrating from city 2 have an impact on the spread of COVID-19 in city 2, and the impact of the inflow of exposed individuals is more significant because of |PRCC(n21)|>|PRCC(p21)|. Therefore, in order to alleviate the situation in severe areas of COVID-19, migration of exposed individuals must be strictly controlled. Table 6 The PRCC values and p-value of the parameters with respect to R01 (left) and R02 (right). Input PRCC values p-value Λ1 0.6513 0 ρ1 −0.5294 0 ϵ1 −0.0348 0.1194 β21 0.1983 0 β11 0.0564 0.0116 δ1 −0.1427 0 n12 −0.3584 0 n21 −0.0286 0.2005 p12 −0.1732 0 p21 −0.0062 0.7806 Input PRCC values p-value Λ2 0.6496 0 ρ2 −0.5223 0 ϵ2 −0.0451 0.0435 β22 0.1896 0 β12 0.052 0.0153 δ2 −0.1691 0 n12 −0.0217 0.3318 n21 −0.2732 0 p12 −0.0048 0.8316 p21 −0.1355 0 Fig. 9 The sensitivity analysis of R01 (left) and R02 (right). 4.4.2 US outbreak In this subsection, the overall spread of COVID-19 in the US is considered first. Then system (10) and system (12) are solved by least squares method [11]. However, beginning 17 May, 2020, the number of confirmed individuals in the US had significantly increased. Emergency situations may have changed government regulations and people’s attitudes, which led to an increase in the number of sick people. Therefore, it is assumed that the emergency starts on 17 May, and the outbreak’s spread in the US is then examined in two stages as follows: (1) 23 January-17 May, 2020; (2) 17 May-17 July, 2020. Therefore, parameter identification is provided in Table 7 based on actual data from 23 January to 17 July 2020. From Fig. 10 and Table 8, it is clear that the fractional-order system (12) is capable of accurately forecasting the confirmed case for the upcoming week. In the meantime, Table 8 shows that, regardless of whether in the first stage or the second stage, the parameter findings of the fractional-order system and integer-order fitting are totally different. Based on Remark 3.2, R0uk=49.84 is very high. From the analysis in Section 4.2.1, it can be found that enhancing the diagnosis rate, reducing contact with infected people and controlling the inflow of foreign population can effectively control the spread of COVID-19. However, the United States is not currently doing anything to limit the influx of foreign population, so it is only considering enhancing nucleic acid testing and reducing contact with infected people to control COVID-19. Like [46], the diagnosed period 1δk of US are as follows: (15) 1δ(t)=1δet≤t3,(1δ0−1δe)e−w(t−t3)+1δet>t3. The meaning of each symbol is similar to that in Section 4.2.3. t3 is 17 July, 2020, which mean increasing the diagnosis rate δ(t) from 17 July, 2020. At the same time, the contact rate βi (i=1,2) is limited by the number of hospitalizations like [46] as follows: (16) βi(t)=βi,logH(t)≤1,βilogH(t),logH(t)>1. It can be seen from Fig. 10 that increasing the diagnosis rate δ(t) and controlling the infection rate βi (i=1,2) can effectively contain COVID-19. Therefore, enhanced nucleic acid testing and limited contact with infected individuals are important to control COVID-19.Table 7 Parameter identification of US. US Integer (first stage) Fractional (first stage) Integer (second stage) Fractional (second stage) Λ 0.3 0.0935 0.3 0.4593 β1 1.056 1.217 4.999 2.641 β2 0.2969 0.4882 1.648×10−7 3.724×10−7 ϵ 0.1791 0.2876 0.0087 0.0077 ρ 0.0281 0.0497 0.0358 0.0272 δ 0.1243 0.2434 0.3975 0.205 λ p1+e−p2(t+p3) p1+e−p2(t+p3) p1+e−p2(t+p3) p1+e−p2(t+p3) κ q1eq2(t−q3)+e−q2(t−q3) q1eq2(t−q3)+e−q2(t−q3) q1+e−q2(t+q3) q1+e−q2(t+q3) Table 8 Estimate the number of confirmed cases within five days in US (×106). Date Real data Fractional Integer 18 July 2.449 2.503 2.397 19 July 2.502 2.541 2.425 20 July 2.534 2.578 2.454 21 July 2.575 2.616 2.482 22 July 2.617 2.655 2.511 Fig. 10 The number of cases in US (without control (left), with control (right)). 4.4.3 US with individual migration This subsection considers the impact of individual migration on COVID-19. We need to preprocess the data to remove data that are less than 0.5% of the current maximum number of confirmed cases. Therefore, the real data after 3 April are selected to identify the parameters of system (14). Similar to the analysis of Section 4.4.2, we consider 17 May, 2020 as the beginning of the emergency, and the COVID-19 spread in New York and Los Angeles into two phases: (1) 3 April-17 May, 2020; (2) 17 May-17 July, 2020. Meanwhile, the recovered data of New York and Los Angeles have not been collected by [1], and then we take hospitalized+recovered individuals as a whole to conduct parameter identification and short-term prediction according to [11]. It can be found from Table 9, Table 10 and Fig. 11 that system (14) can better predict COVID-19. Meanwhile, it can be seen from Fig. 11 that the COVID-19 in New York has been peaked but not in Los Angles. From the analysis of Section 4.4.1, we know that controlling the infection rate, improving the diagnosis rate and controlling the movement of exposed individuals have a significant effect on the control of COVID-19 in US. Therefore, similar to Section 4.4.2, the diagnosis rate δk and the migration rate nkj are utilized as follows: (17) 1δk(t)=1δe,t≤t3,(1δ0−1δe)e−w(t−t3)+1δe,t>t3,and nkj=nkj,log(Hk)<1,nkjlog(Hk),log(Hk)≥1. Meanwhile, the infection rate controlled by the number of hospitalizations is Eq. (16). From Fig. 12, we can seen the fractional-order system (14) with control (Eqs. (16), (17)) in Los Angles can be control quickly but not in New York, which is still an open question and will be discussed later.Table 9 Estimate the number of confirmed cases within five days in New York (×105). New York 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 1.977 1.98 1.983 1.987 1.99 Integer 1.978 1.979 1.98 1.981 1.981 Fractional 1.985 1.986 1.987 1.988 1.989 Table 10 Estimate the number of confirmed cases within five days in Los Angles (×105). Los Angles 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 1.491 1.518 1.549 1.579 1.609 Integer 1.414 1.444 1.464 1.482 1.514 Fractional 1.494 1.517 1.547 1.579 1.608 Fig. 11 The number of cases in New York and Los Angeles with individual movement. Fig. 12 The number of cases in New York and Los Angeles with control Eqs. (16), (17). 5 Conclusion Based on individual migration, a fractional-order SEIHRDP model is proposed with the generalized incidence rate. Meanwhile, some results and effective mitigation measures is suggested to control COVID-19 as follows: (1) The local and global asymptotic stability of the disease-free and endemic equilibrium points are investigated based on the basic reproduction number R0. (2) Based on the real data, it is found that the bilinear incidence rate has a better description of COVID-19 transmission than the saturation incidence rate. Therefore, the bilinear incidence rate is applied in modelling COVID-19. Meanwhile, this is the first time that looked at the impact of the incidence rate in the spread of COVID-19 using real data. (3) By applying the value of PRCC, the sensitivity of the parameters to the basic reproduction number R0k and R0uk are obtained, which is consistent with Remark 3.4. Through the PRCC value, the diagnosis rate, the migration rate and the movement of the infected population are most sensitive to control COVID-19. (4) Multiple peaks have been analyzed for COVID-19 and using four cities in China to show that the fractional-order system (1) works well. Moreover, by increasing the diagnosis rate, it can be found that the third wave of epidemic in Beijing has reached its peak, but the arrival of the next wave of COVID-19 is not ruled out. (5) Analyzing the situation in the United States, it can be seen that system (12) has better predictability than system (10). Meanwhile, by reducing the infection rate and increasing the diagnosis rate, the peak of the epidemic in the US can be accelerated. (6) Results show that the fractional-order system can accurately forecast the real data in the upcoming week when taking into account individual migration between two cities. By limiting the movement of exposed individuals, raising the diagnosis rate, and lowering the infection rate, Los Angeles’ peaks can appear and then decline immediately. Furthermore, this study makes several contributions to predict multi-peak of COVID-19 in China and suggestions on controlling epidemic in the US by changing certain parameters. Nevertheless, this research raises some issues that require more investigation, including how medical and other factors affect the spread of infectious diseases, how to properly administer vaccines, how network topology affects disease transmission and so on. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgment This work is supported by the Natural Science Foundation of Beijing Municipality [grant numbers Z180005], the National Nature Science Foundation of China [grant numbers 61772063] and the Fundamental Research Funds of the Central Universities [grant numbers 2020JBM074]. ==== Refs References 1 The Johns Hopkins University Center for System Science and Engineering, Data of accumulated and newly confirmed cases, recovered case and death case of COVID-19, URL https://github.com/CSSEGISandData/COVID-19. 2 Chan J. Yuan S. Kok K. To K. Chu H. Yang J. A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: a study of a family cluster Lancet 395 10223 2020 514 523 31986261 3 Huang C. Wang Y. Li X. Ren L. Zhao J. Hu Y. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China Lancet 395 10223 2020 497 506 31986264 4 Anderson R. Anderson B. May R. Infectious diseases of humans: Dynamics and control 1992 Oxford University Press 5 Bailey N. The mathematical theory of infectious diseases and its applications 1975 Charles Griffin Company Ltd 6 Upadhyay R. Pal A. Kumari S. Roy P. Dynamics of an SEIR epidemic model with nonlinear incidence and treatment rates Nonlinear Dynam 96 4 2019 2351 2368 7 Zou X. Wang K. Optimal harvesting for a stochastic regime-switching logistic diffusion system with jumps Nonlinear Anal Hybrid 13 2014 32 44 8 Korobeinikov A. Maini P. Nonlinear incidence and stability of infectious disease models Math Med Biol 22 2 2005 113 128 15778334 9 Peng L. Yang W. Zhang D. Zhuge C. Hong L. Epidemic analysis of COVID-19 in China by dynamical modeling 2020 10.1101/2020.02.16.20023465 arXiv preprint arXiv:2002.06563 10 Xu C. Yu Y. Chen Y. Lu Z. Forecast analysis of the epidemics trend of COVID-19 in the USA by a generalized fractional-order SEIR model Nonlinear Dynam 101 3 2020 1621 12634 11 Cheynet E. Generalized SEIR epidemic model (fitting and compution) 2020 GitHub URL http://www.github.com/ECheynet/SEIR 12 Gao D.Z. Cosner C. Cantrell R.S. Beier J.C. Ruan S.G. Modeling the spatial spread of Rift Valley fever in Egypt Bull Math Biol 75 3 2013 523 542 23377629 13 Phaijoo G.R. Gurung D.B. Mathematical study of dengue disease transmission in multi-patch environment Appl Math 7 14 2016 1521 1533 14 Eisenberg C.M. Shuai Z.S. Tien J.H. den Driessche P.V. A cholera model in a patchy environment with water and human movement Math Biosci 246 1 2013 105 112 23958383 15 Julien A. Davis J.R. David H. Richard J. Miller J.M. den Driessche P.V. A multi-species epidemic model with spatial dynamics Math Med Biol 2 2005 129 142 15778332 16 Cheng Z.K.J. Shan J. 2019 Novel coronavirus: where we are and what we know Infection 49 197 2021 17 Lu Z. Yu Y. Chen Y.Q. Ren G. Xu C. Wang S. A fractional-order SEIHDR model for COVID-19 with inter-city networked coupling effects Nonlinear Dynam 101 3 2020 1717 1730 18 Smethurst D. Williams H. Are hospital waiting lists self-regulating? Nature 410 6829 2001 652 653 11287943 19 Meerschaert M. Sikorskii A. Stochastic models for fractional calculus 2011 De Gruyter 20 Angstmann C. Henry B. Mcgann A. A fractional-order infectivity SIR model Physica A 2016 86 93 21 Khan M.A. Atangana A. Modeling the dynamics of novel coronavirus (2019-nCov) with fractional derivative Alex Eng J 59 4 2020 2379 2389 22 Chen Y. Cheng J. Jiang X. Xu X. The reconstruction and prediction algorithm of the fractional TDD for the local outbreak of COVID-19 2020 arXiv preprint arXiv:2002.10302 23 Amjad S.S. Najiroddin S.I. Kottakkaran N.S. A mathematical model of COVID-19 using fractional derivative: Outbreak in India with dynamics of transmission and control Adv Differ Equ 373 2020 1 19 32226454 24 Tang Z. Li X.B. Li H.Q. Prediction of new coronavirus infection based on a modified SEIR model MedRxiv 2020 10.1101/2020.03.03.20030858 25 Zhou P. Ma J. Tang J. Clarify the physical process for fractional dynamical systems Nonlinear Dynam 2020 1 12 26 Li H. Zhang L. Hu C. Jiang Y. Teng Z. Dynamical analysis of a fractional-order predator-prey model incorporating a prey refuge J Appl Math Comput 54 1–2 2017 435 449 27 Singh J. Kumar D. Hammouch Z. Atangana A. A fractional epidemiological model for computer viruses pertaining to a new fractional derivative Appl Math Comput 316 2018 504 515 28 Sierociuk D. Skovranek T. Macias M. Podlubny I. Petras I. Dzielinski A. Diffusion process modeling by using fractional-order models Appl Math Comput 257 2015 2 11 29 Podlubny I. Fractional differential equations 1999 Academic Press 30 Li Y. Chen Y. Podlubny I. Mittag–Leffler stability of fractional order nonlinear dynamic systems Automatica 45 8 2009 1965 1969 31 Tang Q. Teng Z. Jiang H. Global behaviors for a class of multi-group SIRS epidemic models with nonlinear incidence rate Taiwan J Math 19 5 2015 1509 1532 32 Tang Z. Li X. Li H. Prediction of new coronavirus infection based on a modified SEIR model Cold Spring Harbor Lab 2020 1 13 33 Lin M. Huang J. Ruan S. Yu P. Bifurcation analysis of an SIRS epidemic model with a generalized nonmonotone and saturated incidence rate J Differ Equ 267 3 2019 1859 1898 32226129 34 Muroya Y. Enatsu Y. Kuniya T. Global stability for a multi-group SIRS epidemic model with varying population sizes Nonlinear Anal Real 14 3 2013 1693 1704 35 Angstmann C.N. Donnelly I.C. Henry B.I. Langlands T. Continuous-time random walks on networks with vertex- and time-dependent forcing Phys Rev E 88 2 2013 022811 36 Li M. Graef J.R. Wang L.C. Karsai J. Global dynamics of a SEIR model with varying total population size Math Biosci 160 2 1999 191 213 10472754 37 Kheiri H. Jafari M. Stability analysis of a fractional order model for the HIV/AIDS epidemic in a patchy environment J Comput Appl Math 2018 323 339 38 van den Driessche P. Watmough J. Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmission Math Biosci 180 1–2 2002 29 48 12387915 39 Diekmann O. Heesterbeek J. Metz J. On the definition and the computation of the basic reproduction ratio R0 in models for infectious diseases in heterogeneous populations J Math Biol 28 4 1990 365 382 2117040 40 van den Driessche P. Watmough J. Heterogeneous social interactions and the COVID-19 lockdown outcome in a multi-group SEIR model Math Model Nat Phenom 15 36 2020 41 Khajanchi S. Bera S. Roy T.K. Mathematical analysis of the global dynamics of a HTLV-I infection model, considering the role of cytotoxic T-lymphocytes Math Comput Simulat 180 2021 42 Berman A. Plemmons R. Nonnegative matrices in the mathematical sciences 1979 Academic Press 43 Ruan S.G. Wang W.D. Levin S.A. The effect of global travel on the spread of SARS Math Biosci Eng 3 1 2012 205 218 44 Zhang K. Ji Y.P. Pan Q.W. Wei Y.M. Liu H. Sensitivity analysis and optimal treatment control for a mathematical model of Human Papillomavirus infection AIMS Math 5 5 2020 2646 2670 45 Huo H.F. Feng L.X. Global stability for an HIV/AIDS epidemic model with different latent stages and treatment Appl Math Model 37 3 2013 1480 1489 46 Tang B. Xia F. Tang S. Bragazzi N. Li Q. Sun X. The effectiveness of quarantine and isolation determine the trend of the COVID-19 epidemic in the final phase of the current outbreak in China Int J Infect Dis 96 2020 636 647 32689711

36370022

PMC9748871

NO-CC CODE

2022-12-15 23:22:47

no

Indian Pediatr. 2022 Dec 14; 59(11):891

latin-1

Indian Pediatr

10.1007/s13312-022-2653-8

oa_other

==== Front ISA Trans ISA Trans ISA Transactions 0019-0578 1879-2022 ISA. Published by Elsevier Ltd. S0019-0578(22)00637-1 10.1016/j.isatra.2022.12.006 Article The effect mitigation measures for COVID-19 by a fractional-order SEIHRDP model with individuals migration Lu Zhenzhen a Chen YangQuan b Yu Yongguang a⁎ Ren Guojian a Xu Conghui a Ma Weiyuan c Meng Xiangyun a a Department of Mathematics, Beijing Jiaotong University, Beijing, 100044, PR China b Mechatronics, Embedded Systems and Automation Lab, University of California, Merced, CA 95343, USA c School of Mathematics and Computer Science, Northwest Minzu University, Lanzhou, 730000, PR China ⁎ Corresponding author. 14 12 2022 14 12 2022 5 8 2020 22 11 2022 10 12 2022 © 2022 ISA. Published by Elsevier Ltd. All rights reserved. 2022 ISA Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. In this paper, the generalized SEIHRDP (susceptible-exposed-infective-hospitalized-recovered-death-insusceptible) fractional-order epidemic model is established with individual migration. Firstly, the global properties of the proposed system are studied. Particularly, the sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. Secondly, according to the real data in India and Brazil, it can all be concluded that the bilinear incidence rate has a better description of COVID-19 transmission. Meanwhile, multi-peak situation is considered in China, and it is shown that the proposed system can better predict the next peak. Finally, taking individual migration between Los Angeles and New York as an example, the spread of COVID-19 between cities can be effectively controlled by limiting individual movement, enhancing nucleic acid testing and reducing individual contact. Keywords Individual migration Fractional-order epidemic model Peak prediction Sensitivity ==== Body pmc1 Introduction End of 2019 saw the outbreak of the dangerous infectious disease COVID-19, which is brought on by a novel coronavirus. Global public health has been significantly impacted by the 13,837,395 diagnosed cases and the 590,702 death cases as of July 17, 2020 [1]. The quick rise in infection cases suggests that COVID-19 has a much greater ability to spread than MERS-CoV and SARS coronaviruses [2], [3]. A direction for taking the right steps can be provided by a deeper comprehension and insight of the epidemic tendencies. Numerous nations have implemented a variety of mitigating strategies to prevent the spread of COVID-19 since 23 January, 2020. These strategies include home isolation, herd immunity, limiting individual migration, and others. Lack of information on the dynamic mechanism relating to the severity of COVID-19 at the this early stage makes it extremely difficult to limit the spread of COVID-19. However, using a mathematical model, strategies can be measured to serve as a benchmark for determining whether mitigation strategies are adequate. During the modeling process, it is very important to describe how infectious diseases are transmitted between susceptible and infected individuals. Numerous studies have shown that the incidence rate, which measures the infection capacity of a single infected person per unit time, is a crucial tool for describing this process, where susceptible individuals come into contact with infected individuals and are then infected with such a predetermined probability [4], [5], [6], [7]. Meanwhile, Korobeinikov et al. [8] indicated that the stability of the endemic equilibrium point is closely related to the concave of the incidence rate with respect to the infected individuals. Therefore, many researchers established the epidemic model of COVID-19 under various incidence rates  [9], [10], [11], for example, Peng et al. [9] constructed the SEIR (E-exposed) epidemic model and they discovered that COVID-19’s first appearance might be traced to the end of December 2019. A SEIQRD (Q-diagnosed) model was taken into consideration by Xu et al. [10], which has some basic guiding relevance for predicting COVID-19. Besides, individual migration has a crucial effect on the evolution of infectious diseases. With the convenience between cities, individuals move more and more frequent and new infectious diseases develop more rapidly regionally and globally [12]. Numerous deterministic models with multiple patches have been presented in attempt to better understand how individual migration affects the spread of infectious illnesses [13], [14], [15]. Contrary to what was initially reported [16], COVID-19 is in fact spreading from person to person through continuous interpersonal contact [2]. Lu et al. [17] considered a fractional-order SEIHRD (H-hospitalized) model with inter-city networks and they found that COVID-19 could be reduced in low-risk areas, but increased in high-risk areas by restricting communication between cities. Meanwhile, cross-infection among cities are considered, while there is not consider for self-migration [17]. Therefore, it is of great practical significance to include individual migration in different cities or different countries with the modeling COVID-19. Furthermore, the migration of susceptible individuals, exposed individuals, infected individuals are studied in this paper. It is worth noting that the time which patients waits for treatment follows the power law distribution [18], which prompts the use of the Caputo fractional-order derivative [19]. Angstmann et al. [20] discovered how fractional operators naturally appear in their model if the recovery time is a power law distribution after building a SIR epidemic model. Meanwhile, this offers a chronic disease epidemic model in which long-term infected people have little chance of recovering. Based on this statement, several authors have stated that the fractional-order model plays an important role in the process of disease transmission. Khan et al. [21] recounted how individuals, bats, unidentified hosts, and the source of the illness interacted, and considered how crucial the fractional-order system was in preventing the spread of the infection. To predict the spread of COVID-19, Chen et al. [22] developed a fractional-order epidemic model. Amjad et al. [23] built a fractional-order COVID-19 model and calculated the consequences of several mitigation and prevention strategies. Motivated by the above discussion, a fractional-order SEIHRDP epidemic model with individuals movement is established in this paper to study COVID-19. Meanwhile, the number of hospitalizations is the same as confirmed isolation in China, and but in other countries, these two are not equal, which the number of confirmed case is greater than that of hospitalized case. So in order to give a more generalized model, the purpose of this paper is to describe hospitalized individuals in response to the spread of COVID-19. The infectiousness of the incubation time is also taken into consideration, as inspired by [24]. Then, the proposed system’s dynamic behaviors are investigated in order to show the existence and uniqueness of the nonnegative solution, the global asymptotic stability of the disease-free equilibrium, and the uniform persistence, all of which have theoretical implications for future COVID-19 intervention and prevention. Meanwhile, the basic reproduction number with and without individual migration are compared, and it is found that adding individual migration can effectively describe the spread of COVID-19. Furthermore, the sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. Meanwhile, considering India and Brazil, results suggest that the bilinear incidence rate may be more fitted than the saturation incidence rate for stimulating the spread of COVID-19. When individuals movement is not considered, it can be found the proposed fractional-order model can better predict than the integer-order for multi peaks of COVID-19 in China. Meanwhile, when individuals movement is considered, the epidemic in the United States is analyzed and some mitigation measures are carried out to control the development of COVID-19. An implication of the achieved results is the possibility that the United States peaked on 24 November, 2020 (integer-order system) and 1 January, 2021 (fractional-order system), however, the number of infections shows an downward trend after 17 July, 2020 as enhancing nucleic acid detection and reducing the contact rate. Meanwhile, considering measures to limit migration between New York and Los Angeles, and enhance nucleic acid detection and reduce exposure rates, it is evident that there is an immediate increase in confirmed cases before a drop. Based on the above analysis, a generalized fractional-order SEIHRDP epidemic model with individual migration is considered. The main contributions of this study are as follows: • A fractional epidemic model with self-migration is considered, in which the infectivity of exposed individuals and hospitalized individuals are also taken into account. • The global properties of the proposed model are investigated, including the existence and uniqueness of global positive solutions, the local and global stability of disease-free equilibrium points, the persistence of disease transmission. • The sensitivity of parameters to the basic reproduction number are analyzed both theoretically and numerically. • Based on real data, the impact of the incidence rate on modeling COVID-19 is studied in India and Brazil. • Multiple peaks of COVID-19 transmission in China are analyzed by the proposed system. • Individual movement in the spread of COVID-19 in the United States is investigated and the peak are analyzed based on mitigation measures, such as enhanced nucleic acid testing, reduced of individual exposure, and control of individual movement. The rest of this paper is organized as follows. The SEIHRDP fractional-order model with individual movement is developed for COVID-19 in Section 2 and provides some preliminaries. Then dynamic properties of the proposed system are examined in Section 3. The theoretical results are shown using numerical simulations in Section 4. Finally, Section 5 provides the conclusions. 2 System description and preliminaries Fractional-order operator have been determined to have a wide range of uses in the modeling of many dynamic processes, including those in engineering, biology, medicine, and others [25], [26], [27], [28]. In this part, some necessary preliminaries are introduced before the fractional-order epidemic model is presented. 2.1 Preliminaries Definition 2.1 [29] The Caputo fractional-order operator is defined by t0CDtαgt=dαgtdtα=1Γn−α∫0tg(n)(s)t−sα−n+1ds,(n−1<α<n), where g(n)(s) is the nth derivative of g(s) with respect to s. Remark 2.1 If α=n, one has t0CDtαgt=g(n)(t). Lemma 2.1 [30] The Caputo nonlinear system is considered as follows: 0CDtαx(t)=g(x),(α∈(0,1]), with the initial condition x0 . If all eigenvalues of J|x=x∗=∂g∂x|x=x∗ satisfy |arg(λ)|>απ2 , the equilibrium points x∗ are locally asymptotically stable. Lemma 2.2 [31] Suppose X⊂R and the continuous operator T(t):X→X satisfies (1) T(t) is point dissipative in X and compact for t≥0 . (2) there is a finite sequence M={M1,M2,…,Mk} of compact and isolated invariant sets such that (i) Mi∩Mj=0̸ for any i,j=1,2,…,k and i≠j ; (ii) Ω(∂X0)≜∪x∈∂X0ω(x)⊂∪i=1kMi ; (iii) in the case of ∂X0 , no a cycle is formed by any subset of M ; (iv) Ws(Mi)∩X0=0̸ for each i=1,2,…,k . Then T(t) is uniformly persistent in X . 2.2 Graph theory In this paper, a weighted graph ζ=(ϑ,ω,A) will be considered to model the spread of infectious diseases between cities, where ϑ={ϑ1,ϑ2,…,ϑn} denotes the node set and ϑi represents the ith city; ω⊆ϑ×ϑ is the edge set, and if there is individual movement between any two cities, it means that there is a edge between this two cities; matrixes M=[mij]1≤i,j≤n, N=[nij]1≤i,j≤n, P=[pij]1≤i,j≤n and Q=[qij]1≤i,j≤n represent the weighted adjacency matrix of susceptible, exposed, infected and recovered individual, respectively; mij, nij, pij and qij denote the migrate rate of susceptible, exposed, infected and recovered individual from city j to city i with aij≥0 (i≠j) and aii=0 (a=m,n,porq), respectively. Furthermore, based on the directivity of individual migration, the directed graph ζ is studied in this paper. 2.3 System description Starting from 23 January, 2020, the Chinese government has adopted a series of mitigation measures to effectively suppress the spread of COVID-19, such as implementing strict home isolation, restricting various traffic, strengthening nucleic acid testing, establishing shelter hospitals and so on. Meanwhile, other countries around the world have adopted different measures from China, such as social distancing and herd community strategy by British, protecting sensitive compartment from infection by Italy, transferring of critically ill patients with military aircraft by France, etc. Therefore, it is important to establish a generalized model of individual migration to simultaneously quantify the impact of interruption of policies on virus transmission. Moreover, Tang et al. [32] proposed that the exposed individual is infectious of COVID-19. Therefore, a fractional-order SEIHRDP epidemic model with individual migration is considered as follows: (1) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),0CDtαEk=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),0CDtαIk=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),0CDtαHk=δkIk−(λk+κk)Hk,0CDtαRk=λkHk+∑j=1n(qkjRj−qjkRk),0CDtαDk=κkHk,0CDtαPk=ρkSk. with the initial condition (2) Sk(0)=Sk0>0,Ek(0)=Ek0≥0,Ik(0)=Ik0≥0,Hk(0)=Hk0≥0, Rk(0)=Rk0≥0,Dk(0)=Dk0≥0,Pk(0)=Pk0≥0. The specific explanation of system (1) are as follows: • Susceptible Sk: the number of susceptible class within city k at time t. • Exposed Ek: the number of exposed class within city k at time t (neither any clinical symptoms nor high infectivity). • Infectious Ik: the number of infected class within city k at time t (with overt symptoms). • Hospitalized Hk: the number of hospitalized class within city k at time t. • Recovered Rk: the number of recovered class within city k at time t. • Dead Dk: the number of dead class within city k at time t. • Insusceptible Pk: the number of susceptible class who are not exposed to the external community within city k at time t. Meanwhile, the process of disease transmission are as follows: • The susceptible individual Sk contacts with Ek and Ik, and then is infected by β1kSkfk(Ik)+β2kSkgk(Ek), where βik (i=1,2) are the transmission coefficient, fk(Ik) and gk(Ek) are generalized incidence rates. • The parameter Λk is the inflow rate; λk, ϵk, δk and κk represent the recovery, incubation, diagnosis, mortality rate. • The susceptible, exposed, infective and recovered individuals in city j move to city k with probability mkj, nkj, pkj and qkj, respectively. The terms ∑j=1n(mkjSj−mjkSk), ∑j=1n(nkjEj−njkEk), ∑j=1n(pkjIj−pjkIk) and ∑j=1n(qkjRj−qjkRk) represent the movement of Sk, Ek Ik and Rk individual, where ∑j=1nakjWj represents the individuals moving into k city from other cities j (k≠j) and ∑j=1najkWj represents the individuals leaving city k (W=S, E, I, R, respectively, a=m, n, p, q, respectively). • The movement of insuspectible and hospitalized individuals is not considered in this paper. Furthermore, Λk, βik (i=1,2), ρk and ϵk are positive constants; functions δk(t), λk(t), κk(t), mkj(t), nkj(t), pkj(t) and qkj(t) satisfy |δk(t)|≤M1k, |λk(t)|≤M2k, |κk(t)|≤M3k, |mkj(t)|≤M4k, |nkj(t)|≤M5k, |pkj(t)|≤M6k and |qkj(t)|≤M7k for all t≥0 and k,j=1,2,…,n, where M1k, M2k, M3k, M4k, M5k, M6k and M7k are positive constants. The transmission diagram of the generalized SEIHRDP model (1) is shown in Fig. 1. Before presenting the major findings, the following generalized incidence rate hypothesis is put forth: (H):(i)gk(Ek)andfk(Ik)satisfythelocalLipschitzconditionand gk(0)=0,fk(0)=0fork=1,2,…,n;(ii)fk(Ik)isstrictlymonotoneincreasingonIk∈[0,∞)and gk(Ek)isstrictlymonotoneincreasingon Ek∈[0,∞)forallk=1,2,…,n;(iii)fk(Ik)≤akIkforallIk≥0, whereak=fk′(0)forallk=1,2,…,n;(iv)gk(Ek)≤bkEkforallEk≥0, wherebk=gk′(0)forallk=1,2,…,n. Fig. 1 The schematic diagram of SEIHRDP epidemic model with individual migration (i,j=1,2,…,n). Remark 2.2 It should be noted that many current models can be viewed as a special type of system (1) with the hypothesis (H), such as gk(Ek)=bkEk, gk(Ek)=bkEk1+vkEk, fk(Ik)=akIk, fk(Ik)=akIk1+ukIk and others [33] with nonnegative constants ak, bk, uk and vk. Remark 2.3 Compared with [34], the individual movement in this paper can be described as follows: (1) the self-migration of individuals is described in system (1), which is caused by a self-chemotactic-like forcing [35]. However, the cross-infection among cities is considered which is a travel infectious [34]. (2) system (1) describes not only the migration of infected individuals, but also the movement of exposed and recovered individuals. (3) M=[mij]1≤i,j≤n, N=[nij]1≤i,j≤n, P=[pij]1≤i,j≤n and Q=[qij]1≤i,j≤n are not irreducible in this paper, but irreducible in [34]. Then the influence of network structure on disease transmission can be discussed in this paper, such as fully connected network, ring network and centralized network. However, [34] only consider fully connected network. (4) the total population of each city is changed (without considering the decrease in population due to death) in system (1). But in [34], the total population of each city remains constant. 3 System analysis This study explores system (1)’s dynamic analysis. As can be seen, the death class Dk and the insusceptible class Pk have no effect on the susceptible class Sk, exposed class Ek, infected class Ik, hospitalized class Hk, or recovered class Rk of systems (1). Accordingly, the following system is discussed in the next section: (3) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),0CDtαEk=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),0CDtαIk=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),0CDtαHk=δkIk−(λk+κk)Hk,0CDtαRk=λkHk+∑j=1n(qkjRj−qjkRk), with the initial condition (4) Sk(0)=Sk0>0,Ek(0)=Ek0≥0,Ik(0)=Ik0≥0, Hk(0)=Hk0≥0,Rk(0)=Rk0≥0,(k=1,2,…,n). 3.1 Existence and uniqueness of the positive solution The existence, uniqueness, and boundedness of the nonnegative solution for system (3) should be taken into account prior to the numerical process. Therefore, this subsection will be discussed these properties for system (3). Theorem 3.1 For any nonnegative initial condition (4) , there are a unique solution for system (3) and the region Ω={(S1,E1,I1,H1,R1,…,Sn,En,In,Hn,Rn): 0<Si≤Λ¯ρ,0≤Ei≤Λ¯ρ,0≤Ii≤Λ¯ρ, 0≤Hi≤Λ¯ρ,0≤Ri≤Λ¯ρ,i=1,2,…,n} is positively invariant for system (3) , where Λ¯=∑j=1nΛj and ρ=min{ρ1,ρ2,…,ρn} . Proof Let consider the following function: f1k=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),f2k=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),f3k=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),f4k=δkIk−(λk+κk)Hk,f5k=λkHk+∑j=1n(qkjRj−qjkRk). It is obvious that Fk=(f1k,f2k,f3k,f4k,f5k) satisfies the local Lipschitz condition about (Sk,Ek,Ik,Hk,Rk), then system (3) has a unique solution. Next, the nonnegative solution will be analyzed. Consider the following auxiliary system: 0CDtαS_k=−β1kS_kfk(I_k)−β2kS_kgk(E_k)−ρkS_k+∑j=1n(mkjS_j−mjkS_k),0CDtαE_k=β1kS_kfk(I_k)+β2kS_kgk(I_k)−ϵkE_k+∑j=1n(nkjE_j−njkE_k),0CDtαI_k=ϵkE_k−δkI_k+∑j=1n(pkjI_j−pjkI_k),0CDtαH_k=δkI_k−(λk+κk)H_k,0CDtαR_k=λkH_k+∑j=1n(qkjR_j−qjkR_k),Sk_(0)=Ek_(0)=Ik_(0)=Hk_(0)=Rk_(0)=0. Through the comparison theorem, it is not difficult to find that the following auxiliary system has a unique solution (0,0,0,0,0). then the following equation holds: (Sk,Ek,Ik,Hk,Rk)>(0,0,0,0,0). Next, adding all equations gives 0CDtαN≤Λ¯−ρN where N=∑j=1n(Sj+Ej+Ij+Hj+Rj+Dj), Λ¯=∑j=1nΛj and ρ=min{ρ1,ρ2,…,ρn}. Then N(t)≤(N(0)−Λ¯ρ)Eα(−ρtα)+Λ¯ρ. Therefore, the region Ω is positively invariant for system (3).  □ 3.2 Local stability The exploration of the existence and local stability of the disease-free equilibrium point is the focus of this section. Theorem 3.2 There are a unique disease-free equilibrium point E0=(S1∗,0,0,0,0,…,Sn∗,0,0,0,0) for system (3) where S∗=(S1∗,…,Sn∗) , S∗=A−1Λ , Λ=(Λ1,…,Λn) and A=ρ1+∑j≠1nmj1−m12⋯−m1n−m21ρ2+∑j≠2nmj2⋯−m2n⋮⋮⋮⋮−mn1−mn2⋯ρ2+∑j≠nnmjn. Proof Obviously, E0 satisfies the following equation: Λk−ρkSk∗+∑j=1n(mkjSj∗−mjkSk∗)=0, then the above equation can be written as the following matrix form: AS∗=Λ. It can be found that the matrix A is strictly diagonally dominant, and then it follows from [36] that one has A−1≥0. So according to [37], there exists a unique solution S∗=A−1Λ. Therefore, there exists a unique disease-free equilibrium point E0 of system (3). □ The predicted number of secondary cases that a typical infectious individual should create in a community that is totally susceptible is known as the basic reproduction number R0. According to Watmough et al. [38], it can be determined that an infectious disease can commonly infect the community if one diseased individual can typically infect more than one susceptible individual when R0≥1. On the other hand, if R0<1, each infected individual produces less than one new infection, and the infectious diseases can not grow. Thus, it is very important to describe the relationship between the basic reproduction number and the spread of infectious diseases. Here, the basic reproduction number R0 is stated as follows. Theorem 3.3 Under hypothesis H, the basic reproduction number R0 is R0=ρ(F11V11−1−F12V11−1V21V22−1), where matrixes F11=diag(β21b1S1,…,β2nbnSn) , F12=diag(β11a1S1,…,β1nanSn) , V21=diag(−ϵ1,…,−ϵn) , V11=ϵ1+∑j≠1nn1j−n12⋯−n1n−n21ϵ2+∑j≠2nn2j⋯−n2n⋮⋮⋮⋮−nn1−nn2⋯ϵn+∑j≠nnnnj, and V22=δ1+∑j≠1np1j−p12⋯−p1n−p21δ2+∑j≠2np2j⋯−p2n⋮⋮⋮⋮−pn1−pn2⋯δn+∑j≠nnpnj. Proof Let consider the following matrixes: F0=β11S1f1(I1)+β21S1gk(E1)⋮β1nSnfn(In)+β2nSngn(En)0⋮00⋮0andV0=ϵ1E1−∑j=1n(n1jEj−nj1E1)⋮ϵnEn−∑j=nn(nnjEj−njnEn)−ϵ1E1+δ1I1−∑j=1n(p1jIj−pj1I1)⋮−ϵnEn+δnIn−∑j=nn(pnjIj−pjnIn)−δ1I1+(λ1+κ1)H1⋮−δ1In+(λn+κn)Hn. Let u=(E1,…,En,I1,…,In,H1,…,Hn), then take the derivative of F0 and V0 for u at the disease-free equilibrium point E0, respectively, we can see as follows: F=F11F120000000andV=V1100V21V2200V32V33, where F11=diag(β21b1S1∗,…,β2nbnSn∗), F12=diag(β11a1S1∗,…,β1nanSn∗), V21=diag(−ϵ1,…,−ϵn), V33=diag((λ1+κ1),…,(λn+κn)), V32=diag (−δ1,…,−δn), V11=ϵ1+∑j≠1nn1j−n12⋯−n1n−n21ϵ2+∑j≠2nn2j⋯−n2n⋮⋮⋮⋮−nn1−nn2⋯ϵn+∑j≠nnnnj, and V22=δ1+∑j≠1np1j−p12⋯−p1n−p21δ2+∑j≠2np2j⋯−p2n⋮⋮⋮⋮−pn1−pn2⋯δn+∑j≠nnpnj. Then according to [39], the basic reproduction number is as follows: R0=ρ(FV−1)=ρ(F11V11−1−F12V11−1V21V22−1), where ρ(F11V11−1−F12V11−1V21V22−1) is the spectral radius of the matrix (F11V11−1−F12V11−1V21V22−1). □ Remark 3.1 According to [40], the epidemic size ςk=Sk(0)−Sk∗ of city k is defined as the number of individuals affected by the infectious disease, where Sk(0) is initial condition and Sk∗ is the disease-free equilibrium point of susceptible individuals within city k. Remark 3.2 When individual migration is not taken into consideration, it can be calculated from [10] that the basic reproduction number R0uk of city k is R0uk=Sk∗(β2kbkϵk+β1kakδk). Remark 3.3 When individual migration is taken into consideration, R0k of city k is R0k=Sk∗ϵk+∑j⁄=knnkj(β2kbk+β1kakϵkδk+∑j≠knpkj). Remark 3.4 It is easy to see that R0k are not dependent on λk, κk and mkj. Like [41], the other Λk, β1k, β2k, ρk, ϵk, nkj, pkj and δk are calculated as follows: AΛk=ΛkR0k∂R0k∂Λk=1,Aρk=ρkR0k∂R0k∂ρk=−1, Aβ1k=β1kR0k∂R0k∂β1k=β1kakϵkδk+∑j≠knpkjβ2kbk+β1kakϵkδk+∑j≠knpkj,Aβ2k=β2kR0k∂R0k∂β2k=β2kbkβ2kbk+β1kakϵkδk+∑j≠knpkj, Aδk=δkR0k∂R0k∂δk=−1(δk+∑j≠knpkj2)(β2kbk+β1kakϵkδk+∑j≠knpkj),Aϵk=ϵkR0k∂R0k∂ϵk=−1ϵk+∑j≠knnkj(β2kbk+ϵk+∑j≠knnkjδk+∑j≠knpkj), Ankj=nkjR0k∂R0k∂nkj=−1,Apkj=pkjR0k∂R0k∂pkj=−1(δk+∑j≠knpkj2)(β2kbk+β1kakϵkδk+∑j≠knpkj), where AΛk, Aρk, Aβ1k, Aβ2k, Aδk, Aϵk, Ankj and Apkj represent the normalized sensitivity on Λk, ρk, β1k, β2k, δk, ϵk, nkj and pkj, respectively. Through the above calculation found that the increase on Λk, β1k and β2k leads to the increase on R0k, but the increase on ρk, δk, ϵk, nkj and pkj leads to the decrease on R0k. In addition, the movement of susceptible individuals has no impact of R0k, but the movement of exposed and infected individuals is negatively correlated with R0k, and the movement of exposed individuals is more likely to influence the spread of the infectious disease with |Ankj|>|Apkj|. Theorem 3.4 Under hypothesis H, system (3) is locally asymptotically stable at the disease-free equilibrium point E0 if |arg(sF−V)|>απ2 . Proof The following Jacobian matrix at the disease-free equilibrium point E0 is considered: JE0=J11∗00F−V00∗J33, where matrixs J11=−ρ1−∑j=1nmj1m12⋯m1nm21−ρ2−∑j=1nmj2⋯m2n⋮⋮⋮⋮mn1mn2⋯−ρn−∑j=1nmjn, J33=−∑j=1nqj1q12⋯q1nq21−∑j=1nqj2⋯q2n⋮⋮⋮⋮qn1qn2⋯−∑j=1nqjn, F and V see Theorem 3.3. Then if all eigenvalues of the Jacobian matrix JE0 satisfy |arg(si)|>απ2, E0 is locally asymptotically stable and unstable if for some eigenvalues si, |arg(si)|≤απ2. Obviously, J11 and J33 are a nonsingular M-matrix, so J11 and J33 has all eigenvalues with negative real parts according to [42]. Consequently the local stability of E0 depends only on eigenvalues of F−V. Thus, if all eigenvalues of F−V satisfy |arg(sF−V)|>απ2, system (3) is locally asymptotically stable. □ Remark 3.5 If all the eigenvalues of F−V are negative, that is |arg(sF−V)|=π>απ2, system (3) is locally asymptotically stable. Meanwhile, it is obvious that |arg(sF−V)|=π⇔sF−V<0⇔ρ(FV−1)<1⇔R0<1. It can be yielded that if R0<1, the disease-free equilibrium point E0 is locally asymptotically stable of system (3) . 3.3 Global asymptotic stability of the disease-free equilibrium In this subsection, the global asymptotic stability of the disease-free equilibrium point E0 is discussed firstly. Furthermore, the uniform persistence of system (3) is also considered. Theorem 3.5 Under hypothesis (H) and |arg(sF−V)|>απ2 , the disease-free equilibrium point E0 is globally asymptotically stable of system (3) . Proof We use a method similar to the one used in [43]. Firstly, the boundedness of the susceptible class will be analyzed. According to Theorem 3.1 and hypothesis (H), we know Sk, Ek and Ik (k=1,2,…,n) are nonnegative, thus one has (5) 0CDtαSk=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk +∑j=1n(mkjSj−mjkSk)≤Λk−ρkSk+∑j=1n(mkjSj−mjkSk). Let S=(S1,…,Sn), S∗=(S1∗,…,Sn∗), Λ=(Λ1,…,Λn) and A=ρ1+∑j≠1nmj1−m12⋯−m1n−m21ρ2+∑j≠2nmj2⋯−m2n⋮⋮⋮⋮−mn1−mn2⋯ρ2+∑j≠nnmjn, then Eq. (5) can be written in the following matrix: 0CDtαS≤Λ−AS=AS∗−AS, so it is easy to see that the conclusion holds as follows: S(t)≤(S0−S∗)Eα(−Atα)+S∗. Obviously, one has Sk≤Sk∗. Next, the global stability of Ek, Ik and Hk will be discussed. Based on hypothesis (H), one has fk(Ik)≤akIk and gk(Ek)≤bkEk. Then the following auxiliary system is considered: (6) 0CDtαEk¯=β1kSk∗akIk¯+β2kSk∗bkEk¯−ϵkEk¯+∑j=1n(nkjEj¯−njkEk¯),0CDtαIk¯=ϵkEk¯−δkIk¯+∑j=1n(pkjI¯j−pjkIk¯),0CDtαHk¯=δkIk¯−(λk+κk)Hk¯. It is easy to see that (7) 0CDtαW=(F−V)W, where W=(E¯,I¯,H¯), E¯=(E1¯,…,En¯), I¯=(I1¯,…,In¯), H¯=(H¯1,…,H¯n), F and V see Theorem 3.3. Thus, if |arg(sF−V)|>απ2, the above linear system (7) is locally asymptotically stable as well as globally asymptotically stable, that is limt→∞E¯k=limt→∞I¯k=limt→∞H¯k=0. According to the comparison theory and the nonnegative solution of Ek, Ik and Hk, one has limt→∞Ek=limt→∞Ik=limt→∞Hk=0. Based on the above analysis, when t→∞, one has 0CDtαS=AS∗−AS, then one has S(t)→S∗(t→∞). So E0 is globally asymptotically stable if |arg(sF−V)|>απ2. □ Remark 3.6 Similar to Theorem 3.3, it can be concluded that if R0<1, system (3) is globally asymptotically stable at the disease-free equilibrium point E0. Furthermore, the uniform persistence for system (3) is discussed in the following theorem. Theorem 3.6 Under hypothesis (H) and R0>1 , system (3) is uniformly persist, implying there exists a positive constant δ such that lim inft→+∞Sk≥δ,lim inft→+∞Ek≥δ,lim inft→+∞Ik≥δ,lim inft→+∞Hk≥δ,lim inft→+∞Rk≥δ,1≤k≤n. Proof Let consider the following space: X=X1×X2×⋯×Xn,X0=X10×X20×⋯×Xn0,∂X=∂X1×∂X2×⋯×∂Xn, where X0 represents the interior of X, ∂X denotes the boundary of X and Xk={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek≥0,Ik≥0,Hk≥0,Rk≥0}, Xk0={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek>0,Ik>0,Hk>0,Rk>0}, ∂Xk={(Sk,Ek,Ik,Hk,Rk):Sk>0,Ek=0,Ik=0,Hk=0,Rk=0}. Meanwhile, let W(t)=(S1,E1,I1,H1,R1,…,Sn,En,In,Hn,Rn) be the solution of system (3) with initial value W(0)=W0∈X, then W(t)∈X according to Theorem 3.1. For any t≥0, a continuous map F(t):X→X is defined as follows: F(t)W0=W(t). In the following, the uniformly persistent of the map F will be analyzed based on Lemma 2.3. When t=0, one has F(0)W0=W(0), this is F(0)=I where I is the identity matrix. Meanwhile, it can be deduced that the following equation holds: F(t+s)W0=W(t+s)=F(t)W(s)=F(t)F(s)W0, implying F(0)=I. Additionly, it is easy to see that F(t) is C0-semigroup on X, point dissipative and compact in X. Furthermore, consider the following system: 0CDtαSk=Λk−ρkSk+∑j=1n(mkjSj−mjkSk). According to Theorem 3.3, Sk∗ is asymptotically stable, which finds that E0 in ∂X is a global attractor of F(t). Let M={M1}, where M1={E0}. Because of ak=fk′(0) and bkgk′(0), for all ϵ, there exists ϵ¯ that f(ϵ)>(ak−ϵ)ϵ¯andg(ϵ)>(bk−ϵ)ϵ¯. Let the stable set Ws(E0) of a compact invariant set E0 defined by Ws(E0)={Y0∈X:ω(Y0)≠0̸,ω(Y0)∈E0}, where ω(Y0) is ω-limit set through Y0. System (3) has a solution (Sk,Ek,Ik,Hk,Rk) when Ws(E0)∩X0≠0̸, implying Sk→0, Ek→0, Ik→0, Hk→0, Rk→0 (k=1,2,…,n) as t→∞. So there exists a constant τ>0 such that Sk>Sk∗−ϵ, Ek>ϵ, Ik>ϵ, Hk>ϵ and Rk>ϵ for t≥τ. Then according to the monotonicity of fk(Ik) and gk(Ek), one has f(Ik)>f(ϵ)>(ak−ϵ)ϵ¯andg(Ek)>g(ϵ)>(bk−ϵ)ϵ¯. So the following auxiliary system is considered: (8) 0CDtαEk_=β1kSk∗(ak−ϵ)ϵ¯Ik_+β2kSk∗(bk−ϵ)ϵ¯Ek_−ϵkEk_+∑j=1n(nkjEj_−njkEk_),0CDtαIk_=ϵkEk_−δkIk_+∑j=1n(pkjIj_−pjkIk_),0CDtαHk_=δkIk_−(λk+κk)Hk_. It is easy to see from system (8) that (9) 0CDtαW=(F−V)(ϵ¯,ϵ)W, where W=(E_,I_,H_), E_=(E_1,…,E_n), I_=(I_1,…,I_n) and H_=(H_1,…,H_n). Consider the basic reproduction number R0>1, then one has ρ(F11V11−1−F12V11−1V21V22−1)(ϵ¯,ϵ)>1, which results in a contradiction with Ek(t)→0 (t→∞). Hence one has Ws(E0)∩X0=0̸, implying it is uniformly persistent at the operator T(t), so system (3) is uniformly persistent if R0>1.  □ The existence of a positive equilibrium point is implied by the system (3)’s ultimate boundedenss and uniform persistence. As a result, we can derive the following theorem. Theorem 3.7 Under hypothesis (H) and R0>1 , there is at least one endemic equilibrium E∗=(S1∗,E1∗,I1∗,H1∗,R1∗,…,Sn∗,En∗,In∗,Hn∗,Rn∗) of system (3) satisfying Λk−β1kSk∗fk(Ik∗)−β2kSk∗gk(Ek∗)−ρkSk∗+∑j=1n(mkjSj∗−mjkSk∗)=0,β1kSk∗fk(Ik∗)+β2kSk∗gk(Ek∗)−ϵkEk∗+∑j=1n(nkjEj∗−njkEk∗)=0,ϵkEk∗−δkIk∗+∑j=1n(pkjIj∗−pjkIk∗)=0,δkIk∗−(λk−κk)Hk∗=0,λkHk∗+∑j=1n(qkjRj∗−qjkRk∗)=0. 4 Numerical simulation From the previous description, it is clear that E0 is globally asymptotically stable when R0<1 and conversely, system (3) is persistent, which can offer theoretical evidence for further COVID-19 prediction and control. Meanwhile, in order to analyze COVID-19 in different cities, this section is divided into two parts: no restrictions on individual migration and restrictions on individual migration. Furthermore, consider the corresponding integer-order model as follows: (10) dSkdt=Λk−β1kSkfk(Ik)−β2kSkgk(Ek)−ρkSk+∑j=1n(mkjSj−mjkSk),dEkdt=β1kSkfk(Ik)+β2kSkgk(Ek)−ϵkEk+∑j=1n(nkjEj−njkEk),dIkdt=ϵkEk−δkIk+∑j=1n(pkjIj−pjkIk),dHkdt=δkIk−(λk+κk)Hk,dRkdt=λkHk+∑j=1n(qkjRj−qjkRk). 4.1 Data source The Johns Hopkins University Center for System Science and Engineering provided the real data for this study [1]. Data on accumulated and confirmed cases, recovered cases, and COVID-19 death cases were shared by the Johns Hopkins University on January 23, 2020. Assuming that the confirmed individuals mut be hospitalized, one has Hospitalized=Confirmed−Recovered−Death. Hence, we can get the real data of H(t), D(t) and R(t) for different city from 23 January to 17 July, 2020. 4.2 The generalized incidence rate As we know, Korobeinikov et al. [8] indicated that the stability of the endemic equilibrium point for infectious diseases is closely related to the concave of the incidence rate with respect to the infected individuals. Therefore, it is of practical significance to understand the role of different incidence rates in COVID-19. In this section, according to hypothesis (H), the bilinear incidence rate and the saturation incidence rate are discussed as follows: fk(Ik)=Ik,gk(Ek)=Ek. fk(Ik)=Ik1+ukIk,gk(Ek)=Ek1+vkEk. Meanwhile, as the public learns about COVID-19, the recovered rate and the disease-related mortality are time-varying rather than constant. Similar to [11], the best recovered rate λk and the best disease-related mortality κk are selected from the following equation: (11) κk=p1ep2(t−p3)+e−p2(t−p3),p1e(p2(t−p3))2,p1+e(p2(t+p3)),andλk=q11+e−q2(t−q3),q1+e−q2(t+q3), where qi and qi (i=1,2,3) are parameters for κk and λk, respectively. According to the real data reported by [2], the spread of COVID-19 in India and Brazil began on 30 January and 26 February, 2020, as the beginning of the outbreak of India and Brazil in this paper, respectively. According to Matlab function lsqcurvefit [11], the parameter identification results with system (3) and system (10) are depicted in Table 1, Table 2, respectively. Meanwhile, based on Table 1, Table 2, the five days forecast of India and Brazil are shown in Tables 3, 4, Figs. 2, 3, 4, 5, which the solid lines represent simulation results and circles represent real data. The results in Table 1, Table 2 show that the fractional-order system (3) can accurately forecast the real data in the upcoming week, with the real data of currently confirmed cases falling between 95% and 105% of the projected values. Table 1 Parameter identification of India. India Integer (Bilinear) Fractional (Bilinear) Integer (Saturation) Fractional (Saturation) Λ 0.3 0.6245 0.465 0.3 β1 0.3869 1.206 1.156 2.358 β2 0.5133 0.3079 0.3414 0.8 ϵ 0.0023 0.0555 0.0014 0.0692 ρ 0.0264 0.03 0.0188 0.0094 λ p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) κ b1e−q2(t−q3)2 q1e−q2(t−q3)2 q1e−q2(t−q3)2 q1e−q2(t−q3)2 Table 2 Parameter identification of Brazil. Brazil Integer (Bilinear) Fractional (Bilinear) Integer (Saturation) Fractional (Saturation) Λ 0.3 0.5 0.8802 0.5189 β1 1.839 1.082 0.2378 4.804 β2 0.3733 0.928 0.4397 0.3 ϵ 0.0303 0.7905 0.1076 0.9609 ρ 0.0211 0.0214 0.0216 0.0431 δ 0.99 0.2434 0.3975 5.609×10−5 λ p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) p11+e−p2(t−p3) κ q1eq2(t−q3) q1eq2(t−q3) q1eq2(t−q3) q1eq2(t−q3) Table 3 Estimate the number of confirmed cases within five days in India (×105). India 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 3.735 3.906 4.027 4.113 4.263 Integer (bilinear incidence rate) 3.454 3.499 3.52 3.556 3.583 Fractional (bilinear incidence rate) 3.781 3.869 3.936 4.002 4.079 Integer (saturation incidence rate) 3.426 3.476 3.515 3.553 3.578 Fractional (saturation incidence rate) 3.645 3.696 3.741 3.792 3.815 Table 4 Estimate the number of confirmed cases within five days in Brazil (×105). Brazil 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 5.487 5.598 5.242 5.228 5.528 Integer (bilinear incidence rate) 5.864 5.847 5.828 5.806 5.781 Fractional (bilinear incidence rate) 5.502 5.49 5.475 5.458 5.438 Integer (saturation incidence rate) 5.897 5.882 5.864 5.853 5.821 Fractional (saturation incidence rate) 5.857 5.844 5.819 5.793 5.769 Fig. 2 The number of cases in India (Integer-order with the bilinear incidence rate (left), Fractional-order with the bilinear incidence rate (right)). Fig. 3 The number of cases in India (Integer-order with the saturation incidence rate (left), Fractional-order with the saturation incidence rate (right)). Fig. 4 The number of cases in Brazil (Integer-order with the bilinear incidence rate (left), Fractional-order with the bilinear incidence rate (right)). Fig. 5 The number of cases in Brazil (Integer-order with the saturation incidence rate (left), Fractional-order with the saturation incidence rate (right)). 4.3 Restrict individual migration When individual movement is not considered, the parameters satisfy mkj=nkj=pkj=qkj=0. Meanwhile, according to Section 4.2, the bilinear incidence rate is considered in this section. Then based on system (1), the following auxiliary system is considered: (12) 0CDtαS=Λ−β1SIβ2SE−ρS,0CDtαE=β1SI+β2SEϵE,0CDtαI=ϵE−δI,0CDtαH=δI−(λ+κ)H,0CDtαR=λH,0CDtαD=κH. 4.3.1 Sensitivity analysis of parameters in R0uk When individual movement is not taken into consideration in this section, Partial Rank Correlation Coefficients (PRCC) value and Latin hypercube sampling (LHS) [44], which are one of the Monte Carlo (MC) sampling methods established by Mckay in 1979 [45], can be used to account for the sensitivity of the parameter to the basic reproduction number. LHS has the advantage of using fewer iterations than other random sampling techniques and avoiding the clustering phenomenon of sampling [45]. In order to determine which aspects of a certain intervention have the greatest impact on how quickly a new infection spreads, it can be seen from Remark 3.2 that the parameters of system (12) all affect the basic reproduction number to varying degrees, thereby affecting the spread of the infectious disease. We perform LHS on the parameters that appear in R0uk. PRCC are calculated, and a total of 1000 simulations per LHS run are carried out. A uniform distribution is chose as the prior distribution when performing parameter sampling. The parameters Λ, ρ, ϵ, β1, β2 and δ of system (12) are set as input variables, and the basic reproduction number R0uk as the output. The specific process is as follows: (1) There are six parameters that affect the change of R0uk, which are Λ, ρ, ϵ, β1, β2 and δ. Through LSH, [0,1] is divided into 1000 simulations, and 6 × 1000 parameters are generated through random selection on each interval by a uniform distribution. (2) Calculate the basic reproduction number R0uk for each parameter. (3) PRCC is calculated by Matlab function partialcorr. (4) The PRCC’s influence on the basic reproduction number R0uk can increase with increasing PRCC absolute value. However, it is believed that the parameter is not significant if the p value is greater than 0.05. Table 5 lists the PRCC values of the six parameters associated with R0uk and Fig. 6 shows the histogram of PRCC value. From Table 5 and Fig. 6, the following conclusion holds: (1) the parameters Λ, β1 and β2 have a positive influence on R0uk, but ρ, ϵ and δ have a negative influence, which is consistent with Remark 3.4; (2) the positive impact of birth rate Λ is the most obvious with PRCC(Λ)=0.5868; (3) the positive impact of the transmission rate β2 for the exposed population is more obvious than that of the infected population with PRCC(β2)>PRCC(β1). That is, the greater the transmission coefficient of the exposed population, the greater the value of the basic reproduction number R0uk, and then the greater the number of people infected with COVID-19. Therefore, it is more critical to limit exposed individual. However, because exposed individual doed not show any symptoms, identifying them is very difficult, which is a key reason for the spread of COVID-19; (4) the diagnosis rate δ has more greater negative impact on R0uk. That is to say, enhancing nucleic acid detection can effectively reduce R0ui, thereby reducing the number of infected people; (5) from the p-value, it can be found that the p-values of all parameters are less than 0.05, so they all have a significant impact on the basic reproduction number R0uk. Therefore, based on the above analysis, it can be obtained that controlling the influx of foreign population and enhancing nucleic acid detection are the most effective measures to control COVID-19. Meanwhile, home isolation can also control COVID-19. Therefore, this evidence confirms the effectiveness of Chinese government’s interruption policies, such as home isolation, prohibition of the inflow of foreign population, and enhancing nucleic acid detection, which may provide a good reference for the other countries. Table 5 The PRCC values and p-value of the parameters with respect to R0uk. Input PRCC values p-value Λ 0.5868 0 ρ −0.5363 0 ϵ −0.1368 0 β2 0.1035 3.5×10−6 β1 0.0847 1.448×10−4 δ −0.4362 0 Fig. 6 The sensitivity analysis of R0uk. 4.3.2 China’s second outbreak From the analysis in Section 4.3.1, it can be found that enhancing the diagnosis rate and controlling the inflow of foreign population can effectively control the spread of the epidemic. For China, individual migration has been strictly restricted at the beginning of COVID-19. Therefore, the impact of enhanced diagnosis rate will be only considered in this section. Due to the increase in public awareness and the development of detection technology, the time from onset to diagnosis is gradually shortened. Additionally, despite the use of the nucleic acid test method, the number of confirmed cases climbed significantly and peaked in early February 2020 as a result of the use of the CT diagnosis method. As a result, it is assumed that starting on 12 February, 2020, China’s diagnosis rate can reach and remain at its highest level. However, the third COVID-19 wave has been occurring in Beijing since the end of June 2020. Beijing has said that starting on 17 June, 2020, nucleic acid could be more readily detected. As a result, a new distribution, rather than the max level dated June 17, now governs the diagnostic rate. Similar to [46], the following piecewise function are described the diagnosed period of two and three peaks: (13) 1δk=(1δ0−1δe)e−w1t+1δe,t<t1,1δe,t≥t1,and 1δk=(1δ0−1δe)e−w1t+1δe,t<t1,1δe,t1≤t≤t2,(1δe−1δf)e−w2(t−t2)+1δf,t>t2, where δ0, δe (δe>δ0), w1, w2 and δf are similar to [46], t1 is 13 February, 2020, t2 is 17 June, 2020. Meanwhile, similar to [11], the best recovered rate λk and the best disease-related mortality κk are selected from Eq. (11). Then system (12) and system (10) are solved by predictor–correctors scheme and least squares method [11] by the real data from 23 January to 17 July, which 17 March, 2020 is considered as the beginning of the emergency in Heilongjiang, Shanghai and Guangdong, and 17 June, 2020 are considered as the beginning of the emergency in Beijing, respectively. From Fig. 7, Fig. 8, the fractional-order system (12) is found to fit the real data more accurately than the integer-order system (10) does. and COVID-19 in Beijing, Shanghai reaches its highest peak in a short time but there may be fourth wave peak, however, Heilongjiang and Guangdong are only two peaks and the third wave of epidemic peaks will not occur in a short time (current policies remain unchanged). Therefore, under the condition of restricting the migration of individuals, the fractional system (12) can better simulate the multi-peak problem of COVID-19, and the strengthening of nucleic acid detection can predict the new wave in advance, which provides a theoretical basis for the control of the epidemic. Fig. 7 The number of cases in Beijing and Shanghai. Fig. 8 The number of cases in Guangdong and Heilongjiang. 4.4 Individual migration As of 17 July, 2020, the United States has a total of 3,647,715 confirmed cases, 139266 deaths and 1,107,204 recovery cases. It is urgent to formulate reasonable and effective mitigation measures. Thus in this section, based on the sensitivity analysis of parameter to R0k, the effect mitigation measures are provided to control the development of COVID-19 in US. 4.4.1 Sensitivity analysis of parameters in R0k Similar to Section 4.3.1, consider two cities to examine the sensitivity of parameters to R0k (k=1,2). Then when n=2, system (3) can be simplified as follows: (14) 0CDtαS1=Λ1−β11S1I1−β21S1E1−ρ1S1+(m12S2−m21S1),0CDtαE1=β11S1I1+β21S1E1−ϵ1E1+(n12E2−n21E2),0CDtαI1=ϵ1E1−δ1I1+(p12I2−p21I1),0CDtαH1=δ1I1−(λ1+κ1)H1,0CDtαR1=λ1H1+(q12R2−q21R1),0CDtαS2=Λ2−β12S2I2−β22S2E2−ρ2S2+(m21S1−m12S2),0CDtαE2=β12S1I2+β22S2E2−ϵ1E2+(n21E1−n12E1),0CDtαI2=ϵ2E2−δ2I2+(p21I1−p12I2),0CDtαH2=δ2I2−(λ2+κ2)H2,0CDtαR2=λ2H2+(q21R1−q12R2). It can be found from Remark 3.4 that there exists 16 parameter of the basic reproduction number R0k (k=1,2), and then the 16 parameters are set as input variables, and R0k as the output. Similar to Section 4.3.1, Table 6 lists the PRCC values and Fig. 9 shows the histogram of PRCC value. According to Table 6 and Fig. 9, it can be found that the following conclusion holds: (1) the movement of susceptible individuals mkj (k,j=1,2) does not affect R0k; (2) the sensitivity of the parameter to R0k (k=1,2) is same as that of Section 4.2.1, except for n12, n21, p12 and p21; (3) considering the basic reproduction number R01 of city 1, the p-value of p21 is large than 0.05, which means that infected individuals migrating from city 1 have a significant impact on COVID-19 in city 1. But exposed and infected individuals migrating to city 1 have an impact on the spread of COVID-19 in city 1, and the impact of the inflow of exposed individuals is more significant because of |PRCC(n12)|>|PRCC(p12)|; (4) contrary to the situation in city 1, the p-value of and p12 is large than 0.05, which means that infected individuals migrating from city 2 have a significant impact on the spread of disease in city 2. But exposed and infected individuals migrating from city 2 have an impact on the spread of COVID-19 in city 2, and the impact of the inflow of exposed individuals is more significant because of |PRCC(n21)|>|PRCC(p21)|. Therefore, in order to alleviate the situation in severe areas of COVID-19, migration of exposed individuals must be strictly controlled. Table 6 The PRCC values and p-value of the parameters with respect to R01 (left) and R02 (right). Input PRCC values p-value Λ1 0.6513 0 ρ1 −0.5294 0 ϵ1 −0.0348 0.1194 β21 0.1983 0 β11 0.0564 0.0116 δ1 −0.1427 0 n12 −0.3584 0 n21 −0.0286 0.2005 p12 −0.1732 0 p21 −0.0062 0.7806 Input PRCC values p-value Λ2 0.6496 0 ρ2 −0.5223 0 ϵ2 −0.0451 0.0435 β22 0.1896 0 β12 0.052 0.0153 δ2 −0.1691 0 n12 −0.0217 0.3318 n21 −0.2732 0 p12 −0.0048 0.8316 p21 −0.1355 0 Fig. 9 The sensitivity analysis of R01 (left) and R02 (right). 4.4.2 US outbreak In this subsection, the overall spread of COVID-19 in the US is considered first. Then system (10) and system (12) are solved by least squares method [11]. However, beginning 17 May, 2020, the number of confirmed individuals in the US had significantly increased. Emergency situations may have changed government regulations and people’s attitudes, which led to an increase in the number of sick people. Therefore, it is assumed that the emergency starts on 17 May, and the outbreak’s spread in the US is then examined in two stages as follows: (1) 23 January-17 May, 2020; (2) 17 May-17 July, 2020. Therefore, parameter identification is provided in Table 7 based on actual data from 23 January to 17 July 2020. From Fig. 10 and Table 8, it is clear that the fractional-order system (12) is capable of accurately forecasting the confirmed case for the upcoming week. In the meantime, Table 8 shows that, regardless of whether in the first stage or the second stage, the parameter findings of the fractional-order system and integer-order fitting are totally different. Based on Remark 3.2, R0uk=49.84 is very high. From the analysis in Section 4.2.1, it can be found that enhancing the diagnosis rate, reducing contact with infected people and controlling the inflow of foreign population can effectively control the spread of COVID-19. However, the United States is not currently doing anything to limit the influx of foreign population, so it is only considering enhancing nucleic acid testing and reducing contact with infected people to control COVID-19. Like [46], the diagnosed period 1δk of US are as follows: (15) 1δ(t)=1δet≤t3,(1δ0−1δe)e−w(t−t3)+1δet>t3. The meaning of each symbol is similar to that in Section 4.2.3. t3 is 17 July, 2020, which mean increasing the diagnosis rate δ(t) from 17 July, 2020. At the same time, the contact rate βi (i=1,2) is limited by the number of hospitalizations like [46] as follows: (16) βi(t)=βi,logH(t)≤1,βilogH(t),logH(t)>1. It can be seen from Fig. 10 that increasing the diagnosis rate δ(t) and controlling the infection rate βi (i=1,2) can effectively contain COVID-19. Therefore, enhanced nucleic acid testing and limited contact with infected individuals are important to control COVID-19.Table 7 Parameter identification of US. US Integer (first stage) Fractional (first stage) Integer (second stage) Fractional (second stage) Λ 0.3 0.0935 0.3 0.4593 β1 1.056 1.217 4.999 2.641 β2 0.2969 0.4882 1.648×10−7 3.724×10−7 ϵ 0.1791 0.2876 0.0087 0.0077 ρ 0.0281 0.0497 0.0358 0.0272 δ 0.1243 0.2434 0.3975 0.205 λ p1+e−p2(t+p3) p1+e−p2(t+p3) p1+e−p2(t+p3) p1+e−p2(t+p3) κ q1eq2(t−q3)+e−q2(t−q3) q1eq2(t−q3)+e−q2(t−q3) q1+e−q2(t+q3) q1+e−q2(t+q3) Table 8 Estimate the number of confirmed cases within five days in US (×106). Date Real data Fractional Integer 18 July 2.449 2.503 2.397 19 July 2.502 2.541 2.425 20 July 2.534 2.578 2.454 21 July 2.575 2.616 2.482 22 July 2.617 2.655 2.511 Fig. 10 The number of cases in US (without control (left), with control (right)). 4.4.3 US with individual migration This subsection considers the impact of individual migration on COVID-19. We need to preprocess the data to remove data that are less than 0.5% of the current maximum number of confirmed cases. Therefore, the real data after 3 April are selected to identify the parameters of system (14). Similar to the analysis of Section 4.4.2, we consider 17 May, 2020 as the beginning of the emergency, and the COVID-19 spread in New York and Los Angeles into two phases: (1) 3 April-17 May, 2020; (2) 17 May-17 July, 2020. Meanwhile, the recovered data of New York and Los Angeles have not been collected by [1], and then we take hospitalized+recovered individuals as a whole to conduct parameter identification and short-term prediction according to [11]. It can be found from Table 9, Table 10 and Fig. 11 that system (14) can better predict COVID-19. Meanwhile, it can be seen from Fig. 11 that the COVID-19 in New York has been peaked but not in Los Angles. From the analysis of Section 4.4.1, we know that controlling the infection rate, improving the diagnosis rate and controlling the movement of exposed individuals have a significant effect on the control of COVID-19 in US. Therefore, similar to Section 4.4.2, the diagnosis rate δk and the migration rate nkj are utilized as follows: (17) 1δk(t)=1δe,t≤t3,(1δ0−1δe)e−w(t−t3)+1δe,t>t3,and nkj=nkj,log(Hk)<1,nkjlog(Hk),log(Hk)≥1. Meanwhile, the infection rate controlled by the number of hospitalizations is Eq. (16). From Fig. 12, we can seen the fractional-order system (14) with control (Eqs. (16), (17)) in Los Angles can be control quickly but not in New York, which is still an open question and will be discussed later.Table 9 Estimate the number of confirmed cases within five days in New York (×105). New York 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 1.977 1.98 1.983 1.987 1.99 Integer 1.978 1.979 1.98 1.981 1.981 Fractional 1.985 1.986 1.987 1.988 1.989 Table 10 Estimate the number of confirmed cases within five days in Los Angles (×105). Los Angles 18 Jul 19 Jul 20 Jul 21 Jul 22 Jul Real data 1.491 1.518 1.549 1.579 1.609 Integer 1.414 1.444 1.464 1.482 1.514 Fractional 1.494 1.517 1.547 1.579 1.608 Fig. 11 The number of cases in New York and Los Angeles with individual movement. Fig. 12 The number of cases in New York and Los Angeles with control Eqs. (16), (17). 5 Conclusion Based on individual migration, a fractional-order SEIHRDP model is proposed with the generalized incidence rate. Meanwhile, some results and effective mitigation measures is suggested to control COVID-19 as follows: (1) The local and global asymptotic stability of the disease-free and endemic equilibrium points are investigated based on the basic reproduction number R0. (2) Based on the real data, it is found that the bilinear incidence rate has a better description of COVID-19 transmission than the saturation incidence rate. Therefore, the bilinear incidence rate is applied in modelling COVID-19. Meanwhile, this is the first time that looked at the impact of the incidence rate in the spread of COVID-19 using real data. (3) By applying the value of PRCC, the sensitivity of the parameters to the basic reproduction number R0k and R0uk are obtained, which is consistent with Remark 3.4. Through the PRCC value, the diagnosis rate, the migration rate and the movement of the infected population are most sensitive to control COVID-19. (4) Multiple peaks have been analyzed for COVID-19 and using four cities in China to show that the fractional-order system (1) works well. Moreover, by increasing the diagnosis rate, it can be found that the third wave of epidemic in Beijing has reached its peak, but the arrival of the next wave of COVID-19 is not ruled out. (5) Analyzing the situation in the United States, it can be seen that system (12) has better predictability than system (10). Meanwhile, by reducing the infection rate and increasing the diagnosis rate, the peak of the epidemic in the US can be accelerated. (6) Results show that the fractional-order system can accurately forecast the real data in the upcoming week when taking into account individual migration between two cities. By limiting the movement of exposed individuals, raising the diagnosis rate, and lowering the infection rate, Los Angeles’ peaks can appear and then decline immediately. Furthermore, this study makes several contributions to predict multi-peak of COVID-19 in China and suggestions on controlling epidemic in the US by changing certain parameters. Nevertheless, this research raises some issues that require more investigation, including how medical and other factors affect the spread of infectious diseases, how to properly administer vaccines, how network topology affects disease transmission and so on. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgment This work is supported by the Natural Science Foundation of Beijing Municipality [grant numbers Z180005], the National Nature Science Foundation of China [grant numbers 61772063] and the Fundamental Research Funds of the Central Universities [grant numbers 2020JBM074]. ==== Refs References 1 The Johns Hopkins University Center for System Science and Engineering, Data of accumulated and newly confirmed cases, recovered case and death case of COVID-19, URL https://github.com/CSSEGISandData/COVID-19. 2 Chan J. Yuan S. Kok K. To K. Chu H. Yang J. A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: a study of a family cluster Lancet 395 10223 2020 514 523 31986261 3 Huang C. Wang Y. Li X. Ren L. Zhao J. Hu Y. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China Lancet 395 10223 2020 497 506 31986264 4 Anderson R. Anderson B. May R. Infectious diseases of humans: Dynamics and control 1992 Oxford University Press 5 Bailey N. The mathematical theory of infectious diseases and its applications 1975 Charles Griffin Company Ltd 6 Upadhyay R. Pal A. Kumari S. Roy P. Dynamics of an SEIR epidemic model with nonlinear incidence and treatment rates Nonlinear Dynam 96 4 2019 2351 2368 7 Zou X. Wang K. Optimal harvesting for a stochastic regime-switching logistic diffusion system with jumps Nonlinear Anal Hybrid 13 2014 32 44 8 Korobeinikov A. Maini P. Nonlinear incidence and stability of infectious disease models Math Med Biol 22 2 2005 113 128 15778334 9 Peng L. Yang W. Zhang D. Zhuge C. Hong L. Epidemic analysis of COVID-19 in China by dynamical modeling 2020 10.1101/2020.02.16.20023465 arXiv preprint arXiv:2002.06563 10 Xu C. Yu Y. Chen Y. Lu Z. Forecast analysis of the epidemics trend of COVID-19 in the USA by a generalized fractional-order SEIR model Nonlinear Dynam 101 3 2020 1621 12634 11 Cheynet E. Generalized SEIR epidemic model (fitting and compution) 2020 GitHub URL http://www.github.com/ECheynet/SEIR 12 Gao D.Z. Cosner C. Cantrell R.S. Beier J.C. Ruan S.G. Modeling the spatial spread of Rift Valley fever in Egypt Bull Math Biol 75 3 2013 523 542 23377629 13 Phaijoo G.R. Gurung D.B. Mathematical study of dengue disease transmission in multi-patch environment Appl Math 7 14 2016 1521 1533 14 Eisenberg C.M. Shuai Z.S. Tien J.H. den Driessche P.V. A cholera model in a patchy environment with water and human movement Math Biosci 246 1 2013 105 112 23958383 15 Julien A. Davis J.R. David H. Richard J. Miller J.M. den Driessche P.V. A multi-species epidemic model with spatial dynamics Math Med Biol 2 2005 129 142 15778332 16 Cheng Z.K.J. Shan J. 2019 Novel coronavirus: where we are and what we know Infection 49 197 2021 17 Lu Z. Yu Y. Chen Y.Q. Ren G. Xu C. Wang S. A fractional-order SEIHDR model for COVID-19 with inter-city networked coupling effects Nonlinear Dynam 101 3 2020 1717 1730 18 Smethurst D. Williams H. Are hospital waiting lists self-regulating? Nature 410 6829 2001 652 653 11287943 19 Meerschaert M. Sikorskii A. Stochastic models for fractional calculus 2011 De Gruyter 20 Angstmann C. Henry B. Mcgann A. A fractional-order infectivity SIR model Physica A 2016 86 93 21 Khan M.A. Atangana A. Modeling the dynamics of novel coronavirus (2019-nCov) with fractional derivative Alex Eng J 59 4 2020 2379 2389 22 Chen Y. Cheng J. Jiang X. Xu X. The reconstruction and prediction algorithm of the fractional TDD for the local outbreak of COVID-19 2020 arXiv preprint arXiv:2002.10302 23 Amjad S.S. Najiroddin S.I. Kottakkaran N.S. A mathematical model of COVID-19 using fractional derivative: Outbreak in India with dynamics of transmission and control Adv Differ Equ 373 2020 1 19 32226454 24 Tang Z. Li X.B. Li H.Q. Prediction of new coronavirus infection based on a modified SEIR model MedRxiv 2020 10.1101/2020.03.03.20030858 25 Zhou P. Ma J. Tang J. Clarify the physical process for fractional dynamical systems Nonlinear Dynam 2020 1 12 26 Li H. Zhang L. Hu C. Jiang Y. Teng Z. Dynamical analysis of a fractional-order predator-prey model incorporating a prey refuge J Appl Math Comput 54 1–2 2017 435 449 27 Singh J. Kumar D. Hammouch Z. Atangana A. A fractional epidemiological model for computer viruses pertaining to a new fractional derivative Appl Math Comput 316 2018 504 515 28 Sierociuk D. Skovranek T. Macias M. Podlubny I. Petras I. Dzielinski A. Diffusion process modeling by using fractional-order models Appl Math Comput 257 2015 2 11 29 Podlubny I. Fractional differential equations 1999 Academic Press 30 Li Y. Chen Y. Podlubny I. Mittag–Leffler stability of fractional order nonlinear dynamic systems Automatica 45 8 2009 1965 1969 31 Tang Q. Teng Z. Jiang H. Global behaviors for a class of multi-group SIRS epidemic models with nonlinear incidence rate Taiwan J Math 19 5 2015 1509 1532 32 Tang Z. Li X. Li H. Prediction of new coronavirus infection based on a modified SEIR model Cold Spring Harbor Lab 2020 1 13 33 Lin M. Huang J. Ruan S. Yu P. Bifurcation analysis of an SIRS epidemic model with a generalized nonmonotone and saturated incidence rate J Differ Equ 267 3 2019 1859 1898 32226129 34 Muroya Y. Enatsu Y. Kuniya T. Global stability for a multi-group SIRS epidemic model with varying population sizes Nonlinear Anal Real 14 3 2013 1693 1704 35 Angstmann C.N. Donnelly I.C. Henry B.I. Langlands T. Continuous-time random walks on networks with vertex- and time-dependent forcing Phys Rev E 88 2 2013 022811 36 Li M. Graef J.R. Wang L.C. Karsai J. Global dynamics of a SEIR model with varying total population size Math Biosci 160 2 1999 191 213 10472754 37 Kheiri H. Jafari M. Stability analysis of a fractional order model for the HIV/AIDS epidemic in a patchy environment J Comput Appl Math 2018 323 339 38 van den Driessche P. Watmough J. Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmission Math Biosci 180 1–2 2002 29 48 12387915 39 Diekmann O. Heesterbeek J. Metz J. On the definition and the computation of the basic reproduction ratio R0 in models for infectious diseases in heterogeneous populations J Math Biol 28 4 1990 365 382 2117040 40 van den Driessche P. Watmough J. Heterogeneous social interactions and the COVID-19 lockdown outcome in a multi-group SEIR model Math Model Nat Phenom 15 36 2020 41 Khajanchi S. Bera S. Roy T.K. Mathematical analysis of the global dynamics of a HTLV-I infection model, considering the role of cytotoxic T-lymphocytes Math Comput Simulat 180 2021 42 Berman A. Plemmons R. Nonnegative matrices in the mathematical sciences 1979 Academic Press 43 Ruan S.G. Wang W.D. Levin S.A. The effect of global travel on the spread of SARS Math Biosci Eng 3 1 2012 205 218 44 Zhang K. Ji Y.P. Pan Q.W. Wei Y.M. Liu H. Sensitivity analysis and optimal treatment control for a mathematical model of Human Papillomavirus infection AIMS Math 5 5 2020 2646 2670 45 Huo H.F. Feng L.X. Global stability for an HIV/AIDS epidemic model with different latent stages and treatment Appl Math Model 37 3 2013 1480 1489 46 Tang B. Xia F. Tang S. Bragazzi N. Li Q. Sun X. The effectiveness of quarantine and isolation determine the trend of the COVID-19 epidemic in the final phase of the current outbreak in China Int J Infect Dis 96 2020 636 647 32689711

0

PMC9748873

NO-CC CODE

2022-12-15 23:22:47

no

Indian Pediatr. 2022 Dec 14; 59(11):894

latin-1

Indian Pediatr

10.1007/s13312-022-2656-5

oa_other

==== Front Indian Pediatr Indian Pediatr Indian Pediatrics 0019-6061 0974-7559 Springer India New Delhi 2654 10.1007/s13312-022-2654-7 Correspondence COVID-19 and Tuberculosis in Children Tripathy Saroj Kumar 1 Das Sarthak drsarthakdas2222@gmail.com 1 Mane Sushant 2 Pustake Manas pustakemanas@gmail.com 2 1 grid.413618.9 0000 0004 1767 6103 Department of Pediatrics, All India Institute of Medical Sciences, Deoghar, Jharkhand India 2 Centre of Excellence for Pediatric TB, Grant Government Medical College and Sir JJ Group of Hospitals, Byculla, Mumbai, Maharashtra India 14 12 2022 2022 59 11 892892 © Indian Academy of Pediatrics 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. issue-copyright-statement© Indian Academy of Pediatrics 2022 ==== Body pmc ==== Refs References 1. Mane SS Janardhanan J Pustake M Outcome of COVID-19 in children with tuberculosis: Single-center experience Indian Pediatr 2022 59 617 19 10.1007/s13312-022-2574-6 35962655 2. Chansaenroj J Yorsaeng R Posuwan N Long-term specific IgG response to SARS-CoV-2 nucleocapsid protein in recovered COVID-19 patients Sci Rep 2021 11 23216 10.1038/s41598-021-02659-4 34853374 3. Christophers B Gallo Marin B Oliva R Trends in clinical presentation of children with COVID-19: A systematic review of individual participant data Pediatr Res 2022 91 494 501 10.1038/s41390-020-01161-3 32942286 References 1. Mane SS Janardhanan J Pustake M Outcome of COVID-19 in children with tuberculosis: Single-center experience Indian Pediatr 2022 59 617 19 10.1007/s13312-022-2574-6 35962655 2. Mane SS Pustake M Authors’ reply Indian Pediatr 2022 59 883

36370024

PMC9748874

NO-CC CODE

2022-12-15 23:22:47

no

Indian Pediatr. 2022 Dec 14; 59(11):892

utf-8

Indian Pediatr

10.1007/s13312-022-2654-7

oa_other

==== Front Biologia (Bratisl) Biologia (Bratisl) Biologia 0006-3088 1336-9563 Springer International Publishing Cham 1269 10.1007/s11756-022-01269-3 Original Article Effect of zinc nanoparticles on the growth and biofortification capability of mungbean (Vigna radiata) seedlings Sorahinobar Mona m.sorahinobar@alzahra.ac.ir 1 Deldari Tooba 1 Nazem Bokaeei Zahra 1 Mehdinia Ali 2 1 grid.411354.6 0000 0001 0097 6984 Department of Plant Sciences, Faculty of Biological Sciences, Alzahra University, Tehran, Iran 2 Iranian National Institutes for Oceanography and Atmospheric Science, Tehran, Iran 14 12 2022 110 4 6 2022 8 11 2022 © The Author(s), under exclusive licence to Plant Science and Biodiversity Centre, Slovak Academy of Sciences (SAS), Institute of Zoology, Slovak Academy of Sciences (SAS), Institute of Molecular Biology, Slovak Academy of Sciences (SAS) 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Zinc insufficiency is a nutritional trouble worldwide, especially in developing countries. In the current study, an experiment was conducted to evaluate the effect of supplementation of MS media culture with different concentrations of ZnO nanoparticles (NPs) (0, 10, 20, 40, 80, and 160 ppm) on growth, nutrient uptake, and some physiological parameters of 7-days-old mung bean seedlings. ZnO NPs enhanced the Zn concentration of mung bean from 106.41 in control to more than 4600 µg/g dry weight in 80 and 160 ppm ZnO NPs treated seedlings. Our results showed that ZnO NPs in the concentration range from 10 to 20 ppm had a positive influence on growth parameters and photosynthetic pigments. Higher levels of ZnO NPs negatively affected seedling’s growth by triggering oxidative stress which in turn caused enhancing antioxidative response in seedlings including polyphenol oxidase and peroxidase activity as well as phenolic compounds and anthocyanine contents. Considering the positive effects of ZnO NPs treatment on mungbean seedlings growth, micronutrents, protein and shoot phenolics content, 20 ppm is recommended as the optimal concentration for biofortification. Our findings confirm the capability of ZnO NPs in the remarkable increase of Zn content of mungbean seedlings which can be an efficient way for plant biofortification and dealing with environmental stress. Supplementary information The online version contains supplementary material available at 10.1007/s11756-022-01269-3. Keywords Antioxidative response Biofortification Mung bean Nanoparticles Oxidative stress Phenolic compounds Photosynthetic pigments Zinc oxide ==== Body pmcIntroduction It is now apparent that a nutritional deficiency of zinc in humans is widespread, affecting nearly 2 billion individuals worldwide (Prasad 2008). Zinc has a critical role in homeostasis, immune system function, oxidative stress, apoptosis, aging, and significant disorders of great public health interest are associated with zinc deficiency. Reports confirm that dietary supplements of zinc are efficient in the control and treatment of covid 19 (reviewed by Giacalone et al. 2021). Insufficient intake of zinc from food is a major contributor to zinc deficiency in humans which somehow is related to the low Zn content of agricultural soil (Tabrez et al. 2022). Hence there is an urgent need to increase the zinc content and bioavailability in food especially in developing and poor countries (Welch and Graham 2004; Zhao and McGrath 2009). Crops biofortification offers a sustainable solution to reduce malnutrition and deal with human diseases. Use of nanotechnology to fortify the crops in human societies has received much attention in recent years (El-Ramady et al. 2021; Khan et al. 2021a, b). Crops nano-biofortification could be achieved by seed priming (Rizwan et al. 2019), soil and foliar application (Du et al. 2019; Semida et al. 2021), and cultivation of plants in media rich of candidate nutrient using nanomaterials. In this regard, various aspects should be considered including: (i) finding the appropriate type and shape of nanomaterials, (ii) determination of optimal dose (without negative effects on plant growth and physiology), and (iii) investigation of probable negative or positive effect on the absorption and accumulation of other nutrients. Mung bean seeds are used up by humans and its hay is used for animal feedstuff that its products can be back to humans through meat, milk, and dairy products. About 6 million-hectare area of the world is under mung bean cultivation (8.5% of the world’s pulse area). Biofortification of crops like mung bean (Vigna radiate (L.) R. Wilczek) which contains carbohydrates, proteins, calcium, β-carotene, iron, and a lot of micronutrients can be a suitable strategy, especially for poor communities to cope with a nutritional deficiency. Haider et al. (2021) and Rani et al. (2022) confirmed the capability of using ZnO NPs for the biofortification of mung beans in the field and hydroponic greenhouse environments. These works have focused on zinc biofortification and failed to address the effect of ZnO NPs treatments on other nutrients uptake. Additionally, more precise studies are required to clarify the effect of ZnO NPs on plant physiology and to find the optimal dose for zinc biofortification. Mung bean sprouts are now part of the diet of people in many countries as an appetizer and salad. One of the objectives of the current study was to evaluate the capability of the ZnO NPs in mung bean sprout biofortification, performance, and nutrient quality. Like humans, zinc is crucial for the enzyme functioning in plants and it is very important for plant growth and development. Zn deficiency in plants is one of the major concerns globally. In recent years, the West Asia region has been severely affected by the effects of climate change, such as successive droughts, the expansion of salinized land and dust storms, which have affected agricultural production. Many reports confirm Zn supplementation could alleviate the negative effects of abiotic stresses on plants (Khan et al. 2004; Hussein and Abou-Baker 2018; Thounaojam et al. 2012; Sattar et al. 2021). Micronutrients nanoparticles because of their small size and high surface-to-volume ratio are considered a good candidate for plant uptake and biofortification. The application of appropriate concentrations of zinc oxide nanoparticles (ZnO NPs) with positive effects on plant physiology can improve the growth and protection of different plant species (Mukherjee et al. 2016; Subbaiah et al. 2016; Venkatachalam et al. 2017). A high concentration of ZnO NPs can cause phytotoxicity in plants (Salah et al. 2015; Wang et al. 2016) and produce oxidative stress, which is common in plants in response to environmental stresses. To cope with oxidative stress, plants increase their enzymatic and non-enzymatic anti-oxidants content which helps hinder reactive oxygen species. Some of those non-enzymatic antioxidants are secondary metabolites like phenolic compounds which in turn have nutritional value for humans as antioxidants. Zinc NPs could be a potential high-performed fertilizer for enhancing plant yield and quality (Du et al. 2019). Accordingly, another aspect of the present study was to investigate the effect of ZnO NPs on the growth and physiology of mung bean seedlings. In the current study, we investigated the effects of different concentrations of ZnO NPs on seedling growth, biochemistry, and physiological response of V. radiate. Materials and methods Plant material and culture conditions Seeds of mung bean (V. radiata) were surface sterilized in 1% (v/v) sodium hypochlorite solution for 15 min, followed by three washes with sterile distilled water. Ten sterilized seeds were placed into Petri dishes containing 10 ml of MS basal medium (Murashige and Skoog 1962), 7% agarose, pH = 5.7 and 0, 10, 20, 40, 80, and 160 ppm of ZnO NPs (10–30 nm, nearly spherical, US-NANO). The stock water suspension of ZnO NPs, before application in media culture, was placed for 30 min in the Elmasonic S80(H) Ultrasonic Cleaner (37 kHz of ultrasonic frequency, of 150 W of effective ultrasonic power) (Elma Schmidbauer GmbH, Singen, Germany) to achieve better dispersion of particles. Petri dishes were placed in a tissue culture room at a temperature of 25 °C, relative humidity of 55%, and a 16-h photoperiod with a light intensity of 2300 lx. Seedlings obtained after 7 days of growth were used for experiments. The biometrical data of 7-days-old seedlings were collected as mean values of the shoot and root fresh and dry weight (mg) and length (cm). For physiological and biochemical assays 7 days old seedlings were frozen in liquid nitrogen and kept at -70 °C and used for the experiments. Multielement analysis 0.5 g of powdered freeze-dried mung bean shoot samples was placed in polyethylene tube with 8 ml of nitric acid (65%) for 12 h at room temperature. Then 2 ml perchloric acid was added to the samples and kept at 80 °C for 1 h and 150 °C for 3 h. The solution was filtered using Whatman No. 42 filter paper and diluted to 25 mL with ultrapure water and stored in the dark before analysis. ICP-MS (HP-4500, USA) equipped with an Asx-520 auto-sampler was used to define elements concentration. Pigment contents The total contents of carotenoids, chlorophylls a and b were measured according to the procedure described by Linchenthaler and Wellburn (1983). Briefly, fresh samples (0.2 g) were homogenized in 80% acetone (Merck, Germany) and the absorbance was recorded spectrophotometrically at 470, 646, and 663 nm. The results were calculated based on the following equations: Chla(μg/ml)=12.26A663-2.79A646Chlb(μg/ml)=21.50A646-5.10A663ChlTotal(μg/ml)=Chla+ChlbCarotenoids(μg/ml)=(1000A470-3.27Chla-104Chlb)/229 Anthocyanin content was extracted using 1% HCl v/v acidified methanol. Fresh samples were homogenized in the extraction solution, centrifuged at 18 000 × g at 4 °C for 15 min, and stored in darkness for 5 h at 5 °C. The amount of anthocyanin was quantified at 550 nm spectrophotometrically (Abdel Latef et al. 2020). Malondialdehyde and hydrogen peroxide content Malondialdehyde (MDA) content was determined based on Heath and Packer (1968). Fresh tissues were extracted with 0.1% TCA, and 0.5 ml of the supernatant was mixed with 1 ml of thiobarbituric acid (0.5%) in TCA (20%). After heating the solution at 95 °C for 25 min, the absorbance was read spectrophotometrically at 532 and 600 nm. The amount of MDA was calculated by an extinction coefficient of 155 mM− 1 cm− 1. For H2O2 quantification, the 0.5 ml of extract (0.1% TCA) was mixed with 0.5 ml potassium phosphate buffer (10 mM, pH 7.0) and 1 ml KI (1 M). The absorbance was measured at 390 nm based on the Velikova et al. (2000) method. Protein content and antioxidant enzyme activity 0.2 g of fresh samples was ground in 1 M Tris-HCl (pH 6.8) at 4 °C. The supernatant was separated at 12 000 × g centrifugation for 15 min and used for protein and enzyme assays. Protein content was quantified with bovine serum albumin as the standard (Bradford 1976). Antioxidant enzyme assay For measurement of peroxidase (POX) activity, 50 µl of the extract was mixed with 0.1 ml benzidine, 0.2 ml H2O2 (3%), and 0.2 M acetate buffer (pH 4.8), and the activity was defined at 530 nm (Abeles and Biles 1991). Polyphenol oxidase (PPO; E.C. 1.14.18.1) activity was measured based on the method described by Raymond et al. (1993). The reaction mixture contained 0.2 ml of pyrogallol (20 mM), 2.5 ml of 200 mM sodium phosphate buffer (pH 6.8), and 50 µl of enzyme extract. The enzyme activity was recorded at 430 nm. Total phenolic compounds content For extraction, 0.4 g of dried powder of samples was placed in 80% (v/v) methanol for 48 h and then was put in an ultrasonic bath for 20 min. After centrifugation (10 min at 11 000 × g), the supernatant was utilized for the detection of total phenolic compounds. For determination of total phenolic content, 100 µl of the extract was mixed with 500 µl of Folin-Ciocalteu reagent and 400 µl of sodium carbonate. Then, the reaction solution was kept for 30 min at room temperature. The absorbance was read spectrophotometrically at 630 nm (Singleton and Rossi 1965). Statistical analyses The experiment was set up in a completely randomized design. For experimental treatment, 10 repetitions were used with 10 seeds each (1 Petri dish). For biometrical analysis at least 20 plants were assayed in each treatment. For physiological and biochemical assays 25 seedlings were pooled and used as one replication of triplicate. Each data point was an average of three or five replicates. Presented data were analyzed by one way ANOVA (analysis of variance) using SPSS (version 21). The significance of differences was determined according to Duncan’s multiple range tests at the 0.05 level of probability. The graphs were designed with Graphpad Prism version 6 and Microsoft Excel version 2010. Principal component analysis (PCA) and Pearson correlation test were conducted using publicly available Past3.16 software. Results Zn content significantly increased with the increase of ZnO NPs concentration in media culture, on average, from 106.41 in control to more than 4 600 µg/g dry weight in 80 and 160 ppm ZnO NPs treated seedlings. No significant difference in Zn content was observed in the samples treated with 10 and those treated with 20 ppm ZnO NPs. Interestingly, with the exception of Zinc, the lowest concentrations of the all assayed elements were observed in 10 ppm ZnO NPs treated samples. On the other hand, the highest levels of calcium, magnesium, phosphorus, manganese, iron, cobalt, copper, and molybdenum were observed in the 40 ppm ZnO NPs treated samples (Table 1). Although none of the evaluated elements showed a positive or negative correlation with zinc, a strong correlation (≥ 90%) between K with Mg and P, Mg with P, Fe and Cu, P with K, Mn with Fe and Co, Fe with P, and Mn and Co were observed (Supplementary Fig. 1). Table 1 Nutrient content (µg/g dry weight) in shoot of mung bean (V. radiata) seedlings subjected to 0 (control), 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles Control 10 20 40 80 160 K 30504.01 ± 110.12e 26343.34 ± 80.39f 36993.42 ± 98.70a 36682.3 ± 160.04b 34903.01 ± 140.23c 33015.55 ± 90.04d P 7776.04 ± 38.39e 6253.08 ± 42.45f 9463.62 ± 56.14c 10912.7 ± 105.14a 10596.2 ± 89.19b 9364.67 ± 43.12d Ca 2899.2 ± 50.10d 2365.56 ± 25.09e 2966.73 ± 60.15c 3821.63 ± 34.56a 3645.18 ± 17.20b 3807.94 ± 80.78a Mg 1754.26 ± 40.04d 1353.49 ± 22.15e 1840.01 ± 13.18c 2075.26 ± 35.19a 1940.8 ± 28.40b 1815.57 ± 43.89c Na 792.88 ± 20.40d 686.38 ± 12.81e 1022.61 ± 30.08b 1014.43 ± 14.20b 926.45 ± 16.04c 1298.53 ± 22.31a Mn 430.56 ± 14.06d 250.55 ± 8.19f 390.77 ± 20.86c 573.72 ± 19.18a 501.23 ± 31.12b 312.95 ± 25.57e Fe 154.58 ± 10.63b 114.66 ± 19.02c 158.89 ± 21.74b 206.42 ± 13.94a 183.71 ± 8.11a 153.70 ± 9.12b Zn 106.41 ± 15.01e 815.66 ± 24.12d 867.34 ± 37.67d 2665.91 ± 84.88c 4745.59 ± 30.15a 4644 ± 50.04b Cu 33.91 ± 4.10a 7.52 ± 1.03c 29.91 ± 6.01b 34.49 ± 2.19a 30.84 ± 4.90a 26.66 ± 3.02b Mo 57.83 ± 11.02c 30.24 ± 4.20e 34.88 ± 5.09e 182.45 ± 8.15a 129.75 ± 7.50b 44.36 ± 3.50d Co 0.35 ± 0.01d 0.26 ± 0.01f 0.40 ± 0.09c 0.54 ± 0.08a 0.45 ± 0.05b 0.32 ± 0.03e Different letters indicate means that are significantly different at (P ≤ 0.05) Seedling growth ZnO NPs treatment induced changes in the growth biomarkers of V. radiata seedlings (Fig. 1). The use of ZnO NPs in the concentration range from 10 to 20 ppm had a positive influence on growth. The fresh and dry weight of shoot and root increased up to 20 ppm and then decreased at 40, 80, and 160 ppm. Shoot and root length also increased significantly following ZnO NPs treatment up to 20 ppm, while the length of seedlings was negatively affected at higher concentrations. In summary, the highest and lowest rate of seedling biomass including plant height as well as fresh and dry weight was observed in 20 ppm and 160 ZnO NPs treated samples respectively (Supplementary Fig. 2). Fig. 1 Relative changes in biomass from control of the 7-day-old mung bean (V. radiata) seedlings in response to 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles Photosynthetic pigments Chl a, Chl b, Chl T and carotenoids content significantly increased (P ≤ 0.05) up to 20 ppm ZnO NPs and then markedly reduced with the increase of ZnO concentrations in media culture (Fig. 2). Fig. 2 The content of photosynthetic pigments: chlorophyll a, chlorophyll b, total chlorophyll and carotenoids in shoots of 7 days old mungbean (V. radiata) seedlings cultured on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at (P ≤ 0.05) Protein content The shoot soluble protein content was observed to be the highest at 40 to160 ZnO NPs treated samples (Fig. 3). In the root, total soluble protein content significantly increased up to 40 ppm then significantly decreased under higher concentrations (80 and 160 ppm) of ZnO NPs (P ≤ 0.05). Fig. 3 The root and shoot total protein content of mungbean (V. radiata) seedlings grown on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at P ≤ 0.05 Lipid peroxidation and H2O2 content The content of MDA (as a bio-indicator of lipid peroxidation) and H2O2 (as the most stable ROS) in the shoot samples of ZnO treated samples were significantly higher than that of controls (Fig. 4). The highest content of H2O2 and MDA in shoot and root samples treated with ZnO NPs were observed at 160 and 40 ppm respectively. Root H2O2 content in response to ZnO NPs treatment was highly increased as compared to shoot. However, lipid peroxidation in shoot was higher which probably can be justified by higher anti-oxidants in root. Fig. 4 The content of A H2O2 and B malondialdhyde(MDA) in root and shoot of mungbean (V. radiata) seedlings grown on MS media containing various levels of zinc oxide nanaoparticles (0, 10, 20, 40, 80 and 160 ppm). Different letters indicate means that are significantly different at P ≤ 0.05 Enzymatic antioxidant assay ZnO NPs treatment induced the activities of POX and PPO of V. radiate in both shoot and root (Fig. 5). The results showed the activity of POX in the shoot was considerably lower than in root samples. The highest activity levels of POX were observed in the shoot and root samples collected from seedlings grown on 20 ppm and 160 ppm of ZnO NPs, respectively. The highest activity level of PPO activity in both root and shoot was observed at 160 ppm ZnO NPs treated samples. Fig. 5 The activities of A proxidase and B polyphenol oxidase enzymes in root and shoot of samples of mungbean seedlings grown on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at P ≤ 0.05 Non-enzymatic antioxidants The effect of ZnO nanoparticles on total phenolic compounds and anthocyanin was presented in Fig. 6. The treatment with ZnO NPs increased the total phenolic content of both shoot and root. The highest phenolic compounds induction was observed in shoot and root samples grown on 20 and 160 with the elevation rate of 207% and 160% percent as compared to control. Additionally, the highest level of the shoot and root anthocyanin content was observed at 160 and 80 ppm ZnNPs treated samples respectively (Fig. 6). No significant differences in anthocyanin content were observed in 20 to 160 and 40 to 160 ZnO NPs treated samples of shoot and root respectively. The principal component analysis (PCA) was performed on Z-scored transformed data of assayed data. PCA showed that 79% variation between samples could be explained by two (PC1 = 57.60 and PC2 = 21.98%) principal components (Fig. 4). The first component (PCA1) separates the 0, 10, and 20 ppm ZnO NPs treated samples from other species primarily based on and shoot zinc, calcium, H2O2, MDA, and chlorophyll content as well as root fresh weight (Fig. 7). The second component (PCA2) separated 40 and 60 ppm ZnO NPs treated samples from other species primarily based on cobalt, potassium, phenolic compounds contents as well as PPO and POX activities. These results confirmed that plant nutritional status is highly affected under ZnO NPs treatment followed by photosynthesis and antioxidative capability. Fig. 6 The content of A phenolic compounds and B anthocyanins in root and shoot of mungbean seedlings grown on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at P ≤ 0.05 Fig. 7 Principal Component Analysis (PCA) of the z-score transformed mean of all assayed parameters of mungbean seedlings grown on MS media containing 0 (Control), 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. First two axes shown are 57% and 21% of the variation in composition Discussion The increased zinc content of shoot in response to ZnO NPs supplementation confirmed that the application of zinc nanoparticles can be an effective way to increase the zinc content of mung bean seedlings. Considering that the recommended daily intake of zinc in the diet is between 8 and 12 mg, with the consumption of 2 g of 7 days old mung bean seedlings, germinated and grown on MS medium containing 80 ppm of zinc oxide nanoparticles, this need can be met. Thus, according to the tremendous effect of ZnO NPs treatment on the zinc content of seedlings (about 45 times more than that of control), this method could lead to an augmentation that could meet the daily needs of adults. Although the lowest concentration of ZnO NPs used in this study caused a more than 7-fold increase in zinc content of seedlings shoot, however it had a negative effect on the content of other analyzed elements. This finding, along with its positive effect on plant biomass, shows that the positive effect of zinc is related to its own nutritional value, not its effect on the absorption of other elements (at least the elements that have been studied). Because at least a part of the nutrients required for the seedling’s shoot is supplied from the seed storage, it is expected that the content of elements in the shoot samples treated with ZnO NPs should not be less than that of the control samples. Therefore, it seems that the negative effect of zinc (at 10 ppm) on the content of the studied elements is related to its effect on the translocation of elements from seed to shoot not to the absorption of other elements from the medium culture. Generally, the solubility of Zn in the media culture and soil increases with decrease in pH (Salinitro et al., 2020). It seems that the low pH of MS media culture increased the bioavailability of Zn for plant uptake. In plants, the interaction between Zn and other plant nutrients do exist and both positive and negative interactions are reported (Prasad et al. 2016). Regarding the positive effect of zinc on plant growth, two points can be mentioned: Firstly, there are some reports that confirm the positive effect of zinc on tryptophan (as a precursor of auxin) and gibberellic acid content in plants which as plant growth regulators can promote plant growth (Mašev and Kutáček 1966; Wang et al. 2021). Secondly, our findings of increased leaves content of Chla, Chlb, and carotenoid (up to 20 ppm) are a good indicator of the positive impact of zinc on plant photosynthesis. In the current study seedlings’ growth (biomass) showed a strong zinc concentration-dependent trend. The positive impact of low concentration (10 and 20 ppm) and negative impact of high concentration (40 to 160 ppm) of ZnO NP media culture supplementation on seedling growth may be related to its nutritional value on one hand and toxicological effects on the other hand. Leaves photosynthetic pigments as well as root MDA and H2O2 content can be considered as bioindicators for the above-mentioned positive and negative effects. Zinc is an essential nutrient for plants and it has a role in protein synthesis and activity of several enzymes, as well as membrane integrity, metabolic reactions, water uptake, transport, and gene expressions (Rudani et al. 2018). Hence, an increase in Zn level (at least up to 715% which was observed in 20 ppm ZnO NP supplemented media) was beneficial for the plants, leading to growth improvement. The positive effect of Zn application on plant chlorophyll content and photosynthesis was reported in different plants such as Triticum aestivum L., Brassica juncea (L.) Czern., and Gossypium hirsutum L. (Wu et al. 2015; Khan et al. 2016; Sultana et al. 2016). Similar to our findings the positive impact of low (≥ 20 ppm) concentration of ZnO NPs on plant growth and protein content was reported in Zea mays L. (Sabir et al. 2020), and Lycopersicon esculentum Mill. (Faizan et al. 2018). The negative effect of a higher concentration of ZnO NPs on plant growth was reported in the other studies (Zhang et al. 2015; Javed et al. 2017; Tymoszuk and Wojnarowicz 2020; Rani et al. 2022). ZnO NP toxicity (at a concentration of 40 to 160 ppm) might be due to the perturbed homeostasis of Zn and its impacts on other elements’ homeostasis (Srivastav et al. 2021). A strong correlation (~ 98%) between shoot Zn and MDA is a signal of lipid peroxidation and probable membrane damage (Supplementary Fig. 1). The finding is an indication of the creation of oxidative stress due to excessive accumulation of zinc as reported in the other plants (Remans et al. 2012; Feigl et al. 2015). Researchers have discussed about the mechanisms of nanoparticle-induced oxidative stress including light activation of electron hole pairs, active electronic configurations and functional groups generation on the surface of nanoparticles, and active redox cycling on the surface of nanoparticle (reviewed by Saliani et al. 2016). There are evidences that confirm the smaller nanoparticles have higher density of surface defects and provide greater free electrons and holes which leads to the generation of ROS (Singh et al. 2021).To eliminate excess ROS, plants activate enzymatic and non-enzymatic antioxidant defense systems. In this regard, the activation of secondary plant metabolism and synthesis of phenolic compounds like anthocyanins is a crucial way to scavenge ROS (Sheteiwy et al. 2016; Mittler 2017; Xu and Rothstein 2018). In the current study root PPO activity and anthocyanin content showed a significant correlation with internal Zn content (Supplementary Fig. 1). It seems that this strategy could restrain ROS accumulation in the root. A similar response of increased phenolic content to ZnO NP treatment was reported in Brassica nigra (L.) K.Koch, potato, and Coriandrum sativum L. (Marichali et al. 2014; Zafar et al. 2016; Raigond et al. 2017). However, it seems that change in the activity of PPO and POX along with the increased content of phenolic compounds has not been enough to inhibit oxidative damage in the seedling shoot (Wang et al. 2018). This observation is probably due to the fact that the photosynthesis apparatus of plant leaves are among the main site of ROS generation (Khorobrykh et al. 2020). Comparing the content of assayed minerals in the samples, confirmed that the seedlings grown on the 40 ppm ZnO NPs, accumulated the highest amount of elements (a total of 58175.35 µg/g). Meanwhile, the highest seedling dry-weight biomass was found at 20 ppm ZnO NPs treatment. Considering the content of mineral nutrients in the total biomass produced in 7 day time period, it is determined that the highest mineral level is found in 20 ppm (2118.72) followed by 40 ppm (1602.73), 10 ppm (1324.52), 80 ppm (1308.87), control (1267.81), and 160 ppm (1021.70). These data confirm the advantage of samples grown on 20 ppm ZnO NPs over others. Conclusion Taken together our findings confirm the capability of ZnO NP in zinc biofrotification of mung bean seedlings. Despite the significantly increased content of many micro and macronutrients especially under 40 ppm ZnO NPs supplemented medium, its stressful effects were observed as a decrease in growth and photosynthetic pigments and an increase in the content of H2O2 and MDA. Considering the positive effects of ZnO NP on growth, photosynthetic pigments, protein content, activity, and content of enzymatic and non-enzymatic antioxidants in mung bean seedlings, a concentration of 20 ppm is recommended as the optimal concentration. Supplementary information Below is the link to the electronic supplementary material.Supplementary Fig. 1 Pearson correlation analysis of all assayed variables in mungean seedlings germinated and grown on MS media supplemented with 0, 10, 20, 40, 80, and 160 ppm ZnO NPs. (JPG 4.13 MB) Supplementary Fig. 2 Growth biomarkers of mungbean seedlings grown on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at P ≤ 0.05. (JPG 300 KB) Acknowledgements The financial support was provided by Alzahra University. Author contributions Mona Sorahinobar and Zahra Nazem Bokee designed experiments with the assistance of Ali Mehdinia, Toba Deldari carried out experiments, Mona Sorahinobar analyzed experimental results and data and wrote the manuscript. Declarations Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Ethical approval This article does not contain any studies with human participants or animals performed by any of the authors. Consent for publication Not applicable. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abdel Latef AAH Dawood MF Hassanpour H Rezayian M Younes NA Impact of the static magnetic field on growth, pigments, osmolytes, nitric oxide, hydrogen sulfide, phenylalanine ammonia-lyase activity, antioxidant defense system, and yield in lettuce Biology 2020 9 7 172 10.3390/biology9070172 32709036 Abeles FB Biles CL Characterization of Peroxidases in Lignifying Peach Fruit Endocarp Plant Physiol 1991 95 1 269 273 10.1104/pp.95.1.269 16667963 Bradford MM A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding Anal Biochem 1976 72 1-2 248 254 10.1016/0003-2697(76)90527-3 942051 Du W Yang J Peng Q Liang X Mao H Comparison study of zinc nanoparticles and zinc sulphate on wheat growth: from toxicity and zinc biofortification Chemosphere 2019 227 109e116 10.1016/j.chemosphere.2019.03.168 30986592 El-Ramady H Abdalla N Elbasiouny H Elbehiry F Elsakhawy T Omara AED Nano-biofortification of different crops to immune against COVID-19: A review Ecotoxicol Environ Saf 2021 222 112500 10.1016/j.ecoenv.2021.112500 34274837 Faizan M Faraz A Yusuf M Khan ST Hayat S Zinc oxide nanoparticle-mediated changes in photosynthetic efficiency and antioxidant system of tomato plants Photosynthetica 2018 56 2 678 686 10.1007/s11099-017-0717-0 Feigl G Lehotai N Molnár A Ördög A Rodríguez-Ruiz M Palma JM Kolbert Z Zinc induces distinct changes in the metabolism of reactive oxygen and nitrogen species (ROS and RNS) in the roots of two Brassica species with different sensitivity to zinc stress Ann Bot 2015 116 4 613 625 10.1093/aob/mcu246 25538112 Giacalone A Marin L Febbi M Tovani-Palone MR Current evidence on vitamin C, D, and zinc supplementation for COVID-19 prevention and/or treatment Electron J Gen Med 2021 18 5 em311 10.29333/ejgm/11099 Haider MU Hussain M Farooq M Ul-Allah S Ansari MJ Alwahibi MS Farooq S Zinc biofortification potential of diverse mungbean [V. radiata (L.) Wilczek] genotypes under field conditions PLoS ONE 2021 16 6 e0253085 10.1371/journal.pone.0253085 34161364 Heath RL Packer L Photoperoxidation in isolated chloroplasts Arch Biochem Biophys 1968 125 1 189 198 10.1016/0003-9861(68)90654-1 5655425 Hussein MM Abou-Baker NH The contribution of nano-zinc to alleviate salinity stress on cotton plants R Soc Open Sci 2018 5 8 171809 10.1098/rsos.171809 30224982 Javed R Usman M Yücesan B Zia M Gürel E Effect of zinc oxide (ZnO) nanoparticles on physiology and steviol glycosides production in micropropagated shoots of Stevia rebaudiana Bertoni Plant Physiol Biochem 2017 110 94 99 10.1016/j.plaphy.2016.05.032 27246994 Khan MK Pandey A Hamurcu M Gezgin S Athar T Rajput VD Gupta OP Minkina Insight into the prospects for nanotechnology in wheat biofortification Biology 2021 10 11 1123 10.3390/biology10111123 34827116 Khan ST, Malik A, Alwarthan A, Shaik MR (2021b) The enormity of the zinc deficiency problem and available solutions; an overview. Arab J Chem 103668. 10.1016/j.arabjc.2021.103668 Khan R Gul S Hamayun M Shah M Sayyed A Ismail H Gul H Effect of foliar application of zinc and manganese on growth and some biochemical constituents of Brassica junceae grown under water stress J Agric Environ Sci 2016 16 984 997 10.5829/idosi.aejaes.2016.16.5.12923 Khan MU Qasim M Jamil M Effect of Zn on starch content of paddy and zinc content of soil, leaf and root of rice grown in calcareous soils Int J Agric Biol 2004 6 6 1132 1135 Khorobrykh S Havurinne V Mattila H Tyystjärvi E Oxygen and ROS in photosynthesis Plants 2020 9 1 91 10.3390/plants9010091 31936893 Linchenthaler HK Wellburn AR Determination of total carotenoides and clorophyll a and b of leaf extracts in diferent solventes Biochem Soc Transac 1983 11 1591 1592 10.1042/bst0110591 Mašev N Kutáček M The effect of zinc on the biosynthesis of tryptophan, andol auxins and gibberellins in barley Biol Plant 1966 8 2 142 151 10.1007/BF02930623 Marichali A Dallali S Ouerghemmi S Sebei H Hosni K Germination, morpho-physiological and biochemical responses of coriander (Coriandrum sativum L.) to zinc excess Ind Crops Prod 2014 55 248 257 10.1016/j.indcrop.2014.02.033 Mittler R ROS are good Trends Plant Sci 2017 22 1 11 19 10.1016/j.tplants.2016.08.002 27666517 Mukherjee A Sun Y Morelius E Tamez C Bandyopadhyay S Niu G Gardea-Torresdey JL Differential toxicity of bare and hybrid ZnO nanoparticles in green pea (Pisum sativum L.): a life cycle study Front Plant Sci 2016 6 1242 10.3389/fpls.2015.01242 26793219 Murashige T Skoog F A revised medium for rapid growth and bio assays with tobacco tissue cultures Physiologia Plantarum 1962 15 3 473 497 10.1111/j.1399-3054.1962.tb08052.x Prasad AS Zinc in human health: effect of zinc on immune cells Mol Med 2008 14 5 353 357 10.2119/2008-00033.Prasad 18385818 Prasad R Shivay YS Kumar D Interactions of zinc with other nutrients in soils and plants-A Review Ind J Fertilisers 2016 12 5 16 26 Raigond P Raigond B Kaundal B Singh B Joshi A Dutt S Effect of zinc nanoparticles on antioxidative system of potato plants J Environ Biol 2017 38 3 435 439 10.22438/jeb Rani N Kumari K Sangwan P Barala P Yadav J Hooda V Nano-iron and nano-zinc induced growth and metabolic changes in Vigna radiata Sustainability 2022 14 14 8251 10.3390/su14148251 Raymond J Rakariyatham N Azanza JL Purification and some properties of polyphenoloxidase from sunflower seeds Phytochemistry 1993 34 4 927 931 10.1016/S0031-9422(00)90689-7 Remans T Opdenakker K Guisez Y Carleer R Schat H Vangronsveld J Cuypers A Exposure of Arabidopsis thaliana to excess Zn reveals a Zn-specific oxidative stress signature Environ Exp Bot 2012 84 61 71 10.1016/j.envexpbot.2012.05.005 Rizwan M Ali S Ali B Adrees M Arshad M Hussain A Zia ur Rehman M Waris AA Zinc and iron oxide nanoparticles improved the plant growth and reduced the oxidative stress and cadmium concentration in wheat Chemosphere 2019 214 269 277 10.1016/j.chemosphere.2018.09.120 30265934 Rudani L Vishal P Kalavati P The importance of zinc in plant growth-A review Int Res J Nat Sci 2018 5 2 38 48 Sabir S Zahoor MA Waseem M Siddique MH Shafique M Imran M Muzammil S Biosynthesis of ZnO nanoparticles using bacillus subtilis: characterization and nutritive significance for promoting plant growth in Zea mays L Dose-Response 2020 18 3 1559325820958911 10.1177/1559325820958911 32973419 Salah SM Yajing G Dongdong C Jie L Aamir N Qijuan H Jin H Seed priming with polyethylene glycol regulating the physiological and molecular mechanism in rice (Oryza sativa L.) under nano-ZnO stress Sci Rep 2015 5 1 1 14 10.1038/srep14278 Saliani M Jalal R Goharshadi EK Mechanism of oxidative stress involved in the toxicity of ZnO nanoparticles against eukaryotic cells Nanomed J 2016 3 1 1 14 10.7508/nmj.2016.01.001 Salinitro M van der Ent A Tognacchini A Tassoni A Stress responses and nickel and zinc accumulation in different accessions of Stellaria media (L.) Vill. in response to solution pH variation in hydroponic culture Plant Physiol Biochem 2020 148 133 141 10.1016/j.plaphy.2020.01.012 31958680 Sattar A Wang X Ul-Allah S Sher A Ijaz M Irfan M Skalicky M Foliar application of zinc improves morpho-physiological and antioxidant defense mechanisms, and agronomic grain biofortification of wheat (Triticum aestivum L.) under water stress Saudi J Biol Sci 2021 29 3 1699 1706 10.1016/j.sjbs.2021.10.061 35280547 Semida WM Abdelkhalik A Mohamed GF Abd El-Mageed TA Abd El-Mageed SA Rady MM Ali EF Foliar application of zinc oxide nanoparticles promotes drought stress tolerance in eggplant (Solanum melongena L.) Plants 2021 10 2 421 10.3390/plants10020421 33672429 Sheteiwy MS Fu Y Hu Q Nawaz A Guan Y Li Z Hu J Seed priming with polyethylene glycol induces antioxidative defense and metabolic regulation of rice under nano-ZnO stress Environ Sci Pollut Res 2016 23 19 19989 20002 10.1007/s11356-016-7170-7 Singh RD Kumbhakar P Pal S Khannam SK Kumbhakar P Investigating morphology-dependent antibacterial property of ZnO nanoparticles and developing an insight into oxidative stress generation and cellular response Biologia 2021 6 4 1339 1348 10.1007/s11756-021-00690-4 Singleton VL Rossi JA Colorimetry of total phenolics with phosphomolybdic-phosphotungstic acid reagents Amer J Enol Viticult 1965 16 3 144 158 Srivastav A Ganjewala D Singhal RK Rajput VD Minkina T Voloshina M Shrivastava M Effect of ZnO nanoparticles on growth and biochemical responses of wheat and maize Plants 2021 10 12 2556 10.3390/plants10122556 34961025 Subbaiah LV Prasad TNV Krishna TG Sudhakar P Reddy BR Pradeep T Novel effects of nanoparticulate delivery of zinc on growth, productivity, and zinc biofortification in maize (Zea mays L.) J Agric Food Chem 2016 64 19 3778 3788 10.1021/acs.jafc.6b00838 27089102 Sultana S Naser H Shil NC Akhter S Begum RA Effect of foliar application of zinc on yield of wheat grown by avoiding irrigation at different growth stages Bangladesh J Agric Res 2016 41 2 323 334 10.23910/IJBSM/2016.7.5.1645b Tabrez S Khan AU Hoque M Suhail M Khan MI Zughaibi TA Biosynthesis of ZnO NPs from pumpkin seeds’ extract and elucidation of its anticancer potential against breast cancer Nanotechnol Rev 2022 11 1 2714 2725 10.1515/ntrev-2022-0154 Thounaojam TC Panda P Mazumdar P Kumar D Sharma GD Sahoo L Sanjib P Excess copper induced oxidative stress and response of antioxidants in rice Plant Physiol Biochem 2012 53 33 39 10.1016/j.plaphy.2012.01.006 22306354 Tymoszuk A Wojnarowicz J Zinc oxide and zinc oxide nanoparticles impact on in vitro germination and seedling growth in Allium cepa L Materials 2020 13 12 2784 10.3390/ma13122784 32575606 Velikova V Yordanov I Edreva A Oxidative stress and some antioxidant systems in acid rain-treated bean plants Plant Science 2000 151 1 59 66 10.1016/S0168-9452(99)00197-1 Venkatachalam P Jayaraj M Manikandan R Geetha N Rene ER Sharma NC Sahi SV Zinc oxide nanoparticles (ZnONPs) alleviate heavy metal-induced toxicity in Leucaena leucocephala seedlings: a physiochemical analysis Plant Physiol Biochem 2017 110 59 69 10.1016/j.plaphy.2016.08.022 27622846 Wang XP Li QQ Pei ZM Wang SC Effects of zinc oxide nanoparticles on the growth, photosynthetic traits, and antioxidative enzymes in tomato plants Biol Plant 2018 62 4 801 808 10.1007/s10535-018-0813-4 Wang X Yang X Chen S Li Q Wang W Hou C Wang S Zinc oxide nanoparticles affect biomass accumulation and photosynthesis in Arabidopsis Front Plan Sci 2016 6 1243 10.3389/fpls.2015.01243 Wang Y Meng Y Ma Y Liu L Wu D Shu X Lai Q Combination of high Zn density and low phytic acid for improving Zn bioavailability in Rice (Oryza stavia L.) grain Rice 2021 14 1 1 12 10.1186/s12284-021-00465-0 33409869 Welch RM Graham RD Breeding for micronutrients in staple food crops from a human nutrition perspective J Exp Bot 2004 55 396 353 364 10.1093/jxb/erh064 14739261 Wu S Hu C Tan Q Li L Shi K Zheng Y Sun X Drought stress tolerance mediated by zinc-induced antioxidative defense and osmotic adjustment in cotton (Gossypium hirsutum) Acta Physiol Plant 2015 37 8 1 9 10.1007/s11738-015-1919-3 Xu Z Rothstein SJ ROS-Induced anthocyanin production provides feedback protection by scavenging ROS and maintaining photosynthetic capacity in Arabidopsis Plant Signal Behav 2018 13 3 e1451708 10.1080/15592324.2018.1451708 29533127 Zafar H Ali A Ali JS Haq IU Zia M Effect of ZnO nanoparticles on Brassica nigra seedlings and stem explants: growth dynamics and antioxidative response Front Plant Sci 2016 7 535 10.3389/fpls.2016.00535 27148347 Zhang W Ebbs SD Musante C White JC Gao C Ma X Uptake and accumulation of bulk and nanosized cerium oxide particles and ionic cerium by radish (Raphanus sativus L.) J Agric Food Chem 2015 63 2 382 390 10.1021/jf5052442 25531028 Zhao FJ McGrath SP Biofortification and phytoremediation Curr Opin Plant Biol 2009 12 3 373 380 10.3389/fpls.2015.00136 19473871

0

PMC9748875

NO-CC CODE

2022-12-15 23:22:47

no

Biologia (Bratisl). 2022 Dec 14;:1-10

utf-8

Biologia (Bratisl)

10.1007/s11756-022-01269-3

oa_other

==== Front Biologia (Bratisl) Biologia (Bratisl) Biologia 0006-3088 1336-9563 Springer International Publishing Cham 1269 10.1007/s11756-022-01269-3 Original Article Effect of zinc nanoparticles on the growth and biofortification capability of mungbean (Vigna radiata) seedlings Sorahinobar Mona m.sorahinobar@alzahra.ac.ir 1 Deldari Tooba 1 Nazem Bokaeei Zahra 1 Mehdinia Ali 2 1 grid.411354.6 0000 0001 0097 6984 Department of Plant Sciences, Faculty of Biological Sciences, Alzahra University, Tehran, Iran 2 Iranian National Institutes for Oceanography and Atmospheric Science, Tehran, Iran 14 12 2022 110 4 6 2022 8 11 2022 © The Author(s), under exclusive licence to Plant Science and Biodiversity Centre, Slovak Academy of Sciences (SAS), Institute of Zoology, Slovak Academy of Sciences (SAS), Institute of Molecular Biology, Slovak Academy of Sciences (SAS) 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Zinc insufficiency is a nutritional trouble worldwide, especially in developing countries. In the current study, an experiment was conducted to evaluate the effect of supplementation of MS media culture with different concentrations of ZnO nanoparticles (NPs) (0, 10, 20, 40, 80, and 160 ppm) on growth, nutrient uptake, and some physiological parameters of 7-days-old mung bean seedlings. ZnO NPs enhanced the Zn concentration of mung bean from 106.41 in control to more than 4600 µg/g dry weight in 80 and 160 ppm ZnO NPs treated seedlings. Our results showed that ZnO NPs in the concentration range from 10 to 20 ppm had a positive influence on growth parameters and photosynthetic pigments. Higher levels of ZnO NPs negatively affected seedling’s growth by triggering oxidative stress which in turn caused enhancing antioxidative response in seedlings including polyphenol oxidase and peroxidase activity as well as phenolic compounds and anthocyanine contents. Considering the positive effects of ZnO NPs treatment on mungbean seedlings growth, micronutrents, protein and shoot phenolics content, 20 ppm is recommended as the optimal concentration for biofortification. Our findings confirm the capability of ZnO NPs in the remarkable increase of Zn content of mungbean seedlings which can be an efficient way for plant biofortification and dealing with environmental stress. Supplementary information The online version contains supplementary material available at 10.1007/s11756-022-01269-3. Keywords Antioxidative response Biofortification Mung bean Nanoparticles Oxidative stress Phenolic compounds Photosynthetic pigments Zinc oxide ==== Body pmcIntroduction It is now apparent that a nutritional deficiency of zinc in humans is widespread, affecting nearly 2 billion individuals worldwide (Prasad 2008). Zinc has a critical role in homeostasis, immune system function, oxidative stress, apoptosis, aging, and significant disorders of great public health interest are associated with zinc deficiency. Reports confirm that dietary supplements of zinc are efficient in the control and treatment of covid 19 (reviewed by Giacalone et al. 2021). Insufficient intake of zinc from food is a major contributor to zinc deficiency in humans which somehow is related to the low Zn content of agricultural soil (Tabrez et al. 2022). Hence there is an urgent need to increase the zinc content and bioavailability in food especially in developing and poor countries (Welch and Graham 2004; Zhao and McGrath 2009). Crops biofortification offers a sustainable solution to reduce malnutrition and deal with human diseases. Use of nanotechnology to fortify the crops in human societies has received much attention in recent years (El-Ramady et al. 2021; Khan et al. 2021a, b). Crops nano-biofortification could be achieved by seed priming (Rizwan et al. 2019), soil and foliar application (Du et al. 2019; Semida et al. 2021), and cultivation of plants in media rich of candidate nutrient using nanomaterials. In this regard, various aspects should be considered including: (i) finding the appropriate type and shape of nanomaterials, (ii) determination of optimal dose (without negative effects on plant growth and physiology), and (iii) investigation of probable negative or positive effect on the absorption and accumulation of other nutrients. Mung bean seeds are used up by humans and its hay is used for animal feedstuff that its products can be back to humans through meat, milk, and dairy products. About 6 million-hectare area of the world is under mung bean cultivation (8.5% of the world’s pulse area). Biofortification of crops like mung bean (Vigna radiate (L.) R. Wilczek) which contains carbohydrates, proteins, calcium, β-carotene, iron, and a lot of micronutrients can be a suitable strategy, especially for poor communities to cope with a nutritional deficiency. Haider et al. (2021) and Rani et al. (2022) confirmed the capability of using ZnO NPs for the biofortification of mung beans in the field and hydroponic greenhouse environments. These works have focused on zinc biofortification and failed to address the effect of ZnO NPs treatments on other nutrients uptake. Additionally, more precise studies are required to clarify the effect of ZnO NPs on plant physiology and to find the optimal dose for zinc biofortification. Mung bean sprouts are now part of the diet of people in many countries as an appetizer and salad. One of the objectives of the current study was to evaluate the capability of the ZnO NPs in mung bean sprout biofortification, performance, and nutrient quality. Like humans, zinc is crucial for the enzyme functioning in plants and it is very important for plant growth and development. Zn deficiency in plants is one of the major concerns globally. In recent years, the West Asia region has been severely affected by the effects of climate change, such as successive droughts, the expansion of salinized land and dust storms, which have affected agricultural production. Many reports confirm Zn supplementation could alleviate the negative effects of abiotic stresses on plants (Khan et al. 2004; Hussein and Abou-Baker 2018; Thounaojam et al. 2012; Sattar et al. 2021). Micronutrients nanoparticles because of their small size and high surface-to-volume ratio are considered a good candidate for plant uptake and biofortification. The application of appropriate concentrations of zinc oxide nanoparticles (ZnO NPs) with positive effects on plant physiology can improve the growth and protection of different plant species (Mukherjee et al. 2016; Subbaiah et al. 2016; Venkatachalam et al. 2017). A high concentration of ZnO NPs can cause phytotoxicity in plants (Salah et al. 2015; Wang et al. 2016) and produce oxidative stress, which is common in plants in response to environmental stresses. To cope with oxidative stress, plants increase their enzymatic and non-enzymatic anti-oxidants content which helps hinder reactive oxygen species. Some of those non-enzymatic antioxidants are secondary metabolites like phenolic compounds which in turn have nutritional value for humans as antioxidants. Zinc NPs could be a potential high-performed fertilizer for enhancing plant yield and quality (Du et al. 2019). Accordingly, another aspect of the present study was to investigate the effect of ZnO NPs on the growth and physiology of mung bean seedlings. In the current study, we investigated the effects of different concentrations of ZnO NPs on seedling growth, biochemistry, and physiological response of V. radiate. Materials and methods Plant material and culture conditions Seeds of mung bean (V. radiata) were surface sterilized in 1% (v/v) sodium hypochlorite solution for 15 min, followed by three washes with sterile distilled water. Ten sterilized seeds were placed into Petri dishes containing 10 ml of MS basal medium (Murashige and Skoog 1962), 7% agarose, pH = 5.7 and 0, 10, 20, 40, 80, and 160 ppm of ZnO NPs (10–30 nm, nearly spherical, US-NANO). The stock water suspension of ZnO NPs, before application in media culture, was placed for 30 min in the Elmasonic S80(H) Ultrasonic Cleaner (37 kHz of ultrasonic frequency, of 150 W of effective ultrasonic power) (Elma Schmidbauer GmbH, Singen, Germany) to achieve better dispersion of particles. Petri dishes were placed in a tissue culture room at a temperature of 25 °C, relative humidity of 55%, and a 16-h photoperiod with a light intensity of 2300 lx. Seedlings obtained after 7 days of growth were used for experiments. The biometrical data of 7-days-old seedlings were collected as mean values of the shoot and root fresh and dry weight (mg) and length (cm). For physiological and biochemical assays 7 days old seedlings were frozen in liquid nitrogen and kept at -70 °C and used for the experiments. Multielement analysis 0.5 g of powdered freeze-dried mung bean shoot samples was placed in polyethylene tube with 8 ml of nitric acid (65%) for 12 h at room temperature. Then 2 ml perchloric acid was added to the samples and kept at 80 °C for 1 h and 150 °C for 3 h. The solution was filtered using Whatman No. 42 filter paper and diluted to 25 mL with ultrapure water and stored in the dark before analysis. ICP-MS (HP-4500, USA) equipped with an Asx-520 auto-sampler was used to define elements concentration. Pigment contents The total contents of carotenoids, chlorophylls a and b were measured according to the procedure described by Linchenthaler and Wellburn (1983). Briefly, fresh samples (0.2 g) were homogenized in 80% acetone (Merck, Germany) and the absorbance was recorded spectrophotometrically at 470, 646, and 663 nm. The results were calculated based on the following equations: Chla(μg/ml)=12.26A663-2.79A646Chlb(μg/ml)=21.50A646-5.10A663ChlTotal(μg/ml)=Chla+ChlbCarotenoids(μg/ml)=(1000A470-3.27Chla-104Chlb)/229 Anthocyanin content was extracted using 1% HCl v/v acidified methanol. Fresh samples were homogenized in the extraction solution, centrifuged at 18 000 × g at 4 °C for 15 min, and stored in darkness for 5 h at 5 °C. The amount of anthocyanin was quantified at 550 nm spectrophotometrically (Abdel Latef et al. 2020). Malondialdehyde and hydrogen peroxide content Malondialdehyde (MDA) content was determined based on Heath and Packer (1968). Fresh tissues were extracted with 0.1% TCA, and 0.5 ml of the supernatant was mixed with 1 ml of thiobarbituric acid (0.5%) in TCA (20%). After heating the solution at 95 °C for 25 min, the absorbance was read spectrophotometrically at 532 and 600 nm. The amount of MDA was calculated by an extinction coefficient of 155 mM− 1 cm− 1. For H2O2 quantification, the 0.5 ml of extract (0.1% TCA) was mixed with 0.5 ml potassium phosphate buffer (10 mM, pH 7.0) and 1 ml KI (1 M). The absorbance was measured at 390 nm based on the Velikova et al. (2000) method. Protein content and antioxidant enzyme activity 0.2 g of fresh samples was ground in 1 M Tris-HCl (pH 6.8) at 4 °C. The supernatant was separated at 12 000 × g centrifugation for 15 min and used for protein and enzyme assays. Protein content was quantified with bovine serum albumin as the standard (Bradford 1976). Antioxidant enzyme assay For measurement of peroxidase (POX) activity, 50 µl of the extract was mixed with 0.1 ml benzidine, 0.2 ml H2O2 (3%), and 0.2 M acetate buffer (pH 4.8), and the activity was defined at 530 nm (Abeles and Biles 1991). Polyphenol oxidase (PPO; E.C. 1.14.18.1) activity was measured based on the method described by Raymond et al. (1993). The reaction mixture contained 0.2 ml of pyrogallol (20 mM), 2.5 ml of 200 mM sodium phosphate buffer (pH 6.8), and 50 µl of enzyme extract. The enzyme activity was recorded at 430 nm. Total phenolic compounds content For extraction, 0.4 g of dried powder of samples was placed in 80% (v/v) methanol for 48 h and then was put in an ultrasonic bath for 20 min. After centrifugation (10 min at 11 000 × g), the supernatant was utilized for the detection of total phenolic compounds. For determination of total phenolic content, 100 µl of the extract was mixed with 500 µl of Folin-Ciocalteu reagent and 400 µl of sodium carbonate. Then, the reaction solution was kept for 30 min at room temperature. The absorbance was read spectrophotometrically at 630 nm (Singleton and Rossi 1965). Statistical analyses The experiment was set up in a completely randomized design. For experimental treatment, 10 repetitions were used with 10 seeds each (1 Petri dish). For biometrical analysis at least 20 plants were assayed in each treatment. For physiological and biochemical assays 25 seedlings were pooled and used as one replication of triplicate. Each data point was an average of three or five replicates. Presented data were analyzed by one way ANOVA (analysis of variance) using SPSS (version 21). The significance of differences was determined according to Duncan’s multiple range tests at the 0.05 level of probability. The graphs were designed with Graphpad Prism version 6 and Microsoft Excel version 2010. Principal component analysis (PCA) and Pearson correlation test were conducted using publicly available Past3.16 software. Results Zn content significantly increased with the increase of ZnO NPs concentration in media culture, on average, from 106.41 in control to more than 4 600 µg/g dry weight in 80 and 160 ppm ZnO NPs treated seedlings. No significant difference in Zn content was observed in the samples treated with 10 and those treated with 20 ppm ZnO NPs. Interestingly, with the exception of Zinc, the lowest concentrations of the all assayed elements were observed in 10 ppm ZnO NPs treated samples. On the other hand, the highest levels of calcium, magnesium, phosphorus, manganese, iron, cobalt, copper, and molybdenum were observed in the 40 ppm ZnO NPs treated samples (Table 1). Although none of the evaluated elements showed a positive or negative correlation with zinc, a strong correlation (≥ 90%) between K with Mg and P, Mg with P, Fe and Cu, P with K, Mn with Fe and Co, Fe with P, and Mn and Co were observed (Supplementary Fig. 1). Table 1 Nutrient content (µg/g dry weight) in shoot of mung bean (V. radiata) seedlings subjected to 0 (control), 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles Control 10 20 40 80 160 K 30504.01 ± 110.12e 26343.34 ± 80.39f 36993.42 ± 98.70a 36682.3 ± 160.04b 34903.01 ± 140.23c 33015.55 ± 90.04d P 7776.04 ± 38.39e 6253.08 ± 42.45f 9463.62 ± 56.14c 10912.7 ± 105.14a 10596.2 ± 89.19b 9364.67 ± 43.12d Ca 2899.2 ± 50.10d 2365.56 ± 25.09e 2966.73 ± 60.15c 3821.63 ± 34.56a 3645.18 ± 17.20b 3807.94 ± 80.78a Mg 1754.26 ± 40.04d 1353.49 ± 22.15e 1840.01 ± 13.18c 2075.26 ± 35.19a 1940.8 ± 28.40b 1815.57 ± 43.89c Na 792.88 ± 20.40d 686.38 ± 12.81e 1022.61 ± 30.08b 1014.43 ± 14.20b 926.45 ± 16.04c 1298.53 ± 22.31a Mn 430.56 ± 14.06d 250.55 ± 8.19f 390.77 ± 20.86c 573.72 ± 19.18a 501.23 ± 31.12b 312.95 ± 25.57e Fe 154.58 ± 10.63b 114.66 ± 19.02c 158.89 ± 21.74b 206.42 ± 13.94a 183.71 ± 8.11a 153.70 ± 9.12b Zn 106.41 ± 15.01e 815.66 ± 24.12d 867.34 ± 37.67d 2665.91 ± 84.88c 4745.59 ± 30.15a 4644 ± 50.04b Cu 33.91 ± 4.10a 7.52 ± 1.03c 29.91 ± 6.01b 34.49 ± 2.19a 30.84 ± 4.90a 26.66 ± 3.02b Mo 57.83 ± 11.02c 30.24 ± 4.20e 34.88 ± 5.09e 182.45 ± 8.15a 129.75 ± 7.50b 44.36 ± 3.50d Co 0.35 ± 0.01d 0.26 ± 0.01f 0.40 ± 0.09c 0.54 ± 0.08a 0.45 ± 0.05b 0.32 ± 0.03e Different letters indicate means that are significantly different at (P ≤ 0.05) Seedling growth ZnO NPs treatment induced changes in the growth biomarkers of V. radiata seedlings (Fig. 1). The use of ZnO NPs in the concentration range from 10 to 20 ppm had a positive influence on growth. The fresh and dry weight of shoot and root increased up to 20 ppm and then decreased at 40, 80, and 160 ppm. Shoot and root length also increased significantly following ZnO NPs treatment up to 20 ppm, while the length of seedlings was negatively affected at higher concentrations. In summary, the highest and lowest rate of seedling biomass including plant height as well as fresh and dry weight was observed in 20 ppm and 160 ZnO NPs treated samples respectively (Supplementary Fig. 2). Fig. 1 Relative changes in biomass from control of the 7-day-old mung bean (V. radiata) seedlings in response to 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles Photosynthetic pigments Chl a, Chl b, Chl T and carotenoids content significantly increased (P ≤ 0.05) up to 20 ppm ZnO NPs and then markedly reduced with the increase of ZnO concentrations in media culture (Fig. 2). Fig. 2 The content of photosynthetic pigments: chlorophyll a, chlorophyll b, total chlorophyll and carotenoids in shoots of 7 days old mungbean (V. radiata) seedlings cultured on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at (P ≤ 0.05) Protein content The shoot soluble protein content was observed to be the highest at 40 to160 ZnO NPs treated samples (Fig. 3). In the root, total soluble protein content significantly increased up to 40 ppm then significantly decreased under higher concentrations (80 and 160 ppm) of ZnO NPs (P ≤ 0.05). Fig. 3 The root and shoot total protein content of mungbean (V. radiata) seedlings grown on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at P ≤ 0.05 Lipid peroxidation and H2O2 content The content of MDA (as a bio-indicator of lipid peroxidation) and H2O2 (as the most stable ROS) in the shoot samples of ZnO treated samples were significantly higher than that of controls (Fig. 4). The highest content of H2O2 and MDA in shoot and root samples treated with ZnO NPs were observed at 160 and 40 ppm respectively. Root H2O2 content in response to ZnO NPs treatment was highly increased as compared to shoot. However, lipid peroxidation in shoot was higher which probably can be justified by higher anti-oxidants in root. Fig. 4 The content of A H2O2 and B malondialdhyde(MDA) in root and shoot of mungbean (V. radiata) seedlings grown on MS media containing various levels of zinc oxide nanaoparticles (0, 10, 20, 40, 80 and 160 ppm). Different letters indicate means that are significantly different at P ≤ 0.05 Enzymatic antioxidant assay ZnO NPs treatment induced the activities of POX and PPO of V. radiate in both shoot and root (Fig. 5). The results showed the activity of POX in the shoot was considerably lower than in root samples. The highest activity levels of POX were observed in the shoot and root samples collected from seedlings grown on 20 ppm and 160 ppm of ZnO NPs, respectively. The highest activity level of PPO activity in both root and shoot was observed at 160 ppm ZnO NPs treated samples. Fig. 5 The activities of A proxidase and B polyphenol oxidase enzymes in root and shoot of samples of mungbean seedlings grown on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at P ≤ 0.05 Non-enzymatic antioxidants The effect of ZnO nanoparticles on total phenolic compounds and anthocyanin was presented in Fig. 6. The treatment with ZnO NPs increased the total phenolic content of both shoot and root. The highest phenolic compounds induction was observed in shoot and root samples grown on 20 and 160 with the elevation rate of 207% and 160% percent as compared to control. Additionally, the highest level of the shoot and root anthocyanin content was observed at 160 and 80 ppm ZnNPs treated samples respectively (Fig. 6). No significant differences in anthocyanin content were observed in 20 to 160 and 40 to 160 ZnO NPs treated samples of shoot and root respectively. The principal component analysis (PCA) was performed on Z-scored transformed data of assayed data. PCA showed that 79% variation between samples could be explained by two (PC1 = 57.60 and PC2 = 21.98%) principal components (Fig. 4). The first component (PCA1) separates the 0, 10, and 20 ppm ZnO NPs treated samples from other species primarily based on and shoot zinc, calcium, H2O2, MDA, and chlorophyll content as well as root fresh weight (Fig. 7). The second component (PCA2) separated 40 and 60 ppm ZnO NPs treated samples from other species primarily based on cobalt, potassium, phenolic compounds contents as well as PPO and POX activities. These results confirmed that plant nutritional status is highly affected under ZnO NPs treatment followed by photosynthesis and antioxidative capability. Fig. 6 The content of A phenolic compounds and B anthocyanins in root and shoot of mungbean seedlings grown on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at P ≤ 0.05 Fig. 7 Principal Component Analysis (PCA) of the z-score transformed mean of all assayed parameters of mungbean seedlings grown on MS media containing 0 (Control), 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. First two axes shown are 57% and 21% of the variation in composition Discussion The increased zinc content of shoot in response to ZnO NPs supplementation confirmed that the application of zinc nanoparticles can be an effective way to increase the zinc content of mung bean seedlings. Considering that the recommended daily intake of zinc in the diet is between 8 and 12 mg, with the consumption of 2 g of 7 days old mung bean seedlings, germinated and grown on MS medium containing 80 ppm of zinc oxide nanoparticles, this need can be met. Thus, according to the tremendous effect of ZnO NPs treatment on the zinc content of seedlings (about 45 times more than that of control), this method could lead to an augmentation that could meet the daily needs of adults. Although the lowest concentration of ZnO NPs used in this study caused a more than 7-fold increase in zinc content of seedlings shoot, however it had a negative effect on the content of other analyzed elements. This finding, along with its positive effect on plant biomass, shows that the positive effect of zinc is related to its own nutritional value, not its effect on the absorption of other elements (at least the elements that have been studied). Because at least a part of the nutrients required for the seedling’s shoot is supplied from the seed storage, it is expected that the content of elements in the shoot samples treated with ZnO NPs should not be less than that of the control samples. Therefore, it seems that the negative effect of zinc (at 10 ppm) on the content of the studied elements is related to its effect on the translocation of elements from seed to shoot not to the absorption of other elements from the medium culture. Generally, the solubility of Zn in the media culture and soil increases with decrease in pH (Salinitro et al., 2020). It seems that the low pH of MS media culture increased the bioavailability of Zn for plant uptake. In plants, the interaction between Zn and other plant nutrients do exist and both positive and negative interactions are reported (Prasad et al. 2016). Regarding the positive effect of zinc on plant growth, two points can be mentioned: Firstly, there are some reports that confirm the positive effect of zinc on tryptophan (as a precursor of auxin) and gibberellic acid content in plants which as plant growth regulators can promote plant growth (Mašev and Kutáček 1966; Wang et al. 2021). Secondly, our findings of increased leaves content of Chla, Chlb, and carotenoid (up to 20 ppm) are a good indicator of the positive impact of zinc on plant photosynthesis. In the current study seedlings’ growth (biomass) showed a strong zinc concentration-dependent trend. The positive impact of low concentration (10 and 20 ppm) and negative impact of high concentration (40 to 160 ppm) of ZnO NP media culture supplementation on seedling growth may be related to its nutritional value on one hand and toxicological effects on the other hand. Leaves photosynthetic pigments as well as root MDA and H2O2 content can be considered as bioindicators for the above-mentioned positive and negative effects. Zinc is an essential nutrient for plants and it has a role in protein synthesis and activity of several enzymes, as well as membrane integrity, metabolic reactions, water uptake, transport, and gene expressions (Rudani et al. 2018). Hence, an increase in Zn level (at least up to 715% which was observed in 20 ppm ZnO NP supplemented media) was beneficial for the plants, leading to growth improvement. The positive effect of Zn application on plant chlorophyll content and photosynthesis was reported in different plants such as Triticum aestivum L., Brassica juncea (L.) Czern., and Gossypium hirsutum L. (Wu et al. 2015; Khan et al. 2016; Sultana et al. 2016). Similar to our findings the positive impact of low (≥ 20 ppm) concentration of ZnO NPs on plant growth and protein content was reported in Zea mays L. (Sabir et al. 2020), and Lycopersicon esculentum Mill. (Faizan et al. 2018). The negative effect of a higher concentration of ZnO NPs on plant growth was reported in the other studies (Zhang et al. 2015; Javed et al. 2017; Tymoszuk and Wojnarowicz 2020; Rani et al. 2022). ZnO NP toxicity (at a concentration of 40 to 160 ppm) might be due to the perturbed homeostasis of Zn and its impacts on other elements’ homeostasis (Srivastav et al. 2021). A strong correlation (~ 98%) between shoot Zn and MDA is a signal of lipid peroxidation and probable membrane damage (Supplementary Fig. 1). The finding is an indication of the creation of oxidative stress due to excessive accumulation of zinc as reported in the other plants (Remans et al. 2012; Feigl et al. 2015). Researchers have discussed about the mechanisms of nanoparticle-induced oxidative stress including light activation of electron hole pairs, active electronic configurations and functional groups generation on the surface of nanoparticles, and active redox cycling on the surface of nanoparticle (reviewed by Saliani et al. 2016). There are evidences that confirm the smaller nanoparticles have higher density of surface defects and provide greater free electrons and holes which leads to the generation of ROS (Singh et al. 2021).To eliminate excess ROS, plants activate enzymatic and non-enzymatic antioxidant defense systems. In this regard, the activation of secondary plant metabolism and synthesis of phenolic compounds like anthocyanins is a crucial way to scavenge ROS (Sheteiwy et al. 2016; Mittler 2017; Xu and Rothstein 2018). In the current study root PPO activity and anthocyanin content showed a significant correlation with internal Zn content (Supplementary Fig. 1). It seems that this strategy could restrain ROS accumulation in the root. A similar response of increased phenolic content to ZnO NP treatment was reported in Brassica nigra (L.) K.Koch, potato, and Coriandrum sativum L. (Marichali et al. 2014; Zafar et al. 2016; Raigond et al. 2017). However, it seems that change in the activity of PPO and POX along with the increased content of phenolic compounds has not been enough to inhibit oxidative damage in the seedling shoot (Wang et al. 2018). This observation is probably due to the fact that the photosynthesis apparatus of plant leaves are among the main site of ROS generation (Khorobrykh et al. 2020). Comparing the content of assayed minerals in the samples, confirmed that the seedlings grown on the 40 ppm ZnO NPs, accumulated the highest amount of elements (a total of 58175.35 µg/g). Meanwhile, the highest seedling dry-weight biomass was found at 20 ppm ZnO NPs treatment. Considering the content of mineral nutrients in the total biomass produced in 7 day time period, it is determined that the highest mineral level is found in 20 ppm (2118.72) followed by 40 ppm (1602.73), 10 ppm (1324.52), 80 ppm (1308.87), control (1267.81), and 160 ppm (1021.70). These data confirm the advantage of samples grown on 20 ppm ZnO NPs over others. Conclusion Taken together our findings confirm the capability of ZnO NP in zinc biofrotification of mung bean seedlings. Despite the significantly increased content of many micro and macronutrients especially under 40 ppm ZnO NPs supplemented medium, its stressful effects were observed as a decrease in growth and photosynthetic pigments and an increase in the content of H2O2 and MDA. Considering the positive effects of ZnO NP on growth, photosynthetic pigments, protein content, activity, and content of enzymatic and non-enzymatic antioxidants in mung bean seedlings, a concentration of 20 ppm is recommended as the optimal concentration. Supplementary information Below is the link to the electronic supplementary material.Supplementary Fig. 1 Pearson correlation analysis of all assayed variables in mungean seedlings germinated and grown on MS media supplemented with 0, 10, 20, 40, 80, and 160 ppm ZnO NPs. (JPG 4.13 MB) Supplementary Fig. 2 Growth biomarkers of mungbean seedlings grown on MS media containing 0, 10, 20, 40, 80 and 160 ppm zinc oxide nanaoparticles. Different letters indicate means that are significantly different at P ≤ 0.05. (JPG 300 KB) Acknowledgements The financial support was provided by Alzahra University. Author contributions Mona Sorahinobar and Zahra Nazem Bokee designed experiments with the assistance of Ali Mehdinia, Toba Deldari carried out experiments, Mona Sorahinobar analyzed experimental results and data and wrote the manuscript. Declarations Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Ethical approval This article does not contain any studies with human participants or animals performed by any of the authors. Consent for publication Not applicable. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Abdel Latef AAH Dawood MF Hassanpour H Rezayian M Younes NA Impact of the static magnetic field on growth, pigments, osmolytes, nitric oxide, hydrogen sulfide, phenylalanine ammonia-lyase activity, antioxidant defense system, and yield in lettuce Biology 2020 9 7 172 10.3390/biology9070172 32709036 Abeles FB Biles CL Characterization of Peroxidases in Lignifying Peach Fruit Endocarp Plant Physiol 1991 95 1 269 273 10.1104/pp.95.1.269 16667963 Bradford MM A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding Anal Biochem 1976 72 1-2 248 254 10.1016/0003-2697(76)90527-3 942051 Du W Yang J Peng Q Liang X Mao H Comparison study of zinc nanoparticles and zinc sulphate on wheat growth: from toxicity and zinc biofortification Chemosphere 2019 227 109e116 10.1016/j.chemosphere.2019.03.168 30986592 El-Ramady H Abdalla N Elbasiouny H Elbehiry F Elsakhawy T Omara AED Nano-biofortification of different crops to immune against COVID-19: A review Ecotoxicol Environ Saf 2021 222 112500 10.1016/j.ecoenv.2021.112500 34274837 Faizan M Faraz A Yusuf M Khan ST Hayat S Zinc oxide nanoparticle-mediated changes in photosynthetic efficiency and antioxidant system of tomato plants Photosynthetica 2018 56 2 678 686 10.1007/s11099-017-0717-0 Feigl G Lehotai N Molnár A Ördög A Rodríguez-Ruiz M Palma JM Kolbert Z Zinc induces distinct changes in the metabolism of reactive oxygen and nitrogen species (ROS and RNS) in the roots of two Brassica species with different sensitivity to zinc stress Ann Bot 2015 116 4 613 625 10.1093/aob/mcu246 25538112 Giacalone A Marin L Febbi M Tovani-Palone MR Current evidence on vitamin C, D, and zinc supplementation for COVID-19 prevention and/or treatment Electron J Gen Med 2021 18 5 em311 10.29333/ejgm/11099 Haider MU Hussain M Farooq M Ul-Allah S Ansari MJ Alwahibi MS Farooq S Zinc biofortification potential of diverse mungbean [V. radiata (L.) Wilczek] genotypes under field conditions PLoS ONE 2021 16 6 e0253085 10.1371/journal.pone.0253085 34161364 Heath RL Packer L Photoperoxidation in isolated chloroplasts Arch Biochem Biophys 1968 125 1 189 198 10.1016/0003-9861(68)90654-1 5655425 Hussein MM Abou-Baker NH The contribution of nano-zinc to alleviate salinity stress on cotton plants R Soc Open Sci 2018 5 8 171809 10.1098/rsos.171809 30224982 Javed R Usman M Yücesan B Zia M Gürel E Effect of zinc oxide (ZnO) nanoparticles on physiology and steviol glycosides production in micropropagated shoots of Stevia rebaudiana Bertoni Plant Physiol Biochem 2017 110 94 99 10.1016/j.plaphy.2016.05.032 27246994 Khan MK Pandey A Hamurcu M Gezgin S Athar T Rajput VD Gupta OP Minkina Insight into the prospects for nanotechnology in wheat biofortification Biology 2021 10 11 1123 10.3390/biology10111123 34827116 Khan ST, Malik A, Alwarthan A, Shaik MR (2021b) The enormity of the zinc deficiency problem and available solutions; an overview. Arab J Chem 103668. 10.1016/j.arabjc.2021.103668 Khan R Gul S Hamayun M Shah M Sayyed A Ismail H Gul H Effect of foliar application of zinc and manganese on growth and some biochemical constituents of Brassica junceae grown under water stress J Agric Environ Sci 2016 16 984 997 10.5829/idosi.aejaes.2016.16.5.12923 Khan MU Qasim M Jamil M Effect of Zn on starch content of paddy and zinc content of soil, leaf and root of rice grown in calcareous soils Int J Agric Biol 2004 6 6 1132 1135 Khorobrykh S Havurinne V Mattila H Tyystjärvi E Oxygen and ROS in photosynthesis Plants 2020 9 1 91 10.3390/plants9010091 31936893 Linchenthaler HK Wellburn AR Determination of total carotenoides and clorophyll a and b of leaf extracts in diferent solventes Biochem Soc Transac 1983 11 1591 1592 10.1042/bst0110591 Mašev N Kutáček M The effect of zinc on the biosynthesis of tryptophan, andol auxins and gibberellins in barley Biol Plant 1966 8 2 142 151 10.1007/BF02930623 Marichali A Dallali S Ouerghemmi S Sebei H Hosni K Germination, morpho-physiological and biochemical responses of coriander (Coriandrum sativum L.) to zinc excess Ind Crops Prod 2014 55 248 257 10.1016/j.indcrop.2014.02.033 Mittler R ROS are good Trends Plant Sci 2017 22 1 11 19 10.1016/j.tplants.2016.08.002 27666517 Mukherjee A Sun Y Morelius E Tamez C Bandyopadhyay S Niu G Gardea-Torresdey JL Differential toxicity of bare and hybrid ZnO nanoparticles in green pea (Pisum sativum L.): a life cycle study Front Plant Sci 2016 6 1242 10.3389/fpls.2015.01242 26793219 Murashige T Skoog F A revised medium for rapid growth and bio assays with tobacco tissue cultures Physiologia Plantarum 1962 15 3 473 497 10.1111/j.1399-3054.1962.tb08052.x Prasad AS Zinc in human health: effect of zinc on immune cells Mol Med 2008 14 5 353 357 10.2119/2008-00033.Prasad 18385818 Prasad R Shivay YS Kumar D Interactions of zinc with other nutrients in soils and plants-A Review Ind J Fertilisers 2016 12 5 16 26 Raigond P Raigond B Kaundal B Singh B Joshi A Dutt S Effect of zinc nanoparticles on antioxidative system of potato plants J Environ Biol 2017 38 3 435 439 10.22438/jeb Rani N Kumari K Sangwan P Barala P Yadav J Hooda V Nano-iron and nano-zinc induced growth and metabolic changes in Vigna radiata Sustainability 2022 14 14 8251 10.3390/su14148251 Raymond J Rakariyatham N Azanza JL Purification and some properties of polyphenoloxidase from sunflower seeds Phytochemistry 1993 34 4 927 931 10.1016/S0031-9422(00)90689-7 Remans T Opdenakker K Guisez Y Carleer R Schat H Vangronsveld J Cuypers A Exposure of Arabidopsis thaliana to excess Zn reveals a Zn-specific oxidative stress signature Environ Exp Bot 2012 84 61 71 10.1016/j.envexpbot.2012.05.005 Rizwan M Ali S Ali B Adrees M Arshad M Hussain A Zia ur Rehman M Waris AA Zinc and iron oxide nanoparticles improved the plant growth and reduced the oxidative stress and cadmium concentration in wheat Chemosphere 2019 214 269 277 10.1016/j.chemosphere.2018.09.120 30265934 Rudani L Vishal P Kalavati P The importance of zinc in plant growth-A review Int Res J Nat Sci 2018 5 2 38 48 Sabir S Zahoor MA Waseem M Siddique MH Shafique M Imran M Muzammil S Biosynthesis of ZnO nanoparticles using bacillus subtilis: characterization and nutritive significance for promoting plant growth in Zea mays L Dose-Response 2020 18 3 1559325820958911 10.1177/1559325820958911 32973419 Salah SM Yajing G Dongdong C Jie L Aamir N Qijuan H Jin H Seed priming with polyethylene glycol regulating the physiological and molecular mechanism in rice (Oryza sativa L.) under nano-ZnO stress Sci Rep 2015 5 1 1 14 10.1038/srep14278 Saliani M Jalal R Goharshadi EK Mechanism of oxidative stress involved in the toxicity of ZnO nanoparticles against eukaryotic cells Nanomed J 2016 3 1 1 14 10.7508/nmj.2016.01.001 Salinitro M van der Ent A Tognacchini A Tassoni A Stress responses and nickel and zinc accumulation in different accessions of Stellaria media (L.) Vill. in response to solution pH variation in hydroponic culture Plant Physiol Biochem 2020 148 133 141 10.1016/j.plaphy.2020.01.012 31958680 Sattar A Wang X Ul-Allah S Sher A Ijaz M Irfan M Skalicky M Foliar application of zinc improves morpho-physiological and antioxidant defense mechanisms, and agronomic grain biofortification of wheat (Triticum aestivum L.) under water stress Saudi J Biol Sci 2021 29 3 1699 1706 10.1016/j.sjbs.2021.10.061 35280547 Semida WM Abdelkhalik A Mohamed GF Abd El-Mageed TA Abd El-Mageed SA Rady MM Ali EF Foliar application of zinc oxide nanoparticles promotes drought stress tolerance in eggplant (Solanum melongena L.) Plants 2021 10 2 421 10.3390/plants10020421 33672429 Sheteiwy MS Fu Y Hu Q Nawaz A Guan Y Li Z Hu J Seed priming with polyethylene glycol induces antioxidative defense and metabolic regulation of rice under nano-ZnO stress Environ Sci Pollut Res 2016 23 19 19989 20002 10.1007/s11356-016-7170-7 Singh RD Kumbhakar P Pal S Khannam SK Kumbhakar P Investigating morphology-dependent antibacterial property of ZnO nanoparticles and developing an insight into oxidative stress generation and cellular response Biologia 2021 6 4 1339 1348 10.1007/s11756-021-00690-4 Singleton VL Rossi JA Colorimetry of total phenolics with phosphomolybdic-phosphotungstic acid reagents Amer J Enol Viticult 1965 16 3 144 158 Srivastav A Ganjewala D Singhal RK Rajput VD Minkina T Voloshina M Shrivastava M Effect of ZnO nanoparticles on growth and biochemical responses of wheat and maize Plants 2021 10 12 2556 10.3390/plants10122556 34961025 Subbaiah LV Prasad TNV Krishna TG Sudhakar P Reddy BR Pradeep T Novel effects of nanoparticulate delivery of zinc on growth, productivity, and zinc biofortification in maize (Zea mays L.) J Agric Food Chem 2016 64 19 3778 3788 10.1021/acs.jafc.6b00838 27089102 Sultana S Naser H Shil NC Akhter S Begum RA Effect of foliar application of zinc on yield of wheat grown by avoiding irrigation at different growth stages Bangladesh J Agric Res 2016 41 2 323 334 10.23910/IJBSM/2016.7.5.1645b Tabrez S Khan AU Hoque M Suhail M Khan MI Zughaibi TA Biosynthesis of ZnO NPs from pumpkin seeds’ extract and elucidation of its anticancer potential against breast cancer Nanotechnol Rev 2022 11 1 2714 2725 10.1515/ntrev-2022-0154 Thounaojam TC Panda P Mazumdar P Kumar D Sharma GD Sahoo L Sanjib P Excess copper induced oxidative stress and response of antioxidants in rice Plant Physiol Biochem 2012 53 33 39 10.1016/j.plaphy.2012.01.006 22306354 Tymoszuk A Wojnarowicz J Zinc oxide and zinc oxide nanoparticles impact on in vitro germination and seedling growth in Allium cepa L Materials 2020 13 12 2784 10.3390/ma13122784 32575606 Velikova V Yordanov I Edreva A Oxidative stress and some antioxidant systems in acid rain-treated bean plants Plant Science 2000 151 1 59 66 10.1016/S0168-9452(99)00197-1 Venkatachalam P Jayaraj M Manikandan R Geetha N Rene ER Sharma NC Sahi SV Zinc oxide nanoparticles (ZnONPs) alleviate heavy metal-induced toxicity in Leucaena leucocephala seedlings: a physiochemical analysis Plant Physiol Biochem 2017 110 59 69 10.1016/j.plaphy.2016.08.022 27622846 Wang XP Li QQ Pei ZM Wang SC Effects of zinc oxide nanoparticles on the growth, photosynthetic traits, and antioxidative enzymes in tomato plants Biol Plant 2018 62 4 801 808 10.1007/s10535-018-0813-4 Wang X Yang X Chen S Li Q Wang W Hou C Wang S Zinc oxide nanoparticles affect biomass accumulation and photosynthesis in Arabidopsis Front Plan Sci 2016 6 1243 10.3389/fpls.2015.01243 Wang Y Meng Y Ma Y Liu L Wu D Shu X Lai Q Combination of high Zn density and low phytic acid for improving Zn bioavailability in Rice (Oryza stavia L.) grain Rice 2021 14 1 1 12 10.1186/s12284-021-00465-0 33409869 Welch RM Graham RD Breeding for micronutrients in staple food crops from a human nutrition perspective J Exp Bot 2004 55 396 353 364 10.1093/jxb/erh064 14739261 Wu S Hu C Tan Q Li L Shi K Zheng Y Sun X Drought stress tolerance mediated by zinc-induced antioxidative defense and osmotic adjustment in cotton (Gossypium hirsutum) Acta Physiol Plant 2015 37 8 1 9 10.1007/s11738-015-1919-3 Xu Z Rothstein SJ ROS-Induced anthocyanin production provides feedback protection by scavenging ROS and maintaining photosynthetic capacity in Arabidopsis Plant Signal Behav 2018 13 3 e1451708 10.1080/15592324.2018.1451708 29533127 Zafar H Ali A Ali JS Haq IU Zia M Effect of ZnO nanoparticles on Brassica nigra seedlings and stem explants: growth dynamics and antioxidative response Front Plant Sci 2016 7 535 10.3389/fpls.2016.00535 27148347 Zhang W Ebbs SD Musante C White JC Gao C Ma X Uptake and accumulation of bulk and nanosized cerium oxide particles and ionic cerium by radish (Raphanus sativus L.) J Agric Food Chem 2015 63 2 382 390 10.1021/jf5052442 25531028 Zhao FJ McGrath SP Biofortification and phytoremediation Curr Opin Plant Biol 2009 12 3 373 380 10.3389/fpls.2015.00136 19473871

0

PMC9748876

NO-CC CODE

2022-12-15 23:22:47

no

TVZ verpleegkd prakt wet. 2022 Dec 14; 132(6):8-12

latin-1

oa_other

==== Front Curr Psychol Curr Psychol Current Psychology (New Brunswick, N.j.) 1046-1310 1936-4733 Springer US New York 4132 10.1007/s12144-022-04132-5 Article The dark triad and depressive symptoms among chinese adolescents: moderated mediation models of age and emotion regulation strategies http://orcid.org/0000-0003-3630-3460 Shen Ke 673472766@qq.com Henan Police College, 450001 Zhengzhou, China 14 12 2022 110 21 8 2022 20 11 2022 5 12 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Dark Triad has been found to relate with depressive symptoms, but the underlying mechanism was rarely investigated. In the present study, we examined the mediating effect of two emotion regulation strategies (cognitive reappraisal and expressive suppression) and the moderating effect of gender and age. 709 Chinese adolescents aged 12–18 years (M = 14.54, SD = 1.70;55.3% girls) filled out the Dirty Dozen, Emotion Regulation Questionnaire and the 10-item of Center for Epidemiological Studies Depression Scale. As expected, Dark Triad was related with increased levels of depression and emotion regulation strategies acted as mediators in these associations. Suppression mediated the relationship between Machiavellianism and depression. Reappraisal and suppression mediated the link between psychopathy and depression. Reappraisal mediated the association between narcissism and depression. Besides, age moderated the effect of Machiavellianism on reappraisal, suggesting Machiavellianism had a negative impact on reappraisal for younger adolescents, but not for older adolescents. Age also moderated the effect of reappraisal on depression, indicating the negative effect of reappraisal on depression was stronger for younger adolescents than for older adolescents. These results address how Dark Triad traits affect depression via emotion regulation strategies and indicate the effectiveness of Dark Triad and emotion regulation strategies may change across age groups. Keywords Dark triad Depression Emotion regulation strategies Cognitive reappraisal Expressive suppression Age ==== Body pmcIntroduction Depression is characterized by sustained negative affect, which not only has adverse effects on individual’s mental health and social functioning, but also results in huge economic costs for society (McTernan et al., 2013). Due to its wide and severe influences, depression has become a global public health issue. From the view of personality psychology, individuals with depressive symptoms seem to exhibit certain personality characteristics. The relation of personality and mental disorders has been discussed for a long time (Krueger & Eaton, 2010), and Big Five traits (neuroticism, extraversion, conscientiousness, agreeableness, and openness to experience) is the most concentrated subject. In a meta-analysis from 175 empirical studies, patients with depression report higher levels of neuroticism, and lower levels of extraversion and conscientiousness than health people (Kotov et al., 2010). Besides Big Five traits, more and more researchers started to explore the relationship between depression and the “dark side” of personality, Dark Triad traits. The Dark Triad is a group of socially malevolent personality traits, which includes three separate but related subclinical traits—Machiavellianism, psychopathy, and narcissism (Paulhus & Williams, 2002). Although these traits share the tendencies toward dishonesty, coldness, and selfishness, they still are different concept. Machiavellianism is characterized by apathy, utilitarianism, cheat and manipulation. Psychopathy refers to high impulsivity, thrill-seeking and low empathy. Narcissism is identified as grandiosity, dominance, and superiority (Paulhus &Williams, 2002). Based on life history theory, individuals high in Dark Triad traits apply a fast life strategy exhibited by short-term mating, risk-taking tendency and indifference to social morality (Jonason et al., 2010), in order to gain more reproductive and living resources. Thus, Dark Triad often correlated with aggressive, unethical and antisocial behavior (Sijtsema et al., 2019), and negative mental health outcomes such as depressive symptoms (Gómez-Leal et al., 2019). Quiet a few of studies have examined the association between Dark Triad and depression among adults (e.g., Bonfá-Araujo et al., 2021; Gómez-Leal et al., 2019; Shih et al., 2019). Except for several evidences (Hansen et al., 2013; Willemsen et al., 2011), the majority of researchers observed that Machiavellian and psychopathic individuals appeared to exhibit higher levels of depression (Gómez-Leal et al., 2019; Gogola et al., 2021; Lyons et al., 2019; Mojsa-Kaja et al., 2021; Shih et al., 2019). As for narcissism, many studies found narcissism seem a “bright” trait of the Dark Triad, and correlated with less signs of depression (Bonfá-Araujo et al., 2021; Lyons et al., 2019; Shih et al., 2019). Whereas, recent findings during the COVID-19 pandemic showed individuals high in narcissism reported increased depressive symptoms (Gogola et al., 2021; Mojsa-Kaja et al., 2021). Although the association between Dark Triad and depression has been investigated by many studies, participants in these studies mainly recruited from adults, less from other age groups, such as adolescents. Some research have found adolescents with Dark Triad traits could obtain high peer status (De Bruyn, & Cillessen, 2006). But high peer status could not protect them from experiencing emotional pain due to less secure relationships (Jonason et al ., 2010). Conversely, Dark Triad traits were likely to predict their bad mental health and adverse well-being outcomes, like depressive symptoms (Jonason et al., 2015). In the development of depression, adolescence is regarded as a crucial period (Nelemans et al., 2016). Intervention during this period can reduce the severity or persistence of depression disorders (De Girolamo et al., 2012). Thus, the association between Dark Triad and depression is need to be examined in adolescents, in order to figure out risking factors in the process of prevention and treatment of depression. Additionally, the underlying mechanism of the Dark Triad-depression association is still unclear. In particular, there is a lack of research to investigate factors moderating or mediating this association. And most researchers have examined the correlation of Dark Triad and depression in Western societies, few in Asian countries, especially in China. According to Lewis-Fernández and Kleinman (1994), culture had significant influences on the personality and psychopathology. The management of emotions was a major currency of relationships and a hallmark of health in Chinese communities. So, it is necessary to explore the Dark Triad-depression relationship and the underlying factors, such as emotion regulation, between them among Chinese adolescents. Emotion regulation, referring to the processes through which individuals influence their own emotional experiences and expressions using a variety of cognitive and behavior strategies (Gross, 1998). Emotion regulation deficits, particularly inappropriate habitual use of emotion regulation strategies, are considered as key components of depressive disorders (Liu & Thompson, 2017) and Dark Triad (Zeigler-Hill & Vonk, 2015). Based on the biopsychosocial model of health psychology (Sarafino & Smith, 2012), coping strategies can mediate the relationship between dispositional characteristics, such as Dark Triad traits, and health-related outcomes. In line with this model, a latest study demonstrated maladaptive cognitive emotion regulation strategies mediate the relationship between Dark Triad and stress, anxiety and depression experienced during the COVID-19 pandemic (Mojsa-Kaja et al., 2021). Hence, it is plausible to regard emotion regulation strategies as mediators between Dark Triad and depression. As the most widely studied emotion regulation strategies, cognitive reappraisal and expressive suppression have distinct effects on depression. Expressive suppression, referring to the inhibition of outward emotional expression to control emotional responses, was regarded as a maladaptive behavioral strategy resulting in adverse emotional, social and cognitive consequences (Gross, 2002; Gross & John, 2003; John & Gross, 2004) including increased depressive symptoms (Sai et al., 2016). In contract, cognitive reappraisal, defined as an attempt to change one’s subjective evaluation of the emotion-eliciting situation to alter its emotional impact, was considered as an adaptive cognitive strategy which is correlated with positive emotional experiences, good interpersonal functioning, and high well-being (Gross & John, 2003; John & Gross, 2004) and effectively relieved depressive symptoms (Sai et al., 2016). In addiction, Dark Triad traits have been shown to be correlated with cognitive reappraisal and expressive suppression. In a latest meta-analysis, psychopathy was significantly correlated with less use of reappraisal and more use of suppression (Walker et al., 2022). A recent work revealed that Machiavellians were habituated to more use of suppression, and narcissists were accustomed to less use of suppression (Akram & Stevenson, 2021). Thus, cognitive reappraisal and expressive suppression may serve as mediating roles in the relationship between Dark Triad and depression. Moreover, there are gender differences in Dark Triad traits, emotion regulation strategies and depressive symptoms. Men tend to report higher scores in Dark Triad traits (Muris et al., 2017), and prefer to use more expressive suppression than women (Gross & John, 2003). Girls reported to use more reappraisal and less suppression than boys in Chinese adolescents (Zhao et al., 2014). Yet, female adolescents often reported more symptoms of depression than males (Granrud et al., 2019; Shorey et al., 2021). This may be due to gender differences in the mediating effect of emotion regulation strategies. For example, in the relationship between trait forgiveness and depression, the mediating effect of cognitive reappraisal was significantly greater for girls than for boys (Zhang et al., 2020). So, the mediating effect of emotion regulation strategies between Dark Triad traits and depression are possible to be moderate by gender. Age differences also existed in Dark Triad, emotion regulation strategies and depression. With age increased, the level of Dark Triad traits have decreased markedly in Japanese samples, and especially in women (Kawamoto, Shimotsukasa, & Oshio, 2020). From childhood to adolescence, boys’ depression scores changed little, but girls reported a significantly increased level of depression (Twenge & Nolen-Hoeksema, 2002). Men turn to reappraisal from suppression with age, further enhancing positive mood, reducing negative mood and improving their mental health (Masumoto, Taishi, & Shiozaki, 2016), but women turn to suppression as they grow older (Nolen-Hoeksema & Aldao, 2011). Given that empirical research of the relations between Dark Triad and depression yielded mixed results, the current study aims to examine the relationship between Dark Triad traits and depression in a Chinese adolescent sample. In addition, considering about the close connections among Dark Triad, emotion regulation and depression, we investigate whether emotion regulation strategies act as mediators between Dark Triad traits and depression and whether gender and age play moderating roles in the mediation model. According to prior studies, We hypothesized that (1) Dark Triad traits will related with increased levels of depression; (2) Dark Triad traits could correlated with more depressive symptoms via less use of cognitive reappraisal and more use of expressive suppression; (3) gender and age will moderate the mediating effect of emotion regulation strategies between Dark Triad traits and depression. Method Participants 730 students were recruited from two middle and high schools in Henan province, China. After eliminating invalid and extreme data, the sample included 709 adolescents (55.3% girls) aged 12–18 years (M = 14.54, SD = 1.70). Students completed this survey during the normal class period, and they were informed that their responses were anonymous and confidential. Informed consents were obtained from students and their parents. Measures Dark triad The Dark Triad was measured using the Dirty Dozen (DD; Jonason & Webster, 2010). It is a 12- item scale with three subscales: Machiavellianism (4 items, e.g., “I tend to manipulate others to get my way”), psychopathy (4 items, e.g.,“I tend to lack remorse”) and narcissism (4 items, e.g.,“I tend to want others to admire me”). Participants were asked to rate statements on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree), and higher scores represented higher levels of Dark Triad traits. DD has been confirmed to have acceptable structural validity across eight world regions (Rogoza et al., 2020). In the present study, the Cronbach’s α of Machiavellianism, psychopathy and narcissism subscales were 0.83, 0.62 and 0.84 respectively. Emotion regulation strategies Emotion regulation strategies were assessed using Emotion Regulation Questionnaire (ERQ; Gross & John, 2003). It is a 10-item questionnaire with two subscales: cognitive reappraisal (6 items, e.g., “I control my emotions by changing the way I think about the situation I’m in”) and expressive suppression (4 items, e.g., “I control my emotions by not expressing them”). Participants were asked to rate items on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree), and higher scores represented more use of reappraisal or suppression. ERQ has great validity and reliability in Chinese adolescents and young adults (Wang et al., 2020). In the current sample, the Cronbach’s α of reappraisal and suppression subscales were 0.84 and 0.72 respectively. Depressive symptoms Depressive symptoms were measured using the 10-item of Center for Epidemiological Studies Depression Scale (CESD-10) (Andresen et al., 1994). Participants were asked to rate each statement during the past week on a 4-point Likert scale: 0 (rarely or none of the time, < 1 day); 1 (some or few of the time, 1–2 days); 2 (occasionally or a moderate amount of the time, 3–4 days); 3 (most or all of the time, 5–7 days). All items is calculated to yield a global score, and higher scores indicate more symptoms of depression. CESD-10 has been validated and widely used in the China Health and Retirement Longitudinal Study (Cheng et al., 2016). The Cronbach’s α of CESD-10 was 0.80 in the present sample. Data analyses Descriptive analysis and Pearson correlation analysis of all variables were conducted using SPSS 26.0. Parallel multiple mediation analyses were performed to examine the mediating effects of emotion regulation strategies between Dark Triad and depression by using the PROCESS macro for SPSS (Hayes, 2013). PROCESS model 4 was used to test three mediation models with an independent variable (Machiavellianism, psychopathy or narcissism), two mediators (cognitive reappraisal and expressive suppression) and a dependent variable (depression). Moderated mediation analyses were performed to examine the moderating effects of gender and age by using the PROCESS model 59. We set bootstrapping to 50,000 samples and a 95% confidence interval to determine statistical significance. If the confidence interval of indirect effect exclusive of zero, the mediating effect was considered as significant. According to age, the participants were divided into two groups: the younger group (age 12 to 14) and the older group (age 15 to 18). Results Common method deviation test The Harman single factor test was used to examine common method biases. We found 7 factors of which eigenvalues were greater than 1 without rotation, and the first factor could explain 18.84% of the variation, which indicated that the current study was not influenced by common method biases. Gender and age differences As showed in Table 1, boys reported significantly higher scores on all of Dark Triad traits and more use of expressive suppression than girls. In terms of age, older students showed a significantly higher level of Narcissism than younger students. Table 1 Gender and age differences tests of all variables (M ± SD). Variables Total Boys(306) Girls(403) t Younger(361) Older(348) t 1.Machiavellianism 1.72 ± 0.91 1.98 ± 1.04 1.53 ± 0.76 6.31* * * 1.73 ± 0.95 1.72 ± 0.88 0.22 2.Psychopathy 2.22 ± 1.04 2.42 ± 1.06 2.07 ± 1.00 4.51* * * 2.18 ± 1.05 2.26 ± 1.02 -0.98 3.Narcissism 3.31 ± 1.54 3.49 ± 1.52 3.18 ± 1.54 2.62* * 3.13 ± 1.50 3.51 ± 1.55 -3.34* * 4.Cognitive reappraisal 4.60 ± 1.16 4.62 ± 1.13 4.60 ± 1.19 0.24 4.69 ± 1.17 4.52 ± 1.15 1.93 5.Expressive suppression 3.99 ± 1.37 4.13 ± 1.31 3.89 ± 1.40 2.26* 4.08 ± 1.34 3.91 ± 1.39 1.63 6.Depression 1.01 ± 0.52 1.03 ± 0.53 0.99 ± 0.52 0.83 0.97 ± 0.56 1.04 ± 0.49 -1.72 Note: * p < .05, * * p < .01, * * * p < .001 Correlation analyses Person correlations (Table 2) suggested that Machiavellianism was positively related with expressive suppression and depression. Psychopathy were negatively related with cognitive reappraisal and positively related with expressive suppression and depression. Narcissism was negatively related with cognitive reappraisal and positively related with depression. As expected, cognitive reappraisal was negatively related with depression, and expressive suppression was positively related with depression. Table 2 Correlations between all variables Variables 1 2 3 4 5 6 1.Machiavellianism - 0.40** 0.38** -0.07 0.14** 0.26** 2.Psychopathy - 0.36** -0.20** 0.18** 0.30** 3.Narcissism - -0.09* 0.04 0.17** 4.Cognitive reappraisal - 0.07 -0.28** 5.Expressive suppression - 0.26** 6.Depression - Note: * p < .05, * * p < .01. Mediation analyses The direct and indirect effects of Dark Triad traits on depression through reappraisal and suppression were presented in Table 3. In model 1, the direct effect of Machiavellianism on depression (β = 0.22, p < .001) and the indirect effect via suppression (β = 0.04, 95% CI [0.02, 0.06]) was significant, indicating that suppression partially mediated the relationship between Machiavellianism and depression. Machiavellianism predicted higher depressive symptoms partially through more use of suppression. In model 2, the direct effect of psychopathy on depression (β = 0.22, p < .001) and the indirect effects via reappraisal (β = 0.05, 95% CI [0.03, 0.08]) and suppression (β = 0.04, 95% CI [0.02, 0.07]) was significant, suggesting that reappraisal and suppression partially mediated the relationship between psychopathy and depression. Psychopathy predicted more depressive symptoms partially through less use of reappraisal and more use of suppression. In model 3, the direct effect of narcissism on depression (β = 0.12, p < .01) and the indirect effects via reappraisal (β = 0.02, 95% CI [0.00, 0.05]) was significant, showing that reappraisal partially mediated the relationship between narcissism and depression. Narcissism predicted more depressive symptoms partially through less use of reappraisal. Table 3 Effects of Dark Triad traits on depression via emotion regulation strategies Model Variables Effect of IV on M Effect of M on DV Total effect Direct effect Indirect effect 95% CI Model1 IV:Mac DV:Dep 0.27*** 0.22*** M1:CR -0.08 -0.27*** 0.02 [-0.00, 0.04] M2:ES 0.15*** 0.23*** 0.04 [0.02, 0.06] Model2 IV:Psy DV:Dep 0.31*** 0.22*** M1:CR -0.20*** -0.24*** 0.05 [0.03, 0.08] M2:ES 0.19*** 0.22*** 0.04 [0.02, 0.07] Model3 IV:Nar DV:Dep 0.15*** 0.12** M1:CR -0.09* -0.27*** 0.02 [0.00, 0.05] M2:ES 0.04 0.25*** 0.01 [-0.01, 0.03] Note: IV = independent variable; Dep = Depression; Mac = Machiavellianism; Psy = Psychopathy; Nar = Narcissism; CR = Cognitive reappraisal, ES = Expressive suppression. * p < .05, * * p < .01, * * * p < .001. Moderated mediation analyses Three moderated mediation models were illustrated in Figs. 1, 2 and 3. Gender was not observed to moderate any associations in these models. Whereas, age was found to moderate the effect of Machiavellianism on reappraisal (Fig. 1). The negative effect of Machiavellianism on reappraisal was significant for younger students (β =-0.17, p < .01) but non-significant for older students (β = 0.03, p > .05). Age was found to moderate the effect of reappraisal on depression (Figs. 1, 2 and 3). The negative effect of reappraisal on depression was stronger for younger students (β =-0.35, p < .001) than for older students (β = -017, p < .01) . Fig. 1 A moderated mediation model with Machiavellianism as a predictor. (CR=Cognitive reappraisal, ES=Expressive suppression. * * * p < .001.) Fig. 2  A moderated mediation model with psychopathy as a predictor. (CR=Cognitive reappraisal, ES=Expressive suppression.) Fig. 3  A moderated mediation model with Narcissism as a predictor (CR=Cognitive reappraisal, ES=Expressive suppression.) Discussion Although Dark Triad traits have been related with depressive symptoms among adults (Bonfá-Araujo et al., 2021; Gómez-Leal et al., 2019; Mojsa-Kaja et al., 2021; Shih et al., 2019), the relationships have not been examined in adolescents and the underlying mechanism was rarely investigated. In this study, moderated mediation models were carried out to investigate the underlying mechanism between Dark Triad and depression with a Chinese adolescents sample. The results determined the positive associations between Dark Triad traits and depression, the mediating effect of emotion regulation strategies and the moderating role of age in these associations . As expected, Machiavellian adolescents reported more symptoms of depression. This findings correspond with most prior studies (Bonfá-Araujo et al., 2021; Gómez-Leal et al., 2019; Shih et al., 2019), but a cluster-analysic study indicated that high Machiavellianism coexisted with less symptoms of depression (Bianchi & Mirkovic, 2020). This discordance may be interpreted by potential moderators. For example, ability emotional intelligence has been proved to moderate the Machiavellianism-depression association in male. Machiavellian men with high emotional intelligence displayed less depression, but Machiavellian men with low emotional intelligence displayed more depression (Bianchi et al., 2020). Also, in line with previous results (Bonfá-Araujo et al., 2021; Gómez-Leal et al., 2019; Shih et al., 2019), psychopathic adolescents exhibited increased levels of depression. However, particular works suggested the opposite evidences, indicating that depressed male prisoners scored lower in the Psychopathy Checklist-Revised (PCL-R) (Hansen et al., 2013; Willemsen et al., 2011). This incongruent result may be accounted for two facets of psychopathy, clinical and subclinical psychopathy. The PCL-R is developed and validated on criminal samples, and includes lots of items involving criminality. It is adapted for measuring clinical psychopathic personality in prisoners, not subclinical psychopathy. When the Levenson Self-Report Psychopathy Scale (LSRP) was employed to measure subclinical psychopathy among male offenders, psychopathy exhibited a positive correlation with depression (Pennington et al., 2015), which was consistent with results in normal populations. Consistent with our hypothesis, adolescents high in narcissism experienced more depressive symptoms. With regard to the relation of narcissism and depression, many researchers discovered narcissism was correlated with reduced depressive symptoms and may serve on a protective factor for depression disorder (Bonfá-Araujo et al., 2021; Lyons et al., 2019; Shih et al., 2019), while others found narcissism was correlated with increased depressive symptoms (Gogola et al., 2021; Mojsa-Kaja et al., 2021). This contradiction could be explained by two subdimensions of narcissism, vulnerability and grandiosity, which have different associations with depressive symptoms. Vulnerable narcissism is commonly regarded as a risk factor for depression (Papageorgiou, Denovan & Dagnall, 2019), but grandiose narcissism seems to be associated with a lower level of depression (Papageorgiou et al., 2019). Considering the narcissism scale of Dirty Dozen used in this study displayed more relation with the vulnerability dimension (Maples et al., 2014), the positive association between narcissism and depression we found is reasonable. In regard to the mediating effects, our work proved that adolescents high in Dark Triad traits displayed improper use of emotion regulation strategies, bringing out an increased risk of depression. To be specific, Machiavellian individuals tended to apply expressive suppression leading to their higher level of depressive symptoms. For Machiavellians, manipulation of others requires hiding their own emotions to get what they want, leading to more use of expressive suppression (Walker et al., 2022). Although they can obtain short-term benefits by suppressing emotional expressions, their mental health will be impaired in the long run. Psychopaths are inclined to use expressive suppression and less likely to use cognitive reappraisal, which result in more symptoms of depression in them. This finding corresponds to the meta-analysis results (Walker et al., 2022), it may indicate that people high in psychopathy are unable to reappraise the adverse situations, and switch to apply suppression to regulate their own emotion processes. This habit of emotion regulation will further cause negative outcomes, such as depressive disorders. Narcissists are prone to less use of cognitive reappraisal which can give rise to their depressive symptoms. As for narcissists, they tend to seek admire, attention, and special favors from others. In China, adolescents high on narcissism need to achieve outstanding academic performance in order to obtain applause or reward from teachers and parents, and to gain the superiority status among peers. It is difficult to get prestige or status in the class by other ways. Thus, when facing stressful situations such as failure in the exam, narcissistic students can not have a different way of looking at them, suggesting less use of cognitive reappraisal accompanying with a increased risk of depression. Unexpectedly, gender was not observed to moderate any associations in mediation models. Although we observed that boys displayed higher levels of Dark Triad traits and more use of expressive suppression, they didn’t reported more symptoms of depression than girls. Considering the previous finding (Bianchi et al., 2020), there might exist some variables, such as ability emotional intelligence, moderating the relationship between Dark Triad and depression among boys. These moderators can make Dark Triad traits become protective factors of depression in boys. Researchers should further explore potential moderators in order to explain inconsistent results of gender differences in Dark Triad, emotion regulation strategies and depression. Age was found to moderate the effect of Machiavellianism on reappraisal and the effect of reappraisal on depression. Increased with age, the adverse impact of Machiavellianism on cognitive reappraisal was greatly lessened to non-significant. That is probably because younger adolescents are going through the transition from childhood to adolescence with dramatic biological and psychological changes (Götz et al., 2020). During this period, negative attitudes such as suspiciousness, mistrust, negativity, and cynicism may have significantly negative impact (Zhu et al., 2021). Then, the adverse impact disappeared with maturity. Similarly, the weakening effect of cognitive reappraisal on depression was significantly attenuated with age. According to past studies (Nolen-Hoeksema & Aldao, 2011; Vannucci et al., 2018), the effectiveness of adaptive emotion regulation strategies may be susceptible to developmental changes in social contexts and stressor types. For instance, cognitive reappraisal was negatively related with depression under uncontrollable stress, while cognitive reappraisal was positively related with depression under controllable stress (Troy et al., 2013). Rather than gender differences, researchers have paid less attention to age changes in the use and effectiveness of emotion regulation strategies, which was not helpful to choose the most appropriate strategies to relive negative emotions at different ages, especially for children and adolescents. So, it is essential to investigated the developmental changes in the effect of emotion regulation strategies on mental health. Implications This work confirmed the relationship of Dark Triad traits and depression in adolescents. Consistent with most results from adults, adolescents with higher levels of Dark Triad traits experienced increased symptoms of depression. This result enriched the profile of Dark Triad traits, and provided more evidence for the passive impact of Dark Triad on mental health. For the first time, this study examined the mediating effect of suppression and reappraisal between Dark Triad and depression. More specifically, suppression mediated the Machiavellianism-depression association; reappraisal and suppression mediated the psychopathy-depression link; and reappraisal mediated the narcissism-depression association. These results provided a better understanding of the potential mechanism of Dark Triad on mental health. For different Dark Triad traits, we should take pertinent intervention focusing on effective emotion regulation strategies to reduce negative emotional states. Significantly, we observed the moderating role of age between reappraisal and depression in adolescents. This finding is in keeping with the previous work (Vannucci et al., 2018), suggesting the effectiveness of emotion regulation strategies, such as cognitive reappraisal, may change across developmental age. Given the effect of reappraisal on depression is relatively weak for older adolescents, psychological teachers or counselors should consider more adaptive strategies and suitable therapies for them, such as interpersonal therapy to improve social functioning and mindfulness training aimed at paying attention to the present moment. Limitations and future directions Whereas, several limitations should be noticed in this study. First, the Dirty Dozen (DD) we used can not cover the different facets of Dark Triad traits which may have distinct effect on depression, such as two subdimensions of narcissism (vulnerability and grandiosity). The more comprehensive measurements should be applied to explored the relation of Dark Triad traits and mental disorders. Second, our analysis only included two emotion regulation strategies (cognitive reappraisal and expressive suppression), but they were insufficient to fully explain the relationships between Dark Triad and depression. In future, researchers should concerned about more cognitive and behavior strategies in the processes of emotion regulation. Third, the 12–18 age range of adolescents in the current study was not wide, so the moderating effects of age we observed may be not significant in other age groups, such as children and adults. A sample covering more ages was needed to explore the developmental changes in the use and effectiveness of Dark Triad and emotion regulation strategies. Fourth, although some of the correlations in this study were significant, their sizes were relatively small. For instance, the correlation coefficient between narcissism and reappraisal is -0.09 (p < .05). This probably was due to the large sample size. So the relationship between narcissism and reappraisal need to be further explored. Finally, this study did not address the influence of specific negative events such as COVID-19 pandemic (Ren et al., 2021) and bullying experience (Ngo et al., 2021) on adolescents’ mental health. Considering about previous findings during COVID-19 pandemic (Gogola et al., 2021; Mojsa-Kaja et al., 2021) and related evidences of bullying (Geel et al., 2017), it is necessary to investigate the underlying mechanisms of Dark Triad and depression related to specific negative events. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request Declarations Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards Informed consent Informed consent was obtained from all participants and their parents included in the current study Conflict of interest None Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Andresen EM Malmgren JA Carter WB Patrick DL Screening for depression in well older adults: evaluation of a short form of the CES-D (center for epidemiologic Studies Depression Scale) American journal of preventive medicine 1994 10 2 77 84 10.1016/S0749-3797(18)30622-6 8037935 Akram U Stevenson JC Self-disgust and the dark triad traits: the role of expressive suppression Personality and Individual Differences 2021 168 110296 10.1016/j.paid.2020.110296 Bianchi R Mirkovic D Is machiavellianism associated with depression? A cluster-analytic study Personality and Individual Differences 2020 152 109594 10.1016/j.paid.2019.109594 Bianchi R Patthey N Mirkovic D Lemaître B Schlegel K Machiavellian males with high emotional intelligence exhibit fewer depressive symptoms Personality and Individual Differences 2020 158 109867 10.1016/j.paid.2020.109867 Bonfá-Araujo B Lima-Costa AR Baptista MN Hauck-Filho N Depressed or satisfied? The relationship between the dark triad traits, depression, and life satisfaction Current Psychology 2021 10.1007/s12144-021-01726-3 Cheng HG Chen S McBride O Phillips MR Prospective relationship of depressive symptoms, drinking, and tobacco smoking among middle-aged and elderly community-dwelling adults: results from the China Health and Retirement Longitudinal Study (CHARLS) Journal of affective disorders 2016 195 136 143 10.1016/j.jad.2016.02.023 26895091 De Bruyn EH Cillessen AH Popularity in early adolescence: Prosocial and antisocial subtypes Journal of Adolescent Research 2006 21 6 607 627 10.1177/0743558406293 De Girolamo G Dagani J Purcell R Cocchi A McGorry P Age of onset of mental disorders and use of mental health services: needs, opportunities and obstacles Epidemiology and Psychiatric Sciences 2012 21 1 47 57 10.1017/S2045796011000746 22670412 Geel MV Goemans A Toprak F Vedder P Which personality traits are related to traditional bullying and cyberbullying? A study with the big five, Dark Triad and sadism Personality and Individual Differences 2017 106 231 235 10.1016/j.paid.2016.10.063 Gogola AM Debski P Goryczka A Gorczyca P Piegza M The Dark Triad of Personality’s relationship with compliance towards COVID-19 pandemic recommendations along with anxiety and depressive symptoms in polish Citizens International Journal of Environmental Research and Public Health 2021 18 5478 10.3390/ijerph18105478 34065401 Gómez-Leal R Megías-Robles A Gutiérrez-Cobo MJ Cabello R Fernández- Abascal EG Fernández-Berrocal P Relationship between the dark triad and depressive symptoms PeerJ 2019 7 e8120 10.7717/peerj.8120 31803535 Götz FM Bleidorn W Rentfrow PJ Age differences in Machiavellianism across the life span: evidence from a large-scale cross-sectional study Journal of personality 2020 88 5 978 992 10.1111/jopy.12545 32145085 Granrud MD Steffenak AK Theander K Gender differences in symptoms of depression among adolescents in Eastern Norway: results from a cross-sectional study Scandinavian Journal of Public Health 2019 47 2 157 165 10.1177/1403494817715379 28669279 Gross JJ The emerging field of emotion regulation: an integrative review Review of General Psychology 1998 2 3 271 299 10.1037/1089-2680.2.3.271 Gross JJ Emotion regulation: affective, cognitive, and social consequences Psychophysiology 2002 39 3 281 291 10.1017/S0048577201393198 12212647 Gross JJ John OP Individual differences in two emotion regulation processes: implications for affect, relationships, and well-being Journal of Personality and Social Psychology 2003 85 2 348 362 10.1037/0022-3514.85.2.348 12916575 Hansen AL Stokkeland L Pallesen S Johnsen BH Waage L The relationship between the psychopathy checklist-revised and the MMPI-2: a pilot study Psychological Reports 2013 112 2 445 457 10.2466/03.09.PR0.112.2.445-457 23833874 Hayes AF Introduction to mediation, moderation, and conditional process analysis: a regression-based approach 2013 New York, NY Guilford Press John, O. P., & Gross, J. J. (2004). Healthy and unhealthy emotion regulation: Personality processes, individual differences, and life span development. Journal of Personality, 72(6), 1301–1334. 10.1111/j.1467-6494.2004.00298.x Jonason PK Webster GD The dirty dozen: a concise measure of the dark triad Psychological Assessment 2010 22 2 420 432 10.1037/a0019265 20528068 Jonason PK Li NP Buss DM The costs and benefits of the Dark Triad: implications for mate poaching and mate retention tactics Personality and Individual Differences 2010 48 4 373 378 10.1016/j.paid.2009.11 Jonason PK Koenig BL Tost J Living a fast life: the Dark Triad and life history theory Human Nature 2010 21 428 442 10.1007/s12110-010-9102-4 Jonason PK Baughman HM Carter GL Parker PD Dorian Gray without his portrait: psychological, social, and physical health costs associated with the Dark Triad Personality and Individual Differences 2015 78 5 13 10.1016/j.paid.2015.01.008 Kawamoto T Shimotsukasa T Oshio A Cross-sectional age differences in the Dark Triad traits in two japanese samples Psychology and Aging 2020 35 1 91 96 10.1037/pag0000399 31556647 Kotov R Gamez W Schmidt F Watson D Linking “big” personality traits to anxiety, depressive, and substance use disorders: a meta-analysis Psychological Bulletin 2010 136 5 768 821 10.1037/a0020327 20804236 Krueger RF Eaton NR Personality traits and the classification of mental disorders: toward a more complete integration in DSM–5 and an empirical model of psychopathology Personality Disorder 2010 1 2 97 118 10.1037/a0018990 Lewis-Fernández R Kleinman A Culture, personality, and psychopathology Journal of Abnormal Psychology 1994 103 1 67 71 10.1037/0021-843X.103.1.67 8040483 Liu DY Thompson RJ Selection and implementation of emotion regulation strategies in major depressive disorder: an integrative review Clinical psychology review 2017 57 183 194 10.1016/j.cpr.2017.07.004 28739277 Lyons M Evans K Helle S Do “dark” personality features buffer against adversity? The associations between cumulative life stress, the Dark Triad, and mental distress SAGE Open 2019 9 1 1 13 10.1177/2158244018822383 34290901 Maples JL Lamkin J Miller JD A test of two brief measures of the dark triad: the dirty dozen and short dark triad Psychological assessment 2014 26 1 326 331 10.1037/a0035084 24274044 Masumoto K Taishi N Shiozaki M Age and gender differences in relationships among emotion regulation, mood, and mental health Gerontology and Geriatric Medicine 2016 2 1 8 10.1177/2333721416637022 McTernan WP Dollard MF LaMontagne AD Depression in the workplace: an economic cost analysis of depression-related productivity loss attributable to job strain and bullying Work and Stress 2013 27 4 321 338 10.1080/02678373.2013.846948 Mojsa-Kaja J Szklarczyk K González-Yubero S Palomera R Cognitive emotion regulation strategies mediate the relationships between Dark Triad traits and negative emotional states experienced during the COVID-19 pandemic Personality and Individual Differences 2021 181 111018 10.1016/j.paid.2021.111018 Muris P Merckelbach H Otgaar H Meijer E The malevolent side of human nature Perspectives on Psychological Science 2017 12 2 183 204 10.1177/1745691616666070 28346115 Nelemans SA Branje SJT Hale WW Goossens L Koot HM Oldehinkel AJ Meeus WHJ Discrepancies between perceptions of the parent–adolescent relationship and early adolescent depressive symptoms: an illustration of polynomial regression analysis Journal of Youth and Adolescence 2016 45 2049 2063 10.1007/s10964-016-0503-5 27230118 Ngo AT Nguyen LH Dang AK Hoang MT Nguyen THT Vu GT Bullying experience in urban adolescents: prevalence and correlations with health-related quality of life and psychological issues Plos One 2021 16 6 e0252459 10.1371/journal.pone.0252459 34101739 Nolen-Hoeksema, S., & Aldao, A. (2011). Gender and age differences in emotion regulation strategies and their relationship to depressive symptoms. Personality and Individual Differences,51(6), 704–708. 10.1016/j.paid.2011 Papageorgiou KA Denovan A Dagnall N The positive effect of narcissism on depressive symptoms through mental toughness: narcissism may be a dark trait but it does help with seeing the world less grey European Psychiatry 2019 55 74 79 10.1016/j.eurpsy.2018.10.002 30390475 Paulhus DL Williams KM The dark triad of personality: narcissism, Machiavellianism, and psychopathy Journal of Research in Personality 2002 36 6 556 563 10.1016/S0092-6566(02)00505-6 Pennington CR Cramer RJ Miller HA Anastasi JS Psychopathy, Depression, and anxiety as predictors of suicidal ideation in offenders Death Studies 2015 39 288 295 10.1080/07481187.2014.991953 25551785 Ren Z Xin Y Wang Z Liu D Ho RCM Ho CSH What factors are most closely Associated with Mood Disorders in Adolescents during the COVID-19 pandemic? A cross-sectional study based on 1,771 adolescents in Shandong Province, China Frontiers in Psychiatry 2021 12 728278 10.3389/fpsyt.2021.728278 34603106 Rogoza R Zemojtel-Piotrowska M Jonason PK Piotrowski J Campbell KW Gebauer JE Structure of Dark Triad dirty dozen across eight world regions Assessment 2020 28 4 1125 1135 10.1177/1073191120922611 32484407 Sai L Luo S Ward A Sang B Development of the tendency to use emotion regulation strategies and their relation to depressive symptoms in chinese adolescents Frontiers in Psychology 2016 7 1222 10.3389/fpsyg.2016.01222 27597834 Sarafino EP Smith TW Health psychology: Biopsychosocial interactions 2012 7 New York, NY Wiley Shorey S Ng ED Wong CH Global prevalence of depression and elevated depressive symptoms among adolescents: a systematic review and meta-analysis The British journal of clinical psychology 2021 61 2 287 305 10.1111/bjc.12333 34569066 Shih S Chi N Wu C Wang S When dark meets blue: the relations between dark triad personality and depression symptoms Current Psychology 2019 40 6110 6117 10.1007/s12144-019-00549-7 Sijtsema J Garofalo C Jansen K Klimstra T Disengaging from Evil: Longitudinal Associations between the Dark Triad, Moral Disengagement, and antisocial behavior in adolescence Journal of Abnormal Child Psychology 2019 47 1351 1365 10.1007/s10802-019-00519-4 30737660 Troy AS Shallcross AJ Mauss IB A person-by-situation approach to emotion regulation: cognitive reappraisal can either help or hurt, depending on the context Psychological Science 2013 24 12 2505 2514 10.1177/0956797613496434 24145331 Twenge JM Nolen-Hoeksema S Age, gender, race, socioeconomic status, and birth cohort difference on the children’s depression inventory: a meta-analysis Journal of Abnormal Psychology 2002 111 4 578 588 10.1037/0021-843X.111.4.578 12428771 Vannucci A Flannery KM McCauley Ohannessian C Age-varying associations between coping and depressive symptoms throughout adolescence and emerging adulthood Development and Psychopathology 2018 30 2 665 681 10.1017/S0954579417001183 29692294 Walker, S. A., Olderbak, S. G., Gorodezki, J., Zhang, M., Ho, C., & MacCann, C. (2022). Primary and secondary psychopathy relate to lower cognitive reappraisal: A meta-analysis of the Dark Triad and emotion regulation processes. Personality and Individual Differences, 187, 111394. 10.1016/j.paid Wang D Yuan B Han H Wang C Validity and reliability of emotion regulation questionnaire (ERQ) in chinese rural-to-urban migrant adolescents and young adults Current Psychology 2020 41 2346 2353 10.1007/s12144-020-00754-9 Willemsen J Vanheule S Verhaeghe P Psychopathy and lifetime experiences of depression Criminal behaviour and mental health 2011 21 4 279 294 10.1002/cbm.812 21469239 Zeigler-Hill V Vonk J Dark personality features and emotion dysregulation Journal of Social and Clinical Psychology 2015 34 8 692 704 10.1521/jscp.2015.34.8.692 Zhu X Wang F Geng Y Machiavellianism on quality of life: the role of lifestyle, age, gender, social support Personality and Individual Differences 2021 173 110609 10.1016/j.paid.2020.110609 Zhang L Lu J Li B Wang X Shangguan C Gender differences in the mediating effects of emotion-regulation strategies: forgiveness and depression among adolescents Personality and Individual Differences 2020 163 110094 10.1016/j.paid.2020.110094 Zhao X Zhang R Zheng K Gender differences in emotion regulation strategies in adolescents Chinese Journal of Clinical Psychology 2014 22 05 849 854 10.16128/j.cnki.1005-3611.2014.05.067

0

PMC9748877

NO-CC CODE

2022-12-15 23:22:47

no

Curr Psychol. 2022 Dec 14;:1-10

utf-8

Curr Psychol

10.1007/s12144-022-04132-5

oa_other

==== Front Curr Psychol Curr Psychol Current Psychology (New Brunswick, N.j.) 1046-1310 1936-4733 Springer US New York 4132 10.1007/s12144-022-04132-5 Article The dark triad and depressive symptoms among chinese adolescents: moderated mediation models of age and emotion regulation strategies http://orcid.org/0000-0003-3630-3460 Shen Ke 673472766@qq.com Henan Police College, 450001 Zhengzhou, China 14 12 2022 110 21 8 2022 20 11 2022 5 12 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Dark Triad has been found to relate with depressive symptoms, but the underlying mechanism was rarely investigated. In the present study, we examined the mediating effect of two emotion regulation strategies (cognitive reappraisal and expressive suppression) and the moderating effect of gender and age. 709 Chinese adolescents aged 12–18 years (M = 14.54, SD = 1.70;55.3% girls) filled out the Dirty Dozen, Emotion Regulation Questionnaire and the 10-item of Center for Epidemiological Studies Depression Scale. As expected, Dark Triad was related with increased levels of depression and emotion regulation strategies acted as mediators in these associations. Suppression mediated the relationship between Machiavellianism and depression. Reappraisal and suppression mediated the link between psychopathy and depression. Reappraisal mediated the association between narcissism and depression. Besides, age moderated the effect of Machiavellianism on reappraisal, suggesting Machiavellianism had a negative impact on reappraisal for younger adolescents, but not for older adolescents. Age also moderated the effect of reappraisal on depression, indicating the negative effect of reappraisal on depression was stronger for younger adolescents than for older adolescents. These results address how Dark Triad traits affect depression via emotion regulation strategies and indicate the effectiveness of Dark Triad and emotion regulation strategies may change across age groups. Keywords Dark triad Depression Emotion regulation strategies Cognitive reappraisal Expressive suppression Age ==== Body pmcIntroduction Depression is characterized by sustained negative affect, which not only has adverse effects on individual’s mental health and social functioning, but also results in huge economic costs for society (McTernan et al., 2013). Due to its wide and severe influences, depression has become a global public health issue. From the view of personality psychology, individuals with depressive symptoms seem to exhibit certain personality characteristics. The relation of personality and mental disorders has been discussed for a long time (Krueger & Eaton, 2010), and Big Five traits (neuroticism, extraversion, conscientiousness, agreeableness, and openness to experience) is the most concentrated subject. In a meta-analysis from 175 empirical studies, patients with depression report higher levels of neuroticism, and lower levels of extraversion and conscientiousness than health people (Kotov et al., 2010). Besides Big Five traits, more and more researchers started to explore the relationship between depression and the “dark side” of personality, Dark Triad traits. The Dark Triad is a group of socially malevolent personality traits, which includes three separate but related subclinical traits—Machiavellianism, psychopathy, and narcissism (Paulhus & Williams, 2002). Although these traits share the tendencies toward dishonesty, coldness, and selfishness, they still are different concept. Machiavellianism is characterized by apathy, utilitarianism, cheat and manipulation. Psychopathy refers to high impulsivity, thrill-seeking and low empathy. Narcissism is identified as grandiosity, dominance, and superiority (Paulhus &Williams, 2002). Based on life history theory, individuals high in Dark Triad traits apply a fast life strategy exhibited by short-term mating, risk-taking tendency and indifference to social morality (Jonason et al., 2010), in order to gain more reproductive and living resources. Thus, Dark Triad often correlated with aggressive, unethical and antisocial behavior (Sijtsema et al., 2019), and negative mental health outcomes such as depressive symptoms (Gómez-Leal et al., 2019). Quiet a few of studies have examined the association between Dark Triad and depression among adults (e.g., Bonfá-Araujo et al., 2021; Gómez-Leal et al., 2019; Shih et al., 2019). Except for several evidences (Hansen et al., 2013; Willemsen et al., 2011), the majority of researchers observed that Machiavellian and psychopathic individuals appeared to exhibit higher levels of depression (Gómez-Leal et al., 2019; Gogola et al., 2021; Lyons et al., 2019; Mojsa-Kaja et al., 2021; Shih et al., 2019). As for narcissism, many studies found narcissism seem a “bright” trait of the Dark Triad, and correlated with less signs of depression (Bonfá-Araujo et al., 2021; Lyons et al., 2019; Shih et al., 2019). Whereas, recent findings during the COVID-19 pandemic showed individuals high in narcissism reported increased depressive symptoms (Gogola et al., 2021; Mojsa-Kaja et al., 2021). Although the association between Dark Triad and depression has been investigated by many studies, participants in these studies mainly recruited from adults, less from other age groups, such as adolescents. Some research have found adolescents with Dark Triad traits could obtain high peer status (De Bruyn, & Cillessen, 2006). But high peer status could not protect them from experiencing emotional pain due to less secure relationships (Jonason et al ., 2010). Conversely, Dark Triad traits were likely to predict their bad mental health and adverse well-being outcomes, like depressive symptoms (Jonason et al., 2015). In the development of depression, adolescence is regarded as a crucial period (Nelemans et al., 2016). Intervention during this period can reduce the severity or persistence of depression disorders (De Girolamo et al., 2012). Thus, the association between Dark Triad and depression is need to be examined in adolescents, in order to figure out risking factors in the process of prevention and treatment of depression. Additionally, the underlying mechanism of the Dark Triad-depression association is still unclear. In particular, there is a lack of research to investigate factors moderating or mediating this association. And most researchers have examined the correlation of Dark Triad and depression in Western societies, few in Asian countries, especially in China. According to Lewis-Fernández and Kleinman (1994), culture had significant influences on the personality and psychopathology. The management of emotions was a major currency of relationships and a hallmark of health in Chinese communities. So, it is necessary to explore the Dark Triad-depression relationship and the underlying factors, such as emotion regulation, between them among Chinese adolescents. Emotion regulation, referring to the processes through which individuals influence their own emotional experiences and expressions using a variety of cognitive and behavior strategies (Gross, 1998). Emotion regulation deficits, particularly inappropriate habitual use of emotion regulation strategies, are considered as key components of depressive disorders (Liu & Thompson, 2017) and Dark Triad (Zeigler-Hill & Vonk, 2015). Based on the biopsychosocial model of health psychology (Sarafino & Smith, 2012), coping strategies can mediate the relationship between dispositional characteristics, such as Dark Triad traits, and health-related outcomes. In line with this model, a latest study demonstrated maladaptive cognitive emotion regulation strategies mediate the relationship between Dark Triad and stress, anxiety and depression experienced during the COVID-19 pandemic (Mojsa-Kaja et al., 2021). Hence, it is plausible to regard emotion regulation strategies as mediators between Dark Triad and depression. As the most widely studied emotion regulation strategies, cognitive reappraisal and expressive suppression have distinct effects on depression. Expressive suppression, referring to the inhibition of outward emotional expression to control emotional responses, was regarded as a maladaptive behavioral strategy resulting in adverse emotional, social and cognitive consequences (Gross, 2002; Gross & John, 2003; John & Gross, 2004) including increased depressive symptoms (Sai et al., 2016). In contract, cognitive reappraisal, defined as an attempt to change one’s subjective evaluation of the emotion-eliciting situation to alter its emotional impact, was considered as an adaptive cognitive strategy which is correlated with positive emotional experiences, good interpersonal functioning, and high well-being (Gross & John, 2003; John & Gross, 2004) and effectively relieved depressive symptoms (Sai et al., 2016). In addiction, Dark Triad traits have been shown to be correlated with cognitive reappraisal and expressive suppression. In a latest meta-analysis, psychopathy was significantly correlated with less use of reappraisal and more use of suppression (Walker et al., 2022). A recent work revealed that Machiavellians were habituated to more use of suppression, and narcissists were accustomed to less use of suppression (Akram & Stevenson, 2021). Thus, cognitive reappraisal and expressive suppression may serve as mediating roles in the relationship between Dark Triad and depression. Moreover, there are gender differences in Dark Triad traits, emotion regulation strategies and depressive symptoms. Men tend to report higher scores in Dark Triad traits (Muris et al., 2017), and prefer to use more expressive suppression than women (Gross & John, 2003). Girls reported to use more reappraisal and less suppression than boys in Chinese adolescents (Zhao et al., 2014). Yet, female adolescents often reported more symptoms of depression than males (Granrud et al., 2019; Shorey et al., 2021). This may be due to gender differences in the mediating effect of emotion regulation strategies. For example, in the relationship between trait forgiveness and depression, the mediating effect of cognitive reappraisal was significantly greater for girls than for boys (Zhang et al., 2020). So, the mediating effect of emotion regulation strategies between Dark Triad traits and depression are possible to be moderate by gender. Age differences also existed in Dark Triad, emotion regulation strategies and depression. With age increased, the level of Dark Triad traits have decreased markedly in Japanese samples, and especially in women (Kawamoto, Shimotsukasa, & Oshio, 2020). From childhood to adolescence, boys’ depression scores changed little, but girls reported a significantly increased level of depression (Twenge & Nolen-Hoeksema, 2002). Men turn to reappraisal from suppression with age, further enhancing positive mood, reducing negative mood and improving their mental health (Masumoto, Taishi, & Shiozaki, 2016), but women turn to suppression as they grow older (Nolen-Hoeksema & Aldao, 2011). Given that empirical research of the relations between Dark Triad and depression yielded mixed results, the current study aims to examine the relationship between Dark Triad traits and depression in a Chinese adolescent sample. In addition, considering about the close connections among Dark Triad, emotion regulation and depression, we investigate whether emotion regulation strategies act as mediators between Dark Triad traits and depression and whether gender and age play moderating roles in the mediation model. According to prior studies, We hypothesized that (1) Dark Triad traits will related with increased levels of depression; (2) Dark Triad traits could correlated with more depressive symptoms via less use of cognitive reappraisal and more use of expressive suppression; (3) gender and age will moderate the mediating effect of emotion regulation strategies between Dark Triad traits and depression. Method Participants 730 students were recruited from two middle and high schools in Henan province, China. After eliminating invalid and extreme data, the sample included 709 adolescents (55.3% girls) aged 12–18 years (M = 14.54, SD = 1.70). Students completed this survey during the normal class period, and they were informed that their responses were anonymous and confidential. Informed consents were obtained from students and their parents. Measures Dark triad The Dark Triad was measured using the Dirty Dozen (DD; Jonason & Webster, 2010). It is a 12- item scale with three subscales: Machiavellianism (4 items, e.g., “I tend to manipulate others to get my way”), psychopathy (4 items, e.g.,“I tend to lack remorse”) and narcissism (4 items, e.g.,“I tend to want others to admire me”). Participants were asked to rate statements on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree), and higher scores represented higher levels of Dark Triad traits. DD has been confirmed to have acceptable structural validity across eight world regions (Rogoza et al., 2020). In the present study, the Cronbach’s α of Machiavellianism, psychopathy and narcissism subscales were 0.83, 0.62 and 0.84 respectively. Emotion regulation strategies Emotion regulation strategies were assessed using Emotion Regulation Questionnaire (ERQ; Gross & John, 2003). It is a 10-item questionnaire with two subscales: cognitive reappraisal (6 items, e.g., “I control my emotions by changing the way I think about the situation I’m in”) and expressive suppression (4 items, e.g., “I control my emotions by not expressing them”). Participants were asked to rate items on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree), and higher scores represented more use of reappraisal or suppression. ERQ has great validity and reliability in Chinese adolescents and young adults (Wang et al., 2020). In the current sample, the Cronbach’s α of reappraisal and suppression subscales were 0.84 and 0.72 respectively. Depressive symptoms Depressive symptoms were measured using the 10-item of Center for Epidemiological Studies Depression Scale (CESD-10) (Andresen et al., 1994). Participants were asked to rate each statement during the past week on a 4-point Likert scale: 0 (rarely or none of the time, < 1 day); 1 (some or few of the time, 1–2 days); 2 (occasionally or a moderate amount of the time, 3–4 days); 3 (most or all of the time, 5–7 days). All items is calculated to yield a global score, and higher scores indicate more symptoms of depression. CESD-10 has been validated and widely used in the China Health and Retirement Longitudinal Study (Cheng et al., 2016). The Cronbach’s α of CESD-10 was 0.80 in the present sample. Data analyses Descriptive analysis and Pearson correlation analysis of all variables were conducted using SPSS 26.0. Parallel multiple mediation analyses were performed to examine the mediating effects of emotion regulation strategies between Dark Triad and depression by using the PROCESS macro for SPSS (Hayes, 2013). PROCESS model 4 was used to test three mediation models with an independent variable (Machiavellianism, psychopathy or narcissism), two mediators (cognitive reappraisal and expressive suppression) and a dependent variable (depression). Moderated mediation analyses were performed to examine the moderating effects of gender and age by using the PROCESS model 59. We set bootstrapping to 50,000 samples and a 95% confidence interval to determine statistical significance. If the confidence interval of indirect effect exclusive of zero, the mediating effect was considered as significant. According to age, the participants were divided into two groups: the younger group (age 12 to 14) and the older group (age 15 to 18). Results Common method deviation test The Harman single factor test was used to examine common method biases. We found 7 factors of which eigenvalues were greater than 1 without rotation, and the first factor could explain 18.84% of the variation, which indicated that the current study was not influenced by common method biases. Gender and age differences As showed in Table 1, boys reported significantly higher scores on all of Dark Triad traits and more use of expressive suppression than girls. In terms of age, older students showed a significantly higher level of Narcissism than younger students. Table 1 Gender and age differences tests of all variables (M ± SD). Variables Total Boys(306) Girls(403) t Younger(361) Older(348) t 1.Machiavellianism 1.72 ± 0.91 1.98 ± 1.04 1.53 ± 0.76 6.31* * * 1.73 ± 0.95 1.72 ± 0.88 0.22 2.Psychopathy 2.22 ± 1.04 2.42 ± 1.06 2.07 ± 1.00 4.51* * * 2.18 ± 1.05 2.26 ± 1.02 -0.98 3.Narcissism 3.31 ± 1.54 3.49 ± 1.52 3.18 ± 1.54 2.62* * 3.13 ± 1.50 3.51 ± 1.55 -3.34* * 4.Cognitive reappraisal 4.60 ± 1.16 4.62 ± 1.13 4.60 ± 1.19 0.24 4.69 ± 1.17 4.52 ± 1.15 1.93 5.Expressive suppression 3.99 ± 1.37 4.13 ± 1.31 3.89 ± 1.40 2.26* 4.08 ± 1.34 3.91 ± 1.39 1.63 6.Depression 1.01 ± 0.52 1.03 ± 0.53 0.99 ± 0.52 0.83 0.97 ± 0.56 1.04 ± 0.49 -1.72 Note: * p < .05, * * p < .01, * * * p < .001 Correlation analyses Person correlations (Table 2) suggested that Machiavellianism was positively related with expressive suppression and depression. Psychopathy were negatively related with cognitive reappraisal and positively related with expressive suppression and depression. Narcissism was negatively related with cognitive reappraisal and positively related with depression. As expected, cognitive reappraisal was negatively related with depression, and expressive suppression was positively related with depression. Table 2 Correlations between all variables Variables 1 2 3 4 5 6 1.Machiavellianism - 0.40** 0.38** -0.07 0.14** 0.26** 2.Psychopathy - 0.36** -0.20** 0.18** 0.30** 3.Narcissism - -0.09* 0.04 0.17** 4.Cognitive reappraisal - 0.07 -0.28** 5.Expressive suppression - 0.26** 6.Depression - Note: * p < .05, * * p < .01. Mediation analyses The direct and indirect effects of Dark Triad traits on depression through reappraisal and suppression were presented in Table 3. In model 1, the direct effect of Machiavellianism on depression (β = 0.22, p < .001) and the indirect effect via suppression (β = 0.04, 95% CI [0.02, 0.06]) was significant, indicating that suppression partially mediated the relationship between Machiavellianism and depression. Machiavellianism predicted higher depressive symptoms partially through more use of suppression. In model 2, the direct effect of psychopathy on depression (β = 0.22, p < .001) and the indirect effects via reappraisal (β = 0.05, 95% CI [0.03, 0.08]) and suppression (β = 0.04, 95% CI [0.02, 0.07]) was significant, suggesting that reappraisal and suppression partially mediated the relationship between psychopathy and depression. Psychopathy predicted more depressive symptoms partially through less use of reappraisal and more use of suppression. In model 3, the direct effect of narcissism on depression (β = 0.12, p < .01) and the indirect effects via reappraisal (β = 0.02, 95% CI [0.00, 0.05]) was significant, showing that reappraisal partially mediated the relationship between narcissism and depression. Narcissism predicted more depressive symptoms partially through less use of reappraisal. Table 3 Effects of Dark Triad traits on depression via emotion regulation strategies Model Variables Effect of IV on M Effect of M on DV Total effect Direct effect Indirect effect 95% CI Model1 IV:Mac DV:Dep 0.27*** 0.22*** M1:CR -0.08 -0.27*** 0.02 [-0.00, 0.04] M2:ES 0.15*** 0.23*** 0.04 [0.02, 0.06] Model2 IV:Psy DV:Dep 0.31*** 0.22*** M1:CR -0.20*** -0.24*** 0.05 [0.03, 0.08] M2:ES 0.19*** 0.22*** 0.04 [0.02, 0.07] Model3 IV:Nar DV:Dep 0.15*** 0.12** M1:CR -0.09* -0.27*** 0.02 [0.00, 0.05] M2:ES 0.04 0.25*** 0.01 [-0.01, 0.03] Note: IV = independent variable; Dep = Depression; Mac = Machiavellianism; Psy = Psychopathy; Nar = Narcissism; CR = Cognitive reappraisal, ES = Expressive suppression. * p < .05, * * p < .01, * * * p < .001. Moderated mediation analyses Three moderated mediation models were illustrated in Figs. 1, 2 and 3. Gender was not observed to moderate any associations in these models. Whereas, age was found to moderate the effect of Machiavellianism on reappraisal (Fig. 1). The negative effect of Machiavellianism on reappraisal was significant for younger students (β =-0.17, p < .01) but non-significant for older students (β = 0.03, p > .05). Age was found to moderate the effect of reappraisal on depression (Figs. 1, 2 and 3). The negative effect of reappraisal on depression was stronger for younger students (β =-0.35, p < .001) than for older students (β = -017, p < .01) . Fig. 1 A moderated mediation model with Machiavellianism as a predictor. (CR=Cognitive reappraisal, ES=Expressive suppression. * * * p < .001.) Fig. 2  A moderated mediation model with psychopathy as a predictor. (CR=Cognitive reappraisal, ES=Expressive suppression.) Fig. 3  A moderated mediation model with Narcissism as a predictor (CR=Cognitive reappraisal, ES=Expressive suppression.) Discussion Although Dark Triad traits have been related with depressive symptoms among adults (Bonfá-Araujo et al., 2021; Gómez-Leal et al., 2019; Mojsa-Kaja et al., 2021; Shih et al., 2019), the relationships have not been examined in adolescents and the underlying mechanism was rarely investigated. In this study, moderated mediation models were carried out to investigate the underlying mechanism between Dark Triad and depression with a Chinese adolescents sample. The results determined the positive associations between Dark Triad traits and depression, the mediating effect of emotion regulation strategies and the moderating role of age in these associations . As expected, Machiavellian adolescents reported more symptoms of depression. This findings correspond with most prior studies (Bonfá-Araujo et al., 2021; Gómez-Leal et al., 2019; Shih et al., 2019), but a cluster-analysic study indicated that high Machiavellianism coexisted with less symptoms of depression (Bianchi & Mirkovic, 2020). This discordance may be interpreted by potential moderators. For example, ability emotional intelligence has been proved to moderate the Machiavellianism-depression association in male. Machiavellian men with high emotional intelligence displayed less depression, but Machiavellian men with low emotional intelligence displayed more depression (Bianchi et al., 2020). Also, in line with previous results (Bonfá-Araujo et al., 2021; Gómez-Leal et al., 2019; Shih et al., 2019), psychopathic adolescents exhibited increased levels of depression. However, particular works suggested the opposite evidences, indicating that depressed male prisoners scored lower in the Psychopathy Checklist-Revised (PCL-R) (Hansen et al., 2013; Willemsen et al., 2011). This incongruent result may be accounted for two facets of psychopathy, clinical and subclinical psychopathy. The PCL-R is developed and validated on criminal samples, and includes lots of items involving criminality. It is adapted for measuring clinical psychopathic personality in prisoners, not subclinical psychopathy. When the Levenson Self-Report Psychopathy Scale (LSRP) was employed to measure subclinical psychopathy among male offenders, psychopathy exhibited a positive correlation with depression (Pennington et al., 2015), which was consistent with results in normal populations. Consistent with our hypothesis, adolescents high in narcissism experienced more depressive symptoms. With regard to the relation of narcissism and depression, many researchers discovered narcissism was correlated with reduced depressive symptoms and may serve on a protective factor for depression disorder (Bonfá-Araujo et al., 2021; Lyons et al., 2019; Shih et al., 2019), while others found narcissism was correlated with increased depressive symptoms (Gogola et al., 2021; Mojsa-Kaja et al., 2021). This contradiction could be explained by two subdimensions of narcissism, vulnerability and grandiosity, which have different associations with depressive symptoms. Vulnerable narcissism is commonly regarded as a risk factor for depression (Papageorgiou, Denovan & Dagnall, 2019), but grandiose narcissism seems to be associated with a lower level of depression (Papageorgiou et al., 2019). Considering the narcissism scale of Dirty Dozen used in this study displayed more relation with the vulnerability dimension (Maples et al., 2014), the positive association between narcissism and depression we found is reasonable. In regard to the mediating effects, our work proved that adolescents high in Dark Triad traits displayed improper use of emotion regulation strategies, bringing out an increased risk of depression. To be specific, Machiavellian individuals tended to apply expressive suppression leading to their higher level of depressive symptoms. For Machiavellians, manipulation of others requires hiding their own emotions to get what they want, leading to more use of expressive suppression (Walker et al., 2022). Although they can obtain short-term benefits by suppressing emotional expressions, their mental health will be impaired in the long run. Psychopaths are inclined to use expressive suppression and less likely to use cognitive reappraisal, which result in more symptoms of depression in them. This finding corresponds to the meta-analysis results (Walker et al., 2022), it may indicate that people high in psychopathy are unable to reappraise the adverse situations, and switch to apply suppression to regulate their own emotion processes. This habit of emotion regulation will further cause negative outcomes, such as depressive disorders. Narcissists are prone to less use of cognitive reappraisal which can give rise to their depressive symptoms. As for narcissists, they tend to seek admire, attention, and special favors from others. In China, adolescents high on narcissism need to achieve outstanding academic performance in order to obtain applause or reward from teachers and parents, and to gain the superiority status among peers. It is difficult to get prestige or status in the class by other ways. Thus, when facing stressful situations such as failure in the exam, narcissistic students can not have a different way of looking at them, suggesting less use of cognitive reappraisal accompanying with a increased risk of depression. Unexpectedly, gender was not observed to moderate any associations in mediation models. Although we observed that boys displayed higher levels of Dark Triad traits and more use of expressive suppression, they didn’t reported more symptoms of depression than girls. Considering the previous finding (Bianchi et al., 2020), there might exist some variables, such as ability emotional intelligence, moderating the relationship between Dark Triad and depression among boys. These moderators can make Dark Triad traits become protective factors of depression in boys. Researchers should further explore potential moderators in order to explain inconsistent results of gender differences in Dark Triad, emotion regulation strategies and depression. Age was found to moderate the effect of Machiavellianism on reappraisal and the effect of reappraisal on depression. Increased with age, the adverse impact of Machiavellianism on cognitive reappraisal was greatly lessened to non-significant. That is probably because younger adolescents are going through the transition from childhood to adolescence with dramatic biological and psychological changes (Götz et al., 2020). During this period, negative attitudes such as suspiciousness, mistrust, negativity, and cynicism may have significantly negative impact (Zhu et al., 2021). Then, the adverse impact disappeared with maturity. Similarly, the weakening effect of cognitive reappraisal on depression was significantly attenuated with age. According to past studies (Nolen-Hoeksema & Aldao, 2011; Vannucci et al., 2018), the effectiveness of adaptive emotion regulation strategies may be susceptible to developmental changes in social contexts and stressor types. For instance, cognitive reappraisal was negatively related with depression under uncontrollable stress, while cognitive reappraisal was positively related with depression under controllable stress (Troy et al., 2013). Rather than gender differences, researchers have paid less attention to age changes in the use and effectiveness of emotion regulation strategies, which was not helpful to choose the most appropriate strategies to relive negative emotions at different ages, especially for children and adolescents. So, it is essential to investigated the developmental changes in the effect of emotion regulation strategies on mental health. Implications This work confirmed the relationship of Dark Triad traits and depression in adolescents. Consistent with most results from adults, adolescents with higher levels of Dark Triad traits experienced increased symptoms of depression. This result enriched the profile of Dark Triad traits, and provided more evidence for the passive impact of Dark Triad on mental health. For the first time, this study examined the mediating effect of suppression and reappraisal between Dark Triad and depression. More specifically, suppression mediated the Machiavellianism-depression association; reappraisal and suppression mediated the psychopathy-depression link; and reappraisal mediated the narcissism-depression association. These results provided a better understanding of the potential mechanism of Dark Triad on mental health. For different Dark Triad traits, we should take pertinent intervention focusing on effective emotion regulation strategies to reduce negative emotional states. Significantly, we observed the moderating role of age between reappraisal and depression in adolescents. This finding is in keeping with the previous work (Vannucci et al., 2018), suggesting the effectiveness of emotion regulation strategies, such as cognitive reappraisal, may change across developmental age. Given the effect of reappraisal on depression is relatively weak for older adolescents, psychological teachers or counselors should consider more adaptive strategies and suitable therapies for them, such as interpersonal therapy to improve social functioning and mindfulness training aimed at paying attention to the present moment. Limitations and future directions Whereas, several limitations should be noticed in this study. First, the Dirty Dozen (DD) we used can not cover the different facets of Dark Triad traits which may have distinct effect on depression, such as two subdimensions of narcissism (vulnerability and grandiosity). The more comprehensive measurements should be applied to explored the relation of Dark Triad traits and mental disorders. Second, our analysis only included two emotion regulation strategies (cognitive reappraisal and expressive suppression), but they were insufficient to fully explain the relationships between Dark Triad and depression. In future, researchers should concerned about more cognitive and behavior strategies in the processes of emotion regulation. Third, the 12–18 age range of adolescents in the current study was not wide, so the moderating effects of age we observed may be not significant in other age groups, such as children and adults. A sample covering more ages was needed to explore the developmental changes in the use and effectiveness of Dark Triad and emotion regulation strategies. Fourth, although some of the correlations in this study were significant, their sizes were relatively small. For instance, the correlation coefficient between narcissism and reappraisal is -0.09 (p < .05). This probably was due to the large sample size. So the relationship between narcissism and reappraisal need to be further explored. Finally, this study did not address the influence of specific negative events such as COVID-19 pandemic (Ren et al., 2021) and bullying experience (Ngo et al., 2021) on adolescents’ mental health. Considering about previous findings during COVID-19 pandemic (Gogola et al., 2021; Mojsa-Kaja et al., 2021) and related evidences of bullying (Geel et al., 2017), it is necessary to investigate the underlying mechanisms of Dark Triad and depression related to specific negative events. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request Declarations Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards Informed consent Informed consent was obtained from all participants and their parents included in the current study Conflict of interest None Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Andresen EM Malmgren JA Carter WB Patrick DL Screening for depression in well older adults: evaluation of a short form of the CES-D (center for epidemiologic Studies Depression Scale) American journal of preventive medicine 1994 10 2 77 84 10.1016/S0749-3797(18)30622-6 8037935 Akram U Stevenson JC Self-disgust and the dark triad traits: the role of expressive suppression Personality and Individual Differences 2021 168 110296 10.1016/j.paid.2020.110296 Bianchi R Mirkovic D Is machiavellianism associated with depression? A cluster-analytic study Personality and Individual Differences 2020 152 109594 10.1016/j.paid.2019.109594 Bianchi R Patthey N Mirkovic D Lemaître B Schlegel K Machiavellian males with high emotional intelligence exhibit fewer depressive symptoms Personality and Individual Differences 2020 158 109867 10.1016/j.paid.2020.109867 Bonfá-Araujo B Lima-Costa AR Baptista MN Hauck-Filho N Depressed or satisfied? The relationship between the dark triad traits, depression, and life satisfaction Current Psychology 2021 10.1007/s12144-021-01726-3 Cheng HG Chen S McBride O Phillips MR Prospective relationship of depressive symptoms, drinking, and tobacco smoking among middle-aged and elderly community-dwelling adults: results from the China Health and Retirement Longitudinal Study (CHARLS) Journal of affective disorders 2016 195 136 143 10.1016/j.jad.2016.02.023 26895091 De Bruyn EH Cillessen AH Popularity in early adolescence: Prosocial and antisocial subtypes Journal of Adolescent Research 2006 21 6 607 627 10.1177/0743558406293 De Girolamo G Dagani J Purcell R Cocchi A McGorry P Age of onset of mental disorders and use of mental health services: needs, opportunities and obstacles Epidemiology and Psychiatric Sciences 2012 21 1 47 57 10.1017/S2045796011000746 22670412 Geel MV Goemans A Toprak F Vedder P Which personality traits are related to traditional bullying and cyberbullying? A study with the big five, Dark Triad and sadism Personality and Individual Differences 2017 106 231 235 10.1016/j.paid.2016.10.063 Gogola AM Debski P Goryczka A Gorczyca P Piegza M The Dark Triad of Personality’s relationship with compliance towards COVID-19 pandemic recommendations along with anxiety and depressive symptoms in polish Citizens International Journal of Environmental Research and Public Health 2021 18 5478 10.3390/ijerph18105478 34065401 Gómez-Leal R Megías-Robles A Gutiérrez-Cobo MJ Cabello R Fernández- Abascal EG Fernández-Berrocal P Relationship between the dark triad and depressive symptoms PeerJ 2019 7 e8120 10.7717/peerj.8120 31803535 Götz FM Bleidorn W Rentfrow PJ Age differences in Machiavellianism across the life span: evidence from a large-scale cross-sectional study Journal of personality 2020 88 5 978 992 10.1111/jopy.12545 32145085 Granrud MD Steffenak AK Theander K Gender differences in symptoms of depression among adolescents in Eastern Norway: results from a cross-sectional study Scandinavian Journal of Public Health 2019 47 2 157 165 10.1177/1403494817715379 28669279 Gross JJ The emerging field of emotion regulation: an integrative review Review of General Psychology 1998 2 3 271 299 10.1037/1089-2680.2.3.271 Gross JJ Emotion regulation: affective, cognitive, and social consequences Psychophysiology 2002 39 3 281 291 10.1017/S0048577201393198 12212647 Gross JJ John OP Individual differences in two emotion regulation processes: implications for affect, relationships, and well-being Journal of Personality and Social Psychology 2003 85 2 348 362 10.1037/0022-3514.85.2.348 12916575 Hansen AL Stokkeland L Pallesen S Johnsen BH Waage L The relationship between the psychopathy checklist-revised and the MMPI-2: a pilot study Psychological Reports 2013 112 2 445 457 10.2466/03.09.PR0.112.2.445-457 23833874 Hayes AF Introduction to mediation, moderation, and conditional process analysis: a regression-based approach 2013 New York, NY Guilford Press John, O. P., & Gross, J. J. (2004). Healthy and unhealthy emotion regulation: Personality processes, individual differences, and life span development. Journal of Personality, 72(6), 1301–1334. 10.1111/j.1467-6494.2004.00298.x Jonason PK Webster GD The dirty dozen: a concise measure of the dark triad Psychological Assessment 2010 22 2 420 432 10.1037/a0019265 20528068 Jonason PK Li NP Buss DM The costs and benefits of the Dark Triad: implications for mate poaching and mate retention tactics Personality and Individual Differences 2010 48 4 373 378 10.1016/j.paid.2009.11 Jonason PK Koenig BL Tost J Living a fast life: the Dark Triad and life history theory Human Nature 2010 21 428 442 10.1007/s12110-010-9102-4 Jonason PK Baughman HM Carter GL Parker PD Dorian Gray without his portrait: psychological, social, and physical health costs associated with the Dark Triad Personality and Individual Differences 2015 78 5 13 10.1016/j.paid.2015.01.008 Kawamoto T Shimotsukasa T Oshio A Cross-sectional age differences in the Dark Triad traits in two japanese samples Psychology and Aging 2020 35 1 91 96 10.1037/pag0000399 31556647 Kotov R Gamez W Schmidt F Watson D Linking “big” personality traits to anxiety, depressive, and substance use disorders: a meta-analysis Psychological Bulletin 2010 136 5 768 821 10.1037/a0020327 20804236 Krueger RF Eaton NR Personality traits and the classification of mental disorders: toward a more complete integration in DSM–5 and an empirical model of psychopathology Personality Disorder 2010 1 2 97 118 10.1037/a0018990 Lewis-Fernández R Kleinman A Culture, personality, and psychopathology Journal of Abnormal Psychology 1994 103 1 67 71 10.1037/0021-843X.103.1.67 8040483 Liu DY Thompson RJ Selection and implementation of emotion regulation strategies in major depressive disorder: an integrative review Clinical psychology review 2017 57 183 194 10.1016/j.cpr.2017.07.004 28739277 Lyons M Evans K Helle S Do “dark” personality features buffer against adversity? The associations between cumulative life stress, the Dark Triad, and mental distress SAGE Open 2019 9 1 1 13 10.1177/2158244018822383 34290901 Maples JL Lamkin J Miller JD A test of two brief measures of the dark triad: the dirty dozen and short dark triad Psychological assessment 2014 26 1 326 331 10.1037/a0035084 24274044 Masumoto K Taishi N Shiozaki M Age and gender differences in relationships among emotion regulation, mood, and mental health Gerontology and Geriatric Medicine 2016 2 1 8 10.1177/2333721416637022 McTernan WP Dollard MF LaMontagne AD Depression in the workplace: an economic cost analysis of depression-related productivity loss attributable to job strain and bullying Work and Stress 2013 27 4 321 338 10.1080/02678373.2013.846948 Mojsa-Kaja J Szklarczyk K González-Yubero S Palomera R Cognitive emotion regulation strategies mediate the relationships between Dark Triad traits and negative emotional states experienced during the COVID-19 pandemic Personality and Individual Differences 2021 181 111018 10.1016/j.paid.2021.111018 Muris P Merckelbach H Otgaar H Meijer E The malevolent side of human nature Perspectives on Psychological Science 2017 12 2 183 204 10.1177/1745691616666070 28346115 Nelemans SA Branje SJT Hale WW Goossens L Koot HM Oldehinkel AJ Meeus WHJ Discrepancies between perceptions of the parent–adolescent relationship and early adolescent depressive symptoms: an illustration of polynomial regression analysis Journal of Youth and Adolescence 2016 45 2049 2063 10.1007/s10964-016-0503-5 27230118 Ngo AT Nguyen LH Dang AK Hoang MT Nguyen THT Vu GT Bullying experience in urban adolescents: prevalence and correlations with health-related quality of life and psychological issues Plos One 2021 16 6 e0252459 10.1371/journal.pone.0252459 34101739 Nolen-Hoeksema, S., & Aldao, A. (2011). Gender and age differences in emotion regulation strategies and their relationship to depressive symptoms. Personality and Individual Differences,51(6), 704–708. 10.1016/j.paid.2011 Papageorgiou KA Denovan A Dagnall N The positive effect of narcissism on depressive symptoms through mental toughness: narcissism may be a dark trait but it does help with seeing the world less grey European Psychiatry 2019 55 74 79 10.1016/j.eurpsy.2018.10.002 30390475 Paulhus DL Williams KM The dark triad of personality: narcissism, Machiavellianism, and psychopathy Journal of Research in Personality 2002 36 6 556 563 10.1016/S0092-6566(02)00505-6 Pennington CR Cramer RJ Miller HA Anastasi JS Psychopathy, Depression, and anxiety as predictors of suicidal ideation in offenders Death Studies 2015 39 288 295 10.1080/07481187.2014.991953 25551785 Ren Z Xin Y Wang Z Liu D Ho RCM Ho CSH What factors are most closely Associated with Mood Disorders in Adolescents during the COVID-19 pandemic? A cross-sectional study based on 1,771 adolescents in Shandong Province, China Frontiers in Psychiatry 2021 12 728278 10.3389/fpsyt.2021.728278 34603106 Rogoza R Zemojtel-Piotrowska M Jonason PK Piotrowski J Campbell KW Gebauer JE Structure of Dark Triad dirty dozen across eight world regions Assessment 2020 28 4 1125 1135 10.1177/1073191120922611 32484407 Sai L Luo S Ward A Sang B Development of the tendency to use emotion regulation strategies and their relation to depressive symptoms in chinese adolescents Frontiers in Psychology 2016 7 1222 10.3389/fpsyg.2016.01222 27597834 Sarafino EP Smith TW Health psychology: Biopsychosocial interactions 2012 7 New York, NY Wiley Shorey S Ng ED Wong CH Global prevalence of depression and elevated depressive symptoms among adolescents: a systematic review and meta-analysis The British journal of clinical psychology 2021 61 2 287 305 10.1111/bjc.12333 34569066 Shih S Chi N Wu C Wang S When dark meets blue: the relations between dark triad personality and depression symptoms Current Psychology 2019 40 6110 6117 10.1007/s12144-019-00549-7 Sijtsema J Garofalo C Jansen K Klimstra T Disengaging from Evil: Longitudinal Associations between the Dark Triad, Moral Disengagement, and antisocial behavior in adolescence Journal of Abnormal Child Psychology 2019 47 1351 1365 10.1007/s10802-019-00519-4 30737660 Troy AS Shallcross AJ Mauss IB A person-by-situation approach to emotion regulation: cognitive reappraisal can either help or hurt, depending on the context Psychological Science 2013 24 12 2505 2514 10.1177/0956797613496434 24145331 Twenge JM Nolen-Hoeksema S Age, gender, race, socioeconomic status, and birth cohort difference on the children’s depression inventory: a meta-analysis Journal of Abnormal Psychology 2002 111 4 578 588 10.1037/0021-843X.111.4.578 12428771 Vannucci A Flannery KM McCauley Ohannessian C Age-varying associations between coping and depressive symptoms throughout adolescence and emerging adulthood Development and Psychopathology 2018 30 2 665 681 10.1017/S0954579417001183 29692294 Walker, S. A., Olderbak, S. G., Gorodezki, J., Zhang, M., Ho, C., & MacCann, C. (2022). Primary and secondary psychopathy relate to lower cognitive reappraisal: A meta-analysis of the Dark Triad and emotion regulation processes. Personality and Individual Differences, 187, 111394. 10.1016/j.paid Wang D Yuan B Han H Wang C Validity and reliability of emotion regulation questionnaire (ERQ) in chinese rural-to-urban migrant adolescents and young adults Current Psychology 2020 41 2346 2353 10.1007/s12144-020-00754-9 Willemsen J Vanheule S Verhaeghe P Psychopathy and lifetime experiences of depression Criminal behaviour and mental health 2011 21 4 279 294 10.1002/cbm.812 21469239 Zeigler-Hill V Vonk J Dark personality features and emotion dysregulation Journal of Social and Clinical Psychology 2015 34 8 692 704 10.1521/jscp.2015.34.8.692 Zhu X Wang F Geng Y Machiavellianism on quality of life: the role of lifestyle, age, gender, social support Personality and Individual Differences 2021 173 110609 10.1016/j.paid.2020.110609 Zhang L Lu J Li B Wang X Shangguan C Gender differences in the mediating effects of emotion-regulation strategies: forgiveness and depression among adolescents Personality and Individual Differences 2020 163 110094 10.1016/j.paid.2020.110094 Zhao X Zhang R Zheng K Gender differences in emotion regulation strategies in adolescents Chinese Journal of Clinical Psychology 2014 22 05 849 854 10.16128/j.cnki.1005-3611.2014.05.067

36370020

PMC9748878

NO-CC CODE

2022-12-15 23:22:47

no

Indian Pediatr. 2022 Dec 14; 59(11):886-887

latin-1

Indian Pediatr

10.1007/s13312-022-2651-x

oa_other

==== Front J Cancer Surviv J Cancer Surviv Journal of Cancer Survivorship 1932-2259 1932-2267 Springer US New York 1279 10.1007/s11764-022-01279-9 Article Development of a disease conceptual model of patient experience with metastatic colorectal cancer: identification of the most salient symptoms and impacts Guillemin Isabelle isabelle.guillemin@iqvia.com 1 Darpelly Mahesh 2 Wong Brendon 3 Ingelgård Anders 3 Griebsch Ingolf 3 1 grid.434277.1 IQVIA, Paris, France 2 grid.497480.6 IQVIA, Bangalore, India 3 grid.420061.1 0000 0001 2171 7500 Boehringer Ingelheim International GmbH, Ingelheim, Germany 14 12 2022 111 1 8 2022 17 10 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Purpose Patients with metastatic colorectal cancer (mCRC) experience multiple symptoms and impacts affecting their health-related quality of life. However, there is limited data on self-reported experience of the most relevant and bothersome aspects of patients living with mCRC. Methods Semi-structured interviews were conducted in patients with mCRC to identify and understand the signs, symptoms and impacts experienced. Patients were also asked to rate the level of bothersomeness for each concept reported on a scale ranging from 0 (“not bothersome at all”) to 10 (“extremely bothersome”). Verbatim transcripts were analysed following a thematic analysis approach. The most salient concepts were identified (i.e. reported by > 50% of patients with a bothersome rating ≥ 5 out of 10). Results Twenty-five patients (USA; age: 26 to 72 years old) were interviewed. Patients reported 58 signs and symptoms, amongst which 8 were considered salient: fatigue, nausea, neuropathy, diarrhoea, loss of appetite, constipation, weight loss, and abdominal pain; 35 impacts were identified, and 7 were considered salient: reduced ability to work, interference with daily activities, impact on cognitive functioning, financial impact, sleep changes, impact on social life and walking difficulties. The concepts identified helped refine a literature-based disease conceptual model of patient experience with mCRC. Conclusions The interviews provided insights into the most bothersome and salient signs, symptoms and impacts affecting the HRQoL of patients living with mCRC. Implications for cancer survivors There is a need to improve clinical strategies for future clinical development and inform clinical practice decision-making for mCRC survivors. Keyword Metastatic colorectal cancer; Self-reported experience; Health-related quality of life; Interviews; Conceptual model ==== Body pmcIntroduction Colorectal cancer (CRC) ranks as the second most lethal and the third most prevalent malignant tumour worldwide. In 2018, 1.8 million new CRC cases were diagnosed, and 881,000 deaths were reported, which accounted for nearly 10% of new cancer cases and deaths worldwide [1]. Almost a quarter of newly diagnosed CRC are metastatic in nature (mCRC), and up to 50% of CRC will progress to the metastatic stage. The implementation of new systemic treatment regimens have significantly improved overall survival, with median survival for patients with mCRC having increased from 12 to 21 months in the decade from 2010 to 2020 [2]. Similarly, the 5-year survival rate between 2011 and 2017 for mCRC reached 14.0%, from 9.6% in the year 2000 [3]. Gallichio’s et al. recent work confirmed the trend [4], with findings of individuals with mCRC living longer; and about one-fifth of individuals diagnosed with mCRC have lived with the metastatic disease for 10 or more years postdiagnosis. However, the same work suggested that the numbers of individuals living with metastatic cancers, including mCRC, are projected to continue to rise [4], highlighting the need for further investigation to develop effective approaches for medical intervention. The main clinical manifestations of mCRC include rectal bleeding, microcytic anaemia, altered bowel movements (constipation, diarrhoea or faecal incontinence), abdominal pain and fatigue [5, 6]. Patients also experience multiple psychosocial and emotional wellbeing impacts affecting their health-related quality of life (HRQoL) [6]. A previous qualitative study has focused on the development of a conceptual framework for locally recurrent rectal cancer (LRRC) where patient-reported symptoms and sexual, psychological, social and role functioning impacts were highlighted as affecting patients’ HRQoL [7]. Additionally, past qualitative studies on CRC have documented evidence on symptoms affecting patients’ HRQoL, including rectal bleeding, constipation and diarrhoea [8, 9]. HRQoL measures are predictive of survival in treatment-refractory mCRC and can serve as an alternative, but equally important, endpoint in the mCRC population [10]. Nevertheless, there was no or very sparse data available of the experience of patients with mCRC, particularly on the aspects that were most relevant and bothersome to them on an everyday basis. Thus, the objectives of this study were to gain a better understanding of the experience of patients with mCRC and identify the most bothersome of those signs, symptoms and impacts, from the patients themselves. Patients and methods In order to identify the signs, symptoms and impacts most relevant and bothersome to patients experiencing mCRC and to develop a strategy capable of informing future clinical development plans [11] and clinical practice, a literature review and patient interviews were conducted. Literature review A literature review was conducted in April 2020 in PubMed and Cochrane databases to identify and document key patient experiences (signs, symptoms and impacts, jointly referred to as concepts) related to mCRC. A search string was developed to capture patient-centred studies detailing concepts experienced by patients with mCRC and anal cancer, as well as patients’ quality of life, health function and general health perception. Studies providing information on concepts reported by patients with mCRC and detailing aspects of patient experience with the disease or patient reported outcome (PRO) measures were considered eligible for detailed analysis. Amongst those, qualitative studies were prioritized as they provide a description of concepts directly from the patients themselves, with no external bias, intervention or external judgement. Studies focusing exclusively on non-metastatic disease, purely physiological measures or not including either evidence on patient experience or patient reported outcomes were excluded. All titles and abstracts were screened by an experienced researcher with relevance of selected publications confirmed by a second researcher. Details on the study, including design, disease, and methodology, were extracted. Likewise, the concepts reported by patients, as well as the prevalence and attributions to disease or treatment, if available, were also captured. Information available in patient blogs and patient association websites was used to complement and cross-analyse results from the literature search. Five patient association websites focusing on patient experiences with mCRC (Bowel Cancer Australia, Colorectal Cancer Alliance, Bowel Cancer UK, Fight Colorectal Cancer and Colon Cancer Coalition) were analysed. Quotes detailing patients’ experience were collected for cross-checking the reported concepts with those identified from the literature review. Based on the findings from the review, a preliminary disease conceptual model was developed highlighting the disease-related, treatment-related and disease-/treatment-related signs and symptoms and immediate- and general-impacts reported by patients with mCRC and anal cancer. Patient interviews Then, semi-structured patient interviews were conducted to characterize the patient experience of mCRC. The study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice and applicable regulatory requirements. The study and all interview materials received Independent Review Board (IRB) approval. Patient recruitment was conducted by Global Perspectives, via direct contact achieved through its proprietary networking process, including (but not limited to) clinician referrals, social media postings and patient advocacy groups. In order to participate in the interview, participants signalled their availability by contacting Global Perspectives and were required to meet the following predefined selection criteria: diagnosis of Stage IV or mCRC, aged between 18 and 75 years old, residing in the USA (Puerto Rico excluded), able to communicate in English, willing and able to electronically sign an informed consent form, have access to an internet-connected computer or mobile device, have access to a telephone or cell phone and able to complete a 90-min interview. Exclusion criteria included the following: have another malignancy or bowel issue that would confound results; have a history of significant neurological events or medical, psychological, cognitive or mental conditions that would interfere with the ability to participate or confound study results; and having tested positive for COVID-19 in 2 weeks prior to the interview or being exposed to an individual presenting COVID-19 symptomatology. A balance between male and female patients and education level was sought to account for the heterogenicity of experiences due to these factors and reduce selection bias. Concept elicitation phone interviews [12] were conducted by a trained and experienced interviewer using a semi-structured interview guide that ensured the interviewer covered the interview content and the study objectives. The guide was specifically developed for the study, based on the literature review findings. Open-ended questions and probes were used to explore the patient experience with mCRC and whether the concepts reported were disease-related, treatment-related or a combination of both. More specifically, patients were asked about how the disease manifested, including signs, symptoms and domains and ways of their everyday lives that were affected (i.e. impacts) by those signs and symptoms and how this experience has changed over time if applicable. Patients’ experience of fatigue and changes in bowel movements were discussed in detail given the lack of specificity found in the literature. If needed, probes were used to help patients further describe their experience and ensure the objectives of the study were covered. Patients were also asked to provide a “bothersome rating” for each concept, on a scale of 0 to 10, with 0 being not bothersome at all and 10 being extremely bothersome. The top 3 most bothersome signs/symptoms and impacts were discussed in further detail with each patient. Finally, patients were asked about the consequences in their day-to-day living if this symptom/sign or impact improved. Data analysis All interviews were audio-recorded, transcribed verbatim and subject to thematic analysis using qualitative analysis software ATLAS.ti (Germany, version 8.2.3). Before initiating the analysis, a preliminary codebook that captured all concepts and feature types (e.g. severity; frequency) of the interview guide was developed. Two coders, trained in qualitative coding, analysed the transcripts. In a first step, to test for inter-coder agreement (ICA), the two analysts independently coded several transcripts before reaching ICA. After an acceptable level of ICA (> 0.7 as pre-defined) was reached, transcripts were coded by a single coder, and one transcript from each wave was reviewed by the second coder to ensure consistency and quality. Along the analysis, the team, composed of the coders, interviewer, sponsor, scientific lead and project manager, conducted alignment meetings regularly to ensure the codes made sense based on patients’ responses and to alleviate coding discrepancies. Code language and groupings were refined, and the codebook revised as needed to remain up-to-date. Salience of the sign, symptom, and impact concepts were determined using both the total number of patients mentioning a concept and the average bothersome rating reported by patients for the specific concept. If the concept was mentioned by more than 50% of patients (13 patients out of the 25) and the average bothersome rating reported for the concept was equal to or higher than 5 out of 10, the concept was reported as being “salient”. Saturation of concepts was assessed to ensure adequate sample size for a comprehensive and representative picture of the disease that could be generalizable to the adult population of patients with mCRC. “Saturation” is defined as the point at which additional patient interviews do not contribute to the identification of unique concepts or new information [13]. To assess saturation, participants’ transcripts were grouped into 4 waves organized chronologically. Waves 1 to 3 included 6 patients, while wave 4 included 7 patients. Concepts mentioned in each wave of interviews were compared with the concepts mentioned in the previous waves. Saturation was deemed reached when no new concept or new information relevant to the research objective occurred within a set of interviews and confirmed with the following one(s). The preliminary conceptual model obtained from the literature was updated with the identified signs, symptoms and impacts identified from the interviews. Results Literature review and development of the preliminary disease conceptual model The literature review yielded 710 relevant articles of which 32 were selected for full review. Of the 32 articles analysed, 19 were analysed in depth and used to inform the development of the preliminary disease conceptual model. From the 19 articles, 6 were patient interviews/focus groups studies, 2 were literature reviews, 7 relied on medical records analysis, 3 used PRO questionnaires/surveys and 1 presented conclusions from a working group of experts. Out of these studies, 63 signs and symptoms and 22 impacts were identified. Amongst the signs and symptoms, 26 were disease-related, and 13 were treatment-related; 24 could not be attributed solely to disease or to treatment and were considered both disease- and treatment-related. Rectal bleeding, change in bowel habits, abdominal pain, constipation and weight loss were the 5 most prevalent signs and symptoms related to disease solely, while nausea, fatigue, pain and loss of appetite were the 4 most prevalent disease- and treatment-related symptoms, and paraesthesia and hair loss were highlighted as the 2 most prevalent treatment-related symptoms. Prevalent impacts could be divided onto immediate impacts and included interference with daily activities and general impacts, such as reduced ability to work, difficulty caring for children/family issues and financial difficulties. The search of patient association websites confirmed the mentioning of all prevalent concepts identified in the literature. Using the evidence from the literature, a preliminary disease conceptual model of patient’s experience with mCRC was developed. Concept elicitation interviews In total, 25 patients with mCRC were interviewed. Interview duration was between 75 and 90 min. The sociodemographic and clinical characteristics of patients are presented in Table 1. The average patient age was 52 years (range: 26–72) and most patients were female (68%). Almost all patients were White (92%), and 18 (72%) patients had a higher education (bachelor’s/graduate degree, or higher).Table 1 Sociodemographic and clinical characteristics of the patient sample (N = 25) Wave 1 (n = 6) Wave 2 (n = 6) Wave 3 (n = 6) Wave 4 (n = 7) Total (N = 25) (%) Gender (n) Male 2 1 1 4 8 (32%) Female 4 5 5 3 17 (68%) Age (years) Average 50 48 54 54 52 Range 41–63 38–56 43–72 26–67 26–72 Ethnicity (n) White 6 6 6 5 23 (92%) Asian - - - 1 1 (4%) Asian/American - - - 1 1 (4%) Education level (n) High school - - 1 1 2 (8%) College (finished/unfinished) 1 2 - - 3 (12%) Associate’s degree 2 - - - 2 (8%) Bachelor’s/graduate degree 1 3 2 4 10 (40%) Master’s degree 1 1 3 2 7 (28%) PhD 1 - - - 1 (4%) Cancer diagnosis (n) Colon 3 1 4 1 9 (36%) Rectal 1 1 - 1 3 (12%) Colorectal 2 4 2 5 13 (52%) Stage at diagnosis (n) Stage I - 1 1 1 3 (12%) Stage II 1 - - - 1 (4%) Stage III 1 4 1 2 8 (32%) Stage IV 4 1 4 4 13 (52%) Current treatment (n) Chemotherapy 4 4 4 4 16 (64%) No treatment receiving currently 2 2 2 3 9 (36%) Past treatment (n) Chemotherapy 1 - - 1 2 (8%) Chemotherapy and surgery 2 2 5 3 12 (48%) Chemotherapy, radiation and surgery 3 4 1 3 11 (44%) Stage I–III cancer are localized; stage IV cancer is metastatic/advanced Thematic analysis of the transcripts identified 58 unique signs and symptoms from the interviews. Out of the 58 signs and symptoms reported by patients with mCRC, 38 (66%) appeared in wave 1, 13 (22%) in wave 2, 4 (7%) in wave 3 and 3 (5%) in wave 4. Of the 58, 8 signs and symptoms were mentioned by more than 50% of the patients (N > 12). These were fatigue (N = 24, 96%), nausea (N = 21, 84%), neuropathy (N = 21, 84%), diarrhoea (N = 20, 80%), loss of appetite (N = 19, 76%), constipation (N = 18, 72%), weight loss (N = 15, 60%) and abdominal pain (N = 15, 60%). Thirty-seven signs and symptoms were considered highly bothersome by patients (average bothersome rating equal to or higher than 5 out of 10), with defecatory urgency, fever and sexual dysfunction being amongst the most bothersome. Eight signs and symptoms were considered salient: fatigue, nausea, neuropathy, diarrhoea, loss of appetite, constipation, weight loss and abdominal pain (Fig. 1).Fig. 1 Salient signs and symptoms reported by patients with mCRC (N = 25) Patients provided descriptions detailing their experience related to the salient symptoms of mCRC, particularly in terms of their everyday life consequences. Fatigue was attributed by patients both to the treatment and the disease and was described as constant and highly bothersome. Fatigue had implications for patients’ sleep patterns and affected their levels of activity. While most of the patients made a distinction between fatigue, weakness and tiredness, 5 considered fatigue and tiredness to be the same, but different than weakness, and 5 others considered fatigue, weakness and tiredness as the same concept. Patients highlighting a difference between the concepts referred to fatigue as something that would make them unable to stand long enough to do basic activities or would make them fall asleep, while feeling weakness was a result of the inability to exercise or eat well. Patients reported that improvements in fatigue would lead to a reduced need to sleep during the day and increased levels of activity. Finally, some patients described the feeling of fatigue as something that became “normal” for their routine as time passed:So… unfortunately I always will have this kind of… is like kind of my…my normal fatigue. I mean, even when I’m not on treatments now I just consistently felt fatigued ever, you know, probably since 2017 or so. You know, so there’s that constant fatigue... [US_18] Nausea was primarily attributed to the treatments and was characterized as a highly impactful symptom which could lead to vomiting in severe cases. Nausea caused several implications for patients’ quality of life, including difficulties in working and introducing discomfort in social and family settings:When the nausea’s been at its worst, it's about an 8 or a 9. And I think I had to recognize it's been so impactful because of what I just said about the teaching… when I was working in front of the kids, it's like…it was hard to hide the fact that I was carrying around one of those blue vomit bags in my pocket. [US_24] Neuropathy was described by patients as a constant sensation of numbness or tingling, akin to burning or the sensation of pins or needles piercing into the skin. Patients related the experience of pain to neuropathy, and linked impacts on activities of daily living, such as driving or going outdoors to this symptom:…most affected me was the peripheral neuropathy…it was so bad to the point that, if I’d go into the refrigerator, my hands would burn… going outdoors in the winter, any exposed skin would feel…burning... [US_14] Diarrhoea was attributed by patients both to the disease and treatments. Patients reported the significant burden of needing to go to the bathroom unexpectedly several times a day:…started having really frequent bowel movements… Before I started chemo, my baseline was…bowel movement after every meal….I would eat…it would trigger…my GI system….I would have to go….very initially, after my surgery, I was going 6 or 7 times a day. [US_23] Similarly, patients attributed constipation to both the disease and treatments. Patients mentioned that constipation led to the mCRC diagnosis in some cases. When asked for more details about changes in bowel movements, patients described them as a fluctuation between diarrhoea and constipation, often as a result of their treatments, as well as an increased urgency to empty the bowels or a clustering of bowel movements. Patients reported that improvements in bowel symptoms (either constipation or diarrhoea) led to reduced worry about being constantly near a bathroom:I do have a lot of constipation, like I said, for three days, every time I have chemo….before it switches back the other way…I have to take things to combat that … that becomes probably my biggest symptom and problem, is getting that balance with my bowel movements… [US_04] Loss of appetite was reported by patients mainly because of the treatments. Patients mentioned that the chemotherapy treatments would lead to loss of appetite and taste alterations:Yeah, I believe that’s just chemo related. I get a week off from chemo. So next week, my eating will start picking back up and stuff. Then the taste kind of…my taste starts coming back and stuff. [US_20] Weight loss was attributed by patients to the chemotherapy treatments and most specifically to the surgical procedures, with severe weight loss experienced during the recovery from surgery. This weight loss was exponentiated by the lack of activity during recovery and loss of appetite. It was considered a highly bothersome symptom due to the patients’ wish to be healthy and fit to fight the disease:During chemo I lost about 10 pounds. After the first surgery I lost another 10…So when I was first diagnosed I was around 128 pounds. By the time after my first surgery, I dropped down to about 102 pounds... [US_19] Patients attributed abdominal pain to the disease and post-surgery recovery and described it as severe and as having a major impact on their quality of life. Patients mentioned that the frequency of the pain was not constant and would often fluctuate:…my whole abdomen just ached, and it would come in these waves… it was pretty intense…I couldn’t…lay down without discomfort… and when I was at night so I could barely sleep the pain was so bad. [US_18] Thirty-five impacts were experienced and reported by patients during the interviews. Out of the 35 impacts reported by patients with mCRC, 30 (86%) appeared in wave 1, 2 (6%) in wave 2, 2 (6%) in wave 3 and 1 (3%) in wave 4. Out of the 35 impacts reported, 7 impacts were mentioned by more than 50% of the patients (N > 12). These were reduced ability to work (N = 21, 84%), interference with daily activities (N = 14, 56%), cognitive functioning (N = 14, 56%), financial impact (N = 13, 52%), sleep changes (N = 13, 52%), impact on social life (N = 13, 52%) and difficulties with mobility/walking difficulties (N = 12, 48%). Furthermore, 25 impacts were considered highly bothersome by patients (average bothersome rating equal to or higher than 5 out of 10), with family issues, impact on clothing (patients were not able to dress properly mainly due to ostomy), distress and loss of self-identity amongst the most bothersome. Seven impacts were considered salient (Fig. 2). These were reduced ability to work, interference with daily activities, impact in cognitive functioning, financial impact, sleep changes, impact on social life and walking difficulties. Patients provided detailed descriptions of the salient impacts affecting their quality of life.Fig. 2 Salient impacts reported by patients with mCRC (N = 25) Reduced ability to work was attributed by patients to a combination of symptoms including fatigue, nausea, neuropathy and impaired cognitive functioning. Chemotherapy side effects introduced work difficulties for patients. Patients mentioned that their reduced ability to work led to disruptions in their lives:I was a senior executive in the federal government… I essentially took a year off from work…I didn’t have to go through all the work-related stress, and my job was very stressful. But also, it meant that my life was totally disrupted and I was basically sitting home and had nothing I had to do all day. [US_35] Patients reported interference with daily activities which was mainly linked with side effects from the chemotherapy and fatigue. According to patients’ experience, interference with daily activities was related to the inability to cook, do yard work, do chores, engage in sports, reduced ability to drive and difficulty getting out of bed and climbing stairs:When I go on vacation I use a walker, because I get so tired, I have to be able to sit down. I have a housekeeper that they come in twice a month to clean. My husband travels a lot for work and I just can't do everything… I don’t ride my bike anymore… [US_10] Patients reported impacts in cognitive functioning which were linked with side effects from the chemotherapy (e.g. “chemo brain”). Patients reported that cognitive functioning problems had an impact on their work productivity and overall quality of life:Cognition problems is number one, absolutely. I still work. I’m an IT director... I used to be able to multitask, and that is very important in my job. I can’t maintain that level of cognition anymore, so that has been the biggest impact in my life. [US_31] Financial impacts were reported by patients mainly associated with the high treatment costs and with the coverage provided by insurance. Financial problems were exacerbated by patients having to work reduced hours and having to spend savings and retirement funds:….cutting your salary in half is hard. Luckily, I have been able to get disability benefits…..it’s a struggle with bureaucracy. I’m not approved, oh, but you are approved… and as of September 1st, I won’t have health insurance anymore because I’m not working 30 hours a week… [US_18] Impact on sleep was reported mainly in relation to the chemotherapy side effects, abdominal pain and location of cancer metastases. Patients described sleep changes as the inability to have a restful sleep and as waking up multiple times during the night:I have had cancer on my chest wall which is like right around my ribcage on my right side, so …. I couldn’t get comfortable because of the cancer there… So when you can’t get comfortable you can’t really sleep. [US_12] Patients experienced impacts on social life associated with bowel symptoms, reduced levels of energy to participate in social events due and lack of interest in socializing with other people due to the cancer diagnosis:I don’t want people feeling sorry for me. I don’t want people hovering over me like it’s my last day on Earth. I don’t want everybody to do everything for me. I’ve got a mind. I’ve got a body. I can do things… Yes, it has affected relationships. [US_09] Difficulties with mobility and walking were experienced by patients mainly due to symptoms of weakness and lack of physical ability to conduct activities:I could hardly walk. Getting up to go to the bathroom in the middle of the night was impossible, you know, you crawled, but then that turned to a soreness that felt like somebody had taken a bat and beaten the bottom of your feet… [US_36] Final disease conceptual model The evidence generated from the concepts elicited by patients during the interviews was used to refine the preliminary disease conceptual model and to focus this model on the signs, symptoms and impacts experienced by patients with mCRC specifically. Largely, the signs, symptoms and impacts mentioned by the patients during the interviews were consistent with findings from the literature review and that pertained to mCRC (Fig. 3). More specifically, all the concepts that were found mCRC-specific from the literature review and that were elicited during the interviews with the patients with mCRC were identified and informed the development of the final disease conceptual model; newly spontaneously mentioned concept were added; concept were rewording to better classify the symptom according to patient experience (e.g. “paraesthesia” was reworded as “neuropathy” based on patient own words).Fig. 3 Disease conceptual model of the experience of adult patients with metastatic colorectal cancer While a small number of sign-, symptom- and impact-related concepts were still being identified during the last wave of interviews with patients (3 symptoms and 1 impact in wave 4), the concepts which were found to be salient were addressed mainly in the first wave of interviews conducted. The saturation results (for both signs/symptoms and impacts) provide confidence that the concepts included in the final model comprehensively reflect the experience of mCRC patients. Discussion This study aimed to document the experience of patients with mCRC and identify the signs, symptoms and impacts most relevant and bothersome for those patients. Documenting the most relevant and bothersome concepts reported by patients is useful to inform patient centric initiatives while contributing to the innovative development of drugs [11]. An initial review of the literature identified a set of signs, symptoms and impacts characterizing mCRC which informed the development of a preliminary disease conceptual model of mCRC. Rectal bleeding, change in bowel habits, abdominal pain, constipation, weight loss, nausea, fatigue, pain, loss of appetite, paraesthesia, hair loss and impacts including interference with daily activities, reduced ability to work, difficulty caring for children, family issues and financial difficulties were considered most relevant based on the reported prevalence. Most of the evidence pertaining to the most prevalent signs, symptoms and impacts collected in the literature review was confirmed during the concept elicitation interviews. Fatigue, nausea, neuropathy, loss of appetite, constipation, weight loss and abdominal pain were confirmed as salient concepts. The other salient concept of “diarrhoea” reported by patients in the interviews is likely to be included in the symptom “change in bowel habits” identified from the literature which patients characterized both by periods of constipation alternating with periods of diarrhoea. Nevertheless, since patients reported “change in bowel habits” and both “constipation” and “diarrhoea” separately during the interviews, for the purposes of the refinement of the disease conceptual model, “constipation” and “diarrhoea” appear independently of each other. Signs and symptoms reported in this study were in agreement with findings from previous qualitative studies in colorectal cancer. Fatigue, diarrhoea, constipation, abdominal pain and weight loss were reported by the majority of patients in a study exploring reasons for diagnostic delays in colorectal cancer [8]. Neuropathy, nausea and loss of appetite were also highlighted by patients in another qualitative study measuring mindfulness interventions in colorectal cancer patients [14]. Similarly, symptoms such as fatigue, pain, nausea, weight loss and HRQoL impacts have been reported in a broader study where physicians provided their insights on concepts often reported in patients undergoing chemotherapy treatments in various cancer indications, including CRC [15]. The findings from the patient interviews confirm these symptoms as relevant, while also providing evidence on how patients perceive them as bothersome for their HRQoL. The concept elicitation interviews also allowed reduced ability to work, interference with daily activities and financial impact to be confirmed as salient impacts. Additionally, though previously identified and published, but not considered prevalent in the literature, impact on cognitive functioning, sleep changes, impact on social life and walking difficulties were most frequently reported and rated most bothersome by patients during the interviews. Impact on daily activities, financial impact, impacts on occupational status, social life and locomotor difficulties were previously highlighted in a conceptual framework built for LLRC [7]. Sleep disturbance and memory/concentration issues were also reported in the literature [14]. Saturation analysis could conclude with confidence that the signs, symptoms and impacts included in the final disease conceptual model, in particular the salient ones, comprehensively reflected the experience of patients with mCRC. From the eight salient signs and symptoms reported by patients, abdominal pain was the most bothersome concept reported with a score of 8.3 out of 10. Other pain-related concepts such as cramping, pain (cancer related) and rectal pain, change in bowel habits and defecatory urgency, fever, soreness and dryness in the hands and feet, cold sweats and hot flashes, running nose and sexual dysfunction were rated highly bothersome (> 8 out of 10) but mentioned by less than 50% of patients during the interviews. Many of these highly bothersome symptoms, such as pain, the various gastrointestinal symptoms experienced and sexual problems, were previously described in the literature [7, 14], further highlighting their relevance in patients with mCRC. Despite not being salient, considering how bothersome they are according to the patients, they should not be disregarded when defining the endpoint strategy of future clinical development programs and should on the contrary be considered depending on the expected profile and benefits of the treatment. From the seven salient impacts reported by patients, walking difficulties was the most bothersome, with a score of 10 out of 10. This score highlights the impairments in movement experienced by patients and their general lack of independence impacting their quality of life. Impacts in the patients’ cognitive functioning and their reduced ability to work were also considered highly bothersome by patients (> 8 out of 10) for their overall quality of life. Emotional functioning impacts, family impacts and clothing difficulties, were additional impacts considered as highly bothersome (> 8 out of 10) but mentioned by less than 50% of the patients. This information is coherent with the evidence available in the literature describing the emotional burden of cancer for patients [16, 17]. This study may present some limitations. There is a risk for potential selection bias due to the recruitment strategy employed. Since patients were recruited through a combination of clinician referrals, social media posts and patient advocacy groups and had to contact recruiter global perspectives to express their willingness to participate in a qualitative interview, the study sample may have been more attuned to signs, symptoms and impacts of disease than the general mCRC population. Also, the interviews were conducted with 25 subjects, and although this may be considered a small sample size, it is aligned with qualitative research studies, and saturation analysis confirmed the appropriateness of the sample size. Likewise, since all patients had to have access to internet to participate, the potential barriers to internet access, such as educational level, age and socioeconomic status, may limit the generalizability of these findings in comparison to the general mCRC population. Furthermore, patients were all based in the USA, the majority were White (92%), and all apart from one had college-level education, possibly introducing further selection bias and posing a challenge to the generalizability of the findings to different populations. Amongst the strengths of this study, since patients had to confirm their diagnosis with their treating physician, there was solid confirmation that patients had the disease at the time of interview. Patients also reported multiple stages of the disease at diagnosis and multiple metastasis sites which may contribute to the plurality of experiences in reporting the signs, symptoms and impacts of mCRC with significant granularity and specificity. To our knowledge, no conceptual model specific to mCRC was available at the time of the study, only one in LLCR, based on qualitative evidence [7]. Many of the concepts of the LLCR model are also identified in our study. The main advantage of our study is that it considers both the frequency of mentions and the bothersome/impactful ratings reported by patients to establish the main signs, symptoms and impacts experienced in the context of mCRC. Our study provides relevant insights into patients’ experiences of mCRC. This study highlights the signs, symptoms and impacts that are most commonly experienced by patients and, amongst those, the most bothersome and impactful concepts that should be captured and assessed in future clinical development and education programs in mCRC. Their identification will help define patient-reported measurement strategies for clinical trials and development programs aiming to develop innovative drugs capable of reducing patient’s burden with signs, symptoms and impacts of mCRC while increasing patients’ overall quality of life. It will also help inform future development programs and studies to better understand mCRC management in the real world, as well as improve management and decision-making in clinical practice for these patients with mCRC. Acknowledgements The authors of the manuscript would like to thank Tiago Maia and Teresa Obis (IQVIA) for their support for the medical writing, Anna de la Motte (IQVIA) who conducted the patient interviews and performed their analysis and Shweta Bhardwaj who conducted the project throughout its duration. Author contribution Isabelle Guillemin substantially contributed to the conception and design of the work and to the analysis and interpretation of data, contributed drafting the work and revising it critically for important intellectual content, also contributed to the final approval of the version to be published and was accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Mahesh Darpelly contributed substantially to the acquisition, analysis and interpretation of data for the work; also contributed to revising the work critically for important intellectual content and to the final approval of the version to be published; and was accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Brendon Wong, Anders Ingelgård and Ingolf Griebsch contributed substantially to the conception and design of the work and to the analysis and interpretation of data, contributed revising the work critically for important intellectual content, also contributed to the final approval of the version to be published and was accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Funding This study was sponsored by Boehringer Ingelheim. Data availability Data are available upon reasonable request. Declarations Ethical approval The study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice and applicable regulatory requirements. The study and all interview materials received Independent Review Board (IRB) approval. Competing interests Isabelle Guillemin and Mahesh Darpelly are employees of IQVIA. IQVIA conducted the study analyses. IQVIA were paid consultants to Boehringer Ingelheim in connection with the development of this manuscript. Brendon Wong, Anders Ingelgård and Ingolf Griebsch are employees of Boehringer Ingelheim. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials of the study apart from those disclosed. What is already known about this subject? • Patients with metastatic colorectal cancer (mCRC) experience a wide range of signs, symptoms, and impacts, both disease-related, treatment-related, and a combination of both. What does this study add? • This qualitative study is in line with the FDA-related patient-centric research. • The study provides understanding of the patient experience with mCRC and highlights the level of burden of relevant symptoms and impacts through the conduct of concept elicitation interviews with patients with mCRC. • The concepts identified from the interviews helped develop a disease conceptual model of the patient experience with mCRC. How might this impact on clinical development programs and clinical practice? • This qualitative study can inform the patient-reported outcomes (PRO) strategy in future clinical development programs evaluating the benefits of new mCRC therapies by providing evidence for the development/selection of PRO measures capable of assessing signs, symptoms and impacts most relevant for mCRC patients. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Bray F Ferlay J Soerjomataram I Siegel RL Torre LA Jemal A Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries CA Cancer J Clin 2018 68 6 394 424 10.3322/caac.21492 30207593 2. Levin-Sparenberg E Bylsma LC Lowe K Sangare L Fryzek JP Alexander DD A systematic literature review and meta-analysis describing the prevalence of KRAS, NRAS, and BRAF gene mutations in metastatic colorectal cancer Gastroenterol Res 2020 13 5 184 198 10.14740/gr1167 3. Surveillance, epidemiology, and end results program [Internet]. SEER. Available from: https://seer.cancer.gov/index.html. Accessed 19 July 2022. 4. Gallicchio L Devasia TP Tonorezos E Mollica MA Mariotto A Estimation of the number of individuals living with metastatic cancer in the United States JNCI J Natl Cancer Inst. 2022 22 djac158 5. Biller LH Schrag D Diagnosis and treatment of metastatic colorectal cancer: a review JAMA 2021 325 7 669 685 10.1001/jama.2021.0106 33591350 6. Carduff E, Kendall M, Murray SA. Living and dying with metastatic bowel cancer: serial in-depth interviews with patients. Eur J Cancer Care (Engl). 2018;27(1):e12653. 7. Harji DP Koh C Solomon M Velikova G Sagar PM Brown J Development of a conceptual framework of health-related quality of life in locally recurrent rectal cancer Colorectal Dis Off J Assoc Coloproctology G B Irel 2015 17 11 954 964 8. Oberoi DV Jiwa M McManus A Hodder R de Nooijer J Help-seeking experiences of men diagnosed with colorectal cancer: a qualitative study Eur J Cancer Care (Engl) 2016 25 1 27 37 10.1111/ecc.12271 25521505 9. Mosher CE Adams RN Helft PR O’Neil BH Shahda S Rattray NA Family caregiving challenges in advanced colorectal cancer: patient and caregiver perspectives Support Care Cancer Off J Multinatl Assoc Support Care Cancer 2016 24 5 2017 2024 10. Gong J Wu D Chuang J Tuli R Simard J Hendifar A Moving beyond conventional clinical trial end points in treatment-refractory metastatic colorectal cancer: a composite quality-of-life and symptom control end point Clin Ther 2017 39 11 2135 2145 10.1016/j.clinthera.2017.09.015 29079389 11. Patient-focused drug development guidance: methods to identify what is important to patients and select, develop or modify fit-for-purpose clinical outcome assessments [Internet]. U.S. Food and Drug Administration. 2019. Available from: https://www.fda.gov/drugs/news-events-human-drugs/patient-focused-drug-development-guidance-methods-identify-what-important-patients-and-select. Accessed 19 July 2022. 12. Cheng KKF Clark AM Qualitative methods and patient-reported outcomes: measures development and adaptation Int J Qual Methods 2017 16 1 160940691770298 10.1177/1609406917702983 13. Patrick DL Burke LB Gwaltney CJ Leidy NK Martin ML Molsen E Content validity–establishing and reporting the evidence in newly developed patient-reported outcomes (PRO) instruments for medical product evaluation: ISPOR PRO good research practices task force report: part 1–eliciting concepts for a new PRO instrument Value Health J Int Soc Pharmacoeconomics Outcomes Res 2011 14 8 967 977 10.1016/j.jval.2011.06.014 14. Atreya CE Kubo A Borno HT Rosenthal B Campanella M Rettger JP Being Present: A single-arm feasibility study of audio-based mindfulness meditation for colorectal cancer patients and caregivers PLoS ONE 2018 13 7 e0199423 10.1371/journal.pone.0199423 30036361 15. Cella D Paul D Yount S Winn R Chang CH Banik D What are the most important symptom targets when treating advanced cancer? A survey of providers in the National Comprehensive Cancer Network (NCCN) Cancer Invest 2003 21 4 526 535 10.1081/CNV-120022366 14533442 16. Caruso R Nanni MG Riba MB Sabato S Grassi L The burden of psychosocial morbidity related to cancer: patient and family issues Int Rev Psychiatry Abingdon Engl 2017 29 5 389 402 10.1080/09540261.2017.1288090 17. Yates P Family coping: issues and challenges for cancer nursing Cancer Nurs 1999 22 1 63 71 10.1097/00002820-199902000-00012 9990760

36515691

PMC9748881

NO-CC CODE

2022-12-15 23:22:47

no

J Cancer Surviv. 2022 Dec 14;:1-11

utf-8

J Cancer Surviv

10.1007/s11764-022-01279-9

oa_other

==== Front Contemp Probl Ecol Contemp Probl Ecol Contemporary Problems of Ecology 1995-4255 1995-4263 Pleiades Publishing Moscow 8267 10.1134/S1995425522060129 Article Biology of Scutellaria baicalensis Georgi (Lamiaceae) from Different Ecological and Geographical Places of Growth during Introduction Pshenichkina Yu. A. scutel@yandex.ru grid.465435.5 0000 0004 0487 2025 Central Siberian Botanical Garden, Siberian Branch, Russian Academy of Sciences, 630090 Novosibirsk, Russia 14 12 2022 2022 15 6 653658 12 5 2022 20 6 2022 21 6 2022 © Pleiades Publishing, Ltd. 2022, ISSN 1995-4255, Contemporary Problems of Ecology, 2022, Vol. 15, No. 6, pp. 653–658. © Pleiades Publishing, Ltd., 2022.Russian Text © The Author(s), 2022, published in Sibirskii Ekologicheskii Zhurnal, 2022, No. 6, pp. 707–713. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The variability in the development of Scutellaria baicalensis Georgi (Lamiaceae) plants collected from natural habitats (Zabaykalsky krai, Amur oblast, and Primorye) and grown under the same culturing conditions (Novosibirsk) is analyzed. It has been found that interpopulation differences in morphological characteristics of S. baicalensis and the timing of the onset of phenophases developed in nature are preserved under new growing conditions. Data analysis shows the existence of significant differences (t > 3) between the steppe Zabaykalsky and forest Primorye coenopolations (CPs) in plant height, number of pairs of leaves, and number of shoots, both in nature and during introduction. The flowering period in the Zabaykalsky CPs comes earlier than in the Primorye CP, both in nature and in culture. Individuals of Zabaykalsky CPs bloom in early July in culture. The flowering period in the Primorye CP begins at the end of July–August; the seed ripening period is extended, especially in the first years of introduction. It is possible to allow the formation of steppe Zabaykalsky and forest Primorye ecotypes. Keywords: Scutellaria baicalensis Georgi climate seasonal development introduction issue-copyright-statement© Pleiades Publishing, Ltd. 2022 ==== Body pmcThe study of plant behavior under climate change in different parts of the range is important both for understanding microevolutionary processes within species; the possibility of predicting plant development processes; and the practical application in horticulture, agriculture, forestry, and the fight against invasive species (Korsakova, 2019; Park et al., 2019). Introduced plants, entering new conditions of temperature amplitudes, moisture regimes, spectral compositions of light, soil cover, and other abiotic factors, begin to experience unusual effects of seasonal climate rhythms, which cannot but affect their growth, development, and productivity. Scutellaria baicalensis Georgi (family Lamiaceae) is a tap-rooted herbaceous perennial plant with sympodially growing generative monocyclic shoots. The range of the species is Mongolian–Daurian–Manchurian. S. baicalensis is widespread in China (Zhili, Shentung), northeastern Mongolia, Dauria, Manchuria, and Japan (Illyustrirovannaya…, 2009). Three of its fragments enter the territory of Russia: in Eastern Zabaykalsky (Zabaykalsky krai), the Middle Amur Region (Amur oblast), and southwestern Primorye (Primorsky krai). The species is a valuable medicinal plant used both in folk and modern medicine in different countries. Flavonoids are its main active ingredients. S. baicalensis has a wide spectrum of pharmacological activity. It is used as an antihypertensive and sedative drug and as a hemostimulator in antitumor therapy (Goldberg et al., 1994; Razina and Pshenichkina, 1989). Extracts have antioxidant and anti-inflammatory activity (Gao et al, 1999; Yoon et al., 2020) and a hepatoprotective effect (Potapova, 2016), as well as an antibacterial effect against a number of pathogenic species of human microorganisms (Kim et al., 2009). It was also found that the active substances of S. baicalensis in vitro act on the SARS-CoV19 virus, blocking the viral attack and preventing the proliferation of the virus (Song et al., 2020). A number of researchers have studied the ecological, biological, and phytocoenotic features; age structure; and biological and operational stock of species populations, ontogeny, raw and seed productivity, biological activity, and flavonoid content of coenopolations (CPs) of S. baicalensis in Zabaykalsky krai, Amur oblast, and Primorsky krai (Banaeva, 1994; Bukhasheeva et al., 2007; Manyakhin, 2010; Shishmarev, 2012). The study of the development of the species under the same cultural conditions is of particular interest, because the studied CPs were located in different geographical points differing in natural and climatic conditions. The purpose of this study is to analyze the variability of morphological characteristics and the rhythm of seasonal development of S. baicalensis specimens collected in different ecological and geographical conditions of growth during introduction. MATERIAL AND METHODS To study the development under the same culture conditions, plant seeds were collected in natural CPs of S. baicalensis of Zabaykalsky Krai (near the villages of Bishigino, 51°51′12″ N 116°26′28″ E; Borzya 50°23′ N 116°32′ E, and Verkhnie Klyuchi 51°57′41″ N 116°45′56″ E, and the Krasnokamensk railway station 50°04′21″ N 118°13′42″ E), Amur oblast (near the city of Svobodny 51°23′ N 128°08′ E), Primorsky krai (near the village of Komissarovo 44°59′24″ N 131°47′05″ E) and then sown in the experimental plot of the Central Siberian Botanical Garden (CSBG), Novosibirsk (TsSBS 54°49′33″ N 83°06′34″ E). Phenological observations were carried out in 1987–2005 according to (Beideman, 1974). The average daily air temperature and precipitation during the periods of passage of the phenophases of the species were analyzed according to the average long-term data (Archive…, 2022; Weather…, 2022). Location of meteorological stations: Borzya, 50°40′ N 116°50′ E, 676 m a.s.l. (Zabaykalsky krai); Svobodny 51°45′ N 128°10′ E, 200 m a.s.l. (Amur oblast); the settlement of Pogranichny 44°40′ N 131°30′ E, 211 m a.s.l. (Primorsky krai); Novosibirsk planetarium 54°98′ N 83°03′ E, 160 m a.s.l. (Novosibirsk oblast). RESULTS The study areas differ in climatic characteristics (Fig. 1). Fig. 1. Climatic characteristics of the study areas. The climate of Zabaykalsky krai is sharply continental with insufficient moisture, a long duration of sunshine per year, and significant fluctuations in daily and annual air temperatures. Winter is cold and long, with little snow; spring is warm, short, dry, windy; and summer is warm, dry in the first half, with precipitation in the second. The predominance of direct solar radiation causes intense heating of the soil surface, which contributes to the early thawing of the upper horizons. The average annual air temperature in the study area is negative (about –1.5°С). Frosts are possible in early June. About 350 mm of precipitation falls per year. Up to 80% of the annual rainfall falls in July–August. In winter, precipitation is rare; the depth of snow cover does not exceed 10–15 cm. The growing season is about 150 days. Chestnut, chernozem soils predominate, meadow-chernozem, gray forest soils are common, solonchaks, solonetzes and meadow-alkaline soils are often found. Amur oblast is characterized by an ultracontinental climate with monsoon features, a significant amount of sunshine, and large amplitudes of daily and annual temperatures. Winter is cold, with little snow; there are frequent sharp temperature changes associated with the invasion of cyclones in spring; summers are mostly hot. The average annual air temperature in the study area is –0.2°С. The growing season lasts an average of 140 days. The last frosts are possible until the end of April and the first at the end of September. The annual rainfall is over 550 mm. Precipitation falls mainly during the warm period. In winter, the average snow cover reaches 12 cm. Brown forest and podzolic-brown forest soils predominate in the study areas. The climate of the cis-Khanka Plain of Primorsky krai is monsoonal. This region is characterized by uneven precipitation and periodic droughts, especially in spring and the first half of summer, and large fluctuations in seasonal and daily temperatures. Winter is with little snow, spring is cool and dry, and summer is hot. The average annual air temperature is 4.8°C. The last frosts are observed at the end of April, but they are possible in June. The average annual rainfall is over 450 mm. Precipitation often falls in the form of showers, mainly in the second half of July–August. The duration of the growing season is about 188 days. The height of the snow cover is 10–12 cm. Soils are meadow–soddy, meadow gley podzolized, and brown–podzolic. CSBG is located in the forest-steppe zone of Western Siberia in a moderately cool and moderately humid agroclimatic region. The climate of the forest-steppe zone is characterized by sharp continentality with significant fluctuations in seasonal and daily temperatures. Winters are long and cold; summers are short and hot. The average annual air temperature is 2.6°С. The last spring frosts are observed in the first decade of June; the first autumn frosts are observed in the second half of September. The growing season lasts an average of 158 days. The growing season begins in late April–early May. Precipitation in May falls in the form of rain and sleet. Annual rainfall averages about 400 mm. The average snow depth is 35 cm. The soils on the territory of the CSBG are predominantly soddy–podzolic and gray forest (Рh 5.5–6.9) (Rastitel’noe…, 2014). The natural features of Zabaykalsky contributed to the formation of local steppe vegetation. The Zabaykalsky CPs were located in open steppe areas and slopes of hills. Individuals of the Amur and Primorye CPs grew under more mesophytic conditions. The Amur CP Svobodny was confined to the forest-steppe zone. The Primorye CP Komissarovo was confined to the oak woodlands. The formation of generative organs, flowering, and ripening of fruits and seeds are the most important phases of plant development during the growing season. According to the Popov Herbarium of the Central Siberian Botanical Garden, Siberian Branch, Russian Academy of Sciences (CSBG SB RAS) (NSK), Novosibirsk; the Krasnoborov Herbarium, CSBG SB RAS (NS), Novosibirsk; the Krylov Herbarium of Tomsk State University (TK); the Herbarium of the Institute of Biology and Soil Science of the Far Eastern Branch, Russian Academy of Sciences (FEB RAS) (VLA), Vladivostok; collections of employees of the Pacific Institute of Bioorganic Chemistry of the FEB RAS, Vladivostok; the Herbarium of the Regional Chita Museum of Local Lore (Lekarstvennye…, 1990); and the author’s own observations, it was found that the onset of phenophases in individuals of S. baicalensis in Zabaykalsky occurs two weeks earlier than in Primorye. Therefore, plant budding is observed in Zabaykalsky in the second decade of June; flowering is observed in July and until mid-August. In Primorye, plants bud in early July and flowering occurs in mid-July. During the introduction, the following phenological states were identified for individuals of all studied CPs: spring regrowth, budding, flowering, fruiting, and summer–autumn regrowth. Features of the rhythm of development of S. baicalensis individuals in different CPs under the same culture conditions manifested in different periods of onset and duration of individual phenophases (Fig. 2). Fig. 2. Phenological spectra of S. baicalensis during introduction on average for 1987–2005. DISCUSSION It can be seen from the presented data that the climatic conditions of natural habitats and the area of introduction of S. baicalensis differ significantly from each other in the sum of positive temperatures, the amount of precipitation, and the duration of the growing season. Morphological characteristics of specimens of natural and introduced CPs are given in (Pshenichkina and Pshenichkin, 2018). Data analysis shows the existence of significant differences (t > 3) between the steppe Zabaykalsky and forest Primorye CPs in plant height, the number of shoots per individual, and the number of pairs of leaves per shoot, both in nature and during introduction. The height of the plants of the Zabaykalsky CP group averaged 37 cm, the number of shoots was 17–18, and the number of pairs of leaves was 17. Individuals of the Primorye CP Komissarovo were taller (on average 65 cm in height), with a smaller number of shoots (up to 13 pieces) and pairs of leaves (up to 15 pieces). Characteristics of individuals of the Amur CP Svobodny have intermediate values. The ontogenesis of S. baicalensis has four periods and nine age states: seedlings, juvenile, immature, virginal, young and middle-aged generative, senile, and quasi-senile (Banaeva, 2000; Bukhasheeva, 2000). A number of studies have revealed the influence of regional and local climatic factors on the ontogenetic structure of S. baicalensis (Sandanov et al., 2017; Sandanov and Rosbakh, 2019). It was determined that seed renewal in S. baicalensis is significantly reduced with an increase in continentality and, accordingly, an increase in the aridity of the climate, which leads to a decrease in the number of individuals of the pregenerative period and an increase in the proportion of virginal and senile individuals. Our studies have shown that the maximum in the age spectrum in the Zabaykalsky CP group falls on generative young individuals (40–60% of the total number of individuals). The Amur and Primorye CPs have two-peak spectra. The first maximum falls on the immature age group (Svobodny 30%; Komissarovo 17%) and the second—Svobodny for the generative young (27%) and Komissarovo for the generative middle-aged group (33%). The course of ontogeny during introduction is the same for all CPs transferred from different ecological and geographical conditions of growth. However, there is a sharp acceleration of ontogeny in culture caused by a reduction in the length of time an individual stays in a particular age state. These processes are typical for introduced species. As early as in the first year of life in culture, S. baicalensis specimens of all studied CPs bloom and produce seeds, whereas the transition of individuals to the generative state in nature begins at 10–15 years of age. Observations have shown that the timing of the onset of phenological phases, in particular, flowering (in the first year of life, when plants go through the stages of development from a seedling to a generative young state) occurs several days later than in subsequent years for this CP. As can be seen from the phenospectra (Fig. 2), the time from spring regrowth to the beginning of flowering is not the same for different CPs. The shortest period (35 days) is observed in plants of the Zabaykalsky group of CPs. These plants grew in rather harsh conditions and did not require much heat to transition to flowering. The sum of positive temperatures averaged 45°C for Novosibirsk, 43°C for Zabaykalsky krai, 50°C for the Amur oblast, and 40°C for Primorye. The average June temperature is 16.3°C in Novosibirsk, 15.5°C in Zabaykalsky region, 17.8°C in the Amur oblast, and 13.0°C in Primorye. The average air temperature in Primorye reaches 17.5°C only in July, which can explain the later flowering of S. baicalensis there compared to Zabaykalsky krai. Individuals cultured in Zabaykalsky CPs bloom within 7–20 days from the beginning of July. Plants growing in Primorye have developed a greater demand for heat and moisture in the process of evolution, which affected the culture conditions on the duration of the period from regrowth to flowering, which increased to 40 days. Plants of the Primorye CP Komissarovo, both transferred to the culture and growing in nature, bloomed later than the Zabaykalsky plants (in late July–August). Their flowering period was more extended (up to 60 days). Apparently, the selection in the process of evolution in the harsh conditions of Zabaykalsky took place with the shortening of the growing season. Precocity was even more evident in culture. Thus, the flowering of the main shoot in plants of the Zabaykalsky group growing in nature occurs in July, while in culture conditions it happens much earlier (from the end of June). The difference in the timing of flowering in the CPs over the years ranged from 3 to 5 days, depending on weather conditions. A reduction in the duration of phenophases, as well as earlier dates for their onset, were observed during the study for all plants, both Zabaykalsky, Amur, and Primorye. This is explained by the changes in the regional climate caused by global warming. Thus, it was found that the average daily air temperature in Novosibirsk increased by 0.16°C over the period from 1996–2015, the vegetation period lengthened by 12 days, and the active vegetation period of plants increased by 8 days (Fomin and Fomina, 2021). There was a shift in the limits of variability of S. baicalensis in the prefloral period and the duration of the flowering period. However, interpopulation differences in the timing of the onset of phenophases between the steppe Zabaykalsky and forest Primorye CPs persisted with the introduction of S. baicalensis. CONCLUSIONS It has been shown that S. baicalensis specimens collected in different ecological and geographical habitats in culture retain interpopulation differences in some morphological features (plant height, number of pairs of leaves per shoot, and number of shoots per individual) and the timing of the onset of phenophases. The formation of steppe Zabaykalsky and forest Primorye ecotypes can be assumed. The adaptive capabilities of these culturing populations in the new conditions are shown, and the reserve of their hereditary variability is mobilized. ACKNOWLEDGMENTS Material from the Bioresource Scientific Collection of the UNU “Collections of Living Plants in Open and Protected Ground” of the Central Siberian Botanical Garden, Siberian Branch, Russian Academy of Sciences, USU 440534, was used for preparing the article. FUNDING This work was carried out as part of State Task АААА-А21-121011290027-6 under the project “Theoretical and Applied Aspects of Studying the Gene Pools of Natural Plant Populations and Conservation of Plant Diversity outside the Typical Habitat (Ex Situ).” COMPLIANCE WITH ETHICAL STANDARDS This article does not contain any research involving humans or animals as research objects. CONFLICT OF INTERESTS The author declares no conflict of interest. Translated by M. Shulskaya ==== Refs REFERENCES 1 Archive and weather forecast. http://weatherarchive.ru. Cited November 15, 2022. 2 Banaeva, Yu.A., Baikal skullcap (Scutellaria baicalensis Georgi) (ecology, biology, introduction), Extended Abstract Cand. Sci. (Biol.) Dissertation, Novosibirsk, 1994. 3 Banaeva Yu.A. Ontogeny of the Baikal Skullcap (Scutellaria baicalensis Georgi), Ontogeneticheskii atlas lekarstvennykh rastenii 2000 Ioshkar-Ola Ontogenetic Atlas of Medicinal Plants 4 Beideman, I.N., Metodika izucheniya fenologii rastenii i rastitel’nykh soobshchestv (Methods of Studying Plant Phenology and Plant Communities), Novosibirsk: Nauka, 1974. 5 Bukhasheeva, T.G., Ecological and biological features of Scutellaria baicalensis in Transbaikalia, Extended Abstract Cand. Sci. (Biol.) Dissertation, Ulan-Ude, 2000. 6 Bukhasheeva T.G. Sandanov D.V. Aseeva T.A. Shishmarev V.M. Chirikova N.K. Age structure of coenopopulations and raw phytomass of Scutellaria baicalensis Georgi (Lamiaceae) Rast. Resur 2007 43 23 32 7 Fomin E.S. Fomina T.I. Changes in the phenology of perennial plants in Western Siberia against the background of global warming Contemp. Probl. Ecol 2021 14 434 445 10.1134/S199542552105005X 8 Gao Z. Huang K. Yang X. Xu H. Free radical scavenging and antioxidant activities of flavonoids extracted from the radix of Scutellaria baicalensis Georgi Biochim. Biophys. Acta 1999 1472 643 650 10.1016/S0304-4165(99)00152-X 10564778 9 Gol’dberg, V.E., Dygai, A.M., Litvinenko, V.I., Popova, T.P., and Suslov, N.I., Shlemnik baikal’skii. Fitokhimiya i farmakologicheskie svoistva (Scullcap (Scutellaria baikalensis): Phytochemistry and Pharmacological Properties), Tomsk: Tomsk. Univ., 1994. 10 Illyustrirovannaya entsiklopediya rastitel’nogo mira Sibiri (Illustrated Encyclopedia of Flora of Siberia), Artemov, I.A. and Sedel’nikov, V.P., Eds., Novosibirsk: Arta, 2009. 11 Kim Y.-H. Paek J.-Y. Kwon H.-J. Lee J.-W. Yoon O.-H. Han M.-D. Antioxidant and antibacterial activities of ethyl acetate extract from Scutellaria baicalensis Korean J. Food Nutr 2009 22 367 376 12 Korsakova, S.P., Methodological foundations of ecological modeling and forecasting of plant response to climate change, Extended Abstract Doctoral (Biol.) Dissertation, Yalta, 2019. 13 Lekarstvennye rasteniya Chitinskoi oblasti (Iz gerbariya Chitinskogo oblastnogo kraevedcheskogo muzeya): Katalog (Medicinal Plants of the Chita Region (From the Herbarium of the Chita Regional Museum of Local Lore): Catalog), Antonova, G.F., Chita: Oblastnaya Tipografiya, 1990. 14 Manyakhin, A.Yu., Baikal skullcap (Scutellaria baicalensis Georgi) in the south of Primorsky Krai (introduction, composition of flavonoids, biological activity), Extended Abstract Cand. Sci. (Biol.) Dissertation, Vladivostok, 2010. 15 Park I. Jones A. Mazer S.J. PhenoForecaster: A software package for the prediction of flowering phenology Appl. Plant Sci 2019 7 e1230 10.1002/aps3.1230 16 Potapova, A.A., Nephro- and hepatoprotective effect of a dry extract from the Baikal skullcap (Scutellaria baicalensis Georgi) and its water-soluble form in concomitant lesions of the liver and kidneys, Extended Abstract Cand. Sci. (Pharm.) Dissertation, Volgograd, 2016. 17 Pshenichkina Yu.A. Contemp. Probl. Ecol. 2018 11 221 226 10.1134/C1995425518020105 18 Rastitel’noe mnogoobrazie Tsentral’nogo sibirskogo botanicheskogo sada SO RAN (Plant Diversity of the Central Siberian Botanical Garden Sib. Dep. Ross. Acad. Sci.), Koropachinskii, I.Yu. and Banaev, E.V., Novosibirsk: Geo, 2014. 19 Razina T.G. Pshenichkina Yu.A. Dependence of antitumor activity of tincture from rhizomes with roots of Scutellaria baicalensis Georgi on vegetation phase and plant age Rast. Resur 1989 2 247 249 20 Sandanov D.V. Rosbakh S. Demographic structure of Scutellaria baicalensis Georgi depending on climatic gradients and local factors Russ. J. Ecol 2019 50 404 407 10.1134/S1067413619040131 21 Sandanov, D.V., Naidanov, B.B., and Shishmarev, V.M., Influence of regional and local environmental factors on the distribution and population structure of Scutellaria baicalensis Georgi, Vestn. Tomsk. Gos. Univ., 2017, no. 38, pp. 89–103. 10.17223/19988591/38/5 22 Shishmarev, V.M., Ecological and biological assessment of populations of Scutellaria baicalensis Georgi and Pteridium aquilinum (L.) Kuhn in Transbaikalia, Extended Abstract Cand. Sci. (Biol.) Dissertation, Ulan-Ude, 2012. 23 Song J.-W. Long J.-Y. Xie L. Zhang L.-L. Xie Q.-X. Chen H.-J. Deng M. Li X.-F. Applications, phytochemistry, pharmacological effects, pharmacokinetics, toxicity of Scutellaria baicalensis Georgi. and its probably potential therapeutic effects on COVID-19: a review Chin. Med 2020 15 102 10.1186/s13020-020-00384-0 32994803 24 Weather and climate. http://www.pogodaiklimat.ru/ ar-chive.php. Cited November 15, 2022. 25 Yoon E.-J. Lee M.Y. Choi B.I. Lim K.J. Hong S.Y. Park D. Pharmaceutical advantages of Ge-noTX-407, a combination of extracts from Scutellaria baicalensis Root and Magnolia officinalis bark Antioxidants 2020 11 1111 10.3390/antiox9111111

0

PMC9748882

NO-CC CODE

2022-12-15 23:22:47

no

Contemp Probl Ecol. 2022 Dec 14; 15(6):653-658

utf-8

Contemp Probl Ecol

10.1134/S1995425522060129

oa_other

==== Front Contemp Probl Ecol Contemp Probl Ecol Contemporary Problems of Ecology 1995-4255 1995-4263 Pleiades Publishing Moscow 8267 10.1134/S1995425522060129 Article Biology of Scutellaria baicalensis Georgi (Lamiaceae) from Different Ecological and Geographical Places of Growth during Introduction Pshenichkina Yu. A. scutel@yandex.ru grid.465435.5 0000 0004 0487 2025 Central Siberian Botanical Garden, Siberian Branch, Russian Academy of Sciences, 630090 Novosibirsk, Russia 14 12 2022 2022 15 6 653658 12 5 2022 20 6 2022 21 6 2022 © Pleiades Publishing, Ltd. 2022, ISSN 1995-4255, Contemporary Problems of Ecology, 2022, Vol. 15, No. 6, pp. 653–658. © Pleiades Publishing, Ltd., 2022.Russian Text © The Author(s), 2022, published in Sibirskii Ekologicheskii Zhurnal, 2022, No. 6, pp. 707–713. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The variability in the development of Scutellaria baicalensis Georgi (Lamiaceae) plants collected from natural habitats (Zabaykalsky krai, Amur oblast, and Primorye) and grown under the same culturing conditions (Novosibirsk) is analyzed. It has been found that interpopulation differences in morphological characteristics of S. baicalensis and the timing of the onset of phenophases developed in nature are preserved under new growing conditions. Data analysis shows the existence of significant differences (t > 3) between the steppe Zabaykalsky and forest Primorye coenopolations (CPs) in plant height, number of pairs of leaves, and number of shoots, both in nature and during introduction. The flowering period in the Zabaykalsky CPs comes earlier than in the Primorye CP, both in nature and in culture. Individuals of Zabaykalsky CPs bloom in early July in culture. The flowering period in the Primorye CP begins at the end of July–August; the seed ripening period is extended, especially in the first years of introduction. It is possible to allow the formation of steppe Zabaykalsky and forest Primorye ecotypes. Keywords: Scutellaria baicalensis Georgi climate seasonal development introduction issue-copyright-statement© Pleiades Publishing, Ltd. 2022 ==== Body pmcThe study of plant behavior under climate change in different parts of the range is important both for understanding microevolutionary processes within species; the possibility of predicting plant development processes; and the practical application in horticulture, agriculture, forestry, and the fight against invasive species (Korsakova, 2019; Park et al., 2019). Introduced plants, entering new conditions of temperature amplitudes, moisture regimes, spectral compositions of light, soil cover, and other abiotic factors, begin to experience unusual effects of seasonal climate rhythms, which cannot but affect their growth, development, and productivity. Scutellaria baicalensis Georgi (family Lamiaceae) is a tap-rooted herbaceous perennial plant with sympodially growing generative monocyclic shoots. The range of the species is Mongolian–Daurian–Manchurian. S. baicalensis is widespread in China (Zhili, Shentung), northeastern Mongolia, Dauria, Manchuria, and Japan (Illyustrirovannaya…, 2009). Three of its fragments enter the territory of Russia: in Eastern Zabaykalsky (Zabaykalsky krai), the Middle Amur Region (Amur oblast), and southwestern Primorye (Primorsky krai). The species is a valuable medicinal plant used both in folk and modern medicine in different countries. Flavonoids are its main active ingredients. S. baicalensis has a wide spectrum of pharmacological activity. It is used as an antihypertensive and sedative drug and as a hemostimulator in antitumor therapy (Goldberg et al., 1994; Razina and Pshenichkina, 1989). Extracts have antioxidant and anti-inflammatory activity (Gao et al, 1999; Yoon et al., 2020) and a hepatoprotective effect (Potapova, 2016), as well as an antibacterial effect against a number of pathogenic species of human microorganisms (Kim et al., 2009). It was also found that the active substances of S. baicalensis in vitro act on the SARS-CoV19 virus, blocking the viral attack and preventing the proliferation of the virus (Song et al., 2020). A number of researchers have studied the ecological, biological, and phytocoenotic features; age structure; and biological and operational stock of species populations, ontogeny, raw and seed productivity, biological activity, and flavonoid content of coenopolations (CPs) of S. baicalensis in Zabaykalsky krai, Amur oblast, and Primorsky krai (Banaeva, 1994; Bukhasheeva et al., 2007; Manyakhin, 2010; Shishmarev, 2012). The study of the development of the species under the same cultural conditions is of particular interest, because the studied CPs were located in different geographical points differing in natural and climatic conditions. The purpose of this study is to analyze the variability of morphological characteristics and the rhythm of seasonal development of S. baicalensis specimens collected in different ecological and geographical conditions of growth during introduction. MATERIAL AND METHODS To study the development under the same culture conditions, plant seeds were collected in natural CPs of S. baicalensis of Zabaykalsky Krai (near the villages of Bishigino, 51°51′12″ N 116°26′28″ E; Borzya 50°23′ N 116°32′ E, and Verkhnie Klyuchi 51°57′41″ N 116°45′56″ E, and the Krasnokamensk railway station 50°04′21″ N 118°13′42″ E), Amur oblast (near the city of Svobodny 51°23′ N 128°08′ E), Primorsky krai (near the village of Komissarovo 44°59′24″ N 131°47′05″ E) and then sown in the experimental plot of the Central Siberian Botanical Garden (CSBG), Novosibirsk (TsSBS 54°49′33″ N 83°06′34″ E). Phenological observations were carried out in 1987–2005 according to (Beideman, 1974). The average daily air temperature and precipitation during the periods of passage of the phenophases of the species were analyzed according to the average long-term data (Archive…, 2022; Weather…, 2022). Location of meteorological stations: Borzya, 50°40′ N 116°50′ E, 676 m a.s.l. (Zabaykalsky krai); Svobodny 51°45′ N 128°10′ E, 200 m a.s.l. (Amur oblast); the settlement of Pogranichny 44°40′ N 131°30′ E, 211 m a.s.l. (Primorsky krai); Novosibirsk planetarium 54°98′ N 83°03′ E, 160 m a.s.l. (Novosibirsk oblast). RESULTS The study areas differ in climatic characteristics (Fig. 1). Fig. 1. Climatic characteristics of the study areas. The climate of Zabaykalsky krai is sharply continental with insufficient moisture, a long duration of sunshine per year, and significant fluctuations in daily and annual air temperatures. Winter is cold and long, with little snow; spring is warm, short, dry, windy; and summer is warm, dry in the first half, with precipitation in the second. The predominance of direct solar radiation causes intense heating of the soil surface, which contributes to the early thawing of the upper horizons. The average annual air temperature in the study area is negative (about –1.5°С). Frosts are possible in early June. About 350 mm of precipitation falls per year. Up to 80% of the annual rainfall falls in July–August. In winter, precipitation is rare; the depth of snow cover does not exceed 10–15 cm. The growing season is about 150 days. Chestnut, chernozem soils predominate, meadow-chernozem, gray forest soils are common, solonchaks, solonetzes and meadow-alkaline soils are often found. Amur oblast is characterized by an ultracontinental climate with monsoon features, a significant amount of sunshine, and large amplitudes of daily and annual temperatures. Winter is cold, with little snow; there are frequent sharp temperature changes associated with the invasion of cyclones in spring; summers are mostly hot. The average annual air temperature in the study area is –0.2°С. The growing season lasts an average of 140 days. The last frosts are possible until the end of April and the first at the end of September. The annual rainfall is over 550 mm. Precipitation falls mainly during the warm period. In winter, the average snow cover reaches 12 cm. Brown forest and podzolic-brown forest soils predominate in the study areas. The climate of the cis-Khanka Plain of Primorsky krai is monsoonal. This region is characterized by uneven precipitation and periodic droughts, especially in spring and the first half of summer, and large fluctuations in seasonal and daily temperatures. Winter is with little snow, spring is cool and dry, and summer is hot. The average annual air temperature is 4.8°C. The last frosts are observed at the end of April, but they are possible in June. The average annual rainfall is over 450 mm. Precipitation often falls in the form of showers, mainly in the second half of July–August. The duration of the growing season is about 188 days. The height of the snow cover is 10–12 cm. Soils are meadow–soddy, meadow gley podzolized, and brown–podzolic. CSBG is located in the forest-steppe zone of Western Siberia in a moderately cool and moderately humid agroclimatic region. The climate of the forest-steppe zone is characterized by sharp continentality with significant fluctuations in seasonal and daily temperatures. Winters are long and cold; summers are short and hot. The average annual air temperature is 2.6°С. The last spring frosts are observed in the first decade of June; the first autumn frosts are observed in the second half of September. The growing season lasts an average of 158 days. The growing season begins in late April–early May. Precipitation in May falls in the form of rain and sleet. Annual rainfall averages about 400 mm. The average snow depth is 35 cm. The soils on the territory of the CSBG are predominantly soddy–podzolic and gray forest (Рh 5.5–6.9) (Rastitel’noe…, 2014). The natural features of Zabaykalsky contributed to the formation of local steppe vegetation. The Zabaykalsky CPs were located in open steppe areas and slopes of hills. Individuals of the Amur and Primorye CPs grew under more mesophytic conditions. The Amur CP Svobodny was confined to the forest-steppe zone. The Primorye CP Komissarovo was confined to the oak woodlands. The formation of generative organs, flowering, and ripening of fruits and seeds are the most important phases of plant development during the growing season. According to the Popov Herbarium of the Central Siberian Botanical Garden, Siberian Branch, Russian Academy of Sciences (CSBG SB RAS) (NSK), Novosibirsk; the Krasnoborov Herbarium, CSBG SB RAS (NS), Novosibirsk; the Krylov Herbarium of Tomsk State University (TK); the Herbarium of the Institute of Biology and Soil Science of the Far Eastern Branch, Russian Academy of Sciences (FEB RAS) (VLA), Vladivostok; collections of employees of the Pacific Institute of Bioorganic Chemistry of the FEB RAS, Vladivostok; the Herbarium of the Regional Chita Museum of Local Lore (Lekarstvennye…, 1990); and the author’s own observations, it was found that the onset of phenophases in individuals of S. baicalensis in Zabaykalsky occurs two weeks earlier than in Primorye. Therefore, plant budding is observed in Zabaykalsky in the second decade of June; flowering is observed in July and until mid-August. In Primorye, plants bud in early July and flowering occurs in mid-July. During the introduction, the following phenological states were identified for individuals of all studied CPs: spring regrowth, budding, flowering, fruiting, and summer–autumn regrowth. Features of the rhythm of development of S. baicalensis individuals in different CPs under the same culture conditions manifested in different periods of onset and duration of individual phenophases (Fig. 2). Fig. 2. Phenological spectra of S. baicalensis during introduction on average for 1987–2005. DISCUSSION It can be seen from the presented data that the climatic conditions of natural habitats and the area of introduction of S. baicalensis differ significantly from each other in the sum of positive temperatures, the amount of precipitation, and the duration of the growing season. Morphological characteristics of specimens of natural and introduced CPs are given in (Pshenichkina and Pshenichkin, 2018). Data analysis shows the existence of significant differences (t > 3) between the steppe Zabaykalsky and forest Primorye CPs in plant height, the number of shoots per individual, and the number of pairs of leaves per shoot, both in nature and during introduction. The height of the plants of the Zabaykalsky CP group averaged 37 cm, the number of shoots was 17–18, and the number of pairs of leaves was 17. Individuals of the Primorye CP Komissarovo were taller (on average 65 cm in height), with a smaller number of shoots (up to 13 pieces) and pairs of leaves (up to 15 pieces). Characteristics of individuals of the Amur CP Svobodny have intermediate values. The ontogenesis of S. baicalensis has four periods and nine age states: seedlings, juvenile, immature, virginal, young and middle-aged generative, senile, and quasi-senile (Banaeva, 2000; Bukhasheeva, 2000). A number of studies have revealed the influence of regional and local climatic factors on the ontogenetic structure of S. baicalensis (Sandanov et al., 2017; Sandanov and Rosbakh, 2019). It was determined that seed renewal in S. baicalensis is significantly reduced with an increase in continentality and, accordingly, an increase in the aridity of the climate, which leads to a decrease in the number of individuals of the pregenerative period and an increase in the proportion of virginal and senile individuals. Our studies have shown that the maximum in the age spectrum in the Zabaykalsky CP group falls on generative young individuals (40–60% of the total number of individuals). The Amur and Primorye CPs have two-peak spectra. The first maximum falls on the immature age group (Svobodny 30%; Komissarovo 17%) and the second—Svobodny for the generative young (27%) and Komissarovo for the generative middle-aged group (33%). The course of ontogeny during introduction is the same for all CPs transferred from different ecological and geographical conditions of growth. However, there is a sharp acceleration of ontogeny in culture caused by a reduction in the length of time an individual stays in a particular age state. These processes are typical for introduced species. As early as in the first year of life in culture, S. baicalensis specimens of all studied CPs bloom and produce seeds, whereas the transition of individuals to the generative state in nature begins at 10–15 years of age. Observations have shown that the timing of the onset of phenological phases, in particular, flowering (in the first year of life, when plants go through the stages of development from a seedling to a generative young state) occurs several days later than in subsequent years for this CP. As can be seen from the phenospectra (Fig. 2), the time from spring regrowth to the beginning of flowering is not the same for different CPs. The shortest period (35 days) is observed in plants of the Zabaykalsky group of CPs. These plants grew in rather harsh conditions and did not require much heat to transition to flowering. The sum of positive temperatures averaged 45°C for Novosibirsk, 43°C for Zabaykalsky krai, 50°C for the Amur oblast, and 40°C for Primorye. The average June temperature is 16.3°C in Novosibirsk, 15.5°C in Zabaykalsky region, 17.8°C in the Amur oblast, and 13.0°C in Primorye. The average air temperature in Primorye reaches 17.5°C only in July, which can explain the later flowering of S. baicalensis there compared to Zabaykalsky krai. Individuals cultured in Zabaykalsky CPs bloom within 7–20 days from the beginning of July. Plants growing in Primorye have developed a greater demand for heat and moisture in the process of evolution, which affected the culture conditions on the duration of the period from regrowth to flowering, which increased to 40 days. Plants of the Primorye CP Komissarovo, both transferred to the culture and growing in nature, bloomed later than the Zabaykalsky plants (in late July–August). Their flowering period was more extended (up to 60 days). Apparently, the selection in the process of evolution in the harsh conditions of Zabaykalsky took place with the shortening of the growing season. Precocity was even more evident in culture. Thus, the flowering of the main shoot in plants of the Zabaykalsky group growing in nature occurs in July, while in culture conditions it happens much earlier (from the end of June). The difference in the timing of flowering in the CPs over the years ranged from 3 to 5 days, depending on weather conditions. A reduction in the duration of phenophases, as well as earlier dates for their onset, were observed during the study for all plants, both Zabaykalsky, Amur, and Primorye. This is explained by the changes in the regional climate caused by global warming. Thus, it was found that the average daily air temperature in Novosibirsk increased by 0.16°C over the period from 1996–2015, the vegetation period lengthened by 12 days, and the active vegetation period of plants increased by 8 days (Fomin and Fomina, 2021). There was a shift in the limits of variability of S. baicalensis in the prefloral period and the duration of the flowering period. However, interpopulation differences in the timing of the onset of phenophases between the steppe Zabaykalsky and forest Primorye CPs persisted with the introduction of S. baicalensis. CONCLUSIONS It has been shown that S. baicalensis specimens collected in different ecological and geographical habitats in culture retain interpopulation differences in some morphological features (plant height, number of pairs of leaves per shoot, and number of shoots per individual) and the timing of the onset of phenophases. The formation of steppe Zabaykalsky and forest Primorye ecotypes can be assumed. The adaptive capabilities of these culturing populations in the new conditions are shown, and the reserve of their hereditary variability is mobilized. ACKNOWLEDGMENTS Material from the Bioresource Scientific Collection of the UNU “Collections of Living Plants in Open and Protected Ground” of the Central Siberian Botanical Garden, Siberian Branch, Russian Academy of Sciences, USU 440534, was used for preparing the article. FUNDING This work was carried out as part of State Task АААА-А21-121011290027-6 under the project “Theoretical and Applied Aspects of Studying the Gene Pools of Natural Plant Populations and Conservation of Plant Diversity outside the Typical Habitat (Ex Situ).” COMPLIANCE WITH ETHICAL STANDARDS This article does not contain any research involving humans or animals as research objects. CONFLICT OF INTERESTS The author declares no conflict of interest. Translated by M. Shulskaya ==== Refs REFERENCES 1 Archive and weather forecast. http://weatherarchive.ru. Cited November 15, 2022. 2 Banaeva, Yu.A., Baikal skullcap (Scutellaria baicalensis Georgi) (ecology, biology, introduction), Extended Abstract Cand. Sci. (Biol.) Dissertation, Novosibirsk, 1994. 3 Banaeva Yu.A. Ontogeny of the Baikal Skullcap (Scutellaria baicalensis Georgi), Ontogeneticheskii atlas lekarstvennykh rastenii 2000 Ioshkar-Ola Ontogenetic Atlas of Medicinal Plants 4 Beideman, I.N., Metodika izucheniya fenologii rastenii i rastitel’nykh soobshchestv (Methods of Studying Plant Phenology and Plant Communities), Novosibirsk: Nauka, 1974. 5 Bukhasheeva, T.G., Ecological and biological features of Scutellaria baicalensis in Transbaikalia, Extended Abstract Cand. Sci. (Biol.) Dissertation, Ulan-Ude, 2000. 6 Bukhasheeva T.G. Sandanov D.V. Aseeva T.A. Shishmarev V.M. Chirikova N.K. Age structure of coenopopulations and raw phytomass of Scutellaria baicalensis Georgi (Lamiaceae) Rast. Resur 2007 43 23 32 7 Fomin E.S. Fomina T.I. Changes in the phenology of perennial plants in Western Siberia against the background of global warming Contemp. Probl. Ecol 2021 14 434 445 10.1134/S199542552105005X 8 Gao Z. Huang K. Yang X. Xu H. Free radical scavenging and antioxidant activities of flavonoids extracted from the radix of Scutellaria baicalensis Georgi Biochim. Biophys. Acta 1999 1472 643 650 10.1016/S0304-4165(99)00152-X 10564778 9 Gol’dberg, V.E., Dygai, A.M., Litvinenko, V.I., Popova, T.P., and Suslov, N.I., Shlemnik baikal’skii. Fitokhimiya i farmakologicheskie svoistva (Scullcap (Scutellaria baikalensis): Phytochemistry and Pharmacological Properties), Tomsk: Tomsk. Univ., 1994. 10 Illyustrirovannaya entsiklopediya rastitel’nogo mira Sibiri (Illustrated Encyclopedia of Flora of Siberia), Artemov, I.A. and Sedel’nikov, V.P., Eds., Novosibirsk: Arta, 2009. 11 Kim Y.-H. Paek J.-Y. Kwon H.-J. Lee J.-W. Yoon O.-H. Han M.-D. Antioxidant and antibacterial activities of ethyl acetate extract from Scutellaria baicalensis Korean J. Food Nutr 2009 22 367 376 12 Korsakova, S.P., Methodological foundations of ecological modeling and forecasting of plant response to climate change, Extended Abstract Doctoral (Biol.) Dissertation, Yalta, 2019. 13 Lekarstvennye rasteniya Chitinskoi oblasti (Iz gerbariya Chitinskogo oblastnogo kraevedcheskogo muzeya): Katalog (Medicinal Plants of the Chita Region (From the Herbarium of the Chita Regional Museum of Local Lore): Catalog), Antonova, G.F., Chita: Oblastnaya Tipografiya, 1990. 14 Manyakhin, A.Yu., Baikal skullcap (Scutellaria baicalensis Georgi) in the south of Primorsky Krai (introduction, composition of flavonoids, biological activity), Extended Abstract Cand. Sci. (Biol.) Dissertation, Vladivostok, 2010. 15 Park I. Jones A. Mazer S.J. PhenoForecaster: A software package for the prediction of flowering phenology Appl. Plant Sci 2019 7 e1230 10.1002/aps3.1230 16 Potapova, A.A., Nephro- and hepatoprotective effect of a dry extract from the Baikal skullcap (Scutellaria baicalensis Georgi) and its water-soluble form in concomitant lesions of the liver and kidneys, Extended Abstract Cand. Sci. (Pharm.) Dissertation, Volgograd, 2016. 17 Pshenichkina Yu.A. Contemp. Probl. Ecol. 2018 11 221 226 10.1134/C1995425518020105 18 Rastitel’noe mnogoobrazie Tsentral’nogo sibirskogo botanicheskogo sada SO RAN (Plant Diversity of the Central Siberian Botanical Garden Sib. Dep. Ross. Acad. Sci.), Koropachinskii, I.Yu. and Banaev, E.V., Novosibirsk: Geo, 2014. 19 Razina T.G. Pshenichkina Yu.A. Dependence of antitumor activity of tincture from rhizomes with roots of Scutellaria baicalensis Georgi on vegetation phase and plant age Rast. Resur 1989 2 247 249 20 Sandanov D.V. Rosbakh S. Demographic structure of Scutellaria baicalensis Georgi depending on climatic gradients and local factors Russ. J. Ecol 2019 50 404 407 10.1134/S1067413619040131 21 Sandanov, D.V., Naidanov, B.B., and Shishmarev, V.M., Influence of regional and local environmental factors on the distribution and population structure of Scutellaria baicalensis Georgi, Vestn. Tomsk. Gos. Univ., 2017, no. 38, pp. 89–103. 10.17223/19988591/38/5 22 Shishmarev, V.M., Ecological and biological assessment of populations of Scutellaria baicalensis Georgi and Pteridium aquilinum (L.) Kuhn in Transbaikalia, Extended Abstract Cand. Sci. (Biol.) Dissertation, Ulan-Ude, 2012. 23 Song J.-W. Long J.-Y. Xie L. Zhang L.-L. Xie Q.-X. Chen H.-J. Deng M. Li X.-F. Applications, phytochemistry, pharmacological effects, pharmacokinetics, toxicity of Scutellaria baicalensis Georgi. and its probably potential therapeutic effects on COVID-19: a review Chin. Med 2020 15 102 10.1186/s13020-020-00384-0 32994803 24 Weather and climate. http://www.pogodaiklimat.ru/ ar-chive.php. Cited November 15, 2022. 25 Yoon E.-J. Lee M.Y. Choi B.I. Lim K.J. Hong S.Y. Park D. Pharmaceutical advantages of Ge-noTX-407, a combination of extracts from Scutellaria baicalensis Root and Magnolia officinalis bark Antioxidants 2020 11 1111 10.3390/antiox9111111

36370021

PMC9748884

NO-CC CODE

2022-12-15 23:22:47

no

Indian Pediatr. 2022 Dec 14; 59(11):888-889

latin-1

Indian Pediatr

10.1007/s13312-022-2652-9

oa_other

==== Front Nat Rev Rheumatol Nat Rev Rheumatol Nature Reviews. Rheumatology 1759-4790 1759-4804 Nature Publishing Group UK London 892 10.1038/s41584-022-00892-3 Year in Review COVID-19 vaccination in individuals with inflammatory rheumatic diseases Skapenko Alla http://orcid.org/0000-0002-1681-491X Schulze-Koops Hendrik Hendrik.Schulze-Koops@med.uni-muenchen.de grid.5252.0 0000 0004 1936 973X Division of Rheumatology and Clinical Immunology, Department of Internal Medicine IV, Ludwig Maximilians University of Munich, Munich, Germany 14 12 2022 12 © Springer Nature Limited 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Since the start of the SARS-CoV-2 vaccination campaign, our knowledge of the effects of vaccines in people with inflammatory rheumatic diseases has remained incomplete. In particular, the effects of immunomodulatory therapies on vaccine success are poorly understood. Three notable papers from the past year have helped to fill these knowledge gaps. Key advances Although SARS-CoV-2 vaccination seroconversion rates are lower in individuals with inflammatory rheumatic diseases under immunomodulatory therapy, the neutralizing capacity of anti-SARS-CoV-2 antibodies does not differ between affected and non-affected individuals7. In contrast to the humoral immune response, the cellular immune response to SARS-CoV-2 vaccination in patients receiving rituximab is maintained8. After rituximab therapy, determination of peripheral blood B cells may be a means to facilitate successful immunization, as the threshold for a successful immune response is 10 B cells per microlitre of peripheral blood9. Subject terms Infection Autoimmunity ==== Body pmcThe SARS-CoV-2 pandemic has paralysed the world. Faced with the constant fear of severe or even fatal infections with rapid spread throughout the population, unprecedented precautionary measures such as large-scale, strict and prolonged lockdowns and bans on social contact have imposed restrictions on private, social and professional lives. The way out of this stressful and dangerous situation has been achieved by worldwide vaccination campaigns, which have increasingly built up immunity in populations since the beginning of 2021. Studies on the efficacy and safety of the vaccines, some of which are based on new mRNA technology, have shown rapid and strong development of immunity in healthy populations. Unfortunately, individuals with immune systems compromised by disease (such as an inflammatory rheumatic disease (IRD)) or by therapy were explicitly excluded from participation in the large registration trials for the vaccines. The effectiveness and safety of vaccination in these vulnerable individuals was therefore not known. In the spring of 2021, initial observations were published showing that people with IRDs and other immune-mediated inflammatory diseases (IMIDs) have high seroconversion rates, suggesting that vaccines are effective and safe in this population1–3. However, many questions remained that are of great importance in the care of patients with IRDs. For example, it was not clear whether all drugs used for the treatment of IRDs are comparable in terms of vaccine responses. Several observations suggested that rituximab, mycophenolate, cyclophosphamide, methotrexate, abatacept and glucocorticoids, especially at higher doses, would have negative effects on the humoral vaccination response3,4. Other questions related to the persistence of immunity achieved under ongoing immunosuppression, and whether affected individuals should be vaccinated more frequently than those with no immunosuppression. A related issue was whether certain medications should be paused before, during or after vaccination, and notably, whether an immune response could be achieved at all in patients receiving highly potent immunosuppression with rituximab. This question was particularly important, as early evidence indicated that patients on rituximab therapy were at an increased risk of severe and potentially fatal outcomes of SARS-CoV-2 infection5, and that persistent depletion of B cells is associated with an insufficient vaccine response6. Results published in 2022 have provided answers to some of these questions7–9. Wieske et al. analysed humoral immune responses after second and third vaccinations against SARS-CoV-2 in a large cohort of individuals with various IMIDs, including IRDs such as rheumatoid arthritis (RA), spondyloarthritides, soft connective tissue diseases and vasculitides, who were treated with systemically acting immunomodulatory and/or immunosuppressive drugs alone or in combination7. The analysis covered both the concentrations of anti-SARS-CoV-2 immunoglobulins in the sera of the vaccinated individuals and the induction of neutralizing antibodies to the virus. The study had several notable results. The humoral immune responses in these individuals did not vary according to the different diseases. Although concentrations of anti-SARS-CoV-2 serum antibodies were moderately lower in individuals with IRDs than in those without, and they did not increase after a third vaccination, there was no difference in seroconversion rates between individuals with or without IRDs. Importantly, and of particular relevance to people with IRDs, regardless of the concentration of anti-SARS-CoV-2 serum antibodies, the neutralizing capacity (against the original SARS-CoV-2 virus strain WA1/2020) and the ability to generate a rapid and sufficient immune response after re-exposure to antigen did not differ between individuals with or without IRDs. Thus, it reassuringly indicates that the risk of infection and the development of severe disease is not increased for people with IRDs. However, the results also showed that seroconversion is achieved less frequently under ongoing therapy with mycophenolate, spingosine-1-phosphate inhibitors or rituximab than with other treatments, and that for rituximab in particular, even repeated vaccination does little to change this situation. Rituximab poses a particular challenge for vaccination success in people with IRDs. Although, for some patients, it seems possible to change long-term therapy with rituximab to a different therapeutic regimen, thereby enabling successful immunization, this approach is not necessarily possible in other situations such as in remission induction and maintenance of ANCA-associated vasculitis (AAV). However, the recommendation that necessary rituximab therapy should be modified in favour of attaining a vaccination response should also take into consideration the extent of the limitation of vaccination-induced protection under rituximab therapy. After all, for sufficient immunity against viral infection, targeted cellular immunity is important as well as humoral immunity. Although the humoral aspect of the vaccine response is impaired in patients receiving rituximab, evidence suggests that the cellular aspect is not10. In 2022, Jyssum et al. published the results of a study that examined humoral and cellular immunity in patients who were repeatedly vaccinated against SARS-CoV-2 while receiving rituximab therapy for RA8. The results confirmed that patients receiving rituximab therapy were less likely to seroconvert than those on other therapies. The likelihood of seroconversion was a function of the time interval since the last rituximab administration. Interestingly, however, after a second vaccination, 21.8% of individuals had antibodies against SARS-CoV-2, but 53% of individuals had a CD4+ T cell response, and as many as 74% of individuals had a CD8+ T cell response. Although a third vaccination resulted in seroconversion in an additional 16% of people, all individuals studied had detectable T cell responses after the third vaccination, which was given 6–9 months after the last rituximab administration. These data demonstrate the divergent dynamics of humoral and cellular anti-SARS-CoV-2 immune responses in patients receiving rituximab therapy, and they also show that even in the absence of a measurable humoral immune response, a protective T cell response develops in most people. Discontinuation of a medically advisable rituximab therapy, therefore, does not seem to be necessary to address concerns about a lack of immunological protection against SARS-CoV-2. “for sufficient immunity against viral infection, targeted cellular immunity is important as well as humoral immunity” The number of peripheral B cells starts to recover at the earliest 4 months after rituximab treatment. For this reason, the current EULAR recommendations state that vaccination should occur at the earliest 4 months after rituximab administration. However, the interval between treatment and the appearance of the first peripheral B cells is variable. Therefore, the recommendation to vaccinate after only 4 months is not necessarily a safe one, because without B cells there will be no humoral response, and in many patients 4 months may not be long enough for peripheral B cells to reappear. Jyssum et al. identified a correlation between the humoral immune response and the interval from the last rituximab administration8, but they did not identify a robust parameter that would indicate the likelihood of successful vaccination. Results published by Stefanski et al. identified just such a parameter9. They studied antibody and T cell responses to SARS-CoV-2 after vaccination in individuals with RA or AAV on ongoing rituximab therapy, and correlated the responses with peripheral B cell counts. Their findings indicated that a minimum of 10 B cells per microlitre of blood is the threshold above which a sufficient humoral and cellular immune response to SARS-CoV-2 vaccine is established. The peripheral B cell count thus represents an initial biomarker of vaccine success in patients with IRDs on ongoing B cell-depleting therapy. Determination of the B cell count can therefore enable the care of these patients to be objectively controlled while facilitating successful vaccination against SARS-CoV-2. In summary, three studies published in the past year filled notable knowledge gaps relating to the efficacy of SARS-CoV-2 vaccination in individuals with IRDs. Although these individuals were explicitly excluded from the initial approval studies of the vaccines, the results indicate that they can be safely and successfully vaccinated against SARS-CoV-2, thereby enabling them to gain a measure of reassurance that they can achieve immunological protection and a return to regular life, despite the ongoing SARS-CoV-2 endemic. Acknowledgements This work was supported by the Verbundanträge ‘GAIN’ (project 8, 01GM1910C) and ‘COVIM’ (project AP8, 01KX2021), both by the Federal Ministry of Education and Research of Germany; and by the FöFoLe program of the medical faculty of the LMU Munich. Competing interests The authors declare no competing interests. ==== Refs References 1. Geisen UM Immunogenicity and safety of anti-SARS-CoV-2 mRNA vaccines in patients with chronic inflammatory conditions and immunosuppressive therapy in a monocentric cohort Ann. Rheum. Dis. 2021 80 306 311 10.1136/annrheumdis-2021-220272 2. Sattui SE Early experience of COVID-19 vaccination in adults with systemic rheumatic diseases: results from the COVID-19 Global Rheumatology Alliance Vaccine Survey RMD Open 2021 7 e001814 10.1136/rmdopen-2021-001814 34493645 3. Furer V Immunogenicity and safety of the BNT162b2 mRNA COVID-19 vaccine in adult patients with autoimmune inflammatory rheumatic diseases and in the general population: a multicentre study Ann. Rheum. Dis. 2021 80 1330 1338 10.1136/annrheumdis-2021-220647 34127481 4. Sieiro Santos C Immune responses to mRNA vaccines against SARS-CoV-2 in patients with immune-mediated inflammatory rheumatic diseases RMD Open 2022 8 e001898 10.1136/rmdopen-2021-001898 34987093 5. Schulze-Koops H Krueger K Vallbracht I Hasseli R Skapenko A Increased risk for severe COVID-19 in patients with inflammatory rheumatic diseases treated with rituximab Ann. Rheum. Dis. 2021 80 e67 10.1136/annrheumdis-2020-218075 32591357 6. Rehnberg M Vaccination response to protein and carbohydrate antigens in patients with rheumatoid arthritis after rituximab treatment Arthritis Res. Ther. 2010 12 R111 10.1186/ar3047 20529331 7. Wieske L Humoral responses after second and third SARS-CoV-2 vaccination in patients with immune-mediated inflammatory disorders on immunosuppressants: a cohort study Lancet Rheumatol. 2022 4 e338 e350 10.1016/S2665-9913(22)00034-0 35317410 8. Jyssum I Humoral and cellular immune responses to two and three doses of SARS-CoV-2 vaccines in rituximab-treated patients with rheumatoid arthritis: a prospective, cohort study Lancet Rheumatol. 2022 4 e177 e187 10.1016/S2665-9913(21)00394-5 34977602 9. Stefanski AL B cell numbers predict humoral and cellular response upon SARS-CoV-2 vaccination among patients treated with rituximab Arthritis Rheumatol. 2022 74 934 947 10.1002/art.42060 34962360 10. Bonelli MM Mrak D Perkmann T Haslacher H Aletaha D SARS-CoV-2 vaccination in rituximab-treated patients: evidence for impaired humoral but inducible cellular immune response Ann. Rheum. Dis. 2021 80 1355 1356 10.1136/annrheumdis-2021-220408 33958323

36517627

PMC9748886

NO-CC CODE

2022-12-15 23:22:47

no

Nat Rev Rheumatol. 2022 Dec 14;:1-2

utf-8

Nat Rev Rheumatol

10.1038/s41584-022-00892-3

oa_other

==== Front J Fam Econ Issues J Fam Econ Issues Journal of Family and Economic Issues 1058-0476 1573-3475 Springer US New York 9880 10.1007/s10834-022-09880-x Original Paper The Effects on Labor Supply of Living with Older Family Members Needing Assistance with Activities of Daily Living http://orcid.org/0000-0002-8950-8884 Wilcox Virginia vlw@niu.edu 1 http://orcid.org/0000-0003-3862-1091 Sahni Herman 2 1 grid.261128.e 0000 0000 9003 8934 Department of Economics, Northern Illinois University, DeKalb, IL USA 2 grid.252749.f 0000 0001 1261 1616 Department of Finance, Baldwin Wallace University, Berea, OH USA 14 12 2022 119 30 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Using a sample of 18,201 observations of working age respondents drawn from the Medical Expenditure Panel Survey, 1996–2018, this research examined the labor supply effects for younger family members of living with older persons needing assistance with activities of daily living. We report the effects for three labor supply outcomes of younger family members: working hours, full-time work, and occupational flexibility of working hours. Our results indicate that living with an older family member needing assistance significantly reduced younger women’s working hours and the probability of working full-time among younger women, but increased both of these labor outcomes among younger men. In addition, living with an older family member needing help led younger women to work in occupations with significantly larger average variances in working hours. This suggests that these women occupied positions that allowed greater flexibility of working hours. We found little effect on flexibility of working hours for younger men. We conclude that the need for assistance among older family members has important effects on the labor market outcomes of younger family members. Keywords Elder caregiving Labor supply JEL Classification J—Labor and demographic economics J22—Time allocation and labor supply ==== Body pmcIntroduction The number of Americans providing care to ill or needy family members is large and growing as the baby boom cohort ages. In 2020, the AARP and National Alliance for Caregiving (2020) reported survey findings describing the caregiving population: Estimates indicated that in 2020, 16.8% of Americans were caregivers for individuals who were at least 50 years old. Most caregivers of adults cared for a relative, typically an older relative. Caregivers reported averaging about 24 hours of care each week for the care recipient. The picture painted by these numbers clearly shows that many millions of family members are directly or indirectly burdened by the need to care for older family members. While empirical research has shown that employment, whether part-time or full-time, may not have had a large impact on the hours spent by daughters in assisting parents (Dautzenberg et al., 2000; Mazotta et al., 2020), this research asks instead whether living with older family members needing assistance impacts younger family members’ labor supply. The impact on family members’ labor supply is an understudied indirect family cost of elder care. Numerous studies have examined the psychological, physical, and emotional burden borne by caregivers (National Academies of Sciences, Engineering, and Medicine, 2016, Chapter 3). However, fewer studies have examined how the care burden impacts the work activities of family members. Of those studies that examined the impact on family members’ labor supply, many focused solely on the burden of the primary caregiver and ignored the burden on other family members. Because these costs may be significant, in our research we consider the effects of living with older family members needing assistance on the labor market outcomes of both younger women and men rather than limiting our focus to women, who have traditionally been more likely to be caregivers. In this paper we report findings from analyses in which we focused on younger workers who co-resided with older family members. Using a sample of 18,201 observations of working age respondents drawn from the Medical Expenditure Panel Survey, 1996–2018, we estimated the effects of co-residence on the working hours, full-time employment, and working hours flexibility of the younger workers. While our focus on both younger women and men allowed us to observe the possible effects of caregiving by family members, which may have decreased working hours and full-time employment and increased the chosen flexibility of working hours, we also observed an opposite effect for some younger family members who increased their labor supply, possibly in order to afford medical and personal care services for the older family member. Additionally, we found instances in which no impact on the younger family members’ labor supply was observed. This may indicate that the older family member’s need for assistance was mild or that she or he had financial resources to pay for personal care and medical services. We conclude that the need for assistance among older family members has important effects on the labor market outcomes of younger family members. Review of the Literature First, we review the literature describing the costs of caregiving. Although there is a large literature addressing the physical, psychological, emotional and social aspects of the family burden of elder care, particularly that borne by primary caregivers, a smaller number of studies has examined the effects on family members’ labor market activities (National Academies of Sciences, Engineering, and Medicine, 2016, Chapters 3 and 4). A common problem in the earlier literature describing the burdens of caregiving was that the data were often obtained from local surveys or projects (Kemper, 1992; Leon et al., 1998; Max et al., 1995; Rubenstein et al., 2001; Taylor & Sloan, 2000). In these studies, a primary caregiver was often asked to report the amount of time devoted to caregiving activities. Hours spent in caregiving were multiplied by an hourly wage rate to obtain a monetary estimate of caregiver burden. For example, in a cross-sectional study of dementia patient-family caregiver pairs, Leon et al. (1998) measured informal costs of dementia by asking a primary family caregiver to report the amount of time spent on both activities of daily living and on household chores. Reports of caregiving hours were multiplied by national hourly wage rates. The results indicated that cost savings for formal and informal care could be achieved if memory and physical functioning of dementia patients could be improved or if the rate of decline could be delayed. However, the usefulness of these findings was limited because the local nature of the data implies that the findings of the study could not be generalized for the United States population (Franks, 1990). Further, these studies often limited the analyses to labor market effects for a designated caregiver (Carmichael et al., 2010; Ciani, 2012; Leigh, 2010; Van Houtven et al., 2013; Wolf & Soldo, 1994). This is a general problem of the literature describing the labor market effects of illness. Typically, studies in this literature focused on the effects of physical and mental disorders on the labor supply of the afflicted individual and ignored the effects on family members (Max et al., 1995; Marcotte and Wilcox-Gök, 2000, 2001; Skira, 2015). Although some studies analyzed the effect of poor physical health of a specific illness on a family member’s labor supply, these often considered only the effect on the family member expected to be the primary caregiver: Examples include the response in the wife’s labor supply to a decline in the health of her husband, that of a daughter to the health of her disabled, elderly parent, and that of a mother’s labor force participation to the chronic illness of a child (Berger & Fleischer, 1984; Boaz & Muller, 1992; Chang & White-Means, 2006; Ettner, 1995a, 1995b; Muurinen, 1986; Stommel et al., 1994; Whetton-Goldstein et al., 1997; White-Means, 1992). Unfortunately, the exclusive focus of these studies on caregivers’ labor supply means that the findings were subject to selection bias. The study samples were nonrandomly selected because they included only individuals designated as caregivers and excluded other family members. More recent studies of the financial impact of caregiving activities on families with ill members utilized larger national samples (Carmichael et al., 2010; Ciani, 2012; Fevang et al., 2012; Langa et al., 2001; Latif, 2006; Leigh, 2010; Meng, 2013; Mentzakis et al., 2009; Van Houtven et al., 2013). For example, Langa et al. (2001) used the 1993 asset and health dynamics (AHEAD) Study to compare caregiving time and cost for two older groups: elderly with dementia and elderly with normal cognition. Incremental weekly caregiving hours and the associated cost of informal caregiving were greater for elders with dementia than for those without dementia. Among those with dementia, caregiving time and cost increased substantially as cognitive impairment worsened. While this study is one of the best to be found in the literature describing the effects of dementia on family caregiving, even this study aggregated family members’ time and did not allow for substitution of one family member’s time in the labor market for that of another who was engaged in caregiving. Examples of studies that examined labor market effects for multiple family members can be found for other disorders than dementia (Emanuel et al., 2000; Roberts, 1999; Weinberger et al., 1993). Weinberger et al. (1993) found the largest components of family burden associated with Parkinson's disease to be the burden of providing informal caregiving and the loss of earnings due to substituting caregiving activities for market work. Emanuel et al. (2000), a study of terminally ill patients and their caregivers, reported that patients with substantial care needs were significantly more likely to report that they or their family members had to obtain an additional job to deal with the increased economic burden. Finally, two recent studies departed from the usual focus on current labor market effects to estimate longer term labor market outcomes among caregiving women: Schmitz and Westphal (2017) estimated the long-term effects of female caregivers in Germany and Skira (2015) examined the long-term effects among female caregivers in the United States. These studies reported mild short-term effects but found that caregiving significantly reduced the probability of returning to work or increasing work hours after a caregiving spell. Above we reviewed studies that addressed how the need for caregiving may have decreased family members’ labor supply. An alternative effect of co-residing with an older family member needing assistance was for younger family members to have increased labor supply in order to better afford medical and personal care services for the older family member. Relatively few studies have explicitly considered this possibility and the findings were mixed (Berger & Fleischer, 1984; Coile, 2004; Gould, 2004; Roberts, 1999; Siegel, 2006). Using the National Longitudinal Survey of older males, Berger and Fleischer (1984) found that transfer income played a crucial role in a wife’s labor supply response when a husband became ill: When no income transfers were available, the wife increased her market work to replace the lost earnings of the husband. However, as the level of transfer income increased, the wife reduced her labor supply, enabling her to spend more time at home caring for her husband. Roberts (1999) and Coile (2004) found that an illness in the family caused an increase in men’s labor supply and no effect or a decrease in women’s labor supply. Gould (2004) reported that mothers worked fewer hours if a sick child had a condition requiring time-intensive caregiving. She found no evidence of mothers working more hours. Similar to our theoretical approach, Siegel (2006) explained that when a husband's earnings decreased due to illness, the coefficient of the husband's health in models of his wife's labor force participation and hours of work reflected her decision to either decrease her labor supply to provide health care for her husband or increase it to purchase this care in the market. Using data from the Health and Retirement Study, Siegel found that when the husband had physical limitations or a heart condition, the wife’s probability of employment increased, other things equal. However, when the husband’s endogenous earnings are instrumented, the author found that a husband’s ill health, such as a stroke, led an employed wife to decrease work hours. To reconcile these apparently contradictory findings, Siegel suggested that “the functional debilitation that results from a stroke is severe enough to warrant more costly care, i.e., care that costs more per hour than a wife’s market wage” (p. 598). In addition to the impact on the labor supply measures of working hours and full-time employment, in the research reported here we estimated the impact of co-residing with an older family member needing help on younger family members’ choice of flexibility in working hours. Our assumption was that an occupation with a larger variance in usual working hours offered greater choice in working schedules. Empirical evidence from prior studies suggests that higher family demands increased the probability that women chose self-employment, which offered greater flexibility in their working hours (Boden, 1999; Lombard, 2001). Baxter (2011) found that flexible work hours allowed parents to better allocate their time between work time and family time. Among workers who are not self-employed, Golden (2008) noted that “not all occupations lend themselves by nature to flexible scheduling” (p. 87). Despite the lack of broad availability of flexible scheduling across occupations, Golden (2008) found that the distribution of flexible work arrangements was somewhat related to family demands (p. 104). This is echoed in Goldin (2014), who found that greater average earnings of men compared to women was partially attributable to men working in occupations with less flexible work schedules, while women were more likely to work in occupations with greater work schedule flexibility. Although both Golden (2008) and Goldin (2014) focused on family demands due to the presence of young children in the home, the presence of older family members needing assistance with activities of daily living may have similar effects. Labor Supply Model Our empirical research is based on a time allocation model of labor supply (Becker, 1965). Becker’s model, which explicitly recognizes the importance of time costs in economic decision-making, was first applied to family labor supply by Mincer (1962). It is now a standard model in labor economics (Chiappori & Lewbel, 2015) and was applied to caregiving for elder family members as early as 1986 (Muurinen, 1986; White-Means, 1992). Stanfors et al. (2019) reported that this model was the basis for several studies examining the tradeoff among paid work, leisure, and home care for individuals with caregiving responsibilities. The time allocation model is a utility-maximizing model of behavior, implying that family members of the ill are assumed to make decisions that give them the most satisfaction. In addition to the utility function, the model has a production function specifying the alternative activities that a family member may pursue, including but not limited to working in the labor market. The household production functions indicate the time needed to transform goods and services purchased in the marketplace into final consumption goods. For example, a person engaged in giving care to an ill family member might use two hours of her or his time and various store-bought commodities to help the ill family member bathe. The final consumption good produced is a bath. Each individual compares her or his productivity in the home to that in the labor market to determine the optimal allocation of time between home and market work. This is a relatively simple decision if there is only one family member of the ill person. However, in the context of a multi-person family, some individuals will have relatively higher productivity in the labor market and others will have relatively higher productivity in home production. Given institutional restrictions, such as an employer’s required number of hours to be worked each week, the family must choose the optimal allocation of time for family members. Individuals whose relative wage rate is lower than other family members may find that their time is more efficiently spent performing duties in the household (including caregiving for an older family member), leading to a reduction in work hours. This might be a change in hours on the current job, a shift from a full-time to a part-time position, or if hours devoted to caregiving are sufficiently large, the person may leave the labor force altogether. The decision to reduce hours of work or leave the labor force imposes a cost on the family in the form of foregone earnings, though it is important to note that nonmarket activities in the home while caretaking can be highly valued even though they are uncompensated in the labor market. Conversely, family members with relatively higher wage rates may increase their hours of work to earn income needed for supplying services for the ill individual or secure health insurance coverage for the ill family member. This imposes a cost on the family in the form of the value of foregone home activities and leisure. Our hypotheses are drawn from a simple version of a time allocation model describing the younger family members’ labor supply. The older family member’s health (H) is a commodity that is produced in the home by combining market commodities (z) with the caregiver’s time (b). The individual’s utility function is expressed as1 U=UC,H where C = c(x,a) and H = h(z,b,e). The arguments are defined as C = consumption goods and services, H = older family member’s health, x = market goods and services used to produce home consumption goods, Px = price per unit of x, z = market goods and services used to provide care, Pz = price per unit of z, a = time in home production of C, b = time used in caregiving, e = exogenous health endowment of older family member (measures the need of the care receiver, with a higher e indicating a healthier person), T = total time endowment, w = individual’s wage rate, V = family’s unearned income, and n = labor supply hours. Leisure does not specifically enter into the utility function. Following Heckman (1974), all leisure activities can technically be considered consumption activities. Because the separate notation of consumption and leisure is not necessary to derive the desired results, we choose to simply include leisure activities in C. The utility-maximization problem is:2 MaximizeUcx,a,hz,b,e subject to the budget constraint3 wn+V=Pxx+Pzz and the time constraint4 n=T-a-b Combining the budget and time constraints into a single constraint, the Lagrangian is5 L=Ucx,a,hz,b,e+λwT-wa-wb+V-Pxx-Pzz The first order conditions provide the usual results: The marginal utility of time spent doing a equals the marginal utility of time spent in b, the marginal utility per dollar spent on x equals the marginal utility per dollar spent on z, and income is fully exhausted on expenditures. From these conditions, we derive the individual’s demand for hours of work (n) and determine how this will vary with the wage rate (w) and the health of the ill member of the family (e). Alternative responses are possible because care can be provided either directly with a family member's time (b) or indirectly with purchases of services from the market (z). The prediction of the effect of illness of older family members needing assistance with activities of daily living (ADLHELP) on the labor market outcomes of younger family members (LMO) using this model is ambiguous: Illness makes time in health production more valuable, leading to more health production and reduced labor supply (∂LMO/∂ADLHELP < 0). However, if greater earnings are needed to purchase medical inputs for the older family member, there can be an income effect leading to an increase in labor supply (∂LMO/∂ADLHELP > 0). The observed impact will be the net effect of these two opposing forces. For our first hypothesis, we assume that the former effect dominates the latter leading to a decrease in labor supply, other things equal (compared to individuals co-residing with older family members who do not need ADL assistance). H1 The presence in the household of an individual needing ADL assistance (ADLHELP) will cause a decrease in labor supply (LMO) of other family members, other things equal:∂LMO/∂ADLHELP<0 If the value of the family member’s time matters in determining the impact on her or his labor supply, the effect of living with an older family member needing assistance may vary depending upon the wage rate of the younger person. For younger family members with relatively low wage rates, the effect of having an older person in the household who needs assistance may be to decrease labor supply. As a consequence, individuals who have lower wage rates are more likely to decrease work hours when a family member is ill. In contrast, for family members with relatively high wage rates, the effect of co-residing with an older family member who requires assistance may be to increase labor supply. The need for caregiving time or higher earnings to purchase market inputs may impact not only the total time available for work in the labor market, but also the timing of work. To the extent that the caregiving effect dominates the earnings effect, younger family members may desire greater flexibility in their work schedules. Under this assumption, we expect to observe that younger family members will be more likely to choose jobs in occupations with greater work schedule flexibility. This generates our next hypothesis. H2 The presence in the household of an individual suffering from an illness who requires care (ADLHELP) will lead other family members to choose employment in occupations with greater work schedule flexibility, other things equal:∂FLEX/∂ADLHELP>0 Next, we consider the effects that wage rates may have on the response to living with an older family member needing assistance with activities of daily living (ADL). The wage rate is a fundamental determinant of labor supply because it represents the value of an hour of time: Simplistically, if an individual works one hour less in order to engage in an alternative activity, the wage rate is the amount of earnings foregone. The higher the wage rate, the greater the cost of foregoing work hours (Ehrenberg et al., 2023). Following the reasoning above, we expect the impact of living with an older family member needing ADL help on labor supply to decrease with the individual’s wage for both men and women. This leads to our third hypothesis. H3 The effect on labor supply (LMO) due to the presence in the household of an individual requiring care (ADLHELP) will decrease as the wage rate rises, other things equal:∂LMO2/∂ADLHELP·∂w<0 Finally, we examine the effect of other factors, such as social norms regarding work and caretaking, on the labor supply of individuals living with an older family member needing ADL assistance. Prior research has established that average wage rates for women are lower than those for men (Goldin et al., 2017). The lower value of time in the labor market for women may at least partially explain why women have traditionally been more likely to be caretakers than men. In our analyses, we estimate the effect of living with an older family member needing ADL assistance while controlling for the individual’s wage rate and interacting the wage rate variable with the ADL help variable. If the variable representing ADL help continues to be positive and significant for younger women (and not for younger men), this indicates that factors such as social norms may significantly influence the individual’s choices. While our model cannot identify these other factors, by testing whether the inclusion of the wage rate is sufficient to explain the labor supply effects of living with an older family member needing ADL help, it can ascertain their importance. To establish a testable hypothesis, we assume that female family members will experience greater labor market effects than men of living with older family members needing assistance, even while controlling for the wage rate. This leads to our final hypothesis. H4 The presence in the household of an individual suffering from an illness who requires assistance (ADLHELP) will lead to an average effect on labor supply (LMO) of female family members that is more negative than the effect on labor supply of male family members, other things equal:∂LMO/∂ADLHELPWOMEN<∂LMO/∂ADLHELPMEN Note that the condition in H4 is satisfied if the impacts on labor supply for both women and men are negative, but the effect for women is larger in absolute value, or if the impact on women’s labor supply is negative and the impact on men’s labor supply is zero or positive. Hypothesis H4 illustrates that it is possible to observe differing effects of illness on different members of the same family. One family member may reduce her labor supply and other members may compensate by increasing their labor supplies. In fact, we expect family members to self-sort into labor market activity versus home production according to the relative productivities (represented by wage rates) of the members. It is possible that we will reject H4 because the wage rate of the younger family member explains the labor supply effect of living with an older family member needing ADL assistance: That is, we may observe that the labor supply of female family members does not decrease more than that of male family members when the wage rate is controlled for. These hypotheses are tested in our empirical analyses. First, however, we describe the data used for the study in the following section. Data We used observations drawn from 23 years of the medical expenditure panel survey (MEPS) for the empirical analyses (Agency for Healthcare Research and Quality, 1996–2018). For the research reported in this paper, we focused only on individuals co-residing with older family members. Drawing observations from the 1996 through 2018 MEPS data, we obtained a pooled study sample including 30,677 unweighted observations of working age respondents (17,959 women and 12,718 men). Of this preliminary sample, which included both employed and unemployed respondents living in households with at least two family members, 10,185 working women and 8016 working men reported co-residing with older (age 65 or older) family members. Of these, 1513 working women and 1363 working men reported co-residing with older family members needing assistance with activities of daily living (ADL). In our analyses, we compared the labor supply outcomes of the respondents co-residing with older family members who require ADL assistance with those of working age respondents co-residing with older family members not requiring ADL assistance. Independent and Dependent Variables Variables used in the statistical analyses are defined in Table 1. All monetary figures were converted to 2012 dollars. The dependent variables for the analyses represented labor supply decisions of the younger family members. The first measure was usual working hours per week (Hours worked). The second labor supply measure was full-time employment (Full time): This variable was dichotomous with a value = 1 if the person worked for more than 35 hours per week and a value = 0 otherwise.Table 1 Definitions of variables Variable Definition Hours worked Usual hours worked per week (37.373, 10.867) Full time Dummy variable = 1 if younger family member works more than 35 hours per week (0.733, 0.442) Hours variance Variance in usual weekly hours in the younger family member’s occupation (136.665, 42.499) ADL help Dummy variable = 1 if an older family member needs help with activities of daily living (0.158, 0.365) Wage Hourly wage of younger family member ($US 2012) (17.848, 12.541) ADL help* Wage Interaction of ADL help and Wage Other income Household income other than that earned by the younger family member ($US 2012) (55.941, 53.042) Age Individual’s years of age (42.927, 13.458) Age squared Individual’s years of age squared School years Years of schooling (11.291, 4.857) Female Dummy variable = 1 if individual is female Married Dummy variable = 1 if individual is married (0.462, 0.499) Minor children Number of children in the household less than 18 years of age (0.746, 1.199) Poor physical health Dummy variable = 1 if individual rates her/his physical health as excellent or very good (0.179, 0.384) Poor mental health Dummy variable = 1 if individual rates her/his mental health as excellent or very good (0.089, 0.285) White Dummy variable = 1 if individual is White (reference category) (0.674, 0.469) Asian Dummy variable = 1 if individual is Asian (reference category is “white”) (0.104, 0.305) Black Dummy variable = 1 if individual is Black (reference category is “white”) (0.191, 0.393) Other race Dummy variable = 1 if individual is not Asian, Black, or White (reference category is “white”) (0.031, 0.172) Hispanic Dummy variable = 1 if individual is Hispanic (reference category is “not Hispanic”) (0.270, 0.444) Northeast Dummy variable = 1 if individual lives in the Northeast (reference region is the South) (0.166, 0.372) Midwest Dummy variable = 1 if individual lives in the Midwest (reference region is the South) (0.158, 0.365) West Dummy variable = 1 if individual lives in the West (reference region is the South) (0.303, 0.459) South Dummy variable = 1 if individual lives in the South (0.374, 0.484) Year1–Year22 Dummy variables indicating year (1996–2018) Means and standard deviations of 18,201 observations in study sample are reported in parentheses Ideally, we might have examined the effect on hours of work in a sample of self-employed younger family members who can determine their own hours of work. However, the hourly wage for self-employed respondents is frequently missing in the MEPS data. Instead, we considered an alternative measure of flexibility in hours worked: We calculated the population variance in the usual number of working hours for the occupation in which the younger family member was employed (Hours variance). As the MEPS provides nationally representative data, we calculated the variance in the usual number of working hours for each of eight broad occupational categories (management; professional; service; sales; administrative support; farming, fishing, and maintenance; and production, transportation, and material moving) using all MEPS respondents. Average hours worked per week in these occupational groups ranged from 34.13 (sales) to 42.13 (management). The variance in hours worked per week ranged from 165.4 (professional) to 520.6 (farming, fishing, and forestry). Relatively low values of the variance in usual hours were observed for positions in management (175.2), professional (165.4), and administrative support (178.5), while higher variances in usual hours worked per week were observed for jobs in service (293.0), sales (271.2), production, transportation, and material moving (264.3), and farming, fishing, and forestry (520.6). While more narrowly defined occupational groups might better represent the variance in usual hours worked per week for various jobs, we were limited to those categories reported in the MEPS data.1 To describe the functionality of older family members, we used a dummy variable indicating whether the older person required help or supervision with activities of daily living (ADL help), such as bathing, dressing, or getting around the house.2 We also used control variables representing relevant social, demographic, and economic information of the younger family member. These included age, sex, marital status, race, ethnicity, years of education, presence of minor children in the home, regional residence, and other family income. In addition, we controlled for the self-assessed physical and mental health status of the younger family member by including dummy variables indicating whether the family member of the ill person rated his or her own physical or mental health as fair or poor. Because the data is pooled from 23 years of MEPS data, we included a set of year dummy variables to capture year-specific effects in the labor market measures. Finally, for identification in the selection model of the employment decision, we used regional unemployment rates. To allow comparison of the study sample to a more general sample, in Table 2 we provide averages for older family members in our study sample and for the entire MEPS survey sample of older respondents. Averages are reported for each group by whether they reported needing ADL assistance. Thirteen percent of the 15,213 co-residing older family members in our study sample reported needing help with ADLs. Among the older family members in our study sample, those needing ADL assistance differed significantly from those not requiring help. For example, older family members requiring ADL help were older, more likely to be a woman, less likely to be married, had fewer years of education, and were far less likely to be working. Older family members needing ADL help also had poorer physical and mental health than their counterparts. Regarding race and ethnicity, older family members needing ADL help were more likely to be Black or Hispanic and less likely to be White or Asian.Table 2 Average characteristics of older family members with and without need for help with activities of daily living Study samplea Survey sampleb Needing ADL helpc Not needing ADL helpd Needing ADL help Not needing ADL help Age 78.023 71.912 78.487 73.681 Female 0.682 0.525 0.669 0.564 Married 0.283 0.563 0.340 0.537 Poor physical health 0.809 0.368 0.799 0.345 Poor mental health 0.628 0.173 0.580 0.172 School years 8.147 9.665 9.078 10.812 Employed 0.027 0.261 0.024 0.201 White 0.643 0.692 0.726 0.788 Black 0.244 0.186 0.209 0.146 Asian 0.085 0.101 0.045 0.048 Other race 0.028 0.020 0.020 0.018 Hispanic 0.262 0.234 0.170 0.126 N 1985 13,228 9940 78,185 Based on unweighted observations of persons ages 65 and older of the Medical Expenditures Panel Survey, 1996–2018 aAll differences between those who need ADL help and those who do not are statistically significant for α = 0.05 bAll differences between those who need ADL help and those who do not are statistically significant for α = 0.05 except for Asian cAll differences between those in the study sample and the survey sample who need ADL help are statistically significant for α = 0.05 except Female, Employed, and Poor physical health dAll differences between those in the study sample and the survey sample who do not need ADL help are statistically significant for α = 0.05 except Poor mental health In comparison to the above information describing older family members in our study sample, 11.3% of the entire group of older MEPS respondents reported needing help with ADLs. Except for the percentage of older persons who reported being Asian, the relative values of those needing ADL help to those not needing ADL help follow a similar pattern. However, there were differences between the averages for the study sample and those for the MEPS survey sample. These differences permit us to understand how the older family members in our study sample differ from the general population of older persons. The percentage of older women who reported needing ADL help in our study sample was similar to the general MEPS population. The racial and ethnic composition of the two samples differed in that the older family members needing ADL help in the study sample were more likely to be Black, Asian, and Hispanic, while there was a larger percentage of individuals who reported being White in the MEPS survey. The older family members needing ADL help in the study sample reported poorer mental health than those in the MEPS survey sample, although the difference in average physical health between the two groups was not statistically significant. The older family members needing ADL help in the study sample were less likely to be married than older persons in the MEPS survey sample and they also had fewer years of education. The percentage who worked was very low in both samples. Older persons in our study sample who did not need ADL help were slightly younger on average than the sample of older MEPS respondents. They were more likely to be male, married, and employed, and had fewer years of education. Like their counterparts needing ADL help in the study sample, the older family members who did not require ADL assistance were more likely to report being Black, Asian, and Hispanic. Table 3 provides unweighted averages of the characteristics for the younger working women and men in the study sample. The first two columns report the averages for younger family members who co-resided with an older family member needing ADL help. The third and fourth columns report averages for younger family members who lived with an older family member not requiring ADL help. While the pattern of averages describing labor force outcomes in Table 3 is typical for men and women, our purpose was to ascertain whether there were differences in the characteristics of women or men associated with the presence of older family members needing ADL assistance. For this reason, we compared the averages for women (men) who co-resided with an older family member needing ADL help with those for women (men) who co-resided with an older family member not needing ADL help.Table 3 Average characteristics of working age family members Older family member needs ADL help Older family member does not need ADL help Women Men Women Men Hours worked 35.550a,b 39.282 36.189a 38.940 Full time 0.639a,b 0.802 0.691a 0.795 Hours variance 144.311a,b 137.884 135.136 136.671 Age 44.088a,b 41.335b 45.152a 40.089 School years 11.180b 11.231 11.487a 11.072 Wage 16.244a,b 18.204 17.383a 18.747 Other income 56.041 59.009 59.469 58.695 White 0.618b 0.637b 0.676a 0.693 Black 0.237b 0.236b 0.194a 0.169 Asian 0.109 0.090 0.103 0.106 Other race 0.036b 0.037 0.027a 0.033 Hispanic 0.269a 0.289 0.252a 0.289 Married 0.449a,b 0.488b 0.519a 0.386 Minor children 0.798 0.720 0.739 0.748 Poor physical health 0.247a,b 0.196b 0.183a 0.155 Poor mental health 0.108b 0.100 0.086 0.087 Northeast 0.146 0.152 0.173 0.164 Midwest 0.161 0.150 0.159 0.157 West 0.302 0.326 0.292 0.312 N 1513 1363 8672 6653 Based on unweighted observations of persons ages 18 through 64 living in families of at least two persons, Medical Expenditures Panel Survey, 1996–2018 aDifference between women and men (within ADL help group) is statistically significant for α = 0.05 bDifference in average for women between “needs ADL help” and “does not need ADL help” is statistically significant for α = 0.05. Similarly defined for difference in averages for men Younger women who lived with an older family member needing ADL help were less likely to be White and more likely to be Black or Other race than younger women who lived with older family members who did not need help. They were also less likely to be married and reported being in poorer physical and mental health than their female counterparts who lived with older family members not needing assistance. Younger women who lived with older family members needing ADL help had lower hourly wage rates and were less likely to work full-time, but the difference in average hours worked per week was small (0.639 h). However, the variance in usual hours for jobs held by younger women who lived with older family members needing ADL help was significantly greater than that of younger women who lived with healthy older family members. The significant differences between men in the two groups were fewer: Younger men who lived with older family members needing ADL help were less likely to be White and more likely to be Black than those who lived with older family members not needing help. They were more likely to be married, reported poorer physical health, and were on average 15 months older than men who lived with older family members not needing ADL assistance. MEPS was a useful data source for this study: The MEPS data contained information that allowed us to identify and describe older family members co-residing with younger family members; the MEPS data allowed us to include a measure of an older family member’s need for assistance with activities of daily living; and the MEPS data provided measures of labor supply as well as sociodemographic variables.3 Empirical Model This section describes the empirical models used in the analyses. We estimated the effects of co-residing with an older family member needing assistance with activities of daily living on three types of labor market outcomes: working hours per week, full-time employment, and variance in usual working hours in the respondent’s primary job. The ordinary least squares regression analyses of hours worked for an individual i may be represented by the following equation:6 Hours workedi=α+Xiγ+βADL helpi+ℓyeari+εi In Eq. (6), the vector X represents a vector of sociodemographic and economic control variables, which include the hourly wage rate, other family income, age and age squared, years of education, marital status, the presence of minor children in the household, indicators of poor physical health and poor mental health, and race, ethnicity, and region indicators. The variable ADL help is a dummy variable indicating if the older family member needed assistance with activities of daily living. The variable year represents a vector of dummy variables indicating the year from which the data was drawn. The specifications used for the ordinary least square regressions of the variance in usual weekly working hours were identical to Eq. (6), although the dependent variable was logged to normalize the distribution. Similarly, the specifications used for the analyses of the probability of full-time employment were identical to that in Eq. (6) although the analyses conducted were probit regressions. We report the marginal effects from these analyses in our results tables. Also, for comparison purposes, we report estimates for each outcome measure from a baseline model in which the ADL help variable was omitted. As explained in the description of the theoretical model, the effect of living with an older family member who needed ADL help was ambiguous because a younger family member may have responded by either reducing labor supply to engage in caretaking or increasing labor supply to afford purchases of market goods and services for the older family member. For the purpose of establishing a testable hypothesis, we assumed in hypotheses H1 that the caretaking effect dominates. This implies that β < 0 for our analyses of working hours and full-time employment. Finding β ≥ 0 implies that the hypothesis was rejected: That is, the presence of an older family member requiring ADL assistance did not decrease hours worked per week or the probability of full-time work. In our analyses of work schedule flexibility, hypothesis H2 implies that the presence of an older family member needing ADL assistance caused younger family members to seek jobs in which they had greater work schedule flexibility. This flexibility was represented by the logged standard deviation of usual occupational work hours. Thus, we expected the effect of ADL help on work schedule flexibility to be positive: β > 0. To assess whether the hypothesized effect of co-residing with an older family member who required ADL help diminished as the younger family member’s wage rate increased, we estimated a specification of our model in which we interacted the hourly wage rate with the ADL help variable:7 Hours workedi=α+Xiγ+βADL helpi+γyeari+δWagei·ADL helpi+εi We hypothesized in H3 that a higher wage rate would increase the use of market substitutes for own time in providing assistance, other things equal. This implied that the coefficient of the interaction variable would be negative for both women and men: δ > 0. While our multivariate analyses controlled for several sociodemographic factors, they did not control for social norms or other factors that may have influenced younger family members’ responses to living with an older family member needing ADL assistance. Given social norms that may have influenced work and caregiving roles of women and men, for hypothesis H4 we assumed that |βWOMEN| <|βMEN|. That is, we expected to observe larger impacts among younger women than among younger men, even when controlling for the wage rate effects. The issue of self-selection into the labor force may be relevant in this research because it is possible that self-selection into the labor force exists for both women and men. Following Puhani (2000), we assessed the appropriateness of using a correction method for selection bias by first checking for collinearity problems. Puhani (2000) notes that “in the absence of collinearity problems, the full-information maximum likelihood estimator is the preferable to the limited-information two-step method of Heckman (1974), although the latter also gives reasonable results. If, however, collinearity problems prevail, subsample OLS (or the Two-Part Model) is the most robust amongst the simple-to-calculate estimators” (p. 53). We report the results of these tests in the results section. Finally, we focused on the effects of older family members who resided with the younger family members without reference to the possible endogeneity between the younger family member’s co-residence and labor supply. The direction of potential bias due to endogenous co-residence is unknown: While older persons with health problems may have been more likely to reside with working age family members, it is also possible that healthy older persons who co-resided may have assisted with childcare or simply provided housing for the younger family members (Fuller-Thomson & Minkler, 2001; Kanji, 2018; Matsudaira, 2016). In either case, it may have been that the working age family members in households with an older family member in residence differed in observable and unobservable ways from working age family members living in families without an ill older member. To avoid potential endogeneity in the research reported here, we limited our study sample to working younger family members who co-resided with older family members. Results The comparisons of means reported in Table 3 indicate significant differences in several characteristics between younger family members living with older persons requiring ADL assistance and those living with older family members not requiring assistance. To assess the impact of these differences on labor supply outcomes, we report our estimates of the effects of ADL help from multivariate analyses in Tables 4, 5, and 6. Because prior studies have shown that the labor supply behavior of men differed importantly from that of women, in our analyses we followed the usual practice in labor economics of separately analyzing labor outcomes for women and men (Bargain & Peichl, 2016; Killingsworth & Heckman, 1986; Pencavel, 1986). In this section we describe our estimates and discuss the implications for the hypotheses described above.Table 4 Effects of older family member’s need for ADL help on younger family member’s hours worked Women Men Coefficient estimates Coefficient estimates Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Wage 0.176*** 0.176*** 0.178*** 0.095*** 0.095*** 0.106*** ADL help − 0.779* − 0.544 0.070 1.366* ADL help * Wage − 0.014 − 0.071* Other incomea − 4.690* − 4.700* − 4.680* − 1.240 − 1.250 − 1.130 Age 0.984*** 0.991*** 0.992*** 0.864*** 0.864*** 0.863*** Age squared − 0.011*** − 0.011*** − 0.011*** − 0.010*** − 0.010*** − 0.010*** School years 0.043 0.042 0.042 − 0.044 − 0.044 − 0.043 Married 0.014 − 0.008 − 0.008 3.043*** 3.037*** 3.041*** Minor children − 0.097 − 0.100 − 0.101 0.243* 0.244* 0.240* Poor physical health − 0.590† − 0.557† − 0.561† − 0.989** − 0.991** − 0.996** Poor mental health − 1.337** − 1.331* − 1.329** − 1.709** − 1.709** − 1.698** Asian 0.605‡ 0.632† 0.633† − 1.304** − 1.303** − 1.331** Black 0.329 0.359 0.363 − 1.501*** − 1.508*** − 1.521*** Other race 0.564 0.611 0.616 0.008 0.005 − 0.006 Hispanic 0.611* 0.625* 0.626* − 0.620* − 0.621* − 0.613* Midwest − 0.893** − 0.886* − 0.886** − 0.377 − 0.378 − 0.385 Northeast − 1.231*** − 1.248*** − 1.248*** − 1.138** − 1.137** − 1.124** West − 1.006*** − 1.005*** − 1.004*** − 0.822** − 0.823** − 0.826* Constant 12.227*** 12.232*** 12.182*** 20.660*** 20.650*** 20.437*** N 10,185 10,185 10,185 8016 8016 8016 R2 0.0870 0.0877 0.0877 0.1098 0.1098 0.1109 prob > F 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 OLS regressions for sample of working age family members drawn from the Medical Expenditures Panel Survey, 1996–2018. Year dummy variables included. Standard errors clustered by household aOTHINC is divided by 1 million to scale the coefficient estimates *p < 0.05; **p < 0.01; ***p < 0.001; †p ≤ 0.10; ‡p ≤ 0.15 Table 5 Effects of older family member’s need for ADL help on younger family member’s full time status Women Men Marginal effects Marginal effects Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Wage 0.007*** 0.007*** 0.007*** 0.003*** 0.003*** 0.004*** ADL help − 0.050*** − 0.048† − 0.004 0.043* ADL help * Wage − 0.0001 − 0.003** Other incomea − 0.151† − 0.141‡ − 0.141† 0.003 0.003 0.004 Age 0.037*** 0.037*** 0.037*** 0.030*** 0.030*** 0.030*** Age squared − 0.0004*** − 0.0004*** − 0.0004*** − 0.0003*** − 0.0003*** − 0.0003*** School years 0.002‡ 0.002‡ 0.002‡ − 0.002† − 0.002‡ − 0.002† Married − 0.012 − 0.013 − 0.013 0.097*** 0.097*** 0.098*** Minor children − 0.008* − 0.008* − 0.008* 0.006† 0.006‡ 0.006‡ Poor physical health − 0.029* − 0.026* − 0.026* − 0.025* − 0.025* − 0.025* Poor mental health − 0.036* − 0.036* − 0.036* − 0.071*** − 0.071*** − 0.071*** Asian 0.037* 0.037* 0.039* − 0.019 − 0.019 − 0.020‡ Black 0.027* 0.028* 0.028* − 0.016‡ − 0.016‡ − 0.016‡ Other race 0.020 0.022 0.022 0.015 0.015 0.014 Hispanic 0.031** 0.032** 0.032** 0.007 0.007 0.007 Midwest − 0.033** − 0.032** − 0.032** − 0.011 − 0.011 − 0.011 Northeast − 0.023† − 0.024† − 0.024† − 0.022† − 0.022† − 0.022† West − 0.038*** − 0.038*** − 0.038*** − 0.014 − 0.014 − 0.014 N 10,185 10,185 10,185 8016 8016 8016 Pseudo R2 0.0801 0.0818 0.0818 0.1171 0.1172 0.1188 prob > χ2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Probit regressions for sample of working age family members drawn from the Medical Expenditures Panel Survey, 1996–2018. Year dummy variables included. Standard errors clustered by household aOTHINC is divided by 1 million to scale the coefficient estimates *p < 0.05; **p < 0.01; ***p < 0.001; †p ≤ 0.10; ‡p ≤ 0.15 Table 6 Effects of older family member’s need for ADL help on logged variance of average hours worked for younger family member’s occupation Women Men Coefficient estimates Coefficient estimates Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Wage − 0.005*** − 0.005*** − 0.004*** − 0.003*** − 0.003*** − 0.003*** ADL help 0.041*** 0.055*** 0.001 − 0.018 ADL help * Wage − 0.001 0.001* Other incomea − 0.171** − 0.170** − 0.168** − 0.050 − 0.050 − 0.052 Age − 0.007*** − 0.007*** − 0.007*** − 0.008*** − 0.008*** − 0.008*** Age squared 0.0001*** 0.0001*** 0.0001*** 0.0001*** 0.0001*** 0.0001*** School years − 0.010*** − 0.010*** − 0.010*** − 0.003*** − 0.003*** − 0.004*** Married 0.002 0.003 0.003 − 0.025*** − 0.025*** − 0.025*** Minor children 0.014*** 0.014*** 0.014*** 0.001 0.001 0.001 Poor physical health 0.029*** 0.027*** 0.027*** 0.002 0.002 0.002 Poor mental health 0.036*** 0.036*** 0.036*** 0.029* 0.029* 0.029* Asian 0.061*** 0.060*** 0.060*** 0.021* 0.021* 0.021* Black 0.045*** 0.043*** 0.044*** 0.040*** 0.040*** 0.040*** Other race 0.004 0.001 0.001 0.017 0.017 0.017 Hispanic 0.033*** 0.032*** 0.032*** 0.010 0.010 0.010 Midwest 0.007 0.006 0.006 0.031*** 0.031*** 0.032*** Northeast 0.015* 0.016* 0.016* 0.017* 0.017* 0.017* West 0.003 0.003 0.003 0.029*** 0.029*** 0.029*** Constant 4.923*** 4.923*** 4.920*** 4.886*** 4.885*** 4.889*** N 10,185 10,185 10,185 8016 8016 8016 Pseudo R2 0.2191 0.2215 0.2217 0.2092 0.2092 0.2095 prob > χ2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 OLS regressions for sample of working age family members drawn from the Medical Expenditures Panel Survey, 1996–2018. Year dummy variables included. Standard errors clustered by household. Dependent variable is logged for regression analysis aOTHINC is divided by 1 million to scale the coefficient estimates *p < 0.05; **p < 0.01; ***p < 0.001; †p < 0.10; ‡p < 0.15 To determine whether to report estimates for regressions and probit analyses with corrections for selection into the labor force, we first conducted the tests suggested by Puhani (2000) to assess the presence of collinearity problems. The test statistics for our application indicated the presence of significant collinearity. For this reason, we report estimates for the sample of working younger family members without controlling for selection bias. However, we also estimated the models using a Heckman correction for self-selection and found that the estimates were similar in sign with only slight differences in the magnitude of some estimates. We also estimated a two-part model (Duan et al., 1983) and found that the estimates were almost identical to those reported in this paper.4 We report estimates from ordinary least squares regressions of working hours and the logged variance in working hours in Tables 4 and 6 and estimates from probit regressions of full-time employment in Table 5. All variables included in the regressions are defined in Table 1, with the exception that the value of the variable other income has been divided by one million to scale the estimates in Tables 4, 5, 6. In all analyses, we included dummy variables representing the year of the survey response. Because there were multiple observations of individual respondents as well as of respondents in the same household, standard errors of the coefficient estimates were clustered by household. To provide the reader with information describing the statistical significance of estimates that may be close to conventional levels of significance, in Tables 4 through 7 we indicate the significance of coefficient estimates that were ‘weakly significant’ (0.05 < p ≤ 0.10) and ‘nearly significant’ (0.10 < p ≤ 0.15).Table 7 Effects of older family member’s need for ADL help—interactions with marital status Women Men Hours worked Coefficient estimate Coefficient estimate ADL help − 0.431 0.385 Married 0.115 3.155*** ADL help * Married − 0.762 − 0.670 R2 0.0878 0.1100 Prob > F 0.0000 0.0000 Women Men Full time Marginal effect Marginal effect ADL help − 0.041** 0.014 Married − 0.010 0.108*** ADL help * Married − 0.020 − 0.052* Pseudo R2 0.0818 0.1180 Prob > chi2 0.0000 0.0000 Women Men Hours variance Coefficient estimate Coefficient estimate ADL help 0.035** − 0.018† Married 0.001 − 0.032*** ADL help * Married 0.014 0.041** R2 0.2216 0.2099 Prob > F 0.0000 0.0000 N 10,185 8016 Sample of working age family members drawn from the Medical Expenditures Panel Survey, 1996–2018. Year dummy variables included. Standard errors clustered by household. Wage interaction excluded *p < .05; **p < .01; ***p < .001; †p ≤ .10; ‡p ≤ .15; †p ≤ .10 For each of the analyses reported, the low probability of exceeding the critical value of the F statistic or the Wald X2 indicated that our empirical models of labor supply were statistically significant. The first column of estimates in each of Tables 4 through 6 reports the estimates from the baseline analysis that excluded the ADL help variable. These are provided solely to allow comparisons with the other analyses, but we note that the estimated effects of the variables included in Model 1 were stable across all three models. That is, the inclusion of the variables representing the presence of an older family member needing ADL help and the interaction of the ADL help variable with the wage rate had little impact on the estimated effects of the control variables. We therefore focus our discussion of findings on the estimates from Models 2 and 3 which included the ADL help variables. We begin in Table 4 by reporting the estimated effects of living with an older family member needing ADL assistance on the usual working hours of employed younger family members. The coefficient estimate for Model 2 in the second column of Table 4 indicates that the effect of providing ADL help was negative and significant for women. This means that women decreased their working hours in response to living with an older family member needing ADL assistance, other things equal. However, the insignificant estimate for Model 2 in the fifth column suggests that there was no impact on men’s working hours. The marginal effects on full-time employment for Model 2 in the second and fifth columns of Table 5 yield a similar picture: The effect for women was negative and statistically significant. This indicates that residing with an older family member requiring ADL assistance reduced women’s full-time employment by 5.0%, other things equal. Given the average full-time employment rate of 69% for the subsample of co-residing younger women whose older family member did not require ADL assistance, this effect implied a 7.2% decrease in full-time employment. In contrast, for men the estimated marginal effect for Model 2 was statistically insignificant. The findings reported for Model 2 in Tables 4 and 5 support our hypothesis H1 for women: Co-residing with an older family member requiring ADL help negatively impacted labor market outcomes, even when the woman’s wage rate was controlled for in the analysis. The findings also support hypothesis H4: Women’s labor market outcomes were more strongly affected than men’s. Not surprisingly, this means that factors such as social norms played a significant role in determining women’s labor supply response to living with an older family member needing ADL assistance.5 The third and sixth columns of Tables 4 and 5 report the estimated effects from our analyses of Model 3. Model 3 included a variable representing the direct effect of living with an older family member needing ADL assistance as well as a variable representing the interaction of ADL help with the wage rate. The estimated direct effect on women’s working hours of living with an older family member needing ADL help was insignificant, but we found a weakly significant negative effect on the probability of women working full-time. For men the estimated direct effects on working hours and the probability of working full-time were both positive and statistically significant. The interaction variable was included to test the hypothesis (H3) that the direct effect of living with an older family member needing ADL assistance diminished as the individual’s wage rate increased. For both working hours (Table 4) and the probability of working full-time (Table 5), the estimated effect of the interaction variable was statistically insignificant for women. Its inclusion caused the estimate of the direct effect of the ADL help variable to lose significance. However, for men the inclusion in the specification of the interaction variable caused the estimates of the direct effect to become positive and statistically significant, suggesting that men responded to living with an older family member needing ADL help by increasing both work hours and the probability of full-time employment. The coefficient estimate of the interaction variable for men was negative and statistically significant. The negative coefficient estimates for men of the interaction variable for men indicates that the results are consistent with our hypothesis H3: The impact of co-residing with an older family member needing ADL assistance decreased with higher wage rates, other things equal. However, the interpretation differs from our expectation that a negative interaction would imply that a negative direct effect is decreased. Instead, we observed that the negative interaction effect decreased a positive direct effect among men. The pattern of estimated effects indicates that among men the net effect of living with an older family member needing ADL help on working hours depended on the man’s wage rate: The estimated effect was positive for wage rates less than $19.24 but decreasing as the wage rate increased until the effect became negative for wage rates greater than $19.24.6 Converting this value to its $US 2022 value, the breakeven wage rate was $24.33. As reported in Table 5, the estimated marginal effect of the interaction variable for women in Model 3 was zero and its inclusion had no effect on the estimated marginal effect of the ADL help variable. However, the estimated marginal effect of the interaction variable for men in Model 3 was negative and statistically significant. Although its value was very small, its inclusion caused the point estimate of the marginal effect of the ADL help variable to become positive and statistically significant, suggesting that men responded to living with an older family member needing ADL help by increasing the probability of working full-time. Similar to the net effect on hours worked by men, the net effect of living with an older family member needing ADL help on men’s probability of working full-time depended upon the wage rate: The net effect of living with an older family member needing ADL help on men’s probability of working full-time was positive for men working at lower wage rates but decreased as the wage rate increased until the effect became negative for wage rates greater than $14.33. Converting this value to $US 2022 value, the breakeven wage rate was $18.04. Thus, we found a nonlinear impact: At lower wage rates men had a higher probability of working full-time, but this effect declined as the wage rate increased. A possible explanation for the effects on both hours worked and the probability of full-time employment is that men earning lower wage rates worked greater hours in response to the reduced earnings of their female spouses who decreased labor supply when living with an older family member needing ADL help. Table 6 reports the estimated effects of living with an older family member needing ADL assistance on the (logged) variance in usual hours of work per week for the occupational category of the younger family member’s job. In shown in the estimates for Model 2 reported in the second column, we found that women living with an older family member needing ADL help had jobs in occupational categories with greater variance in usual working hours. Consistent with hypothesis H2, this suggests that these women were employed in occupations with more flexibility in working hours. In comparison, the Model 2 estimate for men in the fifth column was statistically insignificant, suggesting that there was no effect on flexibility of men’s work hours. The estimates in the third and sixth columns of Table 6 refer to the Model 3 specification that included an interaction term. Among women, including the interaction term increased the magnitude of the positive direct effect of living with an older family member needing ADL help on the variance in work hours, although the interaction effect was itself not statistically significant. This implies that the net effect of co-residing with an older family member needing ADL help was positive for women. For men, inclusion of the interaction term caused the estimate of the direct effect of ADL help to become negative, although it remained statistically insignificant. The estimated effect of the interaction term was positive and statistically significant, although very small. If we ignore the lack of statistical significance and consider the values of the point estimates, the negative net effect of ADL help among men suggests that living with an older family member requiring ADL help decreased the variance in hours worked for men, but only minimally. While this is consistent with the effects reported in Tables 4 and 5, the lack of statistical significance for the estimate of the direct effect in Model 3 weakens this conjecture. To assess the sensitivity of the reported estimates to the specification of the model, we considered several further specifications: We estimated models including additional variables representing interactions of the ADL help variable with the race (Asian, Black, Other race) and Hispanic variables and with the presence of minor children in the home. The estimated effects of these interaction variables were statistically insignificant and their inclusion had no impact on our target estimates. The only significant interaction was that between the ADL help and Married variables. The estimates from this model, which we report in Table 7, indicate that among men the direct effect of being married increased working hours and the probability of working full-time. This is consistent with long-standing empirical results (Knowles, 2013; Pencavel, 1986). The interaction effect of living with an older family member needing ADL help reduced the probability of working full-time for married men, but the net effect of being married was positive. Similarly, married men worked in occupations with less variance in usual work hours, but the effect of living with an older family member needing ADL assistance counteracted this effect of marriage. In this case, the net effect of being married was slightly positive.7 Discussion and Conclusion In this paper we report the estimated effects on the labor supply of younger individuals of co-residing with an older family member. Our analyses estimated how living with an older family member needing help with activities of daily living (ADL) impacted the working hours, full-time employment, and flexibility of work hours of the younger workers. For our multivariate analyses we used a study sample of 10,185 working women and 8016 working men who reported living with older family members. The data were drawn from 23 years of the medical expenditure panel survey (MEPS). We report several findings that contribute to the literature: First, consistent with our hypotheses, we found that living with an older family member needing ADL assistance had a significant negative impact on the labor supply of younger women: Younger women worked fewer hours per week and were less likely to work full-time, other things equal. These findings are consistent with prior findings in the literature documenting the indirect effects of family illness. Observing these effects for women in our study sample confirms this phenomenon, which is generally attributed to caregiving. For example, Gould (2004) found that mothers worked fewer hours in the labor market when a child had an illness requiring time-intensive caregiving. Similarly, Siegel (2006) reported that a husband’s ill health led an employed wife to reduce hours of work. Second, in contrast to the findings for women, we found that men living with an older family member needing ADL help increased work hours and were more likely to work full-time, other things equal. These findings reinforce prior research such as Roberts (1999) and Coile (2004), which reported that family illness increased men’s labor supply. However, observing this effect among men living with older family members needing ADL assistance represents a significant contribution to this literature. Third, in addition to the effects on work hours and full-time employment, we found that younger women living with older family members needing ADL help worked in occupations with more flexible working hours, other things equal. Thus, it appears that younger women adjusted to the time burden of living with an older family member needing assistance both by reducing labor supply and by working in occupations that permit more flexible work schedules. This effect has been observed among women with greater family demands due to the presence of children, so it is not surprising. For example, Boden (1999) and Lombard (2001) reported that greater family demands led women to choose self-employment, offering greater flexibility in work hours. However, we believe this is the first time that it has been examined using this measure of flexibility and in the context of elder care. In contrast to our finding for women, we found little impact of living with an older family member needing ADL assistance on the flexibility of men’s work hours. While there are no prior studies for direct comparison, this finding is consistent with prior studies that reported men were more likely to be employed in occupations with relatively inflexible work schedules (Golden, 2008; Goldin, 2014). Both studies concluded that women seeking greater work schedule flexibility to manage childcare responsibilities were more likely to work in occupations in which the work schedule, both total hours and specified hours, was more flexible than the occupations in which men worked. While these studies did not address the effects of family illness, the findings are consistent with our finding that women were more likely than men to work in occupations with greater flexibility in work hours in response to living with an older family member needing ADL assistance. Fourth, we found support for another hypothesis we posited which suggested that the effect of living with an older family member needing ADL help would decrease at higher wage rates: We observed that the magnitude of the labor supply effects for men decreased at higher wages. Thus, the positive effect on work hours observed among men living with an older family member needing assistance was lessened among higher wage male workers. It may be that in families with lower earnings, financial needs dictated that income lost due to the wife reducing work hours was compensated for by the husband’s increased hours. If the families of higher wage men had lesser financial constraints, there may have been less need for these men to increase their work hours. In comparison to our finding for men, while a women’s wage rate was a significant determinant of her labor supply, we did not observe a reduction in the negative labor supply effects of living with an older family member needing ADL help among higher wage women. Thus, the evidence suggests that higher wage women did not use greater hours of work and higher earnings to purchase care for the older family member as a substitute for their own caregiving hours. These findings are consistent with evidence reported in several studies that men increased work hours in response to a family illness (Coile, 2004; Emanuel et al., 2000; Roberts, 1999). However, none of these studies considered the relationship between the response in hours of work and the family member’s wage rate. Finally, it is worthwhile to emphasize that our estimates demonstrate that the younger family member’s sex mattered even when the wage rate was controlled for in the analyses: The wage rate had a positive and significant effect on all of our measures of labor supply, but living with an older family member needing ADL help nevertheless had an independent negative impact on women’s labor supply and an independent positive effect on men’s labor supply. This implies that other factors, such as social norms, play a role that extends beyond earning potential in determining a woman’s response to living with an older family member needing assistance. This is consistent with studies that examine the interaction of economic forces and social norms (Burda et al., 2013; Maxwell & Wozny, 2021). Maxwell and Wozny found that social norms about work and home production explain perhaps 40% of the gender gap in time allocation for work in the labor market and work and household production. Social norms vary across cultures. Some cultures have much stronger expectations regarding responsibilities toward older family members than others. While we controlled for sociodemographic factors such as age, education, marital status and children, family income, health, race, Hispanicity, and region of residence, it was beyond the scope of this study to examine how cultural differences associated with race and ethnicity may have differentially influenced expectations about men’s and women’s roles in providing for older family members. With a larger data sample with richer information describing race and ethnicities and specific information describing caregiving activities, a more detailed picture may emerge of these subtle differences. Social norms also change with time. The roles of younger women and men in the family and the workplace have evolved over the 23 years considered in this study. While it is beyond the scope of this study to investigate these changes in depth, a cursory investigation using our study sample suggested that the negative impact on younger women’s labor supply was indeed somewhat reduced over time, although women living with older family members needing ADL help continued to experience significant negative effects on labor supply. In contrast, we found no change in the effect on men’s labor supply. This confirms research suggesting that despite women’s increased participation in the labor force over the past several decades, they continue to typically spend disproportionately more time on unpaid work in the home than men (Ferrant et al., 2014). The contributions of this research are strengthened by our use of a large, multi-year, nationally representative data set to examine the effects for both women and men in the family. Observations of the effects on both men and women provide a more accurate picture of the family burden of illness among older family members. However, despite using several years of data for our analyses we were not able to conduct panel data estimation, which would have permitted individual-specific unobserved factors to be controlled for in the analyses. To the extent that there may have been selection into co-residence, controlling for unobserved heterogeneity may have provided more accurate estimates. We were unable to address this selection because MEPS, like many large-scale data sets, did not include information describing younger family members who did not co-reside with the older individual needing assistance in our study sample. Our data also precluded consideration of the severity of the ADL impairment. Because this may be important in determining the effect on younger family members’ labor supplies, it would be useful to address this in future research using an alternative data source. Our findings have implications for how families handled elder care during the COVID-19 pandemic. In a review of studies focused on family functioning during the pandemic, Andrade et al. (2022) noted that despite some evidence of reallocation of household and childcare tasks in countries affected by the COVID-19 pandemic, gender disparities continued (p.206). With daycare in public spaces closed during the pandemic and families reluctant to bring care workers into the home, younger family members may have been more likely to care for older family members needing assistance themselves. Our findings suggest that younger women, while more likely to be in the labor force and career-oriented in recent decades, were still more likely to provide elder care during the pandemic just as they provided the bulk of child care for children whose schools closed (Igielnik, 2021). The empirical evidence of the family labor supply costs of living with older family members needing care suggests the importance of policy initiatives to support family members providing elder care. Our findings suggest that more public provision of elder care and public financial support of families providing care would reduce the lost working hours among women and ease the financial stress that appears to push men into working more hours. In the last decade, states have started to shift their Medicaid spending toward greater home care support for older citizens, which might provide more support to families with older family members needing assistance, but the level and type of care varies by state (Abrams, 2021). Another option for families is the use of family medical leaves to provide relief to workers needing release time to provide care to older family members, but there are restrictions on eligibility for federal FMLA leaves and state leaves. For example, Klerman et al. (2014) reported that approximately 40 percent of American workers did not qualify for FMLA because of the law’s other restrictions. These limitations are exacerbated for lower income families by their inability to afford unpaid leaves. Paid leaves, which are relatively less available and generally have replacement rates less than 100 percent, are also not a viable option for these families (Bureau of Labor Statistics, 2022). Most promising is a movement among employers toward greater flexibility in work schedules. A flexible workplace could provide employed family members providing care for older family members with the time they need to handle both unexpected events and scheduled events such as doctor’s appointments. Rather than forcing the caregiver to choose a job with a more flexible schedule, possibly incurring losses in wage rates, earnings, and career growth, employers may allow more flexibility across all positions. Ironically, while the COVID-19 pandemic may have increased the immediate family burden of elder care, it increased employers’ interest in flexible work schedules and working from home. Both of these practices may benefit younger workers living with older family members needing assistance, although it remains to be seen whether it will contribute to equalizing the burden of at home work between women and men. Overall, the findings reported in the paper provide information that is important for the design of public health and workplace policies that recognize the full impact of elder caregiving and appropriately target resources to alleviate its effects. To the extent that public and private sector policies support younger family members burdened by living with an older family member needing assistance, they benefit not only the younger individuals, but also the older family members who may delay or avoid entry into a nursing home, and employers who may retain full-time valuable employees. Although not cost-free, these policies have the potential to improve the family lives and workplaces of Americans. Funding No funding was received for conducting this study. Data Availability This study uses publicly available, anonymous data. Declarations Conflict of interest The authors have no competing interests to declare that are relevant to the content of this article. Research Involving in Human and Animal Participants Screening by the NIU Institutional Review Board indicated that it does not involve human subjects and further review was waived. 1 For occupation coding, MEPS groups the 2010 4-digit Census occupation codes into 11 categories. We excluded jobs in military occupations and unclassified occupations, as well as “not in labor force”, leaving eight broad categories. 2 An alternative measure of activities of daily living, the number of activities for which the older family member required assistance, was not available. We also considered including the number of older family members requiring ADL assistance in a household, but found that there were very few households with more than one older family member needing ADL assistance. 3 A drawback of the MEPS survey data was that it was not possible to link older persons to younger family members if they were not co-residing. However, it may be useful for readers to note how our study sample differed from the sample of MEPS respondents who were not co-residing with an older family member: Younger respondents in our study sample (who co-resided with an older family member) were more likely to be Asian and less likely to be married or have minor children in the household than other young MEPS respondents. They were six years older on average, reported six months less schooling, and reported poorer physical health status on average. They were less likely to work in a professional occupation and were more likely to work in a service occupation or in administrative support. Notably, other family income for younger family members in our study sample was significantly greater ($55,191) than that for young respondents not co-residing with an older family member ($46,517). 4 Both the selection bias corrected estimates and the two-part model estimates are available from the authors on request. 5 As social norms have changed over time, the roles of younger women and men in the family and the workplace may have evolved over the 23 years considered in this study. While it is beyond the scope of this study to investigate these changes in depth, a cursory investigation using our study sample suggested that the negative impact on younger women’s labor supply was indeed somewhat reduced over time, although women living with older family members needing ADL help continued to experience significant negative effects on labor supply. In contrast, we found no change in the effect on men’s labor supply. A table of estimated effects is available from the authors. 6 This estimate was obtained by treating the ADL help variable as continuous. 7 Separate analyses of married and unmarried men confirmed that married men living with older family members needing ADL help were significantly less likely to work full-time. They also worked in jobs with greater flexibility in usual hours worked, while unmarried men worked in jobs with less flexibility in usual hours worked. However, the effects on flexibility of usual hours worked were only weakly significant. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References AARP and National Alliance for Caregiving Caregiving in the United States 2020 2020 AARP Abrams, A. (2021). Covid-19 exposed the faults in America’s elder care system. This is our best shot to fix them. Time. https://time.com/6071582/elder-care-after-covid-19/ Agency for Healthcare Research and Quality Medical expenditures panel survey 1996 AHRO Andrade C Gillen M Molina JA Wilmarth MJ The social and economic impact of Covid-19 on family functioning and well-being: Where do we go from here? Journal of Family and Economic Issues 2022 43 205 212 10.1007/s10834-022-09848-x 35669394 Bargain O Peichl A Own-wage labor supply elasticities: Variation across time and estimation methods IZA Journal of Labor Economics 2016 5 10 10.1186/s40172-016-0050-z Baxter J Flexible work hours and other job factors in parental time with children Social Indicators Research 2011 101 239 242 10.1007/s11205-010-9641-4 Becker GS A theory of the allocation of time Economic Journal 1965 75 299 493 517 10.2307/2228949 Berger MC Fleischer BM Husband’s health, wife’s labor supply Journal of Health Economics 1984 3 63 75 10.1016/0167-6296(84)90026-2 10266616 Boaz RF Muller CF Paid work and unpaid help by caregivers of the disabled and frail elders Medical Care 1992 30 2 149 158 10.1097/00005650-199202000-00006 1736020 Boden RJ Flexible working hours, family responsibilities, and female self-employment American Journal of Economics and Sociology 1999 58 1 71 83 10.1111/j.1536-7150.1999.tb03285.x Burda M Hamermesh D Weil P Total work and gender: Facts and possible explanations Journal of Population Economics 2013 26 239 261 10.1007/s00148-012-0408-x Bureau of Labor Statistics Family leave fact sheet 2022 U.S Department of Labor Carmichael F Charles S Hulme C Who will care? Employment participation and willingness to supply informal care Journal of Health Economics 2010 29 182 190 10.1016/j.jhealeco.2009.11.003 20004991 Chang CF White-Means SI Labour supply of informal caregivers International Review of Applied Economics 2006 9 2 192 205 10.1080/758538252 Chiappori PA Lewbel A Gary Becker’s a theory of the allocation of time The Economic Journal 2015 125 583 410 442 10.1111/ecoj.12157 Ciani E Informal adult care and caregivers’ employment in Europe Labour Economics 2012 19 155 164 10.1016/j.labeco.2011.12.001 Coile, C.C. (2004). Health shocks and couples’ labor supply decisions. (NBER Working Paper No.10810). Retrieved from National Bureau of Economics Research website: https://www.nber.org/papers/w10810 Dautzenberg M Diederiks J Philipsen H Stevens F Tan F Vernooij-Dassen M The competing demands of paid work and parent care: middle aged daughters providing assistance to elderly parents Research on Aging 2000 22 165 187 10.1177/0164027500222004 Duan N Manning WG Morris CN Newhouse JP A comparison of alternative models for the demand for medical care Journal of Business & Economic Statistics 1983 1 2 115 126 10.2307/1391852 Ehrenberg RG Smith RS Hallock KF Modern labor economics: Theory and public policy 2023 Routledge Emanuel EJ Fairclough DL Slutsman J Emanuel LL Understanding economic and other burdens of terminal illness: The experience of patients and their caregivers Annals of Internal Medicine 2000 132 6 451 459 10.7326/0003-4819-132-6-200003210-00005 10733444 Ettner SL The impact of ‘parent care’ on female labor supply decisions Demography 1995 32 1 63 80 10.2307/2061897 7774731 Ettner SL The opportunity costs of elder care The Journal of Human Resources 1995 31 1 189 205 10.2307/146047 Ferrant G Pesando LM Nowacka K Unpaid care work: The missing link in the analysis of gender gaps in labour outcomes 2014 OECD Development Centre Fevang E Kverndokk S Røed K Labor supply in the terminal stages of lone parents’ lives Journal of Population Economics 2012 25 1399 1422 10.1007/s00148-012-0402-3 Franks DD Economic contribution of families caring for persons with severe and persistent mental illness Administration and Policy in Mental Health 1990 18 9 18 10.1007/BF00706488 Fuller-Thomson E Minkler M American grandparents providing extensive child care to their grandchildren: Prevalence and profile The Gerontologist 2001 41 2 201 209 10.1093/geront/41.2.201 11327486 Golden L Limited access: Disparities in flexible work schedules and work-at-home Journal of Family and Economic Issues 2008 29 86 109 10.1007/s10834-007-9090-7 Goldin C A grand gender convergence: Its last chapter American Economic Review 2014 104 4 1091 1119 10.1257/aer.104.4.1091 Goldin C Pekkala Kerr S Olivetti C Barth E The expanding gender earnings gap: Evidence from the LEHD-2000 census American Economic Review: Papers & Proceedings 2017 107 5 110 114 10.1257/aer.p20171065 Gould E Decomposing the effects of children’s health on mother’s labor supply: Is it time or money? Health Economics 2004 13 6 525 541 10.1002/hec.891 15185384 Heckman JJ Shadow wages, market wages and labor supply Econometrica 1974 42 4 679 693 10.2307/1913937 Igielnik R A rising share of working parents in the US say it’s been difficult to handle child care during the pandemic 2021 Pew Research Center Kanji S Grandparent care: A key factor in mothers’ labour force participation in the UK Journal of Social Policy 2018 47 3 523 542 10.1017/S004727941700071X Kemper P The use of formal and informal home care by the disabled elderly Health Services Research 1992 27 4 421 451 1399651 Killingsworth MR Heckman JJ Ashenfelter O Layard R Female labor supply: a survey Handbook of labor economics 1986 Elsevier Science Publishers 103 204 Klerman JA Daley K Pozniak A Family and medical leave in 2012: Technical report (2014 revision) 2014 Abt Associates Knowles JA Why are married men working so much? An aggregate analysis of intra-household bargaining and labour supply Review of Economic Studies 2013 80 3 1055 1085 10.1093/restud/rds043 Langa KM Chernew ME Kabeto MU Herzog AR Ofstedal MB Willis RJ Wallace RB Mucha LM Straus WL Fendrick AM National estimates of the quantity and cost of informal caregiving for the elderly with dementia Journal of General Internal Medicine 2001 16 11 770 778 10.1111/j.1525-1497.2001.10123.x 11722692 Latif E Labour supply effects of informal caregiving in Canada Canadian Public Policy Analyse De Politiques 2006 32 413 429 10.3138/Q533-8847-3785-1360 Leigh A Informal care and labor market participation Labour Economics 2010 17 140 149 10.1016/j.labeco.2009.11.005 Leon J Cheng CK Neumann PJ Dementia care: Costs and potential savings Health Affairs 1998 17 6 206 216 10.1377/hlthaff.17.6.206 9916370 Lombard KV Female self-employment and demand for flexible, nonstandard work schedules Economic Inquiry 2001 39 2 214 237 10.1111/j.1465-7295.2001.tb00062.x Marcotte DE Wilcox-Gök V Estimating the indirect costs of mental illness: Recent developments in the United States Social Science & Medicine 2001 53 1 21 27 10.1016/S0277-9536(00)00312-9 11386306 Marcotte DE Wilcox-Gök V Redmon DP Salkever D Sorkin A The labor market effects of mental illness: The case of affective disorders The economics of disability 2000 JAI Press 181 210 Matsudaira JD Economics conditions and the living arrangements of young adults: 1960 to 2011 Journal of Population Economics 2016 29 167 195 10.1007/s00148-015-0555-y Max W Webber P Fox P Dementia: the unpaid burden of caring Journal of Aging and Health 1995 7 2 179 199 10.1177/089826439500700202 10172777 Maxwell NL Wozny N Gender gaps in time use and labor market outcomes: What’s norms got to do with it? Journal of Labor Research 2021 42 56 77 10.1007/s12122-020-09306-3 Mazzotta F Bettio F Zigante V Eldercare hours, work hours and perceived filial obligations Applied Economics 2020 52 21 2219 2238 10.1080/00036846.2019.1687839 Meng A Informal home care and labor-force participation of household members Empirical Economics 2013 44 959 979 10.1007/s00181-011-0537-1 Mentzakis E McNamee P Ryan M Who cares and how much: Exploring the determinants of co-residential informal care Review of the Economics of the Household 2009 7 283 303 10.1007/s11150-008-9047-0 Mincer JJ Lewis HG Labor force participation of married women: A study of labor supply Aspects of labor economics 1962 Princeton University Press 63 106 Muurinen JM The economics of informal care: Labor market effects in the National Hospice Study Medical Care 1986 24 11 1007 1017 10.1097/00005650-198611000-00005 3773575 National Academies of Sciences, Engineering, and Medicine Families caring for an aging America 2016 The National Academies Press Pencavel J Ashenfelter O Layard R Labor supply of men: a survey Handbook of labor economics 1986 Elsevier Science Publishers 3 102 Puhani PA The Heckman correction for sample selection and its critique Journal of Economic Surveys 2000 14 1 53 68 10.1111/1467-6419.00104 Roberts AA The labor market effects of an ill family member Journal of Mental Health Economics and Policy 1999 2 4 183 195 10.1002/(SICI)1099-176X(199912)2:4<183::AID-MHP62>3.0.CO;2-1 Rubenstein LM DeLeo A Chrischilles EA Economic and health-related quality of life considerations of new therapies in Parkinson’s disease PharmacoEconomics 2001 19 7 729 752 10.2165/00019053-200119070-00003 11548910 Schmitz H Westphal M Informal care and long-term labor market outcomes Journal of Health Economics 2017 56 1 17 10.1016/j.jhealeco.2017.09.002 28946010 Siegel MJ Measuring the effect of husband's health on wife's labor supply Health Economics 2006 15 6 579 601 10.1002/hec.1084 16450342 Skira M Dynamic wage and employment effects of elder parent care International Economic Review 2015 56 1 63 93 10.1111/iere.12095 Stanfors M Jacobs JC Neilson J Caregiving time costs and trade-offs: Gender differences in Sweden, the UK, and Canada SSM Population Health 2019 9 100501 10.1016/j.ssmph.2019.100501 31720360 Stommel M Collins CE Given BA The costs of family contributions to the care of persons with dementia The Gerontologist 1994 34 2 199 205 10.1093/geront/34.2.199 8005492 Taylor DH Jr Sloan FA How much do persons with dementia cost Medicare? Journal of the American Geriatrics Society 2000 48 6 639 646 10.1111/j.1532-5415.2000.tb04721.x 10855599 Van Houtven CH Coe NB Skira MM The effect of informal care on work and wages Journal of Health Economics 2013 32 1 240 252 10.1016/j.jhealeco.2012.10.006 23220459 Weinberger M Gold DT Divine GW Cowper PA Hodgson LG Schreiner PJ George LK Expenditures in caring for patients with dementia who live at home American Journal of Public Health 1993 83 3 338 341 10.2105/AJPH.83.3.338 8438969 Whetton-Goldstein K Sloan F Kulas E Cutson T Schenkman M The burden of Parkinson’s disease on society, family, and the individual Journal of the American Geriatrics Society 1997 45 7 844 849 10.1111/j.1532-5415.1997.tb01512.x 9215336 White-Means SI Allocation of labor to informal home health production: Health care for frail elderly, if time permits The Journal of Consumer Affairs 1992 26 1 69 89 10.1111/j.1745-6606.1992.tb00016.x Wolf DA Soldo BJ Married women's allocation of time to employment and care of elderly parents Journal of Human Resources 1994 29 1259 1276 10.2307/146140

0

PMC9748888

NO-CC CODE

2022-12-15 23:22:47

no

J Fam Econ Issues. 2022 Dec 14;:1-19

utf-8

J Fam Econ Issues

10.1007/s10834-022-09880-x

oa_other

==== Front Nat Cardiovasc Res Nat Cardiovasc Res Nature Cardiovascular Research 2731-0590 Nature Publishing Group UK London 194 10.1038/s44161-022-00194-7 Research Briefing Potential POTS association with COVID-19 vaccination weaker than with COVID-19 infection 14 12 2022 12 © Springer Nature Limited 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Postural orthostatic tachycardia syndrome (POTS) has been observed following SARS-CoV-2 infection. In this study, we observed occurences of POTS following COVID-19 vaccination, albeit at a lower rate than following COVID-19 infection. Subject terms Cardiology Infectious diseases ==== Body pmcThe question The COVID-19 pandemic has profoundly affected worldwide health through both acute illness and the long-term effects of infection. Vaccination efforts against COVID-19 have been instrumental in limiting transmission and preventing severe disease. However, anecdotal reports of postural orthostatic tachycardia syndrome (POTS) developing after COVID-19 vaccination have also emerged. POTS is a clinical syndrome that manifests with orthostatic intolerance and postural tachycardia1. Although POTS was previously recognized as a condition that can develop after SARS-CoV-2 infection2, data on its potential development after vaccination have been scarce. As COVID-19 vaccines will continue to be necessary for the foreseeable future, pharmacovigilance regarding off-target effects will improve our ability to engage patients in a balanced discussion on the relative benefits versus the risks of vaccination. Thus, we examined the extent to which new diagnoses of POTS are increased following COVID-19 vaccination in our health system, a large quaternary care system in Los Angeles. The observation We set out to identify the incidence of new POTS-related diagnoses that occur after COVID-19 vaccination. This type of analysis is inherently difficult owing to the lack of ideal controls, multiple confounding factors related to the context of the pandemic and the effect of COVID-19 vaccination on patients’ engagement with the healthcare system. We recognized that pharmaco-epidemiology studies had previously used self-controlled designs, wherein patients serve as their own control3. We also recognized that by comparing the incidence of POTS-related diagnoses against the incidence of common diagnoses that are not POTS-related (for example, urinary tract infections, gastroesophageal reflux disease and hypertension), we could provide a benchmark that can account for healthcare engagement. Using a sequence-symmetry analysis in over 280,000 COVID-19 vaccination records, we compared the odds of POTS-related diagnosis after exposure to COVID-19 versus before exposure, and compared these results with those for ‘common primary care’ diagnoses that are not POTS-related. A similar analysis was performed in over 12,000 SARS-CoV-2 infection records. We found that most POTS-related diagnoses increased after COVID-19 vaccination to a greater degree than non-POTS-related diagnoses (Fig. 1a). The odds of developing POTS-related diagnoses after vaccination compared with before vaccination was 1.33 (1.25–1.41), and the odds ratio of post-vaccine POTS-related versus non-POTS-related diagnoses was 1.10 (1.03–1.17). In the SARS-CoV-2 population, for most conditions studied, post-infection rates were higher than post-vaccination rates (Fig. 1b). For POTS-related diagnoses, the post-infection risk was 5.35 (5.05–5.68) times higher after exposure to SARS-CoV-2 infection than after exposure to vaccination. From these results, we can conclude that POTS might be occurring at a higher-than-expected frequency following COVID-19 vaccination, although at an overall rate lower than the frequency of POTS occurring following SARS-CoV-2 infection.Fig. 1 Odds of diagnoses post-vaccination and post-infection. a, Post-vaccination odds by diagnosis in all patients. b, Post-infection odds by diagnosis in all patients. EDS, Ehlers-Danlos syndrome; GERD, gastroesophageal reflux disease; IDA, iron deficiency anemia; UTI, urinary tract infection. © 2022, Kwan, A. C. The interpretation We hope that our results can offer an evidence-based context of understanding for patients who experience POTS-like symptoms after COVID-19 vaccination, as well as their healthcare providers. As POTS is a frequently under-recognized condition that can cause substantial debility in affected patients4, greater understanding of POTS in general is required to improve overall awareness and rates of appropriate diagnosis and treatment. These study results are not intended to discourage use of the COVID-19 vaccine, especially given the relatively higher risk of developing POTS after SARS-CoV-2 infection. Given the complex design of our study, which was intended to account for potentially important confounders, the estimated risks are reported in relative terms; that is, after exposure relative to before exposure or POTS-related diagnoses relative to non-POTS-related diagnoses. Therefore, deriving precise estimates of risk per exposure is difficult, and follow-up studies are needed to determine the absolute risk of POTS after vaccination. Further studies are also necessary to understand the pathophysiology underlying both POTS occurring after vaccine and POTS occurring after infection. Alan C. Kwan & Susan Cheng Smidt Heart Institute, Cedars Sinai Medical Center, Los Angeles, CA, USA Expert opinion “Kwan and colleagues observed, from a large health organization clinical database, a potential increase in the incidence of POTS following COVID-19 vaccination. Because POTS can be disabling for patients, increased awareness of the existence and increased risks of this complication can help physicians and patients in recognizing this syndrome early, allowing appropriate natural history studies, developing risk prediction tools and potentially evaluating different interventions for future mitigation.” Peter Liu, University of Ottawa Heart Institute, Ottawa, Canada. Behind the paper Our hospital played a substantial role in both care for patients and vaccination efforts in Los Angeles during the initial phases of the COVID-19 pandemic. Through this work, and institutional support for research efforts, we have been able to establish a large number of studies on COVID-19 infection and related therapies. A number of physicians and patient groups subsequently reached out to us with their experiences and observations, which included increased patient presentations with POTS-like symptoms after COVID-19 vaccination. Following up on this information, we developed the initial question about POTS after COVID-19 vaccination. We hope that our results feed back to the community in a positive way and spur further research to solve the issues presented to the medical community as a whole by the COVID-19 pandemic. A.C.K. & S.C. From the editor “The study by Kwan et al. shows that anti-SARS-CoV-2 vaccination can increase the incidence of POTS, although at a much lower frequency than the viral infection itself, and indicates the need to study the link between this dysfunction of the autonomic nervous system and the SARS-CoV-2 spike protein.” Elvira Forte, Associate Editor, Nature Cardiovascular Research This is a summary of: Kwan, A. C. et al. Apparent risks of postural orthostatic tachycardia syndrome diagnoses after COVID-19 vaccination and SARS-CoV-2 COVID infection. Nat. Cardiovasc. Res. 10.1038/s44161-022-00177-8 (2022). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Sheldon RS 2015 heart rhythm society expert consensus statement on the diagnosis and treatment of postural tachycardia syndrome, inappropriate sinus tachycardia, and vasovagal syncope Heart Rhythm 2015 12 e41 e63 10.1016/j.hrthm.2015.03.029 25980576 2. Jamal SM Prospective evaluation of autonomic dysfunction in post-acute sequela of COVID-19 J. Am. Coll. Cardiol. 2022 79 2325 2330 10.1016/j.jacc.2022.03.357 35381331 3. Takeuchi Y Shinozaki T Matsuyama Y A comparison of estimators from self-controlled case series, case-crossover design, and sequence symmetry analysis for pharmacoepidemiological studies BMC Med. Res. Methodol. 2018 18 1 15 10.1186/s12874-017-0457-7 29301497 4. Shaw BH The face of postural tachycardia syndrome - insights from a large cross-sectional online community-based survey J. Intern. Med. 2019 286 434 448 10.1111/joim.12895

0

PMC9748890

NO-CC CODE

2022-12-15 23:22:47

no

Nat Cardiovasc Res. 2022 Dec 14;:1-2

utf-8

Nat Cardiovasc Res

10.1038/s44161-022-00194-7

oa_other

==== Front Indian Pediatr Indian Pediatr Indian Pediatrics 0019-6061 0974-7559 Springer India New Delhi 2650 10.1007/s13312-022-2650-y Clinical Case Letters Concurrent Scrub Typhus and Dengue Fever Mimicking Acute Appendicitis Amritha Jan 1 Raveenthiran Venkatachalam vrthiran@gmail.com 2 1 Department of Pediatrics, Government Cuddalore Medical College, Chidambaram, Tamil Nadu India 2 Department of Pediatric Surgery, Government Cuddalore Medical College, Chidambaram, Tamil Nadu India 14 12 2022 2022 59 11 885886 © Indian Academy of Pediatrics 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. issue-copyright-statement© Indian Academy of Pediatrics 2022 ==== Body pmc ==== Refs References 1. Anitharaj V Gunasekaran D Pradeep J Stephen S Gastrointestinal manifestations of scrub typhus in children and adults from Puducherry and neighbouring Tamil Nadu State, India J Gastrointest Infect 2017 7 1 4 10.5005/jogi-7-1-1 2. Sapkota S Bhandari S Sapkota S Hamal R Dengue and scrub typhus co-infection in a patient presenting with febrile illness Case Rep Infect Dis 2017 2017 6214083 28386493 3. Iqbal N Viswanathan S Remalayam B Pancreatitis and MODS due to scrub typhus and dengue co-infection Trop Med Health 2012 40 19 21 10.2149/tmh.2012-07 22949803 4. Jayasundara B Perera L de Silva A Dengue fever may mislead the surgeons when it presents as an acute abdomen Asian Pacific J Trop Med 2017 10 15 19 10.1016/j.apjtm.2016.12.010 5. Mahajan SK Babu SN Sharma D Scrub typhus presenting as acute abdomen Trop Doct 2011 41 185 6 10.1258/td.2011.110079 21724691 6. Kumarasena L Piranavan P Bandara S A case of dengue fever with acute appendicitis: not dengue fever mimicking appendicitis Sri Lanka J Surg 2014 32 33 35 10.4038/sljs.v32i3.8111 7. Mcfarlane MEC Plummer JM Leake PA Dengue fever mimicking acute appendicitis: A case report Int J Surg Case Rep 2013 4 1032 34 10.1016/j.ijscr.2013.08.017 24096347 8. Thipmontree W Suwattanabunpot K Supputtamonkol Y Spontaneous splenic rupture caused by scrub typhus Am J Trop Med Hyg 2016 95 1284 86 10.4269/ajtmh.16-0089 27698275 9. Chang PH Cheng YP Chang PS A case report and literature review of scrub typhus with acute abdomen and septic shock in a child — the role of leukocytoclastic vasculitis and granulysin Am J Dermatopathol 2018 40 767 71 10.1097/DAD.0000000000001167 29697421 10. Lee CH Lee JH Yoon KJ Peritonitis in patients with scrub typhus Am J Trop Med Hyg 2012 86 1046 48 10.4269/ajtmh.2012.11-0586 22665616 11. Masyeni S Santoso MS Widyaningsih PD Serological cross-reaction and co-infection of dengue and COVID-19 in Asia: Experience from Indonesia Int J Infect Dis 2021 102 152 54 10.1016/j.ijid.2020.10.043 33115680

36370019

PMC9748892

NO-CC CODE

2022-12-15 23:22:47

no

Indian Pediatr. 2022 Dec 14; 59(11):885-886

utf-8

Indian Pediatr

10.1007/s13312-022-2650-y

oa_other

==== Front Indian Pediatr Indian Pediatr Indian Pediatrics 0019-6061 0974-7559 Springer India New Delhi 2649 10.1007/s13312-022-2649-4 Special Article Defensive Medicine in the Context of the Indian Health System Chaudhary Ankit 1 Barwal Vijay Kumar barwalvk@gmail.com 2 1 grid.459475.e 0000 0004 1800 6232 Department of Community Medicine, Dr Rajendra Prasad Government Medical College, Kangra, India 2 grid.414489.4 0000 0004 1768 2079 Department of Community Medicine, Indira Gandhi Medical College, Shimla, Himachal Pradesh India 14 12 2022 2022 59 11 882884 © Indian Academy of Pediatrics 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Defensive medicine; although a recent concept, is slowly beginning to cement its place in the Indian health system. An interaction of multiple factors has paved way for this form of practice. Need for certainty of the diagnosis, lack of hierarchy in medical care, exponential growth of micro/super specializations and private/corporate health institutions, incentive-based practice, increasing incidences of violence against health personnel, rising trend of defamation suites against doctors, bad publicity by media, and interference by elected representatives have jeopardized the situation. This has led to decline in practice of clinical medicine, increased burden of investigations, especially in already compromised public facilities, and high out-of-pocket health expenditure. As much as ethical medical practice, standard patient management protocols, strict protection of interest of medical practitioners by law, responsible role of media and elected representatives are the need of the hour; we need to find ways to accept and incorporate defensive medicine into the modern medicine. Different stakeholders are required to come together and take substantial steps to understand the phenomenon and preserve the art and science of practicing medicine in its true form. Keywords Malpractice Medical Protection Act Protocols Violence against health personnel issue-copyright-statement© Indian Academy of Pediatrics 2022 ==== Body pmc ==== Refs References 1. Chaturvedi S Practicing defensive medicine benefits no one Sushruta Journal of Health Policy and Opinion 2021 14 1 4 10.38192/14.2.5 2. Toker A Shvarts S Perry ZH Clinical guidelines, defensive medicine, and the physician between the two Am J Otolaryngol 2004 25 245 50 10.1016/j.amjoto.2004.02.002 15239030 3. Carroll AE The high costs of unnecessary care JAMA 2017 318 1748 9 10.1001/jama.2017.16193 29136432 4. Miyakis S Karamanof G Liontos M Mountokalakis TD Factors contributing to inappropriate ordering of tests in an academic medical department and the effect of an educational feedback strategy Postgrad Med J 2006 82 823 9 10.1136/pgmj.2006.049551 17148707 5. Sekhar MS Vyas N Defensive medicine: A bane to healthcare Ann Med Health Sci Res 2013 3 295 6 10.4103/2141-9248.113688 23919211 6. Korula RJ. Fee-splitting: The unethical side of medical practice today. Matters of the Heart. March 04, 2017. Accessed February 08, 2022. Available from: https://www.onmanorama.com/lifestyle/health/matters-of-the-heart-fee-splitting-unethical-medical-practice-india.html 7. Ghosh K Violence against doctors: A wake-up call Indian J Med Res 2018 148 130 3 10.4103/ijmr.IJMR_1299_17 30381535 8. Dora SSK Batool H Nishu RI Hamid P Workplace violence against doctors in India: A traditional review Cureus 2020 12 e8706 32699702 9. McClellan MK Do doctors practice defensive medicine? Q J Econ 1996 11 353 90 10. Van RN Auwerx K Debbaut P The effect of Dr. Google on doctor-patient encounters in primary care: A quantitative, observational, cross-sectional study BJGP Open 2017 1 bjgpopen17X100833 10.3399/bjgpopen17X100833 11. Fenton JJ Jerant AF Bertakis KD Franks P The cost of satisfaction: A national study of patient satisfaction, health care utilization, expenditures, and mortality Arch Intern Med 2012 172 405 11 10.1001/archinternmed.2011.1662 22331982 12. Agrawal A Medical negligence: Indian legal perspective Ann Indian Acad Neurol 2016 19 9 14 10.4103/0972-2327.192889 27011622 13. Singh RV. Satyamev Jayate: Aamir targets corrupt doctors. News 18, India. May 27, 2012. Accessed February 10, 2022. Available from: https://www.news18.com/news/indiasatyamev-jayate-aamir-targets-corrupt-doctors-478298.html 14. Markose A Krishnan R Ramesh M Medical ethics J Pharm Bioallied Sci 2016 8 1 4 26957861 15. Fred HL Cutting the cost of health care: The physician’s role Tex Heart Inst J 2016 43 4 6 10.14503/THIJ-15-5646 27047277 16. Srinivisan R. Health care in india-vision 2020 issues and prospects. Accessed February 10, 2022. Available from: https://niti.gov.in/planningcommission.gov.in/docs/reports/genrep/bkpap2020/26_bg2020.pdf 17. Rajivlochan M Clinical audits and the state of record keeping in India Ann Neurosci 2015 22 197 8 10.5214/ans.0972.7531.220402 26527163 18. Anil Kumar G Shweta T Sudip B Amarjeet S Health system strengthening-focusing on referrals: An analysis from India JOJ Nurse Health Care 2017 2 555592 19. Martins C Godycki-Cwirko M Heleno B Brodersen J Quaternary prevention: Reviewing the concept Eur J Gen Pract 2018 24 106 11 10.1080/13814788.2017.1422177 29384397 20. Current Health Expenditure India. World Bank. January 30, 2022. Accessed February 11, 2022. Available from: https://data.worldbank.org/indicator/SH.XPD.CHEX.GD.ZS?locations=IN 21. Vento S Cainelli F Vallone A Defensive medicine: It is time to finally slow down an epidemic World J Clin Cases 2018 6 406 9 10.12998/wjcc.v6.i11.406 30294604 22. Coronavirus-Attacks on health workers to attract up to 7 years in prison. The Hindu, New Delhi. April 22nd 2020. Accessed February 10, 2022. Available from: https://www.thehindu.com/news/national/coronavirus-attacks-on-health-workers-to-attract-up-to-7-year.s-in-pri.son/article31404910.ece 23. Kavadi SN Autonomy for medical institutes in India: A view from history Natl Med J India 2017 30 230 4 10.4103/0970-258X.218680 29162760 24. National Health Policy 2017. Ministry of Health and family Welfare, Government of India. Accessed February 11, 2022. Available from: https://www.nhp.gov.in/nhpfiles/national_health_policy_2017.pdf 25. Al-Balas QAE Al-Balas HAE The ethics of practicing defensive medicine in Jordan: A diagnostic study BMC Med Ethics 2021 22 87 10.1186/s12910-021-00658-8 34229676 26. Attitude, Ethics and Communication (AETCOM) Competencies for the Indian Medical Graduate 2018. Medical Council of India. Accessed August 16, 2022. Available from: https://www.nmc.org.in/wp-content/uploads/2020/01/AETCOM_book.pdf

36370018

PMC9748894

NO-CC CODE

2022-12-15 23:22:48

no

Indian Pediatr. 2022 Dec 14; 59(11):882-884

utf-8

Indian Pediatr

10.1007/s13312-022-2649-4

oa_other

==== Front Indian Pediatr Indian Pediatr Indian Pediatrics 0019-6061 0974-7559 Springer India New Delhi 2639 10.1007/s13312-022-2639-6 Perspective Childhood and Adolescent Anemia Burden in India: The Way Forward Kurpad Anura Viswanath 1 Sachdev Harshpal Singh hpssachdev@gmail.com 2 1 grid.416432.6 0000 0004 1770 8558 Department of Physiology, St John’s Medical College, Bengaluru, Karnataka India 2 grid.419277.e 0000 0001 0740 0996 Pediatrics and Clinical Epidemiology, Sitaram Bhartia Institute of Science and Research, B-16 Qutab Institutional Area, New Delhi, 110 016 India 26 8 2022 2022 59 11 837840 © Indian Academy of Pediatrics 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The burden of anemia in Indian children, based on capillary blood sampling, is believed to be profound and worsening (67.1%) according to the successive National Family Health Surveys (NFHS). This might be an overestimate. The recent Comprehensive National Nutrition Survey of Indian children, that used venous blood sampling, found only less than half (30.7%) the NFHS prevalence, of which only one third was due to iron deficiency (ID). Unfortunately, the apparently worsening NFHS anemia burden estimate has been interpreted as an inadequacy of the present iron supplementation policy. This has led to additional iron supply through mandatory rice fortification. However, the lack of efficacy of iron supplementation appears inevitable, if the true prevalence of iron deficiency anemia is only about 10%. Thus, etiology is a critical consideration when devising appropriate and effective prevention policies. Future policies must focus on precision, thoughtfulness, restraint, and community engagement. Keywords Fortification Hemoglobin Prevalence Iron deficiency issue-copyright-statement© Indian Academy of Pediatrics 2022 ==== Body pmcAcknowledgements AVK and HSS are recipients of the Wellcome Trust/Department of Biotechnology India Alliance Clinical/Public Health Research Centre Grant (IA/CRC/19/1/610006). AVK is also supported by the India Alliance through their Margdarshi Fellowship. Contributors: Conceptualized, drafted and finalized by both authors, who will be equally accountable for the content. Funding: None Competing interests: HSS is a member of the WHO Guideline Development Group on Anemia: Use and interpretation of haemoglobin concentrations for assessing anemia status in individuals and populations. AVK and HSS are members of the National Technical Board on Nutrition and of Task Forces and Expert Groups on anemia constituted by the Department of Biotechnology, Ministry of Health and Family Welfare and the Indian Council of Medical Research. ==== Refs References 1. National Family Health Survey, India. NFHS-4. Accessed July 15, 2022. Available from: http://rchiips.org/nfhs/nfhs4.shtml 2. National Family Health Survey, India. NFHS-5. Accessed July 15, 2022. Available from: http://rchiips.org/nfhs/nfhs5.shtml 3. Ministry of Health and Family Welfare (MoHFW), Government of India, UNICEF and Population Council. 2019. Comprehensive National Nutrition Survey (CNNS) National Report. Accessed July 15, 2022. Available from: https://nhm.gov.in/WriteReadData/1892s/1405796031571201348.pdf 4. Sarna A Porwal A Ramesh S Characterisation of the types of anemia prevalent among children and adolescents aged 1–19 years in India: A population-based study Lancet Child Adolesc Health 2020 4 515 25 32562633 5. Neufeld LM Larson LM Kurpad A Hemoglobin concentration and anemia diagnosis in venous and capillary blood: biological basis and policy implications Ann N Y Acad Sci 2019 1450 172 89 31231815 6. Varghese JS Thomas T Kurpad AV Evaluation of haemoglobin cut-off for mild anemia in Asians — analysis of multiple rounds of two national nutrition surveys Indian J Med Res 2019 150 385 89 31823920 7. World Health Organization. Haemoglobin concentrations for the diagnosis of anemia and assessment of severity. 2011. Accessed July 15, 2022. Available from: https://apps.who.int/iris/bitstream/handle/10665/85839/WHO_NMH_NHD_MNM_11.1_eng.pdf?ua=1 8. Johnson-Spear MA Yip R Hemoglobin difference between black and white women with comparable iron status: justification for race-specific anemia criteria Am J Clin Nutr 1994 60 117 21 8017324 9. Sachdev HS Porwal A Acharya R Haemoglobin thresholds to define anemia in a national sample of healthy children and adolescents aged 1–19 years in India: A population-based study Lancet Glob Health 2021 9 e822 e831 33872581 10. Addo OY Yu EX Williams AM Evaluation of hemoglobin cut-off levels to define anemia among healthy individuals JAMA Netw Open 2021 4 e2119123 34357395 11. Indian Council of Medical Research. Nutrient Requirements and Recommended Dietary Allowance for Indians. A Report of the Expert Group of the Indian Council of Medical Research. ICMR-National Institute of Nutrition, 2010. 12. Ghosh S Sinha S Thomas T Sachdev HS Kurpad AV Revisiting Dietary Iron Requirement and Deficiency in Indian Women: Implications for Food Iron Fortification and Supplementation J Nutr 2019 149 366 71 30753562 13. Indian Council of Medical Research. Nutrient Requirements for Indians. A Report of the Expert Group. ICMR-National Institute of Nutrition, 2020. Accessed July 15, 2022. Available from: https://www.nin.res.in/RDA_Full_Report_2020.html 14. Kulkarni B Peter R Ghosh S Prevalence of iron deficiency and its sociodemographic patterning in Indian children and adolescents: findings from the Comprehensive National Nutrition Survey 2016–18 J Nutr 2021 151 2422 434 34049401 15. Coffey D Geruso M Spears D Sanitation, disease externalities and anemia: Evidence From Nepal Econ J (London) 2018 128 1395 432 16. Mehta U Dey S Chowdhury S The association between ambient PM2.5 exposure and anemia outcomes among children under five years of age in India Environ Epidemiol 2021 5 e125 33778358 17. Kurpad AV Ghosh S Thomas T Perspective: When the cure might become the malady: the layering of multiple interventions with mandatory micronutrient fortification of foods in India Am J Clin Nutr 2021 114 1261 266 34320172 18. Ghosh S, Thomas T, Kurpad AV, Sachdev HS. Is iron status associated with markers of non- communicable disease in Indian children? Preprint, Research Square; December 3, 2021. [preprint] 19. Walter PB Fung EB Killilea DW Oxidative stress and inflammation in iron-overloaded patients with beta-thalas-saemia or sickle cell disease Br J Haematol 2006 135 254 63 17010049 20. Zimmermann MB Fucharoen S Winichagoon P Iron metabolism in heterozygotes for hemoglobin E (HbE), alpha-thalassemia 1, or beta-thalassemia and in compound heterozygotes for HbE/beta-thalassemia Am J Clin Nutr 2008 88 1026 31 18842790 21. Paganini D Zimmermann MB The effects of iron fortification and supplementation on the gut microbiome and diarrhea in infants and children: A review Am J Clin Nutr 2017 106 1688S 1693S 29070552 22. Peña Rosas JP Mithra P Unnikrishnan B Fortification of rice with vitamins and minerals for addressing micronutrient malnutrition Cochrane Database of Systematic Reviews 2019 2019 CD009902 31684687 23. Thankachan P, Selvam S, Narendra AR, et al. There should always be a free lunch: the impact of COVID-19 lockdown suspension of the mid-day meal on nutriture of primary school children in Karnataka, India. BMJ Nutrition Prevention and Health. 2022;e000358. 24. Nemeth E Ganz T Anemia of inflammation Hematol Oncol Clin North Am 2014 28 671 81 25064707 25. Yu EX Addo OY Williams AM Association between anemia and household water source or sanitation in preschool children: the Biomarkers Reflecting Inflammation and Nutritional Determinants of Anemia (BRINDA) project Am J Clin Nutr 2019 112 488S 97S 32743647 26. Kulkarni B Augustine LF Pullakhandam R ‘Screen and treat for anemia reduction (STAR)’ strategy: Study protocol of a cluster randomised trial in rural Telangana, India BMJ Open 2021 11 e052238 27. Kurpad AV, Sachdev HS. Commentary: Time for precision in iron supplementation in children. Int J Epidemiol. 2022; dyac102. [Online ahead of print version].

36036186

PMC9748895

NO-CC CODE

2022-12-15 23:22:48

no

Indian Pediatr. 2022 Aug 26; 59(11):837-840

utf-8

Indian Pediatr

10.1007/s13312-022-2639-6

oa_other

==== Front Educ Inf Technol (Dordr) Educ Inf Technol (Dordr) Education and Information Technologies 1360-2357 1573-7608 Springer US New York 11520 10.1007/s10639-022-11520-8 Article Impact on social capital and learning engagement due to social media usage among the international students in the U.S. http://orcid.org/0000-0002-7736-086X Dong Jianwei jzd0077@auburn.edu Lee Sangah szl0146@auburn.edu Wang Chih-hsuan wangchi@auburn.edu Shannon David M. shanndm@auburn.edu grid.252546.2 0000 0001 2297 8753 Department of Educational Foundations, Leadership, and Technology, Auburn University, 4036Haley Center, 351 W Thach concourse, 36849 Auburn, AL USA 14 12 2022 124 8 8 2022 16 11 2022 7 12 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. International students who pursue their academic goals in United States are prone to difficulties when attempting to build social resources and adjust to the new culture. Social media is a practical means of connection due to its ease of use and accessibility. Previous research has indicated contradictory effects of social media use on academic engagement. In addition to the direct effect, this research examined social media use influences on international students’ learning engagement by mediating social capital and cultural adjustment. A total of 209 international students completed a web-based survey distributed via e-mail and social media between November 2021 and May 2022. Data were analyzed using Structural Equation Model. Results showed that only purposely using social media to collaborate with learning counterparts or materials directly improves international students’ learning engagement. Other uses of social media (e.g., expanding new resources, solidifying close relationships) have no significant direct effects. Nonetheless, they are essential to improving levels of learning engagement via the mediation of bridging capital (social resources attributed to expanding relationships) and students’ cultural adjustment in the U.S. International students’ bonding capital (social resources available through trustworthy relationships) and home cultural retention showed little direct or indirect effects on learning engagement. This study recognizes the importance of social resources and cultural adjustment for international students. Also, this study provides valuable information to educators and administrators, as there is a need to identify the underlying mechanisms to contribute feasible learning intervention approaches and alleviate negative effects for international students. Keywords Social media Social capital Learning engagement Cultural adjustment International students ==== Body pmcIn the U.S., international students comprise approximately 5% of the student population in higher education (Institute of International Education, 2021). Some challenges associated with studying abroad include students leaving their family and friends, learning in a non-native language, and adjusting to a different culture. The ongoing threat of the global COVID-19 pandemic contributed to these difficulties, making it harder for international students to maintain regular connections with instructors and other students and effectively cope with their academic challenges (Kim & Hogge, 2021; Tsai et al., 2020). Therefore, international students are experiencing unprecedented rates of social isolation, psychological stress, cultural conflict (Tsai et al., 2020), and academic difficulty (Jordan & Hartocollis, 2020; Tsai et al., 2020). International students are encouraged to increase their social resources to alleviate stress and to get acclimated to their host countries, promoting both mental health and academic success (Brunsting et al., 2018; Chai et al., 2020, 2022; Cruwys et al., 2021; Sun et al., 2021). Just as social media platforms (e.g., YouTube, Twitter, Instagram, TikTok) have become popular among many people between the ages of 18 and 29 living in the U.S., they have also become integral in the lives of college students (Blasco-Arcas et al., 2013; Cao & Tian, 2022; Chang et al., 2019; Stathopoulou et al., 2019). Due to the popularity of and easy access to social media (Yu et al., 2019), it attracts sojourners such as international students to utilize various applications to achieve multiple communication and entertainment goals (Ali-Hassan et al., 2015). Although there is an increase in researchers exploring the complicated effects of social media usage on learning engagement in higher education, there are many contradictory conclusions regarding the effects social media has on learning engagement (Ansari & Khan, 2020; Cao & Tian, 2020; Chawinga, 2017). One possible reason of the contradictory effects derives from social media’s broad range of applications. A single social media platform can support communication and relaxation, with no need to shift to another platform; this same platform may also enable users’ access to academic courses or allow them to follow experts in a particular field (Ali-Hassan et al., 2015). YouTube is one such platform as it allows users to watch videos for entertainment, learn instructional tutorials, or even access academic courses, depending on the users’ goals and context (Moghavvemi et al., 2018). Additionally, even in the context of intentionally using social media for learning, the influence of time spent on social media on academic performance is unclear. In higher education, some instructors integrate social media platforms with information communication technology (ICT) or learning management systems (LMS), effectively strengthening peer-to-peer discussion, instructor interaction, and the sharing of course-related materials (Al-Rahmi et al., 2018; Ansari & Khan, 2020). The joint efforts among students or students with instructors facilitate collaborative learning approaches (Roberts, 2005), promote learning attitudes (Kabilan et al., 2010), engage with materials and counterparts, and improve student performance (Al-Rahmi et al., 2018). However, these positive results are investigated through collaborative learning, in which social media is one of the integrated approaches to enhance discussion or content sharing (Al-Rahmi et al., 2018; Cao & Tian, 2022; Chawinga, 2017). On the other hand, social media is often associated with negative effects. College students who excessively and unconsciously consume social media reduce their hours of engagement with learning materials. Moreover, social media applications are misused in the classroom (Tindell & Bohlander, 2012) and other learning contexts distracts attention span (Ansari & Khan, 2020). The inconsistent results of social media use on learning engagement among general college students invoke further studies. These studies should investigate how international students utilize social media to achieve academic success. International students’ resources for expanding relationships are substantially limited compared to domestic students. Social media provides easy access and various choices that allow international students to set up global and local communication applications on their devices. In addition to expanding new circles in the U.S., international students can also use their native language to communicate with families and friends at home. As expounded in the following literature review, international students leverage social media to manage their networking on and off campus, implicating their willingness and competence in coping with some of the challenges they face in the U.S. The findings of this study will potentially contribute theoretical and practical implications to improve international students’ adjustment and academic success. Theoretically, this study examines social media’s direct and indirect impact on international students’ learning engagement. The conceptual model used (Fig. 1) brings two additional factors, cultural adjustment, and social capital, as mediators to demonstrate how these mediators complicate the indirect effects. Practically, findings from this study should provide valuable information to educators and administrators, as there is a need to identify the underlying mechanisms that contribute to the feasible learning intervention approaches to alleviate negative effects for international students. Literature review The literature reviewed for the current study is related to (1) college students’ learning engagement and its complexities, (2) the way social media usages inconsistently affect college students’ learning engagement in different contexts, and (3) the way social capital and culture adjustment influence on learning engagement. Following each topic, we discussed our hypothesis model and depicted the direct and mediated conceptual model, as shown in Fig. 1. Fig. 1 Conceptual Model College student learning engagement Student engagement is conceptualized as the effort of students’ participation in academic work, involvement in the learning process, and achievement of expected outcomes (Handelsman et al., 2005). Scholars have widely agreed that student learning engagement is a comprehensive construct that comprises behavioral, affective, and cognitive engagement dimensions (Fredricks et al., 2004). The direct observation of behavioral engagement refers to students’ classroom tasks or coursework involvement, which can be easily identified through indicators/markers (Finn & Zimmer, 2012). Affective engagement refers to emotions, either contentment or stress, generated by students’ interactions with school/course work or people encountered (Pekrun & Linnenbrink-Garcia, 2014). Cognitive engagement refers to students determining motivational goals, mastery orientation, and imposing self-regulation and learning strategies (Cleary & Zimmerman, 2012). The absence of student engagement is seen in burnout, withdrawal, and a lack of motivation, all of which lead to academic failure (Finn & Zimmer, 2012; Skinner, 2016). According to Skinner (2016), two distinct facilitators, self and context, help students stay engaged. The self characterizes internal features, such as self-perception, personality traits, or a sense of belonging, to motivate intentional actions (Baumeister & Leary, 1995; Goodenow, 1993); context broadly refers to the external or social factors and influences from the student’s environment (Tyson & Hill, 2009), presented as relatedness (e.g., family, schools, peers) so that the individual feels supported. International students’ learning engagement, both in self and context, is constantly interwoven with challenges of coping with language competence in specific (Moon et al., 2020), or new environments, which are cross-cultural in general (Al-Oraibi et al., 2022; Chai et al., 2020; Humphrey & Forbes-Mewett, 2021). Moon et al. (2020) explored the challenges in lectures, class discussions, and writing assignments among international students from Korea and China. The result shows that besides language barriers, relationships with instructors and understanding of the assessment standards are essential to differentiate their study experiences. As the explication of the context factor, the feeling of being supported by family, institution, and community positively contributes to academic engagement and achievement. International students being remote to their strong support at home indues the self-perceived social losses (Humphrey & Forbes-Mewett, 2021). Chai et al. (2022) conducted an international students’ study in Japan, suggesting the perceived lack of personal resources leads to mental and physical drain. In contrast, sufficient social resources alleviate negative emotions, advancing academic engagement. Moreover, a sense of inclusion instead of being viewed as “others” encourages international students to participate in peer group discussions and interact with instructors (Blackmore et al., 2021; Chai et al., 2020; Tran & Pham, 2016). The effects of social media on college student learning engagement International students, like many college students in the U.S., grew up in a technologically rich world (Kunka, 2020), and they habitually use social media for various purposes that range from academic tasks to personal communication. Social media is a platform that incorporates international students’ resources to build self and context to engage with students, instructors, and organizations in the U.S. while connecting to family and friends in their home countries (Cao & Tian, 2022; Chang et al., 2019). However, there is rising sentiment against excessive yet passive social media consumption as it is detrimental to academic engagement. College students constantly access social media while studying (Flanigan & Babchuk, 2015). Tindell and Bohlander (2012) found that 92% of undergraduate students who participated in the survey self-reported using their mobile phones to access information unrelated to course materials and texting messages in class. Many college students lack the ability to avoid being distracted by social media. As Flanigan and Kiewra (2018) found, individuals may skillfully enjoy personal communication, gaming, or other entertainment, but they cannot strategically leverage these capabilities in an educational environment. Some scholars have explored ways to facilitate learning while factoring in social media’s notable accessibility and popularity among college students. Drawing from an affordance theory perspective (Gibson, 1986), some researchers describe the use of social media in education as the interaction between actors (users) and artifacts (social media applications) to satisfy needs or achieve goals (Hutchby, 2001; Pozzi, 2014). As a single social media application engages a wide range of users and vice versa, individuals fulfill various tasks without shifting to another social media application. Like YouTube, Twitter, a mini-blog platform, allows students to discuss course topics, complete assignments, and share materials. Chawinga (2017) designed an experimental course at a public university in Malawi that required students to interact with each other and share learning materials via Twitter and blogs using a predefined hashtag. Using Twitter advocated ‘learner-centred’ [sic] pedagogical approach and facilitated cognitive learning engagement with peers and course materials. Instead of singling out an individual application, we treat social media as an assembly with a range of choices (Ali-Hassan et al., 2015; Ansari & Khan, 2020; Cao & Tian, 2022) and focus on the individual purposes and tasks it affords. International students set up various social media choices ranging from globally popular applications to regionally based ones on their mobile devices (Cao & Meng, 2020; Tu, 2018; Yu et al., 2019) to accomplish their tasks. From the technological lens of affordance theory, Ali-Hassan et al. (2015) proposed a three-dimensional (i.e., social, cognitive, hedonic) used to determine the direct and indirect influence of social media on corporate job performance. In the current research, the social dimension is further divided into two subdimensions, to expand future networks (new) and interact with pre-existing relationships (close) to align with international students’ social goals. The following four hypotheses were developed based on the four social media use dimensions (Fig. 2). H1a: International students using social media to expand new resources positively affect learning engagement. H1b: International students using social media to maintain close relationships with family and friends positively affect learning engagement. H1c: International students using social media in a cognitively oriented way to build academic collaborations positively affect learning engagement. H1d: International students using social media to consume hedonic content negatively affect learning engagement. The mediator effect of cultural adjustment and social capital Cultural adjustment For international students, acclimatization to their new environments (e.g., host culture, new institution, new community) helps them cope with feelings of loneliness (Russell et al., 2010; Sawir et al., 2008) or strengthen their sense of belonging (Glass et al., 2015; Tu, 2018; Van Horne et al., 2018). Kim (1988) proposed the Integrative Theory of Cross-Cultural Adaptation that structures the process of engaging in social communication and identity transformation among sojourners, which refers to the individual who grew up in one primary culture and later relocated to the host culture. This multidimensional theory describes the aspects that influence cross-cultural adaptation, such as personal predisposition (e.g., ethnic proximity), host environment (e.g., host conformity pressure), and intercultural transformation (e.g., intercultural identity). In research aimed at international students in the U.S., Kim and her colleagues found that Asian students show lower host communication competence than European students, resulting from insufficient knowledge of host culture and less involvement in intercultural activities (Kim, 2001; Kim et al., 2016). Moreover, sense of belonging is also a factor that influence international students’ adaption in the U.S. According to Van Horne et al. (2018), international students consistently reported lower levels of satisfaction and a weaker sense of belonging than their American counterparts across nine universities in the U.S. On the contrary, those international students reporting a stronger sense of belonging had more interactions with students from other cultures, although the frequency of this interaction varied largely (Van Horne et al. 2018). In this respect, many researchers have reached similar conclusions: international students’ cultural adjustment correlated to their level of engagement with domestic students (Glass & Westmont, 2014), participation in diversity-related cocurricular activities (Glass & Westmont, 2014), and interactions with instructors (Glass et al., 2015; Stathopoulou et al., 2019). In addition to adjusting to the host culture, international students’ retention of their home culture is also important to their cultural integration (Berry et al., 2006; Demes & Geeraert, 2014). Individuals ranking high in both host-culture adjustment and home-culture retention, defined as cultural integration, demonstrated better cultural adjustment in foreign countries (Demes & Geeraert, 2014). Using a sample of 12 international students in the U.S., Tu (2018) investigated their preferred language and tasks using different social media platforms. Students use of their native language to maintain connections with family at home or friends from similar cultures and enjoying entertainment can lower their stress levels (Tu, 2018). Nevertheless, using English on social media is often associated with expanding one’s social circle, engaging in academic tasks, and adjusting to the U.S. This finding can be explained by the affordance perspective (Gibson, 1986) that international students take actions based on social media features that are available to everyone. They can accomplish different tasks driven by their unique emotional needs and academic goals. However, there is limited knowledge of the international students’ uses of social media and its association with cultural integration and learning engagement. We hypothesize that the host or home cultural factors mediate different social media uses influencing their learning engagement (Fig. 2). H2a: International students using social media to expand new social resources will indirectly and positively influence learning engagement by mediating host cultural adjustment. H2b: International students using social media to maintain close relationships with family and friends will indirectly and positively influence learning engagement by mediating home cultural adjustment. H2c: International students using social media in a cognitively oriented way to build academic collaborations will indirectly and positively influence learning engagement by mediating host cultural adjustment. H2d: International students using social media to consume hedonic content will indirectly and positively influence learning engagement by mediating home cultural adjustment. Social capital theory Social capital is conceptualized as the outcome of interpersonal networks and the benefits that are empowered by the relationships integrated online and offline (Coleman, 1988; Putnam, 2000). Bridging and bonding are two distinct components that collectively gauge the amount of social capital. While bridging capital is associated with weak ties that expand connection and spread information to a broader network, bonding offers emotional support and security via the strong ties of relationships (Lin et al., 2012; Putnam, 2000; Tu, 2018). International students’ social capital is essential to their cultural adjustment and the ways they cope with academic challenges. Social capital enables international students to exchange and acquire new information, provides them with support, and strengthens their sense of belonging (Glass & Westmont, 2014; Phua & Jin, 2011; Williams, 2006). Additionally, social capital enhances international students’ engagement at the colleges and universities they attend (Mbawuni & Nimako, 2015) and in the communities in which they live (Glass & Gesing, 2018). Once international students are in a drastically different cultural environment, they must deal with life difficulties and engage with academic work simultaneously; the psychological and sociological stress caused by insufficient resources impedes their cultural adjustment and academic engagement (Cho & Yu, 2015; Chai et al., 2022). International students accumulate social capital by building and maintaining relationships with others in the host countries and in their home countries, respectively (Cao & Meng, 2020; Li & Chen, 2014). According to Glass and Gesing (2018), international students’ relationships fit into the four primary networks: academic programs, campus organizations, residential communities, and family and friends. The first three networks are new to the student and must be established in the U.S., and the latter are usually pre-existing. Social media allows integration of the four primary networks, advancing students’ virtual membership into networks that are beyond their physical reach (Phua & Jin, 2011; Wellman et al., 2001). Repeated social interactions, such as attending classes, joining study groups, and participating in student organizations and residential gatherings, strengthen international students’ sense of belonging (Gomes et al., 2015). Moreover, international students’ participation in such activities increases their chances to interact with students from other cultures as well as people from the local community, offering these students more extensive, dynamic social networks and leading to some students having more social capital than others (Glass & Gesing, 2018). Some researchers found that the process by which international students build bridging and bonding capital is distinct (Cao & Meng, 2020; Lin et al., 2012; Phua & Jin, 2011). Bridging capital results from the development of weak ties (Glass & Gesing, 2018) with new relationships on- or off-campus, access to valuable information, or membership in a professional organization. Meanwhile, bonding capital is the result of strong ties gained through secure and trustworthy pre-existing relationships with family and friends, often from an international student’s home country (Lin et al., 2012). Social capital measures the outcomes that bridging and bonding capital are accumulated in both online and offline contexts (Cao & Meng, 2020; Lin et al., 2012; Phua & Jin, 2011). Online networking increases both bridging and bonding capital while international students stay in the U.S. However, previous studies of social capital suggest that the accumulation of bridging and bonding are imbalanced. The time that international students spend online is associated with bridging capital more than bonding capital. More times of social media use increases bridging capital, but it does not have an equivalent effect on bonding (Lin et al., 2012; Phua & Jin, 2011; Stefanone et al., 2011). The more frequently international students interact with American students, the more their bridging capital is detected, arising from better coping and adjustment to college (Lin et al., 2012). In the current research, we hypothesize that social capital is a mediator between social media uses and students’ learning engagement. In Ali-Hassan et al. (2015)’s research, social capital demonstrated a “nuanced facet” (p. 78) that social media use positively increases job performance only if the specific use increases social capital. In addition to earlier direct effect hypotheses, we proposed that international students’ social media use will indirectly affect their learning engagement through social capital (Fig. 2). H3a: International students using social media to expand new social resources will indirectly and positively influence learning engagement by mediating bridging capital. H3b: International students using social media to maintain close relationships with family and friends will indirectly and positively influence learning engagement by mediating bridging capital. H3c: International students using social media to maintain close relationships with family and friends will indirectly and positively influence learning engagement by mediating bonding capital. H3d: International students using social in a cognitively oriented way to build academic collaborations will indirectly and positively influence learning engagement by mediating bridging capital. Lastly, we explored the paths, including cultural adjustment and social capital, to detect the chain mediation effect (Fig. 2). H4a: International students using social media to expand new social resources will indirectly and positively influence learning engagement by mediating host cultural adjustment and bridging capital. H4b: International students using social media to maintain close relationships with family and friends will indirectly and positively influence learning engagement by mediating host cultural adjustment and bridging capital. Fig. 2 Hypothesis Model Note: A hypothesis model with direct and indirect paths is shown in the figure. There are four types of social media use (e.g., new, close, cognitive, hedonic), two types of cultural adjustment (e.g., host, home), and two types of social capital (e.g., bridging, bonding), direct and indirect influence the latent dependent variable learning engagement. Each number in the hypothesis model is explained. Method Methodology A quantitative survey research method was utilized to investigate a hypothesized model (Fig. 2) that drawn on the relationships between the different uses of social media, bridging and bonding capital, students’ cultural adjustment to their host and home countries, and learning engagement among international students at a university in the U.S. (Dillman, 2014). This study received Institutional Review Board (IRB) approval. Participants The sampling procedure followed a purposeful sampling strategy that targeted eligible international students in higher education in the U.S. A total of 209 participants completed the survey, and all participants were international students enrolled at different colleges throughout the U.S. Participants’ ages ranged from 18 years to 49 years, with a mean average of 28.8 years (SD = 6.5). Among the 209 participants, 42.3% were women (n = 88), and 56.7% were men (n = 118); the remaining 3 students indicated that they “prefer not to say.” Of the students who participated, 46.9% (n = 98) were Asian while 29.7% (n = 62) of the students identified themselves as White, 10% (n = 21) were Black. 8.1% (n = 17) claimed their ethnicity as Hispanic, and 4 indicated that they were “Other.” Most of the students (77.5%, n = 162) reported that they were STEM (science, technology, engineering, and mathematics) majors. Of the 209 participants, 38.2% (n = 80) were enrolled in undergraduate degree programs, 32.1% (n = 67) were pursuing master’s degrees, and 29.7% (n = 62) were enrolled in doctoral programs. Instruments Learning Engagement. Handelsman et al. (2005) developed a college student course engagement questionnaire (SCEQ) measurement. In the current study, we included 16 of 23 items with the factor loading higher than 0.5, measure the individual level of coursework engagement: skills engagement (n = 8, sample item: “Making sure to study on a regular basis”); emotional engagement (n = 3, sample item: “Finding ways to make the course interesting to me”); and participation engagement (n = 5, sample item: “Participating actively in small-group discussions”). Participants were asked to rate each item on a scale of 1 to 5 to indicate the extent to which the learning behavior was applicable to them. On this scale, 1 = not at all characteristic of me to 5 = very characteristic of me. The higher score indicates higher learning engagement. The internal consistency Cronbach’s α for the skills, emotional, and participation engagement subscales was 0.92, 0.81, 0.79, respectively. Social Media Usage. For this study, we adapted 18 items from Ali-Hassan et al.’s (2015) questionnaire measuring social media use and defined 4 subscales in the current research. The subscales featured the use of social media for (1) expanding new social relationships, (2) solidifying existing close relationships, (3) cognitive learning, and (4) hedonic usage for entertainment. A 5-point Likert-type scale was used to describe the frequency of social media use for a particular purpose, with 1 = Never to 5 = Always. Item samples, such as “I use social media to discover people with interests similar to mine” were designed for social media for new networks (n = 5); “I use social media to create content in collaboration with fellow researchers” for social media cognitive (n = 6); and “I enjoy my break during research work,” for social media hedonic (n = 3). An additional item was added to the social media close scale (n = 3) to distinguish two items: one measured close relation in the U.S., and the other specified the close relations from their home countries. Wording was altered for several items, and two items were added to the cognitive use subscale to describe higher education’s academic and research tasks in the U.S. (sample item: I communicate with people who are in leadership roles in my academic/professional field). The higher score indicates more frequent the particular social media use. The internal consistency Cronbach’s α for the four subscales was 0.80, 0.68, 0.90, 0.72, respectively. Social Capital. Williams (2006) developed and validated the online/offline bonding and bridging capital measurement. Some scholars have suggested that the social capital obtained from both online and offline contexts associated with social media use are inseparable because social media integrates online and offline relations (Cao & Meng, 2020). In this study, the source of the two social capitals were derived from online usage (social media), which integrates both online and offline networks. Moreover, general interest in this study was in bonding and bridging capital; therefore, the wording “online/offline” was not specified for each survey question. Participants responded to 10 items each to rate their levels of agreement on a 5-point Likert-type scale by 1 (Strongly disagree) to 5 (Strongly agree) to indicate how often each event occurs to measure bonding and bridging capital. An example of an item on the bonding scale is, “There is someone I can turn to for advice about making very important decisions,” while an example from the bridging scale is, “Interacting with people makes me feel connected to the bigger picture.” A higher mean score indicates a higher social capital value. The internal consistency Cronbach’s α value was at 0.91 and at 0.80 for bridging and bonding, respectively. Cultural Adjustment. The Brief Acculturation Orientation Scale (BAOS), a two-dimensional construct developed by Demes and Geeraert (2014), was used to measure the value of an individual in a host country maintaining the cultural heritage from his or her home country while integrating into the dominant host culture. There were four items for each type of cultural adjustment (i.e., host and home) as aligned with Berry et al. (2006). Participants were asked to indicate their levels of agreement or disagreement on each of the 8 items, with 1 = Strongly disagree to 5 = Strongly agree. On the home cultural adjustment subscale, participants answered questions such as “It is important for me to have friends in my home country.” Likewise, on the host cultural adjustment subscale, a sample question includes, “It is important for me to have friends in the U.S.” The higher score indicates higher cultural adjustment. The Cronbach’s α value of internal consistency was at 0.80 for the host culture and 0.70 for the home culture. Procedure An anonymous online survey hosted by the Qualtrics platform collected data that measured variables and participants’ demographic information from November 2021 to May 2022. International students enrolled at a university in the southeastern region of the U.S. were invited to participate in the study and directed to the online survey via an e-mail list sent from the Office of International at a large Southeastern university. An invitation link was also posted by the social media account of the Journal of International Students. Participants were encouraged to forward the survey link to other international students currently enrolled at higher education institutions in the U.S. Once participants clicked the link to access the survey, they were directed to a page allowing them to grant informed consent, and they were required to click “yes” in response to a screening question to identify whether they were international students. A follow-up reminder e-mail was also sent to the students on this e-mail list, and the social media invitation to participate was reposted after the initial invitation. Data analysis Data were analyzed using the SPSS 28 and Mplus version 8 (Muthen & Muthen, 2017). The SPSS 28 program was used to provide descriptive statistics, correlations, and the reliability of each subscale to summarize the characteristics. The Mplus version 8 was used to examine whether or not the hypothesis model (Fig. 2) for four different types of social media use by international students directly or indirectly influenced their learning engagement by a structural equation model (SEM) path analysis (Hair et al., 2006). To examine the hypothesized model, we applied the equal-weighted composite score variables, the four types of social media use (exogenous variables), the two types of social capital, and the two types of cultural adjustment (endogenous variables), to influence the learning engagement (the latent dependent variable) simultaneously. Deng and Yuan (2022) compared various models using weighted and unweighted path analysis, path analysis with latent variables, and covariance-based SEM. Results suggest that path and partial path analysis are reliable for detecting models. Investigating the social capital or cultural adjustment’s mediator role was essential in elucidating how the social and cultural factors influence the effect of social media usage on learning engagement. The following model fit indices, such as the Chi-square goodness of fit test, comparative fit index (CFI), Tucker-Lewis Index (TLI), and root mean square error of approximation (RMSEA), were used to evaluate the hypothesis model. According to Hair et al. (2006), the general cut-off value for the indices for CFI and TLI should be above 0.95, and the value of RMSEA should be no greater than 0.1. A significant model evaluation result indicates a poor model fit. Therefore, the inadequate model was improved by trimming insignificant paths (Hox & Bechger, 1998). Results Descriptive statistics, the internal consistency reliability of Cronbach’s α, and bivariate correlation coefficients (Berk et al., 1996) of the 11 subscales were summarized in Table 1. Cronbach’s α internal consistency and number of items for those 11 subscales were as follows: skills 0.92 (8 items), emotional 0.81 (3 items), and participation 0.79 (8 items), new 0.80 (5 items), close 0.68 (3 items), cognitive 0.90 (6 items), hedonic 0.72 (3 items), bridging (10 items), bonding 0.80 (10 items), host 0.80 (4 items), home 0.70 (4 items). All values of Cronbach’s α demonstrated adequate reliability in this sample except the close subscale was lower than 0.70 (Berk et al., 1996). The subscales were positively significantly correlated to each other at a moderate to large degree (r = .21 ~ .74, ps < 0.01). All four types of social media use have moderate correlation coefficients with three engagement subscales (0.21 ~ 0.48, ps < 0.01). Both bridging and bonding capital variables showed moderate correlations (r = .43 ~ .57, ps < 0.01). Home cultural adjustment presented lower bivariate correlations (r = .24 ~ .36, ps < 0.01) than host cultural adjustment (r = .39 ~ .43, ps < 0.01). Table 1 Cronbach’s α, Means, SD, and Correlation Variables (N = 209) Cronbach’s α M SD 1 2 3 4 5 6 7 8 9 10 11 1 New 0.80 3.49 0.75 -- 2 Close 0.68 3.56 0.92 0.58** -- 3 Cognitive 0.90 3.31 0.95 0.72** 0.62** -- 4 Hedonic 0.72 3.72 0.90 0.49** 0.54** 0.50** -- 5 Home 0.70 3.82 0.73 0.30** 0.42** 0.45** 0.38** -- 6 Host 0.80 3.93 0.78 0.46** 0.39** 0.43** 0.41** 0.37** -- 7 Bonding 0.80 3.68 0.68 0.47** 0.44** 0.52** 0.48** 0.40** 0.51** -- 8 Bridging 0.91 3.96 0.72 0.42** 0.50** 0.41** 0.44** 0.35** 0.53** 0.63** -- 9 Skills 0.92 3.82 0.74 0.40** 0.37** 0.39** 0.40** 0.24** 0.42** 0.51** 0.57** -- 10 Emotional 0.81 3.73 0.87 0.31** 0.36** 0.45** 0.21** 0.36** 0.39** 0.43** 0.48** 0.74** -- 11 Participation 0.79 3.61 0.80 0.42** 0.35** 0.50** 0.33** 0.26** 0.43** 0.57** 0.50** 0.69** 0.64** -- Note. **p < .01 As shown in Table 2, the test of hypothesized model fit indicated an inadequate fit with observed data. The goodness-of-fit test was statistically significant, χ2 (28) = 202.98, p < .001, χ2/df = 7.25, CFI = 0.795, TLI = 0.642, RMSEA = 0.173. Moreover, most direct and indirect hypotheses paths did not reach statistical significance at the 0.05 critical level. After trimming insignificant paths and re-evaluating model fit sequentially, the final model reached the adequate model fit with the Chi-square value improved up to 156 (from χ2 (28) = 202.98, p < .001 to χ2 (17) = 47.11, p = .001), χ2/df = 2.77, CFI = 0.950, TLI = 0.926, RMSEA = 0.092. The final model path and estimation as shown in Fig. 3; Table 2. Table 2 Direct and Indirect Effects in Hypothesized and Final Model Hypothesized Model Final Model Std. Est. S.E. p Std. Est. S.E. p Engagement ← Skills 0.864 0.042 < 0.001*** 0.878 0.036 < 0.001*** ← Emotional 0.804 0.048 < 0.001*** 0.819 0.043 < 0.001*** ← Participation 0.784 0.04 < 0.001*** 0.786 0.038 < 0.001*** H1a not supported Engagement ← New –0.004 0.109 0.971 -- -- -- H1b not supported Engagement ← Close –0.013 0.102 0.901 -- -- -- H1c supported Engagement ← Cognitive 0.257 0.116 0.027* 0.316 0.082 < 0.001*** H1d not supported Engagement ← Hedonic –0.033 0.106 0.757 -- -- -- H2a not supported Engagement ← Host ← New 0.028 0.029 0.334 Engagement ← Host 0.114 0.101 0.257 -- -- -- Host ← New 0.301 0.098 0.002** 0.456 0.050 < 0.001*** H2b not supported Engagement ← Home ← Close –0.002 0.019 0.931 Engagement ← Home –0.008 0.081 0.926 -- -- -- Home ← Close 0.303 0.083 < 0.001*** -- -- -- H2c not supported Engagement ← Host ← Cognitive 0.016 0.017 0.345 Host ← Cognitive 0.217 0.106 0.041* -- -- -- H2d not supported Engagement ← Home ← Hedo –0.002 0.020 0.931 Home ← Hedo 0.214 0.082 0.009** -- -- -- H3a not supported Engagement ← Bridging← New 0.026 0.033 0.428 -- -- -- Engagement ← Bridging 0.383 0.105 < 0.001*** 0.501 0.075 < 0.001*** Bridging ← New 0.068 0.085 0.422 -- -- -- H3b supported Engagement ← Bridging ← Close 0.081 0.030 0.007** 0.176 0.036 < 0.001*** Bridging← Close 0.316 0.087 < 0.001*** 0.352 0.068 < 0.001*** H3c not supported Engagement ← Bonding ← Close 0.048 0.032 0.136 Engagement ← Bonding 0.219 0.140 0.117 -- -- -- Bonding ← Close 0.327 0.065 < 0.001*** -- -- -- H3d not supported Engagement ← Bridging ← Cognitive 0.001 0.039 0.991 Bridging ← Cognitive 0.001 0.102 0.992 -- -- -- H4a supported Engagement ← Bridging ← Host ← New 0.036 0.016 0.021* 0.092 0.026 < 0.001*** H4b not supported Engagement ← Bonding ← Home ← Close 0.012 0.008 0.134 -- -- -- Note. *p < .05; **p < .01; ***p < .001 The standardized factor loadings of latent variable from skills, emotional, and participation were high (0.86, 0.80, and 0.78, ps < 0.001) shown in Table 2. The endogenous variables bridging capital accounted for the largest amount of variance with R2 = 0.372. The next was bonding capital, host and home cultural adjustment, with R2 = 0.246, R2 = 0.231, and R2 = 0.208, respectively. In the final model (Fig. 3; Table 2), only using social media for cognitive learning had a direct impact on the latent variable engagement (H1c) which showed statistical and positive significance (β = 0.316, p < .001). Using social media to maintain close relationships (H1b) did not predict learning engagement, however, after adding the mediating effect of bridging capital (H3b), the results showed a positive influence on learning engagement (β = 0.176, p < .001). Bridging capital significantly and positively influenced learning engagement. In contrast, bonding capital did not demonstrate the same effect. In hypothesis H4a, using social media for new resources influenced host cultural adjustment, then was mediated by bridging capital to impact engagement. The mediation of host and bridging was positively and statistically significant, but showed a small effect (β = 0.092, p < .001). None of H2a-d was significant. Although the social media uses on cultural adjustment effect were partially held, the two cultural adjustment subscales had no direct effect on learning engagement. The hedonic type of social media use neither directly (H1d) nor indirectly (H2d) affected international students’ learning engagement (β = − 0.033, p = .76). We removed this variable from the final model. Although close relationships had a significant effect on bonding, there was no significant indirect path (H3c, H4b) on learning engagement. Home cultural retainment (β = − 0.008, p = .926) and bonding capital (β = 0.219, p = .117) were two endogenous variables that were removed from the final model due to no significant effect on learning engagement. Fig. 3 Final Model Note: A final model with direct and indirect paths is shown in the figure. *p < .05; **p < .01; ***p < .001. Discussion The SEM path analysis results in the final model indicated that three out of the four uses social media for international students, direct or indirect, improved their learning engagement, except for hedonic use. In terms of the direct effect on engagement, cognitive social media use produced statistical significance (β = 0.316, p < .001), which was consistent with our hypothesis (H1c) and previous research. According to Ansari and Khan (2020), college students in India have diversified social media communication groups and applied their resources to collaborative learning. Among the three cognitive learning activities, interaction with instructors contributed a higher coefficient degree (β = 0.450, p < .001) than knowledge sharing (β = 0.247, p < .001) and peer discussion (β = 0.210, p < .001) on students’ learning engagement. In Ansari and Khan’s (2020) research, the engagement variable data were collected using four statements adapted from Al-Rahmi and his colleagues’ research (2018) regarding factors affecting learning performance through social media in Malaysia. Instead of using a single summation of all items, we applied latent variables that constitute three subconstructs (i.e., skills, emotional, participation) that we adapted from the SCEQ (Handelsman et al., 2005). The complexity of this learning engagement construct has been discussed in the literature review section. Grounded in Skinner’s (2016) concept, the self and context factors are two distinct facilitators affecting the students’ strategies to stay engaged, and these two factors may impact the three subcomponents (i.e., skills, emotional, participation) differently. For example, when college students cognitively apply skillful strategies, but suffer emotionally from the anxiety of being unable to compete with others, their learning engagement summation decreases due to not being able to optimize their participation in learning activities (Schwinger & Stiensmeier-Pelster, 2012). The factor loadings in the hypothesized and final model varied but consistently indicated that skills engagement had the highest factor loading, the second was emotional, followed by participation (Table 2). Studying the mediation effect of bridging capital and host cultural adjustment unraveled two indirect paths leading to the significant improvement of international students’ learning engagement. Bridging was a pivotal factor that appeared in two paths to mediate the effect of two types of social media usage on engagement, respectively. Bridging was mid-high and positively correlated with other exogenous and endogenous variables (Table 1). In the final model estimation (Table 2), the use of social media for close relationships positively predicted bridging capital and improved engagement (H3b). In previous literature, international students’ pre-existing relationships are often associated with strong ties (Lin et al., 2012; Phua & Jin, 2011; Stefanone et al., 2011; Tsai et al., 2020). Little research has discussed how international students’ use of social media to connect with family and friends could increase bridging capital. Cruwys and colleagues (2021) suggested that international students build connections with multiple groups of identities to ensure that resources are available when they encounter unexpected challenges. Pre-existing group memberships increase students’ possibilities in expanding new resources. Our research results indicate that using social media to strengthen communication in close relationships enables international students to simultaneously build weak and strong ties in the U.S. In this respect, we emphasize the significance of using social media to solidify close relationships to increase international students’ bridging and bonding capital and indirectly enhance their learning engagement. With bridging capital as a mediator, the exogenous variable of new networks did not present a significant effect on learning engagement. While the insignificant indirect effect contradicted our hypothesis, the bridging capital’s mediation effect was advanced by adding the host cultural adjustment variable. In our final model, the significant mediation path (H4a) indicated that international students using social media for new resources positively impacted learning engagement through the mediation of their host cultural adjustment in the U.S. and bridging capital simultaneously (Fig. 3). The mediation of host cultural adjustment and bridging capital simultaneously explained the mechanisms that international students expanding their new social networks had a better cultural adjustment in the U.S. which also resulted in higher bridging capital. Our findings are supported by previous research about the benefits of better cultural adjustment. According to Intercultural Adaptation Model (IAM) (Cai & Rodríguez, 1997), individuals encountering cross-cultural communication have varied tendencies to adjust themselves to facilitate understanding or avoid communication. IAM suggests that sojourners’ adaptive effort depends on their previous intercultural experience. In this vein, international students’ use of social media to build new interactions highly relies on their host cultural adjustment, which consequently influences their bridging capital and learning engagement. Individuals with higher levels of bridging capital are usually more open-minded to adjusting to a new environment and more comfortable when their perceptions are challenged (Cao & Meng, 2020; Williams, 2006). Glass and Westmont (2014) proposed that integrated acculturation is a resilience factor and suggested that cross-culture interactions with domestic students and local communities enhance international students’ sense of belonging and adjustment to college in the host culture context. Cross-culture communication buffers negative feelings and provides students with a sense of secure relationships when encountering discrimination, indirectly influencing academic success. Therefore, international students must expand their resources, e.g., make new friends with interests similar to theirs and participate in community activities to build their bridging capital both on and off the campus. Consequently, communication competency, combined with increased cross-cultural sophistication, can foster a lasting positive effect on students’ academic success. Bonding capital is also meaningful for international students to maintain close and assured relationships, but it has no mediation effect in hypothesis testing. Although bonding presented moderate and significant associations with all four uses of social media and the home cultural variable, it failed to predict international students’ learning engagement. Moreover, bridging and bonding capital were moderately to largely correlated with each other (r = .63, p < .01). According to Cao and Meng (2020), international students from China with more contacts in intercultural networks presented higher bridging and higher bonding capital. Our research suggests that international students’ bridging capital plays a more prominent role in improving learning engagement than bonding for international students using social media in the U.S. Therefore, bonding capital and home cultural adjustment are removed from the final model due to little impact on learning engagement. The exogeneous variable related to hedonic use of social media was not included in the final model either. Hedonic use did not significantly lower levels of learning engagement, which contradicted our hypothesis. Instead, hedonic use improved home cultural adjustment, which was partially aligned with the hypothesis. Drawing from previous findings that better home cultural adjustment contributed to international students having better mental health (Chai et al., 2020; 2022; Cruwys et al., 2021; Sun et al., 2021), future researchers can further explore the optimal utility of entertainment to as a form of stress relief and as a way to cope with challenges without significantly interfering learning. Conclusion Drawing from affordance theory, instead of focusing on the positive or negative effects of artifacts (social media platforms) on learning, we shift the attention to the actors (international students) and advocate for the strategic and purposeful use of social media. The aim of this research is to help international students and educators develop strategies and interventions to manage social media tasks and goals to improve student learning. We suggest consideration of both social and cultural factors when studying international students to uncover previous contradicted results by examining indirect effects. Bridging capital is a significant mediator that enables two uses of social media and host cultural adjustment to indirectly enhance learning engagement. Social distancing policies originating during the COVID-19 pandemic limited students’ attendance in activities that build bridging capital. Compared to domestic students, international students have fewer opportunities and channels to reach out. When consider international students’ academic activities, the influence of their social needs and cultural adjustment should not be overlooked. Higher bridging capital and host cultural adjustment to the U.S. resulted in higher levels of academic engagement. More specifically, international students who interact with American peers, faculty, and local residents by participating in community activities or other opportunities have better integration into U.S. society and have more bridging capital. The strengths that facilitate cross-culture communication require contribution from multiple entities. In addition to individual efforts, college administrators should provide more opportunities for international and domestic students to engage in cross-cultural interactions and conversations. For example, an “International Buddy Program” was organized at one university in which international students were paired with domestic students based on their registered interests to encourage cultural diversity and involvement. After being paired, the students meet each other in person. This same institution has provided international students with opportunities to use online conversation platforms to interact with domestic students and improve their communication competency. Through personalized, flexible online learning sessions, international students can improve their language proficiency by conversing in English or sharing experiences about their respective cultures. Activities like these should be organized and funded by universities. Limitations There were several limitations to the current study. First, although the participants were international students with a variety in ethnicity, nationality and race, this study did not divide them based on those differences. The small sample size (n = 209) encouraged treating the participants as a homogenous group. In future research, splitting the group based on these differences may find larger effects in result. Second, 77.5% of students reported that they were STEM (science, technology, engineering, and mathematics) majors, and a disproportionate number of responses were from graduate and doctoral students (61.7%), compared to 38.2% of undergraduate students. This sample distribution may create a response bias that hinders a clearer understanding of the broader population of international undergraduate students. Acknowledgements Not applicable. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Data Availability The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request. Statements and declarations Competing interests The authors declare that they have no financial or personal relationships which may have inappropriately influenced him in writing this article. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Al-Rahmi WM Alias N Othman MS Marin VI Tur G A model of factors affecting learning performance through the use of social media in malaysian higher education Computers & Education 2018 121 59 72 10.1016/j.compedu.2018.02.010 Ali-Hassan H Nevo D Wade M Linking dimensions of social media use to job performance: the role of social capital The Journal of Strategic Information Systems 2015 24 2 65 89 10.1016/j.jsis.2015.03.001 Al-Oraibi A Fothergill L Yildirim M Knight H Carlisle S O’Connor M Blake H Exploring the psychological impacts of COVID-19 social restrictions on international university students: a qualitative study International journal of environmental research and public health 2022 19 13 7631 10.3390/ijerph19137631 35805287 Ansari JAN Khan NA Exploring the role of social media in collaborative learning the new domain of learning Smart Learning Environments 2020 7 1 1 16 10.1186/s40561-020-00118-7 Baumeister, R. F., & Leary, M. R. (1995). The need to belong. Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117(3),497–529. 10.1037/0033-2909.117.3.497 Berk RA Freedman D Pisani R Purves R Statistics Contemporary Sociology 1996 25 4 442 10.2307/2077035 Berry JW Phinney J Sam D Vedder P Immigrant youth in cultural transition: Acculturation, identity and adaptation across national context 2006 Mahwah, NJ Lawrence Erlbaum Blackmore J Tran L Hoang T Chou-Lee M McCandless T Mahoney C Beavis C Rowan L Hurem Affinity spaces and the situatedness of intercultural relations between domestic and international students in two australian schools Educational Review 2021 10.1080/00131911.2022.2026892 Blasco-Arcas L Buil I Hernández-Ortega B Sese FJ Using clickers in class. The role of interactivity, active collaborative learning and engagement in learning performance Computers & Education 2013 62 102 110 10.1016/j.compedu.2012.10.019 Brunsting NC Zachry C Takeuchi R Predictors of undergraduate international student psychosocial adjustment to US universities: a systematic review from 2009–2018 International Journal of Intercultural Relations 2018 66 22 33 10.1016/j.ijintrel.2018.06.002 Cai DA Rodríguez JI Adjusting to cultural differences: the intercultural adaptation model Intercultural Communication Studies 1997 1 2 31 42 Cao C Meng Q Effects of online and direct contact on chinese international students’ social capital in intercultural networks: testing moderation of direct contact and mediation of global competence Higher Education 2020 80 4 625 643 10.1007/s10734-020-00501-w Cao G Tian Q Social media use and its effect on university student’s learning and academic performance in the UAE Journal of Research on Technology in Education 2022 54 1 18 33 10.1080/15391523.2020.1801538i»¿ Chai DS Chae C Lee J International students’ psychological sapital in Japan: Moderated mediation of adjustment and engagement Journal of Student Affairs Research and Practice 2022 59 1 101 115 10.1080/19496591.2021.1943417 Chai DS Van HTM Wang CW Lee J Wang J What do international students need? The role of family and community supports for adjustment, engagement, and organizational citizenship behavior Journal of International Students 2020 10 3 571 589 10.1080/19496591.2021.1943417 Chang CT Tu CS Hajiyev J Integrating academic type of social media activity with perceived academic performance: a role of task-related and non-task-related compulsive internet use Computers & Education 2019 139 157 172 10.1016/j.compedu.2019.05.011 Chawinga WD Taking social media to a university classroom: teaching and learning using Twitter and blogs International Journal of Educational Technology in Higher Education 2017 14 1 1 19 10.1186/s41239-017-0041-6 Cleary TJ Zimmerman BJ Christenson S Reschly A Wylie C A cyclical self-regulatory account of student engagement: theoretical foundations and applications Handbook of research on student Engagement 2012 Boston, MA Springer Cho J Yu H Roles of university support for international students in the United States: analysis of a systematic model of university identification, university support, and psychological well-being Journal of Studies in International Education 2015 19 1 11 27 10.1177/1028315314533606 Coleman JS Social capital in the creation of human capital American Journal of Sociology 1988 94 S95 S120 10.1086/228943 Cruwys T Ng NW Haslam SA Haslam C Identity continuity protects academic performance, retention, and life satisfaction among international students Applied Psychology 2021 70 3 931 954 10.1111/apps.12254 Demes KA Geeraert N Measures matter: scales for adaptation, cultural distance, and acculturation orientation revisited Journal of Cross-Cultural Psychology 2014 45 1 91 109 10.1177/0022022113487590 Deng, L., & Yuan, K. H. (2022). Which method is more powerful in testing the relationship of theoretical constructs? A meta comparison of structural equation modeling and path analysis with weighted composites. Behavior Research Methods, 1(20). 10.3758/s13428-022-01838-z Dillman DA Internet, phone, mail, and Mixed-Mode surveys: the tailored design method 2014 4 New York, NY John Wiley & Sons Finn JD Zimmer KS Christenson S Reschly A Wylie C Student engagement: what is tt? Why does it matter? Handbook of research on student engagement 2012 Boston, MA Springer Flanigan AE Babchuk WA Social media as academic quicksand: a phenomenological study of student experiences in and out of the classroom Learning and Individual Differences 2015 44 40 45 10.1016/j.lindif.2015.11.003 Flanigan AE Kiewra KA What college instructors can do about student cyber-slacking Educational Psychology Review 2018 30 2 585 597 10.1007/s10648-017-9418-2 Fredricks JA Blumenfeld PC Paris AH School engagement: potential of the concept, state of the evidence Review of Educational Research 2004 74 1 59 109 10.3102/00346543074001059 Gibson, J. J. (1986). The ecological approach to visual perception. Hillsdale, NJ: Lawrence. Glass CR Gesing P The development of social capital through international students’ involvement in campus organizations Journal of International Students 2018 8 3 1274 1292 10.5281/zenodo.1254580 Glass CR Kociolek E Wongtrirat R Lynch RJ Cong S Uneven experiences: the impact of student-faculty interactions on international students’ sense of belonging Journal of International Students 2015 5 4 353 367 10.32674/jis.v5i4.400 Glass CR Westmont CM Comparative effects of belongingness on the academic success and cross-cultural interactions of domestic and international students International Journal of Intercultural Relations 2014 38 106 119 10.1016/j.ijintrel.2013.04.004 Gomes, C. (2015). Negotiating everyday life in Australia: Unpacking the parallel society inhabited by Asian international students through their social networks and entertainment media use. Journal of Youth Studies, 18(4), 515-536. 10.1080/13676261.2014.992316 Goodenow C Grady KE The relationship of school belonging and friends’ values to academic motivation among urban adolescent students The Journal of Experimental Education 1993 62 1 60 71 10.1080/00220973.1993.9943831 Hair JF Black WC Babin BJ Anderson RE Tatham RL Multivariate data analysis 2006 6 Upper Saddle River, New Jersey Prentice–Hall, International Inc Handelsman MM Briggs WL Sullivan N Towler A A measure of college student course engagement Journal of Educational Research 2005 98 184 191 10.3200/JOER.98.3.184-192 Hox JJ Bechger TM An introduction to structural equation modeling Family Science Review 1998 11 354 373 10.1080/01688638808402800 Humphrey A Forbes-Mewett H Social value systems and the mental health of international students during the COVID-19 pandemic Journal of International Students 2021 11 S2 58 76 10.32674/jis.v11iS2.3577 Hutchby Ian Technologies, Texts and Affordances Sociology 2001 35 2 441 456 10.1177/S0038038501000219 Institute of International Education (2021). Open Doors 2021. https://opendoorsdata.org/annual-release/international-students/ Jordan, M., & Hartocollis, A. (2020). U.S. rescinds plan to strip visas from international students in online classes The New York Times. https://www.nytimes.com/2020/07/14/us/coronavirus-international-foreign-student-visas.html Kabilan, M. K., Ahmad, N., & Abidin, M. J. Z. (2010). Facebook: An online environment for learning of English in institutions of higher education?. Internet and Higher Education, 13(4), 179–187. Elsevier Ltd. Retrieved September 30, 2022 from https://www.learntechlib.org/p/108382/ Kim, Y. S., & Kim, Y. Y. (2016). Ethnic Proximity and Cross-Cultural Adaptation: A Study of Asian and European Students in the United States. Intercultural Communication Studies, 25(3). Kim E Hogge I Microaggressions against asian international students in therapy Professional Psychology: Research and Practice 2021 52 3 279 289 10.1037/pro0000383 Kim YY Communication and cross-cultural adaptation: an integrative theory 1988 Clevedon, United Kingdom Multilingual Matters Kim YY Becoming Intercultural: an integrative theory of communication and cross-cultural adaptation 2001 Thousand Oaks, CA Sage Publication Kunka BA Twitter in higher education: increasing student engagement Educational Media International 2020 57 4 316 331 10.1080/09523987.2020.1848508 Li X Chen W Facebook or Renren? A comparative study of social networking site use and social capital among chinese international students in the United States Computers in Human Behavior 2014 35 116 123 10.1016/j.chb.2014.02.012 Lin JH Peng W Kim M Kim SY LaRose R Social networking and adjustments among international students New Media & Society 2012 14 3 421 440 10.1177/1461444811418627 Mbawuni J Nimako SG Critical factors underlying students’ choice of institution for graduate programmes: empirical evidence from Ghana International Journal of Higher Education 2015 4 1 120 135 10.5430/ijhe.v4n1p120 Moon CY Zhang S Larke PJ James MC We are not all the same: a qualitative analysis of the nuanced differences between chinese and south korean international graduate students’ experiences in the United States Journal of International Students 2020 10 1 28 49 10.32674/jis.v10i1.770ojed.jis Muthén, L. K., & Muthén, B. O. (2017). Mplus user’s guide (8th ed.). Los Angeles, CA: Muthén & Muthén Pekrun R Linnenbrink-Garcia L International handbook of emotions in education 2014 New York Routledge Phua J Jin SAA “Finding a home away from home”: the use of social networking sites by Asia-Pacific students in the United States for bridging and bonding social capital Asian Journal of Communication 2011 21 5 504 519 10.1080/01292986.2011.587015 Pozzi, G., Pigni, F., & Vitari, C. (2014). Affordance theory in the IS discipline: A review and synthesis of the literature. In AMCIS 2014 Proceedings. https://halshs.archives-ouvertes.fr/halshs-01923663 Putnam RD Bowling alone: the collapse and revival of american community 2000 New York Simon and Schuster Roberts TS Computer-supported collaborative learning in higher education IGI Global 2005 10.4018/978-1-59140-408-8.ch001 Russell J Rosenthal D Thomson G The international student experience: three styles of adaptation Higher Education 2010 60 2 235 249 10.1007/s10734-009-9297-7 Sawir E Marginson S Deumert A Nyland C Ramia G Loneliness and international students: an australian study Journal of Studies in International Education 2008 12 2 148 180 10.1177/1028315307299699 Schwinger M Stiensmeier-Pelster J Effects of motivational regulation on effort and achievement: a mediation model International Journal of Educational Research 2012 56 35 47 10.1016/j.ijer.2012.07.005 Skinner EA Wentzel K Miele D Engagement and disaffection as central to processes of motivational resilience development Handbook of motivation at school 2016 New York Routledge 145 168 Stathopoulou A Siamagka NT Christodoulides G A multi-stakeholder view of social media as a supporting tool in higher education: an educator–student perspective European Management Journal 2019 37 4 421 431 10.1016/j.emj.2019.01.008 Stefanone MA Kwon K Lackaff D The value of online friends: networked resources via social network sites First Monday 2011 10.5210/fm.v16i2.3314 Sun X Hall GCN DeGarmo DS Chain J Fong MC A longitudinal investigation of discrimination and mental health in chinese international students: the role of social connectedness Journal of Cross-Cultural Psychology 2021 52 1 61 77 10.1177/0022022120979625 Tindell DR Bohlander RW The use and abuse of cell phones and text messaging in the classroom: a survey of college students College Teaching 2012 60 1 9 10.1080/87567555.2011.604802 Tran LT Pham L International students in transnational mobility: intercultural connectedness with domestic and international peers, institutions and the wider community Compare: A Journal of Comparative and International Education 2016 46 4 560 581 10.1080/03057925.2015.1057479 Tsai JY Phua J Pan S Yang CC Intergroup contact, COVID-19 news consumption, and the moderating role of digital media trust on prejudice toward Asians in the United States: cross-sectional study Journal of Medical Internet Research 2020 22 9 e22767 10.2196/22767 32924948 Tu H English versus native language on social media: international students’ cultural adaptation in the US Journal of International Students 2018 8 4 1709 1721 10.5281/zenodo.1468074 Tyson DF Linnenbrink-Garcia L Hill NE Regulating debilitating emotions in the context of performance: achievement goal orientations, achievement-elicited emotions, and socialization contexts Human Development 2009 52 6 329 356 10.1159/000242348 Van Horne S Lin S Anson M Jacobson W Engagement, satisfaction, and belonging of international undergraduates at US research universities Journal of International Students 2018 8 1 351 374 10.5281/zenodo.1134313 Wellman B Haase AQ Witte J Hampton K Does the internet increase, decrease, or supplement social capital? Social networks, participation, and community commitment American Behavioral Scientist 2001 45 3 436 455 10.1177/00027640121957286 Williams D On and off the ’net: scales for social capital in an online era Journal of Computer-Mediated Communication 2006 11 2 593 628 10.1111/j.1083-6101.2006.00029.x Yu Q Foroudi P Gupta S Far apart yet close by: social media and acculturation among international students in the UK Technological Forecasting and Social Change 2019 145 493 502 10.1016/j.techfore.2018.09.026

0

PMC9748897

NO-CC CODE

2022-12-15 23:22:48

no

Educ Inf Technol (Dordr). 2022 Dec 14;:1-24

utf-8

Educ Inf Technol (Dordr)

10.1007/s10639-022-11520-8

oa_other

==== Front Gynäkologie Die Gynäkologie 2731-7102 2731-7110 Springer Medizin Heidelberg 5046 10.1007/s00129-022-05046-w Übersichten HIV-Infektion und -Exposition bei Kindern und Jugendlichen 25 Jahre antiretrovirale Prophylaxe und Therapie: Was ist erreicht? HIV infection and exposure in children and adolescents25 years of antiretroviral prophylaxis and treatment: what has been achieved? Baumann Ulrich baumann.ulrich@mh-hannover.de 1 Schulze Sturm Ulf 2 Königs Christoph 3 1 grid.10423.34 0000 0000 9529 9877 Klinik für Pädiatrische Pneumologie, Allergologie und Neonatologie, Bereich Immunologie, Medizinische Hochschule Hannover, Carl-Neuberg-Straße 1, 30625 Hannover, Deutschland 2 grid.13648.38 0000 0001 2180 3484 Klinik für Kinder- und Jugendmedizin, Universitätsklinikum Hamburg-Eppendorf, Martinistr. 52, 20246 Hamburg, Deutschland 3 grid.411088.4 0000 0004 0578 8220 Klinik für Kinder- und Jugendmedizin, Universitätsklinikum Frankfurt, Theodor-Stern-Kai 7, 60596 Frankfurt am Main, Deutschland Redaktion Klaus Diedrich, Lübeck 14 12 2022 111 © The Author(s), under exclusive licence to Springer Medizin Verlag GmbH, ein Teil von Springer Nature 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Hintergrund Seit 1997 steht mit den antiretroviralen, kombinierbaren Wirkstoffen eine effektive Prävention und Behandlung einer Infektion mit dem „human immunodeficiency virus“ (HIV) zur Verfügung. Fragestellung Was wurde in dieser Zeit durch die Transmissionsprophylaxe und Behandlung der HIV-Infektion bei Kindern in Deutschland erreicht? Material und Methoden Darstellung der Entwicklung der Transmissionsprophylaxe und der epidemiologischen Erhebungen des Robert Koch-Instituts, Darstellung der Arzneimittelentwicklung und der Leitlinien zur antiretroviralen Therapie (ART) bei Kindern und Jugendlichen, Vorstellung von Erhebungen der deutschen Kinder-HIV-Kohorte („German pediatric and adolescent HIV cohort“, GEPIC). Ergebnisse Die Einführung der HIV-Testung Schwangerer hat die regelmäßige ART HIV-positiver Schwangerer und die Einführung von transmissionsmindernden Maßnahmen möglich gemacht. Damit konnte das Risiko der vertikalen Transmission von ca. 30 % auf < 1 % gesenkt werden. Gegenwärtig wird untersucht, ob HIV-exponierte Kinder ohne Risiko gestillt werden können. Die ART bei Kindern hat die Überlebensraten sowie die körperliche und die kognitive Entwicklung HIV-positiver Kinder entscheidend verbessert. Neue Wirkstoffe haben ein günstigeres Nebenwirkungsprofil. In Deutschland konnten die WHO-Ziele, bei 90 % der Kinder eine vollständige Suppression der Viruslast zu erzielen, erreicht werden. Späte Diagnosen sind bei Kindern und Jugendlichen weiterhin mit schweren Infektionen verbunden. Die HIV-Infektion ist weiterhin eine Erkrankung mit Stigma geblieben und wird deswegen meist gegenüber den Kindern und ihrer Umwelt geheim gehalten. Schlussfolgerung Mit der jetzt möglichen Transmissionsprophylaxe und Therapie gelingt es, Transmissionen weitgehend zu verhindern und HIV-positiven Kindern bei frühzeitiger Diagnose ein weitgehend gesundes Leben zu ermöglichen. Das Ziel, dass Kinder ihre Krankheit ohne Angst kennen und nennen können, ist nach wie vor nicht erreicht. Background Since 1997 an effective prevention and treatment of infections with human immunodeficiency virus (HIV) have been available in the form of antiretroviral combination therapy. Objective What has been achieved during this time by transmission prophylaxis and treatment of HIV infections in children in Germany? Material and methods Presentation of the development of transmission prophylaxis and the epidemiological surveys of the Robert Koch Institute, presentation of drug development and guidelines for antiretroviral therapy in children and adolescents and presentation of surveys of the German pediatric and adolescent HIV cohort (GEPIC). Results The introduction of HIV testing of pregnant women has made regular antiretroviral treatment of HIV positive pregnant women and the introduction of transmission-reducing measures possible. This has reduced the risk of vertical transmission from about 30% to < 1%. Currently, it is being investigated whether children exposed to HIV can be breastfed without risk. Antiretroviral therapy in children has decisively improved the survival rates and the physical and cognitive development of HIV positive children. New active substances have a more favorable side effect profile. In Germany, the World Health Organization (WHO) target of achieving complete suppression of the viral load in 90% of the children has been achieved. A delayed diagnosis in children and adolescents is still associated with severe infections. An HIV infection remains a disease with a stigma and is therefore usually kept secret from the children and their environment. Conclusion With the transmission prophylaxis and therapy that are now available, it is possible to prevent transmission to a large extent and with a timely diagnosis to enable HIV positive children to lead a largely healthy life. The goal of children being able to know and name their disease without fear has still not been achieved. Schlüsselwörter Schwangere Frauen HIV-Testung Infektionskrankheit, vertikale Transmission Antiretrovirale Therapie GEPIC-Kohorte Keywords Pregnant women HIV testing Infectious disease transmission, vertical Antiretroviral therapy GEPIC cohort ==== Body pmcIm Rahmen der durch das „severe acute respiratory syndrome coronavirus 2“ (SARS-CoV-2) ausgelösten Pandemie ist es um das „human immunodeficiency virus“ (HIV) vergleichsweise still geworden. Dennoch hat die Zahl der Menschen, die in Deutschland mit der HIV-Infektion leben, einen Höchststand erreicht. Für viele ist es überraschend, dass auch Kinder HIV-infiziert sein können. Dies liegt nicht nur an der vergleichsweise geringen Zahl betroffener Kinder, sondern auch an der Geheimhaltung der HIV-Infektion, selbst gegenüber den betroffenen Kindern. Fallbeispiel Ein bisher gesunder männlicher Säugling, reif geboren nach 39 + 6 SSW, wird im Alter von 3½ Monaten in der Kinderarztpraxis vorgestellt. Er leidet seit einigen Tagen an trockenem Husten, dazu an einer neu aufgetretenen Diarrhö sowie an Fieber und einer Tachydyspnoe. Etwa eine Woche zuvor hatte das Kind die zweite Rotavirusimpfung erhalten. Auskultatorisch findet sich ein unauffälliger Befund. Die periphere Sauerstoffsättigung beträgt aber 90 %. Das Kind wird stationär eingewiesen. Nach der stationären Aufnahme verschlechtert sich die respiratorische Situation rasch. Unter Atemhilfe zunächst mit einer „high flow nasal cannula“ (HFNC) und dann mithilfe des „continuous positive airway pressure“ (CPAP) verschlechtert sich der Zustand weiter. Unter dem Bild eines schweren „acute respiratory distress syndrome“ (ARDS) wird das Kind intubiert und mit einer inspiratorischen Sauerstofffraktion (FIO2) von 1,0 und Beatmungsdrücken von bis zu 35 cm H2O inspiratorisch und einem endexspiratorischem Druck von 15 cm H2O beatmet. Radiologisch zeigen sich diffuse bilaterale Infiltrate. Es besteht nur eine milde Konzentrationserhöhung des C‑reaktiven Proteins (CRP). Dennoch wird eine empirische antibiotische Therapie eingeleitet. Die Blutkulturen bleiben ohne Erregernachweis; im Stuhl werden Rotaviren, im Trachealsekret Parainfluenzaviren nachgewiesen. Aufgrund der Schwere des Krankheitsverlaufs bei dem nachgewiesenen Erregerspektrum wird eine Immundefektdiagnostik nach unauffälligem Neugeborenenscreening durchgeführt. Die initiale immunologische Diagnostik zeigt CD4+-T-Zellen im unteren altersbezogenen Normbereich. Erst im weiteren Verlauf fällt die Zahl der CD4+-T-Zellen deutlich ab. Eine ebenfalls durchgeführte „polymerase chain reaction“ (PCR) zum HIV-Nachweis zeigt eine sehr hohe Viruslast von 9.500.000 Kopien/ml. Umgehend wird eine antiretrovirale Therapie (ART) mit den nukleosidischen Reverse-Transkriptase-Inhibitoren Zidovudin und Lamivudin in Kombination mit dem Integraseinhibitor Dolutegravir aufgenommen. Für alle 3 Wirkstoffe sind orale Flüssigformulierungen verfügbar. Die neu gestellte Diagnose einer HIV-Infektion weckt den Verdacht auf eine Pneumonie durch einen opportunistischen Erreger. Wegen des negativen PCR-Befunds bezüglich des Zytomegalie-Virus (CMV) und der erhöhten Lactatdehydrogenase(LDH)-Konzentration im peripheren Blut wird der Verdacht auf eine Pneumocystis-jirovecii-Pneumonie (PJP) gestellt und eine empirische Therapie mit Cotrimoxazol und Prednisolon begonnen. Der Erreger wird später aus dem Magensaft und dem Trachealsekret isoliert. Im weiteren Verlauf entwickelt das Kind einseitig beginnende und sekundär generalisierte Krampfanfälle. Aufgrund des Verdachts auf eine weitere opportunistische Infektion, eine Kryptokokken-Meningitis, wird das Kind empirisch zusätzlich mit Amphotericin B und Fluconazol sowie antikonvulsiv mit Levetiracetam behandelt. Das Kryptokokken-Antigen lässt sich im Serum und im Liquor jedoch nicht nachweisen. Trotz adäquater Behandlung der PJP verschlechtert sich der respiratorische Zustand weiter und macht eine Beatmung mit inhalativem Stickstoffmonoxid (iNO) erforderlich. Erst nach mehreren Wochen setzt eine langsame, aber stetige Besserung ein, die schließlich eine Entwöhnung von der invasiven Beatmung ermöglicht. Nach CPAP-Unterstützung für weitere 2 Wochen und fortgesetzter Sauerstoffgabe über weitere 4 Wochen ist der Zustand des Kindes respiratorisch in Raumluft stabil. Die cMRT zeigt nun eine globale Atrophie mit Marklagerverlust und bifrontal mutmaßlich postischämische Läsionen – passend zu einer HIV-Leukenzephalopathie. Die Mutter war im 1. Trimenon HIV-negativ getestet worden. Sie hat die Infektion also mutmaßlich im weiteren Verlauf der Schwangerschaft akquiriert. Die HIV-Infektion während der Schwangerschaft birgt ein besonders hohes Risiko für eine Mutter-Kind-Übertragung. Ein 2. HIV-Antikörpertest, wie in Großbritannien üblich, hätte mit größerer Wahrscheinlichkeit (> 98 %) die kindliche Infektion verhindern können. Die unerkannte konnatale HIV-Infektion kann innerhalb weniger Monate zu lebensbedrohlichen opportunistischen Infektionen und anderen das „acquired immune deficiency syndrome“ (AIDS) definierenden Ereignissen (Infobox 1) führen. Infobox 1 Auswahl der das „acquired immune deficiency syndrome“ (AIDS) definierenden Erkrankungen im Kindesaltera > 1 invasive bakterielle Infektion innerhalb von 2 Jahren „Human-immunodeficiency-virus“(HIV)-Enzephalopathie Wasting-Syndrom, Kachexie Pneumocystis-jirovecii-Pneumonie (PCP) Zerebrale Toxoplasmose bei Kindern im Alter > 1 Monat Diarrhö > 1 Monat Dauer durch Kokzidien (Cryptosporidium oder Isospora) hervorgerufen Lymphome, einschließlich Lymphomen des Zentralnervensystems (ZNS), Kaposi-Sarkom Progressive multifokale Leukenzephalopathie (PML) Durch Herpes-simplex-Viren (HSV) verursachte mukokutane Ulzera (Dauer > 1 Monat) Bronchitis, Pneumonie oder Ösophagitis durch HSV bei Kindern im Alter > 1 Monat Lymphoide interstitielle Pneumonie (LIP) durch das Epstein-Barr-Virus (EBV) ausgelöst Zytomegalie-Virus (CMV): u. a. Retinitis, Ösophagitis, Kolitis bei Kindern im Alter > 1 Monat Kandidose des Ösophagus oder des Tracheobronchialsystems Extrapulmonale Kryptokokkose Disseminierte oder extrapulmonale Histoplasmose Tuberkulose, atypische Mykobakteriosen aModifiziert nach Caldwell et al. [6] Vertikale Transmission – und wie sie verhindert werden kann Unter den schätzungsweise 91.000 Menschen, die in Deutschland mit einer HIV-Infektion leben, befinden sich gemäß der Erhebung der pädiatrischen HIV-Ambulanzen rund 400 Kinder und Jugendliche [4]. Im Unterschied zu Erwachsenen werden Kinder nahezu ausschließlich während der Schwangerschaft, Geburt oder Stillzeit infiziert (vertikale Transmission). Zur Prävention der kindlichen HIV-Infektion zählten ursprünglich die Sectio-Entbindung, ein Stillverzicht und eine 6‑wöchige Zidovudinprophylaxe beim Säugling, wodurch die Transmissionsrate von etwa 30 % auf weniger als 2 % gesenkt werden konnte. Voraussetzung für diese Maßnahmen war, dass die HIV-Infektion bei der Mutter auch bekannt war. Wegen geringen HIV-Testraten bei Schwangeren wurde 2007 eine verpflichtende Beratung über den HIV-Test in die Mutterschaftsrichtlinien aufgenommen. Im Jahr 2016 war die Testrate Schwangerer auf 92 % gestiegen [21]. Ein deutlicher Rückgang der HIV-Neuinfektionen in Deutschland geborener Kinder war die Folge. Therapie der Mutter Heute beschränken sich perinatale HIV-Transmissionen in Mitteleuropa trotz einer auf 800 angestiegenen jährlichen Zahl der Schwangerschaften HIV-positiver Mütter auf Einzelfälle [20]. Die Transmissionsrate beträgt unter 1 %. Mit den heutigen Wirkstoffen können HIV-positiv getestete Schwangere noch vor der Entbindung eine nichtnachweisbare Viruslast erreichen, was gleichbedeutend mit einem sehr geringen Transmissionsrisiko ist. Je stabiler die Virusreplikation bei der Mutter unterdrückt ist, desto sicherer erscheint eine Prävention der vertikalen Transmission. In einer großen Kohorte von Müttern, die bereits vor der Konzeption eine dauerhaft nichtnachweisbare Virusreplikation aufwiesen, kam es zu keiner einzigen Transmission [19]. Entsprechend wurde in den letzten Jahren der Katalog der übrigen Präventionsmaßnahmen schrittweise verkleinert: vaginale statt Sectio-Entbindung (sofern die HIV-Viruslast < 50 Kopien/ml beträgt) sowie eine auf 2 Wochen verkürzte und unter bestimmten Voraussetzungen sogar ausgesetzte antiretrovirale Postexpositionsprophylaxe beim Säugling. Ist Stillen noch gefährlich? Inzwischen steht auch der Stillverzicht zur Disposition. In ressourcenarmen Ländern wurde das Stillen durch HIV-positive Mütter von der WHO von Anfang an empfohlen, weil die diarrhöbedingte Sterblichkeit bei nichtgestillten Säuglingen höher war als die vertikalen HIV-Transmissionsraten. Entsprechend stammen die meisten Daten zur Transmission gestillter Säuglinge aus den Hochprävalenzländern [12]. Obwohl das HIV-Genom in der Muttermilch auch dann noch nachweisbar ist, wenn die Viruslast im peripheren Blut unter der Nachweisgrenze liegt, sind die Infektionsraten bei gestillten Kindern in Europa so niedrig, dass sich Transmissionsraten kaum berechnen lassen. Die verfügbaren Daten reichen zwar nicht aus, um Stillen als „sicher“ zu betrachten, aber sie stellen kein eindeutiges Gefahrensignal mehr dar, um deshalb vom Stillen abzuraten [9, 15, 27]. Gegenwärtig finden in Deutschland und der Schweiz prospektive Erhebungen statt, um die vergleichsweise wenigen Fälle gestillter Säuglinge mit einer HIV-Infektion in Deutschland auswerten zu können [15, 27]. Entsprechend empfehlen die aktuellen Leitlinien, die Entscheidung für oder gegen das Stillen gemeinsam mit der werdenden Mutter bzw. den werdenden Eltern nach eingehender Aufklärung über Vorbedingungen und Vorteile des Stillens sowie dessen mögliche Risken zu treffen [9]. Gestillte Säuglinge sollen in der pädiatrischen HIV-Nachsorge allerdings häufiger als Ungestillte virologisch untersucht werden (Abb. 1). Nach Erfahrung der Autoren ist die Enttabuisierung des Stillens HIV-positiver Mütter noch nicht in allen Arbeitsbereichen der Behandlung Schwangerer und Wöchnerinnen angekommen. Damit ist es auch Aufgabe der pädiatrischen HIV-Behandler*innen, die geburtshilflichen und neonatologischen Teams mit dieser Neuerung vertraut zu machen. Nachsorge exponierter Kinder Auch für Kinderärzt*innen ist die Betreuung HIV-exponierter Kinder einfacher geworden. Die früher aufwendige Einschätzung des Transmissionsrisikos und die daraus folgende Wahl der prophylaktischen ART des Neugeborenen hat sich auf wenige Kriterien reduziert: die Bestimmung der Viruslast bei der Mutter und das Reifealter des Kindes (Abb. 2). Um eine etwaige Transmission frühzeitig zu entdecken, werden HIV-exponierte Neugeborene nach einem risikoadaptierten Schema auf eine mögliche HIV-Infektion untersucht (Abb. 1). Bis zum Alter von 18 Monaten kann wegen der langen Persistenz maternaler Antikörper im kindlichen Blut allerdings der Antikörpertest nicht angewendet werden. Um eine HIV-Infektion dennoch frühzeitig detektieren zu können, werden bei den exponierten Kindern serielle HIV-PCR-Untersuchungen durchgeführt. Nabelschnurblut sollte nicht untersucht werden, weil das HIV-Genom darin häufig nachgewiesen werden kann, ohne dass die Kinder später HIV-positiv sind. Erstversorgende Kinderärzt*innen stehen vor der Frage, ob die HIV-Exposition ins Vorsorgeheft einzutragen ist. Ein solcher Eintrag birgt die Gefahr, dass die HIV-Infektion der Mutter gegenüber Dritten unbeabsichtigt offenbart wird. Der Hinweis auf eine mütterliche Infektion in der Schwangerschaft in dem dafür vorgesehenen Feld ist ausreichend. Die Qualität der Exponiertennachsorge hängt auch nicht von einem Eintrag in dem Vorsorgeheft ab, sondern von der Planung bei der U1 und U2: Die medikamentöse Prophylaxe ist einzuleiten, und die Nachsorgeuntersuchungen müssen terminiert werden. Um den Müttern eine angemessene Entscheidung für oder gegen das Stillen zu ermöglichen, eine sachgerechte risikoadaptierte Wahl von Geburtsmodus und postnataler Prophylaxe zu sichern und die Umsetzung veralteter Leitlinien zu verhindern, sollten HIV-positive Mütter ihre Kinder in HIV-Schwerpunkt-Kliniken zur Welt bringen. Die früher realistische Erwartung werdender HIV-positiver Mütter, ihr Kind zu infizieren und selbst zu sterben, bevor sie dieses großziehen können, ist Geschichte – oder könnte es zumindest sein. In der Betreuung exponierter Kinder zeigt sich, dass Ängste und Missverständnisse langlebiger sein können als die Viren. Immer wieder scheuen sich HIV-positive Mütter ihre Kinder zu küssen, um sie nicht dabei zu infizieren. Küssen ist aber selbst bei hoher Viruslast nicht infektiös. Infektion verhindert – Probleme gelöst? Trotz der Erfolge der perinatalen Transmissionsprophylaxe sind nicht alle Fragen gelöst. In mehreren Studien wiesen Kinder von HIV-positiven Müttern, die während der Schwangerschaft eine ART einnahmen, im Vergleich zu Kindern HIV-negativer Mütter ein geringeres Geburtsgewicht, schwächere Impfantworten und höhere Raten an Infektionen und Hospitalisierungen auf und ihre motorische und sprachliche Entwicklung war verzögert [11, 18, 29]. Ursächlich sind wahrscheinlich nicht oder nicht nur die während der Schwangerschaft eingenommenen Wirkstoffe. Kinder von Müttern, die bereits vor der Schwangerschaft eine ART erhielten, entwickelten sich besser als Kinder von Müttern, deren ART erst während der Schwangerschaft begonnen wurde [14]. Vor allem im ersten Lebensjahr waren die Morbidität und infektionsbedingte Mortalität HIV-exponierter Kinder im Vergleich zu nichtexponierten Kindern höher. Möglicherweise führt eine Virusreplikation in der Frühphase der Schwangerschaft bereits zu Schäden beim Ungeborenen, unabhängig von dessen tatsächlicher Infektion. Eine Diagnosestellung und Behandlung der HIV-Infektion bei Frauen vor der Schwangerschaft würde demnach nicht nur die Gesundheit der Betroffenen, sondern auch die ihrer zukünftigen Kinder verbessern. Klinischer Verlauf der kindlichen Infektion Der Verlauf der HIV-Infektion bei Kindern, die peri- oder postnatal mit dem HIV infiziert wurden, unterscheidet sich, wie das obige Fallbeispiel illustriert, erheblich von dem Verlauf bei Erwachsenen. Bei Erwachsenen kontrolliert das Immunsystem nach der Infektion innerhalb weniger Wochen die Virusreplikation auf ein relativ geringes Niveau. Anschließend fällt in einem mehrjährigen Zeitraum die Zahl der CD4+-T-Helferzellen nur langsam ab. Meist fühlen sich die Betroffenen in diesem Stadium gesund. Erst wenn die Zahl der T‑Helferzellen deutlich abgefallen ist – als Richtwert gelten 250 CD4+-T-Zellen/µl –, steigt das Risiko für die das „acquired immunodeficiency syndrome“ (AIDS) definierenden Erkrankungen, z. B. Pneumonie durch Pneumocystis jirovecii, Soorösophagitis oder progressive multifokale Leukenzephalopathie (PML). Bei einem weiteren Abfall der T‑Helferzellen auf < 50/µl bestehen hohe Risiken für eine CMV-Retinitis und Kryptokokkose. Dem Immunsystem perinatal infizierter Kinder gelingt es während der Säuglingszeit dagegen kaum, die Virusreplikation zu begrenzen. Ein Fünftel der perinatal infizierten Kinder erleidet bereits im Säuglingsalter HIV-Enzephalitiden [7]. Wegen der geringer kontrollierten Virusreplikation ist bei Kindern die Progression zum Stadium AIDS schneller. Bis zum Alter von 4 Jahren sind bereits bei der Hälfte der perinatal HIV-infizierten Kinder AIDS-definierende Erkrankungen eingetreten [10], bei den anderen kann diese Phase bis in das Jugendalter reichen. Bis dahin kann auch ohne schwere Erkrankungen eine beeinträchtigte Entwicklung auf eine HIV-Infektion hinweisen, z. B. in Form von Gedeihstörung [4], Kleinwuchs [24], verminderter Knochendichte [23] oder verschlechterten Lungenfunktionsparametern [13]. Auch die neurokognitive Entwicklung kann beeinträchtigt sein [17]. Gemäß den Beobachtungen der Autoren stellen dazu die ersten 3 Lebensjahre eine vulnerable Phase dar. Die Höhe der Viruslast in dieser Altersspanne korrelierte negativ mit der neurokognitiven Entwicklung, nicht jedoch die Viruslast in höheren Altersgruppen [28]. Bei Kindern verläuft die Progression zum Stadium AIDS schneller als bei Erwachsenen Unentdeckte HIV-Infektionen bei in Deutschland geborenen Kindern sind wegen des flächendeckenden HIV-Screenings Schwangerer und der wirksamen Postexpositionsprophylaxe eine Seltenheit. Bei Kindern, die in Hochprävalenzregionen geboren wurden und später nach Europa migrieren, können vergleichbare Programme nicht vorausgesetzt werden. Deshalb werden in Deutschland die meisten HIV-Neudiagnosen bei Kindern gestellt, die nicht in Deutschland geboren wurden. Unabhängig vom Geburtsort muss die Diagnose frühzeitig gestellt werden, um schwerste Verläufe mit Todesfolge oder dauerhaften Schäden zu verhindern. Diese treten meist in der frühen Kindheit auf, können sich aber auch erst im Jugendalter manifestieren. Die Infobox 2 fasst klinische Warnzeichen für eine HIV-Infektion zusammen. Infobox 2 Wann an HIV-Infektion denken? Herkunft aus Hochprävalenzregion (Subsahara-Afrika, Osteuropa, Südostasien) Gedeihstörung Kleinwuchs Lymphadenopathie Schwerer Herpes zoster Chronisches Ekzem Ungewöhnlich hohe Immunglobulinspiegel Lymphozytopenie Opportunistische Infektionen und andere AIDS-definierende Erkrankungen Antiretrovirale Therapie Durchführung Mit der ART sollen eine dauerhafte Suppression der Virusreplikation, der Erhalt oder die Wiederherstellung der Immunkompetenz, die Senkung der Morbidität und Mortalität sowie eine möglichst gesunde körperliche und kognitive Entwicklung der Kinder erreicht werden [8]. Wegen der besonderen Vulnerabilität von Säuglingen und Kleinkindern wurde in den von der Pädiatrischen Arbeitsgemeinschaft AIDS (PAAD) formulierten Leitlinien für diese Altersgruppe bereits eine generelle Therapieempfehlung gegeben, als international noch der Beginn der Therapie von der Zahl der CD4+-T-Helferzellen abhängig gemacht wurde [22]. Heute wird international angesichts der wesentlich besseren Verträglichkeit der Wirkstoffe ein umgehender Beginn der ART für alle Altersgruppen empfohlen. Die 3 antiretroviral wirksamen Substanzen aus mindestens 2 verschiedenen Substanzklassen zusammen. Nach aktuellen Empfehlungen wird entweder ein „integrase strand transfer inhibitor“ (INSTI), ein nichtnukleosidischer Reverse-Transkriptase-Inhibitor (NNRTI) oder ein Proteinaseinhibitor (PI) mit 2 nukleosidischen Reverse-Transkriptase-Inhibitoren (NRTI) kombiniert. Der Zulassung des ersten NRTI Zidovudin 1987 folgten inzwischen mehr als 40 weitere Wirkstoffe aus 7 Substanzgruppen für Erwachsene (Abb. 3a). Die Auswahl der für Kinder zugelassenen und in geeigneter Formulierung erhältlichen Wirkstoffe ist dagegen eingeschränkt. Zur Behandlung Neugeborener stehen lediglich 3 Wirkstoffe zur Verfügung, im Alter von 2 Jahren sind es 12 (Abb. 3b und 4). Die Therapie von Kleinkindern gestaltet sich regelmäßig schwierig, weil die verfügbaren Substanzen meist eine niedrige Resistenzschwelle oder einen so schlechten Geschmack haben, dass deren dauerhafte Gabe problematisch sein kann. Für die Behandlung von Jugendlichen sind ab dem Alter von 12 Jahren mehrere Kombinationspräparate zugelassen (Abb. 5), die die einmal tägliche Einnahme von einer Tablette ermöglichen. Diese Konfektionen können die bei Adoleszenten möglicherweise eingeschränkte Adhärenz verbessern. Effektivität Die ART bei Kindern hat seit den späten 90er-Jahren des letzten Jahrhunderts zu erheblich gebesserten Überlebensraten geführt. In einer großen US-amerikanischen Kohorte sank die Mortalität nach Einführung der ART um fast 90 % von 7,2/100 Patientenjahre auf 0,8/100 Patientenjahre [3]. Ein früher Beginn der ART bei noch erhaltener Immunkompetenz senkt die Sterblichkeit besonders effektiv. Europäische und thailändische Daten zeigen, dass die Mortalität um das 1,8- bis 3,5Fache höher ist, wenn mit der Therapie bei Jugendlichen begonnen wird [16]. Die in Deutschland erreichten Behandlungserfolge sind im internationalen Vergleich beachtlich Im Jahr 2014 rief die WHO das globale Ziel aus, die HIV-Infektion bei mindestens 90 % der Betroffenen zu diagnostizieren, davon mindestens 90 % zu therapieren und bei diesen bei mindestens 90 % eine Viruslast unter der Nachweisgrenze zu erreichen [26]. Naturgemäß waren diese Ziele im Hinblick auf Erwachsene konzipiert, die in Deutschland über 99 % der mit einer HIV-Infektion lebenden Menschen ausmachen. Im Jahr 2020 stellte die WHO für Deutschland das Erreichen dieser 90-90-90-Ziele fest. Diese Ziele sind auch für die pädiatrische HIV-Kohorte erreicht. Die GEPIC-Studie (German Pediatric and Adolescent HIV Cohort) der PAAD zeigte eine nahezu vollständige Behandlungsrate (99,5 %) und ein gutes Therapieansprechen unter den in Deutschland behandelten Kindern und Jugendlichen: Es hatten 91 % eine Viruslast < 50 Kopien/ml [5]. Der Therapieerfolg war altersabhängig: Bei Kleinkindern betrug der Anteil 85,7 %, bei Jugendlichen bei 92,1 %. Die in Deutschland erhobenen Zahlen sind im internationalen Vergleich beachtlich. In einer britischen Erhebung betrug der Anteil der Kinder und Jugendlichen mit einer Viruslast < 400 Kopien/ml 66 % [1]. Der erreichte Therapieerfolg kann allerdings rasch verloren gehen (Infobox 3). Infobox 3 Exkurs: HIV-Infektion während der SARS-CoV-2-Pandemie Der mit hohem Aufwand von Behandler*innen und Familien erreichte Therapieerfolg kann rasch verloren gehen. Eine aktuelle Auswertung von Daten der GEPIC-Erhebung zeigt, dass die Zahl der Behandlungen in den pädiatrischen HIV-Ambulanzen während der durch das „severe acute respiratory syndrome coronavirus 2“ (SARS-CoV-2) ausgelösten Pandemie (01.03.2020–31.05.2021) im Vergleich zum unmittelbar vorausgegangenen Zeitraum (01.12.2018–28.02.2020) um 15 % (p = 0,016) zurückging. Während der Pandemie stieg der Anteil von Patient*innen mit grenzwertigem oder positiven Virusnachweis von 30 % auf 47 % (p = 0,011) an, während die Werte der CD4+-T-Helferzellen um rund 2 % (p = 0,015) abfielen. Der Anstieg der mit einer HIV-Infektion verbundenen Viruslast korrelierte mit den einzelnen Coronawellen und den dabei wirksamen Beschränkungen (u. a. virtueller Schulunterricht). Bemerkenswert war, dass diese pandemiebedingten Entwicklungen ausschließlich bei männlichen Jugendlichen nachweisbar waren (Königs C et al. Auswertung der GEPIC-Dokumentation vor und während der SARS-CoV-2-Pandemie; mündl. Mitteilung, 22.07.2022). Die Suppression der Virusreplikation geht mit einer Besserung HIV-bedingter Störungen der Entwicklung einher. Unter ART kommt es u. a. zum Aufholwachstum [24], zu einer verbesserten Lungenfunktion [13] und zu einer verbesserten neurologischen Entwicklung [25]. In Deutschland lebende HIV-positive Kinder hatten unter ART einen zu gesunden Gleichaltrigen vergleichbaren neurokognitiven Entwicklungsstand und Intelligenzquotienten [28]. Unerwünschte Arzneimittelwirkungen Im Vergleich zu den frühen Jahren der ART sind kurz- und langfristige Nebenwirkungen deutlich weniger geworden. Vor 15 Jahren wurden bei 65 % der Jugendlichen Veränderungen des Unterhautfettgewebes (Lipodystrophie) ab dem Tanner-Stadium IV beschrieben [2]. Mithilfe neuer Wirkstoffe kann eine Lipodystrophie weitgehend vermieden werden. Doch es gibt bei den neuen, als besonders gut verträglich geltenden Wirkstoffen neue Profile von unerwünschten Arzneimittelwirkungen, deren Relevanz noch nicht abschließend einzuschätzen ist. Bei Erwachsenen, die mit Tenofoviralafenamid, Dolutegravir oder Bictegravir behandelt wurden, fiel eine signifikante Gewichtszunahme auf [3]. Dieser Effekt konnte inzwischen auch bei Kindern und Jugendlichen in der deutschen GEPIC-Kohorte gezeigt werden [4]. In der Auswertung wurden nur Kinder und Jugendliche mit einem normalen Immunstatus berücksichtigt, um ein Aufholwachstum nicht als pathologisch zu werten. Es wird jetzt darauf zu achten sein, ob die ungewollte Gewichtszunahme die Therapieadhärenz bei Jugendlichen gefährdet. Umgang mit der Diagnose Muss die Diagnose verschwiegen werden? Die HIV-Infektion ist die einzige Erkrankung im Kindesalter, die den betroffenen Kindern regelmäßig verschwiegen wird. Während die Behandlung der HIV-Infektion große Fortschritte gemacht hat und sie nicht mehr den Charakter einer tödlichen Infektion besitzt, ist die Überwindung ihrer Stigmatisierung langsamer. Die Angst vor Ansteckung durch HIV-positive Kinder ist an Kindergärten, Schulen und in Vereinen sowohl bei Eltern anderer Kinder als auch bei Mitarbeiter*innen verbreitet. Ein Bekanntwerden der Diagnose beinhaltet das Risiko der Ausgrenzung bis hin zu Ausschluss und Verweis. Das Bekanntwerden der Diagnose beinhaltet noch immer das Risiko der Ausgrenzung Damit Kinder ihre HIV-Diagnose nicht unbedacht zum Ausdruck bringen und eine unkontrollierte Angstreaktion in der Umgebung auslösen, ist es in 2022 immer noch angemessen, den Familien das Verschweigen der Diagnose zu raten. Die Autoren empfehlen, den Kindern durch einen Ersatzbegriff zu erläutern, warum sie behandelt werden, um ihnen eine Einordnung des Grundes ihrer Behandlung zu ermöglichen, und gleichzeitig zu vermeiden, dass in der Umgebung der Verdacht auf ein delikates Geheimnis entsteht. Hier hat die Metapher einer Krankheit der „Körperpolizei“ weite Verbreitung gefunden und in mehr als 20 Jahren zu keinen „Unfällen“ geführt. Einige Eltern gehen allerdings mit der HIV-Diagnose offen nach außen um, ohne dadurch eine Ausgrenzung zu erfahren. Wahrscheinlich ist diese Strategie erfolgreicher, weil das offene und sachliche Ansprechen eine Einordnung der HIV-Infektion als „ungefährlich“ ermöglicht, bevor Ängste aufgrund mangelnden Wissens entstehen. Diagnoseaufklärung und sexuelle Aktivität Die HIV-Diagnose kann und soll den Kindern nicht dauerhaft verschwiegen werden. Im Alter zwischen 10 und 12 Jahren sind Kinder in der Lage, das Wissen um ihre eigene HIV-Infektion zu verstehen und es für sich zu behalten. Im Aufklärungsgespräch sollte dem Kind einleitend und unter Wertschätzung mitgeteilt werden, warum es zu diesem Gespräch kommt („Du bist jetzt alt genug, dass wir Dir ein Geheimnis anvertrauen können“), bevor erläutert wird, worin dieses Geheimnis besteht. Aufklärungen sollten vor Beginn der Pubertät geschehen, damit die Diagnosemitteilung nicht zu einem Störfaktor in der turbulenten Phase der sexuellen Entwicklung wird. Die Bedeutung der HIV-Infektion als Gefahrenmoment bei sexueller Aktivität ist nicht mehr gegeben – unter der Voraussetzung einer wirksamen ART. Die Autoren empfehlen weiter nachdrücklich, Präservative zu benutzen, aber nur noch, um Schwangerschaften und eine Ansteckung mit anderen sexuell übertragbaren Krankheiten zu vermeiden. Versorgungsstruktur in Deutschland Wegen der Seltenheit der kindlichen HIV-Infektion und der Besonderheiten ihrer Therapie (die Leitlinien umfassen 127 Seiten [8, 9]) findet die Behandlung der rund 400 positiven Kinder und Jugendlichen sowie der exponierten Säuglinge nahezu ausschließlich an Schwerpunktambulanzen von Universitäten und anderen Maximalversorgern statt. Diese Ambulanzen teilen mit den anderen Ambulanzen der spezialärztlichen Versorgung die nichtkostendeckende Finanzierung. Da die HIV-Behandlung von Kindern weitgehend ambulant erfolgt, gibt es auch keine Querfinanzierung aus dem stationären Sektor. Wie bei anderen seltenen Erkrankungen ist der Aufwand zur Fortbildung, bezogen auf die Patientenzahl, erheblich. Wahrscheinlich höher als bei den meisten seltenen Erkrankungen sind die Anteile von Kindern mit nichteuropäischer Herkunft (80 %) und von alleinerziehenden (50 %) oder nicht mehr vorhandenen Eltern (35 %, Zahlen der HIV-Ambulanz an der Medizinischen Hochschule Hannover). Psychosoziale Mitarbeiter*innen sind im ambulanten Sektor nicht finanziert und stehen in Deutschland nur in wenigen Ambulanzen drittmittelfinanziert zur Verfügung. Patientenselbsthilfen, die bei der Vernetzung der betroffenen Familien helfen und öffentlich Spenden einwerben, gibt es aus verständlichen Gründen nicht. Ausblick und Forschungsbedarf Kombinationspräparate für eine einmal tägliche Einnahme haben bei Erwachsenen und mittlerweile auch bei Jugendlichen die medikamentöse Therapie sehr vereinfacht. Kinder unter 12 Jahren können von dieser Entwicklung bisher kaum profitieren (Abb. 3). Besonders aufwendig ist die Entwicklung von Arzneimitteln, die von Kindern unter 6 Jahren eingenommen werden können, weil keine geringer dosierten Festkonfektionen verwendet werden können. Gegenwärtig prüft die Europäische Arzneimittelagentur die Zulassung des ersten antiretroviralen Arzneimittels als Kombinationspräparat für Kinder unter 6 Jahren. Es besteht aus einer 3 Wirkstoffe enthaltenden suspendierbaren Tablette. Entwicklungen dieser Art könnten dazu beitragen, dass die 90-90-90-Ziele der WHO auch bei Vorschulkindern erreicht werden. Mit HIV infizierte Kinder und Jugendliche sehen unverändert einer lebenslangen ART entgegen Während Behandler und Eltern dankbar auf die Erfolge der ART sehen, wünschen sich HIV-positive Jugendliche v. a., von der Krankheit in Ruhe gelassen zu werden. Kürzlich (für Erwachsene) zugelassene Depotwirkstoffe ermöglichen es, die Behandlung auf Injektionen in 2‑monatigen Abständen zu reduzieren. Wirkstoffe für eine einmal jährliche Injektion sind in der Entwicklung. Realistische Konzepte für eine Heilung sind dagegen weiterhin nicht erkennbar. Die provirale DNA müsste gentechnisch aus jeder infizierten Körperzelle entfernt werden. Damit sehen HIV-infizierte Kinder und Jugendliche unverändert einer lebenslangen ART entgegen. Eine Forschungsaufgabe wird die Erfassung und ggf. Behandlung möglicher Langzeiteffekte der ART sein. Gegenwärtig wird auch diskutiert, ob die verbleibende Virusreplikation unterhalb der heutigen Nachweisgrenze zu einer Langzeitaktivierung des Immunsystems und einer Störung der Integrität von Endothel- und Epithelverbänden führt [30]. Weiterführende Informationen Kontakt mit HIV-Behandler*innen: www.kinder-aids.de Fazit für die Praxis Die Prävention einer Infektion mit dem „human immunodeficiency virus“ (HIV) bei Kindern setzt die HIV-Testung und die Behandlung Schwangerer voraus. Werdende Mütter, die HIV-positiv sind, sollten in Kliniken, die im Umgang mit HIV-exponierten Säuglingen erfahren sind, entbunden werden. Unter der Voraussetzung einer dauerhaft nichtnachweisbaren Virusreplikation ist Stillen durch HIV-positive Mütter möglich. Die HIV-Infektion der Mutter ist nicht in das Vorsorgeheft des Kindes einzutragen. Bei Kindern HIV-positiver Mütter ist der HIV-Antikörpertest im ersten Lebensjahr immer positiv und nicht aussagekräftig. Die HIV-Infektion bei Kindern sollte so früh wie möglich behandelt werden. Bei Kindern unter 6 Jahren kann die Einnahme von Flüssigarzneimitteln wegen deren schlechten Geschmacks schwierig sein. Kinder sollten vor der Pubertät über ihre Diagnose aufgeklärt werden. Von Kindern, die HIV-positiv sind, geht keine Ansteckungsgefahr aus. Sie können uneingeschränkt alle Gemeinschaftseinrichtungen besuchen. Einhaltung ethischer Richtlinien Interessenkonflikt C. Königs gibt an, ein Vortragshonorar von MSD erhalten zu haben. U. Baumann und U. Schulze Sturm geben an, dass kein Interessenkonflikt besteht. Für diesen Beitrag wurden von den Autor/-innen keine Studien an Menschen oder Tieren durchgeführt. Bei den hier aufgeführten GEPIC-Daten handelt es sich um Originaldaten, erhoben aus pseudonymisierten Registerdaten. Damit wurden in diesem Beitrag Daten verarbeitet, welche im Zusammenhang mit der Behandlung von Menschen erhoben wurden. Für die erhobenen Daten gelten die Genehmigungen der jeweiligen Ethikkommissionen. QR-Code scannen & Beitrag online lesen ==== Refs Literatur 1. Asad H Collins IJ Goodall RL Crichton S Hill T Doerholt K Mortality and AIDS-defining events among young people following transition from paediatric to adult HIV care in the UK HIV Med 2021 22 8 631 642 10.1111/hiv.13096 33939876 2. Aurpibul L Puthanakit T Lee B Mangklabruks A Sirisanthana T Sirisanthana V Lipodystrophy and metabolic changes in HIV-infected children on non-nucleoside reverse transcriptase inhibitor-based antiretroviral therapy Antivir Ther 2007 12 8 1247 1254 10.1177/135965350701200811 18240864 3. Bourgi K Jenkins CA Rebeiro PF Palella F Moore RD Altoff KN Weight gain among treatment-naïve persons with HIV starting integrase inhibitors compared to non-nucleoside reverse transcriptase inhibitors or protease inhibitors in a large observational cohort in the United States and Canada J Int AIDS Soc 2020 23 4 e25484 10.1002/jia2.25484 32294337 4. GEPIC (2021). Entwicklung von Gewicht und BMI HIV-positiver Kinder und Jugendlicher nach Therapiewechsel. Deutsch-österreichischer AIDS-Konferenz, 26. März 2021 5. Braun M Situation HIV-infizierter Kinder und junger Erwachsener in Deutschland 2022 Frankfurt Universitätsklinikum der Goethe-Universität 6. Caldwell M Oxtoby MJ Simonds RJ 1994 revised classification system for human immunodeficiency virus infection in children less than 13 years of age; official authorized addenda: human immunodeficiency virus infection codes and official guidelines for coding and reporting ICD-9-CM MMWR Morb Mortal Wkly Rep 1994 43 12 1 19 7. Cooper ER Hanson C Diaz C Mendez H Abboud R Nugent R Encephalopathy and progression of human immunodeficiency virus disease in a cohort of children with perinatally acquired human immunodeficiency virus infection. Women and Infants Transmission Study Group J Pediatr 1998 132 5 808 812 10.1016/S0022-3476(98)70308-7 9602190 8. Deutsche AIDS-Gesellschaft. Deutsch-Österreichische Leitlinien Zur Antiretroviralen Therapie der HIV-Infektion bei Kindern und Jugendlichen. AWMF Leitlinie 2019, Registernummer 048-011. 9. Deutsche AIDS-Gesellschaft e. V. Deutsch-Österreichische Leitlinien Zur HIV-Therapie in der Schwangerschaft und bei HIV-exponierten Neugeborenen. AWMF Leitlinie 2020, Registernummer 055-002. 10. Diaz C Hanson C Cooper ER Read JS Watson J Mendez HA Disease progression in a cohort of infants with vertically acquired HIV infection observed from birth: the women and Infants transmission study (WITS) J Acquir Immune Defic Syndr Hum Retrovirol 1998 18 3 221 228 10.1097/00042560-199807010-00004 9665498 11. Evans C Humphrey JH Ntozini R Prendergast AJ HIV-exposed uninfected infants in Zimbabwe: insights into health outcomes in the pre-antiretroviral therapy era Front Immunol 2016 7 190 10.3389/fimmu.2016.00190 27375613 12. Flynn PM Taha TE Cababasay M Fowler MG Mofenson LM Owor M Prevention of HIV-1 transmission through breastfeeding: efficacy and safety of maternal antiretroviral therapy versus infant nevirapine prophylaxis for duration of breastfeeding in HIV-1-infected women with high CD4 cell count (IMPAACT PROMISE): a randomized, open-label, clinical trial J Acquir Immune Defic Syndr 2018 77 4 383 392 10.1097/QAI.0000000000001612 29239901 13. Githinji LN Gray DM Zar HJ Lung function in HIV-infected children and adolescents Pneumonia 2018 10 6 10.1186/s41479-018-0050-9 29984134 14. Goetghebuer T Smolen KK Adler C Das J McBride T Smits G Initiation of antiretroviral therapy before pregnancy reduces the risk of infection-related hospitalization in human immunodeficiency virus-exposed uninfected infants born in a high-income country Clin Infect Dis 2019 68 7 1193 1203 10.1093/cid/ciy673 30215689 15. Haberl L Audebert F Feiterna-Sperling C Gillor D Jakubowski P Jonsson-Oldenbüttel C Not recommended, but done: breastfeeding with HIV in Germany AIDS Patient Care STDS 2021 35 2 33 38 10.1089/apc.2020.0223 33571048 16. Judd A Chappell E Turkova A Le Coeur S Noguera-Julian A Goetghebuer T Long-term trends in mortality and AIDS-defining events after combination ART initiation among children and adolescents with perinatal HIV infection in 17 middle- and high-income countries in Europe and Thailand: A cohort study PLoS Med 2018 15 1 e1002491 10.1371/journal.pmed.1002491 29381702 17. Judd A Le Prevost M Melvin D Arenas-Pinto A Parrott F Winston A Cognitive function in young persons with and without perinatal HIV in the AALPHI cohort in england: role of non-HIV-related factors Clin Infect Dis 2016 63 10 1380 1387 10.1093/cid/ciw568 27581764 18. Labuda SM Huo Y Kacanek D Patel K Huybrechts K Jao J Rates of hospitalization and infection-related hospitalization among human Immunodeficiency virus (HIV)-exposed uninfected children compared to HIV-unexposed uninfected children in the United States, 2007–2016 Clin Infect Dis 2020 71 2 332 339 10.1093/cid/ciz820 31504291 19. Mandelbrot L Tubiana R Le Chenadec J Dollfus C Faye A Pannier E No perinatal HIV-1 transmission from women with effective antiretroviral therapy starting before conception Clin Infect Dis 2015 61 11 1715 1725 26197844 20. Marcus U Beck N Infektionen mit dem humanen Immundefizienzvirus bei Kindern in Deutschland, 1999–2016 Monatsschr Kinderheilkd 2022 170 5 403 411 10.1007/s00112-020-00865-4 21. Marcus U HIV infections and HIV testing during pregnancy, Germany, 1993 to 2016 Eur Surv Bull 2019 24 48 1 10 22. Niehues T Baumann U Buchholz B Dünsch D Edelhäuser M Funk M Empfehlungen zur antiretroviralen Therapie bei HIV-infizierten Kindern 2006 – Vollständig überarbeitetes und aktualisiertes Konsensus-Statement der Pädiatrischen Arbeitsgemeinschaft AIDS (PAAD) und der Deutschen Gesellschaft für Pädiatrische Infektiologie (DGPI) Monatsschr Kinderheilkunde 2022 154 6 563 573 23. Rosso R Vignolo M Parodi A Di Biagio A Sormani MP Bassetti M Bone quality in perinatally HIV-infected children: role of age, sex, growth, HIV infection, and antiretroviral therapy Aids Res Hum Retroviruses 2005 21 11 927 932 10.1089/aid.2005.21.927 16386108 24. Traisathit P Urien S Le Coeur S Srirojana S Akarathum N Kanjanavanit S Impact of antiretroviral treatment on height evolution of HIV infected children BMC Pediatr 2019 19 1 287 10.1186/s12887-019-1663-8 31421667 25. van Wyhe KS Laughton B Cotton MF Meintjes EM van der Kouwe A Boivin MJ Cognitive outcomes at ages seven and nine years in South African children from the children with HIV early antiretroviral (CHER) trial: a longitudinal investigation J Int AIDS Soc 2021 24 7 e25734 10.1002/jia2.25734 34259393 26. HIV/Aids: Erste Ziele in Deutschland erreicht. https://unric.org/de/hiv03122021/. Zugegriffen: 22. Juli 2022 27. Wagner N Crisinel P Kahlert C Martinez De Tejada B Breastfeeding for HIV-positive mothers in Switzerland: are we ready to discuss? Rev Med Suisse 2020 16 712 2050 2054 33112519 28. Weber V Radeloff D Reimers B Salzmann-Manrique E Bader P Schwabe D Neurocognitive development in HIV-positive children is correlated with plasma viral loads in early childhood Medicine 2017 96 23 e6867 10.1097/MD.0000000000006867 28591025 29. Wedderburn CJ Weldon E Bertran-Cobo C Rehman AM Stein DJ Gibb DM Early neurodevelopment of HIV-exposed uninfected children in the era of antiretroviral therapy: a systematic review and meta-analysis Lancet Child Adolesc Health 2022 6 6 393 408 10.1016/S2352-4642(22)00071-2 35483380 30. Zicari S Sessa L Cotugno N Ruggiero A Morrocchi E Concato C Immune activation, inflammation, and non-AIDS co-morbidities in HIV-infected patients under long-term ART Viruses 2019 11 3 1 19 10.3390/v11030200

0

PMC9748901

NO-CC CODE

2022-12-15 23:22:48

no

Gyn&#x000e4;kologie. 2022 Dec 14;:1-11

utf-8

oa_other

==== Front Support Care Cancer Support Care Cancer Supportive Care in Cancer 0941-4355 1433-7339 Springer Berlin Heidelberg Berlin/Heidelberg 7457 10.1007/s00520-022-07457-w Research A randomized, double-blinded feasibility trial of educational materials for hiccups in chemotherapy-treated patients with cancer Ehret Christopher J. 1 Le-Rademacher Jennifer 12 Storandt Michael H. 3 Martin Nichole 1 Rajotia Arush 1 Jatoi Aminah jatoi.aminah@mayo.edu 1 1 grid.66875.3a 0000 0004 0459 167X Department of Oncology, Mayo Clinic, 200 First Street SW, Rochester, MN USA 2 grid.66875.3a 0000 0004 0459 167X Division of Clinical Trials and Biostatistics, Mayo Clinic, Rochester, MN USA 3 grid.66875.3a 0000 0004 0459 167X Department of Medicine, Mayo Clinic, Rochester, MN USA 14 12 2022 2023 31 1 3026 5 2022 12 11 2022 © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Purpose Chemotherapy can cause hiccups but few randomized controlled trials have focused on hiccups. This trial examined the feasibility of such research. Methods This single-institution, multi-site trial used phone recruitment for patients: (1) 18 years or older, (2) able to speak/read English, (3) with a working e-mail address, (4) with hiccups 4 weeks prior to contact, and (5) with ongoing oxaliplatin or cisplatin chemotherapy. The primary outcome was feasibility. Patients were randomly assigned to one of two sets of educational materials, each of which discussed hiccups and palliative options. The experimental materials were almost identical to the standard materials but provided updated content based on the published medical literature. At 2 weeks, patients responded by phone to a 5-item verbally administered questionnaire. Results This trial achieved its primary endpoint of recruiting 20 eligible patients within 5 months; 50 patients were recruited in 3 months. Among the 40 patients who completed the follow-up questionnaire, no statistically significant differences between arms were observed in hiccup incidence since initial contact, time spent reviewing the educational materials, and the troubling nature of hiccups. Twenty-five patients tried palliative interventions (13 in the experimental arm and 12 in the standard arm), most commonly drinking water or holding one’s breath. Eleven and 10 patients, respectively, described hiccup relief after such an intervention. Conclusions Clinical trials for chemotherapy-induced hiccups are feasible and could address an unmet need. Keywords Trial Hiccups Education Randomized Feasibility AJ is the Betty J. Foust, M.D. and Parents' Professor of Oncology00000 Jatoi Aminah issue-copyright-statement© Springer-Verlag GmbH Germany, part of Springer Nature 2023 ==== Body pmcIntroduction Commonly prescribed chemotherapy drugs, such as oxaliplatin and cisplatin, appear to cause hiccups, particularly when administered with dexamethasone [1–3]. Chemotherapy-induced hiccups are estimated to occur in 15–40% of patients [2, 4]. Although typically self-limited, hiccups can recur and thereby generate a cumulative burden of symptomatology [5]. These recurrent episodes of hiccups can be particularly concerning because each bout puts patients at risk for sleep deprivation, fatigue, aspiration, and other related complications [5]. Despite their problematic nature, hiccups are understudied. To our knowledge, only 3 published randomized, controlled trials—the gold standard for testing the efficacy of an intervention—have focused on hiccup palliation specifically [6–8]. First, Zhang and others conducted a randomized, 30-patient placebo-controlled trial in stroke patients and reported that baclofen resulted in a statistically significant reduction in hiccups [6]. Similar favorable findings from Ramirez and others were reported in a 4-patient, double-blinded, placebo-controlled, crossover study with baclofen [7]. Finally, Wang and Wang conducted a placebo-controlled trial with metoclopramide in 36 patients who had intractable hiccups, reporting that this agent resulted in a statistically significant reduction in hiccups [8]. Although the above three trials are clinically important, they underscore the limitations of research on hiccups. First, with respective sample sizes of 30, 4, and 36 patients, these randomized trials are perhaps too small to provide definitive conclusions. Second, these trials point to a paucity of research specifically related to chemotherapy-induced hiccups, as none of these trials addressed this cyclical cause of hiccups. Third, to our knowledge, these are the only randomized trials to have been conducted for hiccups over the span of decades. Lastly, although recent observational studies have suggested that interventions, such as a forced inspiratory suction tool and a novel sour-tasting oral lollipop formulation might palliate hiccups, these other remedies have yet to be tested within the context of the gold standard of a randomized, double-blinded trial [9, 10]. The trial reported here was conducted to assess the feasibility of a randomized, double-blinded trial for hiccups in chemotherapy-treated patients with cancer. This trial tested a standard set of educational materials (control arm) versus an updated set of educational materials (experimental arm) [9, 10]. The demonstration of feasibility of conducting clinical trials for hiccups could conceivably lead to more research, the rigorous testing of newer novel interventions, and palliation for cancer patients who suffer from recurrent chemotherapy-induced hiccups. Methods Trial design and oversight The Mayo Clinic Institutional Review Board (IRB) approved this trial (#21–012,057), which relied on the Mayo Clinic’s multi-site catchment area of Minnesota, Arizona, Florida, Wisconsin, and Iowa. This trial was registered on www.clinicatrials.gov (NCT05313685). The trial used patient phone contact and electronic communication for recruitment and capture of endpoint data. Eligibility criteria Patient eligibility criteria were comprised of the following: (1) 18 years or older, (2) able to speak and read in English, (3) has an e-mail address, (4) patient-reported hiccups 4 weeks prior to patient contact, and (5) planned ongoing chemotherapy with either oxaliplatin or cisplatin. The rationale for this last criterion is the assumption that chemotherapy-induced hiccups would recur with ongoing chemotherapy (Fig. 1).Fig. 1 Consort diagram. The consort diagram shows how eligibility criteria gave rise to the cohorts and shows dropouts per arm Recruitment The electronic medical record was searched every weekday to identify potentially eligible patients, who were then contacted by phone with a caller identification that showed Mayo Clinic as the caller. Upon phone contact, an IRB-approved script was used to describe the trial to each potential enrollee. If the patient expressed interest, a study team member confirmed eligibility, including patient-reported hiccups within the required interval, obtained oral consent, and then acquired the patient’s signature on a Health Insurance Portability Accountability Act (HIPAA) form either electronically or via regular mail. Educational intervention, arm assignment, double-blinding, and acquisition of patient-reported data The interventions were two different sets of educational materials that consisted of 5–6 pages of double-spaced content that explained hiccup pathophysiology, causes, risk factors, complications, and palliation (prescribed medical and interventional therapy as well home remedies) [11]. Each educational intervention was provided to patients electronically as a Portable Document Format (PDF) document. Patients were assigned one of two versions: a publically available, Mayo Clinic-written document which was considered the standard intervention versus the same document, which was considered the experimental intervention and included the following wording of updated information on more recent interventions for hiccups [9–11]:Just about everyone has had hiccups, but some people get them more often and are more troubled by them. If you feel hiccups have been difficult for you, it would be important for you to contact your team of healthcare providers to let them know. This document provides educational information on hiccups. A few recent updates that you could try at home include the following. A recent large study suggested that interventions, such as drinking from a straw may help with hiccups. This has not been proven for sure. Some smaller studies and reports suggest that trying a sour food, such as sucking on a piece of lemon, might help hiccups. This has not been proven for sure. Participants were randomly assigned to one of these two interventions. A computer-generated list of random numbers was used to determine arm allocation. Based on the randomization code, each PDF file was assigned a number, which, in turn, was assigned consecutively to each enrolled patient. Each PDF file was constructed and labeled to make it impossible to discern arm allocation; the study team member, who recruited patients, who e-mailed the PDF file to patients, and who acquired secondary outcome measures, remained blinded throughout all patient interactions. Similarly, in view of the nature of the intervention, each patient was unable to discern arm allocation throughout the trial. Patients were then called by phone within 14 days of receipt of their allocated educational materials. This interval was mandated per protocol based on previous data that suggest optimal patient recall of symptoms within this time frame [12]. Patients were then asked to complete a verbally administered questionnaire to assess the secondary endpoints, as described in further detail below. The study team attempted to contact patients shortly after they had received chemotherapy, as chemotherapy was presumed to be the cause of the hiccups. Outcome measures The primary outcome was feasibility. Based on review of the published literature and institutional patient volumes, the study team defined enrollment of 20 eligible patients within 5 months of trial initiation as success and thereafter allowed for enrollment of up to 50 total patients [6–8]. Enrollment was defined as the acquisition of oral consent and a patient signature on a submitted HIPAA document. Secondary endpoints were derived from a 5-item questionnaire, which had undergone multiple iterations for face validity by study team members and others and which relied, in part, on a numeric rating scale [13–15]: 1) “Have you had hiccups since you signed up for this study?,” which required a “yes” or “no” response; (2) “How much time did you spend reviewing the educational materials on hiccups?,” with 5 response options (1 = little to no time to 5 = a lot of time), as per a numeric rating scale; (3) “How troubling were your hiccups since you signed up for this study?,” with 5 response options (1 = little to no trouble to 5 = severe hiccups); (4) “Since you signed up for this study, what did you try to stop or prevent hiccups?” with choices of “nothing,” “drinking from a straw,” “tasting something sour,” “medication from a healthcare provider” (to be specified), and “other” (to be specified); and (5) “Since you signed up for the study, what helped stop or prevent your hiccups the most?” (choices were the same as question 4). Patients provided phone responses, which an investigator recorded on case report forms. Descriptive patient demographics were also acquired during phone encounters and during medical record review. Analyses The primary endpoint was feasibility, as defined above. Because 30% of patients are likely to report symptoms as less troubling based on a placebo effect, report of less troubling symptoms in 70% of patients was considered clinically meaningful in the experimental arm and therefore worthy of further testing of these educational materials in a larger trial [16, 17]. Fifty patients enabled detection of a type 1 error of < 0.05 with 80% power. Questionnaire numeric response categories 1–2 and 3–5 were combined for questions 2 and 3, and a 2-tailed Fisher’s exact test was used to assess differences between arms. A p-value < 0.05 was considered statistically significant. Results Demographics A total of 50 patients were enrolled. The median age was 63 years (range: 24, 85 years). Thirty-eight patients (76%) were men. All received concomitant dexamethasone. Demographics as per arm appear in Table 1.Table 1 Demographics; n = 50 DEMOGRAPHIC Number of patients with EXPERIMENTAL educational materials n = 25 (%) Number of patients with REGULAR educational materials n = 25 (%) Median age in years (range) 68 (24, 85) 61 (36, 80) Sex Male 17 (68) 21 (84) Female 8 (32) 4 (16) Cancer type Colorectal 7 (28) 6 (24) Genitourinary 6 (24) 2 (8) Pancreatic 1 (4) 6 (24) Gastroesophageal 2 (8) 4 (16) Head and neck 2 (8) 3 (12) Cholangiocarcinoma 3 (12) 2 (8) Lung 2 (8) 1 (4) Gynecological 1 (4) 1 (4) Other 1 (4) 0 Chemotherapy** Oxaliplatin + other chemotherapy agents 12 (48) 16 (64) Cisplatin + other chemotherapy agents 11 (44) 5 (20) Cisplatin (single agent) 2 (8) 4 (16) *Numbers in parentheses refer to percentages unless otherwise specified **All patients received concomitant dexamethasone Primary endpoint The first 20 patients were enrolled within 24 days. All 50 patients were enrolled within 3 months (Fig. 2).Fig. 2 Patient recruitment over time. The first 20 eligible patients were recruited within the first month of the trial. All 50 eligible patients were recruited within 3 months Secondary endpoints Among the 40 patients who completed the follow-up questionnaire, per the study protocol, no statistically significant differences in hiccup incidence were observed between arms. Similarly, no statistically significant differences were observed between arms in time spent reviewing the educational materials and in the troubling nature of hiccups (Table 2).Table 2 Secondary patient-reported outcomes QUESTION EXPERIMENTAL educational materials n = 19 (%) REGULAR educational materials n = 21 (%) p-value Have you had hiccups since you signed up for this study?   Yes 16 (84) 17 (81) 0.79   No 3 (16) 4 (19) How much time did you spend reviewing the educational materials on hiccups?   Little to no time (categories 1 and 2) 13 (68) 9 (42) 0.10   A lot of time (categories 3, 4, and 5) 6 (32) 12 (57) How troubling were your hiccups since you signed up for this study?   Little to no trouble (categories 1 and 2) 15 (79) 17 (81) 0.94   Severe hiccups (categories 3, 4, and 5) 4 (21) 4 (19) A subgroup of patients reported their hiccups as severe. Among patients assigned to the experimental arm, one patient rated his hiccup severity with a 5, or “severe hiccups.” Three in the experimental arm provided a rating of 3 for hiccup trouble/severity. In the standard arm, 2 patients rated their hiccup trouble/severity with a 5, one with a 4, and one with a 3. For hiccup palliation, 25 patients tried various interventions, (namely, 13 in the experimental arm and 12 in the standard arm). Some of these interventions had been described in the educational materials (Table 3). Eleven and 10 patients, respectively, described hiccup relief with the tried interventions.Table 3 Patient-attempted interventions for hiccup palliation Trial arm 1 = experimental 2 = regular* Interventions attempted Did the intervention(s) help? 1 Small sips of water with no straw Yes 1 Breathing exercises Yes 1 Standing upright, eating ice, drinking water with no straw No 1 Slow eating Yes 1 Eating slowly Yes 1 Holding one’s breath, taking a deep breath, walking Yes, walking 1 Holding one’s breath Yes 1 Breathing exercises Yes 1 Gabapentin, olanzapine, holding breath Yes, to olanzapine 1 sips of water No 1 Baclofen, eating peanut butter, drinking water with one’s head back Yes, to baclofen 1 Tasting something sour, baclofen, chlorpromazine, positional drinking, exercising (pushups), holding one’s breath, breathing into a bag Yes, to exercise (pushups) 1 Drinking water without a straw Yes 2 Drinking water with no straw and holding breath No 2 Drinking water with no straw and taking peppermint Yes 2 Sipping water with no straw, placing head between legs No 2 Limiting carbonated beverages, eating more slowly Yes, to limiting carbonated beverages 2 Drinking from a straw Yes 2 Ondansetron, prochlorperazine Yes 2 Holding one’s breath, focusing on something else Yes 2 Drinking from a straw, holding one’s breath Yes, to drinking from a straw 2 Drinking from a straw, holding one’s breath Yes, to holding one’s breath 2 Holding one’s breath yes 2 Prilosec, prochlorperazine, positional changes, drinking water Yes, to prochlorperazine, and positional changes 2 Drinking from a straw, chewing gum Yes, to chewing gum *Each row denotes the response from a specific patient; the other 15 patients responded with “nothing” with respect to trying an intervention Discussion This 50-patient clinical trial is perhaps the largest, completed randomized double-blinded trial that focuses on hiccup palliation [6–8]. Of note, Go and others conducted a 65-patient trial on rotational corticosteroids, but that 14-site trial appeared to be aimed more at understanding whether dexamethasone was truly a cause of hiccups (“to prove the inequality of hiccup intensity”) as opposed to testing palliative interventions for hiccups [18]. The current single-institution trial did in fact achieve its primary endpoint of feasibility, thereby demonstrating that clinical trials for hiccup palliation can be conducted expeditiously, particularly in patients with cancer. Although the secondary trial results did not suggest that the experimental educational materials for hiccups are more effective at palliation than the standard educational materials and therefore not worthy of further testing, some patients did resort to the interventions in their assigned educational materials, albeit with mixed palliative success. Nonetheless, the important lessons gleaned from the current trial are that chemotherapy-treated patients experience hiccups, most appear to develop recurrent hiccups after further chemotherapy, some report their hiccups to be severe, and a notable proportion are willing to participate in a clinical trial that focuses on hiccup palliation. Taken together, these findings underscore the feasibility—as well as perhaps the need—for further research to palliate hiccups in chemotherapy-treated patients with cancer. One can only speculate on reasons for this paucity of research on hiccups. The assumption that everyone—including babies in utero—experience hiccups makes this symptomatology less likely to be viewed as pathologic or notably troubling both by patients and healthcare providers. Additionally, the cyclical nature of hiccups, which seem to occur shortly after chemotherapy, might result in patients’ forgetting to discuss their hiccups 2–4 weeks later at an upcoming clinic visit, particularly if other more pressing matters, such as cancer status, are to be discussed. Although hiccups might not be patients’ gravest issue in the setting of cancer, this trial does suggest hiccups are bothersome—particularly for some—to the point that patients are willing to enroll in research aimed at their palliation. This feasibility trial provides at least six lessons that might guide future research. First, it is noteworthy that patients with less troubling hiccups (graded as 1–2 of 5) were interested in enrolling in this trial. Although these patients’ hiccups might not be as distressing, perhaps the inconvenience of hiccups might sway patients with even milder symptoms to consider trial participation. Thus, future trials might consider the implementation of broad eligibility criteria. Second, this apparent trivialization of hiccups on the part of both patients and healthcare providers suggests a need to contact patients, ask about hiccups, and, if present, offer trial enrollment. This direct approach, which did not rely on referral from healthcare providers or on medical record review for hiccup documentation, likely accounts for the success of the current trial and should be considered for future hiccup trials. Third, importantly, most patients—over 80%—who reported hiccups at baseline and were therefore eligible for the trial, did in fact experience recurrent hiccups. This finding suggests that, in patients who are receiving chemotherapy, prior hiccups are a predictor of future hiccups and should be considered an eligibility criterion for future trials. Fourth, despite a short follow-up of 2 weeks, the dropout rate in this trial is high at 20%. The fact that these patients have cancer and are therefore at risk for other emerging issues—such as being diagnosed with symptomatic COVID-19, as occurred in one patient—speaks to the challenges of conducting this research in ill patients. In this context, future clinical trials should plan for high dropout rates of 20% of greater. Fifth, if such educational materials are used in future research, it might be important to gain a better sense of whether patients are actually reading the materials and implementing what they have learned from such materials. Assessment materials that acquire such patient-reported information would be important. Furthermore, researchers might focus on the burden-related aspects of such materials—their word count and complexity of content—to make sure that such materials are truly easy for patients to read and understand. Finally, timing is important. In view of the episodic nature of hiccups after chemotherapy, future interventions should be readily available to patients, in the same manner as antiemetics. The success of an intervention would need to enable expeditious access to the intervention at the time hiccups start. Similarly, the transient nature of hiccups is such that the success of an intervention should be assessed within the context of a control arm, as tested here, as otherwise the fleeing nature of hiccups might be erroneously attributed to the success of an intervention. Importantly, it should be noted that future hiccup interventions, which are particularly complicated, difficult to access, or medications fraught with severe side effects, might not accrue patients as readily as the trial reported here. In summary, this trial demonstrates the feasibility of conducting clinical trials in chemotherapy-treated patients with hiccups or at risk for hiccups. This trial provides an important platform to facilitate further research of an understudied adverse event with the goal of lessening the suffering and morbidity of patients with cancer. Acknowledgements Dr. Jatoi is the Betty J. Foust, M.D. and Parents’ Professor; this endowment helped fund this research. Author contribution All the authors contributed equally to this research, helped formulate the study design, gather the data, analyze the data, and write the manuscript. Funding This work was funded by institutional funds. Data availability The data and material associated with this work will not be available due to confidentiality. Code availability N/A. Declarations Ethics approval This study was approved by the Mayo Clinic Institutional Review Board. Consent to participate All patients provided appropriate consent prior to participating in this research. Consent for publication All patients were aware of the intention to publish this research. Conflict of interest Dr. Jatoi had served as a consultant for Meterhealth. Competing interests The authors declare no competing interests. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Hosoya R Uesawa Y Ishii-Nozawa R Kagaya H Analysis of factors associated with hiccups based on the Japanese Adverse Drug Event Report database PLoS One 2017 12 2 e0172057 10.1371/journal.pone.0172057 28196104 2. Liaw CC Wang CH Chang HK Wang HM Huang JS Lin YC Chen JS Cisplatin-related hiccups: male predominance, induction by dexamethasone, and protection against nausea and vomiting J Pain Symptom Manage 2005 30 4 359 366 10.1016/j.jpainsymman.2005.08.008 16256900 3. Ehret CJ, Le-Rademacher JG, Martin N, Jatoi A (2021) Dexamethasone and hiccups: a 2000-patient, telephone-based study. BMJ Support Palliat Care bmjspcare-2021–003474. 10.1136/bmjspcare-2021-003474 4. Ehret C Young C Ellefson CJ Aase LA Jatoi A Frequency and symptomatology of hiccups in patients with cancer: using an on-line medical community to better understand the patient experience Am J Hosp Palliat Care 2022 39 2 147 151 10.1177/10499091211006923 33792359 5. Hendrix K Wilson D Kievman MJ Jatoi A Perspectives on the medical, uality of life, and economic consequences of hiccups Curr Oncol Rep 2019 21 12 113 10.1007/s11912-019-0857-4 31858286 6. Zhang C Zhang R Zhang S Xu M Zhang S Baclofen for stroke patients with persistent hiccups: a randomized, double-blind, placebo-controlled trial Trials 2014 22 15 295 10.1186/1745-6215-15-295 7. Ramírez FC Graham DY Treatment of intractable hiccup with baclofen: results of a double-blind randomized, controlled, cross-over study Am J Gastroenterol 1992 87 12 1789 1791 1449142 8. Wang T Wang D Metoclopramide for patients with intractable hiccups: a multicentre, randomised, controlled pilot study Intern Med J 2014 44 12a 1205 1209 10.1111/imj.12542 25069531 9. Alvarez J Anderson JM Snyder PL Mirahmadizadeh A Godoy DA Fox M Seifi A Evaluation of the forced inspiratory suction and swallow tool to stop hiccups JAMA Netw Open 2021 4 6 e2113933 10.1001/jamanetworkopen.2021.13933 34143196 10. Ehret CJ, Jatoi A (2022) Establishing the groundwork for clinical trials with Hiccupops® for hiccup palliation. Am J Hosp Palliat Care 39(10):1210–1214. 10.1177/10499091211063821 11. https://www.mayoclinic.org/diseases-conditions/hiccups/diagnosis-treatment/drc-20352618. Accessed 18 Apr 2022 12. Lawson A Tan AC Naylor J Harris IA Is retrospective assessment of health-related quality of life valid? BMC Musculoskelet Disord 2020 21 1 415 10.1186/s12891-020-03434-8 32605559 13. Yesilyurt M Faydali S Evaluation of patients using numeric pain-rating scales Int J Caring Sci 2021 14 890 897 14. Hjermstad MJ Fayers PM Haugen DF Studies comparing numerical rating scales, verbal rating scales, and visual analogue scales for assessment of pain intensity in adults: a systematic literature review J Pain Symptom Manage 2011 41 1073 1093 10.1016/j.jpainsymman.2010.08.016 21621130 15. Lee MK Schalet BD Cella D Establishing a common metric for patient-reported outcomes in cancer patients: linking patient reported outcomes measurement information system (PROMIS), numerical rating scale, and patient-reported outcomes version of the common terminology criteria for adverse events (PRO-CTCAE) J Patient Rep Outcomes 2020 4 106 10.1186/s41687-020-00271-0 33305344 16. Junior PNA Barreto CMN Cubero DIG del Giglio A The efficacy of placebo for the treatment of cancer-related fatigue: a systematic review and meta-analysis Support Care Cancer 2020 28 1755 1764 10.1007/s00520-019-04977-w 31302766 17. Chvetozoff G Tannock IF Placebo effects in oncology J National Cancer Inst 2003 95 19 29 10.1093/jnci/95.1.19 18. Go SI Koo DH Kim ST Song HN Kim RB Jang JS Oh SY Lee KH Lee SI Kim SG Park LC Lee SC Park BB Ji JH Yi SY Lee YG Yun J Bruera E Hwang IG Kang JH Antiemetic corticosteroid rotation from dexamethasone to methylprednisolone to prevent dexamethasone-induced hiccup in cancer patients treated with chemotherapy a randomized, single-blind, crossover phase III trial Oncologist 2017 22 11 1354 1361 10.1634/theoncologist.2017-0129 28687626

36515742

PMC9748902

NO-CC CODE

2022-12-15 23:22:48

no

Support Care Cancer. 2023 Dec 14; 31(1):30

utf-8

Support Care Cancer

10.1007/s00520-022-07457-w

oa_other

==== Front Curr Psychol Curr Psychol Current Psychology (New Brunswick, N.j.) 1046-1310 1936-4733 Springer US New York 4092 10.1007/s12144-022-04092-w Article A four-level meta-analytic review of the relationship between social media and well-being: a fresh perspective in the context of COVID-19 http://orcid.org/0000-0003-3858-8391 Wong Joax joax.wong.2019@socsc.smu.edu.sg Yi Poh Xin Quek Frosch Y. X. Lua Verity Y. Q. Majeed Nadyanna M. Hartanto Andree andreeh@smu.edu.sg grid.412634.6 0000 0001 0697 8112 School of Social Sciences, Singapore Management University, 90 Stamford Road, Level 4, Singapore, 178903 Singapore 14 12 2022 115 28 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Social media, one of the most pervasive forms of technology, has been widely studied in relation to the mental health and well-being of individuals. However, the current literature on social media and well-being has provided mixed and inconclusive findings, thus creating a polarizing view of social media. These mixed findings continue to extend into the pandemic, with researchers debating over the effects of social media in the new norms of social isolation. In light of these inconclusive findings, the aim of our meta-analysis was to synthesize previous research data in order to have a holistic understanding of the association between social media and well-being, particularly in the present context of COVID-19. The current meta-analysis systematically investigated 155 effect sizes from 42 samples drawn from 38 studies published during the COVID-19 pandemic (N = 43,387) and examined the potential moderators in the relationship between social media and well-being, such as the different operationalizations of social media usage and demographics. Overall, our study found that the relationship between social media usage and well-being was not significant in the context of COVID-19. Additionally, the impact of various moderators on the relationship between social media and well-being was found to vary. We discuss the various theoretical, methodological and practical implications of these findings and highlight areas where further research is necessary to shed light on the complex and nuanced relationship between social media and well-being. Keywords Social media Well-being Meta-analysis Covid-19 ==== Body pmcSince its inception in the 1990s, social media has been adopted by more than half of the 7.7 billion people in the world, and the number is projected to continue increasing (GlobalWebIndex, 2015; Pew Research Center, 2021; Statista Research Department, 2021b). Indeed, in the last decade alone, social media platforms have almost tripled their total user base, from 970 million in 2010 to more than 4.72 billion users in 2021 (GlobalWebIndex, 2015; Kemp, 2021; Pew Research Center, 2021; Statista Research Department, 2020). Social media, which was traditionally only used as a simple communication platform, has now evolved into a crucial platform for the creation and transmission of user-generated content, amongst various other uses (Hunter, 2020; Kaplan & Haenlein, 2010; Kim & Johnson, 2016; Lardo et al., 2017). In light of its pervasiveness throughout society, laypersons and researchers alike have taken an interest in the psychological effects of social media (David et al., 2018; Perloff, 2014; Saiphoo & Vahedi, 2019; Vogel et al., 2015). In particular, one psychological implication of social media that has been widely studied is related to the mental health and well-being of individuals. Researchers have extensively covered the implications of social media on well-being, with many cross-sectional and correlational trend analyses finding a negative correlation between social media usage and well-being (Brunborg & Andreas, 2019; Ivie et al., 2020; Keles, et al., 2020; Lin et al., 2016; McPherson et al., 2006; Milani et al., 2009; Park et al., 2014). Interestingly, many meta-analytic studies on the relationship between social media usage and depressive symptoms reported small but significant effect sizes of r = 0.11 (Cunningham et al., 2021), r = 0.11 (Ivie et al., 2020) and r = 0.17 (Vahedi & Zannella, 2021) respectively. This suggests that the effect sizes are typically small but still significantly predict lower well-being. Indeed, recent meta-analytic studies have also reported that higher levels of social media usage were associated with lower psychological well-being (Huang, 2017), decreased self-esteem (Liu & Baumeister, 2016), increased loneliness (Song et al., 2014), as well as an increase in depressive symptoms (Cunningham et al., 2021; Ivie et al., 2020). Against this backdrop of a possible link between social media and depressive symptoms, researchers have even coined the term “Facebook depression” as a consequence of social media usage (O'Keeffe & Clarke-Pearson, 2011), further cementing the negative psychological impacts of social media on well-being. Several lines of reasoning and empirical evidence have been presented to explain why social media usage is associated with lower well-being. One explanation is that the strength of relationships forged through social media might be weaker due to lower quality conversations that lack depth (Kraut et al., 1998; Lee, 2009). This is because communication on social media takes place online, thus lacking the necessary human touch and quality needed to provide the same benefits as real-life interactions (Christensen, 2018; Lee et al., 2011; Reich et al., 2012; Yang et al., 2014). Indeed, studies have shown that face-to-face interactions with family and friends are associated with higher well-being (Adams et al., 2011; Sullivan, 1953) whereas communication via social media is associated with lower levels of well-being (Hunt et al., 2018; Lee, 2009; Newson et al., 2021). Additionally, researchers have also posited that social media might cause individuals to experience reduced social interaction (Hunt et al., 2018; Neto et al., 2015). This is also known as the “displacement hypothesis”, whereby increased time spent using social media reduces the time available for real-life interactions (Dunbar, 2016; Kraut et al., 1998). This view has been supported by several studies that have found a positive association between social media usage and loneliness (Hunt et al., 2018; Neto et al., 2015). Others argue that social media platforms (e.g., Instagram, Facebook) provide ample opportunities for peer comparison about self-presentation and image among individuals (Mascheroni et al., 2015). The facilitation of upward social comparison through social media can increase the psychological distress of individuals and result in stress, anxiety and lower levels of self-esteem (Chen & Lee, 2013; Feinstein et al., 2013; Fox & Vendemia, 2016). Lastly, other studies have also found that social media adversely affects well-being due to the high exposure of negative stimuli and experience such as derogatory content, cyberbullying, and unhealthy social comparison (Vogel et al., 2014; Whittaker & Kowalski, 2015). Despite the seemingly negative implications of social media, a smaller body of research has also shown that social media usage is associated with higher well-being (Bucci et al., 2019; Ellison et al., 2007; Stern, 2008; Naslund et al., 2016). For instance, the social compensation hypothesis posits that online communication will benefit people who are socially anxious and isolated as they may feel more at ease when developing friendships online in a safe environment (Barker, 2009; Ellison et al., 2007; Zywica & Danowski, 2008). Another line of argument stems from the stimulation hypothesis, which posits that online communication stimulates communication with existing friends, leading to mostly positive outcomes and stronger friendships overall (Nesi et al., 2018; Valkenburg & Peter, 2007). Indeed, studies have reported that social media usage was associated with an increase in social capital which is linked to positive psychological effects (Chan, 2015; Chen & Li, 2017; Nieminen et al., 2010). Consistent with this, an increase in self-disclosure, communication and friending via social media was also found to have a positive effect on psychological well-being (Chen & Li, 2017). Other studies also found that social media usage was associated with a higher quality of friendship (Ellison et al., 2007; Wang et al., 2019) and social support (Bucci et al., 2019; Naslund et al., 2016), which are both related to higher levels of well-being (Asante, 2012; Chu et al., 2010; Cuadros & Berger, 2016). Taken together, the association between social media and well-being still remains a polarizing issue due to the mixed findings of both positive and negative correlations. Several explanations have been hypothesized to explain the mixed findings regarding the link between social media and well-being. One of the plausible theories attributes the mixed findings to the duality of social media ― the freedom of expression and different functionalities on social media can create a myriad of potential harm and benefits to an individual (Baccarella et al., 2018; Pavlíček, 2013). Indeed, one researcher coined it “the social media see-saw” (Weinstein, 2018) suggesting that it is a balancing act between both positive and negative impacts of using social media and that social media does not consist of only one of these effects. Another reason for the mixed findings could be due to the different operationalizations of social media. For example, some studies operationalized social media usage as a coping mechanism (Eden et al., 2020; Teresa et al., 2021) while others did not (Alam et al., 2021; Riehm et al., 2020). Moreover, many studies utilised subjective and self-report measures instead of objective measures which might be a better indicator of social media usage (McCain & Campbell, 2018). Additionally, different studies also measured social media usage differently (i.e., number of times an individual checks their phone versus total time spent on using their phone). Indeed, the variation in operationalizations and methodologies across the present literature has made it difficult to reach a conclusion whether social media usage is positive or negative for an individual. While there has been extensive research on the topic of social media and well-being, the recent development of the COVID-19 pandemic has warranted a need to investigate how the dynamics of the relationship between social media and well-being has changed, as well as the potential benefits and detrimental effects of utilising social media during COVID-19. During this pandemic, staying at home has become the norm for individuals around the world (Engle et al., 2020). This period of social isolation has accelerated the rate of social media adoption which can be observed through the engagement and growth rates of social media around the world. For example, Tik Tok’s annual user growth rate in the US was 85.3% in 2020 alone (Statista Research Department, 2021a). Furthermore, the average time spent on social media by US users in 2020 also grew from 54 to 65 min per day (Statista Research Department, 2020). The global increase in social media usage during COVID-19 has been commonly attributed to the increasing reliance on social media as a form of coping against loneliness (Cauberghe et al., 2021). Several studies have shown that the pandemic has resulted in an increase in loneliness due to self-isolation and social distancing (Groarke et al., 2020; Killgore et al., 2020) and people are trying to find different ways of coping with it. As mentioned earlier, research conducted before the pandemic found that social media usage is associated with multiple benefits including increases in social capital and social support (Chen & Li, 2017; Wang et al., 2019). Consistent with this, studies conducted during the pandemic have reported that higher social media usage is negatively correlated with loneliness (Groarke et al., 2020). Therefore, people might be increasing their tendency to use social media as a coping mechanism against loneliness, which instead could make social media a potential protective factor that enables individuals to obtain the interaction and social support that they seek during the pandemic. Thus, we hypothesize that there will be a positive correlation between social media and well-being in the context of COVID-19. Additionally, we hypothesize that the tendency to use social media as a coping mechanism will be positively correlated with higher levels of well-being. Considering the high prevalence of social media, as well as the increased reliance on social media to cope with social isolation due to COVID-19, there is a need to re-investigate the link between social media and well-being. Although some research has been conducted in the context of COVID-19, many of those studies had mixed results (Fumagalli et al., 2021; Sun et al., 2020; Masciantonio et al., 2021). Therefore, there exists a need for a meta-analysis to test the conflicting hypotheses about the relationship between social media and well-being. A meta-analytic study can additionally provide the grounds for examining the potential moderators such as different operationalizations of social media and well-being. Additionally, the current meta-analysis aims to shed light on the individual differences in the tendency to use social media as a coping mechanism. Through the synthesis of previous research findings, we will be able to have a clearer picture of the potential boundary conditions of the harms and benefits of social media on well-being in the context of COVID-19. Method Search strategy The present meta-analysis focused on synthesizing findings on the association between social media and well-being in the context of COVID-19. Social media provides an online platform with multiple functionalities (e.g., communication, entertainment, media sharing) that may have an impact on the mental health of individuals. Therefore, the search strategy involved the review of the multiple measures of social media and their various impacts on different aspects of well-being. With that, a literature search was conducted in EBSCOhost ERIC, EBSCOhost PsycINFO, PubMed, Scopus, and Web of Science using the following keywords: ("Social Media" OR "Online Friend" OR "Online Friends" OR "Social Network" OR "Social Networking Sites" OR "Social Media Technology" OR "Online Community" OR Facebook OR Twitter OR Blog* OR Youtube OR Tumblr OR Discord OR Reddit OR Instagram OR Tiktok OR Snapchat OR Pinterest OR LinkedIn OR "Chat Room*" OR "Online Forum*") AND ("wellbeing" OR "well-being" OR "well being" OR "mental health" OR satisfaction OR happiness OR happy OR "positive affect" OR "negative affect" OR mood OR anxiet* OR anxious OR sadness OR "Cantril Ladder" OR lonel* OR self-esteem OR self-efficacy OR depress* OR self-worth OR "quality of life") AND (COVID-19 OR COVID19 OR "corona virus" OR coronavirus OR 2019-nCov OR SARS-CoV-2). Databases were searched for all reports available by June 2021. Furthermore, we conducted manual searches using the keywords “social media” AND COVID-19 in journals related to computers and new media, mental wellness, and mental disorders, namely: (1) Computers in Human Behaviour, (2) Cyberpsychology, Behavior, and Social Networking, (3) Journal of Affective Disorders, (4) Journal of Computer-Mediated Communication, (5) Journal of Mental Health, (6) New Media & Society, and (7) Social Indicators Research. To capture unpublished literature, manual searches were also conducted in ProQuest Dissertations & Theses and Google Scholar using the same keywords "social media" AND COVID-19. Inclusion criteria In total, the search resulted in 2591 potentially eligible records. After removing duplicates using the Mendeley Desktop version 1.19.4 (Mendeley, n.d.), a total of 1704 records were screened for inclusion based on titles and abstracts independently by the first and second author who then discussed and resolved discrepancies. Agreement between the two authors was good, with an average of 98.00% for the abstract screening stage. Based on abstracts, 592 irrelevant records were removed, leaving 1112 full-text records to be screened for inclusion based on the following criteria:Studies were included if they provided quantitative data and specified that the study was conducted during COVID-19), regardless of other methodological characteristics. Studies were also included regardless of their peer review status. Only the peer-reviewed version was kept if two versions of the same study were available (e.g., as part of a thesis and as part of a journal article). There were no restrictions on any sample characteristics such as age or gender. Studies were included as long as they subjectively or objectively measured and reported social media usage (i.e., duration of social media usage, number of times social media was accessed, frequency of social media usage). Subjective measures could be single- or multiple-item self-reports assessing any form of social media usage, while objective measures included screenshots of screen time usage or time spent on social media. Studies were included if they reported at least one measure of well-being. Well-being represents the presence of indicators of psychological adjustment such as life satisfaction or positive affect, and the absence of indicators of psychological maladjustment such as negative affect, or depression (Hartanto et al., 2021a, b; Houben et al., 2015). Common measures of well-being include, but are not limited to, the Satisfaction with Life Scale (Diener et al., 1985), the General Anxiety Disorder Scale (Spitzer et al., 2006), the Center for Epidemiological Studies-Depression Scale (Pinquart & Sörensen, 2003), and the UCLA Loneliness Scale (Russell, 1996). Studies were included if the information necessary to compute effect sizes were reported. If a study was eligible but did not report the appropriate statistics, the original authors of the study were contacted directly to obtain usable data. Out of the 37 authors contacted, 15 authors provided the requested data. The remaining 22 did not respond despite three repeated requests. Based on the examination of the potentially eligible full-text records, 38 records (38 studies) met all criteria and had sufficient data to compute effect sizes (Al-Qahtani et al., 2020; Alam et al., 2021; Aymerich-Franch, 2020; Bonsaksen et al., 2021; Boursier et al., 2020; Chakraborty et al., 2021; Chen et al., 2021a; Chen et al., 2020; Chen et al., 2021b; Chen et al., 2021c; Clavier et al., 2020; Cuara, 2020; Drouin et al., 2020; Eden et al., 2020; Ellis et al., 2020; Fernandes et al., 2020; Fumagalli et al., 2021; Hammad et al., 2021; Hikmah et al., 2020; Ikizer et al., 2021; Krause et al., 2021; Krendl et al., 2021; Lake, 2020; Lemenager et al., 2021; Lisitsa et al., 2020; Magson et al., 2021; Masciantonio et al., 2021; Patabendige et al., n.d.; Reiss et al., 2020; Rens et al., 2021; Riehm et al., 2020; Şentürk et al., 2021; Sewall et al., 2021; Sun et al., 2020; Teresa et al., 2021; Wheaton et al., 2021; Yang et al., 2021; Zhao & Zhou, 2021). 34 studies (89.47%) contributed one sample each, with the remaining 4 studies contributing multiple samples each, providing a total of 42 independent samples with a total unique N of 43,387 (Mdn = 603, M = 1058.22, SD = 1220.59, range = 46–6329). Based on available reports, the range of the mean age of the samples was 10.32–75.2 years (Mdn = 31.70, M = 28.98, SD = 13.29) with a median gender proportion of 67.70% female (M = 65.74%, SD = 15.17%). All eligible studies were conducted from 2020 to 2021, in 26 countries across five continents. The overall selection process of the studies to be included in this meta-analysis is demonstrated in the PRISMA flowchart in Fig. 1 (Moher et al., 2009).Fig. 1 PRISMA flowchart Data extraction The entire coding process was completed independently by the first and second authors who then discussed and resolved discrepancies after the initial coding process. Agreement was generally good, with an average of 99% (ranging from 98 to 100%) agreement between the two authors (see Table 1).Table 1 Agreement rates between coders Variable Trial (n = 7, m = 7, k = 16) Actual (n = 31, m = 35, k = 139) Total (n = 38, m = 42, k = 155) Critical statistical information   Total sample size, N 1.00 0.98 0.98   Type of effect size 1.00 1.00 1.00   Zero-order Pearson correlation, r 0.75 0.91 0.89   Odds ratio, OR 1.00 1.00 1.00   Does effect size need to be reversed? 1.00 1.00 1.00 Record and sample characteristics   Record type 1.00 1.00 1.00   Country 1.00 1.00 1.00   Sample source 1.00 1.00 1.00   Age (mean) 1.00 1.00 1.00   Age (range) 0.88 0.99 0.98   Age (group) 1.00 1.00 1.00   Female proportion 1.00 0.99 0.99 Additional information on social media and well-being   Temporal precedence 1.00 1.00 1.00   Well-being category 1.00 0.99 1.00   Well-being measure 1.00 1.00 1.00   Social media type (subjective/objective) 1.00 1.00 1.00   Social media measure (time spent/number of times accessed) 1.00 1.00 1.00   Social media platform (e.g., Facebook, Twitter) 1.00 1.00 1.00   Social media purpose (used as coping mechanism/non-coping) 0.81 1.00 1.00   Well-being (mean) 1.00 0.94 0.95   Social media (mean times accessed per day) 1.00 1.00 1.00   Social media (mean time spent per day) 1.00 0.99 0.99 Information on moderators   FOMO (mean) 1.00 1.00 1.00   Social media addiction (mean) 1.00 1.00 1.00   Excessive use of social media (mean) 1.00 0.99 0.99   Agreeableness (mean) 1.00 1.00 1.00   Conscientiousness (mean) 1.00 1.00 1.00   Extraversion (mean) 1.00 1.00 1.00   Neuroticism (mean) 1.00 1.00 1.00   Openness (mean) 1.00 1.00 1.00 Overall 0.98 1.00 0.99 n = number of studies, m = number of samples, k = number of effect sizes Firstly, we coded the zero-order Pearson correlation r or unadjusted odds ratio (OR) quantifying the association between social media usage and well-being. Thereafter, the following study characteristics were also coded: (a) publication source of the record (journal articles, unpublished data, dissertations, thesis, book chapters), and (b) country where the study was conducted. Additionally, the following demographic characteristics were coded: (a) the proportion of the sample which was female, and (b) the age range and mean age of the sample. Furthermore, we coded (a) the type of well-being that was assessed (e.g., loneliness, stress), (b) the exact measure used to assess well-being (e.g. Positive and Negative Affect Schedule [PANAS], Generalized Anxiety Disorder [GAD], Depression, Anxiety and Stress Scale [DASS-21]), (c) the social media platform used (e.g., Facebook, Twitter), (d) the mean number of times social media was accessed per day, and (e) the mean time spent on social media in hours per day. Lastly, we also coded for possible moderators such as (a) the source where the sample was retrieved from (community, schools, hospitals) and (b) whether studies measured the tendency to use social media as a coping mechanism. For moderators which were not reported in the results section (e.g., FOMO, social media addiction), this was because there was insufficient data reported in the included studies to conduct an analysis. Meta-analytic approach The main effect size index used was Pearson’s correlation coefficient r, which summarizes the strength of the bivariate relationship between two quantitative variables and is a measure of linear correlation between two sets of data (Allen, 2017). We also investigated overall differences in well-being scores between the groups with social media usage and without social media usage. Thus, another index we used was the OR, which provides an estimate for the relationship between two binary variables (Bland & Altman, 2000). Lastly, we converted OR effect sizes to r so that we could combine all the effect sizes. Effect sizes were coded or otherwise calculated such that positive values indicated an effect consistent with the hypothesis that individuals with higher social media usage will have higher overall well-being scores as compared to individuals with lower social media usage. With some samples completing multiple measures of well-being, it was possible for samples to contribute multiple effect sizes, therefore violating the assumption of independent effect sizes in a meta-analysis. As a result, the overall meta-analytic effect sizes were computed using a four-level meta-analytic approach (Pastor & Lazowski, 2018), with each individual effect size nested within the sample it was retrieved from, which was further nested within the study it was part of. Transparency and openness This meta-analysis’s design and analysis plan were not pre-registered. All data used in the current work has been made publicly available on Researchbox (#683). All analyses were conducted in R version 4.1.1 (R Core Team, 2020) using the meta-analytic package metafor version 3.0–2 (Viechtbauer, 2010) and the package psych version 2.1.3 (Revelle, 2022). Results Overall effect size Two overall meta-analytic effect sizes were calculated. All of the included studies utilised a between-subjects design. The first was calculated in terms of the tendency to use social media as a coping mechanism and well-being scores. The second was calculated in terms of overall social media usage and well-being scores. The two different operationalizations (i.e., actual social media usage versus the tendency to use social media as a coping mechanism) may have a different effect on well-being and thus lead to different associations between social media and well-being. Therefore, it is important to separate the two so that we can have an accurate understanding of the relationship between normal social media usage and well-being, as well as the tendency to use social media as a coping mechanism and well-being. The first effect size was calculated from 5 studies contributing a total of 5 samples and 72 effect sizes. The result suggests that the overall effect size was small, and that individuals with higher tendency to use social media as a coping mechanism had significantly lower well-being as compared to individuals with a lower tendency to use social media as a coping mechanism (transformed r = -0.06, Fisher’s z = -0.06, SEz = 0.02, 95% CIz = [-0.10, -0.02], p = 0.001). The second effect size was calculated from 33 studies contributing a total of 37 samples and 83 effect sizes, with the exclusion of the 5 studies which measured the tendency to use social media as a coping mechanism. These 5 studies were excluded to differentiate actual social media usage with the tendency to use social media as a coping mechanism. The result suggests that there was no significant association between social media usage and well-being (transformed r = -0.06, Fisher’s z = -0.06, SEz = 0.03, 95% CIz = [-0.13, 0.001], p = 0.055). Methodological moderator analyses Operationalization of social media usage We found that the association between social media usage and well-being was small in magnitude and non-significant regardless of the operationalization of social media usage (ps ≥ 0.079; see Table 2). The association between social media usage and well-being was non-significant when social media usage was operationalized as overall time spent on social media and when operationalized as the number of times social media was accessed.Table 2 Results of subgroup analyses in relation to operationalization of social media usage Social Media Variable n m k r z SEz 95% CIz p Social Media Usage Measure   Time Spent 29 31 124 -0.06 -0.06 0.04 [-0.13, 0.007] 0.079   No. Times Accessed 9 11 28 -0.07 -0.07 0.06 [-0.19, 0.05] 0.271 n = number of studies, m = number of samples, k = number of effect sizes, z = Fisher’s z, r = Pearson correlation coefficient, b = slope coefficient, SE = standard error, CI = confidence interval Study quality The effect sizes (145 effect sizes from 37 samples and 34 studies) from peer-reviewed journal articles (95% CI = [-0.14, -0.01]) were not significantly different from the effect sizes (10 effect sizes from 5 samples and 4 studies) from unpublished theses, preprints and dissertations (95% CI = [-0.11, 0.17]) given the overlapped 95% CIs. The results suggest that peer review status was not a significant moderator at least in the current study. Operationalization of well-being We found that regardless of how well-being was operationalized, the association between social media usage and well-being association was small in magnitude (see Fig. 2). Evidence from comparing the 95% CIs obtained suggests that none of the operationalizations of well-being produces social media-well-being relationships that are statistically different from each other (see Fig. 2).Fig. 2 Summary forest plots for operationalizations of well-being. Note. n = number of studies, m = number of samples, k = number of effect sizes. Numbers on the right indicate estimates of meta-analytic effect size in the form of Fisher’s z and corresponding 95% confidence intervals. Squares represent estimates of meta-analytic effect size in the form of Fisher’s z, with the size of each square representing total sample size. Whiskers represent 95% confidence intervals. All outcomes were coded such that positive effect sizes indicate higher levels of well-being for groups with higher social media usage Sample analyses Demographic factors Table 2 provides a summary of moderator analyses (using all of the 38 studies where data was available) with reference to various demographic factors. Subgroup analyses found that the association between social media usage and well-being was small but significant for samples recruited from the community (number of studies = 25, number of samples = 27, number of effect sizes = 113, transformed r = -0.08, Fisher’s z = -0.08, SEz = 0.03, 95% CIz = [-0.15, -0.01], p = 0.016), but not for samples recruited from schools and mixed sources (ps ≥ 0.861; see Table 3). However, moderation analyses did not find age, gender, and sample source to be significant moderators, as evidenced by the overlapping 95% CIs of the categorical variables.Table 3 Results of moderation analyses of demographic variables Moderator n m k b SEb 95% CIb r z SEz 95% CIz p Age (categorical)   < 25 years old 14 15 49 -0.05 -0.08 0.07 [-0.19, 0.08] 0.428   > 25 years old 16 17 84 -0.03 -0.03 0.04 [-0.12, 0.05] 0.420 Age (continuous) 29 32 133 0.002 0.003 [-0.004, 0.008] 0.478 Gender proportion 38 42 154 0.16 0.18 [-0.19, 0.52] 0.367 Sample Source   School 10 12 36 -0.01 -0.01 0.09 [-0.19, 0.16] 0.880   Mixed 2 2 5 -0.01 -0.01 0.08 [-0.18, 0.15] 0.861   Community 25 27 113 -0.08 -0.08 0.03 [-0.15, -0.01] 0.016 n = number of studies, m = number of samples, k = number of effect sizes, z = Fisher’s z, r = Pearson correlation coefficient, b = slope coefficient, SE = standard error, CI = confidence interval Publication bias To rule out potential threats to the validity of the meta-analysis due to small-study effects and selective reporting, we ran Egger’s test (Egger et al., 1997) for publication bias (Sterne & Egger, 2001). A statistically significant p-value (p < 0.05) obtained from Egger’s test would indicate that publication bias was present in the included data. Upon conducting Egger’s test for publication bias, we found that b = 17.88, SEb = 6.59, 95% CIb = [4.97, 30.79], p = 0.007 for the relation between social media usage and well-being, suggesting that publication bias was present in the current meta-analysis and skewed towards a negative effect size (see Fig. 3). This suggests that the negative association between social media usage and well-being found in our study might have a higher likelihood of being biased.Fig. 3 Funnel plot for social media usage. Note. effect size = r Discussion The current literature on social media and well-being has provided mixed and inconclusive findings, thus creating a polarizing view of social media. These mixed findings continue to extend into the pandemic, with researchers debating over the effects of social media in the new norms of social isolation (e.g., quarantines, lockdowns, social distancing). In light of these inconclusive findings, the aim of our meta-analysis was to synthesize previous research data in order to have a holistic understanding of the association between social media and well-being, particularly in the present context of COVID-19. Furthermore, we also considered the effects of various moderators—such as demographics, as well as different operationalizations of social media usage and well-being to potentially explain the mixed findings in the current literature. Overall, our results show that the relationship between social media and well-being was non-significant in the context of COVID-19 which is inconsistent with the majority of the current findings that social media is linked to poorer psychological outcomes. The lack of association found in our study does not support our initial hypothesis of a positive correlation between social media usage and well-being in the context of COVID-19. Nevertheless, our results provide additional support suggesting that the relationship between social media and well-being is not measure-specific across different operationalizations. Indeed, the consistent results across all measures of well-being in our study is in line with past findings which have suggested that different operationalizations of well-being such as negative affectivity, anxiety and depression are highly related (Tanaka-Matsumi & Kameoka, 1986; Ryff, 1989; Sandvik et al., 2009). These results further supported the robustness of the current findings regarding the null effect in the association between social media usage and the different operationalizations of well-being. Similarly, our moderation analyses of demographic variables found that none of the moderators except community were significant. One possible reason for the nonsignificant findings could be due to social media usage being prevalent across the different demographics in our current society (GlobalWebIndex, 2015; Kemp, 2021; Pew Research Center, 2021; Statista Research Department, 2020). As a result of globalization and the rising adoption rate of technology, we speculated that social media usage has become the norm and individuals around the world tend to use social media for relatively similar amounts of time (GlobalWebIndex, 2015; Kemp, 2021; Pew Research Center, 2021; Statista Research Department, 2020). Therefore, this could explain why we did not see differences in effects across different demographic variables. Interestingly, community was the only significant moderator across all demographic variables. One possible reason could be that within the broader community, there might be more individuals who are depressed, lonely, or lack social interaction during COVID-19. These individuals with poorer levels of well-being could have had an over-reliance on social media, thus contributing to an overall negative correlation between social media usage and well-being. Future research should thus conduct more longitudinal studies to explore sample source as a moderator and ascertain directionality of the relationship between social media usage and well-being. Futhermore, our nonsignificant results are contrary to previous meta-analytic findings which have found that social media is significantly correlated with lower levels of well-being and may harm users by exposing them to negative experiences such as unhealthy social comparisons and feelings of inferiority (Vogel et al., 2014; Whittaker & Kowalski, 2015). Additionally, the nonsignificant effect size between social media usage and well-being in our study was inconsistent with previous studies which reported an effect size of r = -0.22 (Marino et al., 2018) and r = -0.20 (Yang et al., 2019) respectively. Similarly, previous meta-analytic studies on the relationship between social media usage and depressive symptoms also reported an effect size of r = 0.11 (Cunningham et al., 2021), r = 0.11 (Ivie et al., 2020) and r = 0.17 (Vahedi & Zannella, 2021) respectively. One possible reason for this nonsignificant relationship could be because the data was collected during pandemic periods. During a pandemic, individuals might experience heightened levels of loneliness due to social isolation. In the context of COVID-19, it is physically impossible to have face-to-face interactions with friends, thus increasing the importance of social capital during COVID-19. One of the few ways that an individual can have meaningful social interactions and obtain social support might be through social media (Hartanto et al., 2020). This supports previous research findings of social media usage being associated with increases in social capital (Chan, 2015; Chen & Li, 2017; Nieminen et al., 2010) and social support (Bucci et al., 2019; Naslund et al., 2016), which are both related to higher levels of well-being (Asante, 2012; Chu et al., 2010; Cuadros & Berger, 2016). Indeed, during pandemic periods, social media might be a potential tool for individuals to obtain the social support and that comfort that they seek, therefore attenuating the negative correlation between social media usage and well-being. Additionally, our study was also interested in exploring the individual differences in the tendency to use social media as a coping mechanism. Contrary to our hypothesis, the tendency to use social media as a coping mechanism was negatively correlated with higher levels of well-being, which suggests that individuals who have a higher tendency to use social media as a form of coping are more likely to experience poorer emotional well-being than individuals who have a lower tendency to use social media as a form of coping. One plausible explanation for this finding could be because social media might not be a healthy coping mechanism. Supporting this line of argument, past findings have shown that an over-reliance on social media as a coping mechanism against life’s stressors might lead to problematic social media use which is associated with lower levels of well-being (Worsley et al., 2018; Kırcaburun et al., 2019). At first glance, the findings for our two hypotheses may appear to be contradictory. We found that during a pandemic, social media may serve as a communication channel for social support and comfort. At the same time, we also found that over-reliance on social media as a coping mechanism against life’s stressors might be problematic. However, it is important to note that relying on social media as a coping mechanism may only lead to negative outcomes when it is excessive (Marttila et al., 2021). During a pandemic, normal usage and reliance of social media in order to stay connected with one’s loved ones may reap positive benefits. Moreover, although our results may suggest that the tendency to use social media as a form of coping is harmful, it is important that we cannot rule out the possibility of reverse causation. For instance, the detrimental physical and psychological impacts arising from COVID-19 factors could have caused individuals to experience lower levels of well-being in general (Stanton et al., 2020). Individuals with poorer levels of well-being could have turned to social media as a form of coping, thus contributing to an overall negative correlation between the tendency to use social media as a coping mechanism and well-being (Hartanto et al., 2021a, b; Song et al., 2014). For example, a previous meta-analysis studying the relationship between loneliness and Facebook usage found that loneliness led people to use Facebook more often (Song et al., 2014). The uncertainty regarding the directionality of the relationship between social media and well-being highlights the importance of longitudinal studies. Indeed, our meta-analysis found that most of the studies on social media usage and well-being were cross-sectional in nature. Therefore, future studies should consider employing longitudinal designs to ascertain the directionality of the relationship. Taken together, our study observes that there was a negative but non-significant correlation between social media and well-being during pandemic times. Our non-significant finding contributes to the present literature by suggesting that the relationship between social media and well-being is dynamic and dependent on different contextual factors. Furthermore, our study has identified several contextual factors whereby social media can be used as a potential avenue of support for dealing with pandemic-related stressors. During a pandemic whereby face-to-face interaction is difficult, social media may serve as a communication channel to provide social support and comfort (Chan, 2015; Chen & Li, 2017; Nieminen et al., 2010). Additionally, we found a negative association between the tendency to use social media as a coping mechanism and well-being, which is line with the present literature (Kırcaburun et al., 2019; Worsley et al., 2018). Moreover, our study also discovered several gaps in the current literature. Through our meta-analysis, we found that there was a lack of sufficient data to conduct analysis on many moderators (e.g., FoMO, problematic social media usage). This highlights the need for future researchers to look at different moderators when conducting studies on social media and well-being. Moreover, most of the studies on social media usage and well-being were cross-sectional in nature which highlights the importance of longitudinal studies to ascertain directionality. There were also great variations in the operationalizations of social media (i.e., tendency to use social media as a coping mechanism versus non-coping) and the methodologies of measuring social media usage (i.e., subjective versus objective measures, total time spent on social media versus number of times social media was accessed) across the studies included in our meta-analysis. Our meta-analysis thus highlights the need for a standardization of methodologies and operationalizations when conducting studies on social media usage. However, our study has one important limitation: We found that publication bias was significant in our study which suggests that our findings might be biased. This highlights the importance of taking into account non-significant findings for future research. In sum, our meta-analysis can serve as an important guideline for future studies to improve its operationalization and methodological rigors as well as to shed light on the potential use of social media as an important avenue of social support during the pandemic. Funding This research was supported by grants awarded to Andree Hartanto by Singapore Management University through research grants from the Ministry of Education Academy Research Fund Tier 1 (21-SOSS-SMU-023) and Lee Kong Chian Fund for Research Excellence. Data availability This meta-analysis’s design and analysis plan were not pre-registered. All data used in the current work has been made publicly available on Researchbox (#683). Declarations Informed Consent No new data was produced in this work. Conflict of Interests On behalf of all authors, the corresponding author states that there is no conflict of interest. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References References marked with an asterisk (*) indicate studies included in the review. Adams RE Santo JB Bukowski WM The presence of a best friend buffers the effects of negative experiences Developmental Psychology 2011 47 6 1786 1791 10.1037/a0025401 21895364 *Alam, M. K., Ali, F. B., Banik, R., Yasmin, S., & Salma, N. (2021). Assessing the mental health condition of home-confined university level students of Bangladesh due to the COVID-19 pandemic. Journal of Public Health, 1–8. 10.1007/s10389-021-01542-w Allen M The SAGE encyclopedia of communication research methods 2017 SAGE Publications *Al-Qahtani, A. M., Elgzar, W. T., & Ibrahim, H. A. F. (2020). COVID-19 pandemic: Psycho-social consequences during the social distancing period among Najran City population. Psychiatria Danubina, 32(2), 280–286. 10.24869/psyd.2020.280 Asante KO Social support and the psychological wellbeing of people living with HIV/AIDS in Ghana African Journal of Psychiatry 2012 15 5 340 345 10.4314/ajpsy.v15i5.42 23044888 *Aymerich-Franch, L. (2020). COVID-19 lockdown: Impact on psychological well-being and relationship to habit and routine modifications [Preprint]. PsyArXiy. 10.31234/osf.io/9vm7r Baccarella CV Wagner TF Kietzmann JH McCarthy IP Social media? It's serious! Understanding the dark side of social media European Management Journal 2018 36 4 431 438 10.1016/j.emj.2018.07.002 Barker V Older adolescents' motivations for social network site use: The influence of gender, group identity, and collective self-esteem Cyberpsychology & Behavior 2009 12 2 209 213 10.1089/cpb.2008.0228 19250021 Bland JM Altman DG The odds ratio British Medical Journal 2000 320 7247 1468 1468 10.1136/bmj.320.7247.1468 10827061 Bonsaksen T Schoultz M Thygesen H Ruffolo M Price D Leung J Geirdal AØ Loneliness and its associated factors nine months after the covid-19 outbreak: A cross-national study International Journal of Environmental Research and Public Health 2021 18 6 1 1 10.3390/ijerph18062841 Boursier V Gioia F Musetti A Schimmenti A Facing loneliness and anxiety during the COVID-19 isolation: The role of excessive social media use in a sample of Italian adults Frontiers in Psychiatry 2020 11 586222 586222 10.3389/fpsyt.2020.586222 33363484 Brunborg GS Andreas JB Increase in time spent on social media is associated with modest increase in depression, conduct problems, and episodic heavy drinking Journal of Adolescence 2019 74 201 209 10.1016/j.adolescence.2019.06.013 31254779 Bucci S Schwannauer M Berry N The digital revolution and its impact on mental health care Psychology and Psychotherapy: Theory, Research and Practice 2019 92 2 277 297 10.1111/papt.12222 Cauberghe V Van Wesenbeeck I De Jans S Hudders L Ponnet K How adolescents use social media to cope with feelings of loneliness and anxiety during COVID-19 lockdown Cyberpsychology, Behavior, and Social Networking 2021 24 4 250 257 10.1089/cyber.2020.0478 33185488 Chakraborty T Kumar A Upadhyay P Dwivedi YK Link between social distancing, cognitive dissonance, and social networking site usage intensity: A country-level study during the COVID-19 outbreak Internet Research 2021 31 2 419 456 10.1108/INTR-05-2020-0281 Chan M Mobile phones and the good life: Examining the relationships among mobile use, social capital and subjective well-being New Media & Society 2015 17 1 96 113 10.1177/1461444813516836 Chao M Xue D Liu T Yang H Hall BJ Media use and acute psychological outcomes during COVID-19 outbreak in China Journal of Anxiety Disorders 2020 74 102248 102248 10.1016/j.janxdis.2020.102248 32505918 Chen W Lee KH Sharing, liking, commenting, and distressed? The pathway between Facebook interaction and psychological distress Cyberpsychology, Behavior, and Social Networking 2013 16 10 728 734 10.1089/cyber.2012.0272 23745614 Chen H-T Li X The contribution of mobile social media to social capital and psychological well-being: Examining the role of communicative use, friending and self-disclosure Computers in Human Behavior 2017 75 958 965 10.1016/j.chb.2017.06.011 Chen IH Chen CY Pakpour AH Griffiths MD Lin CY Internet-related behaviors and psychological distress among schoolchildren during COVID-19 school suspension Journal of the American Academy of Child and Adolescent Psychiatry 2020 59 10 1099 10.1016/j.jaac.2020.06.007 32615153 Chen CY Chen IH O’Brien KS Latner JD Lin CY Psychological distress and internet-related behaviors between schoolchildren with and without overweight during the COVID-19 outbreak International Journal of Obesity 2021 45 3 677 686 10.1038/s41366-021-00741-5 33495523 Chen I-H Chen C-Y Pakpour AH Griffiths MD Lin C-Y Li X-D Tsang HWH Problematic internet-related behaviors mediate the associations between levels of internet engagement and distress among schoolchildren during COVID-19 lockdown: A longitudinal structural equation modeling study Journal of Behavioral Addictions 2021 10 1 135 148 10.1556/2006.2021.00006 33570506 *Chen, Y., Wu, J., Ma, J., Zhu, H., Li, W., & Gan, Y. (2021c). The mediating effect of media usage on the relationship between anxiety/fear and physician–patient trust during the COVID-19 pandemic. Psychology & Health, 1-20. 10.1080/08870446.2021.1900573 Christensen, S. P. (2018). Social media use and its impact on relationships and emotions [Master Thesis, Brigham Young University]. BYUScholarsArchive. http://hdl.lib.byu.edu/1877/etd10188 Chu PS Saucier DA Hafner E Meta-analysis of the relationships between social support and well-being in children and adolescents Journal of Social and Clinical Psychology 2010 29 6 624 645 10.1521/jscp.2010.29.6.624 Clavier T Popoff B Selim J Beuzelin M Roussel M Compere V Veber B Besnier E Association of social network use with increased anxiety related to the COVID-19 pandemic in anesthesiology, intensive care, and emergency medicine teams: Cross-sectional web-based survey study JMIR mHealth and uHealth 2020 8 9 e23153 e23153 10.2196/23153 32924946 Cuadros O Berger C The protective role of friendship quality on the wellbeing of adolescents victimized by peers Journal of Youth and Adolescence 2016 45 9 1877 1888 10.1007/s10964-016-0504-4 27230120 *Cuara, J. M. (2020). Social Media Use and the Relationship between Mental, Social, and Physical Health in College Students [Doctoral dissertation, California Baptist University]. ProQuest Dissertations and Theses. https://www.proquest.com/openview/33a4d12aa4c7ac04c047a64baacb92e6/1?pq-origsite=gscholar&cbl=18750&diss=y Cunningham S Hudson CC Harkness K Social media and depression symptoms: A meta-analysis Research on Child and Adolescent Psychopathology 2021 49 2 241 253 10.1007/s10802-020-00715-7 33404948 David JL Powless MD Hyman JE Purnell DM Steinfeldt JA Fisher S College student athletes and social media: The psychological impacts of Twitter use International Journal of Sport Communication 2018 11 2 163 186 10.1123/ijsc.2018-0044 Diener ED Emmons RA Larsen RJ Griffin S The satisfaction with life scale Journal of Personality Assessment 1985 49 1 71 75 10.1207/s15327752jpa4901_13 16367493 Dougall AL Hyman KB Hayward MC McFeeley S Baum A Optimism and traumatic stress: The importance of social support and coping Journal of Applied Social Psychology 2001 31 2 223 245 10.1111/j.1559-1816.2001.tb00195.x Drouin M McDaniel BT Pater J Toscos T How parents and their children used social media and technology at the beginning of the COVID-19 pandemic and associations with anxiety Cyberpsychology, Behavior and Social Networking 2020 23 11 727 736 10.1089/cyber.2020.0284 32726144 Dunbar RI Do online social media cut through the constraints that limit the size of offline social networks? Royal Society Open Science 2016 3 1 150292 10.1098/rsos.150292 26909163 Ebrahim AH Saif ZQ Buheji M AlBasri N Al-Husaini FA Jahrami H COVID-19 information-seeking behavior and anxiety symptoms among parents OSP Journal of Health Care and Medicine 2020 1 1 1 9 Eden AL Johnson BK Reinecke L Grady SM Media for coping during COVID-19 social distancing: Stress, anxiety, and psychological well-being Frontiers in Psychology 2020 11 3388 10.3389/fpsyg.2020.577639 Egger M Smith GD Schneider M Minder C Bias in meta-analysis detected by a simple, graphical test British Medical Journal 1997 315 7109 629 634 10.1136/bmj.315.7109.629 9310563 Ellis WE Dumas TM Forbes LM Physically isolated but socially connected: Psychological adjustment and stress among adolescents during the initial COVID-19 crisis Canadian Journal of Behavioural Science 2020 52 3 177 187 10.1037/cbs0000215 Ellison N Steinfield C Lampe C The benefits of Facebook “friends”: Exploring the relationship between college students. Use of Online Social Networks and Social Capital’ Journal of Computer-Mediated Communication 2007 12 4 1143 1168 10.1111/j.1083-6101.2007.00367.x Engle, S., Stromme, J., & Zhou, A. (2020). Staying at home: mobility effects of Covid-19. Available at SSRN 3565703. Feinstein BA Hershenberg R Bhatia V Latack JA Meuwly N Davila J Negative social comparison on Facebook and depressive symptoms: Rumination as a mechanism Psychology of Popular Media Culture 2013 2 3 161 10.1037/a0033111 Fernandes B Biswas UN Tan-Mansukhani R Vallejo A Essau CA The impact of COVID-19 lockdown on internet use and escapism in adolescents Revista de Psicología Clínica Con Niños y Adolescentes 2020 7 3 59 65 10.21134/rpcna.2020.mon.2056 Fox J Vendemia MA Selective self-presentation and social comparison through photographs on social networking sites Cyberpsychology, Behavior, and Social Networking 2016 19 10 593 600 10.1089/cyber.2016.0248 27732079 Fumagalli E Dolmatzian MB Shrum L Centennials, FOMO, and loneliness: An investigation of the impact of social networking and messaging/VoIP apps usage during the initial stage of the coronavirus pandemic Frontiers in Psychology 2021 12 620739 620739 10.3389/fpsyg.2021.620739 33633646 GlobalWebIndex (2015). Use of social media by Generation Insights Infographic. GWI. https://www.gwi.com/reports/social-media-across-generations Groarke JM Berry E Graham-Wisener L McKenna-Plumley PE McGlinchey E Armour C Loneliness in the UK during the COVID-19 pandemic: Cross-sectional results from the COVID-19 psychological wellbeing study PLoS ONE 2020 15 9 e0239698 e0239698 10.1371/journal.pone.0239698 32970764 Güngör A Karaman MA Sarı Hİ Çolak TS Investigating the factors related to Coronavirus disease 2019 (COVID-19) on undergraduate students’ interests in coursework International Journal of Psychology and Educational Studies 2020 7 3 1 13 10.17220/ijpes.2020.03.001 Hammad MA Alqarni TM Psychosocial effects of social media on the Saudi society during the Coronavirus Disease 2019 pandemic: A cross-sectional study PLoS ONE 2021 16 3 e0248811 e0248811 10.1371/journal.pone.0248811 33735309 Harkness K Cunningham S Hudson CC Social media and depression symptoms: a meta-analysis Journal of Abnormal Child Psychology 2021 49 2 241 10.1007/s10802-020-00715-7 Hartanto A Yong JC Toh WX Tng GYQ Tov W Cognitive, social, emotional, and subjective health benefits of computer use in adults: A 9-year longitudinal study from the Midlife in the United States (MIDUS) Computers in Human Behavior 2020 104 106179 10.1016/j.chb.2019.106179 Hartanto A Lua VY Quek FY Yong JC Ng MH A critical review on the moderating role of contextual factors in the associations between video gaming and well-being Computers in Human Behavior Reports 2021 4 100135 10.1016/j.chbr.2021.100135 Hartanto A Quek FY Tng GY Yong JC Does social media use increase depressive symptoms? A reverse causation perspective Frontiers in Psychiatry 2021 12 641934 10.3389/fpsyt.2021.641934 33833700 Hayran C Anik L Well-being and fear of missing out (FOMO) on digital content in the time of covid-19: A correlational analysis among university students International Journal of Environmental Research and Public Health 2021 18 4 1 13 10.3390/ijerph18041974 Hikmah K Prisandy L Melinda G Ayatullah MI An online survey: Assessing anxiety level among general population during the Coronavirus disease-19 pandemic in Indonesia Open Access Macedonian Journal of Medical Sciences 2020 8 T1 451 458 10.3889/oamjms.2020.5386 Houben M Van Den Noortgate W Kuppens P The relation between short-term emotion dynamics and psychological well-being: A meta-analysis Psychological Bulletin 2015 141 4 901 10.1037/a0038822 25822133 Huang, C. (2017). Time spent on social network sites and psychological well-being: A meta-analysis. Cyberpsychology, Behavior, and Social Networking, 20(6), 346–354. Hunt MG Marx R Lipson C Young J No more FOMO: Limiting social media decreases loneliness and depression Journal of Social and Clinical Psychology 2018 37 10 751 768 10.1521/jscp.2018.37.10.751 Hunter P The growth of social media in science: Social media has evolved from a mere communication channel to an integral tool for discussion and research collaboration EMBO Reports 2020 21 5 e50550 e50550 10.15252/embr.202050550 32329177 Ikizer G Karanci AN Gul E Dilekler I Post-traumatic stress, growth, and depreciation during the COVID-19 pandemic: Evidence from Turkey European Journal of Psychotraumatology 2021 12 1 1872966 1872966 10.1080/20008198.2021.1872966 34025916 Ivie EJ Pettitt A Moses LJ Allen NB A meta-analysis of the association between adolescent social media use and depressive symptoms Journal of Affective Disorders 2020 275 165 174 10.1016/j.jad.2020.06.014 32734903 Kaplan AM Haenlein M Users of the world, unite! The challenges and opportunities of social media Business Horizons 2010 53 1 59 68 10.1016/j.bushor.2009.09.003 Keles B McCrae N Grealish A A systematic review: The influence of social media on depression, anxiety and psychological distress in adolescents International Journal of Adolescence and Youth 2020 25 1 79 93 10.1080/02673843.2019.1590851 Kemp, S. (2021). Half a billion users joined social in the last year (and Other Facts). Social Media Marketing & Management Dashboard. https://blog.hootsuite.com/simon-kemp-social-media/ Killgore WD Cloonan SA Taylor EC Miller MA Dailey NS Three months of loneliness during the COVID-19 lockdown Psychiatry Research 2020 293 113392 113392 10.1016/j.psychres.2020.113392 32835927 Kim AJ Johnson KK Power of consumers using social media: Examining the influences of brand-related user-generated content on Facebook Computers in Human Behavior 2016 58 98 108 10.1016/j.chb.2015.12.047 Kırcaburun K Kokkinos CM Demetrovics Z Király O Griffiths MD Çolak TS Problematic online behaviors among adolescents and emerging adults: Associations between cyberbullying perpetration, problematic social media use, and psychosocial factors International Journal of Mental Health and Addiction 2019 17 4 891 908 10.1007/s11469-018-9894-8 Krause AE Dimmock J Rebar AL Jackson B Music listening predicted improved life satisfaction in university students during early stages of the COVID-19 pandemic Frontiers in Psychology 2021 11 631033 631033 10.3389/fpsyg.2020.631033 33551940 Kraut R Patterson M Lundmark V Kiesler S Mukopadhyay T Scherlis W Internet paradox: A social technology that reduces social involvement and psychological well-being? American Psychologist 1998 53 9 1017 1031 10.1037//0003-066x.53.9.1017 9841579 Krendl AC Perry BL The impact of sheltering in place during the COVID-19 pandemic on older adults’ social and mental well-being The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences 2021 76 2 e53 e58 10.1093/geronb/gbaa110 32778899 *Lake, C. (2020). Social isolation, fear of missing out, and social media use in deaf and hearing college students [Master Thesis, Rochester Institute of Technology]. RIT Scholar Works. https://scholarworks.rit.edu/theses/10671 Lardo A Dumay J Trequattrini R Russo G Social media networks as drivers for intellectual capital disclosure: Evidence from professional football clubs Journal of Intellectual Capital 2017 18 1 63 80 10.1108/JIC-09-2016-0093 Lee SJ Online communication and adolescent social ties: Who benefits more from Internet use? Journal of Computer-Mediated Communication 2009 14 3 509 531 10.1111/j.1083-6101.2009.01451.x Lee PSN Leung L Lo V Xiong C Wu T Internet communication versus face-to-face interaction in quality of life Social Indicators Research 2011 100 3 375 389 10.1007/s11205-010-9618-3 Lemenager T Neissner M Koopmann A Reinhard I Georgiadou E Müller A Kiefer F Hillemacher T Covid-19 lockdown restrictions and online media consumption in Germany International Journal of Environmental Research and Public Health 2021 18 1 1 13 10.3390/ijerph18010014 Lin LY Sidani JE Shensa A Radovic A Miller E Colditz JB Hoffman BL Giles LM Primack BA Association between social media use and depression among US young adults Depression and Anxiety 2016 33 4 323 331 10.1002/da.22466 26783723 Lin C-Y Broström A Griffiths MD Pakpour AH Investigating mediated effects of fear of COVID-19 and COVID-19 misunderstanding in the association between problematic social media use, psychological distress, and insomnia Internet Interventions : The Application of Information Technology in Mental and Behavioural Health 2020 21 100345 100345 10.1016/j.invent.2020.100345 Lisitsa E Benjamin KS Chun SK Skalisky J Hammond LE Mezulis AH Loneliness among young adults during Covid-19 pandemic: The mediational roles of social media use and social support seeking Journal of Social and Clinical Psychology 2020 39 8 708 726 10.1521/jscp.2020.39.8.708 Liu D Baumeister RF Social networking online and personality of self-worth: A meta-analysis Journal of Research in Personality 2016 64 79 89 10.1016/j.jrp.2016.06.024 Liu D Baumeister RF Yang C Hu B Digital communication media use and psychological well-being: A meta-analysis Journal of Computer-Mediated Communication 2019 24 5 259 273 10.1093/jcmc/zmz013 Liu H Liu W Yoganathan V Osburg V-S COVID-19 information overload and generation Z’s social media discontinuance intention during the pandemic lockdown Technological Forecasting & Social Change 2021 166 120600 10.1016/j.techfore.2021.120600 34876758 Magson NR Freeman JYA Rapee RM Richardson CE Oar EL Fardouly J Risk and protective factors for prospective changes in adolescent mental health during the COVID-19 pandemic Journal of Youth and Adolescence 2021 50 1 44 57 10.1007/s10964-020-01332-9 33108542 Marino C Gini G Vieno A Spada MM The associations between problematic Facebook use, psychological distress and well-being among adolescents and young adults: A systematic review and meta-analysis Journal of Affective Disorders 2018 226 274 281 10.1016/j.jad.2017.10.007 29024900 Marttila E Koivula A Räsänen P Does excessive social media use decrease subjective well-being? A longitudinal analysis of the relationship between problematic use, loneliness and life satisfaction Telematics and Informatics 2021 59 101556 10.1016/j.tele.2020.101556 Mascheroni, G., Vincent, J., & Jimenez, E. (2015). “Girls are addicted to likes so they post semi-naked selfies”: Peer mediation, normativity and the construction of identity online. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 9(1). 10.5817/CP2015-1-5 Masciantonio A Bourguignon D Bouchat P Balty M Rimé B Don’t put all social network sites in one basket: Facebook, Instagram, Twitter, TikTok, and their relations with well-being during the COVID-19 pandemic PLoS ONE 2021 16 3 e0248384 e0248384 10.1371/journal.pone.0248384 33705462 McCain JL Campbell WK Narcissism and social media use: A meta-analytic review Psychology of Popular Media Culture 2018 7 3 308 10.1037/ppm0000137 McPherson M Smith-Lovin L Brashears ME Social isolation in America: Changes in core discussion networks over two decades American Sociological Review 2006 71 3 353 375 10.1177/000312240607100301 Mendeley. (n.d.) . Mendeley Desktop v1.19.4 [Software]. Mendeley. https://www.mendeley.com/autoupdates/installers/1.19.4 Milani L Osualdella D Di Blasio P Quality of interpersonal relationships and problematic Internet use in adolescence CyberPsychology & Behavior 2009 12 6 681 684 10.1089/cpb.2009.0071 19788382 Misra R Crist M Burant CJ Relationships among life stress, social support, academic stressors, and reactions to stressors of international students in the United States International Journal of Stress Management 2003 10 2 137 157 10.1037/1072-5245.10.2.137 Moher D Liberati A Tetzlaff J Altman DG the PRISMA Group Reprint—Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement Physical Therapy 2009 89 9 873 880 10.1093/ptj/89.9.873 19723669 Nakagomi A Shiba K Kondo K Kawachi I Can online communication prevent depression among older people? A longitudinal analysis Journal of Applied Gerontology 2020 41 1 733464820982147 733464820982147 10.1177/0733464820982147 Naslund JA Aschbrenner KA Marsch LA Bartels SJ The future of mental health care: peer-to-peer support and social media Epidemiology and Psychiatric Sciences 2016 25 2 113 122 10.1017/s2045796015001067 26744309 Nesi J Choukas-Bradley S Prinstein MJ Transformation of adolescent peer relations in the social media context: Part 1—A theoretical framework and application to dyadic peer relationships Clinical Child and Family Psychology Review 2018 21 3 267 294 10.1007/s10567-018-0261-x 29627907 Neto R Golz N Polega M Social media use, loneliness, and academic achievement: A correlational study with urban high school students Journal of Research in Education 2015 25 2 28 37 Newson M Zhao Y El Zein M Sulik J Dezecache G Deroy O Tuncgenc B Digital contact does not promote wellbeing, but face-to-face does: A cross-national survey during the Covid-19 pandemic New Media & Society 2021 10.1177/14614448211062164 Nieminen T Martelin T Koskinen S Aro H Alanen E Hyyppä MT Social capital as a determinant of self-rated health and psychological well-being International Journal of Public Health 2010 55 6 531 542 10.1007/s00038-010-0138-3 20361226 O'Keeffe GS Clarke-Pearson K The impact of social media on children, adolescents, and families Pediatrics 2011 127 4 800 804 10.1542/peds.2011-0054 21444588 Oksanen A Oksa R Savela N Mantere E Savolainen I Kaakinen M COVID-19 crisis and digital stressors at work: A longitudinal study on the Finnish working population Computers in Human Behavior 2021 122 106853 10.1016/j.chb.2021.106853 34754137 Oxford, L., Dubras, R., Currey, H., & Nazir, M. (2020). Digital 2020: 3.8 billion people use social media. We Are Social. https://wearesocial.com/blog/2020/01/digital-2020-3-8-billion-people-use-social-media Park N Song H Lee KM Social networking sites and other media use, acculturation stress, and psychological well-being among East Asian college students in the United States Computers in Human Behavior 2014 36 138 146 10.1016/j.chb.2014.03.037 Pastor DA Lazowski RA On the multilevel nature of meta-analysis: A tutorial, comparison of software programs, and discussion of analytic choices Multivariate Behavioral Research 2018 53 1 74 89 10.1080/00273171.2017.1365684 28952787 *Patabendige, M., Gamage, M. M., Weerasinghe, M., & Jayawardane, A. (n.d.). Psychological impact of the COVID-19 pandemic among pregnant women in Sri Lanka. International Journal of Gynaecology & Obstetrics, 15(1), 150–153. 10.1002/ijgo.13335 Pavlíček, A. (2013). Social media–the good, the bad, the ugly. IDIMT-2013, 139. Perloff RM Social media effects on young women’s body image concerns: Theoretical perspectives and an agenda for research Sex Roles 2014 71 11 363 377 10.1007/s11199-014-0384-6 Pew Research Center. (2021). Demographics of social media users and adoption in the United States. Pew Research Center: Internet, Science & Tech. https://www.pewresearch.org/internet/fact-sheet/social-media/ Pinquart M Sörensen S Differences between caregivers and noncaregivers in psychological health and physical health: A meta-analysis Psychology and Aging 2003 18 2 250 10.1037/0882-7974.18.2.250 12825775 Reich SM Subrahmanyam K Espinoza G Friending, IMing, and hanging out face-to-face: Overlap in adolescents' online and offline social networks Developmental Psychology 2012 48 2 356 10.1037/a0026980 22369341 Reiss S Franchina V Jutzi C Willardt R Jonas E From anxiety to action—Experience of threat, emotional states, reactance, and action preferences in the early days of COVID-19 self-isolation in Germany and Austria PLoS ONE 2020 15 12 e0243193 e0243193 10.1371/journal.pone.0243193 33290398 Rens E Smith P Nicaise P Lorant V Van den Broeck K Mental distress and its contributing factors among young people during the first wave of COVID-19: A Belgian survey study Frontiers in Psychiatry 2021 12 575553 575553 10.3389/fpsyt.2021.575553 33584379 Revelle, W. (2022). psych: Procedures for Psychological, Psychometric, and Personality Research. Northwestern University, Evanston, Illinois. R package version 2.2.3, https://CRAN.R-project.org/package=psych Riehm KE Holingue C Kalb LG Bennett D Kapteyn A Jiang Q Veldhuis CB Johnson RM Fallin MD Kreuter F Stuart EA Thrul J Associations between media exposure and mental distress among US adults at the beginning of the COVID-19 pandemic American Journal of Preventive Medicine 2020 59 5 630 638 10.1016/j.amepre.2020.06.008 33011008 Russell, D. W. (1996). UCLA Loneliness Scale (Version 3): Reliability, validity, and factor structure. Journal of Personality Assessment, 66(1), 20–40. Ryff CD Happiness is everything, or is it? Explorations on the meaning of psychological well-being Journal of Personality and Social Psychology 1989 57 6 1069 10.1037/0022-3514.57.6.1069 Saiphoo AN Vahedi Z A meta-analytic review of the relationship between social media use and body image disturbance Computers in Human Behavior 2019 101 259 275 10.1016/j.chb.2019.07.028 Sandvik, E., Diener, E., & Seidlitz, L. (2009). Subjective well-being: The convergence and stability of self-report and non-self-report measures. In Assessing well-being (pp. 119–138). Springer, Dordrecht. Şentürk E Geniş B Menkü BE Cosar B Karayagmurlu A The effects of social media news that users trusted and verified on anxiety level and disease control perception in COVID-19 pandemic Klinik Psikiyatri Dergisi 2021 24 1 23 32 10.5505/kpd.2020.69772 Sewall CJ Goldstein TR Rosen D Objectively measured digital technology use during the COVID-19 pandemic: Impact on depression, anxiety, and suicidal ideation among young adults Journal of Affective Disorders 2021 288 145 147 10.1016/j.jad.2021.04.008 33894615 Song H Zmyslinski-Seelig A Kim J Drent A Victor A Omori K Allen M Does Facebook make you lonely?: A meta analysis Computers in Human Behavior 2014 36 446 452 10.1016/j.chb.2014.04.011 Spitzer RL Kroenke K Williams JB Löwe B A brief measure for assessing generalized anxiety disorder: The GAD-7 Archives of Internal Medicine 2006 166 10 1092 1097 10.1001/archinte.166.10.1092 16717171 Stanton R To QG Khalesi S Williams SL Alley SJ Thwaite TL Fenning AS Vandelanotte C Depression, anxiety and stress during COVID-19: Associations with changes in physical activity, sleep, tobacco and alcohol use in Australian adults International Journal of Environmental Research and Public Health 2020 17 11 4065 10.3390/ijerph17114065 32517294 Statista Research Department. (2020). Number of social network users worldwide from 2017 to 2025 (in billions) [Graph]. Statista. https://www-statista-com.libproxy.smu.edu.sg/statistics/278414/number-of-worldwide-social-network-users/ Statista Research Department. (2021a). U.S. TikTok user growth 2024 [Graph]. Statista. https://www.statista.com/statistics/1100842/tiktok-us-user-growth/ Statista Research Department. (2021b). Topic: Social media use During coronavirus (COVID-19) worldwide. Statista. https://www.statista.com/topics/7863/social-media-use-during-coronavirus-covid-19-worldwide/ Stern, S. (2008). Producing sites, exploring identities: Youth online authorship (pp. 95–118). MacArthur Foundation Digital Media and Learning Initiative. Sterne, J. A., & Egger, M. (2001). Funnel plots for detecting bias in meta-analysis: guidelines on choice of axis. Journal of Clinical Epidemiology, 54(10), 1046–1055. Sullivan HS The interpersonal theory of psychiatry 1953 Norton *Sun, R., Rieble, C., Liu, Y., & Sauter, D. (2020). Connected despite COVID-19: The role of social interactions and social media for wellbeing [Preprint]. PsyArXiy. 10.31234/osf.io/x5k8u Tanaka-Matsumi, J., & Kameoka, V. A. (1986). Reliabilities and concurrent validities of popular self-report measures of depression, anxiety, and social desirability. Journal of Consulting and Clinical Psychology, 54(3), 328. Teresa MT Guss CD Boyd L Thriving during COVID-19: Predictors of psychological well-being and ways of coping PLoS ONE 2021 16 3 e0248591 e0248591 10.1371/journal.pone.0248591 33720985 Vahedi Z Zannella L The association between self-reported depressive symptoms and the use of social networking sites (SNS): A meta-analysis Current Psychology 2021 40 5 2174 2189 10.1007/s12144-019-0150-6 Valkenburg PM Peter J Online communication and adolescent well-being: Testing the stimulation versus the displacement hypothesis Journal of Computer-Mediated Communication 2007 12 4 1169 1182 10.1111/j.1083-6101.2007.00368.x Viechtbauer W Conducting Meta-Analyses in R with the metafor Package Journal of Statistical Software 2010 36 3 1 48 10.18637/jss.v036.i03 Vogel EA Rose JP Roberts LR Eckles K Social comparison, social media, and self-esteem Psychology of Popular Media Culture 2014 3 4 206 10.1037/ppm0000047 Vogel EA Rose JP Okdie BM Eckles K Franz B Who compares and despairs? The effect of social comparison orientation on social media use and its outcomes Personality and Individual Differences 2015 86 249 256 10.1016/j.paid.2015.06.026 Wang W Qian G Wang X Lei L Hu Q Chen J Jiang S Mobile social media use and self-identity among Chinese adolescents: the mediating effect of friendship quality and the moderating role of gender Curr Psychol 2019 40 4479 4487 (2001) 10.1007/s12144-019-00397-5 Weinstein E The social media see-saw: Positive and negative influences on adolescents’ affective well-being New Media & Society 2018 20 10 3597 3623 10.1177/1461444818755634 Wheaton MG Prikhidko A Messner GR Is fear of COVID-19 contagious? The effects of emotion contagion and social media use on anxiety in response to the Coronavirus pandemic Frontiers in Psychology 2021 11 567379 567379 10.3389/fpsyg.2020.567379 33469434 Whittaker E Kowalski RM Cyberbullying via social media Journal of School Violence 2015 14 1 11 29 10.1080/15388220.2014.949377 Worsley JD McIntyre JC Bentall RP Corcoran R Childhood maltreatment and problematic social media use: The role of attachment and depression Psychiatry Research 2018 267 88 93 10.1016/j.psychres.2018.05.023 29886276 Wrzus C Wagner J Hänel M Neyer FJ Social network changes and life events across the life span: A meta-analysis Psychological Bulletin 2013 139 1 53 80 10.1037/a0028601 22642230 Yang C Brown BB Braun MT From Facebook to cell calls: Layers of electronic intimacy in college students’ interpersonal relationships New Media & Society 2014 16 1 5 23 10.1177/1461444812472486 Yang FR Wei CF Tang JH Effect of Facebook social comparison on well-being: A meta-analysis Journal of Internet Technology 2019 20 6 1829 1836 10.3966/160792642019102006013 Yang X Yip BH Mak AD Zhang D Lee EK Wong SY The differential effects of social media on depressive symptoms and suicidal ideation among the younger and older adult population in Hong Kong during the COVID-19 pandemic: Population-based cross-sectional survey study JMIR Public Health and Surveillance 2021 7 5 e24623 10.2196/24623 33835937 Zhao N Zhou G COVID-19 stress and addictive social media use (SMU): Mediating role of active use and social media flow Frontiers in Psychiatry 2021 12 635546 635546 10.3389/fpsyt.2021.635546 33633616 Zywica, J., & Danowski, J. (2008). The faces of Facebookers: Investigating social enhancement and social compensation hypotheses; predicting Facebook™ and offline popularity from sociability and self-esteem, and mapping the meanings of popularity with semantic networks. Journal of Computer-Mediated Communication, 14(1), 1–34. 10.1111/j.1083-6101.2008.01429.x

0

PMC9748903

NO-CC CODE

2022-12-15 23:22:48

no

Curr Psychol. 2022 Dec 14;:1-15

utf-8

Curr Psychol

10.1007/s12144-022-04092-w

oa_other

==== Front Curr Psychol Curr Psychol Current Psychology (New Brunswick, N.j.) 1046-1310 1936-4733 Springer US New York 4109 10.1007/s12144-022-04109-4 Article Longitudinal co-trajectories of depression and alcohol problems in adults during the COVID-19 pandemic Bambrah Veerpal bambrahv@yorku.ca 1 Wardell Jeffrey D. 123 Keough Matthew T. 1 1 grid.21100.32 0000 0004 1936 9430 Department of Psychology, Faculty of Health, York University, Behavioural Sciences Building, 4700 Keele Street, Toronto, Ontario M3J 1P3 Canada 2 grid.155956.b 0000 0000 8793 5925 Institute for Mental Health Policy Research, Centre for Addiction and Mental Health, Toronto, Ontario Canada 3 grid.17063.33 0000 0001 2157 2938 Department of Psychiatry, University of Toronto, Toronto, Ontario Canada 14 12 2022 117 30 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. We examined person-centered heterogeneity in the longitudinal co-development of depression and alcohol problems during the COVID-19 outbreak. We also investigated the risk factors (personality and coping) for being in “higher” relative to “lower” risk subgroups of combined depressive symptoms and alcohol problems. Canadian participants (N = 364, Mage = 32.16, 54.67% male) completed questionnaires four times every three months, starting approximately 2 months after Canada announced its COVID-19 State-of-Emergency. Parallel-process latent class growth analysis found evidence for three latent subgroups: a “moderate increasing depression and alcohol problems” subgroup (Class 1); a “moderate stable depression, moderate decreasing alcohol problems” subgroup (Class 2); and a “low-risk normative” subgroup (with mild depression that was stable and mild alcohol problems that decreased; Class 3). Multinomial logistic regressions found that higher levels of hopelessness, impulsivity, and boredom proneness distinguished Class 1 from Class 3. Further, lower levels of general self-efficacy distinguished Class 1 from Classes 2 and 3. Linear mixed models found that Class 1 increasingly used maladaptive avoidant coping strategies (denial, drugs/alcohol, behavioural disengagement) as the pandemic progressed, whereas Class 2 increasingly used adaptive approach-oriented strategies (planning, seeking emotional support from others). We analyzed longitudinal data to detect classes of individuals with depressive and alcohol-related difficulties during COVID-19 and to characterize the vulnerability factors for increased difficulties. Highlighting the heterogeneity in the co-trajectory of depression and alcohol problems during COVID-19 and the personality and coping factors associated with combined increases in these mental health difficulties can inform treatment practices and bolster peoples’ preparedness and resilience for future pandemics. Supplementary Information The online version contains supplementary material available at 10.1007/s12144-022-04109-4. Keywords COVID-19 Depression Alcohol problems Longitudinal Personality Coping York University Generic Startup Funds ==== Body pmcThe global coronavirus (COVID-19) pandemic has significantly impacted the mental health of people worldwide. Among 1,803 Canadian adults surveyed in late-April 2020, roughly two months after Canada announced its COVID-19 State-of-Emergency, the percentage of respondents with high self-reported depression had increased from 4 percent to 10 percent since the beginning of the pandemic. One-third of respondents with depression additionally reported increased alcohol consumption since the beginning of COVID-19 (Dozois, 2021), and a separate study (McPhee et al., 2020) found that depression severity and risky drinking were greater in May 2020, after social distancing guidelines were introduced, relative to pre-social distancing. Although some studies found increases in alcohol consumption and alcohol problems during the pandemic (e.g., Pollard et al., 2020; Tucker et al., 2022), the results of studies have been mixed, with most finding that certain subgroups, such as those with depressive symptoms, report an increase in their alcohol consumption and alcohol problems (Acuff et al., 2022; Baptist-Mohseni et al., 2022; Capasso et al., 2021; Shield et al., 2022). This underscores the need to study depression and alcohol-related difficulties jointly for a well-rounded understanding of how the pandemic has impacted peoples’ mental health and well-being. The complexity in the comorbidity of depression and alcohol problems Importantly, while much of the literature supports an associative link between depression and alcohol consumption (Khantzian, 1997, 2012), many individuals who struggle with depression symptoms do not problematically consume alcohol (e.g., Pedrelli et al., 2016). Indeed, some studies find a low-to-moderate comorbidity prevalence (4–37 percent) between major depressive disorder and alcohol use disorder within the general population (see Castillo-Carniglia et al., 2019 for a review). Notably, longitudinal research suggests that even over time, the co-trajectories of depression and alcohol problems (the adverse consequences of alcohol consumption) is not uniform or consistent across all adults. In particular, Frohlich and colleagues (2018) and Orui and colleagues (2020) conducted longitudinal studies examining the heterogeneity in the co-trajectories of depressive symptoms and alcohol problems among Canadians after and during university, respectively. In both studies, participants completed measures of depression and alcohol problems over the course of 12–18 months and the authors used parallel-process latent class growth analysis (LCGA) to classify groups of individuals based on the simultaneous growth of these difficulties. This novel analysis has advantages relative to other statistical analyses because it studies the comorbidity of psychological processes over time, which allows researchers to identify distinct unobserved (latent) classes of individuals with the same developmental trajectories (see Muthen & Muthen, 2000). This person-centered analytical approach models heterogeneous longitudinal data by classifying individuals into smaller classes or subgroups with homogenous co-patterns of psychological processes over time. Frohlich and colleagues (2018) and Orui and colleagues (2020) identified multiple latent classes of depression and alcohol problems, most consistently: a “high-risk” (comorbid) class, who had high depression that was stable and high alcohol problems that were stable; a “moderate-risk” (depression-only) class, who had high depression that was stable and low alcohol problems (Orui et al., 2020 found that the latter decreased among participants in this class, whereas Frohlich et al., 2018 found that the latter remained stable among participants in this class); and a “low-risk” (normative) class, who had low depression that was stable and low alcohol problems that decreased. Together, these studies highlight the heterogeneity and hence, complexity in the comorbidity of depression and alcohol problems over time. Unfortunately, at the time of writing the present work, the number of COVID-19 cases and deaths continue to rise, indicating that COVID-19 remains a global health emergency (World Health Organization, 2022). Equally troubling, epidemiological research suggests that the probability of future pandemics that are comparable to COVID-19 in severity is increasing (e.g., Marani et al., 2021). Accordingly, continued scholarship on the impact of the pandemic on peoples’ mental health is important because it can advance prevention and intervention practices and strengthen peoples’ preparedness and resilience during these and other extraordinary times. As depression and problem drinking are clear concerns in the context of COVID-19, the current study meaningfully extended on prior research, using parallel-process LCGA to study the longitudinal co-trajectories of depression and alcohol problems in adults during the pandemic. Risk factors for comorbid depression and alcohol problems While identifying specific subgroups of depression and alcohol problems during the pandemic is imperative, it is equally critical to establish the factors that can increase one’s risk for being in the “high-risk” subgroup of elevated depressive symptoms and alcohol problems, in comparison to the other subgroups. Doing so can elucidate which people are at greater risk for combined depression and alcohol problems and which factors are important targets for prevention strategies and intervention approaches during such stressful circumstances. In their study, Orui et al. (2020) found that higher levels of hopelessness, impulsivity, and anxiety sensitivity differentiated the “high-risk” subgroup from the “low-risk” subgroup; higher levels of hopelessness differentiated the “moderate-risk” subgroup from the “low-risk” subgroup; and higher levels of impulsivity and lower levels of hopelessness differentiated the “high-risk” subgroup from the “moderate-risk” subgroup. In addition to these personality traits explored by Orui et al. (2020), the current study examined other factors and mechanisms that are relevant to the pandemic and that may distinguish the “high-risk” subgroup from other subgroups of combined depression and alcohol problems: boredom proneness, general self-efficacy, external stressors, and coping strategies. Boredom proneness refers to the tendency to frequently and intensely feel bored, which is “the aversive experience of having an unfulfilled desire to engage in satisfying activity” (Fahlman et al., 2013, p. 69). Research during the COVID-19 outbreak, as well as the SARS outbreak, suggest that boredom is one of the most commonly experienced feelings associated with social distancing/quarantine (Barari et al., 2020; Droit-Volet et al., 2020; Reynolds et al., 2008), as well as that boredom proneness impairs peoples’ adherence to such critical public health measures (see Westgate et al., 2022 for a review). Three cross-sectional studies published during COVID-19 suggest a positive relationship between boredom proneness and depression symptoms (McCurdy et al., 2022; Weiss et al., 2022; Yan et al., 2021), while most studies published during COVID-19 suggest that state boredom is positively linked to depressive symptoms and alcohol consumption (e.g., Canadian Centre on Substance Use & Addiction, 2020; Chao et al., 2020; Droit-Volet et al., 2020; Schmits & Glowacz, 2022). Given its characterization, it is plausible that boredom proneness would be a risk factor for “high-risk” depression and alcohol problems. Some work (Danckert et al., 2018) proposes that boredom proneness represents a chronic failure to respond adaptively to the state of boredom, such that people seek unhealthy experiences (e.g., self-harm) to alleviate boredom—even if other negative states are elicited (e.g., Bench & Lench, 2019; Havermans et al., 2015). Other work suggests that boredom proneness represents the general tendency toward maladaptive motivations—irrespective of the state of boredom—such as the desire to act destructively, the desire to act without thinking things through, the desire to avoid one’s emotions, uncertainty (not knowing what to do), and amotivation (not caring to do anything; Bambrah et al., 2020). The pandemic has additionally engendered significant difficulties with self-efficacy, which are the beliefs in one’s abilities to meet given situational demands (Wood & Bandura, 1989). Indeed, several published studies found that self-efficacy was low or declined (compared to before the pandemic) across various samples (e.g., teachers, nurses, mental health workers, first-time parents; Cataudella et al., 2021; Pressley & Ha, 2021; Simonetti et al., 2021; Sun et al., 2020; Xue et al., 2021; Yildirim & Güler, 2020). General self-efficacy, a trait-like dimension of self-efficacy conceptualized as the propensity to view oneself as capable of meeting the demands or challenges of tasks within a wide range of contexts (Chen et al., 2001), has been consistently and positively associated with greater psychological resilience among adults during the pandemic, including less severe depressive symptoms and less hazardous alcohol use (e.g., Dogan-Sander et al., 2021; Kohls et al., 2021; Spoorthy et al., 2020; Volken et al., 2021). As evidence across multiple countries shows that people with low general self-efficacy tend to experience self-doubt when they encounter environmental challenges, think in self-debilitating ways, cope less functionally with stressors, and avoid demands that are perceived as threatening (e.g., Luszczynska et al., 2005), it follows that this trait would be a risk factor for “high-risk” depression and alcohol problems. Additionally, positive links between external stressors and depression and alcohol problems have emerged during the pandemic. For example, research has found that income loss/financial worry, having children under 18 years of age at home, and living alone were related to increased depression, alcohol consumption, and problematic drinking (e.g., binge drinking) at the beginning of the pandemic (e.g., Fancourt et al., 2021; Wardell et al., 2020) and throughout the pandemic (e.g., Acuff et al., 2022; Centre for Addiction and Mental Health, 2020). While these relationships suggest that these external stressors may distinguish “high-risk” from “low-risk” subgroups of depression and alcohol problems, it is alternatively possible that these external stressors are of less importance to comorbidity in comparison to internal factors. Indeed, a longitudinal study of drinking habits among adults over the course of COVID-19 found that a combined increase in alcohol use and related problems was unrelated to the aforementioned external stressors, but was related to pre-pandemic hazardous alcohol use and solitary drinking, as well as possessing a higher alcohol demand prior to COVID-19 (Baptist-Mohseni et al., 2022). Finally, there is a clear relevance to understanding peoples’ coping strategies during stressful circumstances, as in the case of the pandemic (see Kar et al., 2021), which may mechanistically distinguish high- and low-risk subgroups of depression and alcohol problems. To reduce psychological distress, a range of coping styles can be adopted, such as avoidant coping characterized by cognitive or physical efforts to disengage from stressors, problem-focused coping characterized by doing something to modify the source of the stress or problem-solving, and emotion-focused coping characterized by managing and easing the emotional distress that is linked to the situation (Carver et al., 1989). Several studies examining the relationships of depression and alcohol consumption during the pandemic with these coping patterns suggest that these difficulties are positively associated with avoidant coping (e.g., using substances to cope, self-distraction, behavioural disengagement) and negatively associated with approach-oriented problem-focused coping (e.g., positive reframing, active coping) and emotion-focused coping (e.g., religion, low self-blame, low venting; e.g., Chodkiewicz et al., 2020; Gurvich et al., 2021; Shamblaw et al., 2021). Although these studies examined only the one-to-one cross-sectional relations of avoidant, problem-focused, and emotion-focused coping with depression and alcohol misuse, these patterns suggest that increased avoidant coping over time might distinguish a “high-risk” subgroup of depression and alcohol problems from a “low-risk” subgroup. Current study Most prior studies have reported on the overall relations between depressive problems and alcohol problems and posit that individuals with elevated depression have elevated alcohol-related difficulties (Grant et al., 2015). During the pandemic, there has been a limited number of longitudinal findings related to depression and alcohol problems and many studies use cross-sectional designs that fail to examine both depression and alcohol problems as related (but distinct) aspects of psychological adjustment that can both change dynamically over time. Whereas previous studies (Baptist-Mohseni et al., 2022; Leventhal et al., 2022; Tucker et al., 2022) have examined the trajectories of alcohol consumption and alcohol problems over the course of the pandemic, the current four-wave longitudinal study extended upon this work, using data from Baptist-Mohseni et al. (2022) and conducting parallel-process LCGA to characterize the co-trajectories of depression and alcohol problems among Canadians during the first nine months of the COVID-19 outbreak. Based on prior research (i.e., Frohlich et al., 2018; Orui et al., 2020), we expected to find multiple subgroups of depression and alcohol problems, at minimum a high-risk (comorbid) subgroup; a moderate-risk (depression-only) subgroup; and a low-risk (normative) subgroup. Further, informed by previous literature, we expected personality risk factors measured at baseline (namely, anxiety sensitivity, hopelessness, impulsivity, boredom proneness, and low general self-efficacy) to increase the likelihood of combined (i.e., comorbid) depression and alcohol problems over time during the pandemic. Drawing on Baptist-Mohseni et al. (2022), we also examined the relative predictive importance of external stressors (namely, experiencing a loss of income during the pandemic, being a parent who lives with a child under 18 years of age, and living alone). Given the noted sex differences in depression and alcohol difficulties (e.g., Karpyak et al., 2016; Polak et al., 2015; Tucker et al., 2022), we also explored differences between males and females in the co-trajectories of depression and alcohol problems. Finally, based on prior research, we expected that the high-risk depression and alcohol problems subgroup would be distinguished from the other subgroups based on their increasing use of avoidant coping strategies during the pandemic. Method Participant and procedures We received ethics approval from our institutional research ethics board. Data for this study was drawn from larger longitudinal studies that examined addictive behaviours during COVID-19 (see Baptist-Mohseni et al., 2022 and Wardell et al., 2020 for more details). We recruited participants via Prolific (Palan & Schitter, 2018). Adults who live in Canada, report consuming alcohol (i.e., report having more than 1 standard drink in the three months prior to wave 1), and have a high approval rating from prior studies on Prolific completed the current study. Data was collected for wave 1 between April 30th and May 4th of 2020, roughly 2 months after Canada’s COVID-19 State-of-Emergency and strict public health measures (e.g., extensive closures, stay-at-home guidelines) first went into effect. Data was collected for wave 2, wave 3, and wave 4 in July 2020, October 2020, and January 2021, respectively, during which time many public health measures remained in place throughout Canada. Participants were compensated with $13 CAD at each wave. For information on data exclusion, please see Wardell et al. (2020) and Baptist-Mohseni et al. (2022). The final sample comprised of 364 participants (Mage = 32.16, SDage = 9.54; 54.67% male). Of this sample, 294 participants (80.77%) completed wave two, 263 participants (72.25%) completed wave three, and 246 participants (67.58%) completed wave four. Most participants identified as White (65.66%) and as a non-student (75.82%), with the majority possessing a College or University degree (53.85%). Half of the participants resided in Ontario (51.65%). The median income reported by participants was $80,000–$99,000. Nearly a quarter of the sample (24.79%) endorsed hazardous alcohol consumption at wave one (based on a score of 8 or more on the 10-item Alcohol Use Disorders Identification Test; MAUDIT = 5.95, SDAUDIT = 5.08). Table S1 in the Supplemental Materials presents the full demographic characteristics of the sample. Measures See the Supplemental Materials for a full description of the measures administered in the current study, including example items and how each item was rated. Participants’ depression severity was assessed at all four waves using the Patient Health Questionnaire (PHQ-9; Kroenke et al., 2001, α = 0.87–0.90), with the total score corresponding to different levels of severity (i.e., 5–9 = Mild; 10–14 = Moderate; 15–19 = Moderately Severe; 20–27 = Severe). Participants’ alcohol problems was assessed at all four waves using the Short Inventory of Problems-Revised (SIP-2R; Kiluk et al., 2013; α = 0.91–0.95). With respect to personality factors, participants’ anxiety sensitivity, hopelessness, impulsivity, and sensation seeking were assessed at wave 1 using the Substance Use Risk Profile Scale (SURPS; Woicik et al., 2009; α = 0.71–0.89). Participants’ boredom proneness was assessed at wave 1 using the Short Boredom Proneness Scale (SBPS; Struk et al., 2017; α = 0.90). Participants’ general self-efficacy was assessed at wave 1 using the New General Self-Efficacy Scale (NGSES; Chen et al., 2001; α = 0.90). Drawing from Wardell and colleagues (2020) and Baptist-Mohseni and colleagues (2022), we assessed three pandemic-relevant external stressors; specifically, participants indicated if they experienced a loss of income during the pandemic; are a parent who lives with a child under 18; and live alone (each item was rated as “No” = 0 or “Yes” = 1). Finally, participants’ avoidant coping (characterized by facets of self-distraction, denial, substance use, and behavioural disengagement), problem-focused coping (characterized by facets of active coping, using instrumental supports, positive reframing, and planning), and emotion-focused coping (characterized by facets of using emotional supports, venting, humour, acceptance, religion, and self-blame) were assessed at all four waves using the Brief Coping Orientation to Problems Experienced Inventory (Brief-COPE; Carver, 1997; α = 0.71–0.93; see the Supplemental Materials for a description of the three coping facets that were excluded from data analyses due to poor internal reliability). Data analyses First, the data was inspected for outliers and to verify other statistical testing assumptions (e.g., normality). Outlying values were Winsorized (i.e., replaced with the highest value within ± 3.29 standard deviations of a given variable’s the mean score). We additionally conducted independent t-tests to examine potential baseline differences between participants with complete data across all four waves (n = 246) and participants with incomplete data across all four waves (n = 118) on the continuous variables at wave one. Second, parallel-process LCGA was used to identify unique latent (unobserved) classes or subgroups of participants based on their initial levels of depression (PHQ-9) and alcohol problems (SIP-2R) and based on the changes in these difficulties over the first nine months of the pandemic. More specifically, models with one latent class to six latent classes were run and tested consecutively to determine how many classes the data best supports. Three fit indices (Jung and Wickrama, 2007) were used to compare the six models. The sample size-adjusted Bayesian Information Criterion (SA-BIC) is a relative fit index, with lower values suggesting better model fit. As reported by Raftery (1995), a difference of 10 between models in the SA-BIC suggests superior fit. Entropy characterizes a model’s quality for classifying participants into smaller subgroups, with values ≥ 0.80 suggesting that the model’s overall classification quality is good (Ram & Grimm, 2009). To determine whether or not a model with “k” latent subgroups fits the data statistically significantly better than a model with “k – 1” latent subgroups, the parametric bootstrapped likelihood ratio test (BLRT) was used (Nylund et al., 2007). Drawing on Wickrama and colleagues (2016), we additionally examined the class size when selecting the class model, ensuring that the size of the smallest class was not less than 5 percent of the overall sample. Finally, to ensure that the latent classes were unique and theoretically meaningful, we examined all six models visually by graphing the co-trajectories of depression and alcohol problems of each class within each model (Williams & Kibowski, 2016). Third, multinomial logistic regressions were estimated to explore whether personality and pandemic-relevant external stressors predict participants’ membership in the depression-alcohol problems subgroups. The first model examined anxiety sensitivity, hopelessness, impulsivity, and sensation seeking (following Orui et al., 2020), and the second model examined boredom proneness and general self-efficacy. External stressors (i.e., experiencing a loss of income during the pandemic, being a parent who lives with a child under 18, and living alone) were included in these models in order to understand the relative predictive importance of internal versus external factors on class membership. Biological sex was also included as a predictor. Finally, linear mixed models (LMMs) were estimated to explore differences between the depression-alcohol problems classes on avoidant coping, problem-focused coping, and emotion-focused coping over time during the pandemic. The class (subgrouping) variable was represented with dummy coded variables, which were used to create the “Class by Time” interaction terms. Age, biological sex, and race (non-White versus White) were included in all LMMs as covariates. Prior to these LMMs, we investigated the intercept and the slope of each coping strategy outcome variable and we observed variability across participants in both parameter estimates for each outcome variable, which suggests that these parameter estimates are not fixed across participants. Thus, we specified random intercepts and random slopes within each LMM. Further, in each LMM, we initially modelled the linear effect of time and the quadratic effect of time, as well as their respective interactions with class. As the quadratic effect of time and the “quadratic time by class” interaction were not supported across all models (all p’s > 0.05), these two terms were removed from all of the LMMs in order to streamline the interpretation of the linear effect of time and for parsimony. Full information maximum likelihood was used to estimate the parallel-process LCGAs and the LMMs. Results Descriptive statistics Table S2 in the Supplemental Materials presents descriptive data for the continuous variables included in the parallel-process LCGA, multinomial logistic regressions, and LMMs. With respect to the pandemic-relevant external stressors, 41.21% of participants reported experiencing a loss of income during the pandemic, 20.39% of participants identified as parents who live with at least one child under 18 during the pandemic, and 12.91% of participants reported lived alone during the pandemic. The independent t-tests indicated no significant baseline differences between participants who had data at all four waves and participants who had missing data at one or more waves on all continuous wave one measures (all p’s > 0.05). Subgroups for combined depression and alcohol problems Table 1 presents the results for the models consisting of one latent class to six latent classes, estimated using parallel-process LCGA. Across all models, the SA-BIC index decreased, with a difference greater than 10 between each model (i.e., the model with “k” classes) and the previous model (i.e., the model with “k – 1” classes). Across all models, the entropy values were above 0.80, which suggests that the overall classification quality was good, and the parametric BLRT was significant (all p’s < 0.001). However, the sizes for models with four, five, and six classes were small (< 4 percent of the sample). These models also possessed low classification probabilities (specifically < 0.49), which suggests that the models were not accurately classifying participants into the smaller classes/subgroups. Furthermore, models with four, five, and six classes had classes with similar co-trajectories of depression severity and alcohol problems across the study’s four waves, which suggests that the classes were not meaningfully distinct or unique. Accordingly, upon considering all fit statistics, class sizes, and classification probabilities, as well as graphing the co-trajectories for each class in all of the six models, we determined that the three-class solution, which possessed high classification probabilities (> 0.87), was the most interpretable model.Table 1 Fit Information for the Parallel Process Latent Class Growth Models Class #: SA-BIC Entropy Parametric BLRT p-value Smallest Class Size (%) 1 11974.391 N/A N/A 100 2 11671.368 0.973  < .001 8.0 3 11551.286 0.954  < .001 6.3 4 11462.690 0.954  < .001 3.7 5 11395.905 0.914  < .001 3.2 6 11363.622 0.906  < .001 3.0 The fit information of the retained model is bolded The first class (6.32% of participants in the sample, n = 23) had moderate initial levels of both depression and alcohol problems that both significantly increased over the course of the COVID-19 pandemic (i.e., from wave one to wave four). The second class (7.69% of participants in the sample, n = 28) had mild-to-moderate initial levels of depression that remained stable over time and moderate initial levels of alcohol problems that significantly decreased over time. The third class (85.99% of participants in the sample, n = 313) had mild initial levels of both depression and alcohol problems, with the latter significantly decreasing over time. In subsequent sections, we refer to these three classes as followed: “moderate increasing depression and alcohol problems” subgroup (Class 1), “moderate stable depression, moderate decreasing alcohol problems” subgroup (Class 2), and “low-risk normative” subgroup (Class 3). Table 2 and Fig. 1 present the parameter estimates and graphs, respectively, of each latent class.Table 2 Parameter Estimates for Parallel Process Latent Class Growth Model – Three Latent Classes Class: Depression Severity (PHQ-9) Alcohol Problems (SIP-2R) 1 – “Moderate Increasing Depression and Alcohol Problems” (n = 23) Intercept 12.86 (p < .001) 12.87 (p < .001) 95% CI [10.75, 14.98] [11.82, 13.92] Slope 1.19 (p = .001) 1.10 (p < .001) 95% CI [0.47, 1.91] [0.64, 1.56] 2 – “Moderate Stable Depression, Moderate Decreasing Alcohol Problems” (n = 28) Intercept 8.84 (p < .001) 11.38 (p < .001) 95% CI [6.80, 10.88] [10.20, 12.57] Slope 0.28 (p = .386) -2.80 (p < .001) 95% CI [-0.35, 0.91] [-3.25, -2.34] 3 – “Low-risk Normative” (n = 313) Intercept 6.93 (p < .001) 1.26 (p < .001) 95% CI [6.39, 7.47] [0.97, 1.56] Slope 0.11 (p = .202) -0.19 (p = .002) 95% CI [-0.06, 0.29] [-0.31, -0.07] Class 1 = “Moderate Increasing Depression and Alcohol Problems”; Class 2 = “Moderate Stable Depression, Moderate Decreasing Alcohol Problems”; Class 3 = “Low-risk Normative” (mild stable depression, mild decreasing alcohol problems) The bolded values signify the significant results Fig. 1 Longitudinal Co-Trajectories of Depression and Alcohol Problems of the Three Latent Classes Risk factors Next, multinomial logistic regressions were estimated to determine the preditive roles of personality, pandemic-relevant external stressors, and biological sex on participants’ membership in the depression-alcohol problems subgroups. These results are presented in Tables 3 and 4, which report each predictor’s odds ratio with 95% confidence intervals (CIs). A variable was considered to be a predictor of being in the “high-risk” (versus “low-risk”) depression-alcohol problems subgroup if the 95% CI for the odds ratio did not include 1.0.Table 3 Multinomial Logistic Regression Models – Anxiety Sensitivity, Hopelessness, Impulsivity, Sensation Seeking, and External Stressors Baseline Predictors: B SE p OR 95% CI (OR) Model 1: “Moderate Increasing Depression and Alcohol Problems” (versus “Low-risk Normative”)   Intercept -9.85 2.11  < .001   Sex (0 = male; 1 = female) -0.41 0.51 .413 0.66 0.25, 1.78   Anxiety Sensitivity 0.02 0.10 .868 1.02 0.84, 1.23   Hopelessness 0.17 0.06 .004 1.19 1.06, 1.34   Impulsivity 0.44 0.11  < .001 1.55 1.26, 1.91   Sensation Seeking -0.06 0.07 .391 0.94 0.82, 1.08   Income loss 0.92 0.48 .054 2.52 0.98, 6.45   Young child (< 18 years old) -0.11 0.60 .851 0.89 0.28, 2.90   Living alone -0.52 0.73 .481 0.60 0.14, 2.51 “Moderate Stable Depression, Moderate Decreasing Alcohol Problems” (versus “Low-risk Normative”)   Intercept -7.92 1.78  < .001   Sex (0 = male; 1 = female) -0.37 0.44 .399 0.69 0.29, 1.64   Anxiety Sensitivity 0.03 0.09 .699 1.03 0.88, 1.22   Hopelessness 0.10 0.06 .067 1.11 0.99, 1.24   Impulsivity 0.22 0.09 .014 1.25 1.05, 1.49   Sensation Seeking 0.07 0.06 .267 1.07 0.95, 1.21   Income loss 0.70 0.41 .093 2.01 0.89, 4.52   Young child (< 18 years old) -0.13 0.53 .801 0.88 0.31, 2.47   Living alone -0.47 0.67 .489 0.63 0.17, 2.35 “Moderate Increasing Depression and Alcohol Problems” (versus “Moderate Stable Depression, Moderate Decreasing Alcohol Problems”)   Intercept -1.93 2.51 .443   Sex (0 = male; 1 = female) -0.04 0.62 .948 0.96 0.28, 3.25   Anxiety Sensitivity -0.02 0.12 .889 0.98 0.78, 1.24   Hopelessness 0.07 0.07 .317 1.08 0.93, 1.24   Impulsivity 0.22 0.13 .086 1.24 0.97, 1.59   Sensation Seeking -0.13 0.09 .131 0.88 0.74, 1.04   Income loss 0.23 0.59 .700 1.26 0.39, 4.00   Young child (< 18 years old) 0.02 0.73 .977 1.02 0.24, 4.29   Living alone -0.05 0.93 .957 0.95 0.15, 5.89 Statistically significant (p < .05) parameters are bolded Table 4 Multinomial Logistic Regression Models – Boredom Propensity, Self-Efficacy, and External Stressors Baseline Predictors: B SE p OR 95% CI (OR) Model 2: “Moderate Increasing Depression and Alcohol Problems” (versus “Low-risk Normative”)   Intercept -0.95 1.83 .606   Sex (0 = male; 1 = female) -0.33 0.48 .489 0.72 0.28, 1.84   Boredom Propensity 0.06 0.03 .032 1.06 1.01, 1.12   General Self-Efficacy -0.13 0.05 .005 0.88 0.81, 0.96   Income loss 0.68 0.47 .154 1.97 0.78, 4.98   Young child (< 18 years old) 0.44 0.59 .461 1.55 0.49, 4.91   Living alone -0.46 0.73 .523 0.63 0.15, 2.61 “Moderate Stable Depression, Moderate Decreasing Alcohol Problems” (versus “Low-risk Normative”)   Intercept -4.53 1.76 .010   Sex (0 = male; 1 = female) -0.48 0.43 .260 0.62 0.27, 1.42   Boredom Propensity 0.07 0.02 .002 1.07 1.03, 1.12   General Self-Efficacy -0.00 0.05 .948 1.00 0.91, 1.09   Income loss 0.74 0.41 .071 2.10 0.94, 4.69   Young child (< 18 years old) 0.13 0.52 .804 1.14 0.41, 3.12   Living alone -0.39 0.67 .556 0.68 0.18, 2.49 “Moderate Increasing Depression and Alcohol Problems” (versus “Moderate Stable Depression, Moderate Decreasing Alcohol Problems”)   Intercept 3.58 2.37 .131   Sex (0 = male; 1 = female) 0.15 0.61 .809 1.16 0.35, 3.80   Boredom Propensity -0.01 0.03 .702 0.99 0.93, 1.05   General Self-Efficacy -0.12 0.06 .035 0.88 0.79, 0.99   Income loss -0.06 0.59 .915 0.94 0.29, 3.00   Young child (< 18 years old) 0.31 0.73 .676 1.36 0.32, 5.73   Living alone -0.07 0.93 .938 0.93 0.15, 5.77 Statistically significant (p < .05) parameters are bolded In the first model (Table 3), we found that compared to the “low-risk normative” subgroup, the “moderate increasing depression and alcohol problems” subgroup endorsed significantly higher hopelessness and impulsivity, and the “moderate stable depression, moderate decreasing alcohol problems” subgroup endorsed significantly higher impulsivity. The reference subgroup was then changed in order to determine if these personality traits differentiated the “moderate increasing depression and alcohol problems” subgroup from the “moderate stable depression, moderate decreasing alcohol problems” subgroup; we found that these traits did not relate to subgroup membership. In the second model (Table 4), we found that compared to the “low-risk normative” subgroup, the “moderate increasing depression and alcohol problems” subgroup endorsed significantly higher boredom proneness and significantly lower general self-efficacy, and the “moderate stable depression, moderate decreasing alcohol problems” subgroup endorsed significantly higher boredom proneness. Compared to the “moderate stable depression, moderate decreasing alcohol problems” subgroup, the “moderate increasing depression and alcohol problems” subgroup endorsed significantly lower general self-efficacy. In both models (Tables 3 and 4), biological sex, experiencing a loss of income during the pandemic, being a parent who lives with a child under 18, and living alone were unrelated to depression-alcohol problems subgroup membership. Coping strategies Finally, linear mixed models (LMMs) were conducted to explore differences between the depression-alcohol problems subgroups in avoidant, problem-focused, and emotion-focused coping over time during the pandemic. Time was coded as 0, 1, 2, and 3. Age, sex, and race were statistically controlled for. Table 5 and Fig. 2 present the parameter estimates and graphs, respectively, of the Class by Time effects. Using Rights and Sterba’s (2019) Integrative Framework of R-Squared Measures, we report the proportion of total dependent variable variance explained by the LMM (i.e., by predictors through fixed slopes and random slope variation-covariation and by group outcome means through variations in the random intercept).Table 5 Linear Mixed Models – Coping Strategies Predictors: B SE t p 95% CI Model 1: Denial   Intercept 3.86 0.24 16.38  < .001 3.3977, 4.3243   Time (coded as 0, 1, 2, and 3) 0.21 0.10 2.15 .032 .0184, .4005   Age -0.00 0.00 -1.22 .222 -.0054, .0013   Sex (male = 0; female = 1) 0.06 0.09 0.68 .496 -.1153, .2370   Race (non-White = 0; White = 1) -0.03 0.09 -0.27 .786 -.2116, .1600   D1 (Class 1 = 0; Class 3 = 1) -1.30 0.23 -5.66  < .001 -1.7485, -.8472   D2 (Class 1 = 0; Class 2 = 1) -0.92 0.30 -3.07 .002 -1.5035, -.3290   Time*D1 -0.25 0.10 -2.45 .015 -.4433, -.0484   Time*D2 -0.15 0.13 -1.16 .245 -.3949, .1004 Model 2: Substance Use   Intercept 4.80 0.33 14.72  < .001 4.1622, 5.4456   Time (coded as 0, 1, 2, and 3) 0.45 0.12 3.86  < .001 .2185, .6732   Age -0.00 0.00 -1.05 .294 -.0074, .0023   Sex (male = 0; female = 1) 0.14 0.13 1.03 .302 -.1242, .3988   Race (non-White = 0; White = 1) 0.12 0.14 0.88 .382 -.1525, .3979   D1 (Class 1 = 0; Class 3 = 1) -1.86 0.31 -5.92  < .001 -2.4803, -1.2440   D2 (Class 1 = 0; Class 2 = 1) -0.06 0.41 -0.16 .877 -.8687, .7414   Time*D1 -0.48 0.12 -4.01  < .001 -.7132, -.2438   Time*D2 -0.70 0.15 -4.68  < .001 -.9934, -.4050 Model 3: Behavioural Disengagement   Intercept 3.88 0.23 16.68  < .001 3.4249, 4.3398   Time (coded as 0, 1, 2, and 3) 0.43 0.12 3.65  < .001 .1971, .6616   Age -0.00 0.00 -1.61 .109 -.0064, .0006   Sex (male = 0; female = 1) 0.12 0.10 1.22 .224 -.0729, .3096   Race (non-White = 0; White = 1) -0.10 0.10 -0.95 .341 -.2979, .1032   D1 (Class 1 = 0; Class 3 = 1) -1.12 0.22 -5.00  < .001 -1.5545, -.6764   D2 (Class 1 = 0; Class 2 = 1) -0.38 0.29 -1.32 .189 -.9555, .1888   Time*D1 -0.42 0.12 -3.44  < .001 -.6588, -.1790   Time*D2 -0.23 0.15 -1.49 .136 -.5284, .0734 Model 4: Planning   Intercept 4.54 0.31 14.51  < .001 3.9204, 5.1410   Time (coded as 0, 1, 2, and 3) 0.28 0.12 2.32 .021 .0569, .5332   Age -0.00 0.00 -1.74 .082 -.0102, .0006   Sex (male = 0; female = 1) 0.09 0.14 0.65 .519 -.1926, .3932   Race (non-White = 0; White = 1) -0.09 0.15 -0.60 .550 -.3865, .2038   D1 (Class 2 = 0; Class 3 = 1) 0.60 0.30 2.02 .045 .0038, 1.2131   D2 (Class 2 = 0; Class 1 = 1) 0.50 0.43 1.17 .242 -.3322, 1.4577   Time*D1 -0.33 0.13 -2.59 .010 -.5858, -.0884   Time*D2 -0.44 0.19 -2.28 .023 -.8285, -.0700 Model 5: Emotional Supports   Intercept 3.48 0.32 10.74  < .001 2.8452, 4.1193   Time (coded as 0, 1, 2, and 3) 0.25 0.11 2.24 .026 .0305, .4685   Age -0.00 0.00 -0.87 .387 -.0078, 0030   Sex (male = 0; female = 1) 0.63 0.15 4.16  < .001 .3304, .9231   Race (non-White = 0; White = 1) 0.19 0.16 1.18 .241 -.1266, .4984   D1 (Class 2 = 0; Class 3 = 1) 0.75 0.31 2.43 .016 .1442, 1.3602   D2 (Class 2 = 0; Class 1 = 1) 0.48 0.44 1.09 .275 -.3838, 1.3480   Time*D1 -0.29 0.12 -2.50 .013 -.5214, -.0623   Time*D2 -0.11 0.17 -0.64 .523 -.4542, .2328 Class 1 = “Moderate Increasing Depression and Alcohol Problems”; Class 2 = “Moderate Stable Depression, Moderate Decreasing Alcohol Problems”; Class 3 = “Low-risk Normative” (mild stable depression, mild decreasing alcohol problems) Fig. 2 Coping Strategies of the Three Latent Classes With respect to avoidant coping, there was a supported Class (Class 1 versus Class 3) by Time interaction in predicting the use of denial (Table 5, Model 1). Simple slopes analysis indicated that the “moderate increasing depression and alcohol problems” subgroup’s use of denial significantly increased over time (BClass1 = 0.21, SEClass1 = 0.10, pClass1 = 0.032, 95% CIClass1 [0.0184, 0.4005]), whereas denial did not significantly change over time among the “moderate stable depression, moderate decreasing alcohol problems” subgroup (BClass2 = 0.06, SEClass2 = 0.08, pClass2 = 0.436, 95% CIClass2 [-0.0953, 0.2198]) and the “low-risk normative” subgroup (BClass3 = -0.04, SEClass3 = 0.03, pClass3 = 0.149, 95% CIClass3 [-0.0857, 0.0129]). Similarly, there was a supported Class (Class 1 versus Class 2; Class 1 versus Class 3) by Time interaction in predicting the use of substances to cope (Table 5, Model 2). Simple slopes analysis found that the use of drugs and alcohol to cope significantly increased over time among the “moderate increasing depression and alcohol problems” subgroup (BClass1 = 0.45, SEClass1 = 0.12, pClass1 < 0.001, 95% CIClass1 [0.2185, 0.6732]), but significantly decreased over time among the “moderate stable depression, moderate decreasing alcohol problems” subgroup (BClass2 = -0.25, SEClass2 = 0.09, pClass2 = 0.008, 95% CIClass2 [-0.4395, -0.0672]). Among the “low-risk normative” subgroup, the use of substances to cope did not significantly change (BClass3 = -0.03, SEClass3 = 0.03, pClass3 = 0.273, 95% CIClass3 [-0.0910, 0.0257]). Finally, there was a supported Class (Class 1 versus Class 3) by Time interaction in predicting behavioural disengagement (Table 5, Model 3). Simple slopes analysis indicated that behavioural disengagement (i.e., giving up on attempts to cope) significantly increased over time among the “moderate increasing depression and alcohol problems” subgroup (BClass1 = 0.43, SEClass1 = 0.12, pClass1 < 0.001, 95% CIClass1 [0.1971, 0.6616]), but did not significantly change over time among the “moderate stable depression, moderate decreasing alcohol problems” subgroup (BClass2 = 0.20, SEClass2 = 0.10, pClass2 = 0.059, 95% CIClass2 [-0.0103, 0.3934]) and the “low-risk normative” subgroup (BClass3 = 0.01, SEClass3 = 0.03, pClass3 = 0.729, 95% CIClass3 [-0.0495, 0.0705]). The LMMs explained 53.9%, 66.1%, and 51.7% of the total outcome variance in the use of denial, drugs and alcohol, and behavioural disengagement, respectively. With regards to problem-focused coping, there was a supported Class (Class 1 versus Class 2; Class 2 versus Class 3) by Time interaction in predicting planning (Table 5, Model 4). Simple slopes analysis found that the “moderate stable depression, moderate decreasing alcohol problems” subgroup’s use of planning significantly increased over time (BClass2 = 0.28, SEClass2 = 0.12, pClass2 = 0.021, 95% CIClass2 [0.0569, 0.5332]), but that the use of planning did not significantly change over time among the “moderate increasing depression and alcohol problems” subgroup (BClass1 = -0.16, SEClass1 = 0.15, pClass1 = 0.298, 95% CIClass1 [-0.4846, 0.1307]) and the “low-risk normative” subgroup (BClass3 = -0.05, SEClass3 = 0.04, pClass3 = 0.210, 95% CIClass3 [-0.1259, 0.0261]). The LMM explained 52.0% of the total outcome variance in the use of planning. Finally, with respect to emotion-focused coping, the LMM supported a Class (Class 2 versus Class 3) by Time interaction in predicting the use of emotional supports (Table 5, Model 5). Simple slopes analysis indicated that the “moderate stable depression, moderate decreasing alcohol problems” subgroup’s use of emotional supports significantly increased over time (BClass2 = 0.25, SEClass2 = 0.11, pClass2 = 0.026, 95% CIClass2 [0.0305, 0.4685]), whereas this coping strategy did not significantly change over time among the “moderate increasing depression and alcohol problems” subgroup (BClass1 = 0.14, SEClass1 = 0.13, pClass1 = 0.304, 95% CIClass1 [-0.1258, 0.4033]) and the “low-risk normative” subgroup (BClass3 = -0.04, SEClass3 = 0.03, pClass3 = 0.227, 95% CIClass3 [-0.1111, 0.0264]). The LMM explained 66.0% of the total outcome variance in the use of emotional supports. The Class by Time interactions for predicting the other problem-focused and emotion-focused strategies were non-significant (see Table S3 in the Supplemental Materials). Discussion The current study analyzed longitudinal data, using parallel-process LCGA to determine whether distinct and meaningful classes of individuals would emerge based on their co-trajectories of depression and alcohol problems during the COVID-19 pandemic. Equally importantly, the current study sought out to determine the risk factors for combined increases in depression and alcohol problems among some people. To our knowledge, no prior study has examined person-centered changes in both depression and alcohol problems over a long period of time during the COVID-19 pandemic. Consistent with the above-reviewed studies, we found multiple classes of individuals based on their co-trajectories of depression and alcohol problems during the first nine months of the COVID-19 outbreak: a “low-risk normative” subgroup, a “moderate stable depression, moderate decreasing alcohol problems” subgroup, and a “moderate increasing depression and alcohol problems” subgroup. These patterns underscore the heterogeneity in the co-trajectory of depression and alcohol problems during the pandemic, highlighting that depression and alcohol problems remained stable and improved over time for some people, but both significantly worsened over time for other people. The multinomial logistic regressions and LMMs elucidate the risk factors for increasing depression and alcohol problems, which has significant clinical implications. The “moderate increasing depression and alcohol problems” subgroup endorsed higher hopelessness and impulsivity than the “low-risk normative” subgroup. These results suggests depression characteristics (i.e., hopelessness) do not solely contribute to increasing depression and alcohol problems over time, and that impulsivity also plays a unique role, which is consistent with prior work linking impulsivity to both mood- (e.g., Adams et al., 2019; Keough et al., 2016) and alcohol-related difficulties (e.g., Adams et al., 2019; Gonzalez et al., 2011). Further, the “moderate stable depression, moderate decreasing alcohol problems” subgroup also endorsed higher impulsivity than the “low-risk” subgroup, however this result may be accounted for the fact that people in this subgroup had moderate initial levels of alcohol problems (before these problems decreased over time). Overall, these results emphasize the need to further explore the way in which impulsivity exacerbates or maintains peoples’ mood- and alcohol-related difficulties. The “moderate increasing depression and alcohol problems” subgroup also endorsed higher levels of boredom proneness and lower levels of general self-efficacy than the “low-risk normative” subgroup. Tying together boredom proneness and general self-efficacy is the tendency to focus on oneself, which, according to theories of objective self-awareness (Brockmeyer et al., 2015; Duval & Wicklund, 1972; Higgins, 1987; Steenbarger & Aderman, 1979), is associated with negative mood states (e.g., disappointment, dejection) and attempts to escape/avoid self-awareness. A person high in boredom proneness is persistently self-focused—aware of their lack of cognitive engagement but unable to articulate actionable desires and meaningfully engage with their external environment (Eastwood & Bambrah, 2021). Indeed, correlational studies suggest a positive relationship between trait indices of self-directed attention and boredom proneness (Eastwood et al., 2007; Gana et al., 2000; Harris, 2000; Seib & Vodanovich, 1998; von Gemmingen et al., 2003; as cited in Eastwood & Bambrah, 2021). Similarly, a person with low general self-efficacy is often focused on the discrepancy between what they would like to achieve and what they believe they are capable of achieving across contexts; and hence, they often avoid external circumstances/tasks perceived as difficult and they lack the ability to regulate their emotional states in such circumstances/tasks (Bandura, 1999, 2010; Carr, 2004). The current study’s results suggest that, compared to the “normative” subgroup, people with combined increases in depression and alcohol problems might be more self-focused—stuck on themselves, unable to transition into engaging, meaningful, and self-regulatory behaviour(s). Future studies could explicitly test whether or not the relationships of boredom proneness and general self-efficacy with combined increases in depression and alcohol problems involve self-directed attention. Speaking to the findings from the LMMs, the “moderate increasing depression and alcohol problems” subgroup increasingly used avoidant coping over time, specifically the tendency to deny the reality of their problems, to use alcohol or other drugs to make themselves feel better and get through challenges, and to give up on attempts to cope with or address stressors. In tandem with the multinomial logistic regressions, which found that low general self-efficacy uniquely distinguished the “moderate increasing depression and alcohol problems” subgroup from both of the other subgroups, these findings suggests that people with combined increases in depression and alcohol problems not only tend to possess little belief in their ability to manage difficult and stressful circumstances, but also coped with such circumstances during the pandemic in a manner where they avoided thinking about, feeling, and doing difficult things. Importantly, given the pandemic’s enduring nature, these findings are the first to suggest that worsening depression and alcohol problems among a subset of people was associated longitudinally with unhealthy cognitive (denial) and behavioural (substance use, behavioural disengagement) coping efforts to avoid stressors. In contrast, the “moderate stable depression, moderate decreasing alcohol problems” subgroup decreasingly used avoidant coping, specifically substances to cope. They also increasingly used adaptive problem-focused coping skills and emotion-focused coping skills over time, specifically the tendency to plan and strategize the steps to deal with stressors and to seek comfort and emotional support from others. Avoidant coping strategies are associated with a host of psychological difficulties, including depression and problematic alcohol use, whereas problem- and emotion-focused strategies that are characterized by approaching and addressing challenging circumstances and concomitant negative emotions are associated with indices of psychological well-being (e.g., Dias et al., 2012; Eisenberg et al., 2012; Poulus et al., 2020). Notably, increases in depression and alcohol problems were more strongly related to internal factors (hopelessness, impulsivity, boredom proneness, general self-efficacy, and avoidant coping) than external stressors (experiencing a loss of income during the pandemic, being a parent who lives with a child under 18, living alone), which is consistent with the above-reviewed longitudinal findings by Baptist-Mohseni et al. (2022). Elucidating the personality and coping risk factors associated with combined increases in depression and alcohol problems not only underscores who may benefit most from prevention and intervention approaches during the pandemic (and other such stressful circumstances), but also areas of focus for these approaches. Indeed, psychoeducation and treatment approaches that target those prone to hopelessness, boredom, and low self-efficacy and that focus on reducing the use avoidant coping and enhancing the use of approach-oriented coping may help facilitate recovery and preparedness for those at risk for worsening depression alcohol problems. Specifically, approaches that emphasize the impeding nature of denial, substance use, and behavioural disengagement on one’s ability to cope with stressors in the long-term (e.g., that denying the reality of a stressor, using alcohol/drugs, and reducing one’s efforts to address a stressor might permit the stressor to become more serious, making it more difficult to address; Carver et al., 1989), as well as emphasize how these avoidant coping strategies maintain hopelessness, the aversive feeling of being cognitively unengaged with one’s surroundings, and the low belief in one’s ability to achieve goals despite difficult circumstances, may be helpful in this regard. Cognitive behavioural therapy, which seeks to mitigate peoples’ avoidance, escape, and other safety behaviours, as well as increase behavioural activation and engagement in meaningful and valued activities, may be an important avenue for intervention, as it has shown to be effective for people struggling with depression symptoms and problematic drinking (e.g., Pedrelli et al., 2013; Riper et al., 2014). The use of a longitudinal design with four waves of assessment and with a large sample from across Canada, as well as the use of attention checks and validated measures to ensure accurate data on functioning, are key strengths of the current study. However, there are some shortcomings, including the lack of racial, provincial/territorial, and financial representation in our sample as the majority of participants identified as Caucasian, predominantly resided in Ontario, British Columbia, and Alberta, and had a high socioeconomic status. Additionally, we are precluded from extrapolating the current study’s findings and the implications for treating pandemic-related depression and alcohol problems to individuals with severe psychopathology as we did not examine (and control for) pre-existing mental health diagnoses in our community sample. Moreover, there was 32.42% overall attrition across waves and this could have reduced statistical power in the analyses that involved our smaller classes, such as the multinomial logistic regressions. Relatedly, our sample size could be viewed as modest for latent class growth analyses. Larger sample sizes would permit the use of other analytical approaches (see Kamata et al., 2018 for a review of the one- and three-step approaches) that are less likely to yield biased estimates of the relationships of the latent classes with auxiliary variables (i.e., covariates, outcomes) because they retain the classification measurement errors across models. Future studies on pandemic-related depression and alcohol problems that reproduce our results in more diverse and larger groups of participants, as well as that have lower attrition rates, will be important. In spite of these shortcomings, we obtained high classification probabilities in the current study, which, in tandem with prior findings (Frohlich et al., 2018; Orui et al., 2020), reinforces the distinct and meaningful classes of depression and alcohol problems that emerged in the data, as well as the risk factors for people with combined increases in depression and alcohol problems. As noted above, future studies should endeavour to understand exactly how features of personality (e.g., self-directed attention in boredom proneness and general self-efficacy) contribute to increased depression and alcohol problems, so that interventions can be tailored more incisively. Furthermore, longitudinal studies that assess peoples’ depression and alcohol problems beyond the waves that were examined in the current study will be critical in order to continue to understand the persisting and lingering effects of the pandemic on individuals’ psychological functioning. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 85 KB) Authors’ Contributions All authors contributed to the study’s conception and design. Material preparation and data collection were performed by Jeffrey D. Wardell and Matthew T. Keough. Data analyses were performed by Veerpal Bambrah and Matthew T. Keough. The first draft of the manuscript was written and revised by Veerpal Bambrah and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding This research was funded by the York University Generic Startup Funds. Data availability Subject to the ethical requirements of the York University Human Participants Research Committee and the Canadian Tri-Council policy statement on ethical conduct for research involving humans, all data, analysis code, and research materials underlying this study will be made available to researchers upon submission of an approved ethics protocol from their academic or research institution (or equivalent) to the corresponding author. Declarations Ethics approval Ethics approval for this study was obtained from the York University Human Participants Review Committee (Ethics certificate number: e2020-118). Consent to participate Informed consent was obtained from all individual participants. Conflicts of interest The authors have no conflicts of interests to disclose. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Acuff SF Strickland JC Tucker JA Murphy JG Changes in alcohol use during COVID-19 and associations with contextual and individual difference variables: A systematic review and meta-analysis Psychology of Addictive Behaviors 2022 36 1 1 19 10.1037/adb0000796 34807630 Adams T Rapinda KK Frohlich JR O'Connor RM Keough MT Impulsivity moderates the effect of social anxiety on in-lab alcohol craving Addictive Behaviors 2019 97 70 76 10.1016/j.addbeh.2019.05.025 31153094 Bambrah, V., Gurovich, S., Friedman, E., & Eastwood, J. D. (2020). What does trait boredom mean? Is it about how you feel or how you respond to boring situations? Paper presented at the 81st annual convention of the Canadian Psychological Association, Montreal. Bandura A Self-efficacy: The exercise of Control 1999 Freeman Bandura, A. (2010). Self-Efficacy. In The Corsini Encyclopedia of Psychology (4th ed., pp. 1–3). John Wiley & Sons, Inc. Baptist-Mohseni N Morris V Vedelago L Kempe T Rapinda K Mesmer E Bilevicius E Wardell JD MacKillop J Keough MT A longitudinal approach to understanding risk factors for problem alcohol use during the Covid-19 pandemic Alcoholism: Clinical and Experimental Research 2022 46 3 434 446 10.1111/acer.14774 35262939 Barari, S., Caria, S., Davola, A., Falco, P., Fetzer, T., Fiorin, S., Hensel, L., Ivchenko, A., Jachimowicz, J., King, G., Kraft-Todd, G., Ledda, A., MacLennan, M., Mutoi, L., Pagani, C., Reutskaja, E., Roth, C., & Slepoi, F. R. (2020). Evaluating covid-19 public health messaging in Italy: Self-reported compliance and growing mental health concerns. 10.1101/2020.03.27.20042820 Bench SW Lench HC Boredom as a seeking state: Boredom prompts the pursuit of novel (even negative) experiences Emotion 2019 19 2 242 254 10.1037/emo0000433 29578745 Brockmeyer, T., Zimmermann, J., Kulessa, D., Hautzinger, M., Bents, H., Friederich, H.-C., Herzog, W., & Backenstrass, M. (2015). Me, myself, and I: Self-referent word use as an indicator of self-focused attention in relation to depression and anxiety. Frontiers in Psychology, 6. 10.3389/fpsyg.2015.01564 Canadian Centre on Substance Use and Addiction (2020). Canadians under 54 drinking more while at home due to COVID-19 pandemic [Web site]. April 15, 2020. Available at: https://ccsa.ca/canadians-under-54-drinking-more-while-home-due-covid-19-pandemic. Accessed 5 June 2020. Capasso A Jones AM Ali SH Foreman J Tozan Y DiClemente RJ Increased alcohol use during the COVID-19 pandemic: The effect of mental health and age in a cross-sectional sample of social media users in the U.S Preventive Medicine 2021 145 106422 10.1016/j.ypmed.2021.106422 33422577 Carr A Positive psychology: The science of happiness and human strengths 2004 Routledge Carver CS You want to measure coping but your protocol is too long: Consider the Brief-COPE International Journal of Behavioral Medicine 1997 4 1 92 100 10.1207/s15327558ijbm0401_6 16250744 Carver CS Scheier MF Weintraub JK Assessing coping strategies: A theoretically based approach Journal of Personality and Social Psychology 1989 56 2 267 283 10.1037/0022-3514.56.2.267 2926629 Castillo-Carniglia A Keyes KM Hasin DS Cerdá M Psychiatric comorbidities in alcohol use disorder The Lancet Psychiatry 2019 6 12 1068 1080 10.1016/s2215-0366(19)30222-6 31630984 Cataudella S Carta SM Mascia ML Masala C Petretto DR Agus M Penna MP Teaching in times of the COVID-19 pandemic: A pilot study on teachers’ self-esteem and self-efficacy in an Italian sample International Journal of Environmental Research and Public Health 2021 18 15 8211 10.3390/ijerph18158211 34360503 Centre for Addiction and Mental Health (2020). COVID-19 national survey dashboard [Web site]. 2020. Available at: https://www.camh.ca/en/health-info/mental-health-and-covid-19/covid-19-national-survey. Accessed 8 June 2020. Chao M Chen X Liu T Yang H Hall BJ Psychological distress and state boredom during the COVID-19 outbreak in China: The role of meaning in life and media use European Journal of Psychotraumatology 2020 11 1 1769379 10.1080/20008198.2020.1769379 33029315 Chen G Gully SM Eden D Validation of a new general self-efficacy scale Organizational Research Methods 2001 4 1 62 83 10.1177/109442810141004 Chodkiewicz J Talarowska M Miniszewska J Nawrocka N Bilinski P Alcohol consumption reported during the covid-19 pandemic: The initial stage International Journal of Environmental Research and Public Health 2020 17 13 4677 10.3390/ijerph17134677 32610613 Danckert J Mugon J Struk A Eastwood JD Lench HA Boredom: What is it good for? The function of emotions: When and why emotions help us 2018 Springer International Publishing 93 121 Dias C Cruz JF Fonseca AM The relationship between multidimensional competitive anxiety, cognitive threat appraisal, and coping strategies: A multi-sport study International Journal of Sport and Exercise Psychology 2012 10 1 52 65 10.1080/1612197x.2012.645131 Dogan-Sander, E., Kohls, E., Baldofski, S., & Rummel-Kluge, C. (2021). More depressive symptoms, alcohol and drug consumption: Increase in mental health symptoms among university students after one year of the COVID-19 pandemic. Frontiers in Psychiatry, 12. 10.3389/fpsyt.2021.790974 Dozois DJ Anxiety and depression in Canada during the COVID-19 pandemic: A national survey Canadian Psychology/psychologie Canadienne 2021 62 1 136 142 10.1037/cap0000251 Droit-Volet, S., Gil, S., Martinelli, N., Andant, N., Clinchamps, M., Parreira, L., Rouffiac, K., Dambrun, M., Huguet, P., Dubuis, B., Pereira, B., Bouillon, J.-B., & Dutheil, F. (2020). Time and covid-19 stress in the lockdown situation: Time free, dying of boredom and sadness. PLOS ONE, 15(8). 10.1371/journal.pone.0236465 Duval S Wicklund RA A theory of objective self-awareness 1972 Academic Press Eastwood, J. D., & Bambrah, V. (2021). Self-focused but lacking self-insight: The relationship between boredom and self-consciousness. Paper presented at the 4th International Interdisciplinary Boredom Conference, Madrid. Eastwood, J. D., Cavaliere, C., Fahlman, S. A., & Eastwood, A. E. (2007). A desire for desires: Boredom and its relation to alexithymia. Personality and Individual Differences, 42(6), 1035–1045. 10.1016/j.paid.2006.08.027 Eisenberg, S. A., Shen, B.-J., Schwarz, E. R., & Mallon, S. (2012). Avoidant coping moderates the association between anxiety and patient-rated physical functioning in heart failure patients. Journal of Behavioral Medicine, 35(3), 253–261. 10.1007/s10865-011-9358-0 Fahlman SA Mercer-Lynn KB Flora DB Eastwood JD Development and validation of the Multidimensional State Boredom scale Assessment 2013 20 1 68 85 10.1177/1073191111421303 21903683 Fancourt D Steptoe A Bu F Trajectories of anxiety and depressive symptoms during enforced isolation due to covid-19 in England: A longitudinal observational study The Lancet Psychiatry 2021 8 2 141 149 10.1016/s2215-0366(20)30482-x 33308420 Frohlich JR Rapinda KK O'Connor RM Keough MT Examining co-patterns of depression and alcohol misuse in emerging adults following university graduation Addictive Behaviors Reports 2018 8 40 45 10.1016/j.abrep.2018.06.002 30003136 Gana, K., Deletang, B., & Metais, L. (2000). Is boredom proneness associated with introspectiveness? Social Behavior and Personality: An International Journal, 28(5), 499–504. 10.2224/sbp.2000.28.5.499 Gonzalez VM Reynolds B Skewes MC Role of impulsivity in the relationship between depression and alcohol problems among emerging adult college drinkers Experimental and Clinical Psychopharmacology 2011 19 4 303 313 10.1037/a0022720 21480733 Grant BF Goldstein RB Saha TD Chou SP Jung J Zhang H Pickering RP Ruan WJ Smith SM Huang B Hasin DS Epidemiology of DSM-5 alcohol use disorder: Results from the national epidemiologic survey on alcohol and related conditions JAMA Psychiatry 2015 72 8 757 10.1001/jamapsychiatry.2015.0584 26039070 Gurvich, C., Thomas, N., Thomas, E. H. X., Hudaib, A.-R., Sood, L., Fabiatos, K., Sutton, K., Isaacs, A., Arunogiri, S., Sharp, G., & Kulkarni, J. (2021). Coping styles and mental health in response to societal changes during the COVID-19 pandemic. International Journal of Social Psychiatry, 67(5), 540–549. 10.1177/0020764020961790 Harris, M. B. (2000). Correlates and characteristics of boredom proneness and boredom. Journal of Applied Social Psychology, 30(3), 576–598. 10.1111/j.1559-1816.2000.tb02497.x Havermans RC Vancleef L Kalamatianos A Nederkoorn C Eating and inflicting pain out of boredom Appetite 2015 85 52 57 10.1016/j.appet.2014.11.007 25447018 Higgins ET Self-discrepancy: A theory relating self and affect Psychological Review 1987 94 3 319 340 10.1037/0033-295x.94.3.319 3615707 Jung T Wickrama KA An introduction to latent class growth analysis and growth mixture modeling Social and Personality Psychology Compass 2007 2 1 302 317 10.1111/j.1751-9004.2007.00054.x Kamata, A., Kara, Y., Patarapichayatham, C., & Lan, P. (2018). Evaluation of analysis approaches for latent class analysis with auxiliary linear growth model. Frontiers in Psychology, 9. 10.3389/fpsyg.2018.00130 Kar N Kar B Kar S Stress and coping during COVID-19 pandemic: Result of an online survey Psychiatry Research 2021 295 113598 10.1016/j.psychres.2020.113598 33264677 Karpyak VM Biernacka JM Geske JR Abulseoud OA Brunner MD Chauhan M Hall-Flavin DK Lewis KA Loukianova LL Melnyk GJ Onsrud DA Proctor BD Schneekloth TD Skime MK Wittkopp JE Frye MA Mrazek DA Gender-specific effects of comorbid depression and anxiety on the propensity to drink in negative emotional states Addiction 2016 111 8 1366 1375 10.1111/add.13386 27009547 Keough MT Badawi G Nitka D O'Connor RM Stewart SH Impulsivity increases risk for coping-motivated drinking in undergraduates with elevated social anxiety Personality and Individual Differences 2016 88 45 50 10.1016/j.paid.2015.08.036 28066094 Khantzian EJ The self-medication hypothesis of substance use disorders: A reconsideration and recent applications Harvard Review of Psychiatry 1997 4 5 231 244 10.3109/10673229709030550 9385000 Khantzian EJ Reflections on treating addictive disorders: A psychodynamic perspective The American Journal on Addictions 2012 21 3 274 279 10.1111/j.1521-0391.2012.00234.x 22494231 Kiluk BD Dreifuss JA Weiss RD Morgenstern J Carroll KM The short inventory of problems – revised (SIP-R): Psychometric Properties within a large, diverse sample of substance use disorder treatment seekers Psychology of Addictive Behaviors 2013 27 1 307 314 10.1037/a0028445 22642856 Kohls, E., Baldofski, S., Moeller, R., Klemm, S.-L., & Rummel-Kluge, C. (2021). Mental health, social and emotional well-being, and perceived burdens of university students during COVID-19 pandemic lockdown in Germany. Frontiers in Psychiatry, 12. 10.3389/fpsyt.2021.643957 Kroenke K Spitzer RL Williams JB The PHQ-9: Validity of a brief depression severity measure Journal of General Internal Medicine 2001 16 9 606 613 10.1046/j.1525-1497.2001.016009606.x 11556941 Leventhal, A. M., Cho, J., Ray, L. A., Liccardo Pacula, R., Lee, B. P., Terrault, N., Pedersen, E., Lee, J. O., Davis, J. P., Jin, H., Huh, J., Wilson, J. P., & Whaley, R. C. (2022). Alcohol use trajectories among U.S. adults during the first 42 weeks of the Covid‐19 pandemic. Alcoholism: Clinical and Experimental Research. 10.1111/acer.14824 Luszczynska A Gutiérrez-Doña B Schwarzer R General self-efficacy in various domains of human functioning: Evidence from five countries International Journal of Psychology 2005 40 2 80 89 10.1080/00207590444000041 Marani, M., Katul, G. G., Pan, W. K., & Parolari, A. J. (2021). Intensity and frequency of extreme novel epidemics. Proceedings of the National Academy of Sciences, 118(35). 10.1073/pnas.2105482118 McCurdy, A., Stearns, J. A., Rhodes, R. E., Hopkins, D., Mummery, K., & Spence, J. C. (2022). Relationships between physical activity, boredom proneness, and subjective well-being among U.K. adults during the covid-19 pandemic. Journal of Sport & Exercise Psychology, 1–9. 10.1123/jsep.2021-0253 McPhee, M. D., Keough, M. T., Rundle, S., Heath, L. M., Wardell, J. D., & Hendershot, C. S. (2020). Depression, environmental reward, coping motives and alcohol consumption during the covid-19 pandemic. Frontiers in Psychiatry, 11. 10.3389/fpsyt.2020.574676 Muthen B Muthen LK Integrating person-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes Alcoholism: Clinical and Experimental Research 2000 24 6 882 891 10.1111/j.1530-0277.2000.tb02070.x 10888079 Nylund KL Asparouhov T Muthén BO Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study Structural Equation Modeling: A Multidisciplinary Journal 2007 14 4 535 569 10.1080/10705510701575396 Orui K Frohlich JR Stewart SH Sherry SB Keough MT Examining subgroups of depression and alcohol misuse in emerging adults during university: A replication and extension study International Journal of Mental Health and Addiction 2020 19 6 2323 2341 10.1007/s11469-020-00325-w Palan S Schitter C Prolific.ac—a subject pool for online experiments Journal of Behavioral and Experimental Finance 2018 17 22 27 10.1016/j.jbef.2017.12.004 Pedrelli P Borsari B Palm KM Dalton E Fava M Combined MI + CBT for depressive symptoms and binge drinking among young adults: Two case studies Journal of Cognitive Psychotherapy 2013 27 3 235 257 10.1891/0889-8391.27.3.235 25170188 Pedrelli P Shapero B Archibald A Dale C Alcohol use and depression during adolescence and young adulthood: A summary and interpretation of mixed findings Current Addiction Reports 2016 3 1 91 97 10.1007/s40429-016-0084-0 27162708 Polak K Haug NA Drachenberg HE Svikis DS Gender considerations in addiction: Implications for treatment Current Treatment Options in Psychiatry 2015 2 3 326 338 10.1007/s40501-015-0054-5 26413454 Pollard, M. S., Tucker, J. S., & Green, H. D. (2020). Changes in adult alcohol use and consequences during the COVID-19 pandemic in the US. JAMA Network Open, 3(9). 10.1001/jamanetworkopen.2020.22942 Poulus, D., Coulter, T. J., Trotter, M. G., & Polman, R. (2020). Stress and coping in esports and the influence of mental toughness. Frontiers in Psychology, 11. 10.3389/fpsyg.2020.00628 Pressley T Ha C Teaching during a pandemic: United States Teachers' self-efficacy during COVID-19 Teaching and Teacher Education 2021 106 103465 10.1016/j.tate.2021.103465 34754135 Raftery AE Bayesian model selection in Social Research Sociological Methodology 1995 25 111 10.2307/271063 Ram N Grimm KJ Methods and measures: Growth mixture modeling: A method for identifying differences in longitudinal change among unobserved groups International Journal of Behavioral Development 2009 33 6 565 576 10.1177/0165025409343765 23885133 Reynolds DL Garay JR Deamond SL Moran MK Gold W Styra R Understanding, compliance and psychological impact of the SARS quarantine experience Epidemiology and Infection 2008 136 7 997 1007 10.1017/s0950268807009156 17662167 Rights JD Sterba SK Quantifying explained variance in multilevel models: An integrative framework for defining R-squared measures Psychological Methods 2019 24 309 338 10.1037/met0000184 29999378 Riper H Andersson G Hunter SB Wit J Berking M Cuijpers P Treatment of comorbid alcohol use disorders and depression with cognitive-behavioural therapy and Motivational interviewing: A meta-analysis Addiction 2014 109 3 394 406 10.1111/add.12441 24304463 Schmits E Glowacz F Changes in alcohol use during the covid-19 pandemic: Impact of the lockdown conditions and mental health factors International Journal of Mental Health and Addiction 2022 20 2 1147 1158 10.1007/s11469-020-00432-8 33424513 Seib, H. M., & Vodanovich, S. J. (1998). Cognitive correlates of boredom proneness: The role of private selfconsciousness and absorption. The Journal of Psychology, 132(6), 642–652. 10.1080/00223989809599295 Shamblaw AL Rumas RL Best MW Coping during the COVID-19 pandemic: Relations with Mental Health and quality of life Canadian Psychology/psychologie Canadienne 2021 62 1 92 100 10.1037/cap0000263 Shield KD Chrystoja BR Ali S Sohi I Rehm J Nigatu YT Elton-Marshall T Hamilton H Jankowicz D Wells S Changes in alcohol consumption in Canada during the COVID-19 pandemic: Associations with anxiety and self-perception of depression and loneliness Alcohol and Alcoholism 2022 57 2 190 197 10.1093/alcalc/agab055 34387658 Simonetti V Durante A Ambrosca R Arcadi P Graziano G Pucciarelli G Simeone S Vellone E Alvaro R Cicolini G Anxiety, sleep disorders and self-efficacy among nurses during COVID-19 pandemic: A Large Cross-Sectional Study Journal of Clinical Nursing 2021 30 9–10 1360 1371 10.1111/jocn.15685 33534934 Spoorthy MS Pratapa SK Mahant S Mental health problems faced by healthcare workers due to the COVID-19 pandemic–A Review Asian Journal of Psychiatry 2020 51 102119 10.1016/j.ajp.2020.102119 32339895 Steenbarger BN Aderman D Objective self-awareness as a nonaversive state: Effect of anticipating discrepancy reduction Journal of Personality 1979 47 2 330 339 10.1111/j.1467-6494.1979.tb00206.x Struk AA Carriere JS Cheyne JA Danckert J A short boredom proneness scale: Development and psychometric properties Assessment 2017 24 3 346 359 10.1177/1073191115609996 26467085 Sun Y Song H Liu H Mao F Sun X Cao F Occupational stress, mental health, and self-efficacy among community mental health workers: A cross-sectional study during COVID-19 pandemic International Journal of Social Psychiatry 2020 67 6 737 746 10.1177/0020764020972131 33176527 Tucker JS Rodriguez A Green HD Pollard MS Trajectories of alcohol use and problems during the COVID-19 pandemic: The role of social stressors and drinking motives for men and women Drug and Alcohol Dependence 2022 232 109285 10.1016/j.drugalcdep.2022.109285 35033955 Volken T Zysset A Amendola S Klein Swormink A Huber M von Wyl A Dratva J Depressive symptoms in Swiss university students during the COVID-19 pandemic and their correlates International Journal of Environmental Research and Public Health 2021 18 4 1458 10.3390/ijerph18041458 33557193 von Gemmingen, M. J., Sullivan, B. F., & Pomerantz, A. M. (2003). Investigating the relationships between boredom proneness, paranoia, and self-consciousness. Personality and Individual Differences, 34(6), 907–919. 10.1016/s0191-8869(01)00219-7 Wardell, J., Kempe, T., Rapinda, K. K., Single, A. N., Bilevicius, E., Frohlich, J. R., Hendershot, C. S., & Keough, M. T. (2020). Drinking to cope during the COVID-19 pandemic: The role of external and internal stress-related factors in coping motive pathways to alcohol use, solitary drinking, and alcohol problems. Alcoholism: Clinical and Experimental Research. 10.31234/osf.io/8vfp9 Weiss ER Todman MW Maple E Bunn RR Boredom in a time of uncertainty: State and trait boredom’s associations with psychological health during COVID-19 Behavioral Sciences 2022 12 8 298 10.3390/bs12080298 36004869 Westgate, E. C., Buttrick, N., Lin, Y., El Helou, G., Agostini, M., Belanger, J., … Damnjanović, K. (2022). Pandemic boredom: Little evidence that lockdown-related boredom affects risky public health behaviors across 116 countries. 10.31234/osf.io/78kma Wickrama KAS Lee TK O’Neal CW Lorenz FO Higher-order growth curves and mixture modeling with mplus: A practical guide 2016 Routledge Williams G Kibowski F Glenwick D Jason L Latent class analysis and latent profile analysis Handbook of methodological approaches to community-based research 2016 Oxford University Press 143 151 Woicik PA Stewart SH Pihl RO Conrod PJ The substance use risk profile scale: A scale measuring traits linked to reinforcement-specific substance use profiles Addictive Behaviors 2009 34 12 1042 1055 10.1016/j.addbeh.2009.07.001 19683400 Wood R Bandura A Impact of conceptions of ability on self-regulatory mechanisms and complex decision making Journal of Personality and Social Psychology 1989 56 3 407 415 10.1037/0022-3514.56.3.407 2926637 World Health Organization (2022). Coronavirus disease (COVID-19) pandemic. [Web site]. 2020. Available at: https://www.who.int/europe/emergencies/situations/covid-19. Accessed 1 July 2022. Xue A Oros V La Marca-Ghaemmaghami P Scholkmann F Righini-Grunder F Natalucci G Karen T Bassler D Restin T New parents experienced lower parenting self-efficacy during the COVID-19 pandemic lockdown Children 2021 8 2 79 10.3390/children8020079 33498886 Yan L Gan Y Ding X Wu J Duan H The relationship between perceived stress and emotional distress during the covid-19 outbreak: Effects of boredom proneness and coping style Journal of Anxiety Disorders 2021 77 102328 10.1016/j.janxdis.2020.102328 33160275 Yıldırım M Güler A Covid-19 severity, self-efficacy, knowledge, preventive behaviors, and mental health in Turkey Death Studies 2020 46 4 979 986 10.1080/07481187.2020.1793434 32673183

0

PMC9748906

NO-CC CODE

2022-12-15 23:22:48

no

Curr Psychol. 2022 Dec 14;:1-17

utf-8

Curr Psychol

10.1007/s12144-022-04109-4

oa_other

==== Front Chem Phys Lipids Chem Phys Lipids Chemistry and Physics of Lipids 0009-3084 1873-2941 Elsevier B.V. S0009-3084(22)00006-8 10.1016/j.chemphyslip.2022.105178 105178 Research Paper Synthesis and bioactivity of readily hydrolysable novel cationic lipids for potential lung delivery application of mRNAs Pei Yihua Bao Yanjie Sacchetti Cristiano Brady Juthamart Gillard Kyra Yu Hailong Roberts Scott Rajappan Kumar ⁎ Tanis Steven P. ⁎ Perez-Garcia Carlos G. Chivukula Padmanabh Karmali Priya P. Arcturus Therapeutics, 10628 Science Center Drive, Suite 250, San Diego, CA 92121, USA ⁎ Corresponding authors. 3 2 2022 3 2022 3 2 2022 243 105178105178 20 8 2021 27 1 2022 31 1 2022 © 2022 Elsevier B.V. All rights reserved. 2022 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Lipid nanoparticles (LNPs) mediated mRNA delivery has gained prominence due to the success of mRNA vaccines against Covid-19, without which it would not have been possible. However, there is little clinical validation of this technology for other mRNA-based therapeutic approaches. Systemic administration of LNPs predominantly targets the liver, but delivery to other organs remains a challenge. Local approaches remain a viable option for some disease indications, such as Cystic Fibrosis, where aerosolized delivery to airway epithelium is the preferred route of administration. With this in mind, novel cationic lipids (L1-L4) have been designed, synthesized and co-formulated with a proprietary ionizable lipid. These LNPs were further nebulized, along with baseline control DOTAP-based LNP (DOTAP+), and tested in vitro for mRNA integrity and encapsulation efficiency, as well as transfection efficiency and cytotoxicity in cell cultures. Improved biodegradability and potentially superior elimination profiles of L1-L4, in part due to physicochemical characteristics of putative metabolites, are thought to be advantageous for prospective therapeutic lung delivery applications using these lipids. Graphical Abstract ga1 Keywords Lipid nanoparticles mRNA Cationic lipids DOTAP LNP Lung Nebulization ==== Body pmc1 Introduction Drug delivery to the proper cell types and tissues of interest is a major burden in any viral/nonviral therapeutic approach. In recent years, delivery of nonviral mediated RNA (mRNA, siRNA, miRNA etc.) has proven to be successful in the clinic (MacLachlan, 2007, Semple et al., 2010, Jayaraman et al., 2012, Sabnis et al., 2018, Nabhan et al., 2016, Patel et al., 2017). One of the successful platforms of nonviral RNA delivery is using lipid nanoparticles (LNPs), as evidenced by the approval and marketing of Patisiran (ONPATTRO®), to treat transthyretin amyloidosis, (ONPATTRO FDA Label, 2021; ONPATRRO EMA, Summary of opinion, 2018) and lately by the Emergency Use Authorization (EUA) of two of the much-anticipated mRNA vaccines against Covid-19 administered intramuscularly. Much of the therapeutic success with this platform has been attributed to the efficient uptake of these LNPs in the liver through an ApoE/LDLR mediated endocytosis by the hepatocytes (Akinc et al., 2010). There is a compelling need for the therapeutic delivery of LNPs to other organs and cell types. Most recently, there has been some efforts in selectively improving tissue targeting by modifying LNP compositions and properties (Cheng et al. 2020). For example, this paper has demonstrated selective lung delivery when DOTAP was used at a high molar percentage of 50%. Yet, another recent publication has demonstrated nebulized lung delivery of LNPs using high C14PEG2000 molar percentage in the formulation (Lokugamage et al., 2021). In addition to these latest reports, others also have explored targeted delivery of cationic LNPs directly to airway epithelium using cationic lipids as excipients in formulating inhalable LNPs for the efficient deposition of complex RNA therapeutic modalities including, but not limited to, siRNA, miRNA, ASO, mRNA, CRISPR-sgRNA, tRNA and long non-coding RNA (Merkel et al., 2014, Jensen et al., 2012, Jensen et al., 2010, Conti et al., 2014, Kusumoto et al., 2013, Robinson et al., 2018) and for gene delivery applications (Allon et al., 2012). This approach is specifically suitable for administering mRNA encoding functional proteins whose deficit drives diseases such as, for example, Cystic Fibrosis (CF) where restoration of a deficient ion channel represents a major therapeutic development paradigm (Riordan et al., 1989, Welsh, 1990, Chu et al., 1993, Welsh et al., 1992). Therefore, considering CF as a model to further design optimal LNPs using cationic lipids L1-L4, we should consider the factors that will also influence their design such as formulation optimization, nebulization, and physicochemical characterization. Such an approach is needed for better optimal LNPs for lung therapeutics. LNPs principally contain an ionizable lipid that not only drives RNA encapsulation through their pH driven ability to form ion-pair interactions with the anionic phosphate backbone of RNA, but also aids in triggering endosomal escape through membrane destabilization (Kulkarni et al., 2018, Semple et al., 2001, Semple et al., 2010, Viger-Gravel et al., 2018, Lechanteur et al., 2018, Zelphati and Szoka, 1996, Sabnis et al., 2018). Other typical constituents of LNPs are a phospholipid and cholesterol to maintain structural integrity, and an outer polyethylene glycol (PEG) lipid to coat the LNP (Kulkarni et al., 2018, Semple et al., 2001, Leung et al., 2012, Crawford et al., 2011). This outer decoration protects LNPs from the host immune response, imparts serum stability and is cleavable once inside the target cell ( Fig. 1 A). Typically, the ionizable lipids used in LNP formulations are neutral at physiological pH, and hence they possess the characteristics of a neutral liposome. Neutral liposomes of < 100 nm hydrodynamic size have an overwhelming bias towards hepatocytes. Previously, Yan et al. have demonstrated the plasma apolipoprotein E (ApoE) dependent enhanced plasma clearance of neutral liposomes by measuring a 3.6-fold faster rate of clearance vis-à-vis negatively charged liposomes in wild type mice versus ApoE deficient mice (Yan et al., 2005). This was further supported by Akinc et al. (Akinc et al., 2010) by showing the enhanced liver uptake of ionizable LNP (iLNP) compared to cationic LNP (cLNP) in the presence of ApoE. This necessitates alternative strategies for extrahepatic delivery of LNPs. In this regard, lung and lung airway is an organ of great therapeutic relevance that can be targeted by LNPs. An experimentally tested and potentially clinically viable approach is to use cationic LNP (cLNP), formed using cationic lipids such as DOTAP, to target lung airway epithelium. But systemic administration of cLNP has not proven to target lung for non-tumor therapeutic applications. For example, a DOTAP:Cholesterol based cationic liposome has been proven to deliver anticancer p53 gene to lung upon systemic administration (Ramesh et al. 2001). Another gene transfer study (Fletcher et al. 2006) demonstrates a 3-fold stronger gene expression in mouse lungs upon tail vein IV administration of cationic:cholesterol lipoplexes for potential antitumor therapy. Lung delivery of cationic LNPs through systemic administration is limited to lung tumors because of the enhanced permeation and retention (EPR) effect of the tumor mass (Prabhakar et al. 2013). For non-tumor indications, a local cLNP mediated mRNA delivery to airway epithelial cells needs to be carefully designed taking into consideration the unique attributes of the cellular milieu and its reduced capability to metabolize various chemical components of the LNPs.Fig. 1 Overview of the Composition and Properties of a Representative Lipid Nanoparticle (LNP) Targeting the Lung. A: Lipid components will include an ionizable lipid, a phospholipid, cholesterol, and a PEG lipid. LNP-targeting the lung include an additional cationic lipid. A conceptual inside view of an LNP is also represented, where the mRNA is shown as a blue ribbon. B: Physicochemical properties DOTAP and L1-L4 are given. cLogP represent hydrophobicity, ao is the total polar surface area, lc is the carbon chain length, V is the volume occupied by each lipid when it is packed into LNP, P is the critical packing parameter that represents the lamellar structure formed by these lipids. a Calculated using ACD Labs Structure Designer v12.0. cLogP was calculated using ACD Labs version B. b Calculated using Molinspiration Chemoinformatics available on the internet. Å is the unit in Angstrom. Fig. 1 Such a design of LNP for targeting lung airway would involve three key considerations. First, packaging mRNA in a lipid composition that distributes favorably to the airway epithelium as opposed to trafficking to liver must be designed. Second, an ionizable lipid must be included to facilitate endosomal release through formation of destabilizing non‐bilayer structures at acidic pH of the endosome. Third, a reliable route or method of drug administration that ensures lung deposition needs to be applied. Cationic lipids have been determined to be the lipids of choice for lung delivery of plasmid DNA (Wheeler et al., 1996, Bragonzi et al., 1999, Lee et al., 1996), siRNA (Andey et al. 2019), and antisense oligonucleotides (Ma et al., 2002). One of the most often used cationic lipids for this application is 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP, when DOTAP+ is used it refers to LNPs using DOTAP), Fig. 1B). DOTAP was originally developed as a biodegradable substitute for DOTMA, which proved to be toxic due to poor biodegradability owing to the presence of stable ether bonds (Leventis and Silvius, 1990). Even though DOTAP turned out to be a less toxic transfection reagent compared to DOTMA, it is not necessarily an optimal choice. Chemically, DOTAP is a pseudo glyceryl lipid and as such, exists as a racemate or as separate R- and S-enantiomers (Supplementary Figure 1). Unlike the natural glyceryl lipids, the DOTAP typically used in LNP formulations is racemic. Even though siRNA lipoplex formulations with enantiomerically pure R-DOTAP has shown some improved knock down potency compared to S- or racemic DOTAP (Terp et al., 2012), the differential biodegradability profile of the individual enantiomers has not been reported. Also, once any putative DOTAP analogues had enabled the delivery of their cargo to the target epithelial cells they would need to be cleared from the lung. This clearance would likely require ester hydrolysis, and perhaps further metabolism, for all the DOTAP like structure to be ultimately eliminated. We designed four DOTAP like functional analogues (L1-L4) that are achiral, not pseudoglyceryl lipids, and may have improved biodegradability and comparable or better transfection efficiency (Fig. 1B). We surmised that the structural proximity of the ester linkages to the sterically demanding glycerol-like terminus of DOTAP would result in steric congestion that would slow enzymatic hydrolysis of the ester linkages. The 19–20 atom extensions from the branch point carrying the tetra-alkylammonium head group exhibited by DOTAP would be the starting point for modification, with hydrolytically labile ester groups positioned farther from the sterically demanding quaternary carbon. Our design called for ester-transposition, relative to the structure of DOTAP, with a Z-2-alkenol forming an ester with a polar, and symmetrical, 8-oxo-octanedioic acid derived core. The quaternary ammonium moiety was seen as being introduced as an ammonium ethyl ether and as an ammonium-alkyl entity, both derived from the ketone at the 9-position. Physicochemical parameters of the designed cationic lipids would be compared to DOTAP itself and we anticipated that a post in vivo analysis of the performance of our new lipids vs DOTAP+ would point us toward the preferred physicochemical space. Table 1 presents the structure of DOTAP and cationic lipids L1-L4, designed as described above and selected as potential targets along with their associated calculated LogP (c-LogP) values. DOTAP, with a 1-carbon spacer from the branch to the quaternary ammonium nitrogen is associated with a c-LogP = 11.77. Lipid L1, with symmetrical 20-atom extensions from the branch point, ester carbonyls at 8-atoms from the branch and esters formed with (Z)− 2-undecenol and a 2-dimethylaminoethanol based quaternary ammonium head group is similarly lipophilic (c-LogP = 11.34). Lipids L2, L3, and L4 are also derived from symmetrical esters, 8 atoms removed from the branch point, and they differ with respect to the esterifying alcohol (L2 and L4 (Z)− 2 dodecen-1-ol; 3 (Z)− 2-undecen-1-ol) and carbon chain length between the branch point and the quaternary ammonium head group (L2 2-C; L3 and L4 3-C). The c-LogP values for L2, L3, and L4 are 10.85, 10.44, and 11.46, respectively. While L1-L4 are very similar in size and chain extension, the presence of the ether linkage in L1 and the subtly different carbon chain lengths found in the head group and ester moieties of L2-L4 leads to c-LogP values that are similar to DOTAP (L1 and L4) and 1–1.5 log units lower (L2 and L3). The impact of these lipophilicity values will be of interest with respect to formulation into LNPs, cellular transfection, and stability. Further, we wanted to ensure that the critical packing parameters (P) of these modified cationic lipids conformed to the cylindrical shape that enabled the formation of lamellar structure. This imposed restrictions on P to be below 1 and greater than 0.5 (0.5 <P < 1). Our calculations according to the equation P = V/a 0 l c, where V is the molecular volume, a 0 is the polar surface area (PSA) and l c is the carbon chain length, ensured that this was indeed the case (Hsu et al., 2005). In fact, all four new cationic lipids conformed to this requirement (Fig. 1B) with the P of L1 being 0.52 while the value of L2, L3, and L4 identical to that of DOTAP. We report here on the synthesis, biodegradability, transfection efficiency, and cytotoxicity of novel cationic lipids L1-L4 that are putative functional analogs of DOTAP, in our proprietary LNP formulations for prospective lung airway delivery applications.Table 1 Initial and Post Freeze-Thaw Characteristics of LNPs with DOTAP-, DOTAP+ and L1-L4. The physicochemical properties of the LNPs were not much impacted with freeze/thaw and remained within the expected 10% variability. DOTAP- represents formulation without DOTAP, but with ionizable lipid. Data for an initial LNP characterization without the DOTAP- control is given in Supplementary Table 1 to show the consistency of particle properties. Three independent measurements were performed for particle size and PDI. *Data represented as average±std. Table 1 Bulk characterization (Pre-Freeze/Thaw) Characterization Post Freeze/Thaw Lipid Particle Size* (nm) PDI* %Encap mRNA purity Particle Size* (nm) PDI* %Encap mRNA purity (%) DOTAP- 74.7 ± 1.4 0.07 ± 0.01 71.3 84 76.8 ± 0.6 0.07 ± 0.02 73 84 DOTAP+ 74.8 ± 0.2 0.07 ± 0.01 99.8 89 75.5 ± 0.3 0.08 ± 0.01 99.7 91 L1 66.3 ± 0.3 0.09 ± 0.03 100.2 82 68.5 ± 0.2 0.11 ± 0.02 100.2 90 L2 70.3 ± 0.8 0.10 ± 0.03 99.9 87 70.9 ± 0.5 0.10 ± 0.02 100.1 91 L3 67.8 ± 0.8 0.08 ± 0.01 99.9 89 69.6 ± 1.0 0.11 ± 0.00 100 84 L4 68.3 ± 1.6 0.08 ± 0.01 99.8 83 69.5 ± 0.4 0.11 ± 0.01 100 90 2 Materials and methods 2.1 Materials and methods of synthesis 2.1.1 General Starting materials and other reagents were purchased from commercial suppliers and were used without further purification unless otherwise indicated. All reactions were performed under a positive pressure of nitrogen, argon, or with a drying tube, at ambient temperature (unless otherwise stated), in anhydrous solvents, unless otherwise indicated. The reactions were assayed by high-performance liquid chromatography (HPLC) and terminated as judged by the consumption of starting material. 1H NMR spectra were recorded on Varian or Bruker instruments operating at the field strength indicated. 1H NMR spectra are obtained as DMSO-d 6 or CDCl3 solutions as indicated (reported in ppm), using TMS as the reference. Other NMR solvents were used as needed. When peak multiplicities are reported, the following abbreviations are used: s = singlet, d = doublet, t = triplet, m = multiplet, br = broadened, dd = doublet of doublets, dt = doublet of triplets, q = quartet. Coupling constants, when given, are reported in hertz. NMR spectra of individual test compounds can be found in the Supplementary Material. Mass spectra were obtained using liquid chromatography mass spectrometry (LC-MS) on a Schimadzu instrument using atmospheric pressure chemical ionization (APCI) or electrospray ionization (ESI). Mass spectra were also measured by direct injection on a Perkin Elmer PE-SCIEX API-150 instrument or Agilent-TRAP XCT instrument using electrospray (ESI) ionization. All test compounds showed > 95% purity as determined by high-performance liquid chromatography (HPLC). HPLC conditions were as follows: Agilent 1290 Infinity; Halo RP-Amide, 3.00 mm × 100 mm, 2.7 µm, 30 °C; 10% → 90% CH3CN, 0.05% perchloric acid/H2O to 90% → 10% CH3CN, 0.05% perchloric acid/H2O, 30 min run, flow rate 1.0 mL/min, UV detection (λ = 205 nm), or 10% → 90% CH3CN, 0.05% perchloric acid/H2O to 90% → 10% CH3CN, 0.05% perchloric acid/H2O, 30 min run, flow rate 1.0 mL/min, ELSD detection. HPLC analyses of test compounds can be found in the Supporting Information. 2.1.2 Synthesis of cationic lipids L1-L4 2.1.2.1 Synthesis of common intermediate 6 Cationic lipids L1-L4 are prepared from a known common 9-oxo-heptanedioic core unit, diester 6 which was readily prepared as shown in Equation 1. Dialkylation of TOSMIC (Mueller et al., 2004) with ethyl 8-bromo-octanoate provided crude 5, which upon acid hydrolysis, afforded keto-diester 6. For detailed synthesis protocol, please see Supplementary material.(1) Image 1 Equation 1. Synthesis of the common intermediate, diethyl ester of 9-oxoheptanedoic acid (6). 2.1.2.2 Synthesis of L1 The synthesis of DOTAP analog L1 is shown in Scheme 1. Reduction of the keto group of 6 provided alcohol 7 which was alkylated to afford allyl ether 8. Dihydroxylation gave a diol (VanRheenan et al., 1976) which was cleaved in situ to yield aldehyde 9. Reductive amination with dimethylamine gave amine 10 which was hydrolyzed to diacid 11. Esterification with (Z)− 2-undecen-1-ol (Cerutti-Delasalle et al., 2016) provided diester 12. Finally, methylation of 12 and ion exchange (Amberlite® IRA-400(Cl)) afforded L1. For detailed synthesis protocol, please see Supplementary material.Scheme 1 Synthesis of L1: Reagents and Conditions (a) NaBH4, MeOH, rt, 12 h, 98%; (b) Allyl bromide, NaH, DMF:THF (1:1 v/v), 30 oC, 4 h, 35%; (c) i. K2OsO4, NMO, THF:H2O (2:1 v/v); ii. NaIO4, rt; (d) dimethylamine, NaBH3CN, MeOH, RT, 72% for 2 steps; (e) NaOH, EtOH, rt, 2 h; f) Z-2-undecen-1-ol, EDC, DMAP, CH2Cl2, rt, 16 h, 84% for 2 steps; g) CH3I, CH3CN, rt, 3 h; h) Amberlite® IRA-400(Cl), CH3CN:H2O (ca. 65:35 v/v), rt, 4 h, 72% over 2 steps. Scheme 1 2.1.2.3 Synthesis of L2 The synthesis of analog L2 was initiated from keto-diester, as shown in Scheme 2. Diethyl ester 6 was hydrolyzed and the resulting diacid was esterified to provide bis-t-butyl ester 13. Reformatsky reaction with the zincate derived from ethyl bromoacetate followed by dehydration with the Burgess reagent (Burgess et al., 1970) gave chain extended ester 14. Double bond reduction gave saturated ethyl ester 15 which suffered selective ethyl ester reduction upon exposure to LiBHEt3 to provide alcohol 16. Swern oxidation (Omura and Swern, 1978) and reductive amination of the resulting aldehyde afforded dimethyl amine 18. Hydrolysis of bis-t-butyl ester 17 and bis esterification with (Z)− 2-dodecen-1-ol (Yoshida et al., 2007) yielded 18. Finally, the desired quaternary ammonium chloride analog L2 was then realized from the reaction of 18 with CH3Cl. For detailed synthesis protocol, please see Supplementary material.Scheme 2 Synthesis of L2. Reagents and conditions: (a) LiOH, THF, H2O, RT 43%; (b) i. (COCl)2, cat. DMF, CH2Cl2, 0–5 °C to RT, ii. t-BuOH, 40 °C 44%; (c) Ethyl bromoacetate, Zn, TMSCl, Et2O, 30 °C; (d) Burgess reagent, toluene, 30 °C, 61% over 2 steps; (e) H2 (1 atm), Pt2O, CH3CO2H, 40 °C 95%; (f) LiBHEt3, THF, − 5 °C 71%; (g) i. (COCl)2, DMSO, CH2Cl2, − 75 °C, ii. Et3N − 75 °C to RT; (h) (CH3)2NH-HCl, NaOAc, NaBH(OAc)3, CH3CO2H 52% over 2 steps; (i) HCl-dioxane, CH2Cl2, RT; (j) (Z)− 2-dodecen-1-ol, EDCl, DMAP, CH2Cl2, 0 °C to RT 58% over 2 steps; (k) CH3Cl, CH3CN, RT 99%. Scheme 2 2.1.2.4 Synthesis of L3 and L4 Lipids L3 and L4 were prepared from the common intermediate 6 as presented in Scheme 3. Reaction of 6 with 3-butenylmagnesium chloride gave alcohol 19 which was reduced with Et3SiH and BF3 .OEt2 to afford alkene 20. Dihydroxylation and cleavage of the diol, as described for the preparation of 1 in Scheme 1, provided crude aldehyde 21 which was reductively aminated ((CH3)2NH-HCl, NaBH(OAc)3), leading to dimethyl amine 22. Ester hydrolysis provided diacid 23 which was bis-esterified with either (Z)− 2-undecen-1-ol or (Z)− 2-dodecen-1-ol which led to diesters 24 and 25, respectively. Finally, lipids L3 and L4 were prepared upon the methylation of 24 and 25 to CH3Cl. For detailed synthesis protocol, please see Supplementary material.Scheme 3 Synthesis of L3 and L4. Reagents and conditions: (a) 3-butenylmagnesium chloride, THF, − 20 °C; (b) Et3SiH, BF3.Et2O, CH2Cl2, RT, 58% over 2 steps; (c) i. K2OsO4, NMO, THF:H2O (2:1 v/v); ii. NaIO4, rt; (d) (CH3)2NH-HCl, NaOAc, NaBH(OAc)3, THF-CH3CO2H 40% over 2 steps; (e) NaOH, EtOH, RT, crude; (f) (Z)− 2-undecen-1-ol, EDCl, DMAP, CH2Cl2, 0 °C to RT, 24 41% over 2 steps; or (Z)− 2-dodecen-1-ol, EDCl, DMAP, CH2Cl2, 0 °C to RT, 25 46% over 2 steps; (g) CH3Cl, CH3CN, RT, 62% and 85% respectively for L3 and L4. Scheme 3 2.2 Methods 2.2.1 RNA Formulation into Lipid Nanoparticles and nebulization The LNPs were prepared, in a manner similar to that reported (Rajappan et al., 2020, Ramaswamy et al., 2017), by mixing appropriate volumes of lipids in ethanol with an aqueous phase containing TdTomato-mRNA using an Arcturus proprietary mixing module, followed by downstream processing. In all these formulations a proprietary ionizable lipid was also included to facilitate endosomal release of the endocytosed particles. Briefly, lipid excipients including DOTAP or L1-L4, proprietary ionizable lipid (Rajappan et al., 2020), DSPC, cholesterol, and PEG2000-DMG were dissolved in ethanol at a molar ratio of 25:25:10:38.5:1.5. The lipids were rapidly mixed with the aqueous TdTomato-mRNA solution prepared at pH 3.5 citrate buffer at a flow rate ratio of 1:3 (ethanol:citrate v/v) using a proprietary mixing module to keep the ethanol percentage constant at 25%. The formed LNPs were stabilized by sequential dilution with pH 6.0 phosphate buffer, followed by pH 8.0 HEPES buffer. To concentrate the formulation, the diluted formulations were processed with tangential flow filtration (TFF) using PES hollow fiber membrane (100 kDa MWCO, Repligen, USA), and further diafiltered with HEPES buffer. After filtering the formulation with 0.2 µm PES filter, an in-process RNA concentration determination was performed, and the formulation concentration was adjusted to the final target concentration of 1.2 mg/mL with cryoprotectant addition. After sterile filtration, bulk formulation is aseptically filled into glass vials, and stored frozen at − 70 °C. Characterization of final formulation includes particle size and polydispersity (PDI) measurement by dynamic light scattering (ZEN3600, Malvern Instrument, Malvern, United Kingdom), RNA encapsulation and quantification by a fluorometric assay using Ribogreen RNA reagent (Thermo Fisher Scientific, USA), and mRNA purity characterization by capillary electrophoresis using fragment analyzer (Advanced Analytical, USA). Briefly, mRNA was analyzed for size and integrity using a parallel capillary electrophoresis instrument Fragment Analyzer 5400. Each prepared sample is voltage injected into discrete capillaries, arranged in a parallel format. The capillaries contain a separation gel matrix infused with a fluorescent intercalating dye, that is automatically primed into the capillaries prior to each run. During electrophoresis, the nucleic acid fragments in the sample migrate and separate based upon their size, picking up dye along the way. As the separated fragments pass by the detection window, the nucleic acid bound dye is excited by a continuous light source, producing fluorescent emission which is detected by a sensitive CCD detector. The time required to pass through the detection window indicates size and the relative emission signal provides the nucleic acid concentration when compared to a calibrated ladder. ProSize analytical software automatically provides information on mRNA integrity and size distribution. In addition, to more accurately determine purity (integrity) of mRNA the data is transferred to internal Excel sheets (Purity Calculator) designed to perform mRNA peak integration according to principles used by chromatographic integration software. All mRNA samples with initial concentration above 50 µg/mL are diluted to 50 µg/mL of mRNA with RNAse-free water in a total volume of 20 μL or greater and then used for preparation of analytical samples in the 96-wells plates. Any mRNA samples with concentrations lower than 50 µg/mL are not diluted and are used as is for analytical sample preparation. When applicable, it is recommended to concentrate down the samples to meet the concentration requirement. The prepared LNP formulations was diluted with sterile water at a ratio of 1:1 to ensure a physiological osmolality prior to nebulization with an Aerogen Solo vibrating mesh nebulizer (Aerogen, Galway, Ireland, Supplementary Figure 2) to reach 0.6 mg/mL RNA concentration. The aerosol was collected by condensation using ice-cold tube, and post-nebulized samples were subjected to quantification of mRNA encapsulation efficiency and mRNA purity/integrity evaluation. These were used in in vitro cell viability and transfection assays. 2.2.2 Mouse Plasma Stability Lipid stock solution was prepared by dissolution of the lipid in isopropyl alcohol at the concentration of 5 mg/mL. A requisite volume of the lipid-isopropyl alcohol solution was then diluted to 100 µM concentration at a total volume of 1.0 mL with 50:50 (v/v) ethanol / water. Ten microliters of this 100 µM solution was spiked into 1.0 mL of mouse plasma (BioIVT, Cat. No.: MSE00PLNHUNN, CD-1 mouse, anticoagulant: sodium heparin, not filtered) that was pre-warmed to 37 °C and and was stirred at 50 rpm with a magnetic stir bar. The starting concentration of lipids in plasma was thus 1 µM. At time points 0, 15, 30, 45, 60 and 120 min, 0.1 mL of the plasma was withdrawn from the reaction mixture and the protein was precipitated by adding 0.9 mL of ice-cold 4:1 (v/v) acetonitrile/methanol with 1 µg/mL of a selected internal standard lipid added. After filtration through a 0.45 µm 96-well filtering plate, the filtrates were analyzed by LC-MS (Thermo Fisher’s Vanquish UHPLC – LTQ XL linear ion trap Mass Spectrometer); Waters XBridge BEH Shield RP18 2.5 mm (2.1×100 mm) column with its matching guard column. Mobile phase A was 0.1% formic acid in water, and mobile phase B was 0.1% formic acid in 1:1 (v/v) acetonitrile/methanol. Flow rate was 0.5 mL/min. Elution gradient was: Time 0 – 1 min: 10% B; 1–6 min: 10–95% B; 6 – 8.5 min: 95% B; 8.5–9 min: 95–10% B; 9–10 min: 10% B. Mass spectrometry was in positive scanning mode from 600 to 1100 m/z. The peak of the molecular ion of the lipids was integrated in the extracted ion chromatography (XIC) using Xcalibur software (Thermo Fisher). The relative peak area compared to T = 0, after normalization by the peak area of the internal standard, was used as the percentage of the lipid remaining at each time point. T1/2 values were calculated using the first-order decay model. 2.2.3 Cell Transfection Assay / In-Cell Western CFBE41o cells (DF508-Human CFTR) were obtained from Fisher Scientific (Cat# SCC161) and maintained at 37 °C with 5% CO2 in complete Dulbecco's Modified Eagle Medium (Corning Cat# 10–013-CV) supplemented with MEM Non-Essential Amino Acids Solution (Fisher Scientific Cat# 11140050) and 10% Fetal Bovine Serum (HyClone Cat# SH30071.03). For the In Cell Western assay, CFBE cells were seeded in Nunc MicroWell 96-Well Optical-Bottom Plates (Fisher Scientific Cat# 152036) at 17,000 cells/well concentration. 24 h post seeding, CFBE cells were incubated with 1, 0.5, 0.25 µg/mL of tdTomato-expressing LNP formulations, in quadruplicate for each condition. 24 h later, CFBE cells were fixed in 4% PFA for 20 min, permeabilized with 0.1% Triton X-100 and blocked in Odyssey blocking buffer (Licor Cat# 927–50000) for 45 min. Rabbit anti-RFP (Rockland Cat# 600–401–379) antibody was diluted 1:300 in Odyssey blocking buffer and incubated on cells for 1 h. IRDye 800CW Donkey anti-Rabbit IgG Secondary Antibody (Licor Cat# 926–32213) and CellTag 700 (Licor Cat# 926–41090) were diluted 1:7000 and 1:500, respectively, in Odyssey blocking buffer and incubated with CFBE cells for 1 h. Then, whole 96-Well Optical-Bottom Plates were scanned using a Licor Odyssey CLx imaging system. CellTag 700 nm signal was used as a marker of cell viability, with PBS negative samples used as a control, and reported as relative fluorescence units. tdTomato signal, obtained from the 800 nm channel, was instead normalized against the corresponding CellTag 700 nm value, and reported as relative fluorescence units. Licor Image Studio software was used for fluorescence signal analysis, while Microsoft Excel and GraphPad Prism 9 softwares were used for data analysis and visualization. 2.2.4 Flow Cytometry CFBE41o cells (DF508-Human CFTR) were obtained from Fisher Scientific (Cat# SCC161) and maintained at 37 °C with 5% CO2 in complete Dulbecco's Modified Eagle Medium (Corning Cat# 10–013-CV) supplemented with MEM Non-Essential Amino Acids Solution (Fisher Scientific Cat# 11140050) and 10% Fetal Bovine Serum (HyClone Cat# SH30071.03). 100k CFBE cells were plated on 24-well plates (Corning Cat# 07–200–84) and incubated with 0.5 µg/mL LNP formulations or vehicle. After 24 h, cells were harvested using TrypLE (Thermo Fisher Cat# 12605036), washed in DPBS and resuspended in FASC buffer (DPBS, 1% BSA and 2.5 mM EDTA). Quantification of tdTomat% positive cells was measured using FACS analysis (ZE5 Cell Analyzer from Bio-Rad) and FlowJo software. Briefly, cell populations were gated on FSC- Height /SSC- Height scatter plots, and subsequently the singlets identified using FSC- Height /FSC-Width and SSC- Height /SSC-Width scatter plots. Of the singlets population, tdTomato positive cells were gated based on negative controls (vehicle and untreated). 2.2.5 Animal studies Eight-to-week-old, female Balb/C mice were purchased from Jackson Laboratories (Bar Harbour, ME, USA). Mice were housed in a pathogen-free environment in Innovive disposable IVC rodent caging system with a 12 h light/dark cycle and ad libitum access to rodent chow and water. All in vivo procedures were performed in accordance with guidelines established by the Institutional Animal Care and Use Committee (IACUC). Balb/C mice were intra-tracheally dosed with different LNPs (DOTAP-, DOTAP+, L1-L4) carrying a TdTomato mRNA. Twenty-four hours post-dose, mice were anesthetized via 2.5% isoflurane and lungs were processed for immunohistochemistry using a TdTomato primary antibody. For detailed method, please see Supplementary Material. 3 Results and discussion As seen in Supplementary Figure 1, DOTAP is a cationic lipid with a trimethylammonium and a 2,3-dihydroxypropane core that forms asymmetric esters with 2 molecules of oleic acid. This pseudo glyceryl form makes DOTAP chiral with unique chemical and biochemical properties. For example, the chiral nature of the molecule imposes ambiguity on the nature and rate of esterase activity on DOTAP as enzymes do have inherent preferences for cleavage of one chiral form over the other. This is further complicated by the common use of DOTAP as a racemate (vide supra). Hence, we designed cationic lipids L1-L4 which are non-chiral functional analogues of DOTAP. Further, we postulated that ester groups more distal to the branching point bearing the charged quaternary ammonium head group would be subjected to less steric congestion which would likely facilitate biodegradability, resulting in fragments which might be more readily eliminated. Such lipid configuration(s) are known to facilitate rapid elimination of LNPs (Martin et al., 2013). As described above (vide supra), we designed the analogues of the current study to exhibit similar atom extensions from the branch point as does DOTAP. The ester carbonyl was positioned 8 atoms from the branch and similarly sized (Z)− 2-alkenols formed the esters. Different head group modifications were introduced in these designs and they were not anticipated to be the source of significant problems to formulation. The single methylene spacer of DOTAP was replaced in analogs L1-L4 with 2–3 atom spacer groups, in the form of an ether (L1) and all carbon linkages (L2-L4). These changes resulted a cLogP = 11.34 for L1, 10.44 for L2, 10.85 for L3, and 11.46 for L4. The importance of the lipophilicity of DOTAP with respect to formulation and performance of the LNPs containing it, and the possible impact of the range of cLogP values of cationic lipids L1-L4 on the formulation and performance of LNPs containing them would instruct us relative to the appropriateness of our initial designs and may suggest future direction(s). We combined these cationic lipids, whose function is to promote cellular uptake of the LNPs, with proprietary ionizable lipids (Rajappan et al., 2020), that will facilitate endosomal release, in the formulation of novel cationic LNPs and the comparative particle characteristics and their physicochemical properties are shown in Table 1. All formulations with DOTAP or L1-L4 formed particles with < 100 nm in diameter and encapsulation efficiency of TdTomato mRNA > 95%. Also, the purity of mRNA, a measure of the effect of formulation process or lipids themselves have on the integrity of mRNA, remained high in all cases. Formulations prepared without DOTAP (DOTAP-, Table 1) displayed similar particle size and mRNA purity but showed only 71% mRNA encapsulation efficiency. We also tested to see if physicochemical properties of the formulation are affected after freeze-thaw cycle by measuring physicochemical attributes pre and post freezing at − 70 ºC followed by thawing. Three independent measurements were performed for particle size and PDI and reported with average±std. As can be seen from Table 1, all the parameters remained within 10% of the original values for all tested formulations. We also studied the effects of nebulization on mRNA integrity inside the particles. Drug administration through nebulization or powder inhalation to treat pulmonary diseases are proven with other therapeutic modalities, from small molecule drugs like albuterol and Symbicort to large molecules like Afrezza, an inhaled form of insulin (Heinemann et al. 2017). Currently there are no approved LNP mRNA products utilizing inhalation of nebulized particles to deposit LNPs to the lung airway. The prepared LNP formulations were diluted with WFI (water for injection) to ensure osmolality similar to physiological condition and nebulized using a vibrating mesh nebulizer (please see 2.2.1 and Supplementary Figure 2) at 0.6 mg/mL RNA concentration. The aerosol was collected by condensation using ice-cold tube. Since the LNPs are subject to shear-stress during the nebulization process which might be detrimental to the quality of the LNP, encapsulation efficiency and mRNA purity were evaluated for post-nebulized formulations. As noticed from Table 2, for the ones prepared with DOTAP or analogs L1-L4, nebulization neither impacted the mRNA encapsulation efficiency nor affected mRNA integrity to any significant extent. In contrast, DOTAP- formulation showed decreased mRNA encapsulation efficiency after nebulization.Table 2 Characterization of the Impact of Aerosolized LNPs on mRNA Integrity and % Encapsulation. Pre- and post-aerosolized LNPs were characterized for % of encapsulation efficiency and mRNA purity. These physicochemical properties were comparable across DOTAP+ and L1-L4, whereas DOTAP- profile was impacted. Data was consistent across two several repeat studies conducted with different batches of formulations. For more clarity, please see Supplementary Table 2 also. Table 2 Pre-nebulization Post-nebulization Lipid %Encap mRNA purity %Encap mRNA purity (%) DOTAP- 73.0 84 57 74 DOTAP+ 99.7 91 97.5 86 L1 100.2 90 99 82 L2 100.1 91 98.3 84 L3 100 84 99.3 82 L4 100 90 99.2 84 Our next objective was to evaluate the lipids L1-L4 for their biodegradability profile in mouse plasma ( Fig. 2). We also included two ionizable lipids (Lipid Control #1 and #2) with previously established mouse plasma half-lives as positive controls. In this assessment the lipids were incubated with mouse plasma for 2 h (see materials and methods, vide supra) and the percentage of intact lipid remaining at different time points was assessed. As we had expected, DOTAP was quite stable, associated with a T1/2 of > 120 min, while lipids L1-L4 showed much faster degradation profiles as indicated by T1/2 values ranging from 8.0 to 11.5 min (Fig. 2C). At 30 min, almost all L1, L3 and L4 are hydrolyzed, while ~20% L2 and 80% of DOTAP+ remain intact (Fig. 2 B). These data support the notion that changes in the structure of DOTAP+ can lead to improved lipids with better biodegradability profiles, as observed for lipids L1-L4 (Fig. 2). After LNP disruption to release mRNA, individual lipid components need to be metabolized for eventual elimination. This might happen in the plasma because these smaller lipid components are often shuttled into systemic circulation, as opposed to the intact LNP. Hence, plasma stability is an appropriate measure of overall biodegradability and potential for accumulation over time. Hence, the ready plasma degradability of L1-L4 is thought to be an advantage over DOTAP.Fig. 2 Mouse Plasma Biodegradability Profile of DOTAP+ Analog Lipids (L1-L4) in Comparison to DOTAP+. A: Full biodegradability profile in mouse plasma. Samples were collected every 15 min for up to 2 h. B: high level overview of the biodegradability profile up to 30 min (dashed box in A). C: Half-life of the different lipids tested in mouse plasma. L1-L4 have a shorter half-life than DOTAP+ and, therefore, a better biodegradability profile. Fig. 2 After confirming the characteristics of nascent LNPs and consistency of encapsulation efficiency and mRNA integrity across multiple iterations of nebulization, our next objective was to test the efficiency of these cationic LNPs for transfection (no additional transfection reagent was used in these experiments) in Cystic Fibrosis Bronchial Epithelial (CFBE) cells. Different LNPs encapsulating a TdTomato mRNA were used to test this hypothesis. Pre- and post-nebulization fractions were collected and transfected in CFBE cells at a high (1 µg/mL), mid (0.5 µg/mL) and low (0.25 µg/mL) dose. Transfection efficiency measured by FACS showed that DOTAP+ (DOTAP+ indicates DOTAP containing LNP as positive control) and L1-L4 have the highest number of cells transfected when compared to DOTAP- (LNP without DOTAP, Supplementary Figure 3); which is also observed qualitatively by immunocytochemistry for TdTomato in transfected cells (Supplementary Figure 4). Cell viability was also monitored during the process, where the lowest dose (0.25 µg/mL) showed minimal impact in viability comparable at both pre- and post-nebulization fractions across the tested LNPs (Supplementary Figure 5). When post-nebulization fractions are compared, we observed that lipids L3 and L4 had a larger impact on cell viability, which may not necessarily due to the chemical characteristics of these lipids, but to changes in particle characteristics post nebulization, and this warrants further exploration. Relative TdTomato protein quantitation was assessed by fluorescence intensity ( Fig. 3) using In-Cell Western, (ICW) and all LNPs, except the one without DOTAP (DOTAP-), showed dose dependent changes in fluorescent intensity with pre-nebulized LNPs (Fig. 3 A). The data presented by ICW (Fig. 3) and FACS (Supplementary figure 3) shows the same pre- and post-nebulization fraction trends across all LNPs tested, independently of any sensitivity differences that may exist in each assay. Concomitantly, all nebulized LNPs also showed dose dependency (Fig. 3B), albeit quite low intensity for DOTAP- LNP in both cases. Lipids L1-L4 performed similarly to DOTAP+ in terms of TdTomato protein expression in this study. All the data generated with LNP-treated cells were statistically significant at all three doses tested (two-tailed unpaired parametric t-test, Fig. 3). A heat map showing the relative percentage of TdTomato protein levels retained in the post-nebulization fraction compared to the pre-nebulization condition shows both L3 and L4 performing better than DOTAP+ at all three concentrations tested (Fig. 3 C). This is consistent with the similar critical packing parameters (P) calculated for these lipids. Lower performance of L1 may be due to the low P value of 0.52 that is at the threshold for forming stable lamellar structures. While our primary objective was to design better bio-degradable lipids and not necessarily more potent lipids, we were not surprised with these results. These results support the use of L2-L4 with cLogP values that range from 10.85 to 11.46 and P values > 0.60 in the formulation of LNPs that function equivalently to those utilizing DOTAP+ (cLogP = 11.77) and are also readily biodegradable.Fig. 3 Relative TdTomato Protein Quantitation in CFBE cells Transduced with Pre- and Post-Nebulized Lipid Nanoparticles (LNP) Carrying a TdTomato mRNA. A-B: Quantitation of the relative levels of luminescence generated after transfection with three doses (0.25, 0.5 and 1 µg/mL) of formulations (DOTAP-, DOTAP+, L1-L4) carrying a TdTomato mRNA. Pre- (A) and post- (B) nebulization formulations were used in this study, and untransfected and vehicle (LNP buffer) were used as controls. Controls did not show any luminescence. Samples were run in quadruplicates in two independent studies and showed as mean ± SD. A dose response in protein levels was observed in all the formulations and conditions tested. Two-tailed unpaired parametric t-test was used to assess significant TdTomato expression differences (p < 0.05) between DOTAP+ and other formulations, comparing 1 ug/mL dose (*), 0.5 ug/mL dose (**) and 0.25 ug/mL dose (***). C: Heat map showing the relative percentage of TdTomato protein levels retained in the post-nebulization fraction when compared to the pre-nebulization condition. The lighter the blue color the higher the drop in protein levels, being equal or higher than 1.0 considered 100% retained. Fig. 3 To further explore the potential benefit of L1-L4 as novel lipids targeting the lung, we conducted a preliminary in vivo study using wild-type mice that where intra-tracheally dosed with different LNPs (DOTAP-, DOTAP+, L1-L4) carrying a TdTomato mRNA. PBS was used as a control. Twenty-four hours post-dose, lungs were processed for immunohistochemistry for TdTomato ( Fig. 4). A differential expression pattern was observed across the different LNPs, with DOTAP+, L2 and L4 showing the highest levels of TdTomato expression in rodent airways. In contrast, L1 and L3 showed minimal levels of TdTomato. No staining was observed in the control group. Additional studies using an inhaled approach will need to be conducted to further understand these differences, but here we show the first evidence that the design of DOTAP analogs (L1-L4), by changing the structure of DOTAP, and respective cLNPS can have major implications on airway delivery in animal models.Fig. 4 Immunohistochemistry for TdTomato Protein in Mouse Airways after Delivery of LNP Formulations (DOTAP-, DOTAP+, L1-L4) Carrying a TdTomato mRNA. Twenty-four hours post-dose, lungs were processed for immunohistochemistry using a TdTomato primary antibody. Cresyl violet as used as counterstaining. DOTAP+ and L2 and L4 formulations shown higher levels of TdTomato protein when compared across treatment groups and control (PBS). A′-F′ are high power views of A-F, respectively. Scale bars: A-F (250 µm) and A′F’ (50 µm). Fig. 4 4 Conclusion In summary, we have synthesized four novel cationic lipids that are functional analogues (L1-L4) of one of the most used cationic lipids, DOTAP+, and formulated them successfully into cLNPs encapsulating an mRNA. These cLNPs show similar physical chemical characteristic as those utilizing DOTAP. These particles indeed showed excellent physicochemical properties and protected the mRNA cargo as seen from high mRNA encapsulation efficiency and mRNA purity evaluation post a freeze-thaw cycles. Our rational for exploring these lipids were an expected improvement in the biodegradation profile as well as a lack of chiral centers. In a comparison of L1-L4 with DOTAP, we desired comparable or better particle formation characteristics, mRNA release capabilities, as seen from the cognate protein expression, and comparable in vitro profile. While the plasma biodegradabilities of these lipids were demonstrated to be better than that of DOTAP, the other characteristics remained similar. In addition, the cLNPs incorporating L1-L4 were nebulized, and these aerosolized particles also showed comparable TdTomato protein expression levels in vitro, whereas L2 and L4 showed similar delivery profile than DOTAP+ in rodent airways. Thus, with these readily biodegradable analogues, a more extensive set of in vivo studies will be required to further validate L1-L4 for inhaled therapies using relevant animal models for respiratory diseases such as CF. As we move forward with additional designs, we must also be mindful of the need to clear the products of ester cleavage. Hence the physicochemical space occupied by prospective new analogues and the physicochemical characteristics of the fragments of ester hydrolysis will be of importance. Moving into poor physicochemical space for the products of ester hydrolysis could greatly impact the clearance of these entities. Consider the small alteration in the structure of analog L3, as shown in L3a to be illustrative of this concern ( Fig. 5). Hypothetical analogue L3a retains the ester carbonyl at a position 8 atoms from the branch point bearing the tetraalkylammonium head group, approximating the same steric environment used in the design of L3 while it moves the oxygen of the ester to a position more proximal to the tetraalkyl ammonium head group. The calculated physicochemical space occupied by the fragments Metabolite 2 (c-LogP = −0.53, cLogD = −0.91) and Metabolite 4 (c-LogP = −2.08) are illustrative of this concern, as the clearance of tetraalkylammoniums of the size of Metabolite 2 and Metabolite 4 appears to be negatively impacted by lower c-LogP (Hirom et al., 1974, Neef and Meijer, 1984, Yang et al., 2009). In fact, these were part of our considerations for choosing the ester orientations as in L1-L4 as opposed to the hypothetical L3a type orientation. But further PD/PK correlation studies are warranted, and future designs will be guided by such criteria.Fig. 5 Physicochemical Space and Potential Clearance Issues. (A) Compound L3 bio-degrades into metabolite 1 and metabolite 2. (B) Hypothetical alternate design for L3 (shown as L3a) would result in metabolite 4 that is expected to have poorer clearance based on calculated cLogP of − 2.08. Calculated using ACD Labs Structure Designer v12.0. cLogP was calculated using ACD Labs version B. Fig. 5 Declaration of Competing Interest The authors are all employees of Arcturus Therapeutics and may hold equities in the company. The authors declare no other competing financial interests. Appendix A Supplementary material Supplementary material . Acknowledgments Part of the calculations in Table 1 are done using Molinspiration Property Calculation Service freely available on the internet (https://www.molinspiration.com). Appendix A Supplementary data associated with this article can be found in the online version at doi:10.1016/j.chemphyslip.2022.105178. ==== Refs References Akinc A. Querbes W. De S. Qin J. Frank-Kamenetsky M. Narayanannair K.J. Jayaraman M. Rajeev K.G. Cantley W.L. Dorkin J.D. Butler J.S. Qin L. Racie T. Sprague A. Fava E. Zeigerer A. Hope M.J. Zerial M. Sah D.W.Y. Fitzgerald K. Tracy M.A. Manoharan M. Koteliansky V. Fougerolles A. Maier M.A. Targeted delivery of RNAi therapeutics with endogenous and exogenous ligand-based mechanisms Mol. Ther. 18 2010 1357 1364 10.1038/mt.2010.85 20461061 Allon N. Saxena A. Chambers C. Doctor B.P. A new liposome-based gene delivery system targeting lung epithelial cells using endothelin antagonist J. Control Release 160 2012 217 224 10.1016/j.jconrel.2011.10.033 22079949 Andey T. Bora-Singhal N. Chellappan S.P. Singh M. Cationic lipoplexes for treatment of cancer stem cell-derived murine lung tumors Nanomedicine (Lond. ) 18 2019 31 43 https://doi.oreg/10.1016/j.nano.2019.02.007 Bragonzi A. Boletta A. Biffi A. Muggia A. Sersale G. Cheng S.H. Bordignon C. Assael B.M. Conese M. Comparison between cationic polymers and lipids in mediating systemic gene delivery to the lungs Gene Ther. 6 1999 1995 2004 10.1038/sj.gt.3301039 10637451 Burgess E.M. Penton H.R. Jr. Taylor E.A. Synthetic applications of N-carboalkoxysulfamate esters J. Am. Chem. Soc. 92 1970 5224 5226 10.1021/ja00720a041 Cerutti-Delasalle C. Mehiri M. Cagliero C. Rubiolo P. Bicchi C. Meierhenrich U.J. Baldovini N. The (+) - cis- and (+) - trans- olibanic acids: key odorants of frankincense Angew. Chem. Int. Ed. 55 2016 13719 13723 10.1002/anie.201605242 Cheng Q. Wei T. Farbiak L. Johnson L.T. Dilliard S.A. Siegwart D.J. Selective organ targeting (SORT) nanoparticles for tissue-specific mRNA delivery and CRISPR-Cas gene editing Nat. Nanotechnol. 15 2020 313 320 10.1038/s41565-020-0669-6 32251383 Chu C.S. Trapnell B.C. Curristin S. Cutting G.R. Crystal R.G. Genetic basis of variable exon 9 skipping in cystic fibrosis transmembrane conductance regulator mRNA Nat. Genet. 3 1993 151 156 10.1038/ng0293-151 7684646 Conti D.S. Brewer D. Grashik J. Avasarala S. Rocha S.R.P. Poly(amidoamine) dendrimer nanocarriers and their aerosol formulations for siRNA delivery to the lung epithelium Mol. Pharm. 11 2014 1808 1822 10.1021/mp4006358 24811243 Crawford R. Dogdas B. Keough E. Haas R.M. Wepukhulu W. Krotzer S. Burke P.A. Sepp-Lorenzino L. Bagchi A. Howell B.J. Analysis of lipid nanoparticles by Cryo-EM for characterizing siRNA delivery vehicles Int. J. Pharm. 403 2011 237 244 10.1016/j.ijpharm.2010.10.025 20974237 Fletcher S. Ahmad A. Perouzel E. Heron A. Miller A.D. Jorgensen M.R. In vivo studies of dialkynoyl analogues of DOTAP demonstrate improved gene transfer efficiency of cationic liposomes in mouse lung J. Med. Chem. 49 2006 349 357 https://pubs.acs.org/doi/10.1021/jm0507227 16392819 Heinemann L. Baughman R. Boss A. Hompesch M. Pharmacokinetic and Pharmacodynamic Properties of a Novel Inhaled Insulin J. Diabetes Sci. Technol. 11 2017 148 156 10.1177/1932296816658055 27378794 Hirom P.C. Hughes R.D. Millburn P. The Physicochemical Factor Required for the Biliary Excretion of Organic Cations and Anions Biochem. Soc. Trans. 2 1974 327 330 10.1042/bst0020327 Hsu W. Chen H. Liou W. Lin H. Liu W. Mesomorphic complexes of DNA with the mixtures of a cationic surfactant and a neutral lipid Langmuir 21 2005 9426 9431 10.1021/la051863e 16207017 Jayaraman M. Ansell S.M. Mui B.L. Tam Y.K. Chen J. Du X. Butler D. Eltepu L. Matsuda S. Narayanannair J.K. Rajeev K.G. Hafez I.M. Akinc A. Maier M. Tracy M.A. Cullis P.R. Madden T.D. Manoharan M. Hope M.J. Maximizing the potency of siRNA lipid nanoparticles for hepatic gene silencing in vivo. Angew. Chem. Int. Ed. 51 2012 8529 8533 10.1002/anie.201203263 Jensen D.K. Jensen L.B. Koocheki S. Bengtson L. Cun D. Nielsen H.M. Foged C. Design of an inhalable dry powder formulation of DOTAP-modified PLGA nanoparticles loaded with siRNA J. Control Release 157 2012 141 148 10.1016/j.jconrel.2011.08.011 21864597 Jensen D.M.K. Cun D. Maltesen M.J. Frokjaer S. Nielsen H.M. Foged C. Spray drying of siRNA-containing PLGA nanoparticles intended for inhalation J. Control Release 142 2010 138 145 10.1016/j.jconrel.2009.10.010 19840823 Kulkarni J.A. Darjuan M.M. Mercer J.E. Chen S. Meel R. Thewalt J.L. Tam Y.Y.C. Cullis P.R. On the formation and morphology of lipid nanoparticles containing ionizable cationic lipids and siRNA ACS Nano 12 2018 4787 4795 10.1021/acsnano.8b01516 29614232 Kusumoto K. Akita H. Ishitsuka T. Matsumoto Y. Nomoto T. Furukawa R. El-Sayed A. Hatakeyama H. Kajimoto K. Yamada Y. Kataoka K. Harashima H. Lipid envelope-type nanoparticle incorporating a multifunctional peptide for systemic siRNA delivery to the pulmonary endothelium ACS Nano 7 2013 7534 7541 10.1021/nn401317t 23909689 Lechanteur A. Sanna V. Duchemin A. Evrard B. Mottet D. Piel G. Cationic liposomes carrying siRNA: impact of lipid composition on physicochemical properties, cytotoxicity and endosomal escape Nanomater 8 2018 270 10.3390/nano8050270 Lee E.R. Marshall J. Siegel C.S. Jiang C. Nelson S.Y. Nichols M.R. Nietupski J.B. Ziegler R.J. Lane M.B. Wang K.X. Wan N.C. Scheule R.K. Harris D.J. Smith A.E. Cheng S.H. Detailed a n a l y s i s of structures a n d f o r m u l a t i o n s of cationic lipids for efficient g e n e transfer to the l u ng Human Gene Ther. 7 1996 1701 1717 10.1089/hum.1996.7.14-1701 8886841 Leung A.K.K. Hafez I.M. Baoukina S. Belliveau N.M. Zhigaltzev I.V. Afshinmanesh E. Tieleman D.P. Hansen C.L. Hope M.J. Cullis P.R. Lipid nanoparticles containing siRNA synthesized by microfluidic mixing exhibit an electron-dense nanostructured core J. Phys. Chem. C Nanomater. Interfaces 116 2012 18440 18450 10.1021/jp303267y 22962627 Leventis R. Silvius J.R. Interactions of mammalian cells with lipid dispersions containing novel metabolizable cationic amphiphiles Biochim. Biophys. Acta 1023 1 1990 124 132 10.1016/0005-2736(90)90017-i 2317491 Lokugamage M.P. Vanover D. Beyersdorf J. Hatit M.Z.C. Rotolo L. Echeverri E.S. Peck H.E. Ni H. Yoon J.K. Kim Y. Optimization of lipid nanoparticles for the delivery of nebulized therapeutic mRNA to the lungs Nat. Biomed. Eng. 5 2021 1059 1068 10.1038/s41551-021-00786-x 34616046 Ma Z. Zhang J. Alber S. Dileo J. Negishi Y. Stolz D. Watkins S. Huang L. Pitt B. Li S. Lipid-mediated delivery of oligonucleotide to pulmonary endothelium Am. J. Respir Cell Mol. Biol. 27 2002 151 159 10.1165/ajrcmb.27.2.4653 12151306 MacLachlan I. Liposomal formulations for nucleic acid delivery Crooke S.T. Antisense Drug Technologies 2nd ed. 2007 CRC Press Boca Raton, FL 237 270 Martin A.M. Muthusamy J. Shigeo M. Liu J. Barros S. Querbes W. Tam Y.K. Ansell S.M. Kumar V. Qin J. Zhang X. Wang Q. Panesar S. Hutabarat R. Carioto M. Hettinger J. Kandasamy P. Butler D. Rajeev K.G. Pang B. Charisse K. Fitzgerald K. Mui B.L. Du X. Cullis P. Madden T. Hope M.J. Manoharan M. Akinc A. Biodegradable lipids enabling rapidly eliminated lipid nanoparticles for systemic delivery of RNAi therapeutics Mol. Ther. 21 2013 1570 1578 10.1038/mt.2013.124 23799535 Merkel O.M. Rubinstein I. Kissel T. siRNA delivery to the lung: what’s new? Adv. Drug Deliv. Rev. 75 2014 112 128 10.1016/j.addr.2014.05.018 24907426 Mueller R. Yang J. Duan C. Pop E. Geoffroy O.J. Zhang L.H. Huang T. Denisenko S. McCosar B.H. Oniciu D.C. Bisgaier C.L. Pape M.E. Freiman C.D. Goetz B. Cramer C.T. Hopson K. Dasseux J.H. Long hydrocarbon chain keto diols and diacids that favorably alter lipid disorders in vivo J. Med. Chem. 47 2004 6082 6099 10.1021/jm040006p 15537362 Nabhan J.F. Wood K.M. Rao V.P. Morin J. Bhamidipaty S. LaBranche T.P. Gooch R.L. Bozal F. Bulawa C.E. Guild B.C. Intrathecal delivery of frataxin mRNA encapsulated in lipid nanoparticles to dorsal root ganglia as a potential therapeutic for Friedreich’s Ataxia Sci. Rep. 6 2016 1 10 10.1038/srep20019 28442746 Neef C. Meijer D.K.F. Structure-pharmacokinetics relationship of quaternary ammonium compounds Arch Pharmacol 328 1984 111 118 10.1007/BF00512059 Omura K. Swern D. Oxidation of alcohols by “activated” dimethyl sulfoxide. A preparative, steric and mechanistic study Tetrahedron 34 1978 1651 1660 10.1016/0040-4020(78)80197-5 Onpattro, European Medicines Agency. Summary of opinion (initial authorization, 2018). https://www.ema.europa.eu/en/documents/smop-initial/chmp-summary-positive-opinion-onpattro_en.pdf (Accessed 17 August 2021). ONPATTRO (Patisiran) FDA Label, lipid complex injection, for intravenous use. https://www.accessdata.fda.gov/drugsatfda_docs/label/2018/210922s000lbl.pdf (Accessed 17 August 2021). Patel S. Ashwanikumar N. Robinson E. DuRoss A. Sun C. Murphy-Benenato K.E. Mihai C. Almarsson O. Sahay G. Boosting intracellular delivery of lipid nanoparticle-encapsulated mRNA Nano Lett. 17 2017 5711 5718 10.1021/acs.nanolett.7b02664 28836442 Prabhakar U. Maeda H. Jain R.K. Sevick-Muraca E.M. Zamboni W. Farokhzad O.C. Barry S.T. Gabizon A. Grodzinski P. Blakey D.C. Challenges and key considerations of the enhanced permeability and retention effect for nanomedicine drug delivery in oncology Cancer Res. 73 2013 2412 2417 10.1158/0008-5472.CAN-12-4561 23423979 Rajappan K. Tanis S.P. Mukthavaram R. Roberts S. Nguyen M. Tachikawa K. Sagi A. Sablad M. Limphong P. Leu A. Yu H. Chivukula P. Payne J.E. Karmali P. Property Driven Design and Development of Lipids for Efficient Delivery of siRNA J. Med. Chem. 63 2020 12992 13012 10.1021/acs.jmedchem.0c01407 33119286 Ramaswamy S. Tonnu N. Tachikawa K. Limphong P. Vega J.B. Karmali P.P. Chivukula P. Verma I.M. Systemic Delivery of Factor IX messenger RNA for Protein Replacement Therapy Proc. Natl. Acad. Sci. U. S. A. 114 2017 E1941 E1950 10.1073/pnas.1619653114 28202722 Ramesh R. Saeki T. Templeton N.S. Ji L. Stephens L.C. Ito I. Wilson D.R. Wu Z. Branch C.D. Minna J.D. Roth J.A. Successful treatment of primary and disseminated human lung cancers by systemic delivery of tumor suppressor genes using an improved liposome vector Mol. Ther. 3 2001 337 350 10.1006/mthe.2001.0266 11273776 Riordan J.R. Rommens J.M. Kerem B. Alon N. Rozmahel R. Grzelczak Z. Zielenski J. Lok S. Plavsic N. Chou J.L. Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA Science 245 1989 1066 1073 10.1126/science.2475911 2475911 Robinson E. MacDonald K.D. Slaughter K. McKinney M. Patel S. Sun C. Sahay G. Lipid nanoparticle-delivered chemically modified mRNA restores chloride secretion in cystic fibrosis Mol. Ther. 26 2018 2034 2046 10.1016/j.ymthe.2018.05.014 29910178 Sabnis S. Kumarasinghe E.S. Salerno T. Mihai C. Ketova T. Senn J.J. Lynn A. Bulychev A. McFadyen I. Chan J. Almarsson O. Stanton M.G. Benenato K.E. A novel amino lipid series for mRNA delivery: improved endosomal escape and sustained pharmacology and safety in non-human primates Mol. Ther. 26 2018 1509 1519 10.1016/j.ymthe.2018.03.010 29653760 Semple S.C. Akinc A. Chen J. Sandhu A. Mui B.L. Cho C.K. Sah D.W.Y. Stebbing D. Crosley E. Yaworski E. Hafez I.M. Dorkin J.R. Qin J. Lam K. Rajeev K.G. Wong K.F. Jeffs L.B. Nechev L. Eisenhardt M.L. Jayaraman M. Kazem M. Maier M.A. Srinivasulu M. Weinstein M. Chen Q. Alvarez R. Barros S.A. De S. Klimuk S. Borland T. Kosovrasti V. Cantley W.L. Tam Y.K. Manoharan M. Ciufolini M.A. Tracy M.A. Fougerolles A. MacLachlan I. Cullis P.R. Madden T.D. Hope M.J. Rational design of cationic lipids for siRNA delivery Nat. Biotech. 28 2010 172 176 10.1038/nbt.1602 Semple S.C. Klimuk S.K. Harasym T.O. Santos N.D. Ansell S.M. Wong K.F. Maurer N. Stark H. Cullis P.R. Hope M.J. Scherrer P. Efficient encapsulation of antisense oligonucleotides in lipid vesicles using ionizable aminolipids: formation of novel small multilamellar vesicle structures Biochim. Biophys. Acta 1510 2001 152 166 10.1016/S0005-2736(00)00343-6 11342155 Terp M.C. Bauere F. Sugimoto Y. Yu B. Brueggemeier R.W. Lee L.J. Lee R.J. Differential efficacy of DOTAP enantiomers for siRNA delivery in vitro Int. J. Pharm. 430 1–2 2012 328 334 10.1016/j.ijpharm.2012.04.017 22525086 VanRheenan V. Kelly R.C. Cha D.Y. An improved catalytic OsO4 oxidation of olefins to cis-1,2-glycols using tertiary amine oxides as the oxidant Tetrahedron Lett. 23 1976 1973 1976 10.1016/S0040-4039(00)78093-2 Viger-Gravel J. Schantz A. Pinon A.C. Rossini A.J. Schantz S. Emsley L. Structure of lipid nanoparticles containing siRNA or mRNA by dynamic nuclear polarization-enhanced NMR spectroscopy J. Phys. Chem. B 122 2018 2073 2081 10.1021/acs.jpcb.7b10795 29332384 Welsh M.J. Abnormal regulation of ion channels in cystic fibrosis epithelia FASEB J. 4 1990 2718 2725 10.1096/fasebj.4.10.1695593 1695593 Welsh M.J. Anderson M.P. Rich D.P. Berger H.A. Denning G.M. Ostedgaard L.S. Sheppard D.N. Cheng S.H. Gregory R.J. Cystic fibrosis transmembrane conductance regulator: a chloride channel with novel regulation Neuron 8 1992 821 829 10.1016/0896-6273(92)90196-K 1375035 Wheeler C.J. Felgner P.L. Tsai Y.J. Marshall J. Sukhu L. Doh S.G. Hartikka J. Nietupski J. Manthorpe M. Nichols M. Plewe M. Liang X. Norman J. Smith A. Cheng S.H. A novel cationic lipid greatly enhances plasmid DNA delivery and expression in mouse lung Proc. Natl. Acad. Sci. USA. 93 1996 11454 11459 10.1073/pnas.93.21.11454 8876156 Yan X. Kuipers F. Havekes L.M. Havinga R. Dontje B. Poelstra K. Scherphof G.L. Kamps J.A.A.M. The role of apolipoprotein E in the elimination of liposomes from blood by hepatocytes in the mouse Biochem. Biophys. Res. Commun. 328 2005 57 62 10.1016/j.bbrc.2004.12.137 15670750 Yang X. Gandhi Y.A. Suigan D.B. Morris M.E. Prediction of biliary excretion in rats and humans using molecular weight and quantitative structure-pharmacokinetic relationships AAPS J. 11 2009 511 525 10.1208/s12248-009-9124-1 19593675 Yoshida T. Murai M. Abe M. Ichimaru N. Harada T. Nishioka T. Miyoshi H. Crucial Structural Factors and Mode of Action of Polyene Amides as Inhibitors for Mitochondrial NADH-Ubiquinone Oxidoreductase (Complex I) Biochemistry 46 2007 10365 10372 10.1021/bi7010306 17685634 Zelphati O. Szoka F.C. Jr. Mechanism of oligonucleotide release from cationic liposomes Proc. Natl. Acad. Sci. USA. 93 1996 11493 11498 10.1073/pnas.93.21.11493 8876163

35122738

PMC9749011

NO-CC CODE

2022-12-15 23:22:49

no

Chem Phys Lipids. 2022 Mar 3; 243:105178

utf-8

Chem Phys Lipids

10.1016/j.chemphyslip.2022.105178

oa_other

==== Front Health Promot Pract Health Promot Pract HPP sphpp Health Promotion Practice 1524-8399 1552-6372 SAGE Publications Sage CA: Los Angeles, CA 36511091 10.1177/15248399221136533 10.1177_15248399221136533 Resources, Frameworks, and Perspectives Grow502: Centering Community in Medical Education via a Student-Created Organization Focused on Cultivating a Healthy Community https://orcid.org/0000-0002-4126-7575 Udoh Mike Onu MBA, MS 1 Mian Zoha BA 1 Anakwenze Lisa MPH, MS 1 Okeke Chukwudum BA 2 Ziegler Craig PhD 1 Sawning Susan MSSW 1 1 University of Louisville School of Medicine, Louisville, KY, USA 2 Yale School of Public Health, New Haven, CT, USA M. Onu Udoh, University of Louisville School of Medicine, 500 South Preston Street, Louisville, KY 40202, USA; e-mail: moudoh01@louisville.edu 13 12 2022 13 12 2022 15248399221136533© 2022 Society for Public Health Education 2022 Society for Public Health Education This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. The COVID-19 pandemic continues to disproportionately impact communities of color and expose health inequities. Minoritized communities experience disparities in chronic diseases, premature death, and cancer, and gaps continue to widen; systemic injustice in housing, nutrition, and environment are major contributors. At the height of the COVID-19 pandemic and racial justice movement, students at the University of Louisville School of Medicine created Grow502 to speak truth to the challenges of health disparities in Louisville, Kentucky. The goal was to pursue a healthier community by raising awareness. Community leaders and health professionals provided expert consultation. This partnership led to the co-creation of a curriculum focused on education, advocacy, community engagement, and creative media. Grow502 sought to support communities impacted by injustices due to racism, limited health literacy, redlining, and limited green space by creating programming centered on education and empowerment. Effective strategies to reduce disparities involve creating interventions with authentic engagement and context. Grow502 involves community stakeholders as active partners. We continue to intentionally seek effective collaborations and interventions that merge our mission and our diverse communities impacted by health disparities. health disparity social determinants of health health equity health education community intervention creative media health advocacy community engagement edited-statecorrected-proof typesetterts1 ==== Body pmcThe COVID-19 pandemic continues to disproportionately impact communities of color and expose health inequities (Centers for Disease Control and Prevention, 2020; Hooper et al., 2020). Minoritized communities experience disparities in chronic diseases, premature death, and cancer, and gaps continue to widen; systemic injustice in housing, nutrition, and environment are major contributors (National Academies of Sciences, Engineering, and Medicine, 2017; Thornton et al., 2016). Systemic racism and cultural insensitivity within the health care system are also contributors, as studies demonstrate how clinical recommendations built on the falsity of biological differences result in suboptimal care for Black patients (Hoffman et al., 2016); however, newborn survival rates are improved threefold when racially concordant care is provided (Greenwood et al., 2020). To raise awareness among our community, we launched a health disparities series in the Spring of 2021. Context Louisville, Kentucky is a diverse city; however, the community is segregated by race. Due to redlining, a historical practice that puts financial and other services out of reach for residents based on race and/or ethnicity, people of color continue to be clustered primarily in the West End which lacks access to health care, education, and housing. Communities hold the power to impact their health and efforts focused on increasing community self-agency via academic/community partnership interventions are needed to mitigate disparities. Health sciences curricula often fail to center the community as experts or facilitate advocacy (Benabentos et al., 2014; Chen et al., 2017). At the height of the COVID-19 pandemic and racial justice movement, students at the University of Louisville School of Medicine created Grow502 to speak truth to the challenges of health disparities in Louisville, Kentucky. The goal was to pursue a healthier community by raising awareness. Community leaders and health professionals provided expert consultation. This partnership led to the co-creation of a curriculum focused on education, advocacy, community engagement, and creative media. The Louisville Metro Health Equity Report provided a framework that outlined 21 different health outcomes, highlighting differences between race, gender, ethnicity, and zip code (Center for Health Equity Report, 2017). Four strategic teams of medical and public health students worked with community leaders to create the curriculum focusing on five topics and age demographics: (1) Infant Mortality (Infants), (2) Lead Poisoning (Youth), (3) Mental Health (Young Adult), (4) Diabetes/Maternal Mortality (Adults), and (5) Stroke (Elderly). The sessions were open to all in the community with an emphasis on empowering citizens and health professionals. Health Disparities Series and Impact Three medical students and one public health student served as Grow502 Directors. Each Director (education, advocacy, community engagement, creative media) was tasked with managing a project and team for each health disparity. For example, the Director of Education designed and hosted interdisciplinary education panels for Lead Poisoning while the Director of Advocacy similarly created a workshop to build skills to engage the Kentucky legislature. Thirty students from the schools of medicine and public health were recruited to participate in one of the four teams based on their interests. Students were responsible for implementing programs for their respective disparity topics as a team. The creative media team designed all Grow502 marketing and video production. The directors created a standardized weekly schedule for programs, participated in marketing campaigns, and led team meetings to develop materials. Teams developed their presentations and activities with directors, which were reviewed by community experts, advocates, and health care professionals for accuracy and quality. The School of Medicine’s Undergraduate Medical Education Office and local partners supplied resource support. Each week of the curriculum began with an interdisciplinary educational panel that reviewed the basic disease process, epidemiological trends, and case-based problem-solving discussion. Mid-week, a virtual advocacy workshop was offered with discussions on local policies with the goal of developing new skills in health advocacy. End-week, an in-person or virtual community event was held to allow participants to engage with community organizations. Each week, our creative media team implemented a marketing campaign with emails and social media content including but not limited to creative videos highlighting public health data, interviews with local health professionals, cartoons, and poetry. We hoped by providing an early educational event, the gained knowledge would energize the advocacy workshops that followed. An end-of-week community service event cemented the information acquired through personal connections. The education team recruited interdisciplinary panelists for each health disparity. The panel consisted of professionals from different backgrounds (e.g., medicine, social work, law, and patient perspectives). The panel lasted 90 minutes and included background information on local/national trends, followed by a problem-based learning case and discussion points for panelists and attendees, and concluded with a Q&A session and review of local policies and resources. This method oriented attendees to the prevalence of the health disparity on a national scale while making the information actionable and representative of the local community. The advocacy workshops were 60 minutes followed by a short presentation about the topic and its relation to health disparities. These workshops offered students the opportunity to develop advocacy skills. Grow502 session topics included the following: Maternal and Infant Mortality Week Monday—Virtual Interdisciplinary Panel and Case Review: The patient gave birth via C-section with complications and her clinical course was reviewed with attention to physician–patient communication; she suffered and succumbed to her complications which were initially reported to the clinical team, but not promptly acted on. This case highlighted the adversity in obstetrical care experienced by Black/African American women. Discussion questions invited participants to identify sources of bias, impact on communities of color, and prevalence. Participants discussed strategies for prevention, legislation, and community engagement. Wednesday—Virtual Advocacy Workshop on Implicit Bias in Perinatal Care: Legislature staff in support of HB27: An Act relating to implicit bias in perinatal care and HB212: An act relating to child and maternal fatalities in the Commonwealth were contacted and a Kentucky State Representative from the 18th District presented a workshop. Saturday—Maternal Mortality Memorial Wall and Infant Mortality Baby Box Event: Participants were educated on infant and maternal mortality via a short video, offered the chance to submit messages for a memorial wall and to decorate gift boxes for babies. Uplifting messages for mothers of infants that had lost their lives were collected. On the medical campus, baby gift boxes were decorated and filled with items such as baby clothes, breastfeeding educational materials, wet wipes, and diapers. Memorial walls were sent to various community organizations for virtual display (Figure 1). Figure 1 Community Event Hosted by Garden Girl Foods, Whitney Powers (Bottom Left Corner), and Led by Grow502 Director of Community Engagement, Lisa Anakwenze (Top Right Corner), and medical student Katarina Jones (Bottom Right Corner) Diabetes and Stroke Week Monday/Tuesday—Interdisciplinary Panels and Case Reviews: Case 1 involved a male patient with diabetes and multiple other chronic health conditions attempting to establish care with a new primary care physician. The patient’s situation was complex in that he lived in a food desert, lacked transportation, and had limited medical literacy. Discussion questions explored health care costs, food desserts, and the psychology of chronic diseases. Case 2 involved a female patient with stroke-like symptoms who had transportation challenges that prevented timely medical care, as well as poor nutrition, lack of exercise, and diabetes. She had neurological deficits that prevented follow-up. Discussion questions explored connections between poverty, neighborhood development, health-related public service announcements, and transportation in relation to prevention/treatment of stroke. Wednesday—Virtual Advocacy Workshop on Food Deserts and the Implementation of Zoning Districts: Members of Metro Council were contacted to support zoning restrictions that limit the density of businesses such as liquor/tobacco stores. Kroger Corporation was contacted to request that the demands of Feed the West, a racial justice organization that provides fresh food to individuals living with food insecurity, be fulfilled. The President of the Greater Louisville Medical Society presented. Saturday—In-Person Community Event with Garden Girl Food: Teaching kitchen, fitness, and glucose monitoring sessions were offered. Medical students provided health education on nutrition, diabetes prevention/management, and smoking cessation, and led a healthy cooking and exercise workshop. Lead Poisoning Week Monday—Virtual Interdisciplinary Panel and Case Review: The patient was a 5-year-old boy who was having difficulty at school and unexplained somatic ailments. He lived in a home built before 1970 in a lower socioeconomic neighborhood intersected by a highway. He was delayed developmentally and also diagnosed with ADHD but continued to have worsening somatic ailments. Further testing identified elevated lead levels. Discussion questions explored social determinants of health, lead poisoning prevention, and resources for communities. Wednesday—Virtual Advocacy Workshop on Existing Lead Poisoning Regulation in Kentucky: Legislatures were contacted to support modifications to Ky. Rev. Stat. 211.902 and 903 which focused on increasing lead screening standards in Kentucky. An Emergency Medicine faculty clinician presented. Saturday—Virtual Event with Community Activists Discussing Lead Poisoning: Medical students led the programming and provided background on lead poisoning in Louisville. Local researchers, community leaders, and activists shared personal anecdotes and lived experiences with lead poisoning. Mental Health and Substance Use Week Monday—Virtual Interdisciplinary Panel and Case Review: The patient presented to Emergency Psychiatric Services following an acute spell of disorientation and hallucination. The patient had endured numerous psychological and physical traumas and had a substance use disorder. Discussion questions explored the role of substances, adverse childhood events, and addiction recovery on health. Wednesday—Virtual Advocacy Workshop to Petition Kentucky Medicaid to Offer Medication to Ease Opioid Withdrawal: Participants contacted Kentucky Medicaid to advocate for FDA-approved, non-opioid management of opioid withdrawal symptoms without prior authorization. A local endocrinologist who advocates regularly for patients with opioid addiction presented. Saturday—Virtual Training for Opioid Overdose Response with Naloxone: Medical students, pharmacists, and substance use peer educators provided education about substance use disorder. A pharmacist demonstrated how to administer naloxone and kits were provided to attendees. Our Creative Media Director requested feedback from the students who created Grow502 and this was made into a compilation video featured on the Grow502 website (https://www.grow502.org/). Post-session surveys were sent to students, faculty, and community stakeholders; participation was voluntary. The Institutional Review Board of the University of Louisville School of Medicine reviewed and approved this study. Selected feedback is below: ٭ “Great! Love the opportunity to share expertise and collaborate with others in the community.”—Social Worker at University of Louisville Health Stroke Unit ٭ “The event was great on all fronts. The coordination was good and the content was really well-researched and reflected an important issue that does not get discussed nearly enough.”—Assistant Director of the LGBT Center at UofL Health Sciences ٭ “Great opportunity to strengthen the partnership between harm reduction and the medical field.”—Health Education Specialist at the Syringe Services Program at Louisville Metro Public Health ٭ “It was awesome to see diverse groups working on this project. I appreciated the variety of fields/backgrounds of the participants.”—Medical Student ٭ “Amazing work across the board. I felt like Grow 502 really made an impact on the Louisville community.”—Medical Student Discussion Grow502 sought to support communities impacted by injustices due to racism, limited health literacy, redlining, and limited green space by creating programming centered on education and empowerment. Public health data reported in the Louisville Health Equity Report were transformed into an interactive series and facilitated our evidence-based approach. Students acted as pivotal connections between the data and the community throughout the design of the sessions. A final piece to implementation was media engagement and direct action advocacy. Grow502’s programming increased the awareness of disparities and facilitated mitigation. Current and future considerations for Grow502 include hybrid programming based on disparities seen in local communities versus centering on a disease. For example, we are working toward addressing health disparities by collaborating with Feed Louisville to support Louisville’s unhoused population. This year, we held a virtual education panel and case review, helped prepare food to be delivered, and set up a pop-up clinic to provide wound care, dental care, and social services to unhoused individuals (Figure 2). Figure 2  Pop-Up Clinic in April 2022 Hosted by Grow 502, Trager Institute of Health, Kare Mobile, Feed Louisville, and Louisville Metro Public Health Providing public health education is vital to impacting disparities. The education of both community members and health care professionals is important to building a healthier community. Effective strategies to reduce disparities involve creating interventions with authentic engagement and context and intentional advocacy. Grow502 involves community stakeholders as active partners. We continue to intentionally seek effective collaborations and interventions that merge our mission and our diverse communities impacted by health disparities. The authors would like to thank Monica Ann Shaw, MD, MA, Vice Dean of Undergraduate Medical Education for her support and the funding provided by Undergraduate Medical Education Office ORCID iD: M. Onu Udoh https://orcid.org/0000-0002-4126-7575 ==== Refs References Benabentos R. Ray P. Kumar D. (2014). Addressing health disparities in the undergraduate curriculum: An approach to develop a knowledgeable biomedical workforce. CBE—Life Sciences Education, 13 (4 ), 636–640. 10.1187/cbe.14-06-0101 25452486 Center for Health Equity Report. (2017). 2017 health equity report: Uncovering the root causes of our health. https://louisvilleky.gov/government/center-health-equity/health-equity-report Centers for Disease Control and Prevention. (2020, December). What is health equity? U.S. Department of Health and Human Services, Center for Disease Control and Prevention. https://www.cdc.gov/coronavirus/2019-ncov/community/health-equity/racial-ethnic-disparities/index.html Chen F. Overstreet F. Cole A. Kost A. Brown Speights J. (2017). Racial and ethnic health disparities curricula in US Medical Schools: A CERA study. Primer, 1 , 6. 10.22454/PRiMER.2017.1.6 32944692 Greenwood B. Hardeman R. Huang L. Sojourner A. (2020). Physician–patient racial concordance and disparities in birthing mortality for newborns. Proceedings of the National Academy of Sciences of the United States of America, 117 (35 ), 21194–21200. 10.1073/pnas.1913405117 32817561 Hoffman K. M. Trawalter S. Axt J. R. Oliver M. N. (2016). Racial bias in pain assessment and treatment recommendations, and false beliefs about biological differences between blacks and whites. Proceedings of the National Academy of Sciences of the United States of America, 113 (16 ), 4296–4301. 10.1073/pnas.1516047113 27044069 Hooper M. W. Nápoles A. M. Pérez-Stable E. J. (2020). Covid-19 and racial/ethnic disparities. JAMA, 323 (24 ), 2466–2467. 10.1001/jama.2020.8598 32391864 National Academies of Sciences, Engineering, and Medicine. (2017). Communities in action: Pathways to health equity. https://www.ncbi.nlm.nih.gov/books/NBK425844/?report=classic Thornton R. L. Glover C. M. Cené C. W. Glik D. C. Henderson J. A. Williams D. R. (2016). Evaluating strategies for reducing health disparities by addressing the social determinants of health. Health Affairs, 35 (8 ), 1416–1423. 10.1377/hlthaff.2015.1357 27503966

36511091

PMC9749053

NO-CC CODE

2022-12-15 23:22:49

no

Health Promot Pract. 2022 Dec 13;:15248399221136533

utf-8

Health Promot Pract

10.1177/15248399221136533

oa_other

==== Front ACP spacp Asian Journal of Comparative Politics 2057-8911 2057-892X SAGE Publications Sage UK: London, England 10.1177/20578911221141759 10.1177_20578911221141759 Research Article “The pandemic has added to my miseries”: Bangladeshi migrant workers’ social protection revisited https://orcid.org/0000-0003-1826-0583 Rashid Syeda Rozana 95324 University of Dhaka , Bangladesh Ansar Anas 552664 Bonn Center for Dependency and Slavery Studies (BCDSS) , University of Bonn, Germany Md. Khaled Abu Faisal Department of International Relations, 216632 Bangladesh University of Professionals , Bangladesh Syeda Rozana Rashid, Department of International Relations, Faculty of Social Sciences, Dhaka 1000, Bangladesh. Emails: srr21rozana@gmail.com, rozana@du.ac.bd 12 12 2022 12 12 2022 20578911221141759© The Author(s) 2022 2022 University of Niigata Prefecture and SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. The protection of migrant workers has received renewed attention in the wake of the COVID-19 pandemic. This article depicts how unpreparedness, inadequate social security and support services, and pre-existing socio-economic disparities disproportionately impacted Bangladeshi migrant workers during the pandemic. Adopting a qualitative approach based on findings from existing literature and surveys and primary data collected through interviews with returnee Bangladeshi migrants from the Gulf States, the article argues that the dearth of institutional, legal, social, and political understanding of the needs of migrants remains the main impediment to a comprehensive social protection system. The findings call for designing a crisis response and recovery policy, preparing a returnee database and leveraging bilateral, regional, and global processes to ensure migrants’ uninterrupted protection at home and abroad. The article also underscores the importance of a nuanced understanding and practice of gendered social support, and above all, adopting a rights-based approach to labor migration. Bangladesh COVID-19 pandemic gender Gulf labor migration social protection edited-statecorrected-proof typesetterts19 ==== Body pmcIntroduction For decades, international labor migration has played a crucial role in reducing poverty, stimulating economic growth, and improving the socio-economic conditions of both sending and receiving countries. Despite being an important development actor, migrant workers’ vulnerabilities, particularly in times of crisis, remain largely unaddressed in the policy and academic domain (Mingot and Mazzucato, 2018). In many ways, the COVID-19 pandemic unveiled the fragile migration governance and absence of adequate protective mechanisms on both ends of the migratory chain (Jamil and Datta, 2021; Karim et al., 2020). It is now well-recognized that international labor migrants were hit particularly hard in terms of health and socioeconomic indicators, both in their country of origin and destination (Leach et al., 2020; Siddiqui, 2021a). Furthermore, increasingly rigid immigration controls, arbitrary recruitment practices, and employer-tied visa sponsorship largely exclude them from the protections of national labor law. Such ambiguous legal status makes it difficult, if not impossible, for migrants to seek adequate social protection in their host countries (Al-Ali, 2020; Babar, 2020). Amid various state-led support programs, migrants received little attention during the peak of the pandemic in their countries of origin and destination (Gentilini, 2021). This is demonstrated through migrants’ forced deportation, arbitrary dismissal, and exclusion from social service schemes launched by many receiving states in the aftermath of the pandemic (Ansar, 2022; Menon and Vadakepat, 2021). Moreover, the remittance sent by the migrant workers is a ‘crucial lifeline’ for at least 800 million relatives living back home in low- and middle-income countries. This financial support, estimated at US$554 billion in 2019, is vital for the left-behind family members to meet their nutritional, health, educational, and housing needs. Such abrupt disruption of this vital financial supply line has serious implications for the daily lives and sources of livelihood of many migrant households around the world (Leach et al., 2020; UNNM, 2020: 2). Against this backdrop, this article examines the plight of low-wage migrant workers during the pandemic. Using Bangladesh as a case, the various risks faced by migrant workers in the aftermath of COVID-19 and the reasons that led to such disruption are explored in the article. It asks the following questions: What forms of challenges were encountered by men and women migrants during the pandemic? What were the major impediments to migrants’ social safety during the crisis? How can migrants’ social safety be improved to address the future crisis? The aim is to identify gaps in readiness that arose in a crisis and understand how its breadth and scope differ in a ‘crisis’ versus a ‘normal’ situation. There are compelling reasons to revisit the social protection of migrants, particularly in the aftermath of the COVID-19 pandemic. What is distinctive about the pandemic-induced crisis, as compared to previous emergencies, is the scale and scope of its impact. Unprecedented in recent history, the pandemic has had a devastating impact on the lives and health of people throughout the world. In addition, there is a wide discrepancy in migrants’ abilities to endure and recover from the pandemic's various shocks, which warrants detailed scholarly inquiry. With emerging research starting to document the manifold impacts of the pandemic on the lives and livelihoods of Bangladeshi migrants (i.e. Ansar, 2022; Jamil and Datta, 2021; Karim et al., 2020; Siddiqui, 2021a), the larger context of persistent gaps in the existing labor migration policy and social protection mechanisms and crisis preparedness by the states is largely missing from the analysis. While migrants’ rights and related issues in the Global North received increasing attention from scholars (Holzmann and Koettl, 2015; van Ginneken, 2013), a similar understanding of the Global South remains limited. Moreover, existing literature on the Global South focuses on migrants’ lack of access to formal welfare in receiving countries, emphasizing the role of informal mechanisms, i.e. social networks, friends, and families, through which migrants access resources (Rashid, 2016; Sabates-Wheeler and Feldman, 2011; Schiller, 2011). Nevertheless, the informal mechanisms have certain limitations and are deemed insufficient to strengthen migrants’ fallback position in a crisis (Rashid, 2016). Therefore, there is a clear need for renewed attention to the legal and socio-economic barriers migrants experience while trying to access formal support services. This article attempts to fill in these gaps and proposes a durable solution for future crisis preparedness, one in which workers’ welfare lies at the core of the system. The COVID-19 pandemic has prompted policy-oriented research due to the breadth of the pandemic's cross-sectoral implications for migrants’ lives and livelihoods. The logistical, social, and health considerations in the aftermath of the pandemic, therefore, have clearly influenced the methodology of this article, which is centered on research flexibilities and broader collaboration on various fronts for data collection. Given the persistently evolving nature of the pandemic and its consequences, this study does not claim to provide a comprehensive pandemic-led impact analysis on migrant lives. Instead, it attempts to underline the deep-seated flaws and challenges in the existing protection framework for migrants within the South–South migration corridor. Doing so, it expects to bring more clarity and accountability to labor migration governance in the future. The study, therefore, involves a multi-pronged qualitative research method. The thematic analysis presented here is based on both primary and secondary data. Between June 2020 and February 2022, a total of 25 in-depth interviews (IDIs) were carried out, consisting of 20 male and five returnee female migrant workers from the Gulf countries. In addition, the authors conducted 10 focus group discussions (FGDs) with 66 female returnees in female migrant-rich locations, such as Manikganj, Rupganj, Dhaka, Dohar, and Keraniganj. Additionally, 20 key informant interviews (KIIs) with experts and relevant stakeholders in Bangladesh were conducted for data validation. The respondents were informed about the purpose of the study and its strict use in expanding critical scholarship in migration studies and positive policy framing on contract labor migration. For anonymity, the actual names of respondents are changed throughout the study. The primary findings were triangulated and substantiated by secondary sources that consist of newspaper reports, government statistics, NGO reports, and recent academic publications. Particularly, two major studies were instrumental in overall validation of information: first, the Bangladesh Civil Society for Migrants’ (BCSM) study that surveyed 200 households in 21 districts during May and June 2020, and second, a BRAC study on 558 returnee Bangladeshi workers, conducted between 2019 and 2020. The article is organized into six sections. Following the introduction, the second section draws on social protection literature to develop the study's conceptual framework. The third section presents an overview of Bangladeshi labor migration to the Gulf countries. The fourth section examines the predicament that returnee Bangladeshi workers face in the Country of Destination (CoD), as well as in their home country. The fifth section analyzes the factors contributing to protection gaps. The concluding section summarizes important findings and offers policy suggestions. Conceptual framework Social protection refers to “resources and strategies to deal with social risks, such as poverty or obligations and needs of care, which might impede the realization of life chances and well-being” (Bilecen et al., 2019: 1). It is usually targeted at the most vulnerable groups in society, involving various tangible and intangible resources and activities that reduce their social risks (Faist et al., 2014; Sabates-Wheeler and Feldman, 2011). This article considers both formal protection provided by the state, markets, and organizations, as well as informal protection, manifested in interpersonal networks (Mingot and Mazzucato, 2018). Elements of social protection Fundamentals to ensure the social protection of the workers are: social insurance, social assistance, and labor market regulation (ILO 2006). Social insurance provides workers with protection against crisis, social assistance provides support in crisis, whereas labor market regulation indicates basic rights and standard of work. The latter encompasses a wide range of activities, such as social protection laws, bilateral social security agreements, employer–employee relations, and workers’ capacity to invest in social and cultural ties at home (Avato et al., 2010; Rashid, 2016). As far as labor migrants are concerned, a combination of formal rules, regulations, policies, and norms set by the state, as well as informal personal and community support capitalized on by migrants for wellbeing, comprises the idea of social protection (Bilecen and Barglowski, 2014). Social protection and social policies In the rapidly expanding set of policies and programs, social policies are increasingly being seen as a part of social protection that denotes public actions taken in response to levels of vulnerability, risk, and deprivation within a society (Conway et al., 2000). Taking this notion into account, the United Nations (UN) prescribes social protection as “a set of public and private policies and programmes undertaken by societies in response to various contingencies to offset the absence or substantial reduction of income from work; to provide assistance to families with children as well as provide people with basic health and housing” (United Nations, 2000: 4). This conceptualization is significant in identifying the rights, needs, and empowerment to produce an all-encompassing policy framework of social protection (Sabates-Wheeler and Devereux, 2008). Social protection priorities and politics during crisis Research proves that social protection favors those groups with the strongest voice, while ‘new’ poor are also created during a crisis (Barrientos and Hulme, 2008). Studies in the context of financial crises in Latin America and East Asian countries show that short-term emergency responses to the crisis turn out to be dysfunctional in the medium term, but it takes considerable time and political capital to change course (Barrientos and Hulme, 2008). The scale and scope of protection in a crisis period may well vary from a ‘normal period’ of time, as witnessed in the context of migration in the aftermath of the pandemic. Adopting a sustainable social protection initiative in this situation depends on how inclined the political environment is towards the prioritization of social protection. Both scholars and policymakers tend to focus on “definitional debates, policy design and impact evaluations, with relatively little analysis of the ways in which politics shape policy” (Lavers and Hickey, 2016: 388). Gendered social protection The framework of social protection also calls for the investigation of a broader landscape of activities, both formal and informal, and embedded in the hierarchies of sex, age, ethnicity, and class (Amelina, 2016). While many workers with regular job contracts and stable incomes have been exposed to a situation of “absolute unpreparedness” (Lo and Hsieh, 2020), protection needs demand a gender analysis. However, “an isolated view on gender obscures the relationality and internal differentiation of social categories which operate alongside intersectional hierarchies” (Bilecen et al., 2019: 2), since it operates with other markers of difference, such as ethnicity, age, marital and education status, and social class (Ansar, 2022). As such, gender needs to be understood within “the set of mutually constitutive structures and practices which produce gender differentiation, gender inequalities, and gender hierarchy in a given society” (Orloff, 1996: 52). Based on the above conceptualization, the remainder of this article examines the challenges migrants faced in negotiating access to limited provisions for social protection. Labor migration from Bangladesh and the state of migrants’ social protection Labor migration has been recognized as a vital poverty alleviation and development strategy for Bangladesh (Siddiqui, 2016). The Bureau of Manpower, Employment and Training (BMET) data show that, since 1976, more than 14 million Bangladeshis have migrated to the Gulf and other Southeast Asian countries for unskilled and semi-skilled jobs in the infrastructure, manufacturing, and service sectors (BMET, 2022). The statistics, however, do not include the number of returnees. The Bangladeshi migrants predominantly hail from poor rural households with few other means to diversify their earnings (Rashid, 2016). With the outbreak of the COVID-19 pandemic, the country observed a staggering decrease of 69% in its overseas labor migration from 2019 to 2020. A total of only 217,669 Bangladeshis migrated abroad in 2020, compared to 700,159 in 2019 (BMET, 2022). Notably, more than 90 per cent of individuals who migrated in 2019 did so between January and March, prior to the pandemic's impact on the overall migration process (BMET, 2022). The figure increased significantly in 2021, as 485,893 Bangladeshis managed to migrate amid the pandemic-related restrictions and shrinking work opportunities across the globe. Women's migration during the period corresponds with the overall flow. A total of 80,143 women workers migrated from Bangladesh in 2021, compared to 21,934 in 2020 (BMET, 2022). During the pandemic, Bangladesh received a record number of returnee migrants owing to job loss, arbitrary dismissal, and forced deportation. Around 408,000 migrant workers returned to the country in 2020, followed by 64,646 in 2021 while approximately 100,000 new workers, who had completed all the necessary procedures before the COVID-19 outbreak, could not migrate because of air travel restrictions (Siddiqui 2021b: 01). A crucial aspect of labor migration from Bangladesh is the regular remittance flow, which remained persistent despite the pandemic. In 2019, Bangladesh received a little over US$18 billion in total remittances, which increased in 2020, when the country received US$21.7 billion (BMET, 2022). There can be two reasons for continuous remittance flow. First, due to contract termination and abrupt job loss, many migrants from the Gulf States had to return to Bangladesh permanently. Many of them sent their previous savings before their departure. Second, nearly 35% of the total remittances received in 2020 came from North America, Europe, and South East Asia. While the flow reflects some consistency, families traditionally dependent on remittances from the Gulf countries received less during the pandemic than during normal times (BMET, 2022). Legal and institutional arrangements Migrants’ contribution to the national economy and the migrant households’ dependency on remittances has essentially brought about legal and policy changes favoring migration. Bangladesh is the first country in South Asia to adopt an articulated normative framework for migrant workers. Several national laws and policies have been enacted to regulate labor migration from Bangladesh, including the Overseas Employment and Migrants Act (OEMA) 2013, the Wage Earners’ Welfare Fund Rules 2002, the Emigration Rule 2002, and the Recruiting Agent License and Conduct Rules 2002 (ILO, 2013). Over the past five years, the Government of Bangladesh (GoB) has enacted the Expatriate Welfare and Overseas Employment Policy 2016, the Overseas Employment and Migrant Rules 2017, and the Wage Earners’ Welfare Board Act 2018. OEMA articles 29 and 30 of Chapter VII contain directives on the security and welfare of expatriate workers and their left-behind family members. Recently, in its eighth five-year development plan, the Bangladesh government emphasizes, among other issues, the necessity of enhancing the skills base for the development of new overseas labor market opportunities; safeguarding migrants’ human and labor rights; revising current policies to foster migration-sensitive health policies and strategies; and incorporating the needs of women migrant workers into migration policy (GED, 2020). Until the 1990s, overseas migration used to be managed by the Bureau of Manpower Employment and Training (BMET). In 2001, the Ministry of Expatriates’ Welfare and Overseas Employment (MoEWOE) was established to address the overall welfare and equal opportunities for all migrants. Later, several government agencies, including the Wage Earner's Welfare Board (WEWB), Technical Training Centres (TTCs), the District Employment and Manpower Office (DEMO), and Probashi Kallyan Bank (PKB), were established to decentralize migration governance and facilitate the specific needs of the international labor market and migrants. State of migrants’ formal social protection WEWB offers financial support to returnee disabled migrants and the family members of deceased migrants. The GoB introduced two compulsory insurance schemes in 2019, worth Bangladeshi Taka (BDT) 200,000 (US$2135) and BDT500,000 (US$5335), to bring aspirant migrants under Probashi Kormi Beema Nitimala (KII with WEWB Official).1 The schemes are valid for two years and migrant workers aged between 18 and 58 years are eligible to receive a benefit depending on the severity of their injuries. In addition, scholarships and higher education grants are also offered by WEWB to migrants’ left-behind children. The government has also introduced a safe home for women migrants, call centers for migrant workers, legal aid, and counselling services in the country of destination (CoDs) (WEWB, 2022). Currently, 30 labor wings in different Bangladeshi missions are assigned to look after migrant workers’ welfare at CoDs. Following the pandemic, the GoB distributed BDT209 crores (US$2.5 million) among 9215 returnees as cash incentives (Siddiqui, 2021b). Later, BDT500 crores (US$5.5 million) were allocated from public funds to reintegrate returnees. A three-step reintegration program was introduced for returnees, consisting of immediate support to facilitate the safe return of the stranded migrants and an allowance of BDT5000 (US$55) at the airport. Further, midterm support of BDT2 billion was allocated for migrants for a one-time loan from BDT100,000 (US$1066) to a maximum of BDT500,000 (US$5335) at a 4% interest rate (Siddiqui, 2021b). Additionally, the WEWB pledged a BDT25,000 (US$267) subsidy to Saudi Arabia-bound migrants to cover their accommodation costs for compulsory quarantine following entry into Saudi Arabia (Daily Star, 2021c). The GoB thus claims to sufficiently extend its formal social safety nets for migrants during the pandemic. Despite all these efforts, surveys undertaken by civil society organizations and international agencies in 2020 and 2021 reveal that migrants’ access to information and support services was limited due to various challenges. These challenges include a lack of a comprehensive database and information delivery mechanism, inadequate preparedness, and gaps in protection and reintegration policies, which have been detailed in the following. The state of migrants’ protection during the COVID-19 pandemic Drawing on secondary information from survey reports, literature, newspapers, and the first-hand data collected through in-depth interviews with migrant workers, this section elucidates the pandemic-induced plights of migrants. Insecurity and uncertainty at destination With the surge of COVID-19 cases and subsequent mobility restrictions, the lives and livelihoods of Bangladeshi migrants in the Gulf were stuck in limbo. Mobility restrictions helped contain the spread of COVID-19; however, the lockdown also had a detrimental effect on migrants’ living conditions, as it jeopardized their access to money-earning activities and other essential services, including health care (Dhaka Tribune, 2020; Siddiqui, 2021b). Migrants, who usually dwell in cramped dormitory-style labor camps in the Gulf countries, were particularly prone to the COVID-19 virus. The risk of infection in migrant-intensive areas rose due to the absence of systematic quarantine, screening, and contact-tracing measures (BRAC, 2021). As of April 2021, 2729 expatriate Bangladeshis were reported to have died of coronavirus in 23 countries, including the Gulf States, whereas Saudi Arabia alone saw the death of 1228 Bangladeshis (Ejaz, 2021). As macroeconomic conditions, i.e. the drop in oil price, reduced demand in the construction sector, and business closures worsened in the Gulf countries, many Bangladeshi migrants were laid off arbitrarily, often without any notice or benefits. In addition to being without social assistance or other means of income, many workers could not return home (Siddiqui, 2021b). The narratives of the arbitrarily terminated returnee migrants corroborate the scenario outlined above. Bilkis (aged 28), for instance, claimed that her Saudi employer refused to pay her the last four months’ salary before returning to Bangladesh in July 2020, though she had worked for him for five years. The unpaid dues amount to US$1200. She stated:I met every demand of my employer and his family members. However, they did not pay my wages. Each time I asked for my salary, I was beaten and tortured by the employer and other family members. I tried to contact the labor office [of the Bangladesh embassy in Jeddah] and my recruitment agents. No one even picked up my phone call. Moreover, I could not contact the Saudi government, as my employer never allowed me to leave his house before I made it to the airport to return home [to Bangladesh]. Maruf (35) returned home in June 2020 from the United Arab Emirates (UAE) after three years of working in a gas station. He shared his ordeal:On the third day of declaring the lockdown, when I went to my work, the employer simply said I was no longer required, and he would pay the due wages the following day. When I went there to get the wages, he simply refused to pay and threatened me, saying I took money from the cash counter and that if I made any trouble, he would simply lodge a complaint against me to the police. I did nothing wrong, and the allegations were completely false, but the police would never trust me. I also had no proof of my unpaid wages because I usually got paid in cash. Syed (30) has been working in Qatar since 2018. The recruitment agency told him that, in addition to his monthly salary of US$250, he would be provided with free medical insurance and accommodation by the employer. The evening his test report arrived, two men came to his shared company accommodation and removed him from it. As he stated:I was sick, infected by COVID-19, and suddenly became homeless. I had to spend the night at a fellow Bangladeshi's place, and the next morning when I went to the hospital for treatment; they said I had no insurance coverage. My company had that hospital listed in their file, and we were told that we could get free treatment from there, which was eventually a big lie. I had no money to buy food, no place to sleep, and no way to return home. The narratives portray insecurities endured by migrants at CoDs in the absence of legally enforceable contracts, arbitrary dismissal, wage theft, and inaccessibility to support services. The anecdotes also illuminate the loopholes and vacuums engrained in migration practices. Return and reintegration fiasco at home As a result of the COVID-19 outbreak and the associated economic impact, a notable number of Bangladeshi workers were deported from Gulf countries. Even individuals with years of work experience were subjected to the loss of their work permits. Irregular migrants had even greater difficulties deciding to return, as they had to adhere to stringent restrictions governing cross-border mobility (İçduygu, 2020). For undocumented migrant workers, returning meant losing their chances of earning nominal sums to support the family back home and, most importantly, the chances of remigration. Upon return, migrants had to face quarantine debacles at the airport. A survey report shows that 86% of returnees received no assistance after returning to Bangladesh during the first few months of the pandemic (USAID and Winrock International, 2020). Migrants frequently complained of maltreatment at the airport, and a temporary quarantine center was established near the airport. After her deportation from Dubai to Dhaka, Sharmin (37) was taken to the Hajj Camp, the temporary quarantine center. She narrated her painful experience in this way:I asked the authority more than 10 times why the toilet flush was not working. After this repeated query, one of them responded to me, saying the condition [of the camp] is a five-star hotel for us, compared to where we stayed previously and where we will go. It was devastating hearing that response from our fellow countrymen. We have no respect there [in the Gulf], and here [Bangladesh] is also no exception. Many migrants’ attempts to reintegrate into society at home were met with greater stigma and marginalization, due to a lack of preparation and support services upon their abrupt return. For example, according to a survey conducted by BRAC Bangladesh, 29% of returnees reported that their relatives and neighbors did not accept their return and did not behave well with them (BRAC, 2021). Another study conducted by Winrock International on 155 returnees revealed that nearly half (48%) of returnee migrants were maltreated by community leaders and members and by friends or family (USAID and Winrock International, 2020). The reliance on informal support also became untenable. The following cases illustrate this well:When my Kuwaiti Kafeel [local sponsor], under whom I worked for more than 10 years, informed me that he could not pay my salaries and I better go back home for now, I also thought returning home was the best option in the given situation. Nevertheless, returning home, which I perceived as a better alternative, became a nightmare once I landed at the airport. I was treated by them as if I was carrying the virus. Many friends and relatives I always supported financially did not even bother contacting me. My cousin, whom I used to send remittances to, suddenly said no remittances left, although I used to send them almost regularly. When I met the BMET officials to file a complaint, they simply said, “it is a family matter—better solve it through discussion.” (Kabir, 40) Tofazzal (33), who worked on a date plantation for five years in Oman, used to receive relatively fair wages. However, he felt that returning home was the worst decision. As he said:My Kafeel told me that I could stay, but I would not be paid. I would only be provided with food and accommodation for the interim [pandemic lockdown] period. I thought it would be better to go home and stay with my family. However, relatives started acting weird within a few days of my return. They acted as if it was my failure not to be able to stay there. They are happy as long as they get the Riyals from abroad and do not want me around. The issue of wage theft was closely associated with the arbitrary return (Siddiqui, 2021b). Most returnee migrants arrived empty-handed, while many of them were already indebted. Some migrants spent days in misery, unable to pay off their debts and handle living expenses since they had little or no earnings. Mamun (33), who worked in a gas station in Saudi Arabia for five years, claims his employer refused to pay the last three months’ salary, referring to the dire business condition amid the nationwide lockdown. He was fired the next day when he requested that his employer pay him at least a fraction of his wages. His unpaid dues amount to US$1400. Tarana (30), another returnee worker from the UAE, claims that despite following all of her employer's requirements, she was not only denied her rightful wages but also verbally threatened with a fabricated accusation of petty theft to police. The sharp decline in remittances and the subsequent return of the migrants brought financial havoc to households solely dependent on remittances. In the wake of COVID-19, the intermittence and, in some cases, absence of remittance inflows substantially reduced households’ income and, consequently, their ability to meet basic needs (Ansar et al., 2021). At that time, the families resorted to mainly three different types of coping strategies: self-mobilization, which corresponds to the measures adopted by household members to reduce their spending or increase their income from other sources; solidarity-based mobilization, involving aid from friends and close family members in the established tradition of welfare societies; and institutional mobilization, including requests by individuals to the state, civil society, or the market, to overcome financial difficulties (Billah et al., 2021). Returnee reintegration became another significant issue that unveiled gaps in the social protection of the migrants. An IOM survey shows that almost 70% of the migrants repatriated to Bangladesh from abroad between February and June 2020 remained unemployed (IOM, 2020a). A major challenge was access to information regarding reintegration services (Siddiqui, 2021b). Although the GoB allocated a special reintegration fund for returnee migrants, only 2% of the migrants were covered (Shawon, 2021). A BRAC study on 558 returnees found that 47% of returnees remained unemployed even a year after their return (Dhaka Tribune, 2021). Migrants and their families had to resort to informal means of coping with the sudden loss of livelihood, such as seeking support from friends and family or turning to money lenders. Whereas a section of stranded and returnee migrants desired to remigrate to the Gulf countries, this proved difficult for various reasons. Many migrant workers’ permits expired, and some had to wait months to be vaccinated, as it was mandatory to travel abroad. In 2020, over 50,000 prospective migrants and returnees were awaiting flights abroad, especially to the Gulf (Daily Star, 2021). Uncertainty gripped them, as their flights were cancelled and they lost the prospect of employment and opportunities to renew their contracts (Bhuiyan, 2020). They also anticipated considerable financial losses from non-refundable quarantine fees and air tickets (Daily Star, 2021). Since the beginning of 2022, Gulf States have eased the compulsory quarantine requirements for foreign workers. As the GoB decided to provide Pfizer vaccines against COVID-19 to persons travelling to Saudi Arabia and Kuwait, migrant workers had to wait a long time and rely on approved vaccines (Molla, 2021). Meanwhile, globally, the airfare has increased exponentially, making it harder for migrants to afford it and return to work. Shariful (27), a stranded migrant, explains his ordeals as follows:I decided to migrate abroad to avoid other problems [financial uncertainties] and to save a lump sum for the future … I ate up whatever I saved during the past few years abroad. And the suffering that I am going through now is not comparable to anything … The pandemic has added to my miseries only. The ordeal of women migrants Evidence abounds on how the pandemic has intensified existing gender inequalities and created new gender-biased consequences that have disproportionately impacted women migrant workers (Foley and Piper, 2020; Kelly, 2021). Due to the domestic character of their work, seclusion at home, and disconnection from the rest of the community, women migrants were, in general, barred from participating in labor market monitoring and receiving any type of social support in the CoDs. During the pandemic, Bangladeshi women workers had to deal with multiple risks because they were forced to keep working and could not get medical support (Rashid et al., 2021). There was increased demand for health, social, and domestic services in the Gulf. Sharifa (35), a Gulf returnee, stated:When I wanted to return to Bangladesh, my employer insisted that I stay one more year. But I was not willing to do that. My employer made a fake COVID-19-positive certificate so that I could not pass the airport security. Later, I was told that flights had been cancelled. But I knew they had not been. Another returnee, Halima (28), stated: “I had to work regularly, even during the initial surge of the coronavirus. My employer and her family members were infected with coronavirus, but still, I had to come in contact with them and take care of them.” During the pandemic, Bangladesh received 77 dead bodies of women workers from the Gulf countries, though it could not be confirmed whether these women died from coronavirus (Daily Star, 2021b). In 2021 alone, around 50,000 women migrant workers returned to Bangladesh from 21 countries for various reasons, including grave violations of workers’ rights and gender-based violence (Daily Star, 2021c). Since a considerable section of women migrants bear the sole responsibility for their left-behind family, including parents and children, their sudden layoff and return led to profound consequences for their families. The following statement by Sharifa (35) makes this explicit: “Before the pandemic, I used to support my family by sending remittances. Returning to Bangladesh would mean that I would no longer be able to support my family. That was indeed a difficult decision for me.” Adding to their plight, women returnees in the rural areas endured social stigma and community rejection, which often derives from the nature of their work as domestic help (Rashid et al., 2021). Zohora (34), who returned from Jordan in March 2020, explained:I do not know how they [neighbours and community members] came to know about my return home. They locked my main door from outside so that I could not go out. I felt like I was an untouchable person, though I had no symptoms of coronavirus. What obstructs migrants’ social protection? In what follows, the study analyzes the deep-seated structural and operational reasons behind migrants’ protection gaps during crises. Legal, structural, and socio-economic conditions at CoDs The study highlights three stumbling blocks to having a functional social security system at the destination. These include the unavailability of inclusive labor migration policies in CoDs, the absence of compulsory insurance schemes, and the low socio-economic status of the workers. Gulf countries, in particular, have insufficient safety and rights provisions for migrant workers (Alsharif and Malit, 2020). In most Gulf countries, the Kafala system regulates the relationship between employers and migrant workers—a restrictive sponsorship system binding a worker's immigration status to her/his employer (Robinson, 2021). The pandemic, thus, left workers completely dependent upon their employers for their livelihood and residency. It also allowed employers to have near-total control over migrant workers’ incomes, living circumstances, nutrition, ability to work abroad, and even ability to return home. The bulk of Bangladeshi migrant workers in Gulf countries are low-wage, semi-skilled, and working in precarious conditions. Local labor laws favored the employer over the immigrant, further disempowering workers. Despite introducing the ‘employers pay’ model, the UAE and Qatar are not signatories to most international human rights and labor conventions, limiting their obligation to international systems. Interviews with migrants revealed that most Bangladeshi workers in the Gulf countries do not possess any health insurance provisions. Subsequently, high medical costs and limited or no bargaining scope discourage migrants from seeking treatment from public hospitals or clinics. During the pandemic, migrants’ access to health care and social safety was restricted not only by lockdowns and social distancing but also by this systematic violation of their rights. Social policies are usually bound to the ‘nation-states’ and often restricted by legal status. Hence, they are often deprived as significant ‘others’ despite their contribution to the host country's economy (Ansar, et al., 2021; Vullnetari and King, 2008). Both documented and undocumented migrants suffer disproportionately from the adverse health and socio-economic consequences of the COVID-19 pandemic for these reasons (İçduygu, 2020). Like elsewhere, undocumented Bangladeshi workers in the Gulf feared being reported to the immigration authorities and deported if they sought assistance, which reduced their willingness to come forward for screening, testing, contact tracing, or treatment). The system of not allowing forming and joining trade unions is also considered an additional obstacle to raising concerns about their health and safety in the workplace. Migrants’ inclusion in the public health services during the COVID-19 situation was also hindered in cases where governments did not have precise information on the distribution of the migrant population and affected individuals. Interviews with the labor attaché and MoEWOE officials reveal that missions are understaffed compared to the number of migrants. The quality and quantity of support services also depend on the personal effort and initiatives of the service providers. In 2019, a call center was established in the Jeddah mission by the Bangladesh government, which has since then attended to hundreds of calls each month and addressed most of them. Similar services were rarely available elsewhere. Missions often refer to the legal and cultural system that barred them from looking after the welfare of the workers, whereas workers commonly stated that the embassy's services were inadequate or unreachable. The above legal and organizational factors obstruct workers from having “resources and strategies” to deal with the pandemic, which impedes their “life chances and wellbeing” (Bilecen et al., 2019: 1). Located at the bottom of the CoD's society, no tangible and intangible resources and activities were made available for foreign workers to reduce the social risk. The dearth of institutional, social, and political priorities at home At least five factors obstructed migrants’ COVID-time social protection at home: the absence of a systematic database of returnees, inadequate preparedness, inability to provide emergency support services, non-coordination among different agencies, and pre-existing gaps in the reintegration approach. Systematically accumulated return migration data is still unavailable in Bangladesh. Thus, identifying and bringing vulnerable migrants under support services is often cumbersome and untenable. The MoEWOE had to rely on others, i.e. the immigration authority and the CSOs, to trace migrants and provide assistance to the returnees. Bangladesh has yet to develop a standard crisis preparedness policy for migrants. Ineffective monitoring of the health and overall wellbeing of the returnee migrants, insufficient COVID testing services, and poor quarantine facilities are common threads throughout migrants’ stories. That the country has no emergency support framework for migrants was demonstrated during previous crises. After the fall of the Gaddafi regime in Libya, Bangladesh relied on the International Organization for Migration (IOM) for the repatriation of stranded Bangladeshis from war-torn Libya (IOM News, 2011). Recently, it sought the assistance of neighboring India to bring back Bangladeshis from Ukraine after the country encountered a Russian invasion (Hindustan Times, 2022). Despite having a solid legal base, Bangladesh's National Social Protection Strategy does not contain any provision for migrant workers. The country also failed to provide highly required emergency assistance in cash or kind to migrants hard-hit by the pandemic. In effect, the state often endorses the community's perception of migrants as the fortunate section of society and capable of taking care of themselves. As opined by the CSOs, this might be a reason why stimulus packages for migrants and their families arrived late. After sustained advocacy by CSOs, the government allocated BDT200 crores (US$2.4 million) from the Wage Earners’ Welfare Fund created from the compulsory conscription of the outgoing migrants. The newly introduced insurance schemes were no use for the laid-off and deported migrants, as they only cover death and disability. The lack of harmonization of activities among concerned Ministries and authorities appears to be a fundamental flaw in Bangladesh's migrant support services (IOM, 2020b). Coordination between MoEWOE, the civil aviation authority, the immigration authority, and the Ministry of Health was rarely observed in addressing the issues of migrants’ contact tracing, quarantine, vaccination, and remigration. Non-governmental interventions were also sporadic, overlapping, and donor dependent. Over the decades, ‘reintegration’ received scant attention from policymakers compared to ‘migration,’ due to the latter's contribution to national development. It is only more recently that the Bangladeshi government has initiated a returnee reintegration policy for migrants, taking lessons from the pandemic and also curbing the irregular migration to the West. The COVID-19 situation saw hastily made reintegration plans for returnees that proved ineffective in many ways. PKB was late in offering recovery packages because of human resource constraints. Moreover, the terms and conditions of the loans and bureaucratic inefficiency made the effort unwieldy (Ansar et al., 2021). To sum up, the nature and scope of workers’ formal social protection were extended to some extent during the pandemic. Nevertheless, the understanding and commitment to adopt an ‘overall preparedness’ were largely missing in the context. To add to the woes of migrants, the informal social safety nets embedded in social relations and networks also proved unreliable and ineffective during the pandemic. Gendered accessibility to protection In a country like Bangladesh, women's socio-economic status is generally low, and migrant women are doubly vulnerable through being both a woman and a migrant. Therefore, formal social protection should be gender sensitive, considering women's limited access to information regarding migration support services. As the preceding section shows, the live-in Bangladeshi domestic workers had to rely on their employers for information and support, as they could rarely contact the missions abroad. The Kafala system allowed employers an unchallenged authority over employees, further shrinking the latter's protection opportunities. Though the ‘employers’ pay model’ in some CoDs contributed to the reduced cost of women's migration, it provides little safeguard against pre-existing vulnerabilities to being abused and exploited. As elaborated above, very few formal support services are, in effect, available for men and women at CoDs. The support services of Bangladesh missions were, in most cases, either unavailable or inaccessible, particularly to women domestic workers. Women usually resort to the recruitment agents, as they find them their ‘guardians’ or ‘protectors.’ However, they proved to be rather detrimental to the safety and security of the migrant women, as they forced them to carry on the work with the same or another employer in precarious work conditions. The Bangladeshi CSOs had limited social and legal leverage to protect these women. As a CSO interviewee commented, “Migrant women's problems are much-reported and well-discussed now. Yet, very little has been done to eradicate the root causes of women's social risk and vulnerability at CoDs.” In effect, the paucity of a well-designed crisis recovery policy prioritizing the specific needs of women migrants was on full display during the pandemic. The positioning of Bangladeshi women workers at the bottom of the social order makes them incapable of challenging the status quo. As a result, they become the first to drop out of the support system in an emergency, as was evident during the COVID-19 pandemic. Towards a comprehensive social protection system Highlighting the plights of Bangladeshi migrants during the COVID-19 pandemic, the article has uncovered how and why the much-needed protections of workers are affected during a crisis. As demonstrated, despite numerous initiatives and a plethora of rules, laws, and organizations, Bangladeshi migrant workers mostly remained unprotected and mistreated throughout the pandemic. Formal social protection was unavailable, inadequate, and inaccessible during the pandemic. Mobility restrictions, exclusion from stimulus packages, wage theft, arbitrary dismissal from jobs, and forced deportation were common experiences amongst stranded migrants. Upon return, returnees had to withstand mistreatment at the airport and quarantine centers and social stigma and rejection in their native villages. Prolonged uncertainty and unemployment impacted the socio-economic status of migrants and their households. The absence of formal reintegration policies and discontinuation of informal but trusted family support added to their woes, aggravating their psycho-social wellbeing. Digging deeper into the protection gaps, the Bangladesh case also illustrates the lack of institutional, social, and political understanding of the needs of migrants’ welfare. Whereas the Kafala system benefits employers with unfettered and inexplicable control over their employees, the non-existence of insurance coverage coupled with inadequate and inaccessible support services was found to be the major protection hindrances at CoDs. Similarly, the number crunch, information gaps, inadequate preparedness and immediate support services, and insufficient coordination between stakeholders affected returnees’ access to recovery and support services. Above all, well-designed crisis response and recovery plans prioritizing the gender-specific needs of the migrants were absent during the pandemic. The importance of the aforementioned findings and analysis lies in their highlighting of the importance of adopting an inclusive, formalized, and win-win labor arrangement for the future. The quest for protection in a crisis will remain elusive if the system does not acknowledge this gap. The study broadly proposes a five-point recommendation to move the discussion forward. First, an all-encompassing social protection system with well-crafted crisis response and recovery policy is needed. Variations in the needs, abilities, requirements, and expectations of migrant workers across the region call for policies to ensure a coordinated response in times of crisis. Necessary resources—both financial and human—should be allocated for emergency support provisions and crisis preparedness. In addition, migrants’ insurance schemes should be expanded to cover the risks associated with crisis, such as the pandemic. Second, the absence of a systematic database combining relevant information on existing and returnee migrants was felt more than ever. Such a comprehensive database can assist policymakers and agencies in developing successful strategies and appropriate actions. Therefore, in addition to expanding the database of current migrants, a system should be in place to include returnee migrants’ data in national migration statistics. Third, CoDs should categorically address the labor migrants’ welfare provisions in normal and crisis times and abolish discriminatory laws. Migrants’ access to information, health care, and support services at CoDs should be prioritized. All bilateral, regional, and global processes and instruments need to be leveraged to ensure the uninterrupted protection of migrants at CoDs. Fourth, the role of gender and the plight of women migrants have never been as prevalent following the COVID-19 pandemic. A nuanced understanding of gendered social protection is required for policy formulation and negotiation at the bilateral and multilateral forums to safeguard women's position at CoDs and home. Equally important is the inclusion of migrants’ left-behind family members. Last but not least, it is critical to move away from remittance-based labor migration policies and ensure a rights-based approach to migration. Policies and practices need to be in place to pursue quality in migration over quantity. Acknowledgements We are grateful to all migrants who participated in this study, without whom the research would not have been possible. We are also thankful to key informants for their views on the main findings. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article ORCID iD: Syeda Rozana Rashid https://orcid.org/0000-0003-1826-0583 1. A requirement for insurance coverage for workers abroad went into effect on December 19, 2019. For more details, see: https://wewb.portal.gov.bd/site/page/f5714923-ac86-4d79-aeaf-4de920bdc5ce/- (accessed 30 October 2022). ==== Refs References Al-Ali N (2020) Covid-19 and feminism in the Global South: Challenges, initiatives and dilemmas. European Journal of Women's Studies 27 (4 ): 333–347. DOI: 10.1177/1350506820943617. Alsharif F Malit Jr (2020) Migration and the COVID-19 pandemic in the Gulf: A study of foreign expatriate worker communities’ coping attitudes, practices, and future prospects in Dubai and Jeddah. Available at: https://www.kas.de/en/web/rpg/detail/-/content/migration-and-the-COVID-19-pandemic-in-the-gulf (accessed 12 July 2021). Amelina A (2016) Transnationalizing Inequalities in Europe: Sociocultural Boundaries, Assemblages and Regimes of Intersection. New York: Routledge. Ansar A (2022) Bangladeshi women migrants amidst the COVID-19 pandemic: Revisiting globalisation, dependency and gendered precarity in South–South labour migration. Global Networks: 1–14. DOI: 10.1111/glob.12368. Ansar A Sultana M Das RC (2021) In pursuit of resilience: Revisiting migration and gender-based violence amid COVID-19 pandemic. In: Siddiqui T (eds) The Other Face of Globalisation: COVID-19, International Labour Migrants and Left-behind Families in Bangladesh. Dhaka: RMMRU, pp. 133–161. Avato J Koettl J Sabates-Wheeler R (2010) Social security regimes, global estimates, and good practices: The status of social protection for international migrants. World Development 38 (4 ): 455–466. Babar ZR (2020) Migrant Workers Bear the Pandemic’s Brunt in the Gulf. Current History 119 (821 ): 343–348. DOI: 10.1525/curh.2020.119.821.343. Barrientos A Hulme D (2008) Social protection for the poor and poorest: An introduction. In: Barrientos A Hulme D (eds) Social Protection for the Poor and Poorest Concepts, Policies and Politics. London: Palgrave Macmillan, pp. 3–24. Bhuiyan KH (2020) Migrant workers without sponsors unable to return to Saudi Arabia. The Dhaka Tribune. Available at: https://www.dhakatribune.com/world/2020/09/30/migrant-workers-without-sponsors-not-unable-to-return-to-ksa (accessed 11 July 2021). Bilecen B Barglowski K (2014) On the assemblages of informal and formal transnational social protection. Population, Space and Place 21 (3 ): 203–214. Bilecen B Barglowski K Faist T , et al . (2019) Gendered dynamics of transnational social protection. Comparative Migration Studies 7 (47 ): 1–14. Billah M Akash T Ahsan N (2021) Left behind migrant households: Coping with partial or full suspension of Remittances. In: Siddiqui T (eds) The Other Face of Globalisation: COVID-19, International Labour Migrants and Left-behind Families in Bangladesh. Dhaka: RMMRU, pp. 107–132. BMET (2022) Statistical reports. Available at: http://www.old.bmet.gov.bd/BMET/stattisticalDataAction (accessed 15 September 2022). BRAC (2021) A survey on ‘exploring and analysing the socio-economicsocio-economic situation of returnee migrants after one year of their return during covid-19 pandemic’. Dhaka: BRAC Centre. Conway T de Haan A Norton A (2000) Social Protection: New Directions of Donor Agencies. London: Overseas Development Institute (ODI). Daily Star (2019) Govt to introduce insurance schemes for migrant workers. The Daily Star, 11 December. Available at: https://www.thedailystar.net/country/govt-introduce-insurance-schemes-for-bangladeshi-migrant-workers-1838917 (accessed 5 May 2022). Daily Star (2021a) 50,000 female migrants returned during pandemic. The Daily Star, 7 March. Available at: https://www.thedailystar.net/backpage/news/50000-female-migrants-returned-during-pandemic-2056041 (accessed 11 July 2021). Daily Star (2021b) Govt to spend Tk 25,000 to Saudi-bound workers’ bank accounts for hotel quarantine. The Daily Star, 6 June. Available at: https://www.thedailystar.net/bangladesh/news/govt-send-tk-25000-saudi-bound-workers-bank-accounts-hotel-quarantine-2106001 (accessed 11 July 2021). Daily Star (2021c) Resume int’l flights, help migrants. The Daily Star, 14 April. Available at: https://www.thedailystar.net/backpage/news/resume-intl-flights-help-migrants-2077341 (accessed 11 July 2021). Dhaka Tribune (2020) Brac: 87% returnee migrants struggling financially. Dhaka Tribune, 20 May. Available at: https://www.dhakatribune.com/bangladesh/2020/05/22/brac-over-half-of-returnee-migrants-in-need-of-financial-aid-now (accessed 13 May 2021). Dhaka Tribune (2021) Study: Half of migrant workers still unemployed. Dhaka Tribune, 30 April. Available at: https://www.dhakatribune.com/bangladesh/2021/04/30/brac-half-of-migrant-workers-still-unemployed (accessed 15 July 2021). Ejaz R (2021) 1,228 Bangladeshis die of coronavirus in Saudi Arabia. Prothom Alo. Available at: https://en.prothomalo.com/bangladesh/1228-bangladeshis-die-of-coronavirus-in-saudi-arabia (accessed 13 July 2021). Faist T Bilecen B Barglowski K , et al . (2014) Transnational social protection: Migrants’ strategies and patterns of inequalities. Population, Space and Place 21 (3 ): 193–202. Foley L Piper N (2020) COVID-19 and women migrant workers: Impacts and implications. International Organization for Migration. Geneva: (IOM). GED (2020) 8th five-year plan July 2020—June 2025: Promoting Prosperity and Fostering Inclusiveness. Dhaka: Bangladesh Planning Commission. Gentilini U (2021) A game changer for social protection? Six reflections on COVID-19 and the future of cash transfers. Available at: https://blogs.worldbank.org/developmenttalk/game-changer-social-protection-six-reflections-covid-19-and-future-cash-transfers (accessed 18 February 2022). Hindustan Times (2022) Sheikh Hasina thanks PM Modi for rescuing 9 Bangladeshis from war-hit Ukraine. Hindustan Times, 9 March. Available at: https://www.hindustantimes.com/india-news/ukraine-war-bangladesh-pm-hasina-thanks-modi-for-rescuing-9-citizens-101646801740448.html (accessed 10 June 2022). Holzmann R Koettl J (2015) Portability of pension, health, and other social benefits: Facts, concepts, and issues. CESifo Economic Studies 61 (2 ): 377–415. İçduygu A (2020) Stranded irregular migrant workers during the COVID-19 crisis: The question of repatriation. Geneva: IOM. ILO (2006) Social security for all: Investing in global social and economic development: A consultation. Geneva: ILO. ILO (2013) Overseas Employment and Migration Act 2013. Available at: https://www.ilo.org/dyn/natlex/natlex4.detail?p_lang=en&p_isn=95796 (accessed 14 September 2022). IOM (2020b) Bangladesh migration governance framework. Dhaka: IOM. IOM (2020a) Impact of Covid-19 on Bangladesh overseas migrant workers. Dhaka: IOM. IOM News (2011) IOM accelerates repatriation of Bangladeshi migrants fleeing Libya. Available at: https://www.iom.int/news/iom-accelerates-repatriation-bangladeshi-migrants-fleeing-libya (accessed 10 June 2022). Jamil R. Dutta U. (2021) Centering the margins: The precarity of bangladeshi low-income migrant workersduring the time of COVID-19. American Behavioral Scientist 65 (10 ): 1384–1405. DOI: 10.1177/00027642211000397. Karim MR Islam MT and Talukder B (2020) COVID-19′s impacts on migrant workers from Bangladesh: In search of policy intervention. World Development 136 : 1–4. DOI: 10.1016/j.worlddev.2020.105123. Kelly L (2021) Direct and indirect impacts of the covid-19 pandemic on women and girls. Foreign, Commonwealth & Development Office. Available at: https://www.gov.uk/research-for-development-outputs/direct-and-indirect-impacts-of-the-covid-19-pandemic-on-women-and-girls (accessed 20 May 2022). Lavers T Hickey S (2016) Conceptualising the politics of social protection expansion in low income countries: The intersection of transnational ideas and domestic politics. International Journal of Social Welfare 25 (4 ): 388–398. Leach M MacGregor H and Scoones I (2020) Post-pandemic transformations: How and why COVID-19 requires us to rethink development. World Development 138 : 1–11. DOI: 10.1016/j.worlddev.2020.105233. Lo MCM Hsieh HY (2020) The “societalization” of pandemic unpreparedness: Lessons from Taiwan’s COVID response. American Journal of Cultural Sociology 8 (3 ): 384–404.32983457 Menon DV Vadakepat VM (2021) Migration and reverse migration: Gulf-Malayalees’ perceptions during the COVID-19 pandemic. South Asian Diaspora 13 (2 ): 157–177. Mingot E Mazzucato V (2018) Providing social protection to mobile populations: Symbiotic relationships between migrants and welfare institutions. Journal of Ethnic and Migration Studies 44 (13 ): 2127–2143. Molla MAM (2021) Migrants to get Pfizer, Moderna vaccines. The Daily Star. Available at: https://www.thedailystar.net/bangladesh/news/migrant-get-pfizer-moderna-vaccines-2119989 (accessed 11 July 2021). Orloff A (1996) Gender in the welfare state. Annual Review of Sociology 22 (1 ): 51–78. Rashid SR (2016) Uncertain tomorrows: Livelihoods, capital and risk in labour migration from Bangladesh. Dhaka: UPL. Rashid SR , Mohammad H Jahid F , et al. (2021) Facing COVID-19 in countries of destination: Health shocks, income risks, detention, deportation and wage theft. In: Siddiqui T (eds) The Other Face of Globalisation COVID-19, International Labour Migrants and Left-behind Families in Bangladesh. Dhaka: RMMRU, pp. 37–61. Robinson K (2021) What is the Kafala system? Council on Foreign Relations. Available at: https://www.cfr.org/backgrounder/what-kafala-system (accessed 11 July 2021). Sabates-Wheeler R Devereux S (2008) Transformative social protection: The currency of social justice. In: Barrientos A Hulme D (eds) Social Protection for the Poor and Poorest Concepts, Policies and Politics. London: Palgrave Macmillan, pp. 64–84. Sabates-Wheeler R Feldman R (2011) Migration and Social Protection: Claiming Social Rights Beyond Borders. London: Palgrave Macmillan. Schiller NG (2011) A global perspective on migration and development. In: Faist T Fauser M Kivisto P (eds) The Migration-development Nexus. Migration, Diasporas and Citizenship Series. London: Palgrave Macmillan, pp. 29–56. Shawon A (2021) 700C incentive for expats: Disbursement in disarray. Dhaka Tribune. Available at: https://www.dhakatribune.com/bangladesh/2021/03/11/700c-incentive-for-expats-disbursement-in-disarray (accessed 24 November 2021). Siddiqui T (2016) International labor migration and remittance. In: Riaz A Rahman MS (eds) Routledge Handbook of Contemporary Bangladesh. London: Routledge, pp. 197–206. Siddiqui T (2021a) Introduction. In: Siddiqui T (eds) The Other Face of Globalisation: COVID-19, International Labour Migrants and Left-behind Families in Bangladesh. Dhaka: RMMRU, pp. 1–11. Siddiqui T (2021b) Labour migration from Bangladesh 2020 achievements and challenges. Available at: http://www.rmmru.org/newsite/wp-content/uploads/2021/09/Migration-Trends-Report-2020.pdf (accessed 14 August 2021). United Nations (2000) Enhancing social protection and reducing vulnerability in a globalizing world. Commission for Social Development. UNNM (2020) The impact of COVID-19 on family remittances: A lifeline cut for migrant families. Available at https://migrationnetwork.un.org/sites/g/files/tmzbdl416/files/policy_brief-_remittances_in_the_time_of_covid-19.pdf (accessed 15 September 2022). USAID and Winrock International (2020) Situational assessment of labor migrants in Asia: Needs and knowledge during Covid-19. Dhaka: USAID and Winrock International. Van Ginneken W (2013) Social protection for migrant workers: National and international policy challenges. European Journal of Social Security 15 (2 ): 209–221. Vullnetari J King R (2008) ‘Does your granny eat grass?’: On mass migration, care drain and the fate of older people in rural Albania. Global Networks 8 (2 ): 139–171. Wage Earners Welfare Board (WEWB) (2022) Available at: http://www.wewb.gov.bd/ (accessed 10 June 2022).

0

PMC9749054

NO-CC CODE

2022-12-15 23:22:49

no

2022 Dec 12;:20578911221141759

utf-8

oa_other

==== Front Educ Eval Policy Anal Educ Eval Policy Anal EPA spepa Educational Evaluation and Policy Analysis 0162-3737 1935-1062 SAGE Publications Sage CA: Los Angeles, CA 10.3102/01623737221139285 10.3102_01623737221139285 Briefs Teacher Attrition and Mobility in the Pandemic https://orcid.org/0000-0003-4260-4040 Goldhaber Dan American Institutes for Research University of Washington https://orcid.org/0000-0002-5479-4147 Theobald Roddy American Institutes for Research rtheobald@air.org 12 12 2022 12 12 2022 0162373722113928511 3 2022 15 7 2022 28 9 2022 7 10 2022 © 2022 AERA 2022 American Educational Research Association This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. We contextualize the magnitude of teacher attrition during the pandemic, including from the 2020–2021 school year to the 2021–2022 school year, using longitudinal data on teachers in Washington since the 1984–1985 school year. The teacher attrition rate after the 2020–2021 school year (7.3%) increased by almost one percentage point from the attrition rate after the 2019–2020 school year (6.4%), but these rates are well within the range of turnover rates observed during pre-pandemic years. The increase in turnover during the pandemic was also smaller than pre-pandemic differences in turnover between high- and low-poverty classrooms in the state, and these inequities in turnover between high- and low-poverty classrooms decreased during the pandemic relative to pre-pandemic years. retention teacher research policy observational research National Center for Analysis of Longitudinal Data in Education Research https://doi.org/10.13039/100015418 edited-statecorrected-proof typesetterts1 ==== Body pmcTeacher attrition is squarely in the policy spotlight. Major news outlets such as CNN, the New York Times, Newsweek, NPR, the Wall Street Journal, and the Washington Post all have featured recent stories highlighting the impact of the COVID-19 pandemic on teacher shortages. All these news outlets point to teacher attrition as a primary culprit, using phrases like “mass exodus,” “in crisis,” and “Great Resignation” to describe the current state of affairs in the teacher labor market. Below is a small sampling of their characterization of the situation: Burned out teachers are leaving the classroom for jobs in the private sector, where talent-hungry companies are hiring them—and often boosting their pay—to work in sales, software, healthcare and training, among other fields. (Dill, 2022) 55% of [teachers] say they will leave teaching sooner than they had originally planned, according to a poll of its members by the nation’s largest teachers union. (Kamenetz, 2022) The educator profession—a critical cornerstone of American life—is in crisis. (Maxouris & Zdanowicz, 2022) A mass exodus could worsen existing staff shortages in schools and cripple the education system in the U.S. (Rahman, 2022) The coronavirus is vastly exacerbating that shortfall, experts say, by prompting many teachers to leave the profession or take early retirement. (Singer, 2021) But the evidence of increased teacher attrition in these media reports is thin, as they cite anecdotes from specific former teachers or surveys about what current teachers say they might do in the future. Moreover, recent evidence from data collected before the pandemic (Nguyen et al., 2022) finds that only about a third of teachers who report an intent to leave actually do leave their position at the end of the year. So, what do we know at this point about rates of teacher attrition during the pandemic? Contrary to the definitive takes from press reports, the existing large-scale, quantitative evidence suggests that teacher attrition after the first year of the pandemic (i.e., after the 2019–2020 school year) dropped relative to the preceding school years (e.g., Bacher-Hicks et al., 2021; Diliberti & Schwartz, 2021; Goldhaber & Theobald, 2022). This likely reflects teachers hunkering down after the 2019–2020 school year in the midst of the uncertainty of a pandemic (Ayaita & Stürmer, 2020; van Huizen & Alessie, 2019). However, there are reasons to think that things could be different after the 2020–2021 school year and coming into the 2021–2022 school year. For example, surveys of teachers highlight the considerable challenges of teaching during the pandemic (e.g., Diliberti et al., 2021; Kraft et al., 2021), and teacher attrition rates also tend to be inversely related to unemployment rates (Eagan et al., 2022; Goldhaber & Theobald, 2022). In 2021, the economy continued to improve, labor markets were tight, and, arguably, there was less COVID-related economic uncertainty (Baker et al., 2020). All of this suggests that we might expect to see teacher attrition increase coming the 2021–2022 school year. Moreover, given mounting evidence that the COVID-19 pandemic disproportionately impacted learning gains for historically marginalized student groups (Goldhaber et al., 2022), we should be particularly attuned to teacher turnover across different schools serving different populations of students. In this brief, we explore teacher attrition during the pandemic, including from the 2020–2021 school year to the 2021–2022 school year, using publicly available data on the public teaching workforce in Washington. This dataset, the S-275, includes a snapshot of the state’s public school workforce (including teachers and other certificated personnel) from October 1 of each school year since 1984–1985, and does not include data on late teacher hires or short-term substitute teachers. Teacher attrition and mobility data from the last year of available national data of teacher turnover (2011–2012; Goldring et al., 2014) suggest that teacher turnover rates in Washington in this year (15.0%) are comparable to the national estimate (15.8%). The state introduced new employment categories in 2021 to 2022 for teachers funded by federal special purpose aid related to the coronavirus pandemic (i.e., ESSER funds); we consider these as “teachers” for the purposes of this analysis (i.e., teachers who move into one of these ESSER-funded positions not counted as moving into a non-teaching category), though we note that the funding of these positions may be temporary. The stacked bar plots in Figure 1 show the proportion of teachers in different years who at the end of the year: (a) left their schools and the state’s public school workforce entirely (black), (b) left their current teaching position for a nonteaching position (e.g., administration or instructional coach; dark gray) within the state’s public school system, or (c) left their school for another public school teaching position in the state (light gray). We present these three statistics separately because they address different policy questions. From the perspective of the state, the overall attrition rate (i.e., the black and dark gray bars) represents the number of teachers who need to be replaced for the next school year, though attrition from the workforce altogether (black bars) is likely a larger problem than movement into noninstructional positions (dark gray bars) that also play important roles in public schools. But from the perspective of an individual school, all teachers who leave the school (i.e., also including teachers in the light gray bars) need to be replaced next year, so school mobility is also an important factor in terms of school teaching stability and hiring demands. The total teacher turnover rate (i.e., the sum of these three proportions) is shown at the top of each bar. Figure 1. Selected teacher turnover rates in Washington, 1984–1985 through 2020–2021 school years. Consistent with findings from other states from the first 2 years of the pandemic—for example, from Arkansas (Camp et al., 2022) and Massachusetts (Bacher-Hicks et al., 2022)—the overall takeaway from Figure 1 is that the teacher attrition rate after the 2020–2021 school year increased from the attrition rate after the 2019–2020 school year, but these rates are well within the range of turnover rates observed during pre-pandemic years. Specifically, Figure 1 shows four important points: The attrition rate of teachers from the public school workforce after the 2020–2021 school year (7.3%, last set of bars) increased by almost one percentage point from the attrition rate after the 2019–2020 school year (6.4%, next-to-last set of bars), and is also higher than the last pre-pandemic year (6.7%, fourth set of bars). The proportional increases in attrition were relatively consistent for early- and late-career teachers, as well as for teachers of color and White teachers (not shown in Figure 1 but available upon request). Combined with the increased movement of teachers into non-teaching positions in public school districts (which increased from 2.0% after 2019–2020 to 2.7% after 2020–2021), the percentage of teacher “leavers” as defined by federal reports increased by about 1.6 percentage points, which represents a nearly 20% increase in the proportion of teacher leavers compared to the first pandemic year. Rates of school-to-school mobility also increased by nearly a percentage point (to 7.8% from 6.9%). Thus, the total teacher turnover rate (17.8%) increased by 2.5 percentage points in the second pandemic year relative to the first—this is the second largest change in teacher turnover rates seen since 1984 to 1985—and was over one percentage point higher than in the average pre-pandemic year (16.7%, first set of bars). The overall teacher turnover rate after the 2020–2021 school year is well within the range of turnover rates observed during pre-pandemic years (second and third sets of bars), and the rate of attrition from the workforce is the highest since the 2006–2007 school year. Do these statistics support recent assertions of a “Great Resignation” or “Teacher Exodus” from public schools (Rahman, 2022)? These terms are, of course, in the eye of the beholder, but it is simultaneously true that (a) teacher attrition in Washington increased substantially after the second year of the pandemic, resulting in hundreds of additional open teaching positions relative to previous school years and (b) even these increased attrition rates are not inconsistent with what we have seen in past years. The first point is important, as teacher attrition really does predict district staffing challenges. To show this in more concrete terms, we calculate teacher turnover rates by district and compare them to recently collected data by Goldhaber and Gratz (2022) on school district teacher vacancy rates in October 2021. We find that the relationship between district attrition rates after the 2020 –2021 school year and the percentage of open teaching positions in the district in October 2021 is positive and statistically significant (r = 0.23, t = 3.11). In other words, districts that had more teachers leave the workforce after the 2020–2021 school year also had more difficulty hiring teachers by the start of the 2021–2022 school year. And there is no doubt that these staffing challenges are of concern given evidence that teachers hired late tend to be less effective (Papay & Kraft, 2016) and that staffing challenges have led schools to have to close (e.g., Velez, 2021). But it is also important to keep this increase in teacher turnover in perspective. Not only are teacher turnover rates after the 2020–2021 school year lower than turnover rates from the mid-2000s, but the increase in these overall turnover rates is actually smaller than the average difference in turnover rates between high- and low-poverty classrooms in Washington state in a typical school year. To illustrate this, we use data from the three most recent school years to calculate the turnover rates in high-poverty classrooms (defined as the top quartile of the percentage of students in the classroom receiving free or reduced-priced lunch [FRL]) compared to turnover rates in low-poverty (bottom quartile FRL) classrooms. We present these rates in Figure 2. In the most recent pre-pandemic year, the difference in turnover rates between high- and low-poverty classrooms was 3.4 percentage points, a larger difference than we observed in turnover rates between the two pandemic school years. Figure 2. Teacher turnover rates in Washington by classroom poverty level, 2018–2019 through 2020–2021 school years. Figure 2 shows that teacher turnover rates from both types of classrooms increased after the second pandemic year. But it is also notable from Figure 2 that the increase in teacher turnover we document after the second pandemic year was driven disproportionately by increased attrition rates from relatively advantaged classrooms in the state. In other words, consistent with evidence from Massachusetts (Bacher-Hicks et al., 2022), inequities in turnover between high- and low-poverty classrooms decreased during the pandemic relative to pre-pandemic years; in fact, after the most recent year of the pandemic, the difference in overall turnover rates between high- and low-poverty classrooms was less than one percentage point. These trends are consistent for other measures of classroom and school disadvantage (e.g., based on the percent of students of color; results available upon request). In sum, the trends in teacher attrition during the pandemic in Washington suggest that many of the recent media stories about rising teacher attrition rates are accurate in direction but, arguably, not in magnitude. In particular, while there was clearly an uptick in attrition in Washington and other states, we would not characterize the attrition rates in Washington after the second year of the pandemic as a “mass exodus” of teachers. But these results should only be generalized to these contexts, and importantly, they do not imply that we should be unconcerned; we agree with Will (2022) that we should take reports of teacher burnout and dissatisfaction seriously, even if they do not lead to attrition. Instead, we would argue that some of the recent reporting on teacher attrition has mischaracterized the extent to which teacher attrition is driving staffing challenges and has not done enough to highlight some of the long-term issues of differential attrition in different school and classroom settings. The lack of nuance in some media reports is problematic to the degree that policymakers react to these reports, rather than crafting solutions to more specific staffing challenges that have existed in the teacher labor market long before the pandemic. Authors DAN GOLDHABER, PhD, is the director of the National Center for Analysis of Longitudinal Data in Education Research (CALDER) at American Institutes for Research, and also the director of the Center for Education Data and Research (CEDR) at the University of Washington. His research focuses on issues of educational productivity and reform at the K–12 level, the broad array of human capital policies that influence the composition, distribution, and quality of teachers in the workforce, and connections between students’ K–12 experiences and postsecondary outcomes. RODDY THEOBALD, PhD, is a principal researcher in CALDER at American Institutes for Research. His research focuses on teacher education, teacher licensure, special education, and career and technical education. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the National Center for Analysis of Longitudinal Data in Education Research (CALDER), which is funded by a consortium of foundations. For more information about CALDER funders, see www.caldercenter.org/about-calder. We wish to thank the State of Washington Office of Superintendent of Public Instruction (OSPI) for collecting and providing the data we utilize, as well as Matt Barnum for comments that improved the brief. All opinions expressed in this paper are those of the authors and do not necessarily reflect the views of our funders or the institutions to which the author(s) are affiliated. ORCID iDs: Dan Goldhaber https://orcid.org/0000-0003-4260-4040 Roddy Theobald https://orcid.org/0000-0002-5479-4147 ==== Refs References Ayaita A. Stürmer K. (2020). Risk aversion and the teaching profession: An analysis including different forms of risk aversion, different control groups, selection, and socialization effects. Education Economics, 28 (1 ), 4–25. Bacher-Hicks A. Chi O. Orellana A. (2021). COVID-19 and the composition of the Massachusetts teacher workforce. Boston University. Bacher-Hicks A. Chi O. Orellana A. (2022). Two years later: How COVID-19 has shaped the teacher workforce. Boston University. Baker S. R. Bloom N. Davis S. J. Terry S. J. (2020). Covid-induced economic uncertainty (No. w26983). National Bureau of Economic Research. Camp A. Zamarro G. McGee J. (2022). Changes in teachers’ mobility and attrition in Arkansas during the first two years of the COVID-19 pandemic. University of Arkansas. Diliberti M. Schwartz H. (2021). The K-12 pandemic budget and staffing crises have not panned out—Yet. RAND Corporation. Diliberti M. Schwartz H. Grant D. (2021). Stress topped the reasons why public school teachers quit, even before COVID-19. RAND Corporation. Dill K. (2022, February 2). Teachers are quitting, and companies are hot to hire them. The Wall Street Journal. Eagan J. Hwang N. Koedel C. Ladd H. Sorensen L. (2022). Teacher attrition and the business cycle. Teachers College Record. Advance online publication. Goldhaber D. Gratz T. (2022). School district staffing challenges in a rapidly recovering economy (CALDER Flash Brief No. 29-0122). National Center for Analysis of Longitudinal Data in Education Research (CALDER). Goldhaber D. Kane T. J. McEachin A. Morton E. Patterson T. Staiger D. O. (2022). The consequences of remote and hybrid instruction during the pandemic (No. w30010). National Bureau of Economic Research. Goldhaber D. Theobald R. (2022). Teacher attrition and mobility over time. Educational Researcher, 51 (3 ), 235–237. Goldring R. Taie S. Riddles M. Owens C. (2014). Teacher attrition and mobility: Results from the 2012-13 teacher follow-up survey. U.S. Department of Education. Kamenetz A. (2022, February 1). More than half of teachers are looking for the exits, a poll says. NPR. Kraft M. A. Simon N. S. Lyon M. A. (2021). Sustaining a sense of success: The protective role of teacher working conditions during the COVID-19 pandemic. Journal of Research on Educational Effectiveness, 14 (4 ), 727–769. Maxouris C. Zdanowicz C. (2022, February 5). Teachers are leaving and few people want to join the field. Experts are sounding the alarm. CNN. Nguyen T. Bettini E. Redding C. Gilmour A. F. (2022). Comparing turnover intentions and actual turnover in the public sector workforce: Evidence from public school teachers (EdWorkingPaper: 22–537). Annenberg Institute at Brown University. Papay J. Kraft M. (2016). The productivity costs of inefficient hiring practices: Evidence from late teacher hiring. Journal of Policy Analysis and Management, 35 (4 ), 791–817.28966429 Rahman K. (2022, February 17). America’s teacher exodus leaves education system in crisis. Newsweek. Singer N. (2021, January 19). Pandemic teacher shortages imperil in-person schooling. The New York Times. van Huizen T. Alessie R . (2019). Risk aversion and job mobility. Journal of Economic Behavior and Organization, 164 , 91–106. Velez M. (2021, November 9). Seattle and Bellevue schools cancel classes for students Friday due to lack of staff. The Seattle Times. Will M. (2022, February 25). Will there really be a mass exodus of teachers? Education Week.

0

PMC9749055

NO-CC CODE

2022-12-15 23:22:49

no

Educ Eval Policy Anal. 2022 Dec 12;:01623737221139285

utf-8

Educ Eval Policy Anal

10.3102/01623737221139285

oa_other

==== Front Health Educ Behav Health Educ Behav HEB spheb Health Education & Behavior 1090-1981 1552-6127 SAGE Publications Sage CA: Los Angeles, CA 36510857 10.1177/10901981221139168 10.1177_10901981221139168 Original Manuscript Understanding COVID-19 Risk Perceptions and Precautionary Behaviors in Black Chicagoans: A Grounded Theory Approach https://orcid.org/0000-0002-0344-442X Chebli Perla PhD, MPH 1 McBryde-Redzovic Aminah MPH 2 Al-Amin Nadia MPH 2 Gutierrez-Kapheim Melissa MS 2 https://orcid.org/0000-0001-8742-5658 Molina Yamilé PhD, MPH, MS 23 https://orcid.org/0000-0003-1675-1119 Mitchell Uchechi A. PhD, MSPH 2 1 New York University Grossman School of Medicine, New York, NY, USA 2 University of Illinois Chicago, Chicago, IL, USA 3 University of Illinois Cancer Center, Chicago, IL, USA Perla Chebli, Section for Health Equity, Department of Population Health, New York University Grossman School of Medicine, 180 Madison Ave., 8th Fl, New York, NY 10016, USA. Email: Perla.Chebli@nyulangone.org 13 12 2022 13 12 2022 10901981221139168© 2022 Society for Public Health Education 2022 Society for Public Health Education This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Objectives To determine whether actual community-level risk for COVID-19 in the Black community influenced individual perceptions of community-level and personal risk and how self-assessment of personal risk was reflected in the adoption of COVID-19 precautionary behaviors. Methods Semistructured interviews were conducted with 20 Black Chicago adults from February to July 2021. A grounded theory approach was used for the qualitative analysis and initial, focused, and theoretical coding were performed. Results We developed a grounded model consisting of four major themes: (a) Pre-Existing Health Conditions; (b) Presence of COVID-19 Infection in Participant Social Network; (c) COVID-19-Related Information, Participant Trust, and Perceived Personal Risk; and (d) Perceived Higher Burden of COVID-19 in the Black Community. Conclusions Higher perceptions of personal risk were shaped by pre-existing health conditions and experiences with COVID-19 in one’s social network but were not influenced by perceived higher burden of COVID-19 in the Black community. Policy Implications Black adults’ perceptions of their individual risk and precautionary behaviors were not congruent with public health data and recommendations. Therefore, COVID-19 messaging and mitigation should be informed by local community engagement and transparent communication. COVID-19 risk perceptions grounded theory African Americans edited-statecorrected-proof typesetterts1 ==== Body pmcIntroduction The coronavirus pandemic (COVID-19) began spreading in the United States and around the world by January 2020 (Centers for Disease Control and Prevention [CDC], 2022a). By March of 2022, the total number of COVID-related deaths had reached around 6 million globally and close to 1 million in the United States, with an estimated burden of disease of 32,000 disability-adjusted life years (DALYs) (Center for Systems Science and Engineering [CSSE], Johns Hopkins University [JHU], 2022; Fan et al., 2021). While COVID-related morbidity and mortality continue to be a significant source of concern, COVID-19 has also led to substantial life changes and disruption to normal daily activity, all of which necessitated significant lifestyle and behavior change. In particular, the stay-at-home orders, business closures, remote learning and work, as well as mask mandates and social distancing orders have prompted people to rapidly adopt new precautionary behaviors. Decades of public health and behavioral research have shown that behavior change is challenging (Conner & Norman, 2017). Adopting new behaviors or adapting old ones is likely even more difficult while enduring the stress of a global pandemic. These adaptations are particularly important for populations disproportionately affected by the pandemic. The COVID-19 pandemic has highlighted long-standing socioeconomic and health disparities (Ndugga & Artiga, 2021). Racial and ethnic minorities have been disproportionately impacted by COVID-19, putting them at greater risk of morbidity and mortality. For instance, although only 20% of counties in the United States are predominantly Black, they accounted for more than half of all the COVID-19 diagnoses and close to 60% of the deaths (CDC, 2022b; Millett et al., 2020). Even with the availability of the COVID-19 vaccines, Black people are getting vaccinated at lower rates (12%) compared to their White counterparts (60%) and subsequently continue to be at 2.4 times higher risk of hospitalization and 1.6 times higher risk of death from COVID-19 (CDC, 2022c). Given their increased risk of contracting and dying from COVID-19 as well as the history of lower health-care access, economic disadvantage, and health-care mistrust (Abrams & Szefler, 2020; Bogart et al., 2020), it is important to consider factors influencing the COVID-related precautionary behaviors of racial minorities during the pandemic in order to improve the health of these communities and curtail health disparities. One significant predictor of behavior may be perception of risk, a recurrent construct in health behaviors theories (e.g., Health Belief Model, Protection Motivation Theory; Becker, 1974; Rogers, 1975). The motivation to engage in (e.g., vaccination) or avoid (e.g., smoking) certain behaviors is thought to be influenced by individuals’ assessment of the associated probability of health consequences. This pathway was demonstrated empirically in a meta-analysis that showed that risk perception, particularly perceived likelihood and susceptibility, was a significant predictor of adult vaccination uptake (Brewer et al., 2007). Although absolute risk can be thought of as objective, the formation of risk perceptions is subjective and multifaceted; it is the product of personal experiences (e.g., disease diagnoses among family members), exposure to health information (e.g., how the threat is portrayed in the media), or contextual factors (e.g., imminence of threat; Chen & Kaphingst, 2011; Shepperd et al., 2000; Slovic, 1987). In the context of COVID-19, characterized by its unprecedented scale, constant flow of health information, and co-occurring domestic strife, it is important to understand how risk perceptions were formulated and whether they were linked to engagement in precautionary behaviors. This study explores perceptions of COVID-19 risk in a sample of community-dwelling Black adults in Chicago. This setting is significant; Chicago exemplified the profound racial inequities exposed by the pandemic: cases and deaths were concentrated in vulnerable, low-income neighborhoods with a high proportion of Black residents (Kim & Bostwick, 2020; Maroko et al., 2020). These patterns are not surprising, given the legacy of redlining and segregation in Chicago, which are a direct manifestation of structural racism (Nardone et al., 2020). Historically Black neighborhoods were and continue to be disconnected from economic opportunities and investments, leading to high rates of poverty in these areas (Massey et al., 1994). Residents of these neighborhoods are therefore at greater risk for suboptimal health outcomes (Williams & Collins, 2001). Understanding how a community at higher risk collectively appraises risk and uptakes precautionary behaviors is essential to designing and implementing targeted and effective mitigation strategies. The aims of this study are to determine (a) whether actual community-level risk (i.e., the documented disproportional burden of COVID-19 on communities of color and Black communities specifically) influenced individual perceptions of community-level and personal risk and (b) how self-assessment of personal risk was reflected in the adoption of COVID-19 precautionary behaviors (e.g., mask-wearing and vaccine uptake). Methods The research team used a grounded theory approach to understand how individual perceptions of COVID-19 risk were shaped and subsequently influenced the adoption of precautionary behaviors among Black adults in Chicago during the pandemic (Charmaz, 2014). The grounded theory methodology is optimal to generate an understanding of novel phenomena that may not be readily explicated by existing literature and theories. This approach was thus appropriate to understand behavior amid the unfamiliar environment imposed by the COVID-19 pandemic. Setting and Sample All study procedures and materials were approved by the University of Illinois at Chicago Institutional Review Board. This study is part of a larger quantitative survey of community stressors, resources, and mental health among Black and White Chicagoans. Following the survey, we recruited Black participants to engage in in-depth interviews on their pandemic-related experiences. We used a purposeful, heterogeneous sampling strategy to reflect the diversity of the larger survey sample on the dimensions of age, gender, income, and utilization of community resources during the pandemic. To be eligible for interviews, participants had to self-identify as non-Hispanic Black, be 18 years of age or older, currently reside in Chicago, and able to conduct the interview in English. Data Collection The six co-authors developed the interview guide, which was pilot tested and iteratively refined as new concepts emerged from interviews. Exemplar questions include: “At the beginning of the pandemic, what were your beliefs about how badly the coronavirus infection would affect your health? How have your beliefs changed over time?” and “How do you think the Black community has been affected relative to other groups?” Interviews occurred between February and July 2021—for context, the Pfizer and Moderna vaccines were granted emergency use authorization in December 2020. Interviewees participated in a one-time interview over Zoom which lasted 45–60 minutes and were facilitated by trained interviewers (P.C., N.A., and A.M.R.); participants were compensated for their time. After each session, interviewers created field notes which were compiled on a shared project spreadsheet to track emergent and divergent patterns. The interview team met weekly to discuss field notes and refinements to the interview guide. Interviews were recorded, transcribed verbatim, checked for accuracy, and uploaded to Dedoose for qualitative analysis. Data Analysis A team of three coders (P.C., N.A., and A.M.R.) led a constant comparative analysis of the transcripts and each transcript was analyzed by at least two coders to achieve investigator triangulation (Charmaz, 2014). Our coding process followed three iterative phases: initial coding, focused coding, and theoretical coding (Thornberg & Charmaz, 2014). The coding team developed a codebook informed by the data that were iteratively refined as analysis progressed. Codes were then grouped in higher-order categories then relationships between categories were clarified, producing a grounded theoretical model. Coding continued until no more themes were produced (i.e., theoretical saturation). All team members met weekly via Zoom to discuss coding and emergent themes, update the codebook, and resolve discrepancies through discussion until consensus was reached. Results We conducted 20 interviews with Black community members in Chicago; participant characteristics are presented in Table 1. Themes emerged inductively: the resulting grounded theoretical model (Figure 1) integrated four main themes and 13 subthemes that were saturated and inter-related; each theme is labeled as a “path” on the figure and in the section below to illustrate the interconnection of concepts. Illustrative quotes from interviews are presented in Table 2. Table 1. Participants’ Characteristics (n = 20). Characteristics n (%) Mean age 46 years Age group  18–44 8 (40)  45–54 8 (40)  55 and older 4 (20) Gender  Female 14 (70)  Male 6 (30) Sexual orientation  Straight/heterosexual 16 (80)  Gay or Lesbian 2 (10)  Bisexual 2 (10) Marital status  Never married 9 (45)  Married 5 (25)  Divorced/separated 6 (30) Education  High school degree 2 (10)  Some college 7 (35)  College degree or higher 11 (55) Employment  Unemployed 10 (50)  Full time 6 (30)  Part-time 3 (15)  Retired 1 (5) Table 2. Emergent Themes and Subthemes, Corresponding Paths in the Grounded Theoretical Model, and Illustrative Participant Quotes. Subthemes Path Illustrative quotes Theme 1 (A): Presence of COVID-19 in social network  The presence of pre-existing health conditions influences levels of perceived personal risk. A “I was somewhat concerned because of how bad, because they said that for people with underlying health issues, and I, like I said, I’m diabetic, so I was really like, oh my god, you know, I’ve gotta be overly vigilant, I’ve gotta wash my hands all the time, I’ve gotta cover my face all the time. So, that was my biggest fear I guess I would say.”—Participant 008 “I’ll say specifically [my perception of risk] changed because I have a underlying health condition that affects my immune system. So I did not come outside for anything for like six months, I’m sure it was six months. And we would joke and say, if they really ever want to find anybody that really followed the, um, the shutdown order truly, they should just talk to us. Because we. . . Um, the, the ailment is hereditary, so my mother has it and my son has asthma. So the three of us were out of commission.”—Participant 010 “Well, um, I was pretty much just, um, a shut-in because I have all the health issues and so it wasn’t, you know, good for me to be out in the general public.”—Participant 011 Theme 2 (B): Pre-existing health conditions  The presence of COVID in participant social networks influenced levels of perceived personal risk. B “Because my, um, my uncle, um, he’s a healthy guy. He ha- he came down with it, and, um, he was in the hospital. And, um, they said he almost died, and. . . He told me this. I talked to him. . .When I finally was able to speak to him when he in the hospital and he said he was doing better, but he was telling me how he, um. . . uh, he was dying-. . . and that he was, was. . . He thought he was gonna die. You know, he felt that bad, you know, hooked up to machines and all that, you know. So, um, I. . . From that point, I knew it was severe, you know, it was pretty serious. But he. . . God. . . By the grace of God, he overcame it, and he’s, um, on his feet now.”—Participant 012 “I think when I found out my sister had it was when I really took it the most serious. So, um, it just really made me take it a little more seriously and, um, because at that point like I said, I didn’t think it was going to go this far. And then right soon after that, that’s when people here at the medical center started really having a lot of deaths. And so at that point I realized that, um, I’ve gotta start really taking it seriously, um, started ordering masks ‘cause before I just had like a couple masks that I would interchange and I’m like, “No. We need to, uh, order masks off Amazon.”—Participant 017 “Um, in the beginning, I didn’t take it seriously. Uh, I said, “Okay. You know, this is a hoax,” or something like that. But then, as it went along, then, um, people who, who I knew and loved and they passed away from it, it changed me a little bit. So, it, it woke me up a lot. It woke me up a lot.”—Participant 015 Theme 3 (C–I): Information, leadership, and precautionary behaviors  There was a bidirectional relationship between perceived personal risk and precautionary COVID behaviors. C Oh, I was terrified that I would get it. It was something that was very scary. I was- you had to be, you know, extra careful. You couldn’t really interact with anyone. You had to be, um, constantly, you know, washing your hands and since I was also recovering from surgery [. . .] it was very scary for me.”—Participant 011 “And I was always getting sick, you know, for a week. [. . .] When I’d get a cold, I’d be bedridden all . . . you know, really severe . . . So, um, I just thought that if, um, that if I did [. . .] come down with the virus, it wouldn’t last long. Um, but then I started to hear about people dying and all that, so I just really tried to make sure that, um, I was doing the things they said: mask up, social distancing, not going anywhere, staying away from people as long as I could, you know, that kind of thing.”—Participant 012  The perceived lack of unified messaging and information from government, media, and/or healthcare professionals increased their perceived risk of COVID and overall fears and concerns. D “Uh, with, you know. Also, the masks, no masks, you know, no masks, no service. How contagious COVID is, how not. You know, is it airborne? You’re washing your hands. And so, uh, you know, me being afraid for my grandson, because I’m high risk, he’s someone high risk, he suffers from bronchitis. I’m just tying it together for like, uh, the information.”—Participant 013 “Um, because I was like, “No one seems to know what’s going on. They got us walking around in mask. Don’t touch this. Don’t be amongst people.” It’s like, “I don’t know what’s happening and everything.” So, I was concerned.”—Participant 021  Mistrust was related to the perceived lack of unified or consistent messaging about COVID from the government, media, and/or healthcare professionals. E “Having that clinical information, um, even when [Fauci] did, he wasn’t that sure. I didn’t think he should’ve done that. You know, he’s too inconsistent, he has to be more direct and he needed to take more of an aggressive approach. [. . .] You have to protect the general public.”—Participant 013 Um, in the beginning, I didn’t take it seriously. Uh, I said, “Okay. You know, this is a hoax,” or something like that. . . . And then, all of a sudden, when they start going on, they said, um, it’s mostly to affect the older, you know, the older people. And that was nothing. Okay. I can understand that. Then, they turned around and say, “Well, you have to take certain things [. . .] or certain pills or certain things you couldn’t take.” Okay, that was another thing. And then, um, nobody wasn’t too sure about this vaccination, one shot, two shots. And, oh, then, all of a sudden, they said, okay, [. . .] the research time that was put into this is too short. [. . .] But as it went along, and you heard about it, you’re just saying, “Okay, no.” But the thing that shook me up was how many stories that came out of it. And what do you believe and what do you not believe?”—Participant 015 “[What] scares me about it is they like, you gotta get one dose, and then, you get full immunity with another dose and stuff. I’m like, “Oh, my god, it’s too complicated.” —Participant 018  Exposure to perceived misinformation was associated with mistrust in the government, the media, and/or healthcare professionals. F “Um, I would say every time somebody died, calling it a COVID death. Like, I don’t believe that every single person that died during this pandemic died from COVID. Like, they could have died for something else, but they’re going to like, oh, like let’s push the agenda, let’s say this many people die from COVID. I don’t think anybody fakes a death or anything like that, I’m pretty sure all these people did die, but I don’t believe that every single death that they reported is from COVID. I think it’s being put out there to scare people like, yes, let’s get this vaccine, let’s do this.”—Participant 001 “Hmm. Um, to be honest, I was actually at work the day that they announced that there’s a pandemic. I was really scared I didn’t know what really caused it. You know, a lot of people have different theories, um, the government caused it you know, like some people feel like the government literally, I don’t know, infected people with COVID and spread the virus themselves. You know, due to like them pushing COVID on the news and the media and stuff. Meanwhile, there’s a whole bunch of other stuff going on that their not covering. Um, personally, I just I didn’t really know. I just knew that it was a really big issue and the president that we had at the time wasn’t gonna do much. I already knew that. I anticipated that. And so, that made my fear even worse.”—Participant 007 “Um, I probably wouldn’t trust the hospitals. Um, most people that did contract the virus go to the hospital but didn’t come out. We didn’t know what was going on. But, uh, if you just stay clear and if your religious beliefs is there and, uh, you got the shot, let’s say, at a Walgreens or something like that. Or a clinic, urgent care center, um, I think your chances of, your chances of living would be greater than going to the hospital, getting sick by the COVID-19 and surviving that. So I’m not saying that the hospitals was killing people that had the COVID-19 and just writing it off as them having it . . . but something wasn’t there that, uh, didn’t click for me.”—Participant 018  For some, exposure to perceived COVID misinformation influenced precautionary COVID behaviors. G “Yeah, I just be listening to everybody, uh, conversations [. . .], saying stuff. I’m just like, “Oh, really? It’s the vaccine? [. . .] It killed a thousand more people? Oh, okay, okay.” I just be listening. Uh, I wanna be around the third batch of people that get it. I watch the news, so I’ll see if people start walking backwards, having seizures, falling out. It ain’t for me.”—Participant 005 “I mean maybe I’ll change my mind by May cause, um, in May we’re projected to have a vaccine for every adult in the United States. Maybe I’ll change my mind by then, but for right now, I don’t, if it was offered to me today I wouldn’t want it. Well, it’s not like it’s the COVID vaccine specifically. Like, a lot of the vaccines something could happen to you. Like, it might have not been real, but I saw a picture of some people that got a vaccine and like half of their face got paralyzed and that was enough to make me not really want to take part in it.”—Participant 007  Exposure to perceived accurate information influenced personal COVID beliefs and knowledge. H “And then of course now with this one [vaccine] being created so quickly, that was just unbelievable that it was gonna work or that it was safe because we don’t know the long-term effects of it, but with all the scientists and people that are in it, and then they were getting vaccinated on TV from the president, vice president elect at the time, um, that gave me confidence that it will be okay.”—Participant 003 “Uh, I felt 100% comfortable by trying the vaccine. Um, I trust science, I trust my doctor, I trust Dr. Fauci and, um, I, uh, get a flu shot, uh, almost every year. . . Um, I used to be a Dean of Students, uh, at a law school. And I worked in academia for 19 years and I know the flu shot works. Like I would get a flu shot, you know, every year. So when they said, “Oh, we’re getting a vaccine.” You know, the vaccine was coming, I was ready, you know, and I, I was 100%, um, you know, uh, planning on getting the vaccine as soon as I could.”—Participant 014  Exposure to perceived misinformation did not influence personal COVID beliefs and knowledge in so much as most participants acknowledged the existence and prevalence of misinformation but shared that they did not ascribe to such beliefs themselves. H “Cause sometimes you’d hear people say, “Oh, um, the flu has killed more than, uh, the COVID,” and uh, and so I always. . . my response to that would be [. . .] first of all, you have to give this particular virus the chance to see what it’s gonna do, you know. So it’s easier to say, yeah, the flu has killed more people, but the flu’s been around for however long it has been. This is just something that just took place within a year and counting now, so. A lot of people, at least people I’ve spoke to, you know, have their own opinions of what may be and is it for real or not, but I wasn’t one. I believed what the experts were saying, and it’s the same to this day. But, you know, I never looked at it in the aspect of, of I think governments are trying to control our population or whatnot. I never looked at it like that, and quite frankly, when it comes to serious situations as such, I never did look at it like that, or um, formulated thought erratically.”—Participant 002 “I didn’t trust any information that I was getting off the internet or Facebook or any of those places. So I basically relied on what the mayor, governor, and the Illinois public health professional, and some of the stuff from Dr. Fauci and the CDC, basically whatever was on channel 7 or CNN, cause I trust those networks. I relied on that only and nothing off the internet. I didn’t have access to my doctor, so I didn’t get it from them. So basically the news. Well, I had a lot of, uh, things that were flashing up on Facebook and some of the stuff, I just, it was both mace, mainly conspiracy type stuff. And, um, you know, a lot of it was racial slanted against Asian people, which I don’t believe in that. And, um, it was basically blaming them for the reason that the virus was here and, um, some of the information just wasn’t factual. So I didn’t believe in that.”—Participant 003 “So that you’re not a part of the half a million people that, that had to die. Because, um, political rhetoric from leaders that I don’t support made people think that they were not in harm’s way when they were or made people think that were safe when they weren’t. Um, I have a couple of friends who I think highly of otherwise who said, ”Nah, I don’t think I’m gonna get that.” And I’m thinking, “Did you hear about polio? I mean, go get the vaccination.” This doesn’t have to happen. But I’m learning also to take care of my family in my house and, um, be not judgmental, but you know, not be so brash about these kinds of decisions because [. . .] it’s a new decision, you know, and people gotta make it for themselves I guess. So I don’t know that they’re going to just target you when everybody in the whole nation’s gotta get vaccination, right. Um, it’s not a small independent study or let’s see of syphilis really kills black men or people. Yeah. It’s not that right now. That’s not this. Um, you always have to be vigilant but it’s a pandemic. Pan P-A-N. That’s going all around the whole world.”—Participant 010  COVID beliefs and knowledge led to uptake in precautionary COVID behaviors. I “Uh, I believe it’s, um, I believe it’s okay to take because there really is no alternative and I feel like it’s something that, uh, everyone should do in order to not only protect themselves, but to protect others. There’s, uh, no other way to, uh, protect yourself. Or other people, um, spreading or continuing to spread the coronavirus unless you get, um, uh, vaccinated. Well, I hope it’s effective, but I think the reports that are coming out say something about, um, 90%, things like that. To me, it means that it’s not perfect, but it’s well enough. Uh, I think it’s safe. I have taken it, so.”—Participant 011 “[The vaccine] is safe. I haven’t had any side effects and I had it for a couple of months now. [. . .] And, uh, I recommend it. Very effective. [. . .] I think you, you’re due for a second time around just to refresh some things up in your system I believe in about a year or so. I’ll go ahead and take that again but, um, yeah, it’s effective, it’s safe, no side effects.”—Participant 018 Theme 4 (J–M): Societal burden, racism, and barriers effecting perceived COVID burden in the Black community  Institutional racism created structural barriers to access to COVID-related services, products, healthcare, and so on, which in turn presented as a perceived higher COVID burden in the Black community. J “[COVID] is worse [for the Black community], because, uh, I see a lot more deaths reported in the Black community and um, I see a lot more people in the Black community without jobs. And then the Black community, they’re less likely to be vaccinated. Some of it is cause they don’t want to [. . .] And then, um, [the vaccine] is just not widely available in the Black community. Like, you see on the news, a lot of white people have access to it or it seems like it’s easier for them to get it. And I don’t wanna sound like it’s a race war or something, but that’s just how I see it.”—Participant 003 “Oh, man. Well, obviously, our numbers were up compared to other groups as far as people that was getting sick with the virus. And, of course, I think, with anything, whether it’s housing, healthcare, whatever, we become second class when it, you know, when it comes to looking out for us. Yeah, our community always is second, comes second or third behind the majority. [. . .] people of color come second when it comes to health, hospitalization, healthcare, everything.”—Participant 012 “I question, you know, how many Black folks were able to, you know, maybe get a test. You know, was the testing close enough, you know, to their home right? Now, the vaccine is available in more neighborhoods. But honestly when it first came out at the United Center, okay, well, the United Center is not around the corner from everyone. Yeah, it’s around the corner from those people that live on the west side. But what about people that live on the far south side? Right. They put it there first and then it took weeks, you know, for them to get it over here to, you know, Chicago state right? Which is the whole, you know, south side community, right? So where were those people supposed to go to, you know, get the test?”—Participant 014 “One of the ways I think it is because of our history with being oppressed and being misused or whatever, or being experimented on, that definitely put in some extra stress with this happening, especially with the vaccine. I definitely get why we were so afraid or opposed or whatever to get it because of our history because of the way we have been treated.”—Participant 017 “Because people didn’t believe in the pandemic because it, it was hard for people to believe in going to get [. . .] the vaccine and, due to our past history [with the Tuskegee experiment]. So, and things was harder to get to for us. So. . . The vaccine and getting tested was harder to, we already have our bus routes cut and, you know [. . .] in our neighborhood, hospitals and everything is starting to close down. So you have to just spend, people don’t have cars. Some people don’t have, you know, have various ways of getting to it. So, yeah, it’s harder.”—Participant 022  Institutional barriers exposed by COVID had non-health-related consequences for the Black community. Non-health-related consequences contributed to higher perceived COVID burden in the Black community. K “You know, a lot of our, a lot of black children are falling behind because of this pandemic in their education. They, some of the kids that, when the schools shut down and they really haven’t opened back up yet, some children didn’t have access to the internet, so they weren’t able to take their classes at home and, you know, [. . .] or they didn’t have a computer to use to be able to take their classes. So, you know, they missed out on a lot because of that alone. And even though some of them were given computers and some of them were given access to free internet or low-cost internet, it still, you know, the parents weren’t there to really, you know, help them along, you know, to keep them on track. So, they are falling behind. [. . .] and I think that happened with a lot of people, that the kids were not able to keep up for whatever reason, reason it was because they didn’t have the internet or didn’t have a computer of didn’t have anyone to check what they were doing, all of that. So, it’s like, a lot of kids are gonna be, are getting left behind right now.”—Participant 008 “So some, you know, there are quite a few Black folks that don’t travel outside of their neighborhoods, don’t have cars, you know, [. . .] and don’t want to get on two or three buses to get a test, to get the vaccine, you know. [. . .] Um, you know, but my best guess is that some Black people lost their job. [. . .] and, and that’s huge. And then obviously that would affect you being able to pay your bills and pay your rent, and you know, all this other you know, good stuff or whatever. Now you’re facing eviction and you know, how does that work?”—Participant 014  Non-health-related consequences contributed to higher perceived COVID burden in the Black community. Worse, because, uh, I see a lot more deaths reported in the black community and um, I see a lot more people in the black community without jobs. And then the black community, they’re less likely to be vaccinated. Some of it is cause they don’t want to because they’re not (silence). And then, um, it’s just not widely available in the black community. Like, you see on the news, a lot of white people have access to it or it seems like it’s easier for them to get it. And I don’t wanna sound like it’s a race war or something, but that’s just how I see it. – Participant 003  Perceived existing health and social disparities in the African American community were related to perceived higher COVID burden in the Black community. L “Oh, [Black people] have been. . . disproportionately [affected by COVID]. Yeah, yeah, yeah, they’ve been affected heavily. Because lot of people lost their jobs. If you’re low income, or not low income but you have, you know, you get laid off and, you know, you work in a restaurant or retail, those businesses have closed. I mean, not closed, but temporary. Yes. If you’re in the bottom of the economic realm, you know. . .”—Participant 006 “And especially, um, uh, seemed to really hard the African American community. [The] Black community has, uh, really bared most of the brunt of the disease because of the, uh, inherent underlying health problems that we have, meaning the, um, uh, blood pressure, high cholesterol, diabetes, heart disease. And, um, a lot of us don’t really go to the doctor on a regular basis as we should. Some of it is fear. Others, for others it’s a sensibility. You know, for various reasons, it has hit our race a lot harder than other races. I think economically as well ‘cause a lot of Black neighborhoods really suffer poverty or being the working-class poor. And because of the pandemic, a lot of people who work in a lot of those fields are Black, you know, Black people. Like food services and, you know, like restaurants, the travel industry.” –– Participant 011 Note. CDC = Centers for Disease Control and Prevention; CNN = The Cable News Network. Figure 1. Grounded Theoretical Model for Factors Influencing Perceived COVID-19 Personal Risk and Precautionary COVID-19 Behaviors. The letters A-L denote the paths between concepts which are described in the results. Note. Red line indicates that we did not find evidence that participants’ perceived COVID-19 personal risk was influenced by the perceived higher COVID-19 burden in the Black community. Theme 1: Pre-Existing Health Conditions (Path A) Pre-existing health conditions appeared to be positively related to perceived personal risk, such that participants with pre-existing health conditions frequently associated their health status with a perceived increase in personal COVID-19 risk. This theme did not have any associated subthemes. Theme 2: Presence of COVID-19 Infection in Participant Social Network (Path B) For many participants, the presence of COVID-19 infections in their social networks led to an increase in their own perceived personal risk. The pandemic did not appear to be a perceived threat until individuals known to the participant were infected and/or passed away from the virus. This theme did not have any associated subthemes. Theme 3: COVID-19-Related Information, Participant Trust and Perceived Personal Risk (Paths C–I) Exposure to COVID-19-related information, perceptions of a lack of uniformity and consistency in the presentation of COVID-19 information, and mistrust of public health professionals and establishments, government entities, media and news sources, and medical professionals were all related to the uptake of precautionary COVID-19-related behaviors and perceived personal risk. This theme had several interrelated subthemes. A bidirectional relationship existed between perceived personal risk and the uptake of precautionary COVID-19-related behaviors including masking, social distancing, vaccination, hand washing, sanitizer use, and isolation/quarantining (Path C). Despite most of the sample engaging in precautionary behaviors, many participants felt there was a lack of unity and clarity in the messaging around COVID-19 from the government, news media, and medical/public health professionals. For many participants, this confusion directly influenced their perceived personal risk (Path D). This lack of unified messaging was bidirectionally related to feelings of mistrust in public health establishments, the government, news media, and medical professionals to varying degrees (Path E). Relatedly, exposure to both accurate and inaccurate information was also bidirectionally related to mistrust in public health establishments, the government, news media, and medical professionals (Path F). Many participants acknowledged their awareness of perceived accurate versus inaccurate information and chose to distance themselves from belief in anything they perceived as misinformation. Exposure to information directly influenced the uptake of precautionary behaviors for most participants (Path G). Exposure to both accurate and inaccurate information influenced participant COVID-19 beliefs and knowledge in various ways (Path H). For example, some participants reported exposure to information via social media or individuals in their social network and acknowledged that they were not confident in the accuracy of the information. While some participants acknowledged the proliferation of conspiracy theories around the pandemic, some expressly distanced themselves from these theories and others stated that they were unsure what was true or whom to trust. Some participants stated that they intentionally sought out information from places they deemed reliable including the CDC and government officials. COVID-19 beliefs and knowledge, expectedly influenced participants’ uptake of precautionary behaviors (Path I). Theme 4: Perceived Higher Burden of COVID-19 in the Black Community (Paths J–M) Most participants held the belief that there was a perceived higher COVID-19 burden within the Black community, overall. This burden was influenced by perceived disparities in the Black community compared with other racial/ethnic communities, in terms of infection rates, hospitalizations, and deaths as well as institutional racism which created structural barriers to access treatment and COVID-19-related socioeconomic resources. However, most participants did not demonstrate a link between their perceptions of higher COVID-19 burden in the Black community and their own individual perceived personal risk. Within this broader theme, there are several subthemes. Most participants tied institutional racism against the Black community to structural barriers to accessing COVID-related services, supplies, and resources, which lead to negative non-health-related consequences for the Black community including job loss and financial strain, poor schooling for children, and challenges with transportation (Path J). These COVID-related consequences contributed to greater perceived COVID-related burden within the Black community overall, especially when compared with other racial/ethnic groups (Path K). In addition to barriers to accessing services and associated institutional racism, participants identified perceived existing disparities between the Black community and other groups specifically with regards to knowledge and understanding of COVID-19 as well as epidemiological rates in the Black community (Path L). Discussion In this study, we used a grounded theory approach to understand how perceptions of personal COVID-19 risk were related to precautionary behaviors. Our grounded theoretical framework identified four main themes and 13 subthemes. The four main themes focused on pre-existing health conditions (Theme 1), the effects of COVID-19 on social networks (Theme 2), exposure to COVID-related information, the effects of perceived inconsistent messaging from public health and government officials, and mistrust of public health establishments, government, media, and/or medical professionals (Theme 3) and finally, the perceived effects of racial disparities, structural and systemic racism, and perceived COVID-19 burden on the Black community’s experiences of the pandemic (Theme 4). The disproportionately high COVID-19 burden in Black communities did not seem to directly influence personal perceptions of risk in our sample. Participants did not state that their perceptions of a higher COVID-19 burden in the Black community influenced their own perceived personal risk related to COVID-19, often referring to the Black community in othering language such as “they” or “the Black community” and not with inclusive language such as “we.” Rather, infections and deaths in participants’ social network and having a pre-existing health condition more effectively shaped their risk perceptions and subsequent adherence to precautionary behaviors (e.g., mask wearing and vaccine uptake). This observation is consistent with a recent survey showing that U.S. adults make vaccination-related decisions based on individual perceptions of risk rather than “population threat of infection” (Mercadante & Law, 2021). In turn, individual risk perceptions tend to originate from personal experiences; findings from a recent study emphasized that awareness of COVID-19 cases among one’s immediate family members was associated with engagement in precautionary behaviors (Li et al., 2020). Relatedly, an analysis conducted after the 2009/2010 H1N1 pandemic revealed that health-related communications within social networks were positively correlated with the adoption of precautionary behaviors (Lin et al., 2018). Taken together, these findings suggest that a generalized COVID-19 risk communication may not achieve its intended purpose. Instead, localized messaging that focuses on specific social environments may be better suited to convey the magnitude of risk. The historical and contemporary contexts of systemic racism and subsequent mistrust in public health institutions were related to personal risk perceptions of COVID-19. This mistrust has potentially deepened because of inconsistent health messaging, direct attacks on public health institutions by prominent politicians, and the viral spread of misinformation on social media. It is therefore unsurprising that the American public’s confidence in medical science has declined since the beginning of the pandemic, according to a recent Pew survey (Kennedy et al., 2022). Among Black individuals, 28% said they have “a great deal of confidence in medical scientists to act in the public’s best interests” in December 2021, down from 33% in November 2020. These findings align with evidence that Black Americans’ beliefs about COVID-19 were shaped by a perception of implicit bias within health care, long-standing racists practices within the scientific and medical field, conflicting guidance at different levels of government, and a paucity of trusted messengers (Bateman et al., 2021; Carson et al., 2021; Momplaisir et al., 2021; Ordaz-Johnson et al., 2020). While our findings did not unequivocally link mistrust to less uptake of precautionary behaviors, other studies demonstrated a relationship between mistrust and willingness to take the COVID-19 vaccine (Momplaisir et al., 2021). Trust in government and government-led responses are crucial in a crisis situation, wherein adherence to official guidance is key to stemming the spread of a novel virus. In the long-term, authentic reflection and deliberate action around the fundamental causes of mistrust among Black communities are prerequisites to an effective emergency response (Dada et al., 2022). More immediate actions include identifying and empowering trusted entities and messengers within local communities to disseminate scientific evidence and guidance; these may include Black physicians, community organizations, faith-based leaders, and barbershops (Berenbrok et al., 2021; Dada et al., 2022). In addition to mistrust, some participants noted structural barriers to accessing COVID-19 testing and vaccination resources, which may hinder uptake of precautionary behaviors despite accurate knowledge and motivation. Other studies have also found that the location of testing and vaccination sites and associated financial costs presented barriers to access among Black individuals (Bateman et al., 2021; Callaghan et al., 2021). More structurally, COVID-19 vaccine “deserts” have been identified in areas where Black residents live far from medical health centers and have to drive further distances to reach vaccination facilities compared to Whites (Dada et al., 2022; PittWire, 2021). These findings emphasize the importance of understanding the local environmental context and combining health communications strategies with access-enhancing interventions. Non-traditional vaccination strategies have been deployed as alternatives to serve hard-to-reach populations, such as converting parking lots, community centers, businesses, faith-based organizations, and schools or using pop-up or mobile sites (e.g., vans and ambulances) (Dada et al., 2022). The ability to quickly adapt to emergent needs requires a sustained commitment to establish relationships between localities and community organizations and groups to collaboratively identify novel approaches to combat this and future pandemics. Strengths and Limitations We used grounded theory methodology to propose multiple pathways that can influence risk perceptions and precautionary behaviors among Black individuals during the unprecedented COVID-19 pandemic context. We interviewed participants from different age groups, genders, income levels, and community engagement to ensure a diversity of perspectives and lived experiences. Our study has several limitations. First, participation in our study required access to the internet and the ability to connect to Zoom, which may have excluded individuals with digital access and literacy barriers. The study team provided specific instructions to all participants on how to connect to the Zoom platform and made every effort to offer troubleshooting support when possible while also allowing for phone interviews. Second, social desirability bias may have affected participants’ accounts. For example, although participants evoked COVID-19 misinformation unprompted, they frequently distanced themselves from this misinformation by assuring the interviewers that they did not buy into this narrative. This pattern may hint at possible social desirability bias, given that participants recognized that the interviewers worked in public health research. Third, our sample has overall higher educational attainment than the average Black Chicagoan; this could explain the relatively attenuated reports of challenges experienced during the COVID-19 pandemic. Public Health Implication: Relevance to Public Health Emergency Response Our findings suggest that Black adults’ perceptions of their individual risk and precautionary behaviors were not always congruent with public health data and recommendations. Therefore, COVID-19 messaging and mitigation strategies should be informed by (a) local community engagement and (b) transparent communication. First, a one-size-fits-all approach to public health interventions failed to reach disproportionally affected and marginalized communities. Community engagement and community-driven interventions and public health campaigns should be prioritized to understand local contexts and needs, and jointly develop with community partners–targeted interventions that are acceptable and feasible. This strategy aligns with recommendations for COVID-19 vaccine equity generated through focus groups with participants of diverse racial/ethnic backgrounds, including Black Americans. Participants recommended investing in community engagement and community-centered actions that leverage trusted messengers from within each community (Carson et al., 2021). Relatedly, national funding efforts have promoted community-engaged strategies to address COVID-19 disparities (National Institutes of Health, 2021, 2022); these projects should be the bedrock for sustained community partnerships and emergency preparedness. Second, the COVID-19 pandemic was an unprecedented public health emergency, at least in recent history, both in terms of its novelty and spread. Public health information and messaging were constantly refined to reflect emerging data and knowledge. This iterative process of building scientific evidence may seem familiar to and even expected by health and medical professionals, but it caused public disarray and growing mistrust in health agencies. The public had an extraordinary exposure to the scientific method, which should have been coupled with broad education about how science typically unfolds (e.g., why recommendations change over time). Transparent communication about the scientific process and rationale for decision-making should be prioritized in subsequent health emergencies. Coupling these two recommendations for local action and transparent communication may generate the optimal environment for fostering and maintaining trust between communities and the medical and public health entities. The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by a pilot grant received from the University of Illinois Chicago’s Policy & Social Engagement Fellowship through the Institute for Research on Race and Public Policy. ORCID iDs: Perla Chebli https://orcid.org/0000-0002-0344-442X Yamilé Molina https://orcid.org/0000-0001-8742-5658 Uchechi A. Mitchell https://orcid.org/0000-0003-1675-1119 ==== Refs References Abrams E. M. Szefler S. J. (2020). COVID-19 and the impact of social determinants of health. The Lancet Respiratory Medicine, 8 (7 ), 659–661. https://10.1016/S2213-2600(20)30234-4 32437646 Bateman L. B. Schoenberger Y. M. Hansen B. Osborne T. N. Okoro G. C. Speights K. M. Fouad M. N. (2021). Confronting COVID-19 in under-resourced, African American neighborhoods: A qualitative study examining community member and stakeholders’ perceptions. Ethnicity & Health, 26 (1 ), 49–67. https://10.1080/13557858.2021.1873250 33472411 Becker M. H. (1974). The health belief model and sick role behavior. Health Education Monographs, 2 (4 ), 409–419. https://10.1177/109019817400200407 Berenbrok L. Tang S. Coley K. Boccuti C. Guo J. Essien U. Dickson S. Hernandez I. (2021). Access to potential COVID-19 vaccine administration facilities: A geographic information systems analysis. https://s8637.pcdn.co/wp-content/uploads/2021/02/Access-to-Potential-COVID-19-Vaccine-Administration-Facilities-2-2-2021.pdf Bogart L. Ojikutu B. Tyagi K. Klein D. Mutchler M. Dong L. Lawrence S. Thomas D. Kellman S. (2020). COVID-19 related medical mistrust, health impacts, and potential vaccine hesitancy among Black Americans living with HIV. Journal of Acquired Immune Deficiency Syndromes, 86 (2 ), 200–207. https://10.1097/QAI.0000000000002570 Brewer N. T. Chapman G. B. Gibbons F. X. Gerrard M. McCaul K. D. Weinstein N. D. (2007). Meta-analysis of the relationship between risk perception and health behavior. Health Psychology, 26 (2 ), 136–145. https://10.1037/0278-6133.26.2.136 17385964 Callaghan T. Moghtaderi A. Lueck J. A. Hotez P. Strych U. Dor A. Fowler E. F. Motta M. (2021). Correlates and disparities of intention to vaccinate against COVID-19. Social Science & Medicine, 272 , 113638. https://10.1016/j.socscimed.2020.113638 33414032 Carson S. L. Casillas A. Castellon-Lopez Y. Mansfield L. N. Morris D. Barron J. Ntekume E. Landovitz R. Vassar S. D. Norris K. C. Dubinett S. M. Garrison N. A. Brown A. F. (2021). COVID-19 vaccine decision-making factors in racial and ethnic minority communities in Los Angeles, California. JAMA Network Open, 4 (9 ), Article e2127582. https://10.1001/jamanetworkopen.2021.27582 Center for Systems Science and Engineering, Johns Hopkins University. (2022). COVID-19 dashboard. COVID-19 Dashboard. https://coronavirus.jhu.edu/map.html Centers for Disease Control and Prevention. (2022a). CDC museum COVID-19 timeline. CDC Museum COVID-19 Timeline. https://www.cdc.gov/museum/timeline/covid19.html#:~:text=January%2020%2C%202020%20CDC,18%20in%20Washington%20state Centers for Disease Control and Prevention. (2022b). Hospitalization and death by race/ethnicity. https://www.cdc.gov/coronavirus/2019-ncov/covid-data/investigations-discovery/hospitalization-death-by-race-ethnicity.html Centers for Disease Control and Prevention. (2022c). Risk for COVID-19 infection, hospitalization, and death by race/ethnicity. https://www.cdc.gov/coronavirus/2019-ncov/covid-data/investigations-discovery/hospitalization-death-by-race-ethnicity.html Charmaz K. (2014). Constructing grounded theory (2nd ed.). SAGE. Chen L. S. Kaphingst K. A. (2011). Risk perceptions and family history of lung cancer. Community Genetics, 14 (1 ), 26–34. https://10.1159/000294151 Conner M. Norman P. (2017). Health behaviour: Current issues and challenges. Psychology & Health, 32 (8 ), 895–906. https://10.1080/08870446.2017.1336240 28612656 Dada D. Djiometio J. N. McFadden S. M. Demeke J. Vlahov D. Wilton L. Wang M. Nelson L. E. (2022). Strategies that promote equity in COVID-19 vaccine uptake for Black communities: A review. Journal of Urban Health, 99 (1 ), 15–27. https://10.1007/s11524-021-00594-3 35018612 Fan C. Fann J. C. Yang M. Lin T. Chen H. Liu J. Yang K. (2021). Estimating global burden of COVID-19 with disability-adjusted life years and value of statistical life metrics. Journal of the Formosan Medical Association, 120 , S106–S117. https://10.1016/j.jfma.2021.05.019 Kennedy B. Tyson A. Funk C. (2022). Americans’ trust in scientists, other groups declines. Pew Research Center. https://search.proquest.com/docview/2644315175 Kim S. J. Bostwick W. (2020). Social vulnerability and racial inequality in COVID-19 deaths in Chicago. Health Education & Behavior, 47 (4 ), 509–513. https://10.1177/1090198120929677 32436405 Li S. Feng B. Liao W. Pan W. (2020). Internet use, risk awareness, and demographic characteristics associated with engagement in preventive behaviors and testing: Cross-sectional survey on COVID-19 in the United States. Journal of Medical Internet Research, 22 (6 ), Article e19782. https://10.2196/19782 Lin L. McCloud R. F. Jung M. Viswanath K. (2018). Facing a health threat in a complex information environment. Health Education & Behavior, 45 (1 ), 77–89. https://10.1177/1090198117708011 28548547 Maroko A. R. Nash D. Pavilonis B. T. (2020). COVID-19 and inequity: A comparative spatial analysis of New York City and Chicago hot spots. Journal of Urban Health, 97 (4 ), 461–470. https://10.1007/s11524-020-00468-0 32691212 Massey D. S. Gross A. B. Shibuya K. (1994). Migration, segregation, and the geographic concentration of poverty. American Sociological Review, 59 (3 ), 425–445. https://10.2307/2095942 Mercadante A. R. Law A. V. (2021). Will they, or won’t they? Examining patients’ vaccine intention for flu and COVID-19 using the health belief model. Research in Social and Administrative Pharmacy, 17 (9 ), 1596–1605. https://10.1016/j.sapharm.2020.12.012 33431259 Millett G. A. Jones A. T. Benkeser D. Baral S. Mercer L. Beyrer C. Honermann B. Lankiewicz E. Mena L. Crowley J. S. Sherwood J. Sullivan P. S. (2020). Assessing differential impacts of COVID-19 on Black communities. Annals of Epidemiology, 47 , 37–44. https://10.1016/j.annepidem.2020.05.003 32419766 Momplaisir F. Haynes N. Nkwihoreze H. Nelson M. Werner R. M. Jemmott J. (2021). Understanding drivers of coronavirus disease 2019 vaccine hesitancy among Blacks. Clinical Infectious Diseases, 73 (10 ), 1784–1789. https://10.1093/cid/ciab102 33560346 Nardone A. Chiang J. Corburn J. (2020). Historic redlining and urban health today in U.S. cities. Environmental Justice, 13 (4 ), 19–119. https://10.1089/env.2020.0011 National Institutes of Health. (2021). Rapid Acceleration of Diagnostics (RADX). National Institutes of Health Medical Research Initiatives. https://www.nih.gov/research-training/medical-research-initiatives/radx National Institutes of Health. (2022). NIH community engagement alliance (CEAL). National Institutes of Health Community Engagement Alliance. https://covid19community.nih.gov/ Ndugga N. Artiga S. (2021). Disparities in health and health care: 5 key questions and answers. Kaiser Family Foundation. https://www.kff.org/racial-equity-and-health-policy/issue-brief/disparities-in-health-and-health-care-5-key-question-and-answers/ Ordaz-Johnson O. H. Croff R. L. Robinson L. D. Shea S. A. Bowles N. P. (2020). More than a statistic: A qualitative study of COVID-19 treatment and prevention optimization for Black Americans. Journal of General Internal Medicine: JGIM, 35 (12 ), 3750–3752. https://10.1007/s11606-020-06259-2 33021713 PittWire. (2021). Map shows where Black Americans will travel farther than Whites for COVID-19 vaccination. https://www.pitt.edu/pittwire/features-articles/map-shows-where-black-americans-will-travel-farther-whites-covid–vaccination19 Rogers R. W. (1975). A protection motivation theory of fear appeals and attitude change1. The Journal of Psychology, 91 (1 ), 93–114. https://10.1080/00223980.1975.9915803 28136248 Shepperd J. A. Findley-Klein C. Kwavnick K. D. Walker D. Perez S. (2000). Bracing for loss. Journal of Personality and Social Psychology, 78 (4 ), 620–634. https://10.1037/0022-3514.78.4.620 10794370 Slovic P. (1987). Perception of risk. Science, 236 (4799 ), 280–285. https://10.1126/science.3563507 3563507 Thornberg R. Charmaz K. (2014). Grounded theory and theoretical coding. In Flick U. (Ed.), The SAGE handbook of qualitative data analysis (pp. 153). SAGE. https://10.4135/9781446282243.n11 Williams D. R. Collins C. (2001). Racial residential segregation: A fundamental cause of racial disparities in health. Public Health Reports, 116 (5 ), 404–416. https://10.1093/phr/116.5.404 12042604

36510857

PMC9749057

NO-CC CODE

2022-12-15 23:22:49

no

Health Educ Behav. 2022 Dec 13;:10901981221139168

utf-8

Health Educ Behav

10.1177/10901981221139168

oa_other

==== Front Int Sociol Int Sociol ISS spiss International Sociology 0268-5809 1461-7242 SAGE Publications Sage UK: London, England 10.1177/02685809221137783 10.1177_02685809221137783 Article Contextualising planned behaviours to the vaccination against COVID-19 in the European Union https://orcid.org/0000-0001-8618-2252 Sandu Dumitru University of Bucharest, Romania Dumitru Sandu, Faculty of Sociology and Social Work, University of Bucharest, Bucharest 030167, Romania. Email: dumitru.sandu@unibuc.ro 13 12 2022 13 12 2022 02685809221137783© The Author(s) 2022 2022 International Sociological Association This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. The article targets the reasons that are behind behaviour orientation in the vaccination process against COVID-19. The data we are using come from the Flash Eurobarometer 494, collected in May 2021. The key dependent variable puts together vaccination intentions (soon, later on in 2021, undecided, later, never) and the fact of being vaccinated or not. A multivariate and multilevel analysis confirms the validity of an extended theory of planned behaviour in explaining the orientation to the vaccination against COVID-19. The space patterning of the behaviours is highly marked by differences among Old versus the New Member States of the European Union, clusters of countries, urban versus rural areas, and also by a function of trust in relevant institutions, and customs of using vaccination to cope with different diseases as an adult. New questions and hypotheses are generated by multiple comparisons. Résumé L’article s’intéresse aux raisons qui sont à l’origine de l’orientation du comportement dans le processus de vaccination contre le Covid-19. Les données utilisées proviennent de l’enquête Flash Eurobaromètre 494 réalisée en mai 2021. La principale variable dépendante regroupe les intentions de vaccination (bientôt, plus tard en 2021, indécision, plus tard, jamais) et le fait d’être vacciné ou non. Une analyse multiniveau multivariée confirme la validité d’une théorie répandue du comportement planifié pour expliquer l’orientation concernant la vaccination contre le Covid-19. La configuration spatiale des comportements est fortement marquée par les différences entre les anciens et les nouveaux États membres de l’Union européenne, entre des groupes de pays, entre les zones urbaines et rurales, ainsi que par un facteur de confiance dans les institutions concernées et par les habitudes de recours à la vaccination pour lutter contre différentes maladies à l’âge adulte. De nouvelles questions et hypothèses se dégagent des comparaisons multiples. Resumen El artículo aborda las razones que están detrás de la orientación de la conducta en el proceso de vacunación contra la COVID-19. Los datos que se utilizan proceden del Eurobarómetro Flash 494, recogidos en mayo de 2021. La variable dependiente clave recoge conjuntamente la intención de vacunarse (pronto, más tardíamente en 2021, indecisión, más tardíamente, nunca) y el hecho de estar vacunado o no. Un análisis multinivel multivariante confirma la validez de una extendida teoría del comportamiento planificado para explicar la orientación hacia la vacunación contra la COVID-19. El patrón espacial de comportamiento está muy marcado por las diferencias entre los antiguos y los nuevos Estados miembros de la Unión Europea, entre grupos de países, entre áreas urbanas y rurales, y también por el efecto de la confianza en instituciones relevantes y el hábito de usar la vacunación para hacer frente a diferentes enfermedades durante la edad adulta. Nuevas preguntas e hipótesis de investigación han sido generadas a través de las comparaciones múltiples. Comparative analysis COVID-19 multilevel analysis theory of planned behaviour vaccination Mots-clés analyse comparative analyse multiniveau Covid-19 théorie du comportement planifié vaccination edited-statecorrected-proof typesetterts1 ==== Body pmcIntroduction Why have some people been vaccinated against COVID-19 (C19) or planned the action soon, sometime during the year, later, or never? Is there a kind of motivational continuum between very soon and never? Do motivations differ from place to place? And if so, how much and for what reasons? Questions of this kind would deserve to be assumed because only by comparing the answers given in different contexts we could, very likely, advance in understanding the theme. The sociocultural conditions in the C19 vaccination have been addressed through case studies at the national level (Cerda and García 2021) or by comparing the situations between neighbouring or similar countries by their cultural profile (Trent et al., 2021; Wollast et al., 2021). We will extend such an approach to 23 countries in the European Union (EU), using the data of a Flash Eurobarometer survey, from the perspective of planned behaviour theory (Wolff, 2021; Wollast et al., 2021). In line with the standard approach of the theory of planned behaviour (TPB), vaccination practices are well approximated by intentions, and these derive from attitude, pressure from the significant other and perceived control of targeted behaviour (Ajzen, 1991, 2011). In this kind of approach, the key issue is related to the right specification of the prediction model. The theoretical model became better specified in its development from the theory of reasoned action (Fishbein, 1967) which stipulated that attitudes and subjective norms could predict intentions that generate behaviours. Perceived behavioural control was added as a key predictor in the TPB. The model of TPB is, in most applications, a universalist one, in which the context is frequently blurred. In addition, the intention of action and perceived behavioural control (Ajzen, 1991) tend to become unique intermediate variables through which all other factors act on the behaviour. The need for standardisation of context factors is especially emphasised in comparative research on different spatial units (countries, regions, communities). The extended TPB (Conner and Armitage, 1998) is a new sequence in the process of a better specification of the theory to explain planned behaviours. Initially, it targeted extending TPB by adding, as predictors of intentions to act, past behaviour/habits, belief salience, moral norms, self-identity, and affective beliefs. Different versions of extended TPB developed the function of action content and its context. This analysis here supports a variant of extended TPB in the field to include the role of the context of place-time and social space in determining the behaviours of vaccination against C19. Highlighting such contexts at the level of the comparative territorial units can lead to questions and assumptions related to other factors that condition the planned behaviours. Like any contextualisation, it can also favour the substantiation of policies in the field of reference (Gaston, 2017). We expand the TPB by proposing an integrated approach by (1) measuring the dependent variable, (2) the specification of the predictors, and (3) targeting levels of analysis. The dependent variable here is the orientation of the interviewees to vaccinate or not against C19. The usual approaches of TPB are focused on the factors that directly determine the intention for a specified planned behaviour or on the impact of intention on behaviour (see, for example, Buhmann and Brønn, 2018; Conner and Armitage, 1998). Our first integration is to consider behavioural intention and past behaviour of the uptake of the anti-C19 vaccination in the same variable. This is done in two ways, by measuring the latent variable of vaccination orientation as an ordinal or as a nominal/categorical variable (Babbie, 2020). In the first case, the values of the ordinal variable could be no intention to uptake vaccination against C19 (scored 1), sometime in the future (2), undecided (3), in a reasonable and specified time (4), soon (5), or already vaccinated (6). The nominal measure considers the same values of the variable by the six numerical codes but these do not indicate an ordering of the variable values. The procedure of measuring the dependent variable allows for a modification of the research question. In the standard TPB, the questions are (1) on the factors of determination for the action intentions, and (2) on the impact of intention (and perceived control behaviour, in some cases) on the emergent action. In our integrated approach, the research questions are (1) on the specific causal profile that different predictors have on each of the values of the anti-C19 vaccination considered as a nominal variable, and (2) on the same dependent variable in its ordinal version of measurement. Different types of intentions of vaccination anti-C19 and the uptake of vaccine are simply stages in the process of vaccination. Such an approach is specially adjusted for situations when one does not have a clear time separation between structuring the intention and, later on, the act of adopting the planned behaviour as a result of previous intentions. The new approach is also associated with considering specific societies or communities as involved in an innovation diffusion process (Rogers et al., 2005). Such societies are constituted as fields of adopters on non-adopters of vaccines as a health innovation. Those that are already vaccinated are early adopters. They are followed by several waves of later adopters, undecided, and non-adopters of the anti-C19 vaccination (Balas and Chapman 2018). The fields of adopters and non-adopters of the vaccination could be more or less aggregated functions of the available data and analytical interests. The data we use in this analysis comes from the Flesh Eurobarometer 494, implemented in May 2021, in the countries of the EU. Four societies with very small subsamples (Estonia, Cyprus, Malta, and Luxembourg) were omitted from the analysis as we operate not only at the EU level but also country by country. Following the mentioned contextualisation premises, is not only the vaccination intention that counts but also the actual anti-C19 vaccination in the period before the survey moment. In other words, the dependent variable in the analysis includes past behaviour and future intentions in a single qualitative/nominal measured variable. In this way, it will be possible to determine specific influences of the different factors taken into account on some vaccination behaviours as realised or planned (no intention, some time, later in the year 2021, very soon, already vaccinated). Some errors could indeed affect the measurement because some people could declare that they did the anti-C19 vaccine but in fact, they did not. The great number of persons that answered the questionnaire could diminish the impact of such an error in the multivariate analysis that we adopted. Measurement errors in the practice of vaccination against C19 are also possible in the official statistics as an aggregation effect. The use of control variables and multilevel analysis in our case could contribute to diminishing such measurement errors. The next section introduces some survey research to contextualise the data of the Eurobarometer we are using. The following section presents methods and data, results of the analysis and conclusions. Comparative approaches in surveys on C19 vaccination International comparative surveys of vaccination behaviours differ, methodologically speaking, mainly depending on how the dependent variable is measured and how predictors are chosen, with or without a predetermined model. The measurement of the dependent variable may be with or without the inclusion of past C19 vaccination behaviours. Only vaccination intentions, only past behaviours or past behaviours and intentions can be considered in the construction of the dependent variable. Also, in the measurement model, the nominal, ordinal or quantitative estimations of the behaviours or intentions of the anti-C19 vaccination can be taken into account. In the series of predetermined theoretical models used to choose predictors, the most common examples are the TPBs or the Health Beliefs Model. Another criterion in structuring the research methodology on the topic was the use of comparative versus non-comparative approaches. Before the main variants of vaccination intention were identified, surveys investigated attitudes towards a possible vaccine. The first, non-comparative survey was conducted online in Australia, in March 2020, on a representative sample of 1420 Australian adults. The survey tested the attitude of the people towards getting vaccinated, by asking for agreement with the statement that ‘getting myself vaccinated for C19 would be a good way to protect me against infection’ (Seale et al., 2021), on a Likert-type 5-point scale. Significant predictors of this attitude were identified by logistic regression after recording the attitude measure into a dummy variable. Caeteris paribus, the highest probabilities for engaging in the anti-C19 vaccination were for females, 70 or more years old people, those having chronic diseases and private health insurance. The same study mentioned also the share of people intending to be vaccinated in the first wave of pandemic in other countries but without doing systematic comparisons by subsamples, referring to survey results in Denmark, France, Germany, Indonesia, Italy, Portugal, the Netherlands, and the United Kingdom. A comparative survey (Trent et al., 2021) was also conducted online, between July and September 2020, in five major cities in Australia (Sydney, Melbourne), the United States (New York, Phoenix), and the United Kingdom (London). The prediction of the intention to vaccinate, as a dummy variable, was determined by variables related to trust in government/government information about C19, experiences of infection with C19 in personal communities of family or friends, and the perception of risks of infection with the New covid. In the series of socio-demographic predictors, variables related to age categories, gender, income, health status, smoking, and vaccination against influenza are included. Age is the only predictor with a significant impact on the will to vaccinate against the New COVID for each of the five cities in the sense that young people have a lower vaccination propensity than the elderly. Vaccination against influenza in the last 12 months significantly increases the likelihood of vaccination against C19 in four of the five cities. The exception of the lack of impact appears only for the London sample. It is not entirely clear whether the exception is given by reality or by the characteristics of sampling. Another model, close to the previous one, is the one in which the intention to vaccinate anti-C19 is measured not dichotomously but by a continuous variable and predictors are chosen according to predetermined theoretical models. The analysis is based on a survey conducted in five Anglophone countries, namely Australia, the United States, Great Britain, New Zealand, and Canada (Burke et al., 2021). The dependent variable is a factor score that multiplicatively aggregates three indicators related to vaccination intent for protection from C19. The choice of predictors was based on the Health Behaviour Model and the TPB. With the survey conducted before having government C19 vaccination decisions, the questionnaire applied did not request information about behaviours to protect against C19 through vaccination. The research argues that the probability of vaccination is higher for those who trust the information transmitted by the national government on the subject and have a high-level perception of the risks associated with non-vaccination of anti-C19 for themselves and others. The socio-demographic categories that favour a higher probability of vaccination are the elderly, the unemployed but seeking a job, and those who have already been vaccinated for other diseases in the past. Those who were unemployed but were not seeking a job had lower intentions of vaccination. It results that looking for a job stimulated the unemployed to traying also to get vaccinated against C19 as a precondition for employment. Altruistic and collectivist attitudes have also proved favourable to vaccination against C19. Variations in gender, income, religion, religiosity or the knowledge of people who have been ill with C19 do not appear to be significant predictors in the total sample. The TPB is also adopted in the structuring of surveys in which the analysis is made not on vaccination in pandemic conditions but about two behaviours to prevent contamination through C19, namely the limitation of social contacts and repeated hand washing (Wollast et al., 2021). For each of these behaviours, questions about the frequency of behaviours, the degree of structuring of intentions, social norms, attitudes, and the perception of control in carrying out health actions are included in the survey questionnaire. The same questionnaire form is applied online for countries of close culture, that is, France and Belgium. In path analysis, used for data processing, the distinction between behaviours, intentions, social norms, attitudes, and perception of control over actions is maintained. Endogenous variables measured by multiple items are the result of the summative aggregation of some Likert-type scales. Research by representative country surveys that did not use any theory to grounding it is in the case of Robertson et al. (2021) for the Republic of Ireland. The survey on 1600 adult persons, accomplished in January 2021, asked about the probability of deciding to get vaccinated against C19 – definitely yes, probably yes, probably no, definitely no. A logistic regression considered intentions to accept anti-C19 vaccination versus the hesitants as a dependent variable. The hesitants to anti-C19 vaccination were mainly younger women with lower education, having children, and those in the category of ethnic minorities. This was valid in the model running only with demographics as predictors. Another model adding a knowledge test on anti-C19 vaccination and perceived severity of the infection was more efficient to explain the variation of the dependent variable. In this new model, all the demographic predictors lost their statistical explanation (Burke et al., 2021). This is relevant to the fact that demographic variables influence vaccination/hesitancy intentions through the medium of knowledge on the vaccination and perceived severity of C19 infections. The finding could also be relevant to the fact that the knowledge test on anti-C19 vaccination is a very powerful methodological tool. The authors of the study mentioned as a key limitation the fact that the approach considered only intentions, not behaviours as dependent variables. Other possible relevant predictors such as rural-urban residency, past experiences of vaccination against other illnesses, and trust in relevant institutions were not recorded by the survey. Such predictors proved to be efficient in the approach of better-specified models of analysis as is the case with the comparative Anglophone survey presented above (Burke et al., 2021). The analytical construction we present retains basic ideas from the TPB, but it is also dependent on the survey data (Eurobarometer 494 of 2021) with which we work and, inevitably, on different contexts from country to country. It is construction-oriented, simultaneously, by the extensive TPB (Sommer, 2011) but also by the data. One of the basic features of the extensive TPB, in the 2011 Sommer version, includes past behaviour as a predictor of future intentions and behaviour. In the theoretical model that we propose in the analysis, we will resume the idea but with a different approach, to which we will refer in the next section. Past behaviour is considered from two perspectives, generic and particular. Past vaccination behaviour, regardless of the field of disease prevention, appears as a determinant of behavioural intentions. But specific past behaviour, related to the anti-C19 vaccination, is here considered as a constituent value of the nominal variable relating to the intention to vaccinate anti-C19. In the next section, we enter the survey data that we use and the analysis model. We then present the results of the analysis and, finally, the conclusions. Data and theoretical model The data we analyse comes from the Flash Eurobarometer 494, conducted in May 2021 in EU countries. We used the weighted version for 23 countries, with the elimination of very small subsamples, for Luxembourg, Malta, Estonia, and Cyprus (25,912 people of adulthood) to ensure greater stability of the calculation results at the country level. The analysis model operates with the orientation of the anti-C19 vaccination as both an ordinal and nominal variable. In the ordinal version, we considered that the minimum pro-vaccination orientation is in those who declare that they will never get vaccinated to prevent this disease (score 1) and the score for the maximum orientation is 6, granted to those who have already been vaccinated. Intermediate values are expressed in terms of later (2), I do not know (3), sometime in 2021 (4), and soon (5). In this way, between the intention to vaccinate never and I have already been vaccinated, there is a polarity. In the nominal type measurement, the six values are only different, not ordered between them. The option was supported by the fact that past behaviours of anti-C19 vaccination were recorded at the same moment with the intent of being vaccinated. Past behaviour and intentions are considered as values of the same variable measuring the accomplished or announced practice of vaccination. Both of them had specific motivations that were not recorded in a non-panel survey. This is a procedure to integrate vaccination into a time context by reference to the past, near, uncertain, and long-time future. The space integration of the approach is done here by considering institutional and physical space. The contextualisation of the vaccination against C19 is accomplished by a time-space integration of the practice. The first hypothesis (H1) that structures the theoretical model we are working with argues, in line with the planned behaviour theory, that the pro-vaccination behaviour is all the more intense as the anti-C19 pro-vaccination attitude is stronger (Figure 1). Figure 1. Theoretical model of predicting vaccination behaviours against C19. The pro-vaccination orientation index (IPVO) is constructed as a factor score (multiplied by 100), from three indicators related to the agreement of the interviewees with formulations that claim that the anti-C19 vaccination is a civic duty, has more advantages than disadvantages and that vaccination, in general, has contributed to the disappearance of many diseases (Sandu, 2021). To measure the perception of the significant other in terms of anti-C19 vaccination we used as proxies, two confidence indices. The first of these is constructed as a factor score from indicators relating to trust in the government, local authorities and health authorities as sources of information about the anti-C19 vaccination. The second is about trust in online networks used as a source of information about the anti-C19 vaccination. The second hypothesis (H2) argues that increased trust in public administration institutions favours anti-C19 vaccination and the third hypothesis (H3) supports the expectation that increased trust in information on social networks will favour behaviours of refusal or hesitation in the anti-C19 vaccination (Wang and Liu, 2021). The fourth hypothesis (H4) states an increased probability of vaccination/intention of anti-C19 vaccination for people who have been vaccinated at least once before as an adult to prevent another disease (Sandu, 2021). Otherwise formulated, the hypothesis claims that the increased likelihood of anti-C19 vaccination is higher for those who already have the personal habit of being vaccinated as adults. The hypothesis helps to a better specification of the explanatory model. TPB includes this factor as a proxy for perceived behavioural control (Ajzen, 1991). Some of its applications include the mentioned experience in the model (Burke et al., 2021; Conner and Armitage 1998). Some other prediction models omit the factor (Robertson et al., 2021). When one has to compare different countries of the EU, as in our case, the experience is a very useful predictor as it can allow for separating the country from personal experiences about past vaccination experiences. The questionnaire for the flash Eurobarometer 494 that we are using here allows for a differentiation between being or not vaccinated against other diseases as a child or as an adult. We used in this analysis only the information referring to the practice of vaccination as an adult. Finally, in line with other previous research at the European level (Sandu, 2021) focused on the attitude towards the anti-C19 vaccination, it is expected that pro-vaccination intentions and behaviours of vaccination about the same infection by C19 will have a higher probability in more developed territorial units (H5): People from societies of the Old EU tend to accept vaccination anti-C19 more than people from the societies of the New Member States of the EU (H5a); living in urban areas of European society would bring higher rates of vaccination anti-C19 than in rural areas (H5b). We present below the results of the analysis from a descriptive and explanatory point of view. In the strictly descriptive part, we present the way of grouping countries in terms of past behaviour patterns – intentions of vaccination against C19. The similarities that occur between populations in different countries allow the identification of criteria and favour similar attitudes and behaviours in the field of analysis. After identifying the frequency distributions for the six values of the main analysis variable related to the behaviours in the anti-C19 vaccination, we move on to the grouping of countries in terms of similarities between them, depending on the respective dominant behavioural patterns. Subsequently, the analysis returns to the individual level to distinguish models of prediction of vaccination behaviours/intentions. National models for reporting to C19 vaccination The 23 countries that remain under analysis after dropping those for which the samples are very small, in the survey, are far from unique in terms of anti-C19 vaccination behaviours. The first differentiation that appears very clearly is that between the Old and the New EU Member States (Figure 2). Except for Greece, the population of all the other countries of the Old EU is mostly oriented, following H5a, in favour of anti-C19 vaccination to a greater extent than that of the New Member States of the Union. Even if one computes the index of dominant orientation (see details of computation in the footnote of Figure 2) by country levels, not by rural and urban areas, people from Greece have a lower pro-vaccination orientation against C19, closer to that of the people from the New Member States (figures not shown here). For both groups of New and Old Member States, it was found that the pro-vaccination orientation was higher in the big cities compared with that in rural communities. The only exception is the people from Bulgaria, with a lower dominant opinion in favour of anti-C19 vaccination in urban areas compared with rural areas. The trend is in line with H5b and the exception should be addressed by country studies. In major EU cities, for example, the pro-vaccination dominant orientation index was 71% in the May 2021 survey data. For rural communities in the Union, the dominant pro-vaccination orientation was 66%. For all Central and Eastern European countries, these indices were lower than the EU average. Overall, the data in Figure 1 confirm H5 in its variants H5a and H5b. Only in the case of Belgium and Slovakia are situations different from the European trend in the sense that at the level of these countries, the index of the anti-C19 pro-vaccination orientation is higher in rural than in urban areas. Figure 2. Dominant orientation on anti-C19 vaccination by rural and urban areas in EU countries (%). Data source: Flash Eurobarometer 494. The figures in the diagram represent the index of dominant orientation on C19 vaccination = (% positive orientation−% negative orientation) × (100−DK)/100. The positive orientation sums up those that are vaccinated or intend to do it soon or sometime in 2021. The negative orientation is given by those that said that they will never vaccinate or will do it an indefinite later. DK – non-answers. Urban here means living in large cities and rural is the category including villagers and dwellers of towns. For example, the index of dominant orientation on C19 vaccination in urban areas of Spain is 92%, 10 percentage points larger than in rural areas. Countries with very small samples (Malta, Cyprus, Estonia, and Luxembourg) were not included in the analysis. The index of dominant orientation on C19 vaccination for rural Bulgaria is −3. The second level of differentiation can be identified at the level of country clusters (Figure 3). The graph allows the identification of three similarity nuclei, in the sense of groupings of countries with the utmost similarity between the anti-C19 vaccination behavioural profiles, assessed in terms of past intentions and behaviours: Bulgaria–Slovakia, Denmark–Portugal, and Ireland–Sweden–Netherlands–Spain. The largest grouping, in terms of the number of countries, is Ireland–Sweden–Netherlands–Spain. Italy also revolves around this nucleus of similarity. Together, the five countries recorded, in mid-2021, form the European grouping with the strongest intentions of rapid vaccination against C19 (Table 2). Three of the five countries are part of the developed North-Global (Ireland–Sweden–Netherlands), and two, Spain and Italy are in the southern part of the continent. We also keep in mind, for the time being, the information on the possible relevance of the territorial proximity between some of the countries that make up the grouping. Figure 3. Dendrogram of similarities among European countries by intentions and practices of vaccination against C19. Data source: Eurobarometer 494. Hierarchical cluster analysis, furthest neighbour, Pearson correlations as measures of similarity. Profile of each country determined by the share of the country respondents to the question on practices and intention of vaccination against C19: vaccinated already, as soon as possible, later on in 2021, undecided, later, never. Countries with very small samples (Malta, Cyprus, Estonia, and Luxembourg) were not included in the analysis. The six clustering variables are standardised with z scores. The vertical line in the diagram is a marker of the most homogeneous clusters of similarity. The data in Figure 3 and Table 1 make it possible to highlight the fact that most of the countries in the Old EU belong to pro-vaccination groups, in intent and behaviour. To the mentioned grouping of the five countries are added the groupings of similarity Belgium–Germany and Denmark–Finland–Portugal. The only countries under analysis that are part of the Old EU Member States but do not show a strong similarity to that model are Austria, France, and Greece. Austria and France are closer in terms of the behavioural-attitudinal model of anti-C19 vaccination to the countries of the East and Centre of the EU, Hungary, in particular. Greece, for its part, has a behavioural profile closer to Bulgaria–Croatia–Slovakia than to that of the countries of the Old EU. For now, we note, at a descriptive level, again, a confirmation of H5 that formulated the expectation that the intentions and pro-vaccination behaviours-C19 will be much stronger in the Old, compared with the New EU. Table 1. Profiles of European Union countries by patterns of orientation to the vaccination against C19. Types Sub-types Clusters of similar countries by vaccination behaviours ‘When would you like to get vaccinated against COVID-19?’ Never Later Do not know Sometime in 2021 Soon Already vaccinated Controversy societies Polarised between vaccinated and antivaxxers FR AT HU 6.0 3.1 3.7 −2.2 −19.1 12.8 Large share of antivaxxers LV SI 5.4 3.8 3.0 .3 −5.8 −1.2 Between never and undecided LT PL 9.6 .5 10.0 .1 −9.4 −1.9 A high controversy society RO 3.4 3.7 3.8 3.3 −14.2 5.9 Controversy with high share of antivaxxers SK HR BG 12.3 10.3 3.7 5.8 −11.5 −6.8 Antivaxxers orientation Prevalent antivaxxers orientation GR 2.5 4.6 −.9 3.0 −2.9 −2.4 High antivaxxer orientation CZ 5.7 3.9 −.6 −.5 −1.4 −3.3 Provaccination Provaccination orientation DK FI PT −5.4 −.9 −2.3 7.5 4.2 −3.9 High intentions to provaccination IE NL SE ES IT −14.6 −6.7 −6.4 1.3 26.1 −10.9 Unconditional provaccination BE DE −4.7 −5.8 −5.8 −8.7 8.2 6.1 Data source: Eurobarometer 494. The clusters of similar countries by vaccination orientation are derived from the dendrogram in Figure 1. Figures are adjusted standardised residuals in a table crossing cluster of countries with the values of the variable referring to vaccination intentions and past behaviours. A positive sign indicates a positive association between column and row value and the negative one is significant for a disassociation. For a probability of 95%, the threshold for significance is 1.96. For example, the highest positive association to getting vaccinated soon is for the subsample of the survey from Ireland–The Netherlands–Sweden–Italy–Spain as indicated by the value of 26.1. Shaded figures indicate significat positive associations. The most specific profiles of behaviour-intention are for Romania, the Czech Republic, and Greece. Romania is in the same cluster as Latvia, Lithuania, Poland, and Slovenia but as a marginal member, with a rather lower similar profile to the rest of the grouping. The Czech Republic is also in the same cluster as other Eastern European countries (Slovakia, Croatia, and Bulgaria) but, also, as a marginal member, with a lower coefficient of similarity. The same marginal position in the same cluster is specific to Greece which is part of the Old Member States of the EU. Of course, the 10 groupings of countries generated by cluster analysis are with diffuse limits. At this stage of the analysis, it is not clear whether they differ from each other because of their socio-demographic compositions or certain cultural or economic characteristics. We will have an answer to the question when we introduce the survey respondents belonging to those groupings as independent variables in multivariate analysis and multilevel models. Previous findings have operated with country-wide survey data. If the analysis goes down to the individual level, can we talk about country models or a European model of reporting on anti-C19 vaccination? What is the socio-demographic profile of those who said they would get vaccinated soon or never or opted for one of the intermediate variants? What about those who have already been vaccinated? These questions we will answer in the section that follows. Socio-cultural orientations in the anti-C19 vaccination The pro-vaccination attitude leads, as expected according to the first hypothesis, to the adoption of vaccination intentions soon or in the foreseeable future of the survey year (Table 2). A low level of this attitude favours, at the European level, the adoption of the refusal to vaccinate. Only in the grouping of those who vaguely intend to adopt that vaccine the level of that attitude does not matter significantly. Table 2. Predictors of intentions to get vaccinated at the European Union level by categories of practices, May 2021. Predictors ‘When would you like to get vaccinated against COVID-19?’ (reference category ‘do not know’) Never Later Sometime in 2021 Soon Already vaccinated Index of pro-vaccination orientation (IPVO) −0.009 *** −0.001 0.006 *** 0.015 *** 0.016 *** Vaccinated as adult* 0.053 −0.023 0.174 0.411 *** 0.744 *** Index of trust in institutions −0.271 ** 0.262 ** 0.476 *** 0.478 *** 0.442 *** Index of trust in online networks and web 0.065 0.054 −0.004 0.024 −0.143 ** Large city* −0.056 −0.099 0.083 0.250 * 0.112 Age 15–29 years old (yo)* 0.124 0.378 * 0.451 ** 0.264 + −0.485 ** Age 60+ yo* (reference age 30–59 yo) 0.055 0.473 ** 0.016 0.098 1.609 *** Man* 0.250 * 0.342 ** 0.332 ** 0.458 *** 0.274 ** Tertiary education* 0.049 0.106 0.179 0.094 0.024 Still studying* −0.117 −0.163 −0.060 0.053 −0.242 Employee* 0.056 0.402 ** 0.448 *** 0.532 *** 0.684 *** Having with children under 15 years old* 0.022 0.118 0.254 * 0.130 −0.432 *** Constant −1.167 *** −0.270 0.585 ** 0.906 *** 0.486 * Pseudo R2 0.236 N 23,560 Data source: Eurobarometer 494, May 2021. Multinomial logistic regression. The full model that is not presented here included also 23 residence countries (with Finland as a reference category). The coefficients in the table are computed by controlling for the residence country. Pseudo R2 for the model without the country predictors is 0.210. Regression in STATA, using the pweight option. * p < 0.05; **p < 0.01; ***p < 0.001. Why some people are more pro-vaccination oriented, against C19, than others, recording higher IPVO values? It is a question of status, experience, and place. We will summarise shortly the answer to this question, based on the results of multiple regression (that is not presented here) having IPVO as a dependent variable. High pro-vaccination orientation, at the EU level, is specific for older men from cities, being of higher or middle-level education, with high trust in relevant institutions for anti-C19 vaccination, and that lived the experience of being vaccinated before to avoid other illnesses. All these factors have specific, net effects to favour higher values for IPVO. Irrespective of these factors, there are country effects. Living in Spain, Italy, Sweden, and Ireland had the highest positive impact on pro-vaccination against C19. At the other extreme are the countries with the highest negative effects on the same attitude. Here are included mainly central-European countries like Slovenia, Slovakia, and Austria. Living in Latvia had the highest negative effect on the same attitude. Trust in the government, health administration, and local government appears to be one of the strongest factors in favouring the anti-C19 vaccination, in line with the expectations formulated by H2. Symmetrically, distrust in these institutions leads to the refusal of vaccination. Surprisingly, trust in online social networks does act by H3, only for those who have already been vaccinated. Specifically, those who have already been vaccinated tend to have a low level of trust in online networks. For the remaining categories analysed, that factor no longer matters significantly. Such a finding may also be related to the fact that the analysis model was not sufficiently well specified. We do not know, for example, from the basic data of the survey used, which of those interviewed were high-frequency Internet users, associated with social networks. In H4, the expectation is formulated that people who have been vaccinated in the past as adults, for another disease, will be more oriented in favour of anti-C19 vaccination. We do not find that relationship as significant except for those who have already been vaccinated or are going to get vaccinated shortly. That factor appears to be, by presence, a favourable condition for rapid vaccination. His absence, however, does not automatically lead to the blocking of the intention to vaccinate anti-C19. A poorly structured culture of vaccination as a preventive measure is especially specific for Romania, Lithuania, Poland, and Hungary, in the New Member States of the EU, and Italy and Greece from the Old EU. It is in these countries that there is a lower probability of being before vaccinated as an adult for other illnesses, controlling for socio-demographic factors (gender, age, education urban, or rural residence). On the positive extreme, with a high effect on being vaccinated as an adult, irrespective of disease is Portugal. (Results of regression analysis that are not presented here.) The H5a assumption holds that residency in the Old EU Member States favoured a higher likelihood of adopting a positive attitude towards the vaccination against C19. The descriptive data, to which we have already referred, confirmed this expectation. The gap between Old and New Member States of the EU, in terms of pro-vaccination orientation, has multiple sources. Trust in institutions, for example, that are relevant for the vaccination against C19 is systematically higher in the Old Member States (Table 4), except in France. The quality of the institutions that provide the context for anti-C19 vaccination comes, very likely, from the long history of the survival of the low-quality of institutions for the former communist states of the EU (Mishler and Rose, 2001). Social development as measured by life expectancy at birth is also, systematically higher in the Old Member States, with a variation between 81 and 83 years old, in 2021, versus the New Member States with a life expectancy between 71 and 77 years old, according to EUROSTAT sources. The only exception from this point of view is Slovenia, a New Member State, which has a life expectancy that is equal to the value of the index for Germany (81 years old), in 2021. Low life expectancy goes together with lower quality of the health system and higher morbidity rates. Economically, all the New Member States had a gross domestic product (GDP) per capita below the EU average (EUROSTAT data for 2019, not mentioned here). By contrast, 10 out of 14 Old Member States had a GDP per capita higher than the EU average. These are only examples of development gaps between New and Old Member States that could be relevant to population attitudes and behaviours related to vaccination against C19. H5a is also supported by data at the individual level, in multilevel analysis. In Table 3 we present, in simplified form, a picture of the countries for which we have recorded, with data at an individual level, a significant impact of the country of residence on the personal orientation in terms of anti-C19 vaccination. All the eight countries for which there is a positive and significant effect on the intention to vaccinate soon are in the category of the Old EU Member States. The model of adopting the vaccine in the past, until mid-2021, was specific for Germany–Ireland–Belgium, Austria–France and Hungary–Romania. The cultural model of refusing C-19 vaccination is specific to three Eastern European countries (the Czech Republic, Croatia, and Bulgaria) and Greece. Table 3. Specific patterns of vaccination behaviours and intentions. Countries as significant predictors ‘When would you like to get vaccinated against COVID-19 ? ‘ (reference category ‘do not know’) Never Later Sometime in 2021 Soon Already vaccinated Germany + + Ireland + + Belgium + + The Netherlands + Portugal + Italy + Spain + Denmark + Czech Republic + + Greece + + Croatia + Bulgaria + Hungary + Austria + France + Romania + Data source: Eurobarometer 494, May 2021. Simplified table with significant positive regression coefficients (+) using the full model of multinomial regression from Table 2. Example: Living in the Netherlands favours significantly the option of declaring that the person will vaccinate soon, caeteris paribus. Challenges of socio-demographic and institutional contexts Beyond the effects of attitude, trust in institutions relevant to the anti-C19 vaccination and personal health experiences, socio-demographic factors also matter significantly when regression analyses are done country by country. The continental region and country context have an important role to play in differentiating causal or conditioning relationships (Tables 5 and 6 in the Annex containing patterns of multiple regression with the ordinal-dependent variable measuring orientation towards the C19 vaccination). In each of the countries included in the analysis, being over 60 years of age significantly increases the likelihood of C19 vaccination, but more in the countries of the Old EU than in the New Member States. Why is this so? A possible explanation would be the higher average age in the West and the North than in the East. Further analysis is needed. The only exception is Latvia, where being a person over the age of 60 years does not have a significant impact on behavioural orientation towards vaccination. The residential environment itself, controlling the other predictors, only matters to Finland, Latvia, and Romania. In these countries, people in urban areas, in large cities, in particular, tend to support anti-C19 vaccination more than the population in rural communities. It is likely that in these countries social interaction favourable to C19 infection is much stronger in urban than in rural areas compared with the situation in the rest of the EU. It’s a hypothesis generated from data analysis. To be tested, though. The presence of children under the age of 15 in families tends to discourage adult vaccination in 10 of the 23 analysed societies. This effect has a maximum intensity in Poland, the Czech Republic, and Ireland and Finland. We do not know why that is the case. Sampling mode effect or country conditionings? In the total European sample, the status of employees considerably increases the probability of vaccination in fact or as an intention (Table 2). When the analysis is made country by country and with a dependent variable of the ordinal type, the results are different (Table 5 and Table 6 in annexe). Among the countries of the Old EU, only in Italy do we see a significant favouring of the anti-C19 vaccination for those who work. In Eastern Europe, that relationship is more common. It occurs, at a significant level, in descending order of intensity, in Romania, Hungary, Lithuania, and Slovakia. Why only here? Was it only here a higher institutional pressure for the C19 vaccination at the level of employees? Likely, it is not only a supplementary pressure for employees to vaccinate coming from their institutional environment but, also, a higher density of interactions at work, stimulating the employees to be more cautious for their and their family’s health care. Vaccination for employees could be, also, a way to preserve their job in pandemic conditions when finding a job is not so easy. Conclusion and discussion The integrated approach of the vaccination against C19 was promoted, in this analysis, by including past personal vaccination, along with the values of the intention of anti-C19 vaccination, in the same variable, by promoting multilevel analysis, and by expanding the list of predictors. The differentiating lines are not only those between vaccinated–unvaccinated and I will get vaccinated soon versus never. In addition, it was also useful to measure the vaccination orientation both by one nominal and one ordinal variable, allowing for triangulation and sensitivity analysis (Treiman, 2014). Ordinal measurement considered the stages between being vaccinated (score 6) and clear refusal of vaccination against C19 (score 1) as extreme states having between them the ordered states in the series I will vaccinate soon (5), sometime in 2021 (4), do not know (3), later (2). Nominal measurement considered also the same six possible categories of the variable vaccination but the coding numbers here indicate no order. The same technique of regression analysis was used on nominal and ordinal measures of the same measures of vaccine orientation and gave consistent results. The research question was no more, as in the classical TPB, ‘what are the determinants of the intention to adopt certain behaviours?’. Intentions to adopt vaccination against C19 were considered as stages leading to vaccination. The target of the research was to identify significant predictors for each value of the vaccination orientation, using a large representative sample for 23 out of the 27 countries of the EU. The new integrated approach allowed for an understanding of anti-C19 vaccination as an innovation diffusion process (Rogers et al., 2005), distinguishing between early adopters, late adopters, and opponents of vaccination against C19. At the individual level, all five hypotheses are supported by data. Pro-vaccination orientations against C19 tend to be stronger for those with vaccination-friendly attitudes (IPVO), with confidence in the relevant institutions for vaccination against C19, who have the experience of being vaccinated as an adult, against other diseases and reside especially in major cities in the EU Member States. The culture of vaccination in general (as measured by IPVO and prior experiences of vaccination as an adult) and the attitude of the interviewee on relevant institutions towards anti-C19 vaccination are by far more important than education on vaccination behaviour. Education per se, controlling for other factors, has no significant effect on vaccination orientation (see Table 2). This could be an indication that education has mostly an indirect effect on anti-C19 vaccination, through the medium of attitudes, perceptions and trust. Higher education, for example, favours more positive attitudes on vaccination against C19 (higher values of IPVO), and, implicitly or indirectly, more structured behaviours against C-119 vaccination. The fact that the experience of vaccination, for whatever illnesses, favours significantly the belonging to the category of the already vaccinated and to those that are decided to be soon vaccinated is not only a confirmation of the fourth hypothesis. It has, also, a methodological relevance to the fact that the practice of past vaccination is part of a continuum as the same type of causal conditioning appears only for those that are decided to be vaccinated soon. No other categories on the measure of vaccination intention connect significantly to the predictor measuring the culture of vaccination. At a more aggregated level, national contexts, in turn, bring significant differentiation in pro-vaccination guidelines. Germany, Ireland, and Belgium, for example, were, in mid-2021, the countries with the strongest orientation in favour of anti-C19 vaccination both in terms of the share of people already vaccinated and in terms of rapid vaccination intentions. The strongest nucleus of antivaccine against C19 was, also for mid-2021, in Greece, Bulgaria, Croatia, and the Czech Republic. County effects on behaviour-intentions of vaccination against C19 were significantly conditioned by the culture of vaccination as an adult against other illnesses, previous pandemic period. Easter European countries like Romania, Poland, Hungary, and Lithuania, recorded a significantly poorer culture of vaccination to prevent other diseases than C19. Portugal, from the grouping of the Old EU states, is an opposite example of a highly structured culture of vaccination against other illnesses, favouring pro-vaccination against C19. At the space level, we found that the pro-vaccination orientation against C19 is stronger in the Old than in the New Member States of the Union, and the urban area compared with the rural area, with few exceptions discussed in the text. We do not know exactly what mostly counts for the higher anti-C19 vaccination in the Old compared with the New EU. There should be some other factors that were not included in the analysis models. Maybe the quality of the health system or the social contexts with a higher density of population favour more social interactions in the Old compared with the New Member States. Apart from the distinction between the Old and the New Member States, in terms of the index of the dominant orientation of the population concerning the anti-C19 vaccination, the differences in the similarity between groups of countries also operate a lot. The 10 groupings of countries also settle into three broad categories with dominant pro-vaccination, and anti-vaccination orientations and strong controversies on the issue within the countries. The largest group of countries with predominantly pro-vaccination-oriented populations consists of Ireland–Netherlands–Sweden–Spain–Italy. Anti-vaccination guidelines were recorded, with maximum intensities, in mid-2021, in Greece and the Czech Republic. The groups of maximum controversy between the pros and cons of vaccination were in Latvia–Slovenia, Lithuania–Poland, and Romania. The finding is significant for the fact that at the national level there are not only pros and cons in societies but there are also controversial places. Several factors that are unmeasured here could count. The institutional spaces of vaccination with its early or later, good or less satisfactory communication and knowledge processes seem to be essential ones. The ability of a society to learn from population reactions to the vaccination measures and to correct public policies, as a function of these reactions were, very likely, important aspects. There are several uncertainties or weak points in the analysis of these data. The attitude towards anti-C19 vaccination is a logical predictor of the intention to vaccinate against C19, in line with the TPB. What about including, as we did in the article, the attitude on the vaccination against C19 as a predictor of past vaccination against C19 in the multinomial regression? It is possible, at least in theory, to have a successful vaccination not only as a consequence of the previous attitude but, also, the current attitude regarding anti-C19 vaccination as determined by the former practice of ani-C19 vaccination. Even, if possible, such a reverse or circular effect, is rather unlikely because the anti-C19 vaccine was rather new. So, a certain circular loop could function between the two variables but the probability is higher of having a structured attitude favouring anti-C19 vaccination even before the past vaccination against C19. The Eurobarometer survey we are using here was done in May 2021. Unfortunately, we could not have a dynamic image including changes that took place, at the individual level, in terms of anti-C19 vaccination. There is no other survey similar to the flash Eurobarometer 494, including the same key variables for a replication of the analysis. It is, also, a weak point of this analysis the fact that survey data do not allow us to know what kind of vaccination was adopted by the interviewee. It is very likely that not only the culture of vaccination counts but, also, the type of vaccination. Author biography Dumitru Sandu is emeritus professor of sociology at the University of Bucharest. His main publications are on transnational migration, transition sociology, community and regional development, and measures of social capital. Appendix Table 4. Descriptives of key predictors for vaccination against C19. Typology of vaccination orientations in the society Country Vaccinated as adulta Residence in large citya Employeea % 15–29 years old % 60+ years old Index of provaccine orientation Index of trust in institutions relevant for vaccination Mean Mean Mean Row N % Row N % Mean SD Mean SD Controversy societies Polarised between vaccinated and antivaxxers France .78 .22 .39 18.8 33.4 −14.32 101.37 −.11 .94 Hungary .58 .38 .52 15.0 32.8 −36.36 109.81 −.37 .78 Austria .81 .33 .49 26.8 23.8 −28.96 109.47 .04 1.04 Large shares of antivaxxers Latvia .64 .48 .49 25.8 15.0 −69.65 102.77 −.36 .77 Slovenia .67 .19 .52 20.0 21.6 −56.95 106.65 −.39 .69 Between never and undecided Lithuania .45 .50 .49 27.7 18.3 −14.99 105.47 −.20 .87 Poland .55 .37 .51 23.8 22.3 −21.13 101.84 −.37 .79 A high controversy society Romania .42 .45 .57 21.4 16.4 −13.41 106.08 −.07 1.00 Controversy with high share of antivaxxers Slovakia .76 .23 .49 20.3 24.3 −46.75 111.33 −.21 .86 Bulgaria .68 .60 .39 21.1 21.0 −47.35 108.39 −.26 .87 Croatia .66 .37 .47 22.1 26.3 −39.26 98.85 −.40 .70 Antivaxxers orientation Prevalent antivaxxers orientation Greece .60 .52 .41 25.5 12.9 3.62 95.85 −.08 .86 High antivaxxers orientation Czech Rep. .87 .29 .52 18.5 26.8 −20.11 97.41 −.19 .76 Provaccination orientation Provaccination orientation Denmark .71 .35 .48 24.6 27.8 7.02 91.81 .16 1.03 Portugal .90 .36 .46 21.2 24.7 3.80 84.14 .37 1.07 Finland .84 .30 .30 20.6 31.5 10.50 95.55 .44 1.13 High share of soon provaccination Ireland .64 .39 .58 26.1 13.2 13.90 93.21 .21 1.03 Sweden .74 .33 .49 17.3 32.7 40.64 89.59 .56 1.22 The Netherlands .67 .22 .56 17.6 33.8 −7.17 100.10 .31 1.09 Spain .69 .37 .52 17.3 28.1 35.01 82.44 −.02 .91 Italy .56 .30 .33 16.3 29.0 23.58 90.82 .07 .96 Unconditional provaccination Belgium .80 .23 .40 18.2 28.2 −4.82 101.76 .02 1.01 Germany .87 .29 .45 12.9 34.0 −.15 100.07 .09 1.08 Data source: Eurobarometer 494, May 2021. All the descriptive statistics and multivariate analyses are computed by the weighting variable specified in the Flash Eurobarometer EB494 by the data provider. Malta, Cyprus, Luxembourg and Estonia having very small samples of less than 100 persons, were omitted from all the computations. a Dummy variable. Maximum values on columns are marked by shadow. Minimal values on columns are in bold. Shadow mark significant coefficients for p < 0.05. Table 5. Predicting vaccination against C19 in Old Member States of the European Union. Predictors Denmark Finland Portugal Ireland Netherlands Sweden Spain Italy Germany Belgium France Austria Greece Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z IPVO 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 Index of trust in institutions 0.17 0.01 0.10 0.11 0.16 0.02 0.29 0.00 0.11 0.05 0.08 0.17 0.18 0.02 0.06 0.36 0.03 0.62 0.12 0.07 0.12 0.08 0.19 0.00 0.33 0.00 Index of trust in online networks and web −0.13 0.05 −0.13 0.01 −0.13 0.16 −0.16 0.01 −0.07 0.29 −0.03 0.59 −0.16 0.06 −0.05 0.36 −0.07 0.24 −0.10 0.11 −0.16 0.05 −0.24 0.00 −0.20 0.00 Vaccinated as adult* 0.30 0.04 0.75 0.00 0.49 0.04 0.24 0.11 0.32 0.02 0.31 0.05 0.31 0.05 0.32 0.01 0.64 0.00 0.30 0.06 0.25 0.12 0.36 0.04 −0.02 0.88 Age 15–29 years old* −0.21 0.20 −0.33 0.05 −0.41 0.01 −0.29 0.07 −0.43 0.02 −0.43 0.02 0.48 0.02 −0.11 0.59 0.16 0.46 −0.59 0.00 −0.67 0.00 −0.12 0.41 −0.69 0.00 Age 60+ yo* 2.32 0.00 0.85 0.00 2.29 0.00 1.52 0.00 1.90 0.00 1.53 0.00 2.26 0.00 1.31 0.00 1.45 0.00 1.04 0.00 1.25 0.00 1.52 0.00 0.81 0.00 Man* −0.33 0.01 0.16 0.19 0.04 0.80 −0.22 0.13 −0.06 0.65 −0.24 0.07 0.03 0.86 0.00 0.99 −0.09 0.50 −0.14 0.26 −0.06 0.64 0.02 0.87 0.26 0.09 Tertiary education* 0.07 0.66 −0.07 0.66 −0.02 0.89 0.00 1.00 −0.12 0.43 −0.36 0.02 −0.07 0.65 0.02 0.91 0.04 0.79 −0.03 0.86 −0.23 0.19 −0.13 0.38 0.19 0.33 Still studying* −0.26 0.21 0.07 0.73 0.09 0.67 −0.35 0.12 0.05 0.85 −0.75 0.00 −0.88 0.00 −0.13 0.61 −0.12 0.62 −0.06 0.80 0.00 1.00 0.02 0.92 −0.01 0.96 Employee* −0.05 0.74 0.27 0.06 0.22 0.13 −0.05 0.73 −0.03 0.87 0.05 0.78 −0.03 0.84 0.38 0.01 0.30 0.04 −0.15 0.33 0.17 0.30 0.21 0.11 0.25 0.11 Having children under 15 years old* −0.15 0.29 −0.42 0.00 −0.13 0.34 −0.54 0.00 −0.14 0.33 −0.30 0.03 −0.15 0.34 −0.26 0.05 −0.35 0.02 −0.32 0.03 −0.32 0.03 −0.25 0.06 0.01 0.94 Urban residence* 0.11 0.39 0.32 0.02 −0.01 0.97 −0.09 0.56 −0.01 0.97 0.02 0.91 0.17 0.24 0.03 0.80 0.00 0.98 −0.14 0.34 0.13 0.40 0.12 0.36 0.04 0.81 Pseudo R2 0.22 0.17 0.17 0.20 0.19 0.21 0.11 0.15 0.16 0.15 0.19 0.19 0.20 N 1004 1004 1014 1052 1008 1005 1008 1061 1052 1001 1001 1067 1040 Data source: Eurobarometer 494 May 2021. Ordinal logistic regression for each country. Countries of similar profiles by vaccine orientation (Figure 1) or in proximity are placed in proximity columns. IPVO: Index of pro-vaccination orientation. * Dummy variable. Shadow mark significant coefficients for p < 0.05. Table 6. Predicting vaccination against C19 in New Member States of the European Union. Predictors Latvia Slovenia Lithuania Poland Romania Slovakia Bulgaria Croatia Czech Rep Hungary Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z Coef. P > z IPVO 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.02 0.00 0.01 0.00 0.01 0.00 0.01 0.00 Index of trust in institutions 0.40 0.00 0.25 0.03 0.07 0.46 0.17 0.05 0.17 0.02 0.15 0.05 0.18 0.04 0.27 0.00 0.33 0.00 0.26 0.08 Index of trust in online networks and web −0.25 0.00 −0.10 0.27 −0.12 0.08 −0.02 0.77 −0.20 0.00 −0.19 0.00 −0.03 0.58 −0.13 0.03 −0.27 0.00 −0.26 0.00 Vaccinated as adult* 0.31 0.07 0.20 0.18 0.16 0.30 0.50 0.00 −0.05 0.74 −0.04 0.76 0.38 0.01 0.30 0.03 0.40 0.02 0.50 0.00 Age 15–29 years old * 0.03 0.89 −0.30 0.12 −0.28 0.09 −0.57 0.00 −0.19 0.33 −0.74 0.00 −0.16 0.38 −0.46 0.01 −0.61 0.00 −0.22 0.43 Age 60+ yo* 0.41 0.06 1.21 0.00 1.10 0.00 0.43 0.03 0.60 0.01 0.68 0.00 0.52 0.00 0.61 0.00 0.83 0.00 0.94 0.00 Man* −0.08 0.60 0.05 0.74 −0.06 0.70 −0.10 0.44 −0.01 0.97 −0.18 0.16 0.04 0.76 0.00 0.97 −0.02 0.86 0.05 0.78 Tertiary education* 0.17 0.34 0.04 0.79 0.24 0.17 0.03 0.85 0.10 0.57 −0.18 0.21 0.12 0.42 0.20 0.19 −0.16 0.27 0.52 0.00 Still studying* 0.46 0.07 0.10 0.71 0.30 0.28 0.14 0.58 0.30 0.26 0.26 0.28 0.10 0.76 0.16 0.58 0.11 0.52 0.67 0.10 Employee* 0.28 0.07 0.24 0.10 0.53 0.00 0.28 0.06 0.61 0.00 0.29 0.05 0.16 0.24 0.15 0.29 −0.03 0.84 0.60 0.00 Having children under 15 years old* −0.15 0.33 −0.23 0.14 0.07 0.65 −0.48 0.00 0.07 0.62 −0.14 0.30 −0.33 0.02 −0.15 0.27 −0.45 0.00 0.06 0.77 Urban residence* 0.36 0.02 −0.17 0.34 −0.01 0.95 −0.27 0.06 0.32 0.04 0.19 0.22 0.13 0.39 0.04 0.75 −0.03 0.84 0.10 0.54 Pseudo R2 0.18 0.20 0.19 0.16 0.19 0.19 0.19 0.17 0.17 0.23 N 1019 1012 1113 1020 1014 1005 1014 1043 1009 1001 Data source: Eurobarometer 494, May 2021. Ordinal logistic regression for each country. Countries of similar profiles by vaccine orientation (Figure 1) or in proximity are placed in proximity columns. IPVO: Index of pro-vaccination orientation. * Dummy variable. Shadow mark significant coefficients for p < 0.05. Author’s note: The title page of the article, including statements relating to our ethics and integrity policies: original article, not-published. Preliminary form presented in Romanian in Contributors.ro, 19 February 2022. Data availability: Flash Eurobarometer 494, collected data in May 2021. The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Ethical approval: Not the case Funding: The author received no financial support for the research, authorship, and/or publication of this article. Patient consent: Not the case Permission to reproduce material from other sources: All the tables and figures are produced by the author. ORCID iD: Dumitru Sandu https://orcid.org/0000-0001-8618-2252 ==== Refs References Ajzen I (1991) The theory of planned behavior. Organizational Behavior and Human Decision Processes 50 (2 ): 179–211. Ajzen I (2011) The theory of planned behaviour: Reactions and reflections. Psychology & Health 26 (9 ): 1113–1127.21929476 Babbie ER (2020) The Practice of Social Research. Boston, MA: Cengage Learning. Balas EA Chapman WW (2018) Road map for diffusion of innovation in health care. Health Affairs 37 (2 ): 198–204.29401030 Buhmann A Brønn PS (2018) Applying Ajzen’s theory of planned behavior to predict practitioners’ intentions to measure and evaluate communication outcomes. Corporate Communications: An International Journal 3 (23 ): 377–391. Burke PF Masters D Massey G (2021) Enablers and barriers to C19 vaccine uptake: An international study of perceptions and intentions. Vaccine 39 (36 ): 5116–5128.34340856 Cerda AA García LY (2021) Hesitation and refusal factors in individuals’ decision-making processes regarding a coronavirus disease 2019 vaccination. Frontiers in Public Health 9 : 1–14. Conner M Armitage CJ (1998) Extending the theory of planned behavior: A review and avenues for further research. Journal of Applied Social Psychology 28 (15 ): 1429–1464. Fishbein M (1967) Attitude and the prediction of behavior. In Fishbein M (Ed.) Readings in Attitude Theory and Measurement. New York: John Wiley, pp.477–492. Gaston NM (2017) Contextualizing information behaviour: A methodological approach. Journal of Critical Library and Information Studies 1 (1 ): 1–33. Mishler W Rose R (2001) What are the origins of political trust? Testing institutional and cultural theories in post-communist societies. Comparative Political Studies 34 (1 ): 30–62. Robertson DA Mohr KS Barjaková M , et al . (2021) A lack of perceived benefits and a gap in knowledge distinguish the vaccine hesitant from vaccine accepting during the COVID-19 pandemic. Psychological Medicine 1–4. DOI: 10.1017/S0033291721003743. Rogers EM Medina UE Rivera MA , et al . (2005) Complex adaptive systems and the diffusion of innovations. The Innovation Journal: The Public Sector Innovation Journal 10 (3 ): 1–26. Sandu D (2021) Social Worlds of Attitude Towards Anti-COVID-19 Vaccination: Romanians in the European Context. Contributors.ro. Seale H Heywood AE Leask J , et al . (2021) Examining Australian public perceptions and behaviors towards a future COVID-19 vaccine. BMC Infectious Diseases 21 (1 ): 1–9.33390160 Sommer L (2011) The theory of planned behaviour and the impact of past behaviour. International Business & Economics Research Journal (IBER) 10 (1 ): 91–110. Treiman DJ (2014) Quantitative Data Analysis: Doing Social Research to Test Ideas. John Wiley. Trent M Seale H Chughtai AA , et al . (2021) Trust in government, intention to vaccinate and C19 vaccine hesitancy: A comparative survey of five large cities in the United States, United Kingdom, and Australia. Vaccine 40 (17 ): 2498–2505.34218963 Wang Y Liu Y (2021) Multilevel determinants of COVID-19 vaccination hesitancy in the United States: A rapid systematic review. Preventive Medicine Reports 25 : 101673.34934611 Wolff K (2021) C19 vaccination intentions: The theory of planned behaviour, optimistic bias, and anticipated regret. Frontiers in Psychology 12 : 1–7. Wollast R Schmitz M Bigot A , et al . (2021) The theory of planned behaviour during the COVID-19 pandemic: A comparison of health behaviours between Belgian and French residents. PLOS ONE 16 (11 ): e0258320.34735473

0

PMC9749066

NO-CC CODE

2022-12-15 23:22:50

no

Int Sociol. 2022 Dec 13;:02685809221137783

utf-8

Int Sociol

10.1177/02685809221137783

oa_other

==== Front Int J Soc Psychiatry Int J Soc Psychiatry ISP spisp The International Journal of Social Psychiatry 0020-7640 1741-2854 SAGE Publications Sage UK: London, England 36511138 10.1177/00207640221141784 10.1177_00207640221141784 Original Article Social anxiety and depression symptoms in Chinese left-behind children after the lifting of COVID-19 lockdown: A network analysis https://orcid.org/0000-0002-1653-5419 Li Kuiliang 12* Ren Lei 3* Zhang Lei 1 https://orcid.org/0000-0003-0324-8151 Liu Chang 4 Zhao Mengxue 2 Zhan Xiaoqing 1 Li Ling 5 https://orcid.org/0000-0003-2361-7348 Luo Xi 1 Feng Zhengzhi 2 1 Department of Medical English, School of Basic Medical Sciences, Army Medical University, Chongqing, China 2 School of Psychology, Army Medical University, Chongqing, China 3 Department of Military Medical Psychology, Fourth Military Medical University, Xi’ an, China 4 BrainPark, Turner Institute for Brain and Mental Health and School of Psychological Sciences, Monash University, Clayton, Australia 5 College of General Education, Chongqing Water Resources and Electric Engineering College, Chongqing, China Ling Li, College of General Education, Chongqing Water Resources and Electric Engineering College, Chongqing 402160, China. Email: 1628961258@qq.com Xi Luo, Department of Medical English, School of Basic Medical Sciences, Army Medical University, Chongqing 400038, China. Email: luorosi@126.com Zhengzhi Feng, School of Psychology, Army Medical University, Chongqing 400038, China. Email: fzz@tmmu.edu.cn * These authors contributed equally to this work. 13 12 2022 13 12 2022 00207640221141784© The Author(s) 2022 2022 SAGE Publications This article is made available via the PMC Open Access Subset for unrestricted re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the COVID-19 pandemic or until permissions are revoked in writing. Upon expiration of these permissions, PMC is granted a perpetual license to make this article available via PMC and Europe PMC, consistent with existing copyright protections. Background: Returning to social life after the lifting of COVID-19 lockdown may increase risk of social anxiety, which is highly co-morbid with depression. However, few studies have reported the association between them. Aims: To explore the complex relationship between social anxiety and depression symptoms in left-behind children after the lifting of the COVID-19 lockdown. Methods: A cross-sectional survey was conducted 6 months after the lockdown removal. A total of 3,107 left-behind children completed the survey with a mean age of 13.33 and a response rate of 87.77%. Depression and social anxiety severity were assessed by the DSM-5 Patient Health Questionnaire for Adolescents and the DSM-5 Social Anxiety Disorder Questionnaire, respectively. The symptom-level association between the two disorders was examined using network analysis. Results: After the lifting of COVID-19 lockdown, the prevalence of depression and social anxiety in left-behind children was 19.57% and 12.36%, respectively, with a co-morbidity rate of 8.98%. Network analysis showed that “Social tension” and “Social avoidance” had the greatest expected influence; “Humiliation” and “Motor” were bridge symptom nodes in the network. The directed acyclic graph indicated that “Social fright” was at the upstream of all symptoms. Conclusion: Attention should be paid to social anxiety symptoms in left-behind children after the lifting of COVID-19 lockdown. Prevention and intervention measures should be taken promptly to reduce the comorbidity of social anxiety and depression symptoms in the left-behind children after the lifting of lockdown. Depression social anxiety left-behind children adolescents network analysis lockdown removal National Natural Science Foundation of China https://doi.org/10.13039/501100001809 NSFC:81971278 edited-statecorrected-proof typesetterts1 ==== Body pmcIntroduction With the outbreak of COVID-19, many countries especially China have shared plenty of valuable information on disease treatment and control (Chen et al., 2020; Wang, Pan et al., 2020; Zhu et al., 2020). Scientists not only have fought against COVID-19 from the perspective of biomedicine but conducted a great deal of research on mental health. Researchers have attached importance to the mental health of different groups during the COVID-19 pandemic (Almeida et al., 2020; El Hayek et al., 2020; Walton et al., 2020), including medical staff (Abbasi, 2020) and college students (Wang, Hegde et al., 2020), but vulnerable groups, such as children and adolescents, are paid less attention (Holmes et al., 2020). According to WHO reports, 20% of children and adolescents worldwide had mental health problems currently, yet effective treatment coverage was extremely low (WHO, 2022). Children and adolescents are more susceptible to depression and anxiety due to social isolation and fear of the SARS-Cov-2 infection during the COVID-19 pandemic (Loades et al., 2020; Singh et al., 2020). Moreover, they were at a greater risk of social anxiety after the lockdown is lifted (Lim et al., 2022). Among children and adolescents, left-behind children need special attention. Left-behind children are children under the age of 18 who have been separated from at least one parent for more than 6 months (Wang et al., 2019). There are currently hundreds of millions of left-behind children worldwide (Fellmeth et al., 2018), with approximately 61 million in China (Yuan & Wang, 2016). Previous studies have shown that compared with non-left-behind children, left-behind children show more mental health problems due to impaired parent-child relationships, reduced parental support, and deficient parental guidance (Hou et al., 2021; Wang et al., 2019). The current study therefore focuses on the mental health of left-behind children during the COVID-19 pandemic. The evidence above suggests that left-behind children may be more vulnerable to mental illness during the COVID-19 pandemic. Individuals need to resume social activities after the lifting of the lockdown, which may cause an increased risk of social anxiety (Lim et al., 2022; Samantaray et al., 2022). Social anxiety and depression symptoms often co-occurred (Adams et al., 2018), implying that COVID-19 may raise the risk of social anxiety and depression co-morbidity in left-behind children after the lifting of lockdown. However, it remains unclear whether the prevalence of depression-social anxiety comorbidity will also increase. The co-occurrence of these two mental disorders can be explored via network modeling of psychopathology (Jones et al., 2021). Many studies have investigated the network structure of depression and anxiety symptoms during the COVID-19 pandemic (Bai et al., 2021; Liu et al., 2022; Ren et al., 2021). A study identified “depressed” and “nervousness” in adolescents as core symptoms in the network and “depressed” as a bridge symptom connecting the two disorders (Liu et al., 2022), which provides insights into reducing the impact on the mental health of adolescents during the COVID-19 pandemic. However, there is currently no network analysis of social anxiety and depression symptoms during the COVID-19 pandemic. Considering the unique characteristics of left-behind children and the high co-morbidity of social anxiety and depression, this study explored the network structure of social anxiety and depression symptoms in left-behind children after the COVID-19 lockdown was lifted. The following research questions were asked: (1) What are the most important symptoms in the network of social anxiety and depression symptoms? (2) Which symptoms are the bridges between social anxiety and depression symptoms network? (3) Which symptom of social anxiety and depression network has the highest predictive (and potentially causal) priority? Methods Participants and study design A cross-sectional survey was conducted in 32 primary and secondary schools from October to November 2020 in Chongqing, China, and all students over the age of seven were recruited to participate in the study. The survey was carried out by well-trained researchers responsible for giving necessary explanations and instructions. Students participated in the study by filling out a questionnaire in the classroom. Meanwhile, verbal consent has been obtained from their parents, who were notified by schools that the questionnaire was part of a mental health survey. This study has been reviewed and approved by the Medical Ethics Committee of the Department of Medical Psychology, Army Medical University (Project No. CWS20J007). A total of 18,133 questionnaires were collected. First, we excluded 3,854 questionnaires of the participants who did not age between 11 and 17 based on the applicability of the questionnaires or did not indicate age. Among the remaining 14,279 questionnaires, 10,739 filled by the children who were not currently left behind (children who had lived separately from both parents for less than 6 months) were excluded. Among the remaining 3,540 questionnaires, 433 incomplete questionnaires were further excluded. Finally, a total of 3,107 valid questionnaires (87.77% respondents) from left-behind children were included for subsequent statistical analysis. These respondents completed the survey 6 to 7 months after the local lockdown policy was canceled and met the criteria of left-behind children (that is, all respondents were children whose parents had gone out to work for over 6 months after the lockdown was lifted). They received investigations of demographic features, and depression and social anxiety symptoms, which were analyzed using network analysis. Measures Depression symptoms Depression symptoms in adolescents were assessed using the Patient Health Questionnaire for Adolescents (PHQ-A). The scale contains nine items, such as “Little interest or pleasure in doing things” or “Feeling down, depressed, or hopeless.” Participants were asked to choose the item based upon how often each event occurred over the past 2 weeks. Scores for each item range from 0 (“Not at all”) to 3 (“Nearly every day”), with a higher total indicating a higher level of depression. The Cronbach’ α for the current PHQ-A scale was set at 0.902. Social anxiety symptoms Social anxiety was assessed using the Chinese-modified Social Anxiety Scale from the Diagnostic and Statistical Manual of Mental Disorders (DSM-5). The scale consists of 10 items using a 5-point Likert scale, with each item ranging from 0 (“Never”) to 4 (“All of the time”). A higher total score means more severe social anxiety. The Cronbach’ α for this scale was set at 0.923. The Chinese version of assessment tools is from Meilihua Health Systems (Meilihua Health Systems, 2019). The original tools and scoring criteria are available in the Online Assessment Measures of the American Psychiatry Association (American Psychiatric Association, 2013b). Data analysis Graphical LASSO network This network was estimated via Gaussian graphical models (GGMs; Epskamp, Borsboom et al., 2018). Within GGMs, the edges represent partial correlations between two nodes after controlling for all other nodes in the network. The GGMs were estimated based on nonparametric Spearman rho correlation matrices. The regularization of GGMs was carried out using the graphical LASSO (least absolute shrinkage and selection operator) algorithm (Friedman et al., 2008). This regularization process shrinks all edges, and the edges with small partial correlation shrink to zero, so a more interpretative and stable network can be obtained (Epskamp & Fried, 2018; Friedman et al., 2008). Meanwhile, the hyperparameter of the Extended Bayesian information criterion (EBIC) was set to be 0.5, which achieves a good balance between the sensitivity and specificity of extracting real edges (Epskamp & Fried, 2018; Foygel & Drton, 2010). The visualization of the network is derived from the fruchterman-reingold algorithm (Fruchterman & Reingold, 1991). This network was constructed and visualized using the R-package qgraph (Epskamp et al., 2012). In the present network, the node expected influence was calculated via the R-package qgraph (Epskamp et al., 2012). The expected influence is defined as the sum of the value of all edges connecting to a given node. The higher the expected influence, the more important the node is in the network. Moreover, the node bridge expected influence was computed by the R-package networktools (Jones et al., 2021). Bridge expected influence is defined as the sum of the value of all edges connecting a given node with nodes in the other community. Higher bridge expected influence values mean a greater extent for increasing the risk of contagion to other communities (Jones et al., 2021). In the present network, we divided nodes into two communities in advance: one community was depression, which included nine symptoms of depression, and the other was social anxiety, consisting of 10 social anxiety symptoms. We examined the robustness of the network using the R-package bootnet (Epskamp, Borsboom et al., 2018). The accuracy of edge weights was evaluated by calculating 95% confidence interval using a non-parametric bootstrap approach (2000 bootstrap samples) and computing bootstrapped difference test for edge weights. The stabilities of node expected influences and node bridge expected influences were evaluated by computing correlation stability (CS)-coefficient, using a case-dropping bootstrap approach (2,000 bootstrap samples) and computing bootstrapped difference test for them. The value of the CS-coefficient should not be below 0.25 and preferably should be above 0.5 (Epskamp, Borsboom et al., 2018). Directed acyclic graph (DAG) DAG can encode the conditional independence relationships between the nodes in cross-sectional data and identify admissible causal relationships among them (Briganti et al., 2022). Thus, DAG is a directed network, which can reflect the predicted direction of the probabilistic dependence between nodes (McNally, 2016; Moffa et al., 2017). R-package bnlearn and Bayesian hill-climbing algorithm were applied to calculate the DAG for symptoms of depression and social anxiety (Scutari, 2010). The algorithm continuously adds, deletes and reverses the edges’ direction until the best fitting is obtained according to the Bayesian information criterion (BIC) (McNally, Heeren et al., 2017). This includes an iterative process of randomly restarting this procedure with different possible edges connecting different node pairs, perturbing the structure, and applying 50 different random restarts to avoid local maxima. Following previous studies (Bernstein et al., 2017; Blanchard et al., 2021; Heeren et al., 2020; McNally, Mair et al., 2017), we performed 100 perturbations for each restart. To ensure the stability of the DAG, we used the bootstrap approach (10,000 bootstrap samples with replacement) to obtain the final DAG structure, including a two-step process (e.g. Bernstein et al., 2017; Blanchard et al., 2021; Heeren et al., 2020; McNally, Mair et al., 2017). First, we determined how frequently a given edge appeared in the 10,000 bootstrapped DAGs. We then used the optimal cut-point approach of Scutari and Nagarajan (2013) for retaining edges, which obtained DAG with both high sensitivity and specificity. Second, when determining the direction of each edge, if there were edges in the same direction in 51% or more of the 10,000 bootstrapped DAGs, the directional edges would be represented in the final DAG. To make DAG easier to interpret, two visualizations of the resulting outputs were generated, as recommended by previous studies (Bernstein et al., 2017; Blanchard et al., 2021; Heeren et al., 2020). In the first visualization, edge thickness represents the change in the BIC values when that edge is removed from the DAG. In the second one, edge thickness represents directional probabilities. Results Descriptive statistics A total of 3,107 left-behind children were included in the final analysis after the lifting of the lockdown, including 1,615 boys, 1,477 girls, and 15 students not reporting gender in the questionnaires. The mean age was 13.33 years (SD = 1.84). According to the results of the survey, the prevalence of depression among the left-behind children was 19.57% (based on a cut-off score of 10), the prevalence of social anxiety was 12.36% (based on an APA manual score of two) and the comorbidity was 8.98%. Further, 72.66% of the left-behind children with social anxiety also had depression symptoms and 45.89% of them with depression had social anxiety, indicating a very high co-morbidity. More information about the participants can be found in Table 1. All symptom items and codes are shown in Table 2. Table 1. Participant demographics and descriptive statistics of symptoms (n = 3,107). M (SD)/n (%) Age 13.33 (1.84) Gender  Male 1,615 (51.98)  Female 1,477 (47.54)  Missed 15 (0.48) Social anxiety severity (Average total score)  0 1,752 (56.39)  1 971 (31.25)  2 252 (8.11)  3 106 (3.41)  4 26 (0.84) Depression severity (Total raw score)  0–4 1,649 (53.07)  5–9 850 (27.36)  10–14 325 (10.46)  15–19 170 (5.47)  20–27 113 (3.64) Comorbidity of social anxiety and depression Cut-off is 2 for social anxiety and 10 for depression 279 (8.98) Note. The SD is shown in italics to distinguish it from %s in the other brackets. Table 2. Abbreviation, Mean, Standard Deviation of depression, and social anxiety symptoms. Symptoms Abbreviation Mean SD Depression symptoms Feeling down, depressed, irritable, or hopeless D1: Sadness 0.69 0.82 Little interest or pleasure in doing things D2: Anhedonia 0.79 0.87 Trouble falling asleep, staying asleep, or sleeping too much D3: Sleeping problems 0.64 0.91 Poor appetite, weight loss, or overeating D4: Appetite 0.49 0.81 Feeling tired, or having little energy D5: Fatigue 0.80 0.91 Feeling bad about yourself, or that you are a failure or have let yourself or your family down D6: Failure 0.83 0.97 Trouble concentrating on things like school work, reading, or watching TV D7: Concentration 0.73 0.92 Moving or speaking so slowly that other people could have noticed D8: Motor 0.34 0.70 Thoughts that you would be better off dead or of hurting yourself in some way D9: Suicidal ideation 0.37 0.75 Social Anxiety Symptoms Felt moments of sudden terror, fear, or fright in social situations SA1: Social fright 0.68 1.02 Felt anxious, worried, or nervous about social situations SA2: Social tension 0.77 1.06 Had thoughts of being rejected, humiliated, embarrassed, ridiculed, or offending others SA3: Humiliation 0.61 0.95 Felt a racing heart, sweaty, trouble breathing, faint, or shaky in social situations SA4: Social bumping heart 0.48 0.89 Felt tense muscles, felt on edge or restless, or had trouble relaxing in social situations SA5: Social agitation 0.54 0.95 Avoided, or did not approach or enter, social situations SA6: Social avoidance 0.66 1.07 Left social situations early or participated only minimally (e.g. said little, avoided eye contact) SA7: Social involvement 0.75 1.13 Spent a lot of time preparing what to say or how to act in social situations SA8: Social repetition 0.63 1.03 Distracted myself to avoid thinking about social situations SA9: Distraction 0.66 1.05 Needed help to cope with social situations (e.g. alcohol or medications, superstitious objects) SA10: Seek social help 0.29 0.76 Note. SD: Standard Deviation. All symptom scores start at 0. Graphical LASSO network The structure of the network is shown in Figure 1. The node expected influence of the network is shown in Figure 2a. Social anxiety symptoms SA2 (Social tension) and SA6 (Social avoidance), and depression symptom D6 (Failure) demonstrated the highest expected influences, indicating that these three symptoms have the closest association in the network. The CS-coefficient of the node expected influence was 0.75, indicating that the estimation of the node expected influence is adequately stable (Figure S1 in the Supplemental Material). The bootstrapped difference test for node expected influences is shown in Figure S2 in the Supplemental Material. Figure 1. Network structure of social anxiety and depression symptoms. Blue edges represent positive correlations, and red edges represent negative correlations. The thickness of the edge reflects the magnitude of the correlation. The cut-off value is set to be .05. See Table 2 for label names. Figure 2. Network centrality plot of social anxiety and depression symptoms: (a) depicts the expected influence of variables selected in the present network (z-score); (b) depicts the bridge expected influence of variables selected in the present network (z-score). The highest nodes expected influence and bridge expected influences are thickened and highlighted in red. See Table 2 for label names. The node bridge expected influence of the network is shown in Figure 2b. Social anxiety symptom SA3 (Humiliation), and depression symptom D8 (Motor) demonstrated the highest bridge expected influences, indicating that these two symptoms are the bridge symptoms connecting the two symptom communities. Thus, these two symptoms may have the strongest ability to activate/deactivate the co-occurrence of depression and social anxiety symptoms. The CS-coefficient of the node bridge expected influence was 0.75, indicating that the estimation of the node bridge expected influence is adequately stable (Figure S3 in the Supplemental Material). The bootstrapped difference test for node bridge expected influences is shown in Figure S4 in the Supplemental Material. Directed acyclic graph Figure 3a shows the importance of each edge to the overall DAG structure (edge thickness represents the change in the BIC when that edge is removed from the DAG). A thicker edge means that it is more crucial to model fit. The most important edge of DAG structure was SA1-SA2 (change in the BIC value: −803.52) and SA1-SA4 (change in the BIC value: −678.48). The least important edge was D6-D3 (change in the BIC value: −7.58) and D8-SA10 (change in the BIC value: −8.80). The change in BIC values for each edge can be found in Table S1 in Supplemental Materials. Figure 3b shows the directional probability of each edge. A thicker edge means the current direction is in a greater proportion of the bootstrapped DAGs. The thickest edge connected D6 to D8 (0.971; i.e. this edge was pointing in that direction in 9,710 of the 10,000 bootstrapped DAGs). The exact directional probability for each edge in Figure 3b can be found in Table S1 in the Supplemental Material. Structurally, social anxiety symptom SA1 (Social fright) arises at the upstream of the whole DAG, directly influencing the other symptoms of social anxiety and depression (i.e. SA2 [Social tension], SA4 [Social bumping heart], and D1[Sadness]). In addition, depression symptom D1 (Sadness) arises at the upstream of the depression symptoms. Figure 3. Directed acyclic graph (DAG) for symptoms of social anxiety and depression: (a) edge thickness represents the importance of that edge to the overall DAG structure; (b) edge thickness represents the directional probability. See Table 2 for label names. Discussion The left-behind children are at a high risk of developing psychological problems after the COVID-19 lockdown is lifted (Racine et al., 2022). We found that the prevalence of depression in left-behind children was 19.57%, which was much higher than that of a previous study in non-left-behind children (12.33%) (Liu et al., 2021). Moreover, the prevalence of social anxiety among left-behind children after the lifting of the lockdown was 12.36%, yet a previous study suggested that its prevalence was approximately 9% (Burstein et al., 2011). The comorbidity rate of the two diseases in the left-behind children was 8.98%, higher than that in non-left-behind children (4.5%) from a previous study (Klemanski et al., 2017). Previous evidence suggests that individuals with social anxiety disorder in clinical samples had a 30 to 70% probability of developing depression, while individuals with depression had a 15% to 27% probability of developing social anxiety in community and clinical samples (Adams et al., 2016). Our results showed higher comorbidity rates of 72.66% and 45.89%, respectively. These results suggest that the mental health status of left-behind children may be affected after lockdown removal. However, few related studies have been conducted, hence it is of great importance to further study the relationship between depression and social anxiety. It is also important to study the network structure of social anxiety and depression symptoms in left-behind children during the pandemic. However, many network analysis studies only explored the relation between depression and anxiety symptoms in adolescents during the COVID-19 pandemic (Bai et al., 2021; Liu et al., 2022), and the relation between depression and social anxiety remains unclear in left-behind children after the lifting of the lockdown. The only two network analysis studies of social anxiety and depression both used clinical samples (Heeren et al., 2018; Langer et al., 2019). To our knowledge, this is the first study on the network structure of social anxiety and depression symptoms in left-behind children during the pandemic. Our findings would contribute to the prevention and intervention of social anxiety and depression in left-behind children after the COVID-19 lockdown is lifted. SA2 (Social tension) and SA6 (Social avoidance) symptoms were the most prominent symptoms in the social anxiety and depression network of left-behind children after the lifting of the lockdown, consistent with previous studies (Heeren & McNally, 2018; Heeren et al., 2018; Langer et al., 2019) which reported that avoidance and fear of social scenarios are core nodes in the network. These results were in line with our expectations that social isolation may make people socially disconnected (Santini et al., 2020). They need to face more social interaction scenarios after the lockdown is lifted (Lim et al., 2022), which may lead to maladaptation. According to previous theories, problematic beliefs about the social world (e.g. “If people know I am anxious, they will think I am weak”) affect the appraisal of the social situation (e.g. appraising the social situations as more threatening than they actually are), creating anxiety and thus inducing avoidance of threatening situations and causing more generalized social anxiety (Heeren et al., 2020). Furthermore, social situations almost always evoke anxiousness or fear, and individuals typically respond in an avoidant manner (e.g. diverting attention, refusing to participate in activities, or refusing to go to school) according to the DSM-5 diagnostic criteria for social anxiety (American Psychiatric Association, 2013a). These symptoms hence play an important role in maintaining SAD. Besides, we found that SA3 (Humiliation) and D8 (Motor) were bridge symptom nodes in the network. Fear or anxiousness in peer situations is a necessary prerequisite for social anxiety according to DSM-5, as individuals in this situation fear being judged negatively by others (Heimberg et al., 2014), causing impaired self-esteem. Previous studies have shown that left-behind children have lower self-esteem compared with non-left-behind children of the same age (Dai & Chu, 2018), and lower self-esteem is associated with a greater impact of family functioning on their prosocial behavior (Gao et al., 2019). In other words, left-behind children may show lower levels of self-esteem due to impaired family function, resulting in low prosocial behaviors, which in turn cause more pronounced social anxiety symptoms. Experiencing SA3 (Humiliation) may affect motor levels of the individual, and motor often acts as a bridge symptom in comorbidity networks, consistent with the depression and anxiety network in clinical samples (Kaiser et al., 2021). These results suggest that intervention against SA3 (Humiliation) may prevent the progression of social anxiety to depression, and similarly, intervention against D8 (Motor) may prevent the progression of depression symptoms to social anxiety. Interestingly, social anxiety, depression and loneliness are positively correlated with each other (Danneel et al., 2020). Studies during the pandemic have shown that social isolation increases loneliness, stress, and fear, thereby increasing the risk of depression and anxiety (Courtney et al., 2020; Guessoum et al., 2020; Liu et al., 2022). But these findings may be more applicable to the general population, because left-behind children are more adaptable to the loneliness caused by social isolation for they have been separated from their parents for a long time. Previous research also found that left-behind children during the COVID-19 pandemic experienced lower levels of loneliness than non-left-behind children, yet higher levels of depression and anxiety (Wang et al., 2021). Therefore, we believe that loneliness during COVID-19 may not be the most important factor causing depression and anxiety in left-behind children. In contrast, left-behind children need to face more social situations after the lifting of the lockdown, mostly in the school environment (Morrissette, 2021), which may result in maladaptation and increase the risk of social anxiety. To further investigate the relations between the symptoms of the two disorders, we used the DAG to explore the predictive (and potentially causal) priority of these symptoms. Although the current study is a cross-sectional survey that ignores the effect of time, the DAG is able to provide some hypotheses on admissible causal relations (Briganti et al., 2022; Tennant et al., 2021). We found that SA1 (social fright) was at the upstream of all symptoms (social anxiety and depression symptoms), implying that downstream symptoms were dependent on SA1 symptom (McNally, Mair et al., 2017). This means that if left-behind children experience SA1 (social fright), they are more likely to experience symptoms such as SA2 (Social tension), SA4 (Social bumping heart) and D1 (Sadness) than vice versa (McNally, Mair et al., 2017). These results suggest that left-behind children are more likely to develop depression symptoms while they experience social anxiety symptoms of different levels (i.e., social fright or social tension) (Nordahl et al., 2018). This is consistent with the previous findings that when social anxiety co-occurs with depression, social anxiety always precedes depression (Stein et al., 2001). In depression symptoms, D1 (Sadness) was at the upstream of all depression symptoms. Studies have shown that 70% of individuals conceptualize unprovoked grief as a mental illness, especially depression, so we need to pay attention to unprovoked grief experienced by left-behind children, which may be a prerequisite for depression. Interestingly, we found that depression symptoms are not at the lowest level of the DAG (e.g. D1, D6, and D9 have higher levels than SA3, SA9, and SA8), and are ultimately connected to SA10 (Seek social help; the lowest level of the DAG), indicating that depression has high comorbidity with social anxiety. This implies that depressed individuals may seek external means (such as alcohol or drugs) to alleviate symptoms, which may explain why alcohol addiction is present in both social anxiety and depressed adolescents (H. Blumenthal et al.,2015; H. Blumenthal et al., 2010; Johannessen et al., 2017 ). In short, narrowing the range of symptoms caused by social anxiety may block the risk of developing other comorbidities, and special attention should be paid to these upstream symptoms, which may play a key role in preventing social anxiety and depression symptoms in left-behind children. Strengths and limitations For the first time to our knowledge, our study explored the network structure of social anxiety and depression symptoms network in left-behind children after the lifting of the COVID-19 lockdown. We analyzed core and bridge symptoms of social anxiety and depression via the Graphical LASSO network and the predictive (and potentially causal) priority of these symptoms via DAG. The large sample size ensures the relative reliability of the results, and in our study, the stability indicators of the final symptoms network are all excellent. The findings may provide guidance for mental illness prevention and intervention of 900 million left-behind children worldwide during the pandemic (BBC, 2015). Special attention should be paid to left-behind children in low- and medium-income countries, who are more likely to be affected by the pandemic (Barros et al., 2020; Sharpe et al., 2021). Our study has several limitations. First, the data we used were all derived from a cross-sectional survey, therefore dynamic interpretation of symptoms was not possible and the results may be applicable to specific groups. However, the results found by the network analysis may play an important role in preventing the social anxiety and depression symptoms of left-behind children after the lifting of the COVID-19 lockdown. Nonetheless, further follow-up studies are still needed in the future to determine how the symptoms faced by left-behind children in the context of COVID-19 evolve as the pandemic develops. Second, although we used the DAG network to explore the predictive (and potentially causal) priority of these symptoms, we could not confirm the causal relationship due to the limitation of a cross-sectional design. Third, we cannot know whether the left-behind children also had a higher social anxiety level during the lockdown period because we did not obtain data during the lockdown period. Whether the level of social anxiety has changed due to isolation is worthy of further follow-up research. Fourth, it remains unclear whether our findings are applicable to groups with different cultural backgrounds, especially people from Western countries, leaving room for further research. Finally, it should be noted that our survey lasted from October 2020 to November 2020, which is the early stage of the COVID-19 pandemic. Therefore, the interpretation and application of our study results need to take the environmental changes into consideration. For example, under the current dynamic zero-COVID policy, a small-scale short-duration lockdown may be implemented. It is unclear if our results are applicable to this circumstance. We suggest that future studies pay attention to the development of the pandemic and the changes in policies and their dynamic effect on the mental health of left-behind children. Conclusion This study explored the network structure of social anxiety and depression symptoms for the first time in left-behind children after the lifting of the COVID-19 lockdown. In addition to the high prevalence of depression and social anxiety among left-behind children, the comorbidity rates of the two disorders are also quite high. The results of network analysis found that symptoms SA2 (Social tension) and SA6 (Social avoidance) had the highest expected influence; SA3 (Humiliation) and D8 (Motor) were bridge symptoms in the network; and SA1 (Social fright) was also identified as a key priority symptom across the DAG because it was at the upstream of all symptoms. The findings show that special attention should be paid to the prevention and intervention of social anxiety and depression symptoms in left-behind children after the lifting of the lockdown, which may help to reduce social anxiety, depression and their comorbidity. Supplemental Material sj-docx-1-isp-10.1177_00207640221141784 – Supplemental material for Social anxiety and depression symptoms in Chinese left-behind children after the lifting of COVID-19 lockdown: A network analysis Click here for additional data file. Supplemental material, sj-docx-1-isp-10.1177_00207640221141784 for Social anxiety and depression symptoms in Chinese left-behind children after the lifting of COVID-19 lockdown: A network analysis by Kuiliang Li, Lei Ren, Lei Zhang, Chang Liu, Mengxue Zhao, Xiaoqing Zhan, Ling Li, Xi Luo and Zhengzhi Feng in International Journal of Social Psychiatry We thank the boys and girls for participating in the study and Nan Liu and Ting Chen for collecting data. Author contribution: Data acquisition: Kuiliang Li, Mengxue Zhao, Ling Li Formal analysis: Kuiliang Li, Lei Ren Writing: Kuiliang Li, Lei Ren, Lei Zhang, Chang Liu, Xi Luo, Zhengzhi Feng Review and editing: Kuiliang Li, Lei Ren, Lei Zhang, Xiaoqing Zhan, Xi Luo, Zhengzhi Feng. Each author signed a form for disclosure of potential conflicts of interest. None of the authors reported any financial or other conflicts of interest concerning the work described. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the National Natural Science Foundation of China (NSFC: 81971278). The fund was used for software development in the research to provide survey results reports for survey participants and also played a role in data analysis. Ethics statement: This study has been reviewed and approved by the Medical Ethics Committee of the Department of Medical Psychology, Army Medical University (Project No. CWS20J007). Participants were aware of the informed consent before participation in this study, were informed that the survey was anonymous, and were assured that personal information would not be disclosed. ORCID iDs: Kuiliang Li https://orcid.org/0000-0002-1653-5419 Chang Liu https://orcid.org/0000-0003-0324-8151 Xi Luo https://orcid.org/0000-0003-2361-7348 Supplemental material: Supplemental material for this article is available online. ==== Refs References Abbasi J. (2020). Prioritizing physician mental health as COVID-19 marches on. JAMA – Journal of the American Medical Association, 323 , 2235–2236. 10.1001/jama.2020.5205 32432665 Adams G. C. Balbuena L. Meng X. Asmundson G. J. (2016). When social anxiety and depression go together: A population study of comorbidity and associated consequences. Journal of Affective Disorders, 206 , 48–54. 10.1016/j.jad.2016.07.031 27466742 Adams G. C. Wrath A. J. Mondal P. Asmundson G. J. G. (2018). Depression with or without comorbid social anxiety: Is attachment the culprit? Psychiatry Research, 269 , 86–92. 10.1016/j.psychres.2018.08.037 30145307 Almeida M. Shrestha A. D. Stojanac D. Miller L. J. (2020). The impact of the COVID-19 pandemic on women’s mental health. Archives of Women’s Mental Health, 23 , 741–748. 10.1007/s00737-020-01092-2 American Psychiatric Association. (2013a). Diagnostic and Statistical Manual of Mental Disorders (5th ed.). Author. 10.1176/appi.books.9780890425596. American Psychiatric Association. (2013b). DSM-5-TR Online assessment measures. APA. Retrieved January 10, 2022, from https://www.psychiatry.org/psychiatrists/practice/dsm/educational-resources/assessment-measures Bai W. Xi H. T. Zhu Q. Ji M. Zhang H. Yang B. X. Cai H. Liu R. Zhao Y. J. Chen L. Ge Z. M. Wang Z. Han L. Chen P. Liu S. Cheung T. Tang Y. L. Jackson T. An F. Xiang Y. T. (2021). Network analysis of anxiety and depressive symptoms among nursing students during the COVID-19 pandemic. Journal of Affective Disorders, 294 , 753–760. 10.1016/j.jad.2021.07.072 34375200 Barros A. J. D. Wehrmeister F. C. Ferreira L. Z. Vidaletti L. P. Hosseinpoor A. R. Victora C. G. (2020). Are the poorest poor being left behind? Estimating global inequalities in reproductive, maternal, newborn and child health. BMJ Global Health, 5 (1 ), e002229. 10.1136/bmjgh-2019-002229 BBC. (2015). Who are the “left behind children” around the world? Retrieved January 9, 2022, from https://www.bbc.com/news/av/world-33996631 Bernstein E. E. Heeren A. McNally R. J. (2017). Unpacking rumination and executive control: A network perspective. Clinical Psychological Science, 5 (5 ), 816–826. 10.1177/2167702617702717 Blanchard M. A. Roskam I. Mikolajczak M. Heeren A. (2021). A network approach to parental burnout. Child Abuse & Neglect, 111 , 1–12. 10.1016/j.chiabu.2020.104826 Blumenthal H. Ham L. S. Cloutier R. M. Bacon A. K. Douglas M. E. (2015). Social anxiety, disengagement coping, and alcohol-use behaviors among adolescents. Anxiety, Stress and Coping, 29 (4 ), 1–15. 10.1080/10615806.2015.1058366 25616421 Blumenthal H. Leen-Feldner E. W. Frala J. L. Badour C. L. Ham L. S. (2010). Social Anxiety and Motives for Alcohol Use Among Adolescents. Psychology of Addictive Behaviors, 24 (3 ), 529–534. 10.1037/a0019794 20853939 Briganti G. Scutari M. McNally R. J. (2022). A tutorial on Bayesian networks for psychopathology researchers. Psychological Methods. Advance online publication. 10.1037/met0000479 Burstein M. He J. P. Kattan G. Albano A. M. Avenevoli S. Merikangas K. R. (2011). Social phobia and subtypes in the National Comorbidity Survey-adolescent supplement: Prevalence, correlates, and comorbidity. Journal of the American Academy of Child and Adolescent Psychiatry, 50 (9 ), 870–880. 10.1016/j.jaac.2011.06.005 21871369 Chen T. Wu D. Chen H. Yan W. Yang D. Chen G. Ma K. Xu D. Yu H. Wang H. Wang T. Guo W. Chen J. Ding C. Zhang X. Huang J. Han M. Li S. Luo X. . . . Ning Q. (2020). Clinical characteristics of 113 deceased patients with coronavirus disease 2019: Retrospective study. BMJ, 368, 1–12 . 10.1136/bmj.m1091 Courtney D. Watson P. Battaglia M. Mulsant B. H. Szatmari P. (2020). COVID-19 impacts on child and youth anxiety and depression: Challenges and opportunities. The Canadian Journal of Psychiatry, 65 (10 ), 688–691. 10.1177/0706743720935646 32567353 Dai Q. Chu R. X. (2018). Anxiety, happiness and self-esteem of western Chinese left-behind children. Child Abuse & Neglect, 86 , 403–413. 10.1016/j.chiabu.2016.08.002 27568066 Danneel S. Geukens F. Maes M. Bastin M. Bijttebier P. Colpin H. Verschueren K. Goossens L. (2020). Loneliness, social anxiety symptoms, and depressive symptoms in adolescence: Longitudinal distinctiveness and correlated change. Journal of Youth and Adolescence, 49 (11 ), 2246–2264. 10.1007/s10964-020-01315-w 32918664 El Hayek S. Cheaito M. A. Nofal M. Abdelrahman D. Adra A. Al Shamli S. AlHarthi M. AlNuaimi N. Aroui C. Bensid L. Emberish A. M. Larnaout A. Radwan A. Slaih M. Al Sinawi H . (2020). Geriatric Mental Health and COVID-19: An Eye-opener to the situation of the Arab countries in the Middle East and North Africa Region. American Journal of Geriatric Psychiatry, 28 , 1058–1069. 10.1016/j.jagp.2020.05.009 Epskamp S. Borsboom D. Fried E. I. (2018). Estimating psychological networks and their accuracy: A tutorial paper. Behavior Research Methods, 50 (1 ), 195–212. 10.3758/s13428-017-0862-1 28342071 Epskamp S. Cramer A. O. J. Waldorp L. J. Schmittmann V. D. Borsboom D. (2012). Qgraph: Network visualizations of relationships in psychometric data. Journal of Statistical Software, 48 (4 ), 1–18. 10.18637/jss.v048.i04 Epskamp S. Fried E. I. (2018). A tutorial on regularized partial correlation networks. Psychological Methods, 23 (4 ), 617–634. 10.1037/met0000167 29595293 Epskamp S. Waldorp L. J. Mõttus R. Borsboom D. (2018). The Gaussian graphical model in cross-sectional and time-series data. Multivariate Behavioral Research, 53 (4 ), 453–480. 10.1080/00273171.2018.1454823 29658809 Fellmeth G. Rose-Clarke K. Zhao C. Busert L. K. Zheng Y. Massazza A. Sonmez H. Eder B. Blewitt A. Lertgrai W. Orcutt M. Ricci K. Mohamed-Ahmed O. Burns R. Knipe D. Hargreaves S. Hesketh T. Opondo C. Devakumar D. (2018). Health impacts of parental migration on left-behind children and adolescents: A systematic review and meta-analysis. Lancet, 392 (10164 ), 2567–2582. 10.1016/S0140-6736(18)32558-3 30528471 Foygel R. Drton M. (2010). Extended Bayesian Information Criteria for Gaussian Graphical Models. Advances in Neural Information Processing Systems, 23 , 604–612. Friedman J. Hastie T. Tibshirani R. (2008). Sparse inverse covariance estimation with the graphical lasso. Biostatistics, 9 (3 ), 432–441. 10.1093/biostatistics/kxm045 18079126 Fruchterman T. M. J. Reingold E. M. (1991). Graph drawing by force-directed placement. Software Practice and Experience, 21 (11 ), 1129–1164. 10.1002/spe.4380211102 Gao F. Yao Y. Yao C. Xiong Y. Ma H. Liu H. (2019). The status of pro-social tendency of left-behind adolescents in China: How family function and self-esteem affect pro-social tendencies. Frontiers in Psychology, 10 , 1–13. 10.3389/fpsyg.2019.01202.30713512 Guessoum S. B. Lachal J. Radjack R. Carretier E. Minassian S. Benoit L. Moro M. R. (2020). Adolescent psychiatric disorders during the COVID-19 pandemic and lockdown. Psychiatry Research, 291 , 113264. 10.1016/j.psychres.2020.113264 32622172 Heeren A. Bernstein E. E. McNally R. J. (2020). Bridging maladaptive social self-beliefs and social anxiety: A network perspective. Journal of Anxiety Disorders, 74 , 1–9. 10.1016/j.janxdis.2020.102267 Heeren A. Jones P. J. McNally R. J. (2018). Mapping network connectivity among symptoms of social anxiety and comorbid depression in people with social anxiety disorder. Journal of Affective Disorders, 228 , 75–82. 10.1016/j.jad.2017.12.003 29232567 Heeren A. McNally R. J. (2018). Social Anxiety Disorder as a densely interconnected network of fear and avoidance for social situations. Cognitive Therapy and Research, 42 (1 ), 103–113. 10.1007/s10608-017-9876-3 Heimberg R. G. Hofmann S. G. Liebowitz M. R. Schneier F. R. Smits J. A. Stein M. B. Hinton D. E. Craske M. G. (2014). Social anxiety disorder in DSM-5. Depression and Anxiety, 31 , 472–479. 10.1002/da.22231 24395386 Holmes E. A. O’Connor R. C. Perry V. H. Tracey I. Wessely S. Arseneault L. Ballard C. Christensen H. Cohen Silver R. Everall I. Ford T. John A. Kabir T. King K. Madan I. Michie S. Przybylski A. K. Shafran R. Sweeney A. . . . Bullmore E. (2020). Multidisciplinary research priorities for the COVID-19 pandemic: A call for action for mental health science. The Lancet Psychiatry, 7 , 547–560. 10.1016/s2215-0366(20)30168-1 32304649 Hou T. Mao X. Shao X. Liu F. Dong W. Cai W. (2021). Suicidality and its associated factors among students in rural China during COVID-19 pandemic: A Comparative Study of left-behind and Non-Left-Behind Children. Frontiers in Psychiatry, 12 , 708305. 10.3389/fpsyt.2021.708305 34413801 Johannessen E. L. Andersson H. W. Bjørngaard J. H. Pape K. (2017). Anxiety and depression symptoms and alcohol use among adolescents–across sectional study of Norwegian secondary school students. BMC Public Health, 17 (1 ), 1–9. 10.1186/s12889-017-4389-2 28049454 Jones P. J. Ma R. McNally R. J. (2021). Bridge Centrality: A network approach to understanding comorbidity. Multivariate Behavioral Research, 56 (2 ), 353–367. 10.1080/00273171.2019.1614898 31179765 Kaiser T. Herzog P. Voderholzer U. Brakemeier E. (2021). Unraveling the comorbidity of depression and anxiety in a large inpatient sample: Network analysis to examine bridge symptoms. Depression and Anxiety, 38 (3 ), 307–317. 10.1002/da.23136 33465284 Klemanski D. H. Curtiss J. McLaughlin K. A. Nolen-Hoeksema S. (2017). Emotion Regulation and the transdiagnostic role of repetitive negative thinking in adolescents with social anxiety and depression. Cognitive Therapy and Research, 41 (2 ), 206–219. 10.1007/s10608-016-9817-6 28579659 Langer J. K. Tonge N. A. Piccirillo M. Rodebaugh T. L. Thompson R. J. Gotlib I. H. (2019). Symptoms of social anxiety disorder and major depressive disorder: A network perspective. Journal of Affective Disorders, 243 , 531–538. 10.1016/j.jad.2018.09.078 30292147 Lim M. H. Qualter P. Thurston L. Eres R. Hennessey A. Holt-Lunstad J. Lambert G. W. (2022). A Global Longitudinal Study examining social restrictions severity on loneliness, social anxiety, and depression. Frontiers in Psychiatry, 13 , 818030. 10.3389/fpsyt.2022.818030 35418888 Liu R. Chen X. Qi H. Feng Y. Su Z. Cheung T. Jackson T. Lei H. Zhang L. Xiang Y. T. (2022). Network analysis of depressive and anxiety symptoms in adolescents during and after the COVID-19 outbreak peak. Journal of Affective Disorders, 301 , 463–471. 10.1016/j.jad.2021.12.137 34995705 Liu Y. Yue S. Hu X. Zhu J. Wu Z. Wang J. Wu Y. (2021). Associations between feelings/behaviors during COVID-19 pandemic lockdown and depression/anxiety after lockdown in a sample of Chinese children and adolescents. Journal of Affective Disorders, 284 , 98–103. 10.1016/j.jad.2021.02.001 33592433 Loades M. E. Chatburn E. Higson-Sweeney N. Reynolds S. Shafran R. Brigden A. Linney C. McManus M. N. Borwick C. Crawley E. (2020). Rapid systematic review: The impact of social isolation and loneliness on the mental health of children and adolescents in the context of COVID-19. Journal of the American Academy of Child and Adolescent Psychiatry, 59 (11 ), 1218–1239.e3. 10.1016/j.jaac.2020.05.009 32504808 McNally R. J. (2016). Can network analysis transform psychopathology? Behaviour Research and Therapy, 86 , 95–104. 10.1016/j.brat.2016.06.006 27424882 McNally R. J. Heeren A. Robinaugh D. J. (2017). A Bayesian network analysis of posttraumatic stress disorder symptoms in adults reporting childhood sexual abuse. European Journal of Psychotraumatology, 8 , 1341276. 10.1080/20008198.2017.1341276 29038690 McNally R. J. Mair P. Mugno B. L. Riemann B. C. (2017). Co-morbid obsessive-compulsive disorder and depression: A Bayesian network approach. Psychological Medicine, 47 (7 ), 1204–1214. 10.1017/S0033291716003287 28052778 Meilihua Health Systems. (2019). DSM-5 Assessment Scale (Chinese version). Retrieved August 30, 2020, from https://www.mhealthu.com/index.php/list_liangbiao/120/267?page=2 Moffa G. Catone G. Kuipers J. Kuipers E. Freeman D. Marwaha S. Lennox B. R. Broome M. R. Bebbington P. (2017). Using directed acyclic graphs in epidemiological research in psychosis: An Analysis of the role of bullying in psychosis. Schizophrenia Bulletin, 43 (6 ), 1273–1279. 10.1093/schbul/sbx013 28521005 Morrissette M. (2021). School closures and social anxiety during the COVID-19 Pandemic. Journal of the American Academy of Child and Adolescent Psychiatry, 60 , 6–7. 10.1016/j.jaac.2020.08.436 32890669 Nordahl H. Nordahl H. M. Vogel P. A. Wells A. (2018). Explaining depression symptoms in patients with social anxiety disorder: Do maladaptive metacognitive beliefs play a role? Clinical Psychology & Psychotherapy, 25 (3 ), 457–464. 10.1002/cpp.2181 29493054 Racine N. Korczak D. J. Madigan S. (2022). Evidence suggests children are being left behind in COVID-19 mental health research. European Child & Adolescent Psychiatry, 31 (9 ), 1479–1480. 10.1007/s00787-020-01672-8 33151431 Ren L. Wang Y. Wu L. Wei Z. Cui L. B. Wei X. Hu X. Peng J. Jin Y. Li F. Yang Q. Liu X. (2021). Network structure of depression and anxiety symptoms in Chinese female nursing students. BMC Psychiatry, 21 (1 ), 279–312. 10.1186/s12888-021-03276-1 34059013 Samantaray N. N. Kar N. Mishra S. R. (2022). A follow-up study on treatment effects of cognitive-behavioral therapy on social anxiety disorder: Impact of COVID-19 fear during post-lockdown period. Psychiatry Research, 310 , 114439. 10.1016/j.psychres.2022.114439 Santini Z. I. Jose P. E. York Cornwell E. Koyanagi A. Nielsen L. Hinrichsen C. Meilstrup C. Madsen K. R. Koushede V. (2020). Social disconnectedness, perceived isolation, and symptoms of depression and anxiety among older Americans (NSHAP): A longitudinal mediation analysis. The Lancet Public Health, 5 (1 ), e62–e70. 10.1016/S2468-2667(19)30230-0 Scutari M. (2010). Learning Bayesian networks with the bnlearn RPackage. Journal of Statistical Software, 35 (3 ), 1–22. 10.18637/jss.v035.i03 Scutari M. Nagarajan R. (2013). Identifying significant edges in graphical models of molecular networks. Artificial Intelligence in Medicine, 57 (3 ), 207–217. 10.1016/j.artmed.2012.12.006 23395009 Sharpe D. Rajabi M. Chileshe C. Joseph S. M. Sesay I. Williams J. Sait S. (2021). Mental health and wellbeing implications of the COVID-19 quarantine for disabled and disadvantaged children and young people: Evidence from a cross-cultural study in Zambia and Sierra Leone. BMC Psychology, 9 (1 ), 1–15. 10.1186/s40359-021-00583-w 33388086 Singh S. Roy D. Sinha K. Parveen S. Sharma G. Joshi G. (2020). Impact of COVID-19 and lockdown on mental health of children and adolescents: A narrative review with recommendations. Psychiatry Research, 293 , 113429. 10.1016/j.psychres.2020.113429 32882598 Stein M. B. Fuetsch M. Müller N. Höfler M. Lieb R. Wittchen H. U. (2001). Social anxiety disorder and the risk of depression: A prospective community study of adolescents and young adults. Archives of General Psychiatry, 58 (3 ), 251. 10.1001/archpsyc.58.3.251 11231832 Tennant P. W. G. Murray E. J. Arnold K. F. Berrie L. Fox M. P. Gadd S. C. Harrison W. J. Keeble C. Ranker L. R. Textor J. Tomova G. D. Gilthorpe M. S. Ellison G. T. H. (2021). Use of directed acyclic graphs (DAGs) to identify confounders in applied health research: Review and recommendations. International Journal of Epidemiology, 50 (2 ), 620–632. 10.1093/ije/dyaa213 33330936 Walton M. Murray E. Christian M. D. (2020). Mental health care for medical staff and affiliated healthcare workers during the COVID-19 pandemic. European Heart Journal Acute Cardiovascular Care, 9 (3 ), 241–247. 10.1177/2048872620922795 32342698 Wang C. Pan R. Wan X. Tan Y. Xu L. Ho C. S. Ho R. C. (2020). Immediate psychological responses and associated factors during the initial stage of the 2019 coronavirus disease (COVID-19) epidemic among the general population in China. International Journal of Environmental Research and Public Health, 17 (5 ), 1–25. 10.3390/ijerph17051729 Wang F. Lin L. Xu M. Li L. Lu J. Zhou X. (2019). Mental health among left-behind children in rural China in relation to parent-child communication. International Journal of Environmental Research and Public Health, 16 (10 ), 1–10. 10.3390/ijerph16101855 Wang X. Hegde S. Son C. Keller B. Smith A. Sasangohar F. (2020). Investigating mental health of US college students during the COVID-19 pandemic: Cross-sectional survey study. Journal of Medical Internet Research, 22 (9 ), e22817. 10.2196/22817 Wang Y. Liu W. Wang W. Lin S. Lin D. Wang H. (2021). Left-behind children’s social adjustment and relationship with parental coping with children’s negative emotions during the COVID-19 pandemic in China. International Journal of Psychology, 56 (4 ), 512–521. 10.1002/ijop.12754 33739446 WHO. (2022, November 5). Mental health. Retrieved March 5, 2022, from Word Health Organization website: https://www.who.int/health-topics/mental-health#tab=tab_2 Yuan P. Wang L. (2016). Migrant workers: China boom leaves children behind. Nature, 529 (7584 ), 25. 10.1038/529025a Zhu F. C. Li Y. H. Guan X. H. Hou L. H. Wang W. J. Li J. X. Wu S. P. Wang B. S. Wang Z. Wang L. Jia S. Y. Jiang H. D. Wang L. Jiang T. Hu Y. Gou J. B. Xu S. B. Xu J. J. Wang X. W. . . ., Chen W. (2020). Safety, tolerability, and immunogenicity of a recombinant adenovirus type-5 vectored COVID-19 vaccine: A dose-escalation, open-label, non-randomised, first-in-human trial. Lancet, 395 (10240 ), 1845–1854. 10.1016/s0140-6736(20)31208-3 32450106

36511138

PMC9749067

NO-CC CODE

2022-12-15 23:22:50

no

Int J Soc Psychiatry. 2022 Dec 13;:00207640221141784

utf-8

Int J Soc Psychiatry

10.1177/00207640221141784

oa_other

==== Front Chem Eng Sci Chem Eng Sci Chemical Engineering Science 0009-2509 1873-4405 Elsevier Ltd. S0009-2509(21)00314-6 10.1016/j.ces.2021.116749 116749 Article Synthesis of a novel anti-fog and high-transparent coating with high wear resistance inspired by dry rice fields Xiang Juan Liu Xiaoying 1 Liu Yan ⁎ Wang Lilin He Yan Luo Ling Yang Gang Zhang Xiaohong Huang Chengyi Zhang Yanzong ⁎ College of Environment Sciences, Sichuan Agricultural University, Chengdu 611130, China ⁎ Corresponding authors. 1 These authors contributed equally to this work (co-first authors). 8 5 2021 12 10 2021 8 5 2021 242 116749116749 9 2 2021 30 4 2021 5 5 2021 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Graphical abstract During the outbreak of COVID-19, the fogging of goggles was a fatal problem for doctors. At present, there are many ways to prevent fogging by adjusting surface wettability. However, the mechanical properties of most super-hydrophilic antifogging coatings are poor, easy to lose their antifogging properties when encountering fingers or cloth friction. To address this issue, the Konjac Glucomannan was cross-linked with water-soluble silicone fluid to form a binder, then being combined with the modified Ecokimera to prepare an eco-friendly super-hydrophilic coating that possessed excellent super-hydrophilicity, and the water contact angle (WCA) was 2.51 ± 1°. In addition, the WCA is still about 5° after 180 times of antifogging tests. The friction resistance of the coating was as high as 24 m. Moreover, the light transmittance was only reduced by 3%. Besides, they also had the excellent self-cleaning property. After being stored in the laboratory environment for 90 days, it can still maintain the hydrophilic property (WCA is about 5°). In general, the method proposed in this study is low-cost and eco-friendly, and can be widely used in the preparation of antifogging coatings. Keywords Super-hydrophilic Antifogging High wear resistance Anti-fouling Self-cleaning ==== Body pmcNomenclature Abbreviation WCA water contact angle KGM konjac glucomannan HEA hydroxyethyl acrylate TEOS tetraethyl orthosilicate PVA poly vinyl alcohol PAA poly acrylic acid ECO ecokimera DC193 water-soluble silicone fluid TEOA/TEA triethanolamine DI distilled water KOT KGM-DC193-TEOA m-ECO modified ecokimera E-KOT modified ecokimera-KOT Roman symbols rsd component of dispersion force rsh polar force component θ contact angle of water rlv surface tension of water 1 Introduction Medical staff needed to take protective measures to avoid virus infection during COVID-19, and goggles were one of the necessary tools (Lemarteleur et al., 2020, Oldfield and Malwal, 2020). However, the fogging of goggles was a fatal problem in medical work (Bai et al., 2020, Feng et al., 2020, Hu et al., 2020, Sun et al., 2020). Besides, from a security standpoint, the antifogging property was also indispensable in glasses, windshields, etc. (Durán and Laroche, 2019a, Liu and Locklin, 2018, Sun et al., 2020, Tao et al., 2018, Yu et al., 2020). In this case, various antifogging strategies have been proposed by adjusting the surface wettability. For example, the superhydrophobic anti-fog coating could be prepared by airless spray and crystal growth methods (Zhang et al., 2017). Due to the low adhesion of the superhydrophobic surfaces, this strategy requires complex surface structures to be implemented (Liu and Kim, 2014, Wen et al., 2014). Furthermore, it is also technically challenging to produce transparent superhydrophobic coatings (He et al., 2018). So far, constructing super-hydrophilic surfaces seems to be the most promising antifogging strategy (Nam et al., 2017, Walker et al., 2019, Wang et al., 2019a). When the substrate is in the super-hydrophilic state, water droplets spread out rapidly on the surface to form a liquid film layer (Feng et al., 2020, Hsu et al., 2018). As the liquid film has a large surface area and can quickly evaporate, an antifogging effect is produced (Durán and Laroche, 2019b). At present, the raw materials used to prepare the super-hydrophilic coating, such as oxide (Duan et al., 2018, Hikku et al., 2018, Zhong et al., 2017) and the high molecular polymer (Bakshi et al., 2019), could be adopted in various base materials. Zou et al. (2021) found that konjac glucomannan (KGM) and high amylose corn starch combined to improve the mechanical properties of the films. Zhang et al. (2019) combined the modified TiO2 and hydroxyethyl acrylate (HEA) to prepare a super-hydrophilic coating with the light transmittance of more than 90%. Apart from that, the coating retains its hydrophilic properties even after 330 days in an external environment and can also rub 16 m on 600 mesh sandpaper. Joshi et al. (2019) used tetraethyl orthosilicate (TEOS), MgF2, and HNO3 as raw materials and etched the corresponding microstructures on the glass substrate by HF produced by the reaction. Then, it combined with SiO2-MgF2 sol–gel, so that the glass has super-hydrophilic coating by in-situ generation. As for the transparency of the coating, it is as high as 98% (in the visible light region) and can withstand UV for 100 h. Wang et al. (2019) developed self-healing poly (vinyl alcohol) (PVA) / poly (acrylic acid) (PAA) / Ag films with antifogging and antibacterial properties via a simple one-pot strategy. Besides, the films demonstrated the highest 5A adhesion grade and 8H pencil hardness. Although the coatings prepared with these methods are featured with good antifogging properties, they are limited in practical application due to poor mechanical properties. Therefore, the mechanical properties of the super-hydrophilic antifog surface are the core problem of its practical application. In this study, KGM was combined with water-soluble silicone fluid to form a binder, and modified nanoparticle Ecokimera (ECO) was dispersed among them to prepare the super-hydrophilic coating. The unique double-layer structure of the coating has excellent abrasion resistance and higher light transmittance. Therefore, they were expected to be used in goggles, windshields, glasses, anti-fog glass and so on. 2 Experimental section 2.1 Materials Water-soluble silicone fluid (DC193, >99%) was purchased from Dow Corning. Konjac Glucomannan (KGM, ≥95%) was purchased from Hefei Bomei biotechnology Co., Ltd. Triethanolamine (TEOA, ≥85%) was purchased from Chengdu Cologne Chemicals Co., Ltd. Sodium hexametaphosphate ((NaPO3)6, 65–70%) was purchased from Chengdu Cologne Chemicals Co., Ltd. Ecokimera (ECO) was purchased from Dongguan Jinji Environmental Protection Technology Co., Ltd. Glass slide was purchased from Cologne Chemicals Co., Ltd. Distilled water (DI) came from laboratory. Besides, it should be mentioned that all chemicals were used without further purification. 2.2 Fabrication of super-hydrophilic solution and glass The specific experimental process is shown in Fig. 1. Firstly, the ECO was modified by TEOA (Zhang et al., 2019), named as m-ECO. Secondly, DC193, TEOA and DI were mixed at a mass ratio of 1:0. 3:30 at low speed and then marked as solution A. After that, 0.65 wt% KGM solution was prepared at room temperature, stirred for 30 min at 50 ℃ and then labeled as solution B. According to the mass ratio of 10:1:0.05, A, B and (NaPO3)6 were added in turn. After being stirred for 15 min, they were moved to a water bath at 40 ℃ for 15 min to generate a cross-linking reaction for obtaining the KGM-DC193-TEOA solution, abbreviated as KOT. Finally, 1 wt% m-ECO was added. The final solution was acquired by ultrasonic processing for 40 min and got the m-ECO-KOT solution, abbreviated as E-KOT. The glass slides were ultrasonically cleaned in ethanol and deionized water, respectively. Being dried in air, the solution was spin-coated on the substrate at 500 rpm for 10 s and 3000 rpm for 20 s. Then, the sample was naturally air-dried for 12 h.The obtained coating was also deemed as E-KOT. 2.3 Characterization The water contact angle (WCA) was measured using an optical contact angle tester (JCY, FangRui Instrument Co., Ltd, Shanghai, China). 3 μL water drops were deposited on the surfaces for WCA measurements, and at least three different positions were measured on the same sample. The morphologies of the surfaces were observed via a scanning electron microscope (SEM, FEI-INSPECT F50). The surface chemical compositions of the samples were analyzed using an X-ray photoelectron spectrometer (Thermo ESCALAB 250XI) and a Fourier-transform-infrared-spectrometer (FT-IR) instrument (Magna-IR 750, Nicolet 5700, USA). The transmittance of the samples was carried out via the UV–visible spectrophotometer (UV-3000) at normal incidence in the wavelength between 300 and 800 nm. The number of contaminants on the substrate surface was observed by the metallographic microscope (NMM-820TRF). 3 Results and discussion 3.1 Characterization of coatings As shown in Fig. 2 a, the wide peak at 3384.7 cm−1 was attributed to –OH in m-ECO (Dilamian and Noroozi, 2021). The peaks at 2962.9 cm−1, 2885.9 cm−1 and 2821.1 cm−1 were assigned to the stretching vibration of C-H in CH2 (Rekha et al., 2017). The group originally belonging to TEOA was grafted into ECO, showing the success of ECO modification. In Fig. 2 b, the peak at 3365.9 cm−1 disappeared, when the vibration band of –OH was down-shifted to 3138.8 cm−1, which is caused by hydrogen bonding between-OH on m-ECO and –OH in KOT solution (Liu et al., 2020). The peaks at 2965.6 cm−1 and 2867.4 cm−1 were respectively assigned to the stretching vibration of C-H in CH3 and CH2 (Santos et al., 2020, Ye et al., 2016, Yin et al., 2020), which could be extrapolated m-ECO combined with KOT solution. The absorption band at 955.5 cm−1 resulted from the Si-OH tensile vibration of the DC193 (Yu et al., 2018).Fig. 1 Flow diagram of of superhydrophilic coating fabrication. Fig. 2 The FT-IR spectra of m-ECO (a) and E-KOT (b). The XPS survey reveals that Si, C, O and Ti are the main elements of the coating ( Fig. 3 a), of which C and O were jointly determined by DC193, KGM, TEOA and m-ECO. Ti was determined by m-ECO, and Si was determined by DC193 and substrates.Fig. 3 XPS survey spectra of E-KOT (a), high-resolution O 1 s (b), C 1 s (c) and Si 2p (d) of E-KOT. As m-ECO was wrapped in the E-KOT solution, the Ti peak could not be detected ( Fig. 3a). The strong oxygen peak appeared at 532.41 eV ( Fig. 3b), that is to say, the solution contained a lot of oxygen content. One part belonged to -OH and the other part belonged to C-O ( Fig. 3b) (Wang et al., 2019), which means that the solution contains a large number of-OH mainly from m-ECO, conducive to enhance enhancing the intrinsic hydrophilia of the coating. The C1s peak corresponds to C-C at 284.9 eV, C-O at 286.4 eV and C = O at 287.8 eV ( Fig. 3c) (Hu et al., 2016, Songkeaw et al., 2019), or in other words, KGM and DC193 are crosslinked. The Si2p peak corresponds to C-Si at 102.5 eV and SiO2 at 103.4 eV. Among them, SiO2 came from the glass substrate ( Fig. 3d) (Aghaei et al., 2018), thus confirming the bonding of the E-KOT to the glass surfaces. The surface morphology of the substrate before and after treatment was observed by the SEM (Fig. 4). The surface of the original glass was relatively smooth ( Fig. 4 a). However, the sample had a dense granular rough structure ( Fig. 4 b). Then, further magnification revealed a large number of irregular cracks on the coating surface, similar to the structure of a dried rice field ( Fig. 4 c, 4d), which could increase the solid–liquid contact area and promote the liquid absorption of the coating, so as to improve the super-hydrophilic property. Furthermore, it can be seen that there were a large number of prickly pear-like structures on the coating surface with cluster particles overlapping on the surface. The structure is very different from m-ECO, proving that substances in the KOT solution reacted with m-ECO so that it was wrapped (Ao et al., 2017, Kim et al., 2017). ECO, a regular hexagon layer (García-Glez et al., 2017), as a kind of unique titanium dioxide, with strong adsorption and the characteristics of nanoparticles is potential transfer of high-energy translocation reaction (Kim and Chung, 2001), which may generate a redox reaction under the condition of no light and produce strong oxidation of hydrogen and oxygen-free radicals (Kőrösi and Dékány, 2006). Not only can they absorb harmful gas, but also the solution is featured with excellent hydrophilicity. Therefore, ECO can be adopted to construct the rough structure of the bionic super-hydrophilic coating. Besides, as the ECO was covered by TEOA, the morphology of ECO was not observed, which was consistent with the XPS results.Fig. 4 SEM images of the bare glass surfaces (a), SEM images of KOT surface (b), (c) and (d) partial enlarged detail in (b), the illustration shows a dried rice field. 3.2 Super-hydrophilicity and durability of the E-KOT coating The WCA of clean glass and sample glass was measured at room temperature, while the WCA of bare glass was about 30° ( Fig. 5 a). It was hydrophilic due to the existence of some hydroxyl groups on the surface, but could not reach the super-hydrophilicity. By contrast, the WCA of glass treated with E-KOT solution decreased to less than 3° within 1 s ( Fig. 5 b), that is to say, the E-KOT coating possessed excellent super-hydrophilicity. The samples were placed in a petri dish, and the WCA was measured every 15 days to determine the durability of super-hydrophilicity. Fig. 5 c showed that the WCA of the samples was close to 5° after 90 days.Fig. 5 Digital images of the WCA for the bare glass (a) and the E-KOT coating (b); change of static WCAs of the E-KOT coating with time (c); these illustrations are WCA images before and after placing the E-KOT coating for 90 days E-KOT coating. The surface energy is equal to the sum of the components of the dispersion and polar forces (Owens and Wendt, 1969). Water has high polaritywith both the dispersion force component and the polar force component. Therefore, the relationship between the contact angle and surface energy can be expressed as (Owens and Wendt, 1969):(1) 1+cosθ=2γsdγldγlv+2γshγlhγlv where rsd is the component of dispersion force; rshrefers to the polar force component; θ indicates the contact angle of water, while rlvmeans the surface tension of water. From formula (1), it can be inferred that θ decreases as rsh is large. In other words, super hydrophilic materials can be prepared by increasing the polar force component of solid surface. By combining the interpretation of FT-IR, SEM and XPS results, it can be seen that the reason for the super-hydrophilicity of the E-KOT coating may be as follows: firstly, there are much –OH on the surface of the E-KOT coating, which comes from DC193 and increases the polar force component, thus producing affinity to water molecules. Therefore, the inherent hydrophilicity of the coating is enhanced (Zorba et al., 2010). Secondly, wettability is determined by surface chemical composition and microstructure. When the surface has the same elemental composition, the wettability is enhanced by changing the surface roughness. The ECO NPs provide a rough structure on the nanoscale, increasing the surface roughness. High surface roughness facilitates the establishment of ideal conditions for extreme super-hydrophilic behavior (Liu and He, 2008, Zhang et al., 2005). Thirdly, there are cracks similar to a dried rice field in the coating, which can increase the contact area between solid and liquid, thus further promoting the absorption of the coating to liquid and then increasing the hydrophilicity of the glass surface. In this case, it can be found that the synergy of the above mentioned factors resulted in the outstanding super-hydrophilicity of the E-KOT coating. As the inherent characteristics of the coating enable it to be stable and not to be damaged by the external environment, the super-hydrophilicity can stay in the air for a long time. 3.3 Transmittance of the E-KOT coating Different m-ECO content of the coating was synthesized on the glass bases, of which the quality scores of m-ECO were 0, 0.5 wt%, 1 wt%, 2 wt%, and 3 wt%, respectively. As shown in Fig. 6 a, the transparency of the samples remained basically unchanged. The transmittance was measured by using UV–VIS spectrophotometer. Then, it can be found that the transmittance of samples in the visible range (400–760 nm) only decreased by about 3% after adding the m-ECO ( Fig. 6 b), showing that the distribution of m-ECO NPs was uniform in the KOT solution, while that at 303 cm−1 decreased significantly with the increase of m-ECO content ( Fig. 6 b), indicating that the E-KOT coating has good absorption ability under UV light.Fig. 6 Visual effect of the coatings with different m-ECO purities on glass bases (a); the transmittance of the E-KOT coating with different mass fractions (b). According to our experimental results, it can be known that when the m-ECO is less than 1 wt%, the mechanical properties of the coating are poor. Generally speaking, the mechanical properties of the coating increase with the increase of m-ECO content. Nevertheless, the increase in the content of nanoparticles will affect the transparency of the coating. Therefore, the best m-ECO content is 1 wt% that is used in all experiments unless otherwise specified. 3.4 Antifogging and self-cleaning properties of the E-KOT coating The antifogging properties of coated glass and bare glass were shown in Fig. 7 a. Half of the clean glass was coated with the E-KOT solution, while the other half was not treated at all. After that, the sample was put in the refrigerator at 4 ℃ for 1 h and then taken it out to see if the surface fogs. In this case, it was observed that countless small droplets on the untreated glass gathered together, which caused light scattering and resulted in being unable to see the words below clearly. However, the treated surface made the droplets disperse quickly on the surface and then form a film. Therefore, light was unable to scatter, thus achieving the antifogging effect, and the words behind could be seen very clearly. Then, the sample was placed back to the refrigerator for the next test. After 180 cycles, it still has excellent antifogging performance ( Fig. 7 b). In order to determine whether the prepared coating can be applied to other substrates, the coating was rubbed onto the goggles, and the anti-fogging test was carried out in the same way. As shown in Fig. 7 c, it could be seen that the anti-fogging condition was similar to that of the glass substrate. Therefore, the coating can also be adopted on goggles.Fig. 7 Antifogging test of uncoated (left) and E-KOT coated (right) glass bases (a); WCAs with different days (b), these illustrations are WCA of sample before and after antifogging experiments for 180 times; visual effect of the coatings with different m-ECO purities on goggles (c). The surface of glass slides is always easy to be polluted by dust, which affects the transparency and service life. Therefore, the coating is required to be self-cleaning to quickly remove contaminants from the glass surface. In general, super-hydrophilic surfaces have excellent self-cleaning properties. For that reason, the self-cleaning properties of the prepared coating were measured. As was shown in Fig. 8 a and b, the treated and untreated glass was polluted with toner. After the toner was laid flat on the glass slide, it was washed with running water, and the samples after cleaning were observed through the metallographic microscope. Then, it could be clearly seen that the surface of the treated glass had significantly less residual toner than the surface of the untreated glass ( Fig. 8 c, d). The reasons may be as follows: on the one hand, the double-layer structure on the sample surface reduced the contact area between the contaminant and the sample, while on the other hand, it increased the contact area with water and the absorbent capacity (Adachi et al., 2018, Banerjee et al., 2015, Wang et al., 1997). In other words, when the contaminant came into contact with the sample surface, the prickly pear-like micro-nano structure protected the sample from contact with the contaminant. At the same time, the micro-nano structure on the sample surface could absorb more water than the clean glass, thus forming a water film between the sample surface and the pollutants. Under the action of external forces such as gravity or wind force, the water droplets slid out of and carried pollutants away from the surface ( Fig. 8 e), which as two synergistic effects gave the coating excellent self-cleaning properties.Fig. 8 Metallographic microscope images of toner on the sample surface and bare glass surface: fouling processes (a) and (b), bare glass (c) and super-hydrophilic glass after rinsing with water (d); the self-cleaning mechanism of water on the surface of the super-hydrophilic coating (e). 3.5 Mechanical properties of the coating Sandpaper wear tests were performed to determine the mechanical properties of E-KOT coatings. To be specific, the glass was placed on sandpaper (600 mesh) with a weight of 100 g. Then, it was dragged along the ruler of 20 cm and rotated 90°, followed by being again moved for 20 cm to form a cycle (Jin et al., 2017) ( Fig. 9 a, b), which can ensure the transverse and longitudinal wear of the coating surface. It could be seen from Fig. 9 c that the WCA of the glass increased slightly with the increase of wear cycles. After 60 cycles, the WCA of the sample was about 6°. For one thing, the excellent mechanical stability of the sample came from the special structure similar to the combination of dried paddy field and prickly pear, While foranother, the adhesion of KGM provided good adhesion for the coating.Fig. 9 The friction testing of (a) transverse and (b) longitudinal; (c) friction times-dependent variations of the static WCA of the E-KOT coating. SEM images of the same sample with the same multiple at different friction times (4 m, 8 m, 12 m, 16 m, 20 m and 24 m, respectively) (d-i). To further investigate the causes of the E-KOT coating’s super-hydrophilicity loss, the surface morphology change of the coating after every 10 cycles of wear was observed ( Fig. 9 d-i). With the increase of wear times, the surface of the coating gradually became very smooth. Meanwhile, the hydrophilic effect was worse, indicating that the structure similar to the prickly pear determines the mechanical strength of the coating. 3.6 The high wear resistance mechanism of the coating Based on the above test results, the mechanism concerning high wear resistance of the E-KOT coating is propose in Fig. 10 . (1) A large number of hydroxyl groups are connected on the surface of the m-ECO and DC193, and some of them combine with the hydroxyl groups on the glass to form hydrogen bonds, thus enhancing the bonding performance of the coating (Liu et al., 2020). (2) KGM also contains a certain amount of hydroxyl groups. Part of the coating is bound to the hydroxyl group in solution (DC193, m-ECO, etc.), so that the nanoparticles are closely linked to it, which can not only improve the roughness of the coating, but also enhance the superhydrophilicity. In addition to that, some hydroxyl groups bind to the glass surface of the hydroxyl group to form hydrogen bonds, so that the coating’s mechanical properties will be further improved. At the same time, as the KGM itself is characterized with high viscosity, the mechanical wear resistance of the coating is enhanced to a great extent (Fang, 2021). Therefore, the above three reasons together caused the excellent wear resistance of E-KOT coatings.Fig. 10 Mechanism of the coating. 4 Conclusions The high wear-resistant super-hydrophilic coating was successfully prepared by the simple and low-cost method. The friction resistance of the coating was systematically studied. With the increase of friction times, m-ECO NPs and KOT were erased in sequence. Such double-layer structure led to excellent friction resistance. At the same time, the coating had excellent antifogging and super-hydrophilicity properties. The super-hydrophilic coating has a broad application prospect in goggles, automobile windshields, glasses and so on. Furthermore, the coating can also be adopted in anti-ultraviolet materials. CRediT authorship contribution statement Juan Xiang: Conceptualization, Methodology, Formal analysis, Writing - original draft, Writing - review & editing, Methodology, Validation, Data curation, Formal analysis. Xiaoying Liu: Discussion on experimental factors. Yan Liu: Validation. Lilin Wang: Project administration. Yan He: Project administration. Ling Luo: Data curation. Gang Yang: Resources, Supervision. Xiaohong Zhang: Funding acquisition, Resources. Chengyi Huang: Resources, Project administration. Yanzong Zhang: Methodology, Writing - review & editing, Resources, Supervision, Funding acquisition. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments This work was supported by the Science and Technology Department of Sichuan Province (2018JY0457, 2019YFS0502, 2021YFG0275). ==== Refs References Adachi T. Photocatalytic, superhydrophilic, self-cleaning TiO2 coating on cheap, light-weight, flexible polycarbonate substrates Appl. Surf. Sci. 458 2018 917 923 10.1016/j.apsusc.2018.07.172 Aghaei R. Eshaghi A. Aghaei A.A. Durable transparent super-hydrophilic hollow SiO2-SiO2 nanocomposite thin film Mater. Chem. Phys. 219 2018 347 360 10.1016/j.matchemphys.2018.08.039 Ao Y.H. Bao J.Q. Wang P.F. Wang C. A novel heterostructured plasmonic photocatalyst with high photocatalytic activity: Ag@AgCl nanoparticles modified titanium phosphate nanoplates J. Alloy. Compd. 698 2017 410 419 10.1016/j.jallcom.2016.12.231 Bai S. Li X.H. Zhao Y.H. Ren L.X. Yuan X.Y. Antifogging/Antibacterial Coatings Constructed byN-Hydroxyethylacrylamide and Quaternary Ammonium-Containing Copolymers ACS Appl. Mater. Interfaces 12, (10) 2020 12305 12316 10.1021/acsami.9b21871 32068389 Bakshi S. Permanent superhydrophilic surface modification in microporous polydimethylsiloxane sponge for multi-functional applications Journal of Colloid And Interface Science 552 2019 34 42 10.1016/j.jcis.2019.05.028 31102847 Banerjee S. Dionysiou D.D. Pillai S.C. Self-cleaning applications of TiO2 by photo-induced hydrophilicity and photocatalysis Appl. Catal. B 176–177 2015 396 428 10.1016/j.apcatb.2015.03.058 Dilamian M. Noroozi B. Rice straw agri-waste for water pollutant adsorption: Relevant mesoporous super hydrophobic cellulose aerogel Carbohydr. Polym. 251 2021 117016 10.1016/j.carbpol.2020.117016 Duan Z.F. Non-UV activated superhydrophilicity of patterned Fe-doped TiO2 film for anti-fogging and photocatalysis Appl. Surf. Sci. 452 2018 165 173 10.1016/j.apsusc.2018.05.029 Durán I.R. Laroche G. Water drop-surface interactions as the basis for the design of anti-fogging surfaces: Theory, practice, and applications trends Adv. Colloid Interface Sci. 263 2019 68 94 10.1016/j.cis.2018.11.005 30521982 Durán I.R. Laroche G. Current trends, challenges, and perspectives of anti-fogging technology: Surface and material design, fabrication strategies, and beyond Prog. Mater Sci. 99 2019 106 186 10.1016/j.pmatsci.2018.09.001 Fang F. Shear-induced synergistic effects of konjac glucomannan and waxy potato starch on viscosity and gel strength Food Hydrocolloids 114 2021 106540 10.1016/j.foodhyd.2020.106540 Feng K. Universal Antifogging and Antimicrobial Thin Coating Based on Dopamine-Containing Glycopolymers ACS Appl. Mater. Interfaces 12 24 2020 27632 27639 10.1021/acsami.0c07949 32392029 García-Glez J. Trobajo C. Khainakov S.A. Amghouz Z. α-Titanium phosphate intercalated with propylamine: An alternative pathway for efficient europium(III) uptake into layered tetravalent metal phosphates Arabian J. Chem. 10 6 2017 885 894 10.1016/j.arabjc.2016.07.013 He S.J. Formation of superhydrophobic micro-nanostructured iron oxide for corrosion protection of N80 steel Mater. Des. 160 2018 84 94 10.1016/j.matdes.2018.09.002 Hikku G.S. Alkyd resin based hydrophilic self-cleaning surface with self-refreshing behaviour as single step durable coating J. Colloid Interface Sci. 531 2018 628 641 10.1016/j.jcis.2018.07.089 30059914 Hsu W.J. Zeolite-Based Antifogging Coating via Direct Wet Deposition Langmuir 35 7 2018 2538 2546 10.1021/acs.langmuir.8b03738 Hu X.B. Highly transparent superhydrophilic graphene oxide coating for antifogging Mater. Lett. 182 2016 372 375 10.1016/j.matlet.2016.06.125 Hu Y. Wang L. Fang F. Prevention of Fogging of Protective Eyewear for Medical Staff During the COVID-19 Pandemic Journal of emergency nursing 46, (5) 2020 564 566 10.1016/j.jen.2020.05.003 32540099 Jin B. He J. Yao L. Zhang Y. Li J. Rational Design and Construction of Well-Organized Macro-Mesoporous SiO2/TiO2Nanostructure toward Robust High-Performance Self-Cleaning Antireflective Thin Films ACS Appl. Mater. Interfaces 9 20 2017 17466 17475 10.1021/acsami.7b04140 28492300 Joshi D.N. Atchuta S.R. Lokeswara Reddy Y. Naveen Kumar A. Sakthivel S. Super-hydrophilic broadband anti-reflective coating with high weather stability for solar and optical applications Sol. Energy Mater. Sol. Cells 200 2019 110023 10.1016/j.solmat.2019.110023 Kim M. Chung J.G. A Study on the Adsorption Characteristics of Orthophosphates on Rutile-Type Titanium Dioxide in Aqueous Solutions J. Colloid Interface Sci. 233 1 2001 31 37 10.1006/jcis.2000.7225 11112303 Kim Y. Kim H. Lee J. Lee S. Kwon K. Morphological change and photocatalytic activity of titanium phosphates J. Photochem. Photobiol., A 338 2017 146 151 10.1016/j.jphotochem.2017.01.016 Kőrösi L. Dékány I. Preparation and investigation of structural and photocatalytic properties of phosphate modified titanium dioxide Colloids Surf., A 280 1–3 2006 146 154 10.1016/j.colsurfa.2006.01.052 Lemarteleur V. 3D-printed protected face shields for health care workers in Covid-19 pandemic Am. J. Infect. Control 000 2020 1 3 10.1016/j.ajic.2020.08.005 Liu Q. Locklin J. Transparent Grafted Zwitterionic Copolymer Coatings That Exhibit Both Antifogging and Self-Cleaning Properties ACS Omega 3 12 2018 17743 17750 10.1021/acsomega.8b02867 Liu T.L. Kim C.C. Turning a surface superrepellent even to completely wetting liquids Science 346 6213 2014 1096 1100 10.1126/science.1254787 25430765 Liu X.M. He J.H. Superhydrophilic and Antireflective Properties of Silica Nanoparticle Coatings Fabricated via Layer-by-Layer Assembly and Postcalcination The Journal of Physical Chemistry C 113 1 2008 148 152 10.1021/jp808324c Liu Z. Lin D.H. Shen R. Yang X.B. Characterizations of novel konjac glucomannan emulsion films incorporated with high internal phase Pickering emulsions Food Hydrocolloids 109 2020 106088 10.1016/j.foodhyd.2020.106088 Nam E. Antifogging Surface Facilitated by Nanoscale Coatings with Controllable Hydrophobicity and Cross-Linking Density Macromol. Mater. Eng. 302 1 2017 1600199 10.1002/mame.201600199 Oldfield E. Malwal S.R. COVID-19 and Other Pandemics: How Might They Be Prevented? ACS Infectious Diseases 6, (7) 2020 1563 1566 10.1021/acsinfecdis.0c00291 32478500 Owens D.K. Wendt R.C. Estimation of the surface free energy of polymers J. Appl. Polym. Sci. 13 8 1969 1741 1747 10.1002/app.1969.070130815 Rekha S. Martinez A.I. Safeera T.A. Anila E.I. Enhanced luminescence of triethanolamine capped calcium sulfide nanoparticles synthesized using wet chemical method J. Lumin. 190 2017 94 99 10.1016/j.jlumin.2017.05.042 Santos N.L. Physicochemical properties of konjac glucomannan/alginate films enriched with sugarcane vinasse intended for mulching applications Int. J. Biol. Macromol. 165 2020 1717 1726 10.1016/j.ijbiomac.2020.10.049 33069823 Songkeaw P. Onlaor K. Thiwawong T. Tunhoo B. Reduced graphene oxide thin film prepared by electrostatic spray deposition technique Mater. Chem. Phys. 226 2019 302 308 10.1016/j.matchemphys.2019.01.048 Sun G.Q. Fan M.Y. Chen L.L. Luo J. Liu R. Antifogging UV curable coatings based on hierarchical composite particles through electrostatic interactions Colloids Surf. A 589 2020 124458 10.1016/j.colsurfa.2020.124458 Tao C. Formation of zwitterionic coatings with an aqueous lubricating layer for antifogging/anti-icing applications Prog. Org. Coat. 115 2018 56 64 10.1016/j.porgcoat.2017.11.002 Walker C. Transparent Metasurfaces Counteracting Fogging by Harnessing Sunlight Nano Lett. 19 3 2019 1595 1604 10.1021/acs.nanolett.8b04481 30689389 Wang F.J. Superhydrophobic Calcium Aluminate Cement with Super Mechanical Stability Ind. Eng. Chem. Res. 58 24 2019 10373 10382 10.1021/acs.iecr.9b01188 Wang R. Light-induced amphiphilic surfaces Nature 388 6641 1997 431 432 10.1038/41233 Wang W. A facile antifogging/frost-resistant coating with self-healing ability Chem. Eng. J. 378 2019 122173 10.1016/j.cej.2019.122173 Wang Y. Yao L. Ren T.T. He J.H. Robust yet self-healing antifogging/antibacterial dual-functional composite films by a simple one-pot strategy J. Colloid Interface Sci. 540 2019 107 114 10.1016/j.jcis.2019.01.008 30634058 Wen M.X. Wang L. Zhang M.Q. Jiang L. Zheng Y.M. Antifogging and Icing-Delay Properties of Composite Micro- and Nanostructured Surfaces ACS Appl. Mater. Interfaces 6 6 2014 3963 3968 10.1021/am405232e 24602042 Ye S.X. KGM-based magnetic carbon aerogels matrix for the uptake of methylene blue and methyl orange Int. J. Biol. Macromol. 92 2016 116 1174 10.1016/j.ijbiomac.2016.07.106 27370746 Yin J.Y. Ma L.Y. Xie M.Y. Nie S.P. Wu J.Y. Molecular properties and gut health benefits of enzyme-hydrolyzed konjac glucomannans Carbohydr. Polym. 237 2020 116117 10.1016/j.carbpol.2020.116117 Yu N.L. Xiao X.Y. Ye Z.H. Facile preparation of durable superhydrophobic coating with self-cleaning property Surf. Coat. Technol. 347 2018 199 208 10.1016/j.surfcoat.2018.04.088 Yu X.D. Highly durable antifogging coatings resistant to long-term airborne pollution and intensive UV irradiation Mater. Des. 194 2020 108956 10.1016/j.matdes.2020.108956 Zhang M.Q. Wang L. Feng S.L. Zheng Y.M. A Strategy of Antifogging: Air-Trapped Hollow Microsphere Nanocomposites Chem. Mater. 29 7 2017 2899 2905 10.1021/acs.chemmater.6b05139 Zhang X.T. Self-Cleaning Particle Coating with Antireflection Properties Chem. Mater. 17 3 2005 696 700 10.1021/cm0484201 Zhang Y.G. Zhang S.F. Wu S.L. Room-temperature fabrication of TiO2-PHEA nanocomposite coating with high transmittance and durable superhydrophilicity Chem. Eng. J. 371 2019 609 617 10.1016/j.cej.2019.04.038 Zhong H. Hu Y. Wang Y.H. Yang H.X. TiO2/silane coupling agent composed of two layers structure: A super-hydrophilic self-cleaning coating applied in PV panels Appl. Energy 204 2017 932 938 10.1016/j.apenergy.2017.04.057 Zorba V. Chen X. Mao S.S. Superhydrophilic TiO2 surface without photocatalytic activation Appl. Phys. Lett. 96 9 2010 093702 10.1063/1.3291667 Zou, Y.Y. et al., 2021. Formation of high amylose corn starch/konjac glucomannan composite film with improved mechanical and barrier properties. Carbohydrate polymers 251, 117039-117039. DOI: 10.1016/j.carbpol.2020.117039

0

PMC9749070

NO-CC CODE

2022-12-15 23:22:50

no

Chem Eng Sci. 2021 Oct 12; 242:116749

utf-8

Chem Eng Sci

10.1016/j.ces.2021.116749

oa_other

==== Front Comput Biol Med Comput Biol Med Computers in Biology and Medicine 0010-4825 1879-0534 Elsevier Ltd. S0010-4825(22)00043-9 10.1016/j.compbiomed.2022.105251 105251 Article Performance change with the number of training data: A case study on the binary classification of COVID-19 chest X-ray by using convolutional neural networks Imagawa Kuniki ∗ Shiomoto Kohei Tokyo City University, Faculty of Information Technology, 1-28-1 Tamazutsumi, Setagaya-ku, Tokyo, 158-8557, Japan ∗ Corresponding author. 23 1 2022 3 2022 23 1 2022 142 105251105251 25 11 2021 15 1 2022 19 1 2022 © 2022 Elsevier Ltd. All rights reserved. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. One of the features of artificial intelligence/machine learning-based medical devices resides in their ability to learn from real-world data. However, obtaining a large number of training data in the early phase is difficult, and the device performance may change after their first introduction into the market. To introduce the safety and effectiveness of these devices into the market in a timely manner, an appropriate post-market performance change plan must be established at the timing of the premarket approval. In this work, we evaluate the performance change with the variation of the number of training data. Two publicly available datasets are used: one consisting of 4000 images for COVID-19 and another comprising 4000 images for Normal. The dataset was split into 7000 images for training and validation, also 1000 images for test. Furthermore, the training and validation data were selected as different 16 datasets. Two different convolutional neural networks, namely AlexNet and ResNet34, with and without a fine-tuning method were used to classify two image types. The area under the curve, sensitivity, and specificity were evaluated for each dataset. Our result shows that all performances were rapidly improved as the number of training data was increased and reached an equilibrium state. AlexNet outperformed ResNet34 when the number of images was small. The difference tended to decrease as the number of training data increased, and the fine-tuning method improved all performances. In conclusion, the appropriate model and method should be selected considering the intended performance and available number of data. Keywords Performance change Deep learning Convolutional neural network COVID-19 Chest X-ray Medical device regulation ==== Body pmc1 Introduction The extensive interest in artificial intelligence (AI) and machine learning (ML) application is growing in the medical field. This is primarily driven by the impressive progress made by deep learning (DL) as a subset of ML because of the increased computational power and an explosion in the availability of large datasets. The number of research papers on the application of DL to medical fields has increased. For the medical image analysis, the number has dramatically increased since 2015. The number of academic papers in major conferences and journals exceeded 300 by the end of 2016 [1]. Furthermore, more than 60 AI/ML-based medical devices have already been approved by the U.S. Food and Drug Administration (FDA) in the United States [2]. There are several types of AI/ML-based medical devices. Muehlematter et al. [3] reported that most AI/ML-based medical devices approved by the FDA are used in radiology, but they span across various medical specialties, such as cardiovascular and neurology. For example, applications in radiology can be categorized into classification, detection, segmentation, etc., and the required performance varies greatly depending on the modality and target diseases. AI/ML-based medical devices are roughly divided into two types: 1) locked type, which fixes performance prior to marketing and unable to change performance with use; and 2) continuous type, which can change performance by continuously training data after market introduction. To date, several FDA-approved AI/ML-based medical devices are typically locked type, but the FDA announced its marketing authorization for the continuous type on February 2020 [4]. The number of these medical devices is expected to increase in the market in the future. In April 2019, the FDA published a discussion paper for a proposed regulatory framework to account for the iterative nature of AI/ML-based medical devices [5]. The paper describes the total product lifecycle approach. As part of this framework, the necessity of submitting a predetermined change control plan, in which manufacturers are anticipated to perform modifications to performance, inputs, or intended use prior to marketing, is also described. Other jurisdictions are also preparing papers on regulatory guidance, with a concept similar to that in this discussion paper. Therefore, an appropriate post-market performance change plan must be established at the timing of the premarket approval. Since the outbreak of the Coronavirus Disease 2019 (COVID-19), many studies have used convolutional neural networks (CNNs) to detect COVID-19 on chest X-ray (CXR) images. For the COVID-19 diagnosis, testing through viral RNA identification in reverse transcriptase polymerase chain reaction (RT-PCR) is currently recommended. However, chest imaging techniques, such as computed tomography (CT) and chest radiography, are considered as part of the diagnostic workup of patients with suspected or probable COVID-19 disease in case RT-PCR is not available or the results are delayed or initially negative in the presence of symptoms suggestive of COVID-19 [6,7]. Applying ML methods to COVID-19 radiological imaging may improve the diagnosis accuracy compared with the gold-standard RT-PCR while providing a valuable insight into the prognostication of patient outcomes. In particular, chest radiography is widely used, takes less imaging time, and has an accessible diagnostic modality that may be easily brought to the patient's bed. Arising from the success of the ImageNet Large-scale Visual Recognition Challenge (ILSVRC) in 2012 [8], the CNN is a class of artificial neural networks that has become dominant in object recognition, including radiology [9]. Most of the recent papers on COVID-19 diagnosis were conducted based on existing off-the-shelf models and classified images into two [[10], [11], [12], [13]] or three classes [[13], [14], [15]]: COVID-19 and Normal or COVID-19, non-COVID-19 pneumonia, and Normal. These performances such as accuracy achieves exceed 90% by using data augmentation, transfer learning or combining of publicly available datasets. For example, Nayak et al. [16] evaluated the performance of eight pretrained fundamental CNN models, which achieved excellent results in the ILSVRC, to classify COVID-19 and Normal. They also conducted a comparative analysis by considering hyperparameters, such as batch size, learning rate, and optimizer. ResNet-34 [17] exhibited the best performance, followed by AlexNet [18]. Rahaman et al. [19] evaluated the performance of 15 pretrained fundamental CNN models to classify COVID-19, non-COVID-19 pneumonia, and Normal. VGG 19 [20] obtained the highest classification accuracy. Tuan [21] performed two- and three-class tasks to classify COVID-19 using three fine-tuned CNN models (i.e., AlexNet, GoogleNet [22], and SqueezeNet [23]) without data augmentation and achieved a high classification performance in terms of accuracy, sensitivity, specificity, precision, F1 score, and area under the curve (AUC). The results suggested that the fine-tuning of network learning parameters is important because it can help avoid the development of more complex models when existing ones can achieve the same or much better performance. On the contrary, many current studies were conducted based on a small number of training data or a combination of training data without demographic statistics (e.g., age and sex distributions) due to the limited publicly available datasets. These estimated performances may probably be optimistic and misleading because of the high-risk bias based on the non-representative selection of control patients and model overfitting [24,25]. In 2020, Alberto et al. [26] publicly released a large and fully annotated BrixlA dataset of 4703 CXR images related to COVID-19 with additional information on severity, participants' age and sex, and modality manufacturer. A large dataset from the Valencian Region Medical Image Bank containing 3141 CXR and 2239 CT images of patients with COVID-19 along with their radiological findings and locations, pathologies, radiological reports, Digital Imaging and Communications in Medicine (DICOM) metadata, diagnostic antibody tests, etc., was also publicly released [27]. Such large databases are expected to be actively developed and made available to the public in the future. It is hoped that these large datasets with patient demographics will enable the introduction of appropriate AI/ML-based medical devices to the market to support medical decision making through proper regulation. With the abovementioned background, this study evaluates the performance change as the number of training data increases. Two different CNNs, namely AlexNet and ResNet34, with and without a fine-tuning method are used to classify COVID-19 and Normal. A large BrixlA dataset of CXR images related to COVID-19 are used as the training, validation, and test data. The AUC, sensitivity, and specificity are utilized as the performance evaluation items because they are mainly evaluated through FDA premarket approvals [28]. The major outcomes of this study are the following: 1) All performances were rapidly improved as the number of training data were increased and reached an equilibrium state. 2) AlexNet outperformed ResNet34 when training data were small, and the difference between the performance of AlexNet and ResNet34 decreased as the training data increased. 3) The fine-tuned CNNs performed better than CNNs trained from scratch in all training datasets; this effect is particularly noticeable in small training data. 4) The change in the performance of the binary classification for COVID-19 and Normal datasets can be generalized as COVID-19 and non-COVID-19 pneumonia datasets. 2 Material 2.1 Datasets Two independent datasets were used for the evaluation. The first dataset is the BrixlA dataset, which comprises 4703 CXR images of COVID-19 objects taken for both triage and patient monitoring in sub-intensive and intensive care units for 1 month between March 4 and April 4, 2020 at ASST Spedali Civili di Brescia. The images were retrieved from the facility's Picture Archiving and Communication Systems (PACS) and disclosed as DICOM formats with additional information (i.e., severity, patient's age and gender, and modality manufacturer). The second data set is a Chest X-ray14 dataset comprising 112 120 CXR images with 14 diseases and one for the Normal label from 30 805 unique patients. The CXR images were extracted from the PACS database through natural language processing (NLP) at the National Institutes of Health Clinical Center between 1992 and 2015 and disclosed as portable network graphics formats with additional information (i.e., patient ID, age, and gender and view position). The DICOM images in the BrixlA dataset, whose window width (WW) and window center (WC) could not be derived from DICOM Tags, were excluded. The remaining DICOM images were converted from 16-bit into 8-bit through windowing, which changed the picture's appearance to highlight particular structures using the WW and WC derived from the DICOM Tags. Fig. 1 depicts an example of the CXR images from both COVID-19 and Normal classes. We randomly selected 4000 COVID-19 images and 4000 CXR images labeled as Normal from the Chest X-ray14 dataset. The combined dataset, which contained 8000 CXR images, was split into two to separate 7000 images for training and validation and 1000 images for testing. The training and validation data were selected as 16 different datasets (N = 250, 500, 750, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, 6000, 6500, and 7000). The ratio of the COVID-19 and Normal classes had the same proportion in all the training, validation, and test data. Table 1 and Fig. 2 depict the detailed patient demographics and age distribution for each dataset.Fig. 1 Example of CXR images: (a) COVID-19 image without windowing; (b) COVID-19 without windowing derived in the BrixlA dataset; and (c) Normal image from the chest X-ray dataset. Fig. 1 Table 1 Data and patient characteristics. AP: anteroposterior and PA: posteroanterior representing the view position. We cannot find AP and PA in the DICOM Tags, but the ratio is reported as AP (87%) and PA (13%) in the BrixlA dataset. Table 1Database Origin Purpose Label Images Patients Female Male Age AP PA Brixla Traning and Validation Test COVID-19 3500 1804 1060 2440 59 ± 14 – – 500 429 130 370 58 ± 14 – – Chest X-ray 14 Traning and Validation Test Normal 3500 3026 1521 1979 45 ± 17 1229 2271 500 491 223 277 45 ± 17 175 325 Fig. 2 Age distribution for each dataset. Fig. 2 3 Methodology Fig. 3, Fig. 4 show the proposed method for classifying COVID-19 and Normal, which mainly consisted of preprocessing and classification with a fine-tuned CNN. The detailed preprocessing, classification and evaluation are described in the subsequent sections. The 16 training and validation datasets fed to each CNN model with and without a fine-tuning method were evaluated on the common test dataset.Fig. 3 Proposed method for classifying COVID-19 and Normal by using a fine-tuning method, that is AlexNet. Fig. 3 Fig. 4 Proposed method for classifying COVID-19 and Normal by using a fine-tuning method, that is ResNet34. Fig. 4 3.1 Preprocessing Before feeding the CXR images to the system as input, all CXR images were resized to 256 × 256 px and cropped in the center as 224 × 224 px. The grayscale images were converted into a colored format to three channels (RGB: red, green, and blue). The pixel values of the input images were normalized in between ranges 0 and 1 based on the mean and the standard deviation to maintain the numerical stability in the CNN architectures and use a pretrained CNN with ImageNet, which contains 1.4 million images with 1000 classes. 3.2 Classification The AlexNet and ResNet34 models were used for the classification. AlexNet was treated as the first breakthrough in the CNN model architecture, serving as the winner of the 2012 ILSVRC. It adopted an eight-layer network structure consisting of five convolutional layers and three fully connected layers. After each convolution in the five convolutional layers, maximum pooling was performed to reduce the amount of data. Data augmentation and dropout were used to reduce overfitting. Rectified linear units were utilized as an activation function instead of a sigmoid or hyperbolic tangent function. AlexNet had eight layers, and the number of parameters was approximately 60 million. Meanwhile, ResNet was the winner of the 2015 ILSVRC, which enables the training of up to hundreds or thousands of layers and inspired many other models. When the network is too deep, the gradients from where the loss function is calculated easily shrink to zero after several chain rule applications. This result on the weights never updates its values; therefore, no learning is performed. ResNet overcomes this degradation problem by introducing residual connections mapping to fit input from a previous layer to the next layer and achieves compelling performance. ResNet34 had 34 layers and 21.8 million parameters. In addition to the AlexNet and ResNet34 CNN models without pretraining, we utilized a pretrained network, called the fine-tuning method. This method is often applied to radiology studies to replace the fully connected layers of the pretrained model with a new set of fully connected layers to retrain on a given dataset and finetune all the kernels in the pretrained convolutional base by means of backpropagation. All convolutional base layers can be finetuned. Alternatively, some earlier layers can be fixed while fine-tuning the remaining deeper layers. Our method requires the unfreezing of the entire model originally trained on a large-scale labeled dataset called ImageNet and re-training it on CXR images. Data augmentation was not conducted herein. This work is motivated by the observation that early-layer features appear more generic (e.g., edges applicable to various datasets and tasks), while later features progressively become more specific to a particular dataset or task [29,30]. 3.3 Evaluation The AUCs derived from the receiver operating characteristic (ROC) curve, sensitivity, and specificity were used for the evaluation. The ROC curve is graphical display of the true positive rate (TPR) on the y-axis and the false positive rate (FPR) on the x-axis for varying the cut-off points. Both axes are from 0 to 1. The TPR is equal to the sensitivity, while the FPR is equal to ”(1 - specificity)”. The AUC is the definite integral of an ROC curve and an effective and combined measure of sensitivity and specificity that assesses the inherent validity of a diagnostic test. An AUC closer to 1 indicates a better test performance. Sensitivity and specificity were determined by using the cut-off point defined as the Youden Index [31]. The Youden Index is the point on the ROC curve that is farthest from the line of equality (diagonal line). The optimal cut-off value is that at which the determined ”sensitivity + specificity - 1” is maximized. To construct a 95% confidence interval, the standard error was calculated using the Hanley and McNeil method [32] for the AUC and the Wald method [33] for sensitivity and specificity. A nonlinear function, y = a − bx (−c), was also fitted to each data. All results will be described herein to show the relation between each performance and the number of training images. 4 Experiment and results The 16 training and validation datasets were divided into 32 batches. The training of the AlexNet and ResNet34 CNN models with and without a fine-tuning method was conducted by using an Adam optimizer (β1 = 0.9, β2 = 0.999). The network was trained 50 epochs. The learning rate was set to 1 × 105 for AlexNet and 1 × 106 for ResNet34. The hyperparameters were determined based on the comprehensive study of Nayak et al. [16]. To assess the performance of each CNN model, the ROC curve and the AUC on the common test dataset were initially determined through training on the 16 datasets. Fig. 5 shows the relations between the AUC and the number of training images for all CNN models. The sensitivity and the specificity were determined based on the cut-off values determined by using the Youden Index for each ROC curve. Fig. 6 presents an example of an ROC curve and the cut-off values obtained using the AlexNet model for the following number of CXR images: 500, 1000, 2000, and 4000 images. The sensitivity and the specificity for each ROC curve were determined. Fig. 7, Fig. 8 depict the relationship between the sensitivity and the number of training images and between the specificity and the number of training images. Fig. 5, Fig. 7, Fig. 8 show the 95% confidence interval and the fitted nonlinear function, in which parameters a, b, and c were calculated using the open-source Python library used for scientific and technical computing (SciPy; version 1.4.1) with the Levenberg–Marquardt algorism. Table 2 presents parameters a, b, and c.Fig. 5 Relation between the AUC and the number of trained CXR images. Fig. 5 Fig. 6 Example of the ROC curve and the cut-off value defined as the Youden Index. Fig. 6 Fig. 7 Relation between sensitivity and the number of trained CXR images. Fig. 7 Fig. 8 Relation between specificity and the number of trained CXR images. Fig. 8 Table 2 Determined parameters of the nonlinear function: y = a − bx(−c). Table 2Model AUC Sensitivity Specificity a b с A b С a b c AlexNet 1.0072 1.1554 0.5274 1.0643 0.9958 0.2734 1.0176 0.6486 0.3053 Fine-Tuned AlexNet 1.0033 0.0502 0.2715 72.969 72.0594 0.0001 1.0043 1.4673 0.5727 ResNet34 0.9994 70.156 1.0889 115.1554 114.5618 0.0004 0.967 2336.8376 1.6034 Fine-Tuned ResNet34 0.9991 78.6357 1.3567 0.9885 26.3301 1.0009 1.0035 2.94 0.6165 5 Discussion One of the features of AI/ML-based medical devices resides in their ability to learn from real-world data. However, obtaining a large number of training data in the early phase is difficult, and the device performance may change after their first market introduction. To introduce the safety and effectiveness of these devices into the market in a timely manner, an appropriate post-market performance change plan should be established at the timing of the premarket approval, and the real-world performance change must be monitored. In this work, we studied how performance changes when the number of training data is changed. Fig. 9 shows the relations between the AUC and the training sample size and between the increment of the AUC and that of the training sample size by using the nonlinear function obtained from Table 2. All AUCs were rapidly improved as the training data increased and reached an equilibrium state. The 95% confidence interval also decreased as the training data increased. Fang et al. [34] and Samala et al. [35] obtained similar trends for deep learning-based organ auto-segmentation for head-and-neck patients by using CT images and binary classification of malignant and benign masses in digital breast tomosynthesis. Whether the performance of the preapproval stage is in the process of a rapid change or in a steady state must be determined. If the performance is in the process of a rapid change, the manufacturer and regulatory authority should carefully monitor the real-world performance change after the first market introduction.Fig. 9 AUC versus the training sample size and the increment of the AUC versus the increment of the training sample size. Fig. 9 Alternatively, the AUCs for fine-tuned CNNs were better than those for CNNs trained from scratch in all training datasets. This effect is particularly noticeable when the training data are small. Tajbakhsh et al. [36] considered four distinct medical imaging applications in three specialties (radiology, cardiology, and gastroenterology) involving classification, detection, and segmentation under three imaging modalities and compared the performance of deep CNNs trained from scratch and fine-tuned pre-trained CNNs using ImageNet. They concluded that deeply fine-tuned CNNs are useful for medical image analysis, performing equally as CNNs trained from scratch and even outperforming them when limited training data are available. Furthermore, they indicated that the performance gap between deeply fine-tuned CNNs and those trained from scratch widened when the size of training sets was reduced. Our result agrees with their findings, which validates the generality of the mentioned case. Most of the previous studies used publicly available datasets labeled as COVID-19, which are limited to tens to a few hundreds by using data augmentation, transfer learning or combining of datasets. Ran et al. [37] collected over 10 000 CXR images labeled COVID-19 and non-COVID-19 pneumonia from five hospitals and more than 30 clinics by using NLP. In their study, a binary classification was performed by using a modified DenseNet model. The AUC versus the training sample size and the increment of the AUC versus that of the training sample size was confirmed. Their results demonstrated that more than 3000 training samples are needed to achieve an AUC better than 0.90. Moreover, after the training sample size goes beyond 3000, the performance gain with the training sample increase will diminish. Their test dataset was composed of 500 chest radiographs of 500 patients. The COVID-19 and non-COVID-19 pneumonia ratio had the same proportion. A similar trend was observed in our results, where the number of training data for which the gradient disappears was approximately 1600, 400, 2200, and 1100 for AlexNet, Fine-tuned AlexNet, ResNet34, and Fine-tuned ResNet34, respectively (corresponding AUCs: 0.98, 0.99, 0.98, and 0.99, respectively). To generalize our results, additional experiments by using Fine-tuned AlexNet and ResNet34 were conducted to compare the model performance of using the datasets labeled as COVID-19 and Normal with that of using the datasets labeled as COVID-19 and non-COVID-19 pneumonia. Each dataset comprises 2500 CXR images that are randomly selected in the BrixlA and Chest X-ray 14 dataset. The dataset was split as 2000 images for training and validation, and 500 images for testing. Furthermore, training and validation data were selected as 15 different datasets. Fig. 10 shows the relation with AUC for COVID-19 and Normal and COVID-19 and non-COVID-19 pneumonia datasets. Similar trends are observed for both results. The AUC in the equilibrium state of the “pneumonia” dataset was almost the same as that of the “Normal”dataset, indicating good performance. Considering Ibrahim et al. [38], who achieved high performance using automatic detection AlexNet to classify CXR images of COVID-19 and non-COVID-19 viral pneumonia and COVID-19 pneumonia and healthy subjects, our results of binary classification for COVID-19 and Normal can be generalized to classify COVID-19 and non-COVID-19 pneumonia. Additional experiments also indicate that the main reason for the higher performance than Ran et al.’s result is not the difference in the target diseases but the collection of their dataset from multiple medical institutes.Fig. 10 AUC versus the training sample size for COVID-19 and Normal and COVID-19 and non–COVID-19 pneumonia. Fig. 10 Fig. 11 shows the relation of sensitivity, specificity, and training sample size using the nonlinear function obtained from Table 2. The sensitivity and the specificity of AlexNet outperformed ResNet34 with and without the fine-tuning method in the small number of CXR images. In addition, the difference in these performances tended to decrease as the number of training data increased. In other words, if the available data are limited, AlexNet is a more proper model to use compared to ResNet34. D'souza et al. [39] conducted structural analysis and optimization of convolutional neural networks with a small sample size because the number of samples in a dataset can be relatively limited in numerous real-world applications. They trained and tested each structure followed by layer dimension (layer width) optimization using small subsets of these datasets from entirely different data nature (calligraphic, photographic, and microscopic). Their result suggests that “deeper the better” is not always true for CNNs for small datasets, also clearly shows that as the depth increase there is an initial drop in the classification error, but the error soon rises sharply (calligraphic and microscopic). However, this may not always be true, as the microscopic dataset has no clear bias toward deeper or shallower networks. They concluded optimally performing network is largely determined by the data nature. Our result, which 8-layer AlexNet outperformed the 34-layer ResNet34 with the small training dataset, is the same trend of their result especially for calligraphic (The number of training data is 100, 500, and 1000). In the medical fields, comprehensive research is limited in the literature that uses only small datasets without data augmentation, particularly on the relationship between layer and performance. Therefore, further consideration how the number of the layers affects the performance with small number of training data considering target diseases will be needed as a future work. Sensitivity and specificity required by AI/ML-based medical devices vary depending on their intended use. In general practice, high sensitivity is required if the intended use is screening diagnosis. High specificity is required if it is definitive diagnosis. In the real word, the number of available labeled data is limited. Therefore, the appropriate model and method must be selected, and an appropriate post-market performance change plan must be established by considering the intended use and the available real-world labeled data.Fig. 11 Relation between AlexNet and ResNet34 for sensitivity and specificity. Fig. 11 Our study has some limitations. First, our dataset consists of only COVID-19 and Normal, and the ratio was limited to the same proportion. In the real-world data, the training data are expected to include CXR images with many pathologies not limited to COVID-19 and Normal alone. The ratio of COVID-19 and other pathologies, including Normal, may also vary. Therefore, as a future work, we will conduct a study on the performance change when adding many pathologies to the training data and change the ratio in the training data. Second, our study focused on binary classification to classify COVID-19 and Normal. To achieve a more effective classification, it is desirable to validate multi-class classifications, such as COVID-19, other viral, bacteria, and Normal by using CXR images. Although Cohen et al. [40,41] made CXR images labeled as detailed Viral, Bacterial, Fungal, etc. available to the public, the number of images is very limited. These data are continuously being collected from public sources and through indirect collection from hospitals and physicians. Therefore, it is hoped that many of these data will become available in the future. 6 Conclusion Appropriate performance changes must be predicted to manage the performance of medical devices using AI/ML. In this study, we performed binary classification to classify COVID-19 and Normal by using large datasets. In addition, we observed the performance changes (i.e., AUC, sensitivity, and specificity) with the change in the number of training data. In the medical field, comprehensive research is limited in the literature that uses large datasets, particularly on the relationship between performance and training data because building large datasets is costly and burdensome for professionals, and there are concerns about ethical and privacy issues. This paper serves as a fundamental insight for regulators, policy makers, researchers, and manufacturers on how to develop appropriate post-market performance change plans. Declaration of competing interest This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. ==== Refs References 1 Litjens G. Kooi T. Bejnordi B.E. Setio A.A.A. Ciompi F. Ghafoorian M. van der Laak J.A.W.M. van Ginneken B. Sánchez C.I. A survey on deep learning in medical image analysis Med. Image Anal. 42 2017 60 88 Dec 28778026 2 Hamamoto Ryuji Suvarna Kruthi Yamada Masayoshi Kobayashi Kazuma Shinkai Norio Miyake Mototaka Takahashi Masamichi Jinnai Shunichi Shimoyama Ryo Sakai Akira Takasawa Ken Bolatkan Amina Shozu Kanto Ai Dozen Machino Hidenori Takahashi Satoshi Asada Ken Komatsu Masaaki Sese Jun Kaneko Syuzo Application of artificial intelligence technology in oncology: towards the establishment of precision medicine Cancers 12 12 2020 3 Muehlematter Urs J. Daniore Paola Vokinger Kerstin N. Approval of Artificial Intelligence and Machine Learning-Based Medical Devices in the usa and Europe (2015–20): a Comparative Analysis 2021 The Lancet Digital Health 4 U.S. Food and Drug Administration Artificial Intelligence and Machine Learning (Ai/ml) Software as a Medical Device Action Plan 2021 January 5 U.S. Food and Drug Administration Proposed Regulatory Framework for Modifications to Artificial Intelligence/machine Learning (Ai/ml)-based Software as a Medical Device (Samd) - Discussion Paper and Request for Feedback 2019 April 6 World Health Organization Use of Chest Imaging in Covid-19: a Rapid Advice Guide 2020 june 7 American College of RadiologyO Acr Recommendations for the Use of Chest Radiography and Computed Tomography (Ct) for Suspected Covid-19 Infection 2020 MARCH 8 Russakovsky Olga Jia Deng Su Hao Krause Jonathan Satheesh Sanjeev Ma Sean Huang Zhiheng Karpathy Andrej Khosla Aditya Bernstein Michael Imagenet large scale visual recognition challenge Int. J. Comput. Vis. 115 3 2015 211 252 9 Yamashita R. Nishio M. Do R.K.G. Togashi K. Convolutional neural networks: an overview and application in radiology Insights Imaging 9 4 2018 611 629 Aug 29934920 10 Hemdan Ezz El-Din Shouman Marwa A. Karar Mohamed Esmail Covidx-net: A Framework of Deep Learning Classifiers to Diagnose Covid-19 in X-Ray Images vol. 11055 2020 arXiv:2003 11 Ali Narin Kaya Ceren Pamuk Ziynet Automatic Detection of Coronavirus Disease (Covid-19) Using X-Ray Images and Deep Convolutional Neural Networks arXiv 2003 2020 10849 12 Kumar Sethy Prabira Behera Santi Kumari Detection of Coronavirus Disease (Covid-19) Based on Deep Features 2020 MDPI AG 13 Ozturk T. Talo M. Yildirim E.A. Baloglu U.B. Yildirim O. Rajendra Acharya U. Automated detection of COVID-19 cases using deep neural networks with X-ray images Comput. Biol. Med. 121 2020 103792, 06 14 Wang Linda Wong Alexander Covid-net: A Tailored Deep Convolutional Neural Network Design for Detection of Covid-19 Cases from Chest X-Ray Images arXiv 2003 09871 2020 15 Ucar F. Korkmaz D. COVIDiagnosis-Net: deep Bayes-SqueezeNet based diagnosis of the coronavirus disease 2019 (COVID-19) from X-ray images Med. Hypotheses 140 2020 109761 Jul 32344309 16 Nayak S.R. Nayak D.R. Sinha U. Arora V. Pachori R.B. Application of deep learning techniques for detection of COVID-19 cases using chest X-ray images: a comprehensive study Biomed. Signal Process Control 64 2021 102365 Feb 33230398 17 He Kaiming Zhang Xiangyu Ren Shaoqing Sun Jian Deep residual learning for image recognition Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2016 June 18 Krizhevsky Alex Sutskever Ilya Hinton Geoffrey E. Imagenet classification with deep convolutional neural networks Adv. Neural Inf. Process. Syst. 25 2012 1097 1105 19 Rahaman M.M. Li C. Yao Y. Kulwa F. Rahman M.A. Wang Q. Qi S. Kong F. Zhu X. Zhao X. Identification of COVID-19 samples from chest X-Ray images using deep learning: a comparison of transfer learning approaches J. X Ray Sci. Technol. 28 5 2020 821 839 20 Simonyan Karen Zisserman Andrew Very Deep Convolutional Networks for Large-Scale Image Recognition 2014 arXiv:1409.1556 21 Pham Tuan D. Classification of covid-19 chest x-rays with deep learning: new models or fine tuning? Health Inf. Sci. Syst. 9 1 2021 1 11 33235709 22 Szegedy Christian Liu Wei Jia Yangqing Sermanet Pierre Scott Reed Anguelov Dragomir Erhan Dumitru Vincent Vanhoucke Rabinovich Andrew Going Deeper with Convolutions arXiv:1409.4842 2014 23 Iandola Forrest N. Han Song Moskewicz Matthew W. Ashraf Khalid Dally William J. Squeezenet Kurt Keutzer Alexnet-level Accuracy with 50x Fewer Parameters and <0.5mb Model Size 2016 arXiv:1602.07360 24 Wynants L. Van Calster B. Collins G.S. Riley R.D. Heinze G. Schuit E. Bonten M.M.J. Dahly D.L. Damen J.A.A. Debray T.P.A. de Jong V.M.T. De Vos M. Dhiman P. Haller M.C. Harhay M.O. Henckaerts L. Heus P. Kammer M. Kreuzberger N. Lohmann A. Luijken K. Ma J. Martin G.P. McLernon D.J. Andaur Navarro C.L. Reitsma J.B. Sergeant J.C. Shi C. Skoetz N. Smits L.J.M. Snell K.I.E. Sperrin M. Spijker R. Steyerberg E.W. Takada T. Tzoulaki I. van Kuijk S.M.J. van Bussel B. van der Horst I.C.C. van Royen F.S. Verbakel J.Y. Wallisch C. Wilkinson J. Wolff R. Hooft L. Moons K.G.M. van Smeden M. Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal BMJ 369 2020 m1328 04 32265220 25 Roberts Michael Driggs Derek Thorpe Matthew Gilbey Julian Yeung Michael Ursprung Stephan Aviles-Rivero Angelica I. Etmann Christian McCague Cathal Beer Lucian Common pitfalls and recommendations for using machine learning to detect and prognosticate for covid-19 using chest radiographs and ct scans Nat. Mach. Intell. 3 3 2021 199 217 26 Signoroni A. Savardi M. Benini S. Adami N. Leonardi R. Gibellini P. Vaccher F. Ravanelli M. Borghesi A. Maroldi R. Farina D. BS-Net: Learning COVID-19 pneumonia severity on a large chest X-ray dataset Med. Image Anal. 71 2021 102046, 07 27 Vayá Maria de la Iglesia Saborit Jose Manuel Montell Joaquim Angel Pertusa Antonio Bustos Aurelia Cazorla Miguel Galant Joaquin Barber Xavier Orozco-Beltrán Domingo García-García Francisco Bimcv Covid-19+: a Large Annotated Dataset of Rx and Ct Images from Covid-19 Patients arXiv preprint arXiv:2006.01174 2020 28 Wang Lu Wang Hao Xia Chen Wang Yao Tang Qiaohong Li Jiage Zhou Xiao-Hua Toward standardized premarket evaluation of computer aided diagnosis/detection products: insights from fda-approved products Expet Rev. Med. Dev. 17 9 2020 899 918 September 29 Zeiler Matthew D. Fergus Rob Visualizing and understanding convolutional networks European Conference on Computer Vision 2014 Springer 818 833 30 Yosinski Jason Clune Jeff Bengio Yoshua Lipson Hod How Transferable Are Features in Deep Neural Networks? 2014 arXiv:1411.1792 31 Kumar Rajeev Indrayan Abhaya Receiver operating characteristic (roc) curve for medical researchers Indian Pediatr. 48 4 2011 277 287 21532099 32 Hanley J.A. McNeil B.J. A method of comparing the areas under receiver operating characteristic curves derived from the same cases Radiology 148 3 1983 839 843 Sep 6878708 33 Youden William J. Index for rating diagnostic tests Cancer 3 1 1950 32 35 15405679 34 Fang Yingtao Wang Jiazhou Ou Xiaomin Ying Hongmei Hu Chaosu Zhang Zhen Hu Weigang The impact of training sample size on deep learning-based organ auto-segmentation for head-and-neck patients Phys. Med. Biol. 66 18 2021 185012 35 Samala Ravi K. Chan Heang-Ping Hadjiiski Lubomir Helvie Mark A. Richter Caleb D. Cha Kenny H. Breast cancer diagnosis in digital breast tomosynthesis: effects of training sample size on multi-stage transfer learning using deep neural nets IEEE Trans. Med. Imag. 38 3 2018 686 696 36 Tajbakhsh Nima Shin Jae Y. Suryakanth R Gurudu Todd Hurst R. Kendall Christopher B. Gotway Michael B. Liang Jianming Convolutional neural networks for medical image analysis: full training or fine tuning? IEEE Trans. Med. Imag. 35 5 2016 1299 1312 37 Zhang Ran Xin Tie Qi Zhihua Bevins Nicholas B. Zhang Chengzhu Griner Dalton Song Thomas K. Nadig Jeffrey D. Schiebler Mark L. Garrett John W. Diagnosis of coronavirus disease 2019 pneumonia by using chest radiography: value of artificial intelligence Radiology 298 2 2021 E88 E97 32969761 38 Umar Ibrahim Abdullahi Ozsoz Mehmet Serte Sertan Al-Turjman Fadi Shizawaliyi Yakoi Polycarp Pneumonia classification using deep learning from chest X-ray images during Covid-19 Cognitive Computation 2021 1 13 pages 39 D’souza R.N. Huang P.Y. Yeh F.C. Structural analysis and optimization of convolutional neural networks with a small sample size Sci. Rep. 10 1 2020 834 01 31965034 40 Cohen Joseph Paul Morrison Paul Lan Dao Roth Karsten Duong Tim Q. Ghassemi Marzyeh Covid-19 Image Data Collection: Prospective Predictions Are the Future arXiv 2006.11988 2020 41 Cohen Joseph Paul Morrison Paul Dao Lan Covid-19 Image Data Collection 2020 arXiv 2003.11597

35093727

PMC9749084

NO-CC CODE

2022-12-15 23:22:50

no

Comput Biol Med. 2022 Mar 23; 142:105251

utf-8

Comput Biol Med

10.1016/j.compbiomed.2022.105251

oa_other

==== Front Comput Biol Med Comput Biol Med Computers in Biology and Medicine 0010-4825 1879-0534 Elsevier Ltd. S0010-4825(21)00975-6 10.1016/j.compbiomed.2021.105181 105181 Article Horizontal and vertical search artificial bee colony for image segmentation of COVID-19 X-ray images Su Hang a Zhao Dong a∗ Yu Fanhua a Heidari Ali Asghar b1 Zhang Yu a Chen Huiling c∗∗ Li Chengye d Pan Jingye efg Quan Shichao hij∗∗∗ a College of Computer Science and Technology, Changchun Normal University, Changchun, Jilin, 130032, China b School of Surveying and Geospatial Engineering, College of Engineering, University of Tehran, Tehran, Iran c College of Computer Science and Artificial Intelligence, Wenzhou University, Wenzhou, Zhejiang, 325035, China d Department of Pulmonary and Critical Care Medicine, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, 325000, China e Department of Intensive Care Unit, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, Zhejiang, 325000, China f Key Laboratory of Intelligent Treatment and Life Support for Critical Diseases of Zhejiang Provincial, Wenzhou, Zhejiang, 325000, China g Wenzhou Key Laboratory of Critical Care and Artificial Intelligence, Wenzhou, Zhejiang, 325000, China h Department of General Medicine, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, Zhejiang, 325000, China i Department of Big Data in Health Science, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, Zhejiang, 325000, China j Zhejiang Engineering Research Center for Hospital Emergency and Process Digitization, Wenzhou, Zhejiang, 325000, China ∗ Corresponding author. ∗∗ Corresponding author. ∗∗∗ Corresponding author. Department of General Medicine, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, Zhejiang, 325000, China. 1 https://aliasgharheidari.com 3 1 2022 3 2022 3 1 2022 142 105181105181 4 9 2021 20 12 2021 24 12 2021 © 2022 Elsevier Ltd. All rights reserved. 2022 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The artificial bee colony algorithm (ABC) has been successfully applied to various optimization problems, but the algorithm still suffers from slow convergence and poor quality of optimal solutions in the optimization process. Therefore, in this paper, an improved ABC (CCABC) based on a horizontal search mechanism and a vertical search mechanism is proposed to improve the algorithm's performance. In addition, this paper also presents a multilevel thresholding image segmentation (MTIS) method based on CCABC to enhance the effectiveness of the multilevel thresholding image segmentation method. To verify the performance of the proposed CCABC algorithm and the performance of the improved image segmentation method. First, this paper demonstrates the performance of the CCABC algorithm itself by comparing CCABC with 15 algorithms of the same type using 30 benchmark functions. Then, this paper uses the improved multi-threshold segmentation method for the segmentation of COVID-19 X-ray images and compares it with other similar plans in detail. Finally, this paper confirms that the incorporation of CCABC in MTIS is very effective by analyzing appropriate evaluation criteria and affirms that the new MTIS method has a strong segmentation performance. Keywords Disease diagnosis Multi-threshold image segmentation Meta-heuristic COVID-19 Swarm-intelligence ==== Body pmc1 Introduction Image segmentation (IS) is an important research direction in the field of computer vision, which is an important part of image semantic understanding and one of the most difficult problems in image processing [1]. In recent years, more and more image segmentation methods have been proposed [[2], [3], [4], [5]]. For example, multi-level threshold image segmentation (MTIS) [6], wavelet transform-based IS [7], hybrid threshold-based IS [8], multi-atlas-based IS [9], spatial patterns-based IS [10], deep learning-based IS [11], hierarchical clustering-based IS [12], etc. The multi-threshold segmentation method (MTIS) has become a hot technique to study and apply because of its simple implementation, small computational effort, and more stable performance. Peng et al. [13] designed a cell-like P system with the nested structure of three layers to improve MTIS. Zhao et al. [14] proposed an improved ant colony optimizer (RCACO) with a random spare strategy and chaotic intensification strategy to improve the processing efficiency of MTIS. Zhang et al. [15] presented a novel population-based bee foraging algorithm (BFA) to enhance the search efficiency of MTIS. Xing et al. [16] proposed a multi-threshold image segmentation method based on a thermal exchange optimization (TEO) algorithm, used to reduce the algorithm complexity of MTIS. The application of multilevel thresholds to classify image pixels will inevitably improve the segmentation effect of the image. Still, the determination of multilevel thresholds will expand the search space as the number of thresholds itself increases, making the computational complexity much higher. The traditional multi-threshold segmentation method obtains the optimal threshold by exhaustive enumeration, and the computational complexity increases exponentially with the increasing number of thresholds, and the computational efficiency is low. For this reason, scholars have introduced the population intelligence algorithm to find the optimal combination of thresholds, which alleviates the problem of low computational efficiency to a certain extent. Wu et al. [17] presented an improved teaching-learning-based optimization (DI-TLBO) algorithm that increased the stability and accuracy of MTIS. Ewees et al. [18] integrated both the artificial bee colony algorithm and the sine-cosine algorithm (ABCSCA) to enhance the performance of MTIS. Alwerfali et al. [19] proposed a modified spherical search optimizer (SSOSCA) to find the appropriate threshold in MTIS. Abd Elaziz et al. [20] presented an enhanced Harris Hawks Optimizer (HHOSSA) to determine the best threshold of MTIS. Table 1 illustrates the image segmentation method proposed in this paper and compares recent related image segmentation techniques. As we can see, more and more scholars in the field of image processing have paid attention to MTIS in recent years, and many researchers have proposed the use of swarm intelligence optimization algorithm (SIOA) instead of the traditional exhaustive method to find the optimal threshold in order to reduce the computational effort and complexity of traditional multi-threshold image segmentation methods.Table 1 Comparison of work related to image segmentation. Table 1Authors Method Dataset Performance and Contribution The method proposed in this paper CCABC + non-local means + 2D histogram+ 2D Kapur's entropy COVID-19 X-ray images The method has a high threshold search accuracy and image segmentation effect among the same type of segmentation methods tested with real datasets. Peng et al. [13] Membrane computing + fuzzy entropy standard images Membrane computing enhances the accuracy and efficiency of the fuzzy entropy-based segmentation method. Zhao et al. [14] RCACO + Kapur's entropy standard images This method improves the classic Kapur's entropy thresholding segmentation method's segmentation consistency and accuracy. Zhang et al. [15] improved BFA + Otsu benchmark images This method effectively uses image processing, and the target area segmentation is more complete. Xing et al. [16] TEO + gray-level co-occurrence matrix satellite images This method improves the accuracy and robustness of traditional image segmentation methods. Wu et al. [17] DI-TLBO + Otsu + Kapur's entropy X-ray images This method improves the stability of the threshold segmentation method using SIOA and has a better performance among similar methods. Ewees et al. [18] ABCSCA + Otsu benchmark images This method has a stronger threshold search capability and significantly improves the speed and accuracy of image segmentation. Alwerfali et al. [19] SSOSCA + fuzzy entropy Berkeley dataset This method uses SIOA combined with fuzzy entropy to significantly improve the image segmentation method. Abd Elaziz et al. [20] HHOSSA + Kapur's entropy natural gray-scale images This method shows the performance and accuracy of the algorithm by testing on natural datasets. The swarm intelligence optimization algorithm is a random search evolutionary algorithm based on probability [[21], [22], [23], [24]]. These optimizers have found an increasing momentum in the current big data world because they can help decision-makers understand the aspects of their decisions in a more resource-aware way [25,26]. It is an algorithm abstracted by simulating the intelligent collaborative behavior of groups such as insects, herds of animals, and fish [[27], [28], [29], [30], [31]]. Moreover, SIOA is simple and efficient in terms of optimization. And these SIOAs have found its application in many fields such as feature selection [[32], [33], [34], [35], [36]], wind speed prediction [37], engineering design problems [38,39], image segmentation [40], social support domains [[41], [42], [43], [44]], industrial domain [45], the hard maximum satisfiability problem [46,47], disaster prediction [48], renewable energy prediction [49], medical data classification [[50], [51], [52], [53], [54]], complex network [55], bankruptcy prediction [[56], [57], [58]], parameter optimization [[59], [60], [61]], PID optimization control [[62], [63], [64]], fault diagnosis of rolling bearings [65,66], design of power electronic circuit [67,68], detection of foreign fiber in cotton [69,70], and prediction problems in educational field [[71], [72], [73], [74], [75]]. For example, there are particle swarm optimization (PSO) [76], moth search algorithm (MSA) [77], monarch butterfly optimization (MBO) [78], hunger games search (HGS)2 [79], Harris hawks optimization (HHO)3 [80], slime mould algorithm (SMA)4 [81], Runge Kutta optimizer (RUN)5 [82], colony predation algorithm (CPA) [83] and so on. In 2005, Karaboga et al. proposed an SIOA called the artificial bee colony (ABC) [84] algorithm, which has a strong ability to solve the optimal solution. The standard ABC algorithm still suffers from the deficiencies of low optimality finding accuracy and slow convergence speed, especially in solving high-dimensional complex problems. There is still much room for improvement in the algorithm's optimality finding stability and adaptability to jump away from local optimum ability. Therefore, researchers have developed many improved versions of ABC and applied them to various fields. Zou et al. [85] presented a discrete ABC algorithm to address a new multiple automatic guided vehicle dispatching problem. Zhou et al. [86] proposed a dynamic ABC algorithm for a cloud service optimization method. Zhao et al. [87] proposed an improved ABC optimization algorithm to solve the problem of classifying hyperspectral images. Lin et al. [88] proposed a hybrid binary ABC algorithm to solve the maximum set k-covering problem. Ding et al. [89] presented a modified ABC algorithm to improve structural damage identification. Chu et al. [90] proposed a binary superior tracking ABC with dynamic Cauchy mutation to improve the feature selection algorithm. Chen et al. [91] proposed a fireworks explosion-based ABC framework to solve the problem of slow convergence of the ABC algorithm. Boudardara et al. [92] proposed a shrinking ABC to solve a sub-problem of robotic path planning. Yue et al. [93] proposed an improved ABC algorithm to enhance wireless sensor network coverage and connectivity. This paper proposes a new and improved version of ABC by introducing a vertical and horizontal crossover strategy in ABC to improve ABC's efficiency in searching for optimal solutions and the ability to jump out of locally optimal solutions. Further, CCABC is applied to the field of multi-threshold image segmentation to improve image segmentation accuracy while ensuring segmentation efficiency. To demonstrate that CCABC has a better ability to jump out of local optimum and can obtain higher quality solutions more efficiently, this paper conducts comparative experiments on CCABC using 30 test functions of CEC2014. In the algorithm performance experiments, CCABC, original ABC, 5 well-known algorithms, 4 improved algorithms based on original ABC, and 6 other improved algorithms are compared. Furthermore, the comparison results are also analyzed in this paper using Wilcoxon signed-rank test [94] and the Freidman test [95], proving that CCABC outperforms ABC and other similar algorithms. Then, CCABC is likewise compared with multiple similar algorithms in the image segmentation experiments. This paper also utilizes Feature Similarity Index (FSIM) [96], Peak Signal to Noise Ratio (PSNR) [97], and Structural Similarity Index (SSIM) [98] to image segmentation results carefully evaluated. The evaluation results fully demonstrate that the MTIS method based on CCABC does not easily fall into the local optimum trap and thus obtains higher quality segmentation results. The main contributions of this study can be summarized as:● This paper proposes an improved ABC algorithm (CCABC). ● CCABC is applied to COVID-19 multi-threshold image segmentation based on the 2D histogram. ● CCABC has a significant improvement in searching for the optimal solution. ● The segmentation performance of CCABC is verified by comparison with well-known algorithms. The rest of this paper is organized as follows. In Section 2, the work related to image segmentation is described. In Section 3, the original ABC algorithm is described in this paper. In Section 4, this paper mainly describes the proposed CCABC and the improved MTIS (CCABC-MTIS) based on CCABC. In Section 5, comparative experiments on benchmark functions and image segmentation problems are conducted to verify the performance of CCABC. Finally, Section 6 concludes the whole paper and shows the direction of future work. 2 The related works of image segmentation 2.1 Multi-thresholding image segmentation MTIS is a vital image segmentation method, which essentially marks targets with different features in an image with multiple thresholds. Therefore, the most crucial aspect of MTIS is setting the thresholds, which directly determines the effect of the image after segmentation. Among the many MTIS methods, the histogram-based segmentation method is one of the most widely used methods by researchers. The histogram-based segmentation methods can be divided into one-dimensional histogram segmentation and two-dimensional histogram segmentation. In contrast, the one-dimensional histogram-based segmentation methods do not make full use of the spatial information of the image, and the segmentation results are easily disturbed by noise. Abutaleb [99] proposed a two-dimensional histogram-based segmentation method by making full use of the spatial information of the image to combine grayscale images with local mean images. The 2D histogram formed based on grayscale image and local mean image ignores the details of some points and edges, and the traditional 2D histogram uses the exhaustive method to find the optimal threshold, which is very computationally intensive. Therefore, this paper generates a two-dimensional histogram using grayscale images and nonlocal mean images and then introduces SIOA based on Kapur's entropy to help MTIS find the optimal threshold. In the method proposed in this paper, Kapur's entropy is used as the objective function of CCABC, and the efficiency of the swarm intelligence optimization algorithm is used to find the optimal threshold, which greatly reduces the complexity. The detailed flowchart of the method is given in Fig. 1 .Fig. 1 Flowchart of MTIS Fig. 1 2.2 Nonlocal means 2D histogram A new image denoising technique called Nonlocal means 2D histogram was proposed by Buades et al. [100]. It uses redundant information in the image for denoising while maintaining the maximum detailed features of the image. And Nonlocal mean value of pixels is obtained from pixels with similar neighborhood structures in the image, weighted by averaging. Assuming that the grayscale values of pixels p and q of image I are I(p) and I(q), respectively, the nonlocal mean value of image I can be calculated by Eqs. (1)–(4).(1) O(p)=∑q∈II(q)ω(p,q)∑q∈lω(p,q) (2) ω(p,q)=exp−|μ(p)−μ(q)|2σ2 (3) μ(p)=1m×m∑i∈L(p)I(i) (4) μ(q)=1m×m∑i∈L(q)I(i) where O(p) is the filtered value of the nonlocal mean, ω(p,q) is used to calculate the weights of pixel p and pixel q, σ is the standard deviation, μ(p) and μ(q) are the local means, and also L(p) and L(q) are local images of size m×m centered on pixel p and pixel q, respectively. The two-dimensional histogram is formed by combining the grayscale image based on the above method to form the nonlocal mean image. Therefore, the size and magnitude taking range [0,L−1] of the grayscale image G(x,y) must be the same as the size and magnitude taking range [0,L−1] of the nonlocal mean image D(x,y). Then, the final two-dimensional histogram can be obtained after normalizing Eq. (5), which is given in Fig. 2 to generate a two-dimensional histogram based on grayscale images and nonlocal mean images, where images X and Y are from the COVID-19 dataset.(5) Pij=h(i,j)M×N where i represents the value of G(x,y) pixels, j represents the value of D(x,y) pixels, and h(i,j) denotes the number of times the point (i,j) appears on the gray value vector (s,t).Fig. 2 Three-dimension view about 2D histograms of X and Y. Fig. 2 2.3 Kapur's entropy for 2D histogram A two-dimensional planar histogram based on the above two-dimensional histogram is given in Fig. 3 , where {t1,t2…,L−1} denotes the value of the grayscale image and {s1s2…,L−1} denotes the value of the nonlocal average image. Because the main diagonal of the two-dimensional histogram contains enough image information and to simplify the calculation, Kapur's entropy of the subregion on the main diagonal is calculated using Eq. (6). Since CCABC takes Kapur's entropy as the objective function, the optimal solution found by CCABC in {t1,t2…,tn−1} is the optimal threshold.(6) φ(s,t)=−∑i=0s1∑j=0t1PijP1lnPijP1−∑i=t1+1s2∑j=t1+1t2PijP2lnPijP2−…∑i=sL−2sL−1∑tL−2tL−1PijPL−1lnPijPL−1 where P1=∑i=0S1∑j=0t1Pij,P2=∑i=t1+1S2∑j=t1+1t2Pij,PL−1=∑i=sL−2+1SL−1∑j=tL−2+1tL−1Pij.Fig. 3 Two-dimension view about the 2D histogram. Fig. 3 3 An overview of ABC Karaboga et al. [84] proposed the artificial bee colony algorithm (ABC) to solve the multivariate function optimization problem, which simulates the honey harvesting behavior of a bee colony. Like other SIOA, the optimization execution process of ABC is divided into two phases, namely the exploration phase and the exploitation phase. The bees are divided into employed bees, onlooker bees, and scout bees according to the division of labor, where employed bees and scout bees correspond to the exploration phase of the algorithm and onlooker bees correspond to the exploitation phase. In ABC, the bees are divided into three categories: employed bees, onlooker bees, and scout bees, whose main behavior is finding a good food source and exploiting a certain food source. The employed bees are equal to the onlooker bees in terms of number. In addition, the location of each food source represents a candidate solution, which is a vector of dimension D in the algorithm. In contrast, the amount of honey each food source has corresponds to the quality of the problem solution. The process of the ABC algorithm can be divided into four phases: initialization phase, employed bee phase, onlooker bee phase, and scout bee phase. The algorithm model generates N food sources randomly in the initialization phase, as shown in Eq. (7).(7) xi,j=xmin,j+rand(0,1)(xmax,j−xmin,j) Where k∈{1,2,⋯,N},j∈{1,2,⋯,D}, xmin,j and xmax,j denote the minimum and maximum values of the randomly generated food source dimensions, and rand(0,1) denotes a random number uniformly distributed between [0,1]. In this stage and initializing the food sources, the adaptation value is calculated for each food source, and the formula is shown in Eq. (8).(8) fiti={11+fifi≥01+|fi|otherwise where fiti denotes the weight of food source xi and fi denotes the value of the objective function of the corresponding problem with the food source xi as a parameter. In the employed bee phase, each employed bee will be responsible for searching a food source and then generating a new candidate solution by searching around that food source by Eq. (9).(9) vi,j=xi,j+φi,j(xi,j−xk,j) where k∈{1,2,⋯,N},j∈{1,2,⋯,D}, xi denotes the selected food source to be updated, xk denotes the food source different from xi, j denotes the dimension of random selection, and φi,j denotes the random number generated from [−1,1]. After updating the food source, the newly generated candidate solution vi is greedily selected with xi for iterating the optimal food source. In the onlooker bee phase, the onlooker bee informs the employed bee of the information of the food source nectar source by dancing and then calculates the selection probability of the relevant food source based on the information obtained formula shown in Eq. (10).(10) pi=fiti∑j=1Nfiti where pi denotes the probability of being selected for a food source xi and fiti denotes the adaptation value of the food source. It can be seen that the larger the fit value is, the higher the probability that the corresponding food source is selected. Based on this, the food source is selected by roulette, and the same Eq. (9) is used to generate a new candidate solution vi for the selected individuals. vi is replaced if its fitness value is better than xi; otherwise, it remains unchanged. In the scout bee phase, the algorithm model will select the food source i that has not been updated for the longest time, and if its number of not being updated is greater than the predefined limit value, the corresponding employed bee will be converted into a scout bee, and a new food source will be randomly generated to replace food source i by Eq. (7). The pseudocode of ABC is shown in Algorithm 1.Algorithm 1 Pseudo-code of ABCImage 1 4 The proposed CCABC-MTIS method This section describes the vertical crossover optimizer Crisscross optimizer (CSO) and the proposed CCABC in detail. Among them, this paper uses CSO's horizontal crossover search (HCS) and vertical crossover search (VCS) to improve the early search efficiency of the ABC algorithm and the ability to jump out of the local optimum. After that, this paper introduces the improved ABC algorithm into the multi-threshold image segmentation technique to propose the CCABC-MTIS method. 4.1 Crisscross optimizer In 2014, Meng et al. [101] proposed CSO, which mainly used HCS and VCS to complete the whole search process, and experimentally proved that this not only made the algorithm less prone to fall into the local optimum problem but also significantly improved the convergence speed and solution accuracy of the algorithm. After that, Patwal et al. [102] proposed an improved PSO algorithm based on HCS and VCS to improve the ability of PSO search and exploitation. Zhao et al. [103] proposed to use HCS and VCS to solve a variant of ant colony optimization (ACOR) which is prone to fall into local optimum and The problem of low convergence accuracy. Inspired by this, this paper introduces HCS and VCS into ABC to be able to improve its search and exploitation capabilities. 4.1.1 Horizontal crossover search (HCS) HCS mainly acts between two different individuals and crosses the corresponding two individual dimensions arithmetically. Therefore, adding HCS to the ABC algorithm enables the bees to learn from each other and exchange information, thus enabling them to find the optimal solution quickly, thereby improving the convergence speed of the algorithm. Assuming that the n th column dimension of the parent bees xi and xj performs HCS, then the lateral crossover update formula can be expressed as Eq. (11) and Eq. (12).(11) MSin=ε1×xin+(1−ε1)×xjn+c1×(xin−xjn) (12) MSjn=ε2×xjn+(1−ε2)×xin+c2×(xjn−xin) where ε1 and ε2 are random numbers between (0,1), c1 and c2 are random numbers between (−1,1), xin is the n th dimension of the i th bee, xjn is the n th dimension of the j th bee, and MSin and MSjn are the offspring generated by the parent bees xi and xj based on HCS. 4.1.2 Vertical crossover search (VCS) The VCS mainly acts between two different dimensions, arithmetically crossing the corresponding individuals. This may allow some individuals who fall into a local optimum to continue participating in the search, and the individuals who perform the search normally remain as unchanged as possible. VCS is reflected in the ABC algorithm by swapping the dimensions of the bee's positions so that the ABC algorithm does not easily fall into a local optimum. Assuming that the dimension of the nth position of the bee performs VCS, the vertical cross update formula can be expressed as Eq. (13).(13) MSim=ε×xim+(1−ε)×xin where ε is a random number between (0,1), xim and xin are the m th and n th dimensions of the i th bee, and MSim is the offspring of bee xi based on VCS. 4.2 The proposed CCABC The CCABC proposed in this paper is mainly to improve the ability of the original ABC to jump out of the local optimum and the search ability in the early stage, and finally to achieve the purpose of improving the algorithm's ability to find the best. Therefore, this paper proposes a new CCABC algorithm based on the HCS and VCS. In CCABC, after the algorithm performs population initialization, HCS and VCS start to participate in the population iteration to help individuals perform a fast search in the early stage and jump out of local optimum in the middle stage while giving full play to the strong exploitation capability of the ABC algorithm itself in the late iteration to improve the overall performance of the original algorithm. The flow chart of CCABC is shown in Fig. 4 .Fig. 4 Flowchart of CCABC Fig. 4 The pseudocode of CCABC is shown in Algorithm 2.Algorithm 2 Pseudo-code of CCABCImage 2 The complexity of CCABC mainly includes introducing HCS and VCS, employing bees to search for food sources, onlooker bees to develop food sources, and calculating fitness values. First, the complexity level of HCS and VCS is O(T*N∗D+T∗N2). Then, the complexity level of employed bees searching for food sources is O(T∗N). The complexity level of the onlooker bee to exploit the food source is O(T∗N+T∗F), which can be abbreviated as O(T∗N). Finally, the complexity level for both the search and the exploitation of food source fitness values calculation is O(N∗logN). Therefore, the overall complexity level of the CCABC algorithm is O(CCABC)=O((D+N)∗T∗N2∗logN). 5 Experiments and results This section focuses on verifying the algorithm performance of CCABC and evaluating the image quality after segmentation using CCABC-MTIS. To show that CCABC has a better ability to jump out of local optimum and upfront search capability, this paper compares CCABC with six well-known optimization algorithms and six improved algorithms using 30 benchmark functions. Immediately after, to demonstrate the better segmentation performance of CCABC-MTIS in multilevel image segmentation, this paper compares CCABC-MTIS with nine peer methods and evaluates the segmentation results in detail using PSNR, SSIM, and FSIM. 5.1 Experiment setup In the experimental part of the benchmark functions, 30 benchmark functions from CEC2014 are used in this paper to demonstrate the algorithmic performance of CCABC, and Table 2 represents the details of benchmark functions F1–F30. The test value of the benchmark function is the default value.Table 2 Description of the 30 benchmark functions. Table 2Class No. Functions Fi∗=Fi(x∗) Unimodal Functions 1 Rotated High Conditioned Elliptic Function 100 2 Rotated Bent Cigar Function 200 3 Rotated Discus Function 300 Simple Multimodal Functions 4 Shifted and Rotated Rosenbrock's Function 400 5 Shifted and Rotated Ackley's Function 500 6 Shifted and Rotated Weierstrass Function 600 7 Shifted and Rotated Griewank's Function 700 8 Shifted Rastrigin's Function 800 9 Shifted and Rotated Rastrigin's Function 900 10 Shifted Schwefel's Function 1000 11 Shifted and Rotated Schwefel's Function 1100 12 Shifted and Rotated Katsuura Function 1200 13 Shifted and Rotated HappyCat Function 1300 14 Shifted and Rotated HGBat Function 1400 15 Shifted and Rotated Expanded Griewank's plus Rosenbrock's Function 1500 16 Shifted and Rotated Expanded Scaffer's F6 Function 1600 Hybrid Functions 17 Hybrid Function 1 (N = 3) 1700 18 Hybrid Function 2 (N = 3) 1800 19 Hybrid Function 3 (N = 4) 1900 20 Hybrid Function 4 (N = 4) 2000 21 Hybrid Function 5 (N = 5) 2100 22 Hybrid Function 6 (N = 5) 2200 Composition Functions 23 Composition Function 1 (N = 5) 2300 24 Composition Function 2 (N = 3) 2400 25 Composition Function 3 (N = 3) 2500 26 Composition Function 4 (N = 5) 2600 27 Composition Function 5 (N = 5) 2700 28 Composition Function 6 (N = 5) 2800 29 Composition Function 7 (N = 3) 2900 30 Composition Function 8 (N = 3) 3000 In addition, to ensure that the experiments are as fair as possible based on optimization and machine learning works [[104], [105], [106], [107]], all the algorithm comparison experiments in this paper are done under the same conditions as below, with the settings in Table 3 . The paper was analyzed using the mean, variance, Wilcoxon signed-rank test, and Freidman test for the experimental results.Table 3 Parameter settings for benchmark function experiments. Table 3Parameter name Value Population size 30 Maximum number of function evaluations 300,000 Number of tests per algorithm 50 In the experimental part of MTIS, segmented images A, B, C, D, E, F, G, and H are from the COVID-19 dataset. The COVID-19 disease is in the center of attention in recent years [[108], [109], [110]]. This dataset is from a publicly developed GitHub dataset containing chest X-rays and CT images of patients positive or suspected of having COVID-19 or other viral and bacterial pneumonia (MERS, SARS, and ARDS.). All images and data were collected through specialized physicians at each hospital and are under specified licenses. The dataset is publicly available at https://github.com/ieee8023/covid-chestxray-dataset. It has been studied in several works [111,112]. In Fig. 5 shows their original images and the nonlocal average 2D histogram, respectively. To ensure the fairness of the experiment, all the algorithms involved in the comparison were tested under the same conditions. In addition, to accurately reflect each algorithm's performance at different threshold levels, this paper covers both low and high threshold levels as much as possible when setting the threshold levels. In the experiments, the 3, 5, and 7 thresholds represent the low threshold levels, and these 12, 15, and 18 thresholds represent the high threshold levels.Fig. 5 Samples of the segmented images. Fig. 5 Finally, the experimental parameter settings for all algorithms in this paper are default values, and detailed parameter information is given in Table A1 in Appendix A. The experiments were conducted uniformly on the Windows Server 2008R2 operating system to ensure the same environment for all experiments. The main hardware of the device is an Intel(R) Xeon(R) CPUE5-2660v3 (2.60 GHz) and 16 GB of RAM, and the code running software is Matlab2017b. 5.2 CCABC benchmark function validation 5.2.1 Comparison of CCABC and excellent algorithms In this section, CCABC is compared with six basic algorithms and six improved algorithms based on 30 benchmark functions, where the basic algorithms include sine cosine algorithm (SCA) [113], whale optimizer (WOA) [113], salp swarm algorithm (SSA) [113], moth-flame optimization (MFO) [113], multi-verse optimizer (MVO) [113], shortly shown by ABC, SCA, MFO, PSO, SSA, and HHO. Also, the improved algorithms include Nonlinear based chaotic HHO (NCHHO) [114], chaotic BA (CBA) [115], SCA with differential evolution [116], enhanced GWO with a new hierarchical structure (IGWO) [58], improved WOA (IWOA) [117], A-C parametric WOA (ACWOA) [118], adaptive SCA integrated with PSO (ASCA_PSO) [119], shortly all of them shown by NCHHO, CBA, SCADE, IGWO, ACWOA, and ASCA_PSO. The experimental results of comparing CCABC with other peer algorithms are given in Table A2, where AVG denotes the mean value of the algorithm after 50 independent runs and STD denotes its variance. We can initially see that CCABC has the smallest mean and variance in most benchmark function tests by observing the mean and variance. When using CCABC and similar algorithms for benchmark functions, CCABC obtains high-quality solutions and has a high degree of generalizability in optimizing the benchmark function. In Table 4 , this paper further analyzes the benchmark function optimization effect between CCABC and other classes of algorithms using Wilcoxon signed-rank test, where '+' indicates that CCABC outperforms other algorithms, '-' indicates that CCABC underperforms other algorithms, ' = ' indicates that CCABC's performance is equal to other algorithms, Mean indicates the level of overall mean, and Rank denotes the ranking result of the overall mean. Fig. 6 represents a further analysis of the comparative results of the benchmark functions using the Freidman test. The Wilcoxon signed-rank test and Freidman test analysis show that CCABC has the first performance in optimizing both benchmark functions.Table 4 Comparison results of CCABC and excellent algorithms. Table 4Item CCABC ABC SCA MFO PSO SSA HHO +/−/ = ∼ 17/8/5 28/1/1 29/1/0 25/3/2 23/6/1 23/5/2 Mean 2.6 3.0 10.7 9.6 6.5 4.2 5.5 Rank 1 2 12 11 6 3 5 Item ∼ NCHHO CBA SCADE IGWO ACWOA ASCA_PSO +/−/ = ∼ 22/7/1 26/2/2 26/3/1 27/3/0 27/2/1 28/2/0 Mean ∼ 6.6 7.9 10.7 5.3 9.5 8.3 Rank ∼ 7 8 13 4 10 9 Fig. 6 Friedman test results of CCABC and excellent algorithms. Fig. 6 In addition, Fig. 7 shows the convergence curves of CCABC and other similar algorithms on different functions. The convergence curves shown show that CCABC obtains better quality solutions than the other algorithms when optimizing the benchmark functions F6, F11, and F16. However, CCABC converges slightly slower than the other algorithms in the first stage. In optimizing high-quality solutions for F3, F9, F11, and F16, the algorithm has obvious inflection points for jumping out of the local optimum. In optimizing F19 and F25, it can be seen that CCABC obtains high-quality solutions and converges faster than other algorithms. Through the above analysis of the benchmark function results, CCABC can jump out of local optimum solutions, have a solid ability to obtain high-quality solutions and fast convergence speed. Therefore, CCABC is a superior method for optimizing the benchmark function and a very good SIOA.Fig. 7 Convergence curves of CCABC and excellent algorithms. Fig. 7 5.2.2 Analysis of CCABC in the iterative process The proposed CCABC algorithm mainly introduces the horizontal search mechanism and the vertical search mechanism into ABC. Therefore, this section focuses on the performance of CCABC and ABC in the process of optimizing benchmark functions. Fig. 8 shows the results of the qualitative inspection of 23 benchmark functions by CCABC and ABC. The graph in the first column (a) exposes the 3D location distribution of the CCABC search history. The graph in the second column (b) discloses the two-dimensional location distribution of CCABC search history. The graph in the third column (c) discloses the trajectories of CCABC individuals in the first dimension. The graph in the fourth column (d) reveals the average fitness of CCABC, and the graph in the fifth column (e) reveals the convergence curves of CCABC and ABC. Fig. 8 (a) and (b) record the positions and distribution of individuals searching or developing in each iteration. It can be seen that most of the individual search positions are near the optimal solution, and a small number of positions are scattered in the space. This means that after the individuals in the population have searched most of the space, CCABC can still find the location of the optimal solution during the convergence process. In Fig. 8 (c), we can see that the values of the first dimension in the individuals are more volatile and random in the early stage and more smoothly rational in the later stage, which indicates that CCABC has good randomness in the early stage to traverse the space as much as possible, while it does not change much in the later stage of the algorithm and favors exploitation. Fig. 8 (d) shows the average fitness curve, from which it can be seen that the fitness range is wide, and the curve decreases rapidly with the increase of the number of iterations, indicating that CCABC has good convergence characteristics. In addition, it is obvious in the figure that there is a move to expand the search range in the middle of the algorithm to jump out of the local optimum, especially F8 successfully jumps out of the position of the previous local optimum solution. Fig. 8 (e) shows that CCABC is able to find higher-quality solutions faster than ABC.Fig. 8 (a) 3D location distribution of CCABC, (b) 2D location distribution of CCABC, (c) Trajectory of CCABC in the first dimension, (d) Average fitness of CCABC, (e) Convergence curves of CCABC and ABC. Fig. 8 The balance and diversity of CCABC and ABC were analyzed on 30 functions of CEC2014, and the performance of CCABC was further investigated. Fig. 9 (a) and (b) show the equilibrium analysis of CCABC and ABC, and Fig. 9 (c) shows the diversity analysis of CCABC and ABC. We added increment-decrement curves in Fig. 9 (a) and (b). When the value of the mining result is greater than or equal to the mining result, the curve increases. Otherwise, it decreases. When it has a negative value, it is set to zero. Thus, its high value represents extensive search activity, and its low value represents strong exploitation activity. In addition, the duration of the high or low value in the graph reflects the sustained effect of search or exploitation in the search for excellence. The increment-decrement curve is maximized when the search effect is at the same level as the exploitation effect. In Fig. 9 (c), the x-axis represents the number of iterations, and the y-axis represents the diversity. We can see that the population diversity is extensive at the beginning due to random initialization and gradually decreases with the number of iterations. Of course, we also know through the above that the algorithm does not just perform one extensive search, so the diversity of the population fluctuates throughout the search process, especially when many individuals do not find the optimal solution during the exploitation process. Individuals regenerate completely new solutions, such as F15 and F21 in Fig. 9 (c).Fig. 9 (a) The balance analysis of CCABC, (b) The balance analysis of ABC, (c) The diversity analysis of CCABC and ABC. Fig. 9 In the metaheuristic algorithm, the search process and the exploitation process affect each other and can influence the final result of the optimal search, so it is necessary to consider the balance between these two. In Fig. 9 (a) and (b), by observing the incremental-decremental curves, it is evident that the search and exploitation results of CCABC reach superiority significantly earlier than ABC, which indicates that CCABC is able to find a better solution faster in the search phase and can be developed quickly. In addition, it can be seen by the separate search and exploitation curves that CCABC search and exploitation effects are significantly stronger than ABC. 5.2.3 Comparative analysis of CCABC and ABC variants In this section, CCABC is compared with four variants of the ABC algorithm based on 30 benchmark functions, including DEMABC [120], GAABC [121], ISABC [122] and NNMABC [123]. The experimental results of CCABC compared with the variants of the ABC algorithm are shown in Table 5 , where AVG denotes the mean value of the algorithm after 50 independent runs and STD denotes the variance of the algorithm. By looking at the mean and variance, we can initially see that CCABC has the smallest mean and variance in most of the benchmark function tests. Further, In Table 6 , this paper further analyzes the benchmark function optimization effect between CCABC and a variant of the ABC algorithm using the Wilcoxon signed-rank test. can be seen that CCABC is ranked No. 1 and there is a statistically significant difference between them.Table 5 Comparison of CCABC and ABC variants with Wilcoxon test and Friedman test. Table 5Fun Item CCABC DEMABC GAABC ISABC NNMABC F1 AVG 2.747E+06 1.300E+08 8.951E+07 1.278E+08 9.722E+07 STD 1.006E+06 2.519E+06 1.413E+07 4.222E+07 1.715E+08 F2 AVG 2.626E+03 3.244E+08 1.082E+10 2.869E+10 5.952E+09 STD 2.820E+03 3.711E+03 1.126E+09 8.820E+09 1.334E+10 F3 AVG 6.641E+02 3.433E+04 2.815E+04 5.127E+04 2.837E+04 STD 4.618E+02 3.069E+02 1.598E+04 1.383E+04 4.106E+04 F4 AVG 4.860E+02 4.071E+02 1.580E+03 1.268E+03 2.142E+03 STD 2.527E+01 1.688E+01 1.733E+02 1.912E+02 4.632E+03 F5 AVG 5.202E+02 5.200E+02 5.206E+02 5.210E+02 5.203E+02 STD 4.176E-02 2.075E-03 3.925E-02 5.065E-02 2.440E-01 F6 AVG 6.095E+02 6.319E+02 6.250E+02 6.417E+02 6.285E+02 STD 2.425E+00 2.498E+00 1.209E+00 1.411E+00 4.762E+00 F7 AVG 7.000E+02 7.325E+02 7.789E+02 1.102E+03 7.786E+02 STD 6.377E-05 3.918E-02 9.983E+00 8.340E+01 1.655E+02 F8 AVG 8.000E+02 9.676E+02 9.442E+02 1.296E+03 9.108E+02 STD 7.263E-14 2.253E+01 9.153E+00 2.909E+01 7.433E+01 F9 AVG 9.498E+02 1.185E+03 1.099E+03 1.494E+03 1.154E+03 STD 7.814E+00 1.269E+01 1.223E+01 4.721E+01 7.630E+01 F10 AVG 1.001E+03 3.491E+03 4.251E+03 7.701E+03 4.090E+03 STD 8.370E-01 2.844E+02 1.926E+02 3.865E+02 1.479E+03 F11 AVG 2.796E+03 5.353E+03 6.105E+03 8.748E+03 5.534E+03 STD 2.077E+02 3.246E+02 3.169E+02 3.448E+02 1.002E+03 F12 AVG 1.200E+03 1.200E+03 1.201E+03 1.203E+03 1.201E+03 STD 3.469E-02 5.825E-02 9.906E-02 3.699E-01 3.826E-01 F13 AVG 1.300E+03 1.301E+03 1.303E+03 1.311E+03 1.302E+03 STD 2.932E-02 1.609E-02 1.479E-01 1.118E+00 2.278E+00 F14 AVG 1.400E+03 1.400E+03 1.436E+03 1.685E+03 1.421E+03 STD 1.793E-02 9.209E-03 3.715E+00 3.466E+01 5.774E+01 F15 AVG 1.503E+03 1.514E+03 2.461E+03 6.743E+03 5.995E+05 STD 4.909E-01 2.558E+00 2.992E+02 2.940E+03 1.835E+06 F16 AVG 1.609E+03 1.613E+03 1.611E+03 1.614E+03 1.613E+03 STD 3.938E-01 4.817E-01 2.274E-01 1.693E-01 4.591E-01 F17 AVG 6.249E+05 1.885E+06 4.612E+06 8.973E+06 4.446E+06 STD 3.220E+05 7.342E+05 1.368E+06 2.772E+06 1.163E+07 F18 AVG 2.559E+03 2.108E+03 7.078E+07 8.338E+07 1.125E+08 STD 9.569E+02 4.348E+02 2.149E+07 3.711E+07 4.591E+08 F19 AVG 1.905E+03 1.913E+03 1.977E+03 2.034E+03 1.991E+03 STD 9.290E-01 7.154E-01 1.722E+01 3.424E+01 1.427E+02 F20 AVG 7.138E+03 2.227E+04 5.849E+03 1.596E+04 5.796E+04 STD 2.144E+03 1.684E+03 1.480E+03 4.704E+03 2.870E+05 F21 AVG 1.064E+05 2.746E+05 1.129E+06 6.199E+06 1.455E+06 STD 5.625E+04 3.241E+04 3.775E+05 3.034E+06 3.336E+06 F22 AVG 2.380E+03 1.297E+04 2.829E+03 2.803E+03 2.837E+03 STD 7.395E+01 9.172E+01 1.118E+02 1.107E+02 1.909E+02 F23 AVG 2.615E+03 2.632E+03 2.668E+03 2.789E+03 2.707E+03 STD 1.665E-03 9.021E-03 1.137E+01 5.372E+01 2.050E+02 F24 AVG 2.600E+03 2.642E+03 2.625E+03 2.912E+03 2.698E+03 STD 1.291E-02 3.204E+00 7.217E+00 2.492E+01 8.169E+01 F25 AVG 2.700E+03 2.718E+03 2.710E+03 2.788E+03 2.722E+03 STD 0.000E+00 1.652E+00 3.137E+00 2.264E+01 1.739E+01 F26 AVG 2.700E+03 2.800E+03 2.702E+03 2.704E+03 2.713E+03 STD 1.105E-01 5.848E-02 4.243E-01 4.914E-01 2.869E+01 F27 AVG 3.064E+03 3.258E+03 3.176E+03 3.746E+03 3.702E+03 STD 4.019E+01 1.920E+02 1.317E+01 1.921E+02 2.751E+02 F28 AVG 3.649E+03 9.875E+03 4.259E+03 4.783E+03 5.087E+03 STD 3.105E+01 4.773E+02 5.202E+01 1.587E+02 8.883E+02 F29 AVG 4.438E+03 4.109E+03 1.181E+06 2.132E+06 4.451E+06 STD 4.508E+02 2.233E+02 4.541E+05 9.942E+05 1.409E+07 F30 AVG 5.808E+03 3.535E+04 6.266E+04 5.016E+04 2.390E+04 STD 8.399E+02 2.860E+03 1.530E+04 1.235E+04 3.092E+04 Table 6 Comparison results of CCABC with its variants and Wilcoxon ranking. Table 6Item CCABC DEMABC GAABC ISABC NNMABC +/−/ = ∼ 25/4/1 29/1/0 30/0/0 29/0/1 Mean 1.20 2.73 3.00 4.47 3.60 rank 1 2 3 5 4 In addition, Fig. 10 shows the convergence curves of CCABC and ABC algorithm variants on different functions. The convergence curves show that CCABC obtains better solution quality than the other algorithms when optimizing the benchmark functions F22 and F27. Still, CCABC converges slightly slower than the other algorithms in the initial stage. In optimizing F3, F10, and F11, both the high-quality solutions and the algorithm have obvious inflection points for jumping out of the local optimum. By optimizing F1, F17, F19, and F30, it can be seen that CCABC obtains high-quality solutions and converges faster than the other algorithms. Through the above analysis of the benchmark function results, CCABC also has the same large advantage over the other variant algorithms.Fig. 10 Convergence curves of CCABC and ABC variants. Fig. 10 5.3 CCABC for multi-threshold image segmentation This paper compares MTIS experiments for CCABC-MTIS, ABC-MTIS, WOA-MTIS, SCA-MTIS, MVO-MTIS, CLPSO-MTIS, IGWO-MTIS, IWOA-MTIS, and SCADE-MTIS using images A, B, C, D, E, F, G, and H. 5.3.1 Performance evaluation parameters In this section, to accurately evaluate the quality of the segmentation results of each algorithm, we use PSNR, SSIM, and FSIM as the evaluation criteria. The definitions and descriptions of the three evaluation metrics are listed in Table 7 , respectively.Table 7 Performance indicators of the multilevel image segmentation methods. Table 7Evaluation Metrics Core Formula Description Peak Signal to Noise Ratio (PSNR) PSNR=20⋅log10(255RMSE) Image quality evaluation technique based on the error between corresponding pixel points before and after image processing. Structural Similarity Index (SSIM) SSIM=(2μIμSeg+c1)(2σI,Seg+c2)(μI2+μSeg2+c1)(σI2+σSeg2+c2) Image evaluation metrics to evaluate the similarity between images from three perspectives. Feature Similarity Index (FSIM) FSIM=∑I∈ΩSL(X)PCm(X)∑I∈ΩPCm(X) A variant of the SSIM technique that uses feature similarity for image quality evaluation. When we use PSNR to evaluate the results of image segmentation, RMSE is the root mean square error of each pixel, defined as Eq. (14), where M×N denotes the size of the image, Iij denotes the pixel gray value of the original image, and finally, Segij denotes the gray value of the pixel in the segmented image.(14) RMSE=∑i=0M−1∑j=0N−1(Iij−Segij)2M×N Simultaneously, the mean, variance, and Wilcoxon signed-rank test were used to further analyze the FSIM, PSNR, and SSIM evaluation results. 5.3.2 Experimental result analyses Fig. 11 represents the performance of PSNR values of CCABC and other similar segmentation models at different threshold levels. In Fig. 4, the PSNR values are not high when the same type of models perform image segmentation at low threshold levels, which indicates that the segmented image is severely distorted compared to the original image and the noise level of the image is high. This indicates that the noise level and distortion degree of the images segmented by CCABC is the least, and they are most close to the details of the original image.Fig. 11 The PSNR values of CCABC and other similar segmentation models. Fig. 11 The PSNR evaluation criterion analyzes the difference of images before and after segmentation from the perspective of mathematical analysis. At the same time, FSIM and SSIM are evaluation criteria for image quality based on the human visual system (HVS). The general trend is that the higher the segmentation threshold level is, the more obvious the segmentation effect is. Each algorithm model's image segmentation results at the threshold level of 18 are selected as the samples for HVS analysis. The segmentation results under other threshold levels are shown in Table 3, Table 4, Table 5. Therefore, Fig. 12 shows image segmentation results by all algorithms for image H when the threshold level is 18. In Fig. 12, it can be seen from the perspective of HVS that after CCABC segmentation, the image is more consistent and uniform in the regions of the same feature, satisfying the internal consistency; the difference is more significant in the regions adjacent to different features; and the boundaries of the regions are simple and not rough, indicating that the spatial location is equally accurately located. Meanwhile, in Table A4 and Table A5, the FSIM values and SSIM values of CCABC at the threshold level of 18 are higher than those of other models, which indicates that the CCABC segmented image has the highest structural similarity and feature similarity with the original image, which is fully consistent with the results of HVS analysis. In addition, Fig. 13 and Fig. 14 show the box plots of CCABC-MTIS at a low threshold level of 7 and a high threshold level of 18. It can be seen that CCABC-MTIS has the best mean and variance under the three metrics of FSIM, PSNR, and SSIM, and the detailed data can be viewed in the Appendix Tables A3-A5.Fig. 12 The segmented results of H obtained by each algorithm at threshold value 18. Fig. 12 Fig. 13 FSIM, PSNR and SSIM evaluation results for each algorithm at a low threshold level of 7. Fig. 13 Fig. 14 FSIM, PSNR and SSIM evaluation results for each algorithm at a high threshold level of 18. Fig. 14 Table 8, Table 9, Table 10 represent the further analysis of the evaluation results using the Wilcoxon signed-rank test, where Mean indicates the overall mean and Rank indicates the ranking based on the overall mean. From the Wilcoxon signed-rank test analysis of the evaluation results for FSIM, PSNR, and SSIM, the overall average ranking of FSIM, PSNR, and SSIM at all threshold levels was the first. In addition, Fig. 1, Fig. 2, Fig. 3 represent the detailed evaluation results for each algorithm in the FSIM, PSNR, and SSIM evaluation metrics, respectively, where the horizontal axis indicates the name of each algorithm and the vertical axis indicates the threshold value of each segmentation. Fig. 15, Fig. 16, Fig. 17 show the average evaluation results of FSIM, PSNR, and SSIM evaluation metrics for different algorithms, respectively, focusing more on the overall performance of the algorithms. It can be seen in Fig. 1, Fig. 2, Fig. 3 and Fig. 15, Fig. 16, Fig. 17 that CCABC-MTIS has the largest average evaluation results for FSIM, PSNR, and SSIM at the same threshold levels and the same average evaluation results are the largest at all threshold levels. Therefore, the analysis of the evaluation results of FSIM, PSNR, and SSIM can prove the high superiority of the MTIS method based on CCABC.Table 8 The FSIM comparison results of CCABC-MTIS and other methods. Table 8Thresholds CCABC-MTIS ABC-MTIS WOA-MTIS SCA-MTIS MVO-MTIS HHO-MTIS CLPSO-MTIS IGWO-MTIS IWOA-MTIS SCADE-MTIS 3 +/−/ = ∼ 1/0/7 6/0/2 8/0/0 3/0/5 8/0/0 8/0/0 2/1/5 5/0/3 4/0/4 Mean 1.5 3 7.375 6.125 3.875 9.25 8.125 3.625 6.875 5.25 Rank 1 2 8 6 4 10 9 3 7 5 5 +/−/ = ∼ 4/0/4 5/0/3 7/0/1 6/0/2 8/0/0 8/0/0 5/0/3 6/0/2 8/0/0 Mean 1.125 3 5.5 8.625 3 7.875 6.5 3.5 6.875 9 Rank 1 2 5 9 2 8 6 4 7 10 7 +/−/ = ∼ 3/0/5 8/0/0 8/0/0 3/0/5 8/0/0 5/1/2 5/2/1 5/1/2 7/1/0 Mean 1.875 3.75 5.5 8.625 3.5 6.375 6.375 4.75 5.375 8.875 Rank 1 3 6 9 2 7 7 4 5 10 12 +/−/ = ∼ 4/0/4 8/0/0 8/0/0 7/0/1 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 Mean 1 2 4.125 9.625 3 6.125 6.25 6.75 6.75 9.375 Rank 1 2 4 10 3 5 6 7 7 9 15 +/−/ = ∼ 6/0/2 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 Mean 1 2 3.625 9.625 3.375 5.625 6.125 7.125 7.125 9.375 Rank 1 2 4 10 3 5 6 7 7 9 18 +/−/ = ∼ 6/0/2 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 Mean 1 2 3.125 9.375 4.375 5.75 6.625 6.75 6.375 9.625 Rank 1 2 3 9 4 5 7 8 6 10 Table 9 The PSNR comparison results of CCABC-MTIS and other methods. Table 9Thresholds CCABC-MTIS ABC-MTIS WOA-MTIS SCA-MTIS MVO-MTIS HHO-MTIS CLPSO-MTIS IGWO-MTIS IWOA-MTIS SCADE-MTIS 3 +/−/ = ∼ 1/1/6 8/0/0 8/0/0 5/0/3 8/0/0 6/0/2 5/0/3 5/0/3 8/0/0 Mean 1.5 2 6.875 8.625 4.875 8.125 5.5 3.625 5.625 8.25 Rank 1 2 7 10 4 8 5 3 6 9 5 +/−/ = ∼ 3/0/5 7/0/1 8/0/0 5/0/3 8/0/0 8/0/0 7/0/1 8/0/0 8/0/0 Mean 1.125 2.125 5.625 9.625 2.75 7.875 5.75 4.25 6.5 9.375 Rank 1 2 5 10 3 8 6 4 7 9 7 +/−/ = ∼ 4/0/4 8/0/0 8/0/0 4/0/4 8/0/0 8/0/0 7/0/1 8/0/0 8/0/0 Mean 1 2.25 5.125 9.125 2.75 7.75 6.125 4.875 6.25 9.75 Rank 1 2 5 9 3 8 6 4 7 10 12 +/−/ = ∼ 4/0/4 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 Mean 1 2 4.25 9.625 3 6.5 6 6.625 6.625 9.375 Rank 1 2 4 10 3 6 5 7 7 9 15 +/−/ = ∼ 6/0/2 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 Mean 1 2.125 3.375 9.75 3.5 6.125 6 6.875 7 9.25 Rank 1 2 3 10 4 6 5 7 8 9 18 +/−/ = ∼ 7/0/1 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 Mean 1 2 3.125 9.25 4.125 6 6.375 7.125 6.25 9.75 Rank 1 2 3 9 4 5 7 8 6 10 Table 10 The SSIM comparison results of CCABC-MTIS and other methods. Table 10Thresholds CCABC-MTIS ABC-MTIS WOA-MTIS SCA-MTIS MVO-MTIS HHO-MTIS CLPSO-MTIS IGWO-MTIS IWOA-MTIS SCADE-MTIS 3 +/−/ = ∼ 3/0/5 8/0/0 4/0/4 4/0/4 8/0/0 8/0/0 3/2/3 8/0/0 3/0/5 Mean 1.375 2.5 8.875 4.625 5.375 9.625 7.25 3.25 7.5 4.625 Rank 1 2 9 4 6 10 7 3 8 4 5 +/−/ = ∼ 2/0/6 6/0/2 7/0/1 5/0/3 7/0/1 7/0/1 2/1/5 7/0/1 6/0/2 Mean 1.375 2.875 7.5 6.375 4.5 9.125 6.375 3.125 7.125 6.625 Rank 1 2 9 5 4 10 5 3 8 7 7 +/−/ = ∼ 1/0/7 8/0/0 8/0/0 3/0/5 8/0/0 6/1/1 6/2/0 6/1/1 6/1/1 Mean 2 3.25 5.25 7.375 3.5 7.5 5.875 5.375 6.75 8.125 Rank 1 2 4 8 3 9 6 5 7 10 12 +/−/ = ∼ 4/0/4 7/0/1 8/0/0 4/0/4 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 Mean 1 2.125 4.5 9.625 3 6 6.375 6.375 7 9 Rank 1 2 4 10 3 5 6 6 8 9 15 +/−/ = ∼ 6/0/2 6/0/2 8/0/0 7/0/1 7/0/1 8/0/0 8/0/0 8/0/0 8/0/0 Mean 1 2.5 3.125 9.5 3.375 5.75 6.25 7.25 6.75 9.5 Rank 1 2 3 9 4 5 6 8 7 9 18 +/−/ = ∼ 7/0/1 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 8/0/0 Mean 1 2 3.25 9.25 4 5.5 6.875 6.875 6.5 9.75 Rank 1 2 3 9 4 5 7 7 6 10 Fig. 15 The average of FSIM for all threshold levels. Fig. 15 Fig. 16 The average of PSNR for all threshold levels. Fig. 16 Fig. 17 The average of SSIM for all threshold levels. Fig. 17 Table A.6 represents the optimal solution found by each algorithm at each segmentation threshold level, i.e., the optimal KE. we can see that in terms of the optimal KE, most of the values found using CCABC are greater than those of the other algorithms. Moreover, the advantage of CCABC in finding specific segmentation thresholds becomes more and more evident as the number of segmentation thresholds increases starting from 7 threshold-level. Fig. 4, Fig. 5, Fig. 6, Fig. 7, Fig. 8, Fig. 9, Fig. 10, Fig. 11represent the thresholds found by all algorithms on images A-H when the image segmentation level is all 7. The experimental results visually show that the segmentation thresholds vary significantly from one algorithm to another. Fig. 18, Fig. 19 show the convergence curves of CCABC and similar algorithms at low and high threshold levels, respectively, where 7 represents the low threshold level, and 8 represents the high threshold level. It can be seen that CCABC converges faster than other similar algorithms and has a better solving ability.Fig. 18 The convergence curves of all methods at threshold level 7. Fig. 18 Fig. 19 The convergence curves of all methods at threshold level 18. Fig. 19 Fig. 20 represents the box plot of the optimization search time for CCABC and other algorithms. The theoretical algorithm complexity of CCABC has been analyzed in Section 3. Here, the actual time cost of CCABC is analyzed at the 18 threshold-level, which is the most difficult to optimize. The detailed time cost table for CCABC and other algorithms can be found in Table A7. It can be seen that the time spent in the process of segmenting 8 images is relatively constant for all the algorithms. CCABC does not have an advantage over other algorithms in terms of time complexity, but it is within a manageable range. Other algorithms such as HHO, IGWO, and SCADE also have high time complexity but have a considerable advantage over CCABC in terms of optimal search results. In conclusion, the CCABC algorithm has higher performance while the algorithm's complexity is still within a reasonable range.Fig. 20 The box plot of the optimization search time. Fig. 20 Therefore, based on the optimal KE, specific threshold, convergence curve, segmentation results, and the analysis of the evaluation results from FSIM, PSNR, and SSIM, the improved MTIS based on CCABC can obtain better segmentation results, and it is a very excellent segmentation method. Owing to its strong search ability, the proposed CCABC can be applied to tackle other optimization problems such as video coding optimization [124]. Due to significant role of AI in medicine, we will apply the proposed CCABC to handle other datasets [125,126]. CCABC can also be explored to applied to the problems including information retrieval services [[127], [128], [129]], location-based services [130,131], image dehazing [132], kayak cycle phase segmentation [133], and human motion capture [134]. 6 Conclusions and future works To achieve high-quality segmentation of COVID-19 X-ray images, this paper proposes a very effective meta-heuristic algorithm CCABC and an improved MTIS method based on CCABC. CCABC is an improved algorithm based on ABC. In this paper, we introduce HCS and VCS into ABC as a way to improve the search ability and the ability to jump out of local optimum so that CCABC can obtain higher quality solutions. We first conducted comparative experiments using 30 benchmark functions. The comparison experiments above show that CCABC has a broader search performance than ABC. The ability to obtain high-quality solutions is better than ABC. in addition, based on the analysis and comparison of CCABC and its similar algorithms, it can be seen that CCABC also has a better overall ability to jump out of the local optimum trap and obtain higher quality solutions than its peers. Therefore, CCABC is a carefully validated excellent SIOA. Subsequently, we conducted MTIS experiments with CCABC and other similar algorithms and evaluated the segmentation results using FSIM, PSNR, and SSIM. In the preliminary evaluation results, CCABC was shown to have excellent FSIM, PSNR, and SSIM. The evaluation results were further analyzed using the Wilcoxon signed-rank test, and the results also showed that the overall ranking of CCABC was first in the comparison experiments. In addition, by evaluating FSIM, PSNR, and SSIM, we can see that the advantage of CCABC-based MTIS segmentation becomes more obvious as the threshold level increases. Finally, our observation and analysis of KE, specific thresholds, and segmentation results confirm that the specific thresholds found by CCABC are the most reasonable in the comparison experiments, and the segmentation results of CCABC are not only the best for segmenting images as a whole but also the most complete in terms of the details retained. Therefore, the improved MTIS method using CCABC is very excellent. However, the CPU computation for CCABC in MTIS will take longer due to HCS and VCS, resulting in CCABC being more complex than ABC. Still, with the rapid exploitation of parallel computing and high-performance computing techniques, this problem will be solved soon. We will further apply CCABC to more advanced segmentation methods in future work. In addition, in helping COVID-19 diagnosis, we will not be limited to improving image segmentation techniques. Still, we can combine CCABC with fuzzy k-nearest neighbor or support vector machine to build assisted diagnosis models. Finally, CCABC can be applied to medical diagnosis, structural and artificial neural network optimization. We will consider further improving the performance of CCABC in different fields. Declaration of competing interest The authors declare that there is no conflict of interests regarding the publication of article. Appendix A Table A.1 Parametric setups for compared algorithms Table A.1Algorithm Parameters CCABC limit=300 ABC limit=300 SCA A=2 MFO b=1;t=[−1,1];a∈[−1,−2] PSO c1=2;c2=2;vMax=6 SSA c1∈[2, 0];c2∈[0,1];c3∈[0,1] HHO Rabbit Energy=[2,0] WOA a1=[2,0];a2=[−2,−1];b=1 MVO existence probability∈[0.2,1]; travelling distance rate∈[0.6,1] NCHHO c=[2,0];Sinechaoticmap=[1,0]; Inertia factor=0.5 CBA Qmin=0;Qmax=2; SCADE cmin=0.2;cmax=0.8;pCR=0.8 IGWO r1=r2=rand(0,1); C=2×r2; β=ω=10 ACWOA a1=[2,0];a2=[−2,−1];b=1 ASCA_PSO M=4;N=9;c1=2;c2=2;a=2 IWOA b=1;crossover=0.1 CLPSO w=[0.2,0.9];c=1.496 Table A.2 Comparison of CCABC and excellent algorithms Table A.2Fun Item CCABC ABC SCA MFO PSO SSA HHO NCHHO CBA SCADE IGWO ACWOA ASCA_PSO F1 AVG 2.549E+06 4.781E+06 2.356E+08 1.204E+08 8.458E+06 1.603E+06 1.148E+07 8.071E+07 4.157E+06 4.285E+08 1.702E+07 1.411E+08 1.990E+07 STD 1.322E+06 1.663E+06 6.182E+07 1.593E+08 2.371E+06 7.045E+05 5.370E+06 3.904E+07 1.325E+06 8.941E+07 6.380E+06 6.109E+07 1.411E+07 F2 AVG 3.109E+03 4.717E+02 1.676E+10 1.249E+10 1.484E+08 1.255E+04 1.209E+07 4.555E+08 1.211E+04 2.869E+10 2.689E+06 7.398E+09 9.085E+08 STD 3.958E+03 2.710E+02 2.837E+09 6.456E+09 1.355E+07 1.117E+04 2.446E+06 1.791E+08 9.911E+03 3.314E+09 1.403E+06 4.040E+09 1.373E+09 F3 AVG 7.400E+02 1.177E+03 3.915E+04 8.357E+04 9.339E+02 1.380E+03 5.714E+03 3.843E+04 8.941E+03 5.427E+04 6.516E+03 5.081E+04 1.965E+04 STD 5.540E+02 7.907E+02 6.311E+03 4.300E+04 1.259E+02 5.914E+02 2.904E+03 1.111E+04 2.014E+04 6.108E+03 2.692E+03 8.613E+03 6.483E+03 F4 AVG 4.810E+02 4.285E+02 1.382E+03 1.679E+03 4.589E+02 4.874E+02 5.372E+02 7.679E+02 4.947E+02 2.429E+03 5.258E+02 1.189E+03 5.510E+02 STD 2.304E+01 3.002E+01 2.527E+02 1.627E+03 3.264E+01 3.883E+01 5.030E+01 1.280E+02 2.474E+01 4.808E+02 2.952E+01 3.056E+02 8.358E+01 F5 AVG 5.202E+02 5.202E+02 5.209E+02 5.203E+02 5.209E+02 5.201E+02 5.203E+02 5.201E+02 5.202E+02 5.210E+02 5.205E+02 5.208E+02 5.209E+02 STD 3.139E-02 3.912E-02 4.995E-02 1.766E-01 4.815E-02 1.125E-01 1.858E-01 1.629E-01 1.956E-01 5.880E-02 1.274E-01 1.746E-01 5.507E-02 F6 AVG 6.090E+02 6.144E+02 6.343E+02 6.234E+02 6.226E+02 6.186E+02 6.300E+02 6.362E+02 6.411E+02 6.342E+02 6.190E+02 6.337E+02 6.244E+02 STD 2.810E+00 1.266E+00 2.454E+00 3.838E+00 3.008E+00 3.599E+00 3.790E+00 2.838E+00 2.888E+00 2.437E+00 2.734E+00 2.633E+00 3.302E+00 F7 AVG 7.000E+02 7.000E+02 8.351E+02 8.006E+02 7.023E+02 7.000E+02 7.011E+02 7.086E+02 7.000E+02 8.991E+02 7.010E+02 7.417E+02 7.090E+02 STD 1.117E-03 5.875E-05 2.961E+01 5.971E+01 1.452E-01 1.360E-02 1.854E-02 2.629E+00 1.596E-02 4.047E+01 4.786E-02 2.765E+01 1.411E+01 F8 AVG 8.000E+02 8.000E+02 1.042E+03 9.373E+02 9.702E+02 9.075E+02 8.979E+02 9.488E+02 1.006E+03 1.068E+03 8.841E+02 9.985E+02 9.614E+02 STD 7.443E-14 7.079E-14 2.077E+01 3.989E+01 2.071E+01 2.936E+01 1.379E+01 1.714E+01 4.561E+01 1.622E+01 1.666E+01 2.212E+01 2.821E+01 F9 AVG 9.509E+02 9.844E+02 1.174E+03 1.134E+03 1.113E+03 1.016E+03 1.090E+03 1.090E+03 1.163E+03 1.209E+03 1.017E+03 1.130E+03 1.115E+03 STD 7.988E+00 1.202E+01 1.856E+01 4.526E+01 2.610E+01 2.668E+01 2.584E+01 2.131E+01 4.792E+01 1.700E+01 2.150E+01 2.201E+01 3.624E+01 F10 AVG 1.001E+03 1.000E+03 6.948E+03 4.740E+03 5.153E+03 4.523E+03 2.907E+03 4.144E+03 5.480E+03 7.404E+03 3.349E+03 4.790E+03 5.200E+03 STD 8.596E-01 4.147E-01 5.963E+02 8.279E+02 6.593E+02 6.986E+02 7.619E+02 6.562E+02 7.099E+02 3.431E+02 5.906E+02 7.784E+02 6.876E+02 F11 AVG 2.823E+03 3.057E+03 8.035E+03 5.337E+03 5.786E+03 4.508E+03 5.200E+03 5.296E+03 5.804E+03 8.195E+03 4.246E+03 6.255E+03 6.104E+03 STD 2.741E+02 2.268E+02 3.455E+02 6.466E+02 6.240E+02 6.328E+02 6.348E+02 8.228E+02 6.811E+02 2.324E+02 6.910E+02 7.831E+02 8.491E+02 F12 AVG 1.200E+03 1.200E+03 1.203E+03 1.200E+03 1.202E+03 1.200E+03 1.202E+03 1.201E+03 1.201E+03 1.203E+03 1.201E+03 1.202E+03 1.202E+03 STD 3.354E-02 2.847E-02 3.114E-01 2.425E-01 2.539E-01 1.615E-01 4.135E-01 3.600E-01 5.093E-01 3.144E-01 2.888E-01 5.526E-01 3.001E-01 F13 AVG 1.300E+03 1.300E+03 1.303E+03 1.302E+03 1.300E+03 1.301E+03 1.301E+03 1.301E+03 1.300E+03 1.304E+03 1.301E+03 1.302E+03 1.301E+03 STD 2.524E-02 2.814E-02 3.926E-01 1.239E+00 1.030E-01 1.074E-01 1.225E-01 1.189E-01 1.119E-01 3.672E-01 9.749E-02 9.903E-01 1.210E-01 F14 AVG 1.400E+03 1.400E+03 1.444E+03 1.429E+03 1.400E+03 1.400E+03 1.400E+03 1.400E+03 1.400E+03 1.485E+03 1.400E+03 1.419E+03 1.402E+03 STD 1.635E-02 1.562E-02 7.957E+00 1.807E+01 1.395E-01 2.017E-01 1.812E-01 9.965E-01 2.037E-01 1.539E+01 2.595E-01 1.386E+01 3.885E+00 F15 AVG 1.503E+03 1.508E+03 4.655E+03 2.496E+05 1.517E+03 1.508E+03 1.538E+03 1.633E+03 1.561E+03 2.098E+04 1.516E+03 2.027E+03 1.533E+03 STD 3.574E-01 1.313E+00 3.127E+03 4.739E+05 1.420E+00 2.713E+00 7.643E+00 6.485E+01 1.726E+01 8.879E+03 4.663E+00 5.856E+02 7.262E+01 F16 AVG 1.609E+03 1.610E+03 1.613E+03 1.613E+03 1.612E+03 1.611E+03 1.612E+03 1.612E+03 1.613E+03 1.613E+03 1.612E+03 1.612E+03 1.612E+03 STD 3.610E-01 3.833E-01 2.662E-01 4.503E-01 4.621E-01 5.978E-01 4.322E-01 2.945E-01 3.253E-01 2.242E-01 5.250E-01 5.805E-01 3.611E-01 F17 AVG 5.907E+05 2.671E+06 5.751E+06 5.047E+06 2.841E+05 1.102E+05 1.513E+06 6.381E+06 2.257E+05 1.427E+07 8.295E+05 1.555E+07 1.000E+06 STD 3.171E+05 1.063E+06 2.772E+06 9.696E+06 1.338E+05 8.442E+04 1.095E+06 5.183E+06 1.355E+05 6.135E+06 5.526E+05 1.291E+07 1.008E+06 F18 AVG 2.535E+03 2.575E+03 1.558E+08 9.940E+07 2.058E+06 6.316E+03 8.810E+04 2.278E+05 8.883E+03 2.005E+08 2.596E+04 3.647E+07 3.712E+06 STD 8.064E+02 6.382E+02 7.610E+07 2.680E+08 6.108E+05 4.689E+03 4.556E+04 7.803E+05 8.043E+03 1.179E+08 4.153E+04 3.279E+07 1.035E+06 F19 AVG 1.905E+03 1.907E+03 1.994E+03 1.980E+03 1.917E+03 1.917E+03 1.939E+03 1.996E+03 1.936E+03 2.013E+03 1.921E+03 2.014E+03 1.929E+03 STD 9.228E-01 8.647E-01 2.472E+01 6.085E+01 2.606E+00 1.378E+01 4.100E+01 3.622E+01 3.297E+01 1.532E+01 1.924E+01 3.303E+01 2.833E+01 F20 AVG 7.398E+03 9.914E+03 1.595E+04 8.282E+04 2.340E+03 2.325E+03 1.279E+04 3.260E+04 3.588E+03 2.537E+04 2.882E+03 3.243E+04 6.484E+03 STD 2.350E+03 4.591E+03 3.746E+03 8.958E+04 7.602E+01 1.157E+02 6.903E+03 1.897E+04 2.765E+03 7.857E+03 5.840E+02 1.580E+04 3.370E+03 F21 AVG 1.103E+05 2.151E+05 1.384E+06 1.183E+06 1.089E+05 5.748E+04 4.783E+05 1.947E+06 9.921E+04 2.281E+06 3.375E+05 6.306E+06 3.035E+05 STD 7.818E+04 1.238E+05 7.763E+05 3.994E+06 6.109E+04 3.453E+04 3.505E+05 2.253E+06 5.658E+04 1.115E+06 2.698E+05 4.528E+06 2.526E+05 F22 AVG 2.358E+03 2.456E+03 2.968E+03 3.072E+03 2.890E+03 2.571E+03 2.974E+03 3.047E+03 3.430E+03 3.106E+03 2.555E+03 3.072E+03 2.778E+03 STD 7.048E+01 9.252E+01 1.515E+02 2.751E+02 2.096E+02 1.746E+02 2.679E+02 2.476E+02 3.086E+02 1.307E+02 1.789E+02 2.672E+02 2.030E+02 F23 AVG 2.615E+03 2.615E+03 2.671E+03 2.671E+03 2.616E+03 2.615E+03 2.500E+03 2.500E+03 2.616E+03 2.500E+03 2.621E+03 2.529E+03 2.625E+03 STD 7.048E-04 2.969E-01 1.304E+01 4.500E+01 6.728E-01 2.940E-01 0.000E+00 0.000E+00 2.791E-01 0.000E+00 2.418E+00 7.319E+01 5.458E+00 F24 AVG 2.600E+03 2.626E+03 2.600E+03 2.676E+03 2.628E+03 2.642E+03 2.600E+03 2.600E+03 2.674E+03 2.600E+03 2.600E+03 2.600E+03 2.639E+03 STD 1.187E-02 5.285E+00 4.817E-02 2.622E+01 6.498E+00 5.886E+00 8.035E-05 0.000E+00 2.734E+01 2.100E-06 5.202E-03 7.572E-06 7.985E+00 F25 AVG 2.700E+03 2.708E+03 2.726E+03 2.717E+03 2.712E+03 2.712E+03 2.700E+03 2.700E+03 2.731E+03 2.700E+03 2.711E+03 2.700E+03 2.713E+03 STD 0.000E+00 1.275E+00 8.172E+00 7.918E+00 6.427E+00 4.948E+00 0.000E+00 0.000E+00 1.257E+01 0.000E+00 3.192E+00 0.000E+00 5.430E+00 F26 AVG 2.706E+03 2.700E+03 2.702E+03 2.704E+03 2.774E+03 2.701E+03 2.770E+03 2.770E+03 2.720E+03 2.704E+03 2.701E+03 2.756E+03 2.701E+03 STD 2.391E+01 8.979E-02 5.949E-01 1.400E+01 4.434E+01 1.241E-01 4.610E+01 4.609E+01 7.474E+01 4.532E-01 1.347E-01 4.979E+01 1.649E-01 F27 AVG 3.052E+03 3.108E+03 3.477E+03 3.675E+03 3.422E+03 3.446E+03 2.900E+03 2.900E+03 4.000E+03 3.258E+03 3.110E+03 3.724E+03 3.474E+03 STD 3.888E+01 2.712E+00 3.405E+02 1.663E+02 2.800E+02 1.671E+02 0.000E+00 0.000E+00 4.397E+02 2.414E+02 2.964E+00 3.506E+02 2.719E+02 F28 AVG 3.649E+03 3.781E+03 4.816E+03 3.941E+03 6.734E+03 3.827E+03 3.000E+03 3.000E+03 5.250E+03 5.152E+03 3.836E+03 4.190E+03 4.360E+03 STD 4.012E+01 6.740E+01 2.975E+02 1.845E+02 9.907E+02 1.551E+02 0.000E+00 0.000E+00 6.650E+02 8.375E+02 1.664E+02 1.186E+03 3.542E+02 F29 AVG 4.356E+03 3.901E+03 1.283E+07 2.869E+06 5.466E+04 1.501E+06 3.855E+03 3.100E+03 5.311E+07 1.726E+07 1.227E+06 2.427E+07 4.448E+06 STD 4.428E+02 9.831E+01 8.452E+06 3.473E+06 1.156E+05 4.116E+06 3.885E+03 0.000E+00 4.639E+07 9.219E+06 3.678E+06 1.997E+07 6.461E+06 F30 AVG 5.847E+03 5.351E+03 2.207E+05 4.386E+04 1.407E+04 1.106E+04 7.566E+03 3.200E+03 3.700E+04 4.814E+05 2.253E+04 3.407E+05 4.595E+04 STD 8.035E+02 5.890E+02 6.279E+04 4.389E+04 6.473E+03 2.615E+03 6.581E+03 0.000E+00 8.760E+04 1.447E+05 8.708E+03 2.518E+05 5.549E+04 Table A.3 The PSNR comparison of CCABC-MTIS and other methods Table A.3Image Thresholds Item CCABC-MTIS ABC-MTIS WOA-MTIS SCA-MTIS MVO-MTIS HHO-MTIS CLPSO-MTIS IGWO-MTIS IWOA-MTIS SCADE-MTIS A 3 AVG 1.8643E+01 1.8120E+01 1.5904E+01 1.7202E+01 1.7936E+01 1.5461E+01 1.6341E+01 1.7971E+01 1.6999E+01 1.6234E+01 STD 7.0670E-01 9.7337E-01 2.6926E+00 1.9191E+00 1.1901E+00 3.1182E+00 1.3091E+00 1.1591E+00 1.5180E+00 2.0423E+00 5 AVG 1.9988E+01 1.9535E+01 1.8650E+01 1.7511E+01 1.9486E+01 1.8315E+01 1.8675E+01 1.9328E+01 1.8322E+01 1.7664E+01 STD 7.4482E-01 1.1252E+00 2.2891E+00 2.3315E+00 1.1326E+00 2.4037E+00 1.6205E+00 1.9438E+00 2.4584E+00 2.4418E+00 7 AVG 2.2449E+01 2.1753E+01 2.0444E+01 1.8363E+01 2.1711E+01 1.9431E+01 2.0710E+01 2.0780E+01 2.0624E+01 1.9368E+01 STD 9.0046E-01 1.2605E+00 1.7927E+00 1.9870E+00 1.3571E+00 2.0504E+00 1.2672E+00 1.8845E+00 2.9772E+00 3.0777E+00 12 AVG 2.6036E+01 2.5399E+01 2.4204E+01 2.2744E+01 2.4336E+01 2.2780E+01 2.4010E+01 2.3901E+01 2.3489E+01 2.1822E+01 STD 9.1327E-01 1.1898E+00 1.9842E+00 2.4624E+00 1.9106E+00 2.3437E+00 1.8964E+00 2.3732E+00 1.6984E+00 2.4352E+00 15 AVG 2.8156E+01 2.7190E+01 2.6290E+01 2.3562E+01 2.5837E+01 2.5761E+01 2.5673E+01 2.4690E+01 2.4963E+01 2.3866E+01 STD 9.2686E-01 1.0098E+00 2.1618E+00 2.8137E+00 2.2792E+00 2.3160E+00 1.8704E+00 1.8160E+00 2.1298E+00 2.3578E+00 18 AVG 2.9157E+01 2.8891E+01 2.7719E+01 2.5095E+01 2.7011E+01 2.6734E+01 2.6862E+01 2.6687E+01 2.6829E+01 2.4976E+01 STD 8.9739E-01 1.8413E+00 2.1494E+00 2.7725E+00 2.0453E+00 2.2734E+00 1.7370E+00 1.2249E+00 2.2587E+00 1.6579E+00 B 3 AVG 1.7851E+01 1.7376E+01 1.7004E+01 1.5624E+01 1.7178E+01 1.6554E+01 1.6185E+01 1.7474E+01 1.6665E+01 1.5657E+01 STD 5.4214E-01 1.3027E+00 1.3536E+00 1.5492E+00 1.2766E+00 1.9579E+00 1.2589E+00 6.5627E-01 1.5524E+00 1.3473E+00 5 AVG 2.0870E+01 2.0430E+01 1.9240E+01 1.7314E+01 2.0049E+01 1.8109E+01 1.8334E+01 1.9848E+01 1.9463E+01 1.7497E+01 STD 7.4832E-01 9.7277E-01 1.8471E+00 1.7950E+00 1.6109E+00 2.7780E+00 1.6951E+00 1.2449E+00 1.3851E+00 1.8731E+00 7 AVG 2.2028E+01 2.1493E+01 2.0181E+01 1.9047E+01 2.1462E+01 1.8886E+01 2.0181E+01 2.1383E+01 2.0396E+01 1.8691E+01 STD 1.1639E+00 1.6551E+00 2.3375E+00 2.7506E+00 1.5481E+00 3.4491E+00 1.4185E+00 1.3283E+00 2.4092E+00 2.2775E+00 12 AVG 2.6469E+01 2.5801E+01 2.4867E+01 2.1517E+01 2.5334E+01 2.2723E+01 2.4146E+01 2.3671E+01 2.3960E+01 2.2469E+01 STD 9.0550E-01 1.3143E+00 1.9315E+00 2.2938E+00 1.4077E+00 2.3040E+00 1.9154E+00 2.5608E+00 1.7147E+00 2.7696E+00 15 AVG 2.7883E+01 2.7600E+01 2.6968E+01 2.3299E+01 2.6562E+01 2.6069E+01 2.5338E+01 2.4454E+01 2.5174E+01 2.3856E+01 STD 1.0501E+00 8.0615E-01 1.9517E+00 1.7337E+00 1.3085E+00 1.8848E+00 1.7726E+00 2.3969E+00 1.4579E+00 2.4426E+00 18 AVG 2.9168E+01 2.8249E+01 2.7851E+01 2.4838E+01 2.7789E+01 2.6852E+01 2.6632E+01 2.6258E+01 2.6385E+01 2.4243E+01 STD 8.0687E-01 1.5210E+00 2.3724E+00 2.1065E+00 1.4583E+00 1.7640E+00 1.3334E+00 2.1026E+00 1.6734E+00 2.8028E+00 C 3 AVG 1.6839E+01 1.7059E+01 1.3864E+01 1.5129E+01 1.4911E+01 1.4182E+01 1.5272E+01 1.5932E+01 1.4531E+01 1.5456E+01 STD 1.3592E+00 1.2361E+00 1.9805E+00 1.9329E+00 2.0098E+00 2.7902E+00 1.9883E+00 2.3649E+00 2.2732E+00 1.9935E+00 5 AVG 2.1218E+01 2.0018E+01 1.7891E+01 1.7049E+01 1.9797E+01 1.7127E+01 1.8020E+01 1.9524E+01 1.7398E+01 1.7055E+01 STD 1.1123E+00 1.5965E+00 2.5494E+00 2.5204E+00 1.5753E+00 2.5092E+00 2.3310E+00 1.6443E+00 2.6699E+00 2.6079E+00 7 AVG 2.2705E+01 2.2033E+01 2.1065E+01 1.8385E+01 2.1862E+01 1.9254E+01 1.9924E+01 2.0957E+01 2.0206E+01 1.8356E+01 STD 8.0632E-01 1.3281E+00 1.7090E+00 2.1726E+00 1.3938E+00 1.9136E+00 2.1878E+00 1.6911E+00 1.9805E+00 2.5803E+00 12 AVG 2.6783E+01 2.5765E+01 2.4486E+01 2.1077E+01 2.4982E+01 2.4159E+01 2.3866E+01 2.4259E+01 2.3303E+01 2.1778E+01 STD 1.0338E+00 1.4685E+00 2.5691E+00 2.0947E+00 2.0521E+00 2.4447E+00 2.1111E+00 1.6407E+00 2.5423E+00 2.2733E+00 15 AVG 2.8555E+01 2.7608E+01 2.6653E+01 2.3070E+01 2.6888E+01 2.4778E+01 2.5874E+01 2.4965E+01 2.5359E+01 2.3078E+01 STD 1.3637E+00 1.2446E+00 1.7381E+00 2.5561E+00 1.7231E+00 2.6014E+00 1.4169E+00 1.8546E+00 2.5812E+00 3.1397E+00 18 AVG 2.9997E+01 2.9112E+01 2.8234E+01 2.5667E+01 2.8617E+01 2.7400E+01 2.6455E+01 2.6174E+01 2.6973E+01 2.5503E+01 STD 1.0098E+00 1.0997E+00 1.8259E+00 2.1655E+00 1.6688E+00 1.7169E+00 1.7833E+00 1.9058E+00 1.7339E+00 1.8551E+00 D 3 AVG 1.5450E+01 1.5510E+01 1.4923E+01 1.3491E+01 1.5209E+01 1.4801E+01 1.5709E+01 1.5249E+01 1.5084E+01 1.3479E+01 STD 6.3283E-01 9.4591E-01 8.9649E-01 1.9495E+00 8.0522E-01 1.4882E+00 1.3980E+00 7.1293E-01 1.1073E+00 1.9861E+00 5 AVG 1.9365E+01 1.9461E+01 1.8451E+01 1.6016E+01 1.9026E+01 1.7655E+01 1.8404E+01 1.8604E+01 1.8194E+01 1.6345E+01 STD 8.0082E-01 1.5436E+00 1.0920E+00 2.4024E+00 9.7489E-01 1.5412E+00 1.5700E+00 1.5876E+00 1.3677E+00 2.2033E+00 7 AVG 2.2923E+01 2.2648E+01 2.1086E+01 1.7779E+01 2.2426E+01 1.9943E+01 2.0459E+01 2.0215E+01 1.9942E+01 1.8670E+01 STD 1.0957E+00 1.1049E+00 1.9433E+00 2.9425E+00 1.3802E+00 2.3041E+00 1.4998E+00 1.8588E+00 1.8327E+00 1.7464E+00 12 AVG 2.6489E+01 2.5904E+01 2.4496E+01 2.1855E+01 2.4819E+01 2.3463E+01 2.3073E+01 2.3284E+01 2.3615E+01 2.1743E+01 STD 7.5214E-01 1.2327E+00 1.6173E+00 1.5665E+00 1.6412E+00 1.5348E+00 1.4955E+00 2.0332E+00 1.3851E+00 1.9949E+00 15 AVG 2.8102E+01 2.7274E+01 2.6367E+01 2.3138E+01 2.5978E+01 2.4916E+01 2.4879E+01 2.5016E+01 2.4653E+01 2.3719E+01 STD 7.8461E-01 1.0612E+00 1.7388E+00 2.5508E+00 2.0753E+00 1.6196E+00 1.4156E+00 1.3206E+00 1.8662E+00 2.1548E+00 18 AVG 2.9201E+01 2.8322E+01 2.7802E+01 2.5240E+01 2.7012E+01 2.6087E+01 2.6253E+01 2.5587E+01 2.6480E+01 2.4674E+01 STD 9.8706E-01 9.9550E-01 1.7668E+00 2.0425E+00 1.7345E+00 2.3587E+00 1.3259E+00 2.2685E+00 1.5289E+00 2.0450E+00 E 3 AVG 1.7535E+01 1.7214E+01 1.6004E+01 1.4998E+01 1.6749E+01 1.5218E+01 1.6320E+01 1.6803E+01 1.6474E+01 1.6053E+01 STD 7.7020E-01 1.6363E+00 2.1315E+00 1.8718E+00 1.1862E+00 2.0723E+00 1.3633E+00 6.7415E-01 1.3984E+00 1.4255E+00 5 AVG 2.1098E+01 2.0112E+01 1.8820E+01 1.7352E+01 1.9986E+01 1.7525E+01 1.8255E+01 1.9460E+01 1.9120E+01 1.6701E+01 STD 6.6127E-01 1.4538E+00 2.7772E+00 2.0882E+00 1.7685E+00 2.1426E+00 2.3419E+00 1.4631E+00 1.8951E+00 3.1170E+00 7 AVG 2.2110E+01 2.1856E+01 2.0590E+01 1.9223E+01 2.1594E+01 2.0289E+01 2.0611E+01 2.0836E+01 1.9777E+01 1.8455E+01 STD 9.4508E-01 1.0316E+00 1.7439E+00 1.8427E+00 1.0911E+00 3.1825E+00 1.6581E+00 1.6860E+00 2.1538E+00 3.0887E+00 12 AVG 2.6266E+01 2.5664E+01 2.4035E+01 2.1015E+01 2.5434E+01 2.2904E+01 2.3878E+01 2.3517E+01 2.2988E+01 2.1981E+01 STD 9.8603E-01 1.5383E+00 2.9006E+00 2.9169E+00 1.4039E+00 2.2695E+00 1.7835E+00 1.6597E+00 1.8180E+00 2.4837E+00 15 AVG 2.8446E+01 2.7844E+01 2.6104E+01 2.3414E+01 2.6468E+01 2.5176E+01 2.5424E+01 2.5632E+01 2.5002E+01 2.3073E+01 STD 8.6255E-01 1.0790E+00 2.3954E+00 1.7529E+00 1.6368E+00 2.3684E+00 1.7379E+00 1.6878E+00 1.9597E+00 2.1603E+00 18 AVG 2.9595E+01 2.8606E+01 2.7767E+01 2.4881E+01 2.7028E+01 2.6094E+01 2.6030E+01 2.5883E+01 2.6456E+01 2.4715E+01 STD 1.0715E+00 1.4114E+00 1.6880E+00 1.9503E+00 1.9769E+00 2.5611E+00 2.1052E+00 1.6578E+00 2.2058E+00 2.4212E+00 F 3 AVG 1.8186E+01 1.8152E+01 1.6500E+01 1.5819E+01 1.7747E+01 1.5982E+01 1.6364E+01 1.7965E+01 1.6349E+01 1.5809E+01 STD 4.0036E-01 6.6658E-01 2.0391E+00 1.3241E+00 1.1799E+00 2.1177E+00 1.5811E+00 3.8165E-01 1.7867E+00 1.6265E+00 5 AVG 2.1233E+01 2.0814E+01 1.9127E+01 1.7608E+01 2.0669E+01 1.8145E+01 1.8588E+01 1.9884E+01 1.8009E+01 1.7222E+01 STD 1.0968E+00 1.2206E+00 2.3437E+00 1.9010E+00 1.0999E+00 2.8123E+00 1.7414E+00 1.6534E+00 2.4169E+00 2.4254E+00 7 AVG 2.3020E+01 2.2132E+01 2.0687E+01 1.8566E+01 2.1756E+01 2.0082E+01 2.0559E+01 2.1437E+01 2.0945E+01 1.8349E+01 STD 7.6179E-01 9.4401E-01 2.5747E+00 2.1490E+00 1.7709E+00 2.1989E+00 1.7635E+00 1.9342E+00 1.9148E+00 2.6285E+00 12 AVG 2.5926E+01 2.5461E+01 2.3693E+01 2.0859E+01 2.4837E+01 2.4125E+01 2.3727E+01 2.3299E+01 2.3397E+01 2.2873E+01 STD 1.1659E+00 1.1161E+00 2.7404E+00 2.1968E+00 1.6623E+00 1.9159E+00 1.8497E+00 2.0756E+00 1.7766E+00 2.1413E+00 15 AVG 2.8104E+01 2.6815E+01 2.6822E+01 2.3642E+01 2.6476E+01 2.5398E+01 2.5135E+01 2.4846E+01 2.5198E+01 2.4205E+01 STD 1.0128E+00 1.3650E+00 1.5081E+00 2.2948E+00 1.6531E+00 2.3603E+00 1.6243E+00 2.0511E+00 2.2363E+00 2.6740E+00 18 AVG 2.9306E+01 2.8549E+01 2.7618E+01 2.4797E+01 2.7220E+01 2.6260E+01 2.6363E+01 2.6808E+01 2.6460E+01 2.4971E+01 STD 1.3358E+00 1.0321E+00 1.7046E+00 2.3775E+00 1.9502E+00 2.3400E+00 1.4331E+00 1.6421E+00 2.2746E+00 2.0292E+00 G 3 AVG 1.6862E+01 1.6007E+01 1.5596E+01 1.3954E+01 1.5307E+01 1.4654E+01 1.4491E+01 1.4485E+01 1.6595E+01 1.4484E+01 STD 1.7872E+00 2.6668E+00 2.9150E+00 2.4101E+00 2.4416E+00 2.9303E+00 3.0047E+00 2.6995E+00 2.0260E+00 2.4240E+00 5 AVG 1.9578E+01 1.9141E+01 1.8433E+01 1.6590E+01 1.9380E+01 1.7234E+01 1.8044E+01 1.7606E+01 1.7475E+01 1.6569E+01 STD 1.8912E+00 1.5174E+00 2.2728E+00 2.8376E+00 1.7068E+00 2.6630E+00 2.0413E+00 2.3809E+00 2.3397E+00 2.8375E+00 7 AVG 2.3311E+01 2.2258E+01 2.0567E+01 1.8387E+01 2.2363E+01 1.9628E+01 2.0263E+01 2.0159E+01 2.0015E+01 1.8064E+01 STD 1.1347E+00 1.3204E+00 2.4568E+00 2.4281E+00 1.3418E+00 2.4595E+00 1.8870E+00 2.2722E+00 1.8821E+00 3.0270E+00 12 AVG 2.6331E+01 2.6251E+01 2.4957E+01 2.2542E+01 2.5422E+01 2.3753E+01 2.3849E+01 2.3509E+01 2.3860E+01 2.2180E+01 STD 1.0290E+00 9.9881E-01 1.6395E+00 2.0588E+00 1.9726E+00 2.3702E+00 1.8749E+00 1.8142E+00 2.0100E+00 2.0316E+00 15 AVG 2.7902E+01 2.7212E+01 2.6700E+01 2.3545E+01 2.6832E+01 2.4967E+01 2.5476E+01 2.5281E+01 2.4996E+01 2.3929E+01 STD 1.0506E+00 1.3899E+00 2.2517E+00 2.0652E+00 1.4085E+00 2.2916E+00 1.3563E+00 1.7856E+00 1.9299E+00 1.9541E+00 18 AVG 2.9810E+01 2.8786E+01 2.8501E+01 2.5033E+01 2.7026E+01 2.7082E+01 2.7018E+01 2.7093E+01 2.6326E+01 2.4861E+01 STD 7.2629E-01 1.1518E+00 1.5325E+00 1.6655E+00 1.7135E+00 2.0717E+00 1.6410E+00 1.2611E+00 1.9993E+00 2.1156E+00 H 3 AVG 1.6064E+01 1.6752E+01 1.4890E+01 1.3886E+01 1.5489E+01 1.4545E+01 1.5680E+01 1.5925E+01 1.5498E+01 1.4181E+01 STD 1.0813E+00 8.5253E-01 1.4440E+00 2.3479E+00 1.0258E+00 1.9197E+00 1.7858E+00 1.0004E+00 1.6719E+00 2.3045E+00 5 AVG 2.0451E+01 1.9913E+01 1.8028E+01 1.6792E+01 2.0199E+01 1.7774E+01 1.8774E+01 1.8788E+01 1.8544E+01 1.6823E+01 STD 1.0375E+00 1.1686E+00 1.9940E+00 1.9610E+00 1.0013E+00 2.1364E+00 1.3601E+00 2.0979E+00 1.5296E+00 2.0988E+00 7 AVG 2.2785E+01 2.2523E+01 2.0833E+01 1.9380E+01 2.2690E+01 2.0324E+01 1.9991E+01 2.0497E+01 2.0568E+01 1.7918E+01 STD 1.0329E+00 1.2764E+00 1.9141E+00 2.2402E+00 1.5296E+00 2.0348E+00 1.7251E+00 1.7327E+00 2.1779E+00 2.5229E+00 12 AVG 2.6143E+01 2.5828E+01 2.4612E+01 2.1866E+01 2.5180E+01 2.4027E+01 2.3486E+01 2.3716E+01 2.3182E+01 2.1902E+01 STD 1.3980E+00 1.1579E+00 2.0435E+00 2.5515E+00 1.7704E+00 1.8395E+00 1.5527E+00 1.6291E+00 2.5870E+00 2.2088E+00 15 AVG 2.8528E+01 2.7038E+01 2.5988E+01 2.3990E+01 2.6437E+01 2.5566E+01 2.5260E+01 2.4614E+01 2.4767E+01 2.3655E+01 STD 8.9647E-01 1.9624E+00 2.0135E+00 2.3525E+00 2.2377E+00 1.8259E+00 1.6937E+00 1.6629E+00 2.2013E+00 1.7723E+00 18 AVG 2.9583E+01 2.8822E+01 2.8389E+01 2.4816E+01 2.7730E+01 2.7104E+01 2.6846E+01 2.5880E+01 2.6221E+01 2.4950E+01 STD 7.7488E-01 1.1326E+00 1.1780E+00 2.0019E+00 1.5350E+00 2.2860E+00 1.5719E+00 1.6988E+00 1.9682E+00 2.1268E+00 Table A.4 The FSIM comparison of CCABC-MTIS and other methods Table A.4Image Thresholds Item CCABC-MTIS ABC-MTIS WOA-MTIS SCA-MTIS MVO-MTIS HHO-MTIS CLPSO-MTIS IGWO-MTIS IWOA-MTIS SCADE-MTIS A 3 AVG 8.5468E-01 8.3995E-01 7.7179E-01 7.9197E-01 8.3061E-01 7.6209E-01 7.7215E-01 8.3848E-01 7.9309E-01 7.7930E-01 STD 2.2995E-02 2.8749E-02 8.2054E-02 7.5628E-02 3.7543E-02 8.6860E-02 4.0718E-02 2.9791E-02 5.6441E-02 5.5476E-02 5 AVG 8.9228E-01 8.7616E-01 8.4674E-01 8.0883E-01 8.7544E-01 8.3381E-01 8.5162E-01 8.6779E-01 8.3848E-01 8.2069E-01 STD 2.1878E-02 3.4095E-02 7.5598E-02 7.3442E-02 3.4974E-02 7.1671E-02 5.5401E-02 5.1651E-02 7.1098E-02 7.0107E-02 7 AVG 8.0291E-01 7.8274E-01 7.7058E-01 7.3010E-01 7.7585E-01 7.5455E-01 8.9710E-01 8.9249E-01 8.8097E-01 8.4941E-01 STD 2.2103E-02 2.2693E-02 3.0454E-02 3.6014E-02 3.0843E-02 3.1191E-02 3.1731E-02 4.5387E-02 8.1517E-02 8.4456E-02 12 AVG 9.7268E-01 9.6641E-01 9.3677E-01 9.1277E-01 9.4807E-01 9.1348E-01 9.3671E-01 9.2929E-01 9.2919E-01 8.9272E-01 STD 6.3993E-03 1.0560E-02 3.8153E-02 4.9248E-02 2.8800E-02 4.9399E-02 3.3328E-02 4.6097E-02 3.4850E-02 5.2970E-02 15 AVG 9.8172E-01 9.7580E-01 9.5977E-01 9.1815E-01 9.5510E-01 9.5150E-01 9.5230E-01 9.4338E-01 9.4324E-01 9.2853E-01 STD 5.4036E-03 7.8615E-03 2.3887E-02 5.3622E-02 3.0963E-02 3.1739E-02 3.2338E-02 2.7857E-02 3.8626E-02 4.1612E-02 18 AVG 9.8385E-01 9.8133E-01 9.7023E-01 9.3850E-01 9.6305E-01 9.5915E-01 9.6445E-01 9.6492E-01 9.6105E-01 9.3967E-01 STD 4.9314E-03 9.5176E-03 1.8822E-02 5.0959E-02 2.4414E-02 2.5060E-02 1.8235E-02 1.5562E-02 3.1411E-02 2.9291E-02 B 3 AVG 7.0762E-01 7.0197E-01 7.0088E-01 6.9786E-01 7.0304E-01 6.9009E-01 6.7698E-01 7.0689E-01 6.9477E-01 6.9773E-01 STD 4.7342E-03 1.2235E-02 2.3170E-02 1.8212E-02 2.1693E-02 2.7710E-02 2.6987E-02 5.9717E-03 2.2674E-02 2.3718E-02 5 AVG 7.4687E-01 7.4324E-01 7.3200E-01 7.1266E-01 7.3837E-01 7.2637E-01 7.1799E-01 7.4619E-01 7.3883E-01 7.1211E-01 STD 1.9699E-02 2.1747E-02 3.8935E-02 3.9764E-02 3.1805E-02 4.2866E-02 2.7609E-02 2.5823E-02 3.1546E-02 3.5053E-02 7 AVG 8.3642E-01 8.2001E-01 7.9079E-01 7.5325E-01 8.2962E-01 7.6806E-01 7.5052E-01 7.7370E-01 7.6176E-01 7.3611E-01 STD 3.3680E-02 4.8979E-02 5.8764E-02 5.9403E-02 3.7241E-02 7.2336E-02 3.2167E-02 2.9396E-02 3.9788E-02 3.7729E-02 12 AVG 8.8177E-01 8.6596E-01 8.5288E-01 7.8951E-01 8.5937E-01 8.2095E-01 8.3372E-01 8.2860E-01 8.3020E-01 8.1148E-01 STD 2.0538E-02 2.6074E-02 3.8411E-02 3.6263E-02 2.6942E-02 5.2650E-02 3.7511E-02 4.6028E-02 3.9022E-02 5.0649E-02 15 AVG 9.0746E-01 8.9971E-01 8.9219E-01 8.2015E-01 8.8504E-01 8.7375E-01 8.6040E-01 8.4712E-01 8.5708E-01 8.3084E-01 STD 1.7426E-02 1.8752E-02 3.6189E-02 3.2706E-02 2.7195E-02 3.7957E-02 2.8243E-02 3.9773E-02 2.3902E-02 4.8202E-02 18 AVG 9.2387E-01 9.0969E-01 9.0833E-01 8.4665E-01 9.0060E-01 8.8586E-01 8.7422E-01 8.7351E-01 8.7682E-01 8.3985E-01 STD 1.3685E-02 2.5306E-02 3.3659E-02 4.2853E-02 2.5400E-02 3.1694E-02 2.4898E-02 3.5905E-02 2.6784E-02 5.0510E-02 C 3 AVG 7.4918E-01 7.4832E-01 6.8010E-01 7.2984E-01 6.9398E-01 6.8493E-01 7.0254E-01 7.2758E-01 6.8209E-01 7.4338E-01 STD 1.0291E-02 9.0110E-03 4.3131E-02 3.1729E-02 5.2098E-02 4.8873E-02 4.3452E-02 4.6259E-02 4.1244E-02 3.1877E-02 5 AVG 8.2324E-01 7.9616E-01 7.4316E-01 7.4399E-01 7.8325E-01 7.3812E-01 7.5295E-01 7.7701E-01 7.4161E-01 7.4201E-01 STD 3.0393E-02 3.7379E-02 5.5581E-02 4.3842E-02 4.7295E-02 5.7191E-02 4.5586E-02 4.1600E-02 5.1503E-02 4.9257E-02 7 AVG 8.1050E-01 7.9696E-01 7.8529E-01 7.5487E-01 7.9921E-01 7.6113E-01 7.9299E-01 8.1600E-01 8.0530E-01 7.6439E-01 STD 2.5741E-02 3.4161E-02 3.7015E-02 3.1706E-02 3.6690E-02 4.2900E-02 4.4271E-02 3.8298E-02 4.5781E-02 4.9920E-02 12 AVG 9.3413E-01 9.1611E-01 8.8952E-01 8.2102E-01 8.9799E-01 8.8284E-01 8.7082E-01 8.7826E-01 8.6858E-01 8.3606E-01 STD 1.9255E-02 2.5595E-02 5.2418E-02 4.8398E-02 4.1811E-02 5.0645E-02 4.7517E-02 3.4664E-02 4.6595E-02 4.4710E-02 15 AVG 9.5525E-01 9.4344E-01 9.2765E-01 8.5636E-01 9.2862E-01 8.8908E-01 9.0530E-01 8.8671E-01 8.9581E-01 8.5984E-01 STD 1.7880E-02 1.6090E-02 2.9499E-02 5.1933E-02 2.8857E-02 4.8557E-02 2.9042E-02 3.2653E-02 5.0515E-02 5.8963E-02 18 AVG 9.6776E-01 9.5834E-01 9.3889E-01 9.0090E-01 9.4543E-01 9.3137E-01 9.1169E-01 9.1062E-01 9.2247E-01 8.9874E-01 STD 1.0278E-02 1.4356E-02 2.7938E-02 4.6133E-02 2.5212E-02 2.4973E-02 3.4228E-02 3.2313E-02 2.6278E-02 3.3609E-02 D 3 AVG 6.9905E-01 6.9890E-01 6.8366E-01 6.8970E-01 6.9646E-01 6.8294E-01 6.8443E-01 6.9636E-01 6.8940E-01 6.9124E-01 STD 7.5397E-03 8.7657E-03 1.7203E-02 1.8923E-02 1.4793E-02 1.8268E-02 1.6967E-02 1.0032E-02 2.1494E-02 2.0463E-02 5 AVG 7.5803E-01 7.5198E-01 7.3950E-01 7.0561E-01 7.4457E-01 7.2237E-01 7.2746E-01 7.3772E-01 7.2839E-01 7.1756E-01 STD 1.2281E-02 1.5069E-02 2.7355E-02 2.8390E-02 2.3472E-02 2.9714E-02 2.7142E-02 2.3086E-02 2.8326E-02 2.2555E-02 7 AVG 7.8812E-01 7.8375E-01 7.6375E-01 7.3872E-01 7.8780E-01 7.6002E-01 7.5814E-01 7.5926E-01 7.6105E-01 7.2990E-01 STD 2.1854E-02 2.8128E-02 3.3304E-02 4.1774E-02 3.2832E-02 3.7451E-02 2.8597E-02 3.1203E-02 3.3062E-02 3.5579E-02 12 AVG 8.7691E-01 8.6591E-01 8.3690E-01 7.8834E-01 8.4912E-01 8.2705E-01 8.1658E-01 8.2541E-01 8.2880E-01 7.8507E-01 STD 1.3301E-02 2.0158E-02 3.5677E-02 2.9650E-02 2.8266E-02 2.9495E-02 2.8648E-02 3.0950E-02 2.5737E-02 3.5582E-02 15 AVG 9.0742E-01 8.8966E-01 8.7391E-01 8.2130E-01 8.7423E-01 8.4870E-01 8.3695E-01 8.4828E-01 8.4549E-01 8.1810E-01 STD 1.5290E-02 1.9365E-02 3.6847E-02 3.6344E-02 2.7565E-02 3.0335E-02 2.8930E-02 2.6260E-02 3.4888E-02 3.9566E-02 18 AVG 9.2059E-01 9.0692E-01 9.0119E-01 8.4637E-01 8.8455E-01 8.7575E-01 8.6840E-01 8.6725E-01 8.7172E-01 8.4446E-01 STD 1.4566E-02 1.9316E-02 2.5394E-02 3.8597E-02 3.0331E-02 4.1737E-02 2.8116E-02 2.8773E-02 2.3123E-02 3.2463E-02 E 3 AVG 7.2607E-01 7.2239E-01 7.1264E-01 7.1585E-01 7.2874E-01 7.0385E-01 7.0848E-01 7.3522E-01 7.1644E-01 7.2213E-01 STD 1.0429E-02 1.4793E-02 3.0512E-02 2.4150E-02 2.2336E-02 2.9843E-02 2.7843E-02 1.3010E-02 2.5768E-02 2.5846E-02 5 AVG 7.4793E-01 7.3662E-01 7.3868E-01 7.3649E-01 7.3784E-01 7.2330E-01 7.3599E-01 7.4265E-01 7.3619E-01 7.3426E-01 STD 1.5203E-02 2.5995E-02 3.8811E-02 3.4873E-02 2.8978E-02 3.7249E-02 2.5543E-02 2.4524E-02 3.0210E-02 2.8786E-02 7 AVG 9.3602E-01 9.2934E-01 8.9382E-01 8.4338E-01 9.2363E-01 8.8113E-01 7.5841E-01 7.6850E-01 7.4770E-01 7.4623E-01 STD 1.5325E-02 1.8675E-02 3.7918E-02 6.1942E-02 2.4779E-02 7.5843E-02 3.1565E-02 3.2028E-02 3.7649E-02 3.5230E-02 12 AVG 8.6731E-01 8.5609E-01 8.3316E-01 7.8328E-01 8.5520E-01 8.0601E-01 8.2498E-01 8.1952E-01 8.0930E-01 7.9817E-01 STD 2.4777E-02 3.0928E-02 4.8785E-02 4.0060E-02 3.0944E-02 4.4885E-02 3.2036E-02 2.8921E-02 3.7341E-02 3.9598E-02 15 AVG 9.0739E-01 8.9914E-01 8.6960E-01 8.2357E-01 8.7523E-01 8.4872E-01 8.4838E-01 8.5665E-01 8.4694E-01 8.1194E-01 STD 1.7945E-02 1.8391E-02 4.5280E-02 3.1947E-02 2.8501E-02 4.3589E-02 3.5559E-02 2.4774E-02 4.0869E-02 3.3718E-02 18 AVG 9.2719E-01 9.1031E-01 8.9481E-01 8.4536E-01 8.7824E-01 8.6492E-01 8.6206E-01 8.6120E-01 8.7201E-01 8.4201E-01 STD 1.5994E-02 2.5550E-02 3.0774E-02 3.5061E-02 3.4225E-02 4.6677E-02 3.2497E-02 3.3371E-02 2.9166E-02 4.3255E-02 F 3 AVG 7.4019E-01 7.3982E-01 7.2153E-01 7.2285E-01 7.4439E-01 7.1479E-01 7.1206E-01 7.4428E-01 7.2005E-01 7.2435E-01 STD 1.4307E-02 1.6409E-02 3.7862E-02 3.4433E-02 2.1765E-02 3.7925E-02 3.8421E-02 1.9436E-02 3.2831E-02 3.5599E-02 5 AVG 8.0498E-01 7.8659E-01 7.6509E-01 7.5358E-01 7.9348E-01 7.6680E-01 7.5536E-01 7.8434E-01 7.3911E-01 7.5450E-01 STD 1.5270E-02 3.1484E-02 4.9569E-02 4.1445E-02 2.7888E-02 5.5245E-02 3.0154E-02 3.8224E-02 4.2260E-02 4.7529E-02 7 AVG 7.9859E-01 7.8003E-01 7.6760E-01 7.3463E-01 7.7767E-01 7.6992E-01 7.9301E-01 8.2162E-01 8.0524E-01 7.5513E-01 STD 1.6560E-02 2.3344E-02 4.7579E-02 3.8589E-02 3.6636E-02 3.1509E-02 4.4600E-02 4.1870E-02 4.2086E-02 5.3941E-02 12 AVG 9.2546E-01 9.1661E-01 8.7472E-01 8.0762E-01 8.9998E-01 8.8300E-01 8.7083E-01 8.6035E-01 8.6265E-01 8.4955E-01 STD 2.1394E-02 2.1062E-02 5.6984E-02 5.1468E-02 2.9982E-02 4.0026E-02 4.4619E-02 4.9585E-02 4.1093E-02 4.5384E-02 15 AVG 9.5519E-01 9.3884E-01 9.2685E-01 8.6060E-01 9.2671E-01 8.9993E-01 8.9357E-01 8.9225E-01 8.9777E-01 8.7815E-01 STD 1.4693E-02 2.2849E-02 2.5813E-02 5.2716E-02 2.6874E-02 4.4589E-02 3.3277E-02 4.1203E-02 3.8891E-02 5.0418E-02 18 AVG 9.6335E-01 9.5778E-01 9.3738E-01 8.8268E-01 9.2988E-01 9.1011E-01 9.1370E-01 9.2061E-01 9.1897E-01 8.8608E-01 STD 1.6883E-02 1.4026E-02 2.3988E-02 4.4351E-02 3.3887E-02 4.2606E-02 2.8721E-02 2.7264E-02 3.4064E-02 4.2638E-02 G 3 AVG 7.0362E-01 6.7958E-01 6.6877E-01 6.5019E-01 6.5182E-01 6.4902E-01 6.5345E-01 6.4686E-01 6.8264E-01 6.6080E-01 STD 3.3949E-02 5.2080E-02 5.1970E-02 5.0319E-02 5.3157E-02 5.4666E-02 5.6702E-02 5.2774E-02 4.1218E-02 4.5348E-02 5 AVG 7.6842E-01 7.5025E-01 7.4219E-01 7.0235E-01 7.6929E-01 7.1286E-01 7.2747E-01 7.2787E-01 7.1787E-01 7.0733E-01 STD 5.3189E-02 4.5850E-02 5.6828E-02 5.5548E-02 3.9872E-02 6.5523E-02 5.2054E-02 5.8139E-02 6.0421E-02 5.7129E-02 7 AVG 8.6759E-01 8.4131E-01 8.0822E-01 7.7324E-01 8.4747E-01 7.9769E-01 7.9222E-01 7.8705E-01 7.8308E-01 7.4050E-01 STD 2.5508E-02 3.5679E-02 5.2846E-02 4.7022E-02 3.1526E-02 4.6355E-02 4.9290E-02 5.6576E-02 4.3080E-02 5.7932E-02 12 AVG 9.1367E-01 9.1274E-01 8.8553E-01 8.3717E-01 8.9951E-01 8.7111E-01 8.6652E-01 8.6373E-01 8.6610E-01 8.2657E-01 STD 1.9214E-02 1.8769E-02 3.5302E-02 4.5802E-02 3.4795E-02 4.4518E-02 3.4593E-02 4.0261E-02 3.9397E-02 4.8618E-02 15 AVG 9.3924E-01 9.2501E-01 9.1570E-01 8.5671E-01 9.2004E-01 8.8915E-01 8.9212E-01 8.8493E-01 8.7763E-01 8.5867E-01 STD 1.7849E-02 2.5669E-02 3.6155E-02 4.4257E-02 2.2109E-02 3.6477E-02 2.9600E-02 3.2790E-02 3.7050E-02 4.1728E-02 18 AVG 9.5929E-01 9.4624E-01 9.4111E-01 8.8578E-01 9.1556E-01 9.2195E-01 9.1449E-01 9.1902E-01 9.0354E-01 8.7813E-01 STD 9.9453E-03 1.6424E-02 2.3293E-02 3.1954E-02 3.0858E-02 3.3293E-02 2.9972E-02 2.2726E-02 3.9339E-02 4.3236E-02 H 3 AVG 7.3949E-01 7.3460E-01 7.1569E-01 7.2443E-01 7.2639E-01 7.0096E-01 7.1398E-01 7.3701E-01 6.9890E-01 7.2546E-01 STD 1.1029E-02 7.0985E-03 3.4984E-02 2.0006E-02 2.6736E-02 3.7662E-02 3.0872E-02 1.2024E-02 2.6464E-02 2.6924E-02 5 AVG 7.8058E-01 7.5713E-01 7.4077E-01 7.4044E-01 7.6268E-01 7.4099E-01 7.4436E-01 7.6675E-01 7.4992E-01 7.3259E-01 STD 1.5311E-02 2.2671E-02 4.2183E-02 3.2638E-02 2.8584E-02 3.7656E-02 2.9560E-02 4.0270E-02 3.5453E-02 3.7130E-02 7 AVG 8.3915E-01 8.3325E-01 8.0549E-01 7.8046E-01 8.3699E-01 8.0656E-01 7.7065E-01 7.7995E-01 7.8405E-01 7.5785E-01 STD 3.0542E-02 3.6870E-02 5.0327E-02 4.4008E-02 3.5785E-02 4.9238E-02 3.8674E-02 2.7566E-02 3.5714E-02 4.2214E-02 12 AVG 8.9210E-01 8.8391E-01 8.6040E-01 8.1167E-01 8.7006E-01 8.5134E-01 8.3232E-01 8.4110E-01 8.3658E-01 8.1501E-01 STD 2.9484E-02 2.6271E-02 4.8274E-02 4.8526E-02 3.6637E-02 3.3669E-02 3.1897E-02 3.3334E-02 5.0866E-02 4.3248E-02 15 AVG 9.3375E-01 9.0496E-01 8.9171E-01 8.4535E-01 9.0331E-01 8.7989E-01 8.7138E-01 8.5201E-01 8.6956E-01 8.3845E-01 STD 1.6047E-02 3.4619E-02 3.5587E-02 4.1582E-02 3.8088E-02 3.3356E-02 3.2417E-02 3.4367E-02 3.4157E-02 3.4693E-02 18 AVG 9.4361E-01 9.3409E-01 9.2745E-01 8.6150E-01 9.1150E-01 9.0164E-01 8.9376E-01 8.7903E-01 8.8486E-01 8.6466E-01 STD 1.3193E-02 2.0144E-02 1.8949E-02 3.5252E-02 2.6068E-02 3.9409E-02 2.9890E-02 3.1780E-02 3.4659E-02 3.9789E-02 Table A.5 The SSIM comparison of CCABC-MTIS and other method Table A.5Image Thresholds Item CCABC-MTIS ABC-MTIS WOA-MTIS SCA-MTIS MVO-MTIS HHO-MTIS CLPSO-MTIS IGWO-MTIS IWOA-MTIS SCADE-MTIS A 3 AVG 5.7867E-01 5.5611E-01 4.5178E-01 5.2176E-01 5.4187E-01 4.2878E-01 4.6210E-01 5.5321E-01 5.0033E-01 4.7826E-01 STD 3.4749E-02 4.4301E-02 1.3523E-01 9.6392E-02 5.5737E-02 1.6190E-01 6.2335E-02 5.0743E-02 7.6505E-02 9.3886E-02 5 AVG 6.5091E-01 6.2759E-01 5.9199E-01 5.4305E-01 6.2706E-01 5.8284E-01 5.9703E-01 6.2297E-01 5.8222E-01 5.6152E-01 STD 3.5489E-02 5.5653E-02 1.0467E-01 1.1250E-01 5.4679E-02 1.1118E-01 7.7616E-02 8.5848E-02 1.1060E-01 1.0763E-01 7 AVG 7.6665E-01 7.5194E-01 7.3549E-01 7.2962E-01 7.3640E-01 7.1564E-01 6.9127E-01 6.8676E-01 6.7386E-01 6.2159E-01 STD 3.1577E-02 3.9363E-02 5.5082E-02 5.2393E-02 4.7928E-02 7.6972E-02 5.2918E-02 7.7829E-02 1.3854E-01 1.4166E-01 12 AVG 8.7013E-01 8.5274E-01 8.0850E-01 7.5765E-01 8.1752E-01 7.6471E-01 7.9893E-01 7.9468E-01 7.8230E-01 7.2783E-01 STD 2.1840E-02 3.1172E-02 6.3027E-02 8.2995E-02 6.1121E-02 7.9616E-02 6.4516E-02 7.9617E-02 6.4826E-02 8.6540E-02 15 AVG 9.1288E-01 8.9282E-01 8.6477E-01 7.7337E-01 8.5349E-01 8.4372E-01 8.4488E-01 8.2000E-01 8.2820E-01 7.9405E-01 STD 1.6456E-02 2.2019E-02 4.9172E-02 9.7777E-02 6.0576E-02 6.6929E-02 5.7657E-02 5.3190E-02 6.4885E-02 7.0974E-02 18 AVG 9.2644E-01 9.1860E-01 8.9012E-01 8.2130E-01 8.7588E-01 8.6927E-01 8.7372E-01 8.7275E-01 8.6968E-01 8.2436E-01 STD 1.4421E-02 2.9468E-02 4.9598E-02 9.2835E-02 5.1849E-02 5.1009E-02 4.2559E-02 2.9325E-02 6.1421E-02 5.2523E-02 B 3 AVG 6.9155E-01 6.7854E-01 6.5516E-01 6.8122E-01 6.6888E-01 6.2620E-01 6.3501E-01 7.0067E-01 6.5533E-01 6.7629E-01 STD 1.5745E-02 2.9757E-02 8.1257E-02 3.0698E-02 8.1092E-02 1.0536E-01 5.6622E-02 1.2913E-02 7.9663E-02 5.3171E-02 5 AVG 7.1919E-01 7.1526E-01 6.8429E-01 6.9214E-01 6.9503E-01 6.6320E-01 6.8763E-01 7.3470E-01 7.0760E-01 6.8530E-01 STD 1.6720E-02 3.1598E-02 8.4303E-02 5.7804E-02 6.3101E-02 9.1024E-02 3.3063E-02 2.1279E-02 3.4851E-02 5.2850E-02 7 AVG 6.9100E-01 6.8241E-01 6.5940E-01 6.4123E-01 6.9759E-01 6.4132E-01 7.1510E-01 7.4634E-01 7.2149E-01 7.1506E-01 STD 2.8314E-02 4.0034E-02 4.7735E-02 6.0322E-02 3.9878E-02 6.9875E-02 4.3306E-02 3.3709E-02 5.5499E-02 4.4728E-02 12 AVG 8.1887E-01 8.0368E-01 7.9470E-01 7.6567E-01 8.0156E-01 7.7295E-01 7.9706E-01 7.8716E-01 7.9477E-01 7.7749E-01 STD 2.0206E-02 2.6781E-02 4.3917E-02 4.3610E-02 2.7432E-02 6.0547E-02 2.6426E-02 5.8463E-02 2.5627E-02 5.4655E-02 15 AVG 8.4670E-01 8.3836E-01 8.4087E-01 7.8400E-01 8.3612E-01 8.3285E-01 8.1543E-01 8.0406E-01 8.1721E-01 8.0058E-01 STD 2.3096E-02 1.5500E-02 3.1450E-02 3.0979E-02 2.2335E-02 3.1458E-02 2.1305E-02 3.7534E-02 1.8634E-02 3.7918E-02 18 AVG 8.7084E-01 8.5390E-01 8.5149E-01 8.1282E-01 8.5195E-01 8.4366E-01 8.3059E-01 8.2911E-01 8.3127E-01 8.0626E-01 STD 1.3339E-02 2.2638E-02 3.8370E-02 3.0975E-02 2.2424E-02 2.9410E-02 2.4790E-02 3.6184E-02 2.3800E-02 3.6946E-02 C 3 AVG 6.5895E-01 6.5186E-01 5.2981E-01 6.3290E-01 5.7177E-01 5.1504E-01 5.7735E-01 6.2122E-01 5.5437E-01 6.5789E-01 STD 1.8315E-02 1.8531E-02 6.2768E-02 4.6320E-02 6.6199E-02 1.0168E-01 5.4183E-02 6.3635E-02 5.8637E-02 4.2036E-02 5 AVG 6.8228E-01 6.6794E-01 6.1217E-01 6.3252E-01 6.5066E-01 6.0145E-01 6.3005E-01 6.4318E-01 6.1697E-01 6.2981E-01 STD 2.5839E-02 3.6923E-02 5.4237E-02 5.3804E-02 4.1314E-02 6.6486E-02 5.3694E-02 3.6101E-02 5.4420E-02 5.6511E-02 7 AVG 7.3736E-01 7.2807E-01 7.1295E-01 7.1179E-01 7.2149E-01 6.8271E-01 6.6464E-01 6.8524E-01 6.8131E-01 6.4371E-01 STD 2.0610E-02 2.7892E-02 4.6423E-02 3.7153E-02 3.6773E-02 5.4161E-02 3.5462E-02 3.3693E-02 3.8923E-02 6.4276E-02 12 AVG 8.0474E-01 7.8381E-01 7.7061E-01 7.1323E-01 7.7431E-01 7.6539E-01 7.5392E-01 7.5542E-01 7.5444E-01 7.2272E-01 STD 2.4158E-02 3.3710E-02 5.1338E-02 4.2293E-02 4.1878E-02 4.6492E-02 3.8733E-02 3.8607E-02 4.1818E-02 4.5889E-02 15 AVG 8.4802E-01 8.2498E-01 8.1965E-01 7.5144E-01 8.1919E-01 7.8068E-01 7.9722E-01 7.7722E-01 7.8675E-01 7.5819E-01 STD 2.4625E-02 2.5623E-02 3.1346E-02 4.4933E-02 2.8859E-02 4.7470E-02 2.6886E-02 3.3143E-02 5.1240E-02 4.8899E-02 18 AVG 8.7539E-01 8.5908E-01 8.4263E-01 7.9561E-01 8.5265E-01 8.3128E-01 8.1077E-01 8.0397E-01 8.2051E-01 7.9469E-01 STD 1.8056E-02 1.9998E-02 3.2380E-02 4.8895E-02 3.0271E-02 3.4012E-02 3.3721E-02 3.7645E-02 3.4166E-02 3.3611E-02 D 3 AVG 6.9342E-01 6.8722E-01 5.9648E-01 6.7718E-01 6.6693E-01 6.0128E-01 6.4094E-01 6.8515E-01 6.2200E-01 6.8163E-01 STD 1.4469E-02 1.9810E-02 1.0433E-01 5.2629E-02 5.9601E-02 9.7835E-02 7.9040E-02 3.7610E-02 8.7250E-02 5.2638E-02 5 AVG 7.4704E-01 7.2427E-01 7.0491E-01 7.0035E-01 6.9019E-01 6.8814E-01 6.9311E-01 7.2531E-01 7.0557E-01 7.0048E-01 STD 4.0430E-02 5.5802E-02 7.7099E-02 5.3145E-02 6.9898E-02 7.8852E-02 6.5745E-02 6.1982E-02 7.4037E-02 6.9473E-02 7 AVG 7.7004E-01 7.6727E-01 7.4290E-01 7.3619E-01 7.6257E-01 7.4103E-01 7.3909E-01 7.4008E-01 7.1314E-01 7.1999E-01 STD 2.4665E-02 2.2942E-02 4.8488E-02 7.4768E-02 4.0885E-02 4.4109E-02 5.6097E-02 5.1804E-02 6.5950E-02 5.3120E-02 12 AVG 8.2944E-01 8.1394E-01 8.0340E-01 7.8088E-01 8.0607E-01 8.0543E-01 7.8583E-01 7.8314E-01 7.8888E-01 7.5702E-01 STD 1.1865E-02 2.5766E-02 2.9484E-02 3.2796E-02 3.6641E-02 3.2141E-02 3.2671E-02 4.6142E-02 3.8410E-02 5.2316E-02 15 AVG 8.5622E-01 8.4428E-01 8.2970E-01 8.0365E-01 8.2369E-01 8.2083E-01 8.1359E-01 8.2290E-01 8.1502E-01 7.9548E-01 STD 1.4182E-02 1.6310E-02 3.1121E-02 3.0892E-02 3.1192E-02 2.7134E-02 1.5778E-02 2.3226E-02 3.0802E-02 3.6688E-02 18 AVG 8.7306E-01 8.6286E-01 8.5815E-01 8.2224E-01 8.4971E-01 8.4600E-01 8.3801E-01 8.3160E-01 8.3967E-01 8.1918E-01 STD 9.7655E-03 1.4699E-02 2.3031E-02 2.8144E-02 2.1945E-02 2.7801E-02 2.1240E-02 4.0016E-02 1.9429E-02 3.6637E-02 E 3 AVG 7.2772E-01 7.1940E-01 6.6900E-01 7.1415E-01 7.2141E-01 6.4899E-01 6.8921E-01 7.4548E-01 6.9820E-01 7.0457E-01 STD 2.2609E-02 2.6471E-02 8.9038E-02 4.1439E-02 6.8057E-02 9.9420E-02 5.4878E-02 1.0295E-02 6.2621E-02 5.0476E-02 5 AVG 7.5281E-01 7.2824E-01 7.1620E-01 7.3614E-01 7.2849E-01 6.7801E-01 7.3344E-01 7.5339E-01 7.1895E-01 7.3503E-01 STD 1.4053E-02 4.3215E-02 7.8587E-02 3.8477E-02 4.9121E-02 7.8835E-02 3.3751E-02 3.1503E-02 4.6569E-02 3.7415E-02 7 AVG 7.4359E-01 7.3355E-01 6.8211E-01 6.1883E-01 7.2648E-01 6.7387E-01 7.4625E-01 7.6442E-01 7.3243E-01 7.4826E-01 STD 3.7107E-02 4.2067E-02 7.2288E-02 8.0656E-02 4.4149E-02 1.2202E-01 3.4238E-02 3.3003E-02 4.6345E-02 4.2810E-02 12 AVG 8.2519E-01 8.1299E-01 8.0344E-01 7.7894E-01 8.2380E-01 7.8829E-01 8.0049E-01 8.0375E-01 7.8778E-01 7.9290E-01 STD 2.5525E-02 3.1038E-02 3.5507E-02 4.1579E-02 2.1876E-02 3.4467E-02 2.8427E-02 2.4261E-02 3.1940E-02 3.5365E-02 15 AVG 8.5908E-01 8.5102E-01 8.3131E-01 8.1697E-01 8.3824E-01 8.2864E-01 8.2348E-01 8.2944E-01 8.2310E-01 8.0027E-01 STD 1.5262E-02 1.7269E-02 3.3763E-02 2.2848E-02 2.3181E-02 2.9188E-02 2.6739E-02 2.0271E-02 3.3129E-02 2.6964E-02 18 AVG 8.7778E-01 8.6699E-01 8.5460E-01 8.2783E-01 8.4915E-01 8.4482E-01 8.3944E-01 8.4003E-01 8.4026E-01 8.2352E-01 STD 1.3912E-02 1.8916E-02 2.6529E-02 2.3443E-02 2.1912E-02 3.2536E-02 2.1193E-02 2.4260E-02 2.1046E-02 3.0446E-02 F 3 AVG 6.6749E-01 6.7745E-01 6.0256E-01 6.4542E-01 6.5054E-01 6.0616E-01 6.1324E-01 6.6546E-01 5.8334E-01 6.5515E-01 STD 2.0819E-02 2.2285E-02 1.0797E-01 4.9127E-02 6.4971E-02 1.0141E-01 7.4201E-02 2.5306E-02 1.0626E-01 4.8475E-02 5 AVG 6.9876E-01 6.9881E-01 6.6121E-01 6.9058E-01 6.9200E-01 6.6992E-01 6.5605E-01 6.9391E-01 6.5077E-01 6.9140E-01 STD 4.4552E-02 3.9360E-02 8.1132E-02 3.6807E-02 4.4538E-02 9.7269E-02 6.7949E-02 5.9628E-02 6.2322E-02 5.5687E-02 7 AVG 7.5330E-01 7.4758E-01 7.1638E-01 7.2007E-01 7.3636E-01 6.9866E-01 7.0347E-01 7.2662E-01 7.1414E-01 6.8725E-01 STD 2.3774E-02 2.4194E-02 7.4088E-02 4.5955E-02 3.9660E-02 7.8721E-02 4.2906E-02 6.1500E-02 5.4273E-02 5.4880E-02 12 AVG 7.9236E-01 7.8211E-01 7.5882E-01 6.9550E-01 7.7709E-01 7.6503E-01 7.4946E-01 7.4472E-01 7.4080E-01 7.3040E-01 STD 2.8535E-02 2.7107E-02 4.7713E-02 5.0483E-02 3.8022E-02 3.9432E-02 4.2731E-02 4.4175E-02 4.0692E-02 4.5798E-02 15 AVG 8.4570E-01 8.1937E-01 8.2036E-01 7.5407E-01 8.1541E-01 7.9653E-01 7.8789E-01 7.8739E-01 7.9050E-01 7.7538E-01 STD 2.0571E-02 2.3016E-02 3.0452E-02 4.9698E-02 3.5101E-02 4.7812E-02 3.4750E-02 4.7174E-02 4.4526E-02 4.4769E-02 18 AVG 8.7086E-01 8.5521E-01 8.4263E-01 7.8308E-01 8.3205E-01 8.1427E-01 8.1094E-01 8.2539E-01 8.1976E-01 7.8547E-01 STD 2.3433E-02 2.2513E-02 2.9526E-02 4.2203E-02 3.6786E-02 4.5012E-02 3.0926E-02 3.3200E-02 3.6848E-02 3.8536E-02 G 3 AVG 6.2769E-01 5.9565E-01 5.3762E-01 5.8785E-01 5.4657E-01 5.2667E-01 5.4953E-01 5.4805E-01 5.7423E-01 5.9047E-01 STD 4.7366E-02 9.6445E-02 9.3854E-02 5.6528E-02 9.3700E-02 9.0182E-02 9.4927E-02 1.0005E-01 6.8674E-02 4.7509E-02 5 AVG 6.4749E-01 6.3073E-01 6.1259E-01 5.9639E-01 6.3777E-01 5.9572E-01 6.1888E-01 5.9893E-01 5.9533E-01 5.9960E-01 STD 4.8476E-02 4.2783E-02 5.7100E-02 7.3761E-02 4.2863E-02 5.7344E-02 4.3068E-02 5.6103E-02 5.1868E-02 6.1186E-02 7 AVG 7.1605E-01 6.9741E-01 6.7577E-01 6.5275E-01 7.0189E-01 6.6829E-01 6.7123E-01 6.6342E-01 6.6983E-01 6.3043E-01 STD 2.0660E-02 2.9425E-02 4.2691E-02 5.1104E-02 3.3597E-02 4.3774E-02 5.3146E-02 4.4917E-02 3.8250E-02 6.3421E-02 12 AVG 8.1109E-01 8.1082E-01 7.9054E-01 7.5578E-01 8.0014E-01 7.8036E-01 7.7700E-01 7.7845E-01 7.7373E-01 7.4764E-01 STD 2.4778E-02 2.6205E-02 3.8771E-02 3.8661E-02 4.3512E-02 4.5679E-02 3.2747E-02 3.9130E-02 3.3220E-02 4.3377E-02 15 AVG 8.4439E-01 8.3204E-01 8.2775E-01 7.8874E-01 8.3284E-01 8.0892E-01 8.0934E-01 7.9905E-01 7.9934E-01 7.7851E-01 STD 2.2939E-02 2.6082E-02 3.9338E-02 4.6260E-02 2.1829E-02 4.2783E-02 2.5095E-02 3.9251E-02 3.6522E-02 3.5919E-02 18 AVG 8.7535E-01 8.6471E-01 8.6076E-01 8.1074E-01 8.3483E-01 8.5195E-01 8.2892E-01 8.3714E-01 8.2488E-01 7.9812E-01 STD 1.3531E-02 1.8632E-02 2.7852E-02 3.4097E-02 3.2542E-02 3.0728E-02 3.1403E-02 2.8340E-02 3.5154E-02 4.1955E-02 H 3 AVG 7.0748E-01 7.0181E-01 6.3913E-01 6.7283E-01 6.7564E-01 6.0196E-01 6.5342E-01 6.9807E-01 6.2351E-01 6.7263E-01 STD 6.8506E-03 6.4824E-03 7.0011E-02 4.5152E-02 5.7819E-02 9.5877E-02 6.7660E-02 2.8481E-02 6.8694E-02 4.6385E-02 5 AVG 7.2701E-01 7.0928E-01 6.6840E-01 6.9311E-01 7.0238E-01 6.6809E-01 6.8521E-01 7.0505E-01 6.9788E-01 6.7934E-01 STD 1.9006E-02 2.3938E-02 6.7690E-02 3.5509E-02 3.4820E-02 6.6288E-02 3.4686E-02 5.4041E-02 4.3846E-02 5.5992E-02 7 AVG 7.3949E-01 7.3087E-01 7.1575E-01 7.0436E-01 7.3263E-01 7.0173E-01 7.1208E-01 6.9986E-01 7.0891E-01 6.9642E-01 STD 2.1166E-02 4.3498E-02 4.8985E-02 6.8634E-02 4.5503E-02 6.7097E-02 3.6905E-02 5.0028E-02 5.1212E-02 7.6662E-02 12 AVG 8.0032E-01 7.9548E-01 7.8678E-01 7.4498E-01 7.8529E-01 7.7784E-01 7.6552E-01 7.6575E-01 7.7258E-01 7.5170E-01 STD 3.0021E-02 2.6971E-02 4.5365E-02 5.5872E-02 3.0346E-02 3.2109E-02 3.2330E-02 3.3265E-02 3.8439E-02 4.9779E-02 15 AVG 8.4481E-01 8.2131E-01 8.1552E-01 7.8500E-01 8.2736E-01 8.0806E-01 8.0615E-01 7.8697E-01 8.0531E-01 7.7881E-01 STD 1.7584E-02 3.6528E-02 3.3591E-02 3.4006E-02 3.4864E-02 3.0816E-02 3.1940E-02 3.0263E-02 3.5664E-02 3.1607E-02 18 AVG 8.6688E-01 8.5804E-01 8.5153E-01 7.9973E-01 8.4355E-01 8.3854E-01 8.2766E-01 8.0903E-01 8.2149E-01 7.9885E-01 STD 1.2588E-02 1.7556E-02 2.0853E-02 3.0470E-02 2.0605E-02 3.8630E-02 2.4826E-02 2.9432E-02 2.9783E-02 3.8342E-02 Table A.6 The fitness values obtained by each algorithm Table A.6Image Thresholds CCABC ABC WOA SCA MVO HHO CLPSO IGWO IWOA SCADE A 3 3.2379E+01 3.2376E+01 3.2374E+01 3.2106E+01 3.2307E+01 3.2095E+01 3.2186E+01 3.2320E+01 3.2224E+01 3.1915E+01 5 4.4008E+01 4.3465E+01 4.3936E+01 4.1711E+01 4.3954E+01 4.2916E+01 4.2880E+01 4.3769E+01 4.2767E+01 4.2108E+01 7 5.4160E+01 5.3907E+01 5.3931E+01 5.1683E+01 5.4172E+01 5.2873E+01 5.2317E+01 5.3372E+01 5.2203E+01 5.1460E+01 12 7.5545E+01 7.5097E+01 7.3619E+01 6.8010E+01 7.4468E+01 7.3122E+01 7.1414E+01 7.2174E+01 7.2122E+01 6.8362E+01 15 8.6113E+01 8.5329E+01 8.3955E+01 7.7028E+01 8.4731E+01 8.2795E+01 8.0892E+01 8.2368E+01 8.1514E+01 7.6332E+01 18 9.5105E+01 9.4237E+01 9.3568E+01 8.2721E+01 9.2882E+01 9.2196E+01 8.8939E+01 9.0106E+01 9.2296E+01 8.4471E+01 B 3 3.2538E+01 3.2535E+01 3.2529E+01 3.2345E+01 3.2538E+01 3.2463E+01 3.2378E+01 3.2530E+01 3.2465E+01 3.2260E+01 5 4.4276E+01 4.4185E+01 4.4218E+01 4.2951E+01 4.4248E+01 4.3656E+01 4.2812E+01 4.4189E+01 4.3705E+01 4.2613E+01 7 5.4697E+01 5.4154E+01 5.4466E+01 5.1787E+01 5.4528E+01 5.3958E+01 5.2171E+01 5.4490E+01 5.3603E+01 5.0894E+01 12 7.6459E+01 7.5242E+01 7.5695E+01 6.6875E+01 7.5603E+01 7.4968E+01 7.2782E+01 7.2920E+01 7.2568E+01 7.1026E+01 15 8.6131E+01 8.6476E+01 8.6236E+01 7.8721E+01 8.4854E+01 8.4465E+01 8.2515E+01 8.2448E+01 8.2712E+01 7.6185E+01 18 9.5624E+01 9.4743E+01 9.5175E+01 8.4360E+01 9.3465E+01 9.3696E+01 8.9776E+01 9.1127E+01 9.2242E+01 8.4546E+01 C 3 3.2660E+01 3.2338E+01 3.2660E+01 3.2076E+01 3.2645E+01 3.1773E+01 3.2225E+01 3.2605E+01 3.2515E+01 3.2294E+01 5 4.4133E+01 4.3238E+01 4.4209E+01 4.2061E+01 4.3910E+01 4.3311E+01 4.3211E+01 4.3617E+01 4.3230E+01 4.2135E+01 7 5.4419E+01 5.3781E+01 5.4106E+01 5.1670E+01 5.4551E+01 5.3083E+01 5.2801E+01 5.3256E+01 5.3131E+01 5.0267E+01 12 7.6030E+01 7.5304E+01 7.4169E+01 6.9149E+01 7.5797E+01 7.3754E+01 7.2621E+01 7.3300E+01 7.3353E+01 6.7721E+01 15 8.6163E+01 8.6351E+01 8.5866E+01 7.6255E+01 8.6014E+01 8.3312E+01 8.3137E+01 8.1822E+01 8.2389E+01 7.5951E+01 18 9.5809E+01 9.5740E+01 9.4473E+01 8.4126E+01 9.4386E+01 9.2973E+01 9.0121E+01 9.0876E+01 9.1056E+01 8.4441E+01 D 3 3.2006E+01 3.2005E+01 3.2003E+01 3.1736E+01 3.2006E+01 3.1989E+01 3.1873E+01 3.1998E+01 3.1935E+01 3.1644E+01 5 4.4015E+01 4.3714E+01 4.3954E+01 4.2013E+01 4.3964E+01 4.3792E+01 4.2909E+01 4.3643E+01 4.3150E+01 4.1993E+01 7 5.4418E+01 5.3982E+01 5.4147E+01 5.1012E+01 5.4090E+01 5.3170E+01 5.2584E+01 5.3488E+01 5.2890E+01 4.9759E+01 12 7.5748E+01 7.5329E+01 7.4947E+01 6.8591E+01 7.5046E+01 7.4738E+01 7.1183E+01 7.2953E+01 7.2310E+01 6.7728E+01 15 8.6006E+01 8.6240E+01 8.6019E+01 7.7327E+01 8.5696E+01 8.4591E+01 8.2699E+01 8.2660E+01 8.4302E+01 7.7323E+01 18 9.6371E+01 9.5265E+01 9.4804E+01 8.3787E+01 9.5096E+01 9.2151E+01 8.9580E+01 9.0601E+01 9.0986E+01 8.2188E+01 E 3 3.2615E+01 3.2608E+01 3.2611E+01 3.2510E+01 3.2615E+01 3.2526E+01 3.2426E+01 3.2616E+01 3.2580E+01 3.2415E+01 5 4.4411E+01 4.4254E+01 4.4317E+01 4.2846E+01 4.4416E+01 4.3833E+01 4.3624E+01 4.4339E+01 4.4341E+01 4.2912E+01 7 5.4733E+01 5.4725E+01 5.4597E+01 5.1726E+01 5.4626E+01 5.3719E+01 5.2860E+01 5.4080E+01 5.2822E+01 5.1952E+01 12 7.6277E+01 7.5536E+01 7.5049E+01 6.8902E+01 7.5326E+01 7.3780E+01 7.2563E+01 7.5025E+01 7.3099E+01 6.9137E+01 15 8.7038E+01 8.6387E+01 8.5926E+01 7.5918E+01 8.5473E+01 8.3965E+01 8.0900E+01 8.2646E+01 8.3296E+01 7.6781E+01 18 9.5977E+01 9.5540E+01 9.4388E+01 8.6877E+01 9.4400E+01 9.4013E+01 8.8299E+01 8.9902E+01 8.8715E+01 8.3884E+01 F 3 3.2292E+01 3.2291E+01 3.2302E+01 3.1992E+01 3.2303E+01 3.2186E+01 3.2126E+01 3.2297E+01 3.2144E+01 3.2041E+01 5 4.4366E+01 4.3862E+01 4.4177E+01 4.2387E+01 4.4323E+01 4.3789E+01 4.2792E+01 4.4192E+01 4.3563E+01 4.3169E+01 7 5.4445E+01 5.4342E+01 5.4432E+01 5.1564E+01 5.4385E+01 5.3614E+01 5.2533E+01 5.3403E+01 5.2787E+01 5.2056E+01 12 7.4912E+01 7.5262E+01 7.5038E+01 6.9224E+01 7.4865E+01 7.3735E+01 7.1417E+01 7.3573E+01 7.2610E+01 6.7913E+01 15 8.5865E+01 8.5647E+01 8.5454E+01 7.7280E+01 8.4292E+01 8.3536E+01 8.1815E+01 8.1939E+01 8.2275E+01 7.6165E+01 18 9.5500E+01 9.5274E+01 9.3557E+01 8.6031E+01 9.3884E+01 9.3265E+01 9.0765E+01 9.0046E+01 9.0515E+01 8.6786E+01 G 3 3.2790E+01 3.2562E+01 3.2788E+01 3.2345E+01 3.2801E+01 3.2589E+01 3.2727E+01 3.2772E+01 3.2406E+01 3.2146E+01 5 4.4643E+01 4.4645E+01 4.4537E+01 4.2292E+01 4.4316E+01 4.4347E+01 4.3433E+01 4.3708E+01 4.4422E+01 4.2192E+01 7 5.5058E+01 5.5245E+01 5.5091E+01 5.0557E+01 5.4890E+01 5.4307E+01 5.3841E+01 5.3389E+01 5.4121E+01 4.9984E+01 12 7.7011E+01 7.5181E+01 7.5137E+01 6.8725E+01 7.5421E+01 7.4132E+01 7.2757E+01 7.2394E+01 7.4175E+01 6.7608E+01 15 8.7044E+01 8.6391E+01 8.6892E+01 7.6652E+01 8.5814E+01 8.4090E+01 8.1642E+01 8.2771E+01 8.4775E+01 7.9031E+01 18 9.5989E+01 9.5296E+01 9.4713E+01 8.5726E+01 9.3939E+01 9.3189E+01 8.9209E+01 8.9329E+01 9.1520E+01 8.4408E+01 H 3 3.2382E+01 3.2356E+01 3.2382E+01 3.2218E+01 3.2375E+01 3.2313E+01 3.2034E+01 3.2379E+01 3.2152E+01 3.2252E+01 5 4.4340E+01 4.3981E+01 4.4247E+01 4.2712E+01 4.4334E+01 4.3697E+01 4.3012E+01 4.4147E+01 4.3721E+01 4.1737E+01 7 5.4687E+01 5.4595E+01 5.4221E+01 4.9820E+01 5.4696E+01 5.3163E+01 5.2680E+01 5.3599E+01 5.3654E+01 4.9599E+01 12 7.5766E+01 7.5623E+01 7.4364E+01 6.8579E+01 7.5194E+01 7.3299E+01 7.1889E+01 7.3417E+01 7.4093E+01 6.8809E+01 15 8.6553E+01 8.5602E+01 8.5448E+01 7.6990E+01 8.5793E+01 8.4478E+01 8.1322E+01 8.2316E+01 8.2101E+01 7.6192E+01 18 9.5320E+01 9.6354E+01 9.5919E+01 8.3236E+01 9.4552E+01 9.3899E+01 8.9438E+01 8.9907E+01 9.1793E+01 8.3200E+01 Table A.7 Time cost of CCABC and other algorithms (Unit: sec) Table A.7Image CCABC ABC WOA SCA MVO HHO CLPSO IGWO IWOA SCADE A 48.86 24.36 24.20 26.23 24.53 48.23 23.50 35.94 3.76 72.54 B 48.23 23.68 24.04 26.53 24.61 48.02 23.32 36.75 3.70 71.98 C 48.48 23.84 24.32 26.49 24.60 48.80 24.08 36.16 3.79 73.36 D 50.42 24.51 25.56 26.88 24.93 51.05 24.82 36.48 4.01 72.30 E 50.85 25.08 26.09 26.88 25.79 50.12 25.32 37.26 4.26 69.95 F 50.58 25.37 25.77 25.44 25.21 50.96 25.41 37.65 4.44 68.04 G 51.31 24.81 26.12 26.96 25.60 50.72 25.76 38.38 4.36 69.58 H 46.29 23.73 23.94 24.59 23.82 47.47 23.27 34.06 3.77 65.62 Appendix B Fig. B.1 The average of FSIM at each threshold level Fig. B.1 Fig. B.2 The average of PSNR at each threshold level Fig. B.2 Fig. B.3 The average of SSIM at each threshold level Fig. B.3 Fig. B.4 The threshold values of A obtained by each algorithm at level 7 Fig. B.4 Fig. B.5 The threshold values of B obtained by each algorithm at level 7 Fig. B.5 Fig. B.6 The threshold values of C obtained by each algorithm at level 7 Fig. B.6 Fig. B.7 The threshold values of D obtained by each algorithm at level 7 Fig. B.7 Fig. B.8 The threshold values of E obtained by each algorithm at level 7 Fig. B.8 Fig. B.9 The threshold values of F obtained by each algorithm at level 7 Fig. B.9 Fig. B.10 The threshold values of G obtained by each algorithm at level 7 Fig. B.10 Fig. B.11 The threshold values of H obtained by each algorithm at level 7 Fig. B.11 Acknowledgments This research is supported by the 10.13039/501100004731 Natural Science Foundation of Zhejiang Province (LZ22F020005), 10.13039/501100001809 National Natural Science Foundation of China (62076185, U1809209, 81873949), Medical Innovation Discipline of Zhejiang Province (Critical Care Medicine, Y2015), 10.13039/501100007194 Wenzhou Science and Technology Bureau (2018ZG016, Y20210097), Wenzhou Key Technology Breakthrough Program on Prevention and Treatment for COVID-19 Epidemic (ZG2020012), University-Industry Collaborative Education Program. Minitry of Education. PRC. (202002236013). It is also supported by the Joint fund of Science & Technology Department of Liaoning Province and 10.13039/501100011259 State Key Laboratory of Robotics , China (2020-KF-22-08), the “Thirteenth Five-Year” Science and Technology Project of Jilin Provincial Department of Education (JJKH20200829KJ), Science and Technology Research Project of Jilin Provincial Education Department (JJKH20210888KJ), 10.13039/501100008344 Changchun Normal University Ph.D. Research Startup Funding Project (BS [2020]), the 5G Network-based Platform for Precision Emergency Medical Care in Regional Hospital Clusters. MIIT.PRC (2020NO.78) 2 https://aliasgharheidari.com/HGS.html. 3 https://aliasgharheidari.com/HHO.html. 4 https://aliasgharheidari.com/SMA.html. 5 https://aliasgharheidari.com/RUN.html. ==== Refs References 1 Zhou W. Lv Y. Lei J. Yu L. Global and local-contrast guides content-aware fusion for RGB-D saliency prediction IEEE Transactions on Systems, Man, and Cybernetics: Systems 51 2021 3641 3649 2 Zhao S. Wang P. Heidari A.A. Chen H. Turabieh H. Mafarja M. Li C. Multilevel threshold image segmentation with diffusion association slime mould algorithm and Renyi's entropy for chronic obstructive pulmonary disease Comput. Biol. Med. 134 2021 104427 3 Liu L. Zhao D. Yu F. Heidari A.A. Ru J. Chen H. Mafarja M. Turabieh H. Pan Z. Performance optimization of differential evolution with slime mould algorithm for multilevel breast cancer image segmentation Comput. Biol. Med. 138 2021 104910 4 Zhao S. Wang P. Heidari A.A. Chen H. He W. Xu S. Performance optimization of salp swarm algorithm for multi-threshold image segmentation: comprehensive study of breast cancer microscopy Comput. Biol. Med. 139 2021 105015 5 Chen C. Wang X. Heidari A.A. Yu H. Chen H. Multi-threshold image segmentation of maize diseases based on elite comprehensive particle swarm optimization and Otsu Front. Plant Sci. 12 2021 6 Abd Elaziz M. Oliva D. Ewees A.A. Xiong S.W. Multi-level thresholding-based grey scale image segmentation using multi-objective multi-verse optimizer Expert Syst. Appl. 125 2019 112 129 7 Gao J.Q. Wang B.B. Wang Z.Y. Wang Y.F. Kong F.Z. A wavelet transform-based image segmentation method Optik 2020 208 8 Liang Y.F. Sun L. Ser W. Lin F. Tay E.Y. Gan E.Y. Thng T.G. Lin Z.P. Hybrid threshold optimization between global image and local regions in image segmentation for melasma severity assessment Multidimens. Syst. Signal Process. 28 2017 977 994 9 Zheng Q. Wu Y.H. Fan Y. I. Alzheimer's dis neuroimaging, integrating semi-supervised and supervised learning methods for label fusion in multi-atlas based image segmentation Front. Neuroinf. 12 2018 10 Zhao X.H. Jiang Y.C. Wang W.Q. SAR image segmentation based on analysing similarity with clutter spatial patterns Electron. Lett. 2016 52 11 Chen L.C. Papandreou G. Kokkinos I. Murphy K. Yuille A.L. DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs IEEE Trans. Pattern Anal. Mach. Intell. 40 2018 834 848 28463186 12 Huang L. Yao B.X. Chen P.D. Ren A.P. Xia Y. Superpixel segmentation method of high resolution remote sensing images based on hierarchical clustering J. Infrared Millim. Waves 39 2020 263 272 13 Peng H. Wang J. Perez-Jimenez M.J. Optimal multi-level thresholding with membrane computing Digit. Signal Process. 37 2015 53 64 14 Zhao D. Liu L. Yu F. Heidari A.A. Chen H. Chaotic random spare ant colony optimization for multi-threshold image segmentation of 2D Kapur entropy Knowl. Base Syst. 216 2020 106510 15 Zhang Z.C. Yin J.Q. Bee foraging algorithm based multi-level thresholding for image segmentation Ieee Access 8 2020 16269 16280 16 Xing Z.K. Jia H.M. An improved thermal exchange optimization based GLCM for multi-level image segmentation Multimed. Tool. Appl. 79 2020 12007 12040 17 Wu B. Zhou J.X. Ji X.Y. Yin Y.J. Shen X. An ameliorated teaching-learning-based optimization algorithm based study of image segmentation for multilevel thresholding using Kapur's entropy and Otsu's between class variance Inf. Sci. 533 2020 72 107 18 Ewees A.A. Abd Elaziz M. Al-Qaness M.A.A. Khalil H.A. Kim S. Improved artificial bee colony using sine-cosine algorithm for multi-level thresholding image segmentation Ieee Access 8 2020 26304 26315 19 Alwerfali H.S.N. Al-qaness M.A.A. Abd Elaziz M. Ewees A.A. Oliva D. Lu S.F. Multi-level image thresholding based on modified spherical search optimizer and fuzzy entropy Entropy 22 2020 20 Abd Elaziz M. Heidari A.A. Fujita H. Moayedi H. A competitive chain-based Harris Hawks Optimizer for global optimization and multi-level image thresholding problems Appl. Soft Comput. 95 2020 21 Zhan Z.-H. Wang Z.-J. Jin H. Zhang J. Adaptive distributed differential evolution IEEE Trans. Cybern. 50 2019 4633 4647 22 Liu X.-F. Zhan Z.-H. Lin Y. Chen W.-N. Gong Y.-J. Gu T.-L. Yuan H.-Q. Zhang J. Historical and heuristic-based adaptive differential evolution IEEE Transactions on Systems, Man, and Cybernetics: Systems 49 2018 2623 2635 23 Sun G. Li C. Deng L. An adaptive regeneration framework based on search space adjustment for differential evolution Neural Comput. Appl. 2021 1 17 24 Yin F. Xue X. Zhang C. Zhang K. Han J. Liu B. Wang J. Yao J. Multifidelity genetic transfer: an efficient framework for production optimization SPE J. 2021 1 22 25 Song S. Wang P. Heidari A.A. Wang M. Zhao X. Chen H. He W. Xu S. Dimension decided Harris hawks optimization with Gaussian mutation: balance analysis and diversity patterns Knowl. Base Syst. 2020 106425 10.1016/j.knosys.2020.106425 26 Wang X. Chen H. Heidari A.A. Zhang X. Xu J. Xu Y. Huang H. Multi-population following behavior-driven fruit fly optimization: a Markov chain convergence proof and comprehensive analysis Knowl. Base Syst. 210 2020 106437 27 Moayedi H. Mosavi A. Synthesizing multi-layer perceptron network with ant lion biogeography-based dragonfly algorithm evolutionary strategy invasive weed and league champion optimization hybrid algorithms in predicting heating load in residential buildings Sustainability 13 2021 3198 28 Moayedi H. Mosavi A. Suggesting a stochastic fractal search paradigm in combination with artificial neural network for early prediction of cooling load in residential buildings Energies 14 2021 1649 29 Moayedi H. Mosavi A. An innovative metaheuristic strategy for solar energy management through a neural networks framework Energies 14 2021 1196 30 Moayedi H. Mosavi A. Electrical power prediction through a combination of multilayer perceptron with water cycle ant lion and satin bowerbird searching optimizers Sustainability 13 2021 2336 31 Seifi A. Ehteram M. Singh V.P. Mosavi A. Modeling and uncertainty analysis of groundwater level using six evolutionary optimization algorithms hybridized with ANFIS, SVM, and ANN, Sustainability 12 2020 4023 32 Zhang Y. Liu R. Wang X. Chen H. Li C. Boosted binary Harris hawks optimizer and feature selection Eng. Comput. 2020 1 30 33 Hu J. Chen H. Heidari A.A. Wang M. Zhang X. Chen Y. Pan Z. Orthogonal learning covariance matrix for defects of grey wolf optimizer: insights, balance, diversity, and feature selection Knowl. Base Syst. 213 2021 106684 34 Zhang X. Xu Y. Yu C. Heidari A.A. Li S. Chen H. Li C. Gaussian mutational chaotic fruit fly-built optimization and feature selection Expert Syst. Appl. 141 2020 112976 35 Li Q. Chen H. Huang H. Zhao X. Cai Z. Tong C. Liu W. Tian X. An enhanced grey wolf optimization based feature selection wrapped kernel extreme learning machine for medical diagnosis Comput. Math. Methods Med. 2017 2017 36 Liu T. Hu L. Ma C. Wang Z.-Y. Chen H.-L. A fast approach for detection of erythemato-squamous diseases based on extreme learning machine with maximum relevance minimum redundancy feature selection Int. J. Syst. Sci. 46 2015 919 931 37 Chen M. Zeng G. Lu K. Weng J. A two-layer nonlinear combination method for short-term wind speed prediction based on ELM, ENN, and LSTM IEEE Internet Things J. 6 2019 6997 7010 38 Yu H. Li W. Chen C. Liang J. Gui W. Wang M. Chen H. Dynamic Gaussian bare-bones fruit fly optimizers with abandonment mechanism: method and analysis Eng. Comput. 2020 1 29 39 Nautiyal B. Prakash R. Vimal V. Liang G. Chen H. Improved Salp Swarm Algorithm with mutation schemes for solving global optimization and engineering problems Eng. Comput. 2021 1 23 40 Zhao D. Liu L. Yu F. Heidari A.A. Wang M. Liang G. Muhammad K. Chen H. Chaotic random spare ant colony optimization for multi-threshold image segmentation of 2D Kapur entropy Knowl. Base Syst. 2020 106510 10.1016/j.knosys.2020.106510 41 Ahmed K. Babers R. Darwish A. Hassanien A.E. Swarm-based Analysis for Community Detection in Complex Networks, Big Data Analytics 2018 CRC Press 213 230 42 Ahmed K. Hassanien A.E. Ezzat E. An efficient approach for community detection in complex social networks based on elephant swarm optimization algorithm Handbook of Research on Machine Learning Innovations and Trends vol. 2 2017 43 Ahmed K. Hassanien A. Ezzat E. Tsai P.W. An adaptive approach for community detection based on chicken swarm optimization algorithm Genet. Evol. Comput. 536 2016 281 288 44 Ahmed K. Hafez A.I. Hassanien A.E. A Discrete Krill Herd Optimization Algorithm for Community Detection, 11th International Computer Engineering Conference (ICENCO) 2015 2016 IEEE 45 Ahmed K. Hassanien A.E. Ezzat E. Bhattacharyya S. Swarming Behaviors of Chicken for Predicting Posts on Facebook Branding Pages 2018 International Conference on Advanced Machine Learning Technologies and Applications 46 Zeng G.-q. Lu Y.-z. Mao W.-J. Modified extremal optimization for the hard maximum satisfiability problem J. Zhejiang Univ. - Sci. C 12 2011 589 596 47 Zeng G. Lu Y. Dai Y. Wu Z. Mao W. Zhang Z. Zheng C.J. Backbone guided extremal optimization for the hard maximum satisfiability problem Int. J. Innov. Comput. Inf. Contr. 8 2012 8355 8366 48 Ahmed K. Ewees A.A. Hassanien A.E. Prediction and management system for forest fires based on hybrid flower pollination optimization algorithm and adaptive neuro-fuzzy inference system 2017 Eighth International Conference on Intelligent Computing and Information Systems (ICICIS) 2017 IEEE 299 304 49 Ahmed K. Ewees A.A. Abd El Aziz M. Hassanien A.E. Gaber T. Tsai P.-W. Pan J.-S. A hybrid krill-ANFIS model for wind speed forecasting International Conference on Advanced Intelligent Systems and Informatics 2016 Springer 365 372 50 Hu L. Li H. Cai Z. Lin F. Hong G. Chen H. Lu Z. A new machine-learning method to prognosticate paraquat poisoned patients by combining coagulation, liver, and kidney indices PLoS One 12 2017 e0186427 51 Li C. Hou L. Sharma B.Y. Li H. Chen C. Li Y. Zhao X. Huang H. Cai Z. Chen H. Developing a new intelligent system for the diagnosis of tuberculous pleural effusion Comput. Methods Progr. Biomed. 153 2018 211 225 52 Zhao X. Zhang X. Cai Z. Tian X. Wang X. Huang Y. Chen H. Hu L. Chaos enhanced grey wolf optimization wrapped ELM for diagnosis of paraquat-poisoned patients Comput. Biol. Chem. 78 2019 481 490 30501982 53 Huang H. Zhou S. Jiang J. Chen H. Li Y. Li C. A new fruit fly optimization algorithm enhanced support vector machine for diagnosis of breast cancer based on high-level features BMC Bioinf. 20 2019 1 14 54 Basha S.H. Tharwat A. Ahmed K. Hassanien A.E. A Predictive Model for Seminal Quality Using Neutrosophic Rule-Based Classification System, International Conference on Advanced Intelligent Systems and Informatics 2018 Springer 495 504 55 Zhao C. Zhong S. Zhong Q. Shi K. Synchronization of Markovian complex networks with input mode delay and Markovian directed communication via distributed dynamic event-triggered control Nonlinear Anal.: Hybrid Syst. 36 2020 100883 56 Zhang Y. Liu R. Heidari A.A. Wang X. Chen Y. Wang M. Chen H. Towards Augmented Kernel Extreme Learning Models for Bankruptcy Prediction: Algorithmic Behavior and Comprehensive Analysis 2020 Neurocomputing 10.1016/j.neucom.2020.1010.1038 57 Yu C. Chen M. Cheng K. Zhao X. Ma C. Kuang F. Chen H. SGOA: annealing-behaved grasshopper optimizer for global tasks Eng. Comput. 2021 10.1007/s00366-00020-01234-00361 58 Cai Z. Gu J. Luo J. Zhang Q. Chen H. Pan Z. Li Y. Li C. Evolving an optimal kernel extreme learning machine by using an enhanced grey wolf optimization strategy Expert Syst. Appl. 138 2019 112814 59 Shen L. Chen H. Yu Z. Kang W. Zhang B. Li H. Yang B. Liu D. Evolving support vector machines using fruit fly optimization for medical data classification Knowl. Base Syst. 96 2016 61 75 60 Wang M. Chen H. Yang B. Zhao X. Hu L. Cai Z. Huang H. Tong C. Toward an optimal kernel extreme learning machine using a chaotic moth-flame optimization strategy with applications in medical diagnoses Neurocomputing 267 2017 69 84 61 Wang M. Chen H. Chaotic multi-swarm whale optimizer boosted support vector machine for medical diagnosis Appl. Soft Comput. 88 2020 105946 62 Zeng G.-Q. Lu K.-D. Dai Y.-X. Zhang Z.-J. Chen M.-R. Zheng C.-W. Wu D. Peng W.-W.J.N. Binary-coded extremal optimization for the design of PID controllers Neurocomputing 138 2014 180 188 63 Zeng G.-Q. Chen J. Dai Y.-X. Li L.-M. Zheng C.-W. Chen M.-R.J.N. Design of fractional order PID controller for automatic regulator voltage system based on multi-objective extremal optimization Neurocomputing 160 2015 173 184 64 Zeng G.-Q. Xie X.-Q. Chen M.-R. Weng J. Adaptive population extremal optimization-based PID neural network for multivariable nonlinear control systems Swarm Evol. Comput. 44 2019 320 334 65 Deng W. Liu H. Xu J. Zhao H. Song Y.J. Measurement, an Improved Quantum-Inspired Differential Evolution Algorithm for Deep Belief Network 2020 IEEE Transactions on Instrumentation & Measurement 10.1109/TIM.2020.2983233 66 Zhao H. Liu H. Xu J. Deng W.J. Measurement, Performance Prediction Using High-Order Differential Mathematical Morphology Gradient Spectrum Entropy and Extreme Learning Machine 2019 IEEE Transactions on Instrumentation & Measurement 10.1109/TIM.2019.2948414 67 Liu X.-F. Zhan Z.-H. Zhang J. Resource-aware distributed differential evolution for training expensive neural-network-based controller in power electronic circuit IEEE Transact. Neural Networks Learn. Syst. 2021 10.1109/TNNLS.2021.3075205 68 Zhan Z.-H. Liu X.-F. Zhang H. Yu Z. Weng J. Li Y. Gu T. Zhang J. Cloudde: A heterogeneous differential evolution algorithm and its distributed cloud version IEEE Trans. Parallel Distr. Syst. 28 2016 704 716 69 Zhao X. Li D. Yang B. Ma C. Zhu Y. Chen H. Feature selection based on improved ant colony optimization for online detection of foreign fiber in cotton Appl. Soft Comput. 24 2014 585 596 70 Zhao X. Li D. Yang B. Chen H. Yang X. Yu C. Liu S. A two-stage feature selection method with its application Comput. Electr. Eng. 47 2015 114 125 71 Wei Y. Lv H. Chen M. Wang M. Heidari A.A. Chen H. Li C. Predicting entrepreneurial intention of students: an extreme learning machine with Gaussian barebone Harris hawks optimizer IEEE Access 8 2020 76841 76855 72 Zhu W. Ma C. Zhao X. Wang M. Heidari A.A. Chen H. Li C. Evaluation of sino foreign cooperative education project using orthogonal sine cosine optimized kernel extreme learning machine IEEE Access 8 2020 61107 61123 73 Lin A. Wu Q. Heidari A.A. Xu Y. Chen H. Geng W. Li C. Predicting intentions of students for master programs using a chaos-induced sine cosine-based fuzzy K-Nearest neighbor classifier Ieee Access 7 2019 67235 67248 74 Tu J. Lin A. Chen H. Li Y. Li C. Predict the entrepreneurial intention of fresh graduate students based on an adaptive support vector machine framework Math. Probl Eng. 2019 1 16 2019 75 Wei Y. Ni N. Liu D. Chen H. Wang M. Li Q. Cui X. Ye H. An improved grey wolf optimization strategy enhanced SVM and its application in predicting the second major Math. Probl Eng. 2017 1 12 2017 76 Kennedy J. Particle swarm optimization Proc. of 1995 IEEE Int 4 2011 1942 1948 77 Wang G.-G. Moth search algorithm: a bio-inspired metaheuristic algorithm for global optimization problems Memet. Comput. 10 2018 151 164 78 Feng Y. Deb S. Wang G.-G. Alavi A.H. Monarch butterfly optimization: a comprehensive review Expert Syst. Appl. 168 2021 114418 79 Yang Y. Chen H. Heidari A.A. Gandomi A.H. Hunger games search: visions, conception, implementation, deep analysis, perspectives, and towards performance shifts Expert Syst. Appl. 2021 114864 80 Heidari A.A. Mirjalili S. Faris H. Aljarah I. Mafarja M. Chen H. Harris hawks optimization: algorithm and applications Future Gen. Comput. Syst. Int. J. Esci. 97 2019 849 872 81 Li S. Chen H. Wang M. Heidari A.A. Mirjalili S. Slime mould algorithm: a new method for stochastic optimization Future Generat. Comput. Syst. 111 2020 300 323 82 Ahmadianfar I. Asghar Heidari A. Gandomi A.H. Chu X. Chen H. RUN beyond the metaphor: an efficient optimization algorithm based on Runge Kutta method Expert Syst. Appl. 2021 115079 83 Tu J. Chen H. Wang M. Gandomi A.H. The colony predation algorithm JBE 18 2021 674 710 84 Karaboga D. Basturk B. A powerful and efficient algorithm for numerical function optimization: artificial bee colony (ABC) algorithm J. Global Optim. 39 2007 459 471 85 Zou W.Q. Pan Q.K. Meng T. Gao L. Wang Y.L. An effective discrete artificial bee colony algorithm for multi-AGVs dispatching problem in a matrix manufacturing workshop Expert Syst. Appl. 2020 161 86 Zhou K. Wen Y.Z. Wu W.Y. Ni Z.Y. Jin T.G. Long X.J. Cloud service optimization method based on dynamic artificial ant-bee colony algorithm in agricultural equipment manufacturing Math. Probl. Eng. 2020 2020 87 Zhao C.J. Zhao H.D. Wang G.Z. Chen H. Improvement SVM classification performance of hyperspectral image using chaotic sequences in artificial bee colony Ieee Access 8 2020 73947 73956 88 Lin G. Xu H.P. Chen X. Guan J. An effective binary artificial bee colony algorithm for maximum set k-covering problem Expert Syst. Appl. 2020 161 89 Ding Z.H. Fu K.S. Deng W. Li J. Lu Z.R. A modified Artificial Bee Colony algorithm for structural damage identification under varying temperature based on a novel objective function Appl. Math. Model. 88 2020 122 141 90 Chu X.H. Li S.X. Gao D. Zhao W. Cui J.S. Huang L.Y. A binary superior tracking artificial bee colony with dynamic Cauchy mutation for feature selection Complexity 2020 2020 91 Chen X. Wei X. Yang G.X. Du W.L. Fireworks explosion based artificial bee colony for numerical optimization Knowl. Base Syst. 2020 188 92 Boudardara F. Gorkemli B. Solving artificial ant problem using two artificial bee colony programming versions Appl. Intell. 50 2020 3695 3717 93 Yue Y.G. Cao L. Luo Z.Q. Hybrid artificial bee colony algorithm for improving the coverage and connectivity of wireless sensor networks Wireless Pers. Commun. 108 2019 1719 1732 94 García S. Fernández A. Luengo J. Herrera F. Advanced nonparametric tests for multiple comparisons in the design of experiments in computational intelligence and data mining: experimental analysis of power Inf. Sci. 180 2010 2044 2064 95 Derrac J. García S. Molina D. Herrera F. A practical tutorial on the use of nonparametric statistical tests as a methodology for comparing evolutionary and swarm intelligence algorithms Swarm Evol. Comput. 1 2011 3 18 96 Zhang L. Zhang L. Mou X. Zhang D. FSIM: a feature similarity Index for image quality assessment IEEE Trans. Image Process. 20 2011 2378 2386 21292594 97 Huynh-Thu Q. Ghanbari M. Scope of validity of PSNR in image/video quality assessment Electron. Lett. 44 2008 800 801 98 Zhou W. Bovik A.C. Sheikh H.R. Simoncelli E.P. Image quality assessment: from error visibility to structural similarity IEEE Trans. Image Process. 13 2004 600 612 15376593 99 Abutaleb A.S. Automatic thresholding of gray-level pictures using two-dimensional entropy Comput. Vis. Graph Image Process 47 1989 22 32 100 Buades A. Coll B. Morel J. A non-local algorithm for image denoising 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) vol. 62 2005 60 65 101 Meng A.B. Chen Y.C. Yin H. Chen S.Z. Crisscross optimization algorithm and its application Knowl. Base Syst. 67 2014 218 229 102 Patwal R.S. Narang N. Crisscross PSO algorithm for multi-objective generation scheduling of pumped storage hydrothermal system incorporating solar units Energy Convers. Manag. 169 2018 238 254 103 Zhao D. Liu L. Yu F. Heidari A.A. Chen H. Ant colony optimization with horizontal and vertical crossover search: fundamental visions for multi-threshold image segmentation Expert Syst. Appl. 167 2020 114122 104 Liu R. Wang X. Lu H. Wu Z. Fan Q. Li S. Jin X. SCCGAN: style and characters inpainting based on CGAN Mobile Network. Appl. 26 2021 3 12 105 Zhong L. Fang Z. Liu F. Yuan B. Zhang G. Lu J. Bridging the theoretical bound and deep algorithms for open set domain adaptation IEEE Transact. Neural Networks Learn. Syst. 2021 1 15 106 Chen J. Zou Q. Li J. DeepM6ASeq-EL: prediction of human N6-methyladenosine (m6A) Sites with LSTM and ensemble learning Front. Comput. Sci. 16 2022 1 7 107 Zhang Y. Shi X. Zhang H. Cao Y. Terzija V. Review on deep learning applications in frequency analysis and control of modern power system Int. J. Electr. Power Energy Syst. 136 2022 107744 108 Uyar Cankay T. Besenek M. What do we over-look during covid-19 pandemic? an adolescent stroke case presumed conversion disorder Psychiatr. Danub. 32 2020 300 302 32796802 109 Salopek-Žiha D. Hlavati M. Gvozdanović Z. Gašić M. Placento H. Jakić H. Klapan D. Šimić H. Differences in distress and coping with the COVID-19 stressor in nurses and physicians Psychiatr. Danub. 32 2020 287 293 32796800 110 Ćosić K. Popović S. Šarlija M. Kesedžić I. Impact of human disasters and COVID-19 pandemic on mental health: potential of digital psychiatry Psychiatr. Danub. 32 2020 25 31 32303026 111 Liu L. Zhao D. Yu F. Heidari A.A. Li C. Ouyang J. Chen H. Mafarja M. Turabieh H. Pan J. Ant colony optimization with Cauchy and greedy Levy mutations for multilevel COVID 19 X-ray image segmentation Comput. Biol. Med. 136 2021 104609 112 Zhang Q. Wang Z. Heidari A.A. Gui W. Shao Q. Chen H. Zaguia A. Turabieh H. Chen M. Gaussian barebone salp swarm algorithm with stochastic fractal search for medical image segmentation: a COVID-19 case study Comput. Biol. Med. 139 2021 104941 113 Mirjalili S. Dong J.S. Lewis A. Nature-inspired Optimizers: Theories, Literature Reviews and Applications 2019 Springer 114 Aad A. Ass B. Smc D. Kzg E. Nonlinear-based Chaotic Harris Hawks Optimizer: Algorithm and Internet of Vehicles Application 2021 Applied Soft Computing 115 Adarsh B.R. Raghunathan T. Jayabarathi T. Yang X.-S. Economic dispatch using chaotic bat algorithm Energy 96 2016 666 675 116 Nenavath H. Jatoth R.K. Hybridizing sine cosine algorithm with differential evolution for global optimization and object tracking Appl. Soft Comput. 62 2018 1019 1043 117 Tubishat M. Abushariah M.A.M. Idris N. Aljarah I. Improved whale optimization algorithm for feature selection in Arabic sentiment analysis Appl. Intell. 49 2019 1688 1707 118 Elhosseini M.A. Haikal A.Y. Badawy M. Khashan N. Biped robot stability based on an A–C parametric Whale Optimization Algorithm J. Comput. Sci. 31 2019 17 32 119 Issa M. Hassanien A. Oliva D. Helmi A. Ziedan I. Alzohairy A. ASCA-PSO: Adaptive sine cosine optimization algorithm integrated with particle swarm for pairwise local sequence alignment Expert Syst. Appl. 99 2018 56 70 120 Gao W.F. Liu S.Y. A modified artificial bee colony algorithm Comput. Oper. Res. 39 2012 687 697 121 Zhong X. Jiang S. Song H. ABCGA Algorithm for the Two Echelon Vehicle Routing Problem 2017 ABCGA Algorithm for the Two Echelon Vehicle Routing Problem 122 Hanbay K. Talu M.F. Segmentation of SAR images using improved artificial bee colony algorithm and neutrosophic set Appl. Soft Comput. 21 2014 433 443 123 Ilmawan D.M.H. Warsito B. Sugito S. Penerapan Artificial Neural Network Dengan Optimasi Modified Artificial Bee Colony Untuk Meramalkan Harga Bitcoin Terhadap Rupiah 2020 124 Zhou Y. Xu G. Tang K. Tian L. Sun Y. Video coding optimization in AVS2 Inf. Process. Manag. 59 2022 102808 125 Xu Q. Guo Q. Wang C.-X. Zhang S. Wen C.-B. Sun T. Peng W. Chen J. Li W.-H. Network differentiation: a computational method of pathogenesis diagnosis in traditional Chinese medicine based on systems science Artif. Intell. Med. 118 2021 102134 126 Zhang M. Chen Y. Susilo W. PPO-CPQ: A privacy-preserving optimization of clinical pathway query for e-healthcare systems IEEE Internet Things J. 7 2020 10660 10672 127 Wu Z. Li R. Xie J. Zhou Z. Guo J. Xu X. A user sensitive subject protection approach for book search service J. Assoc. Inf. Sci. Technol. 71 2020 183 195 128 Wu Z. Shen S. Lian X. Su X. Chen E. A dummy-based user privacy protection approach for text information retrieval Knowl. Base Syst. 195 2020 105679 129 Wu Z. Shen S. Zhou H. Li H. Lu C. Zou D. An effective approach for the protection of user commodity viewing privacy in e-commerce website Knowl. Base Syst. 220 2021 106952 130 Wu Z. Li G. Shen S. Cui Z. Lian X. Xu G. Constructing dummy query sequences to protect location privacy and query privacy in location-based services World Wide Web 24 2021 25 49 131 Wu Z. Wang R. Li Q. Lian X. Xu G. A location privacy-preserving system based on query range cover-up for location-based services IEEE Trans. Veh. Technol. 2020 69 132 Zhang X. Wang T. Wang J. Tang G. Zhao L. Pyramid channel-based feature attention network for image dehazing Comput. Vis. Image Understand. 197 2020 103003 133 Qiu S. Hao Z. Wang Z. Liu L. Liu J. Zhao H. Fortino G. Sensor combination selection strategy for kayak cycle phase segmentation based on body sensor networks IEEE Internet Things J. 2021 10.1109/JIOT.2021.3102856 134 Qiu S. Zhao H. Jiang N. Wu D. Song G. Zhao H. Wang Z. Sensor network oriented human motion capture via wearable intelligent system Int. J. Intell. Syst. 37 2021 1646 1673 10.1002/int.22689

35016099

PMC9749108

NO-CC CODE

2022-12-15 23:22:50

no

Comput Biol Med. 2022 Mar 3; 142:105181

utf-8

Comput Biol Med

10.1016/j.compbiomed.2021.105181

oa_other

==== Front Cell Host Microbe Cell Host Microbe Cell Host & Microbe 1931-3128 1934-6069 Elsevier Inc. S1931-3128(22)00530-3 10.1016/j.chom.2022.10.017 Correction SARS-CoV-2 human T cell epitopes: Adaptive immune response against COVID-19 Grifoni Alba Sidney John Vita Randi Peters Bjoern Crotty Shane Weiskopf Daniela Sette Alessandro ∗ ∗ Corresponding author 14 12 2022 14 12 2022 14 12 2022 30 12 17881788 © 2022 Elsevier Inc. 2022 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcMain text (Cell Host & Microbe 29, 1076–1092; July 14, 2021) In the originally published version of this article, the supplemental table contained numbered references, while the main article contained Harvard-style references. An updated version of the table with Harvard-style references is now available online. The authors and production staff apologize for any confusion.

36521443

PMC9749374

NO-CC CODE

2022-12-15 23:22:53

no

Cell Host Microbe. 2022 Dec 14; 30(12):1788

utf-8

Cell Host Microbe

10.1016/j.chom.2022.10.017

oa_other

==== Front Psychiatry Res Case Rep Psychiatry Res Case Rep Psychiatry Research Case Reports 2773-0212 The Authors. Published by Elsevier B.V. S2773-0212(22)00077-3 10.1016/j.psycr.2022.100085 100085 Article A new form of checking obsessive-compulsive disorder in physicians: Another consequence of the COVID-19 pandemic. A case series. Hurtado María M. 1⁎ Macías María 2 Morales-Asencio José Miguel 3 1 Mental Health Unit, Regional University Hospital, Málaga, Spain, Instituto de Investigación Biomédica de Málaga (IBIMA) 2 Mental Health Unit, Regional University Hospital, Málaga, Spain 3 Universidad de Málaga, Faculty of Health Sciences, Málaga, Spain, Instituto de Investigación Biomédica de Málaga (IBIMA) ⁎ Corresponding Author: María M. Hurtado, Regional University Hospital, Plaza del Hospital Civil s/n, Málaga 29009, Spain. +34 951952833 14 12 2022 14 12 2022 100085© 2022 The Authors. Published by Elsevier B.V. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The current article provides information that facilitates early identification of a new form of checking obsessive-compulsive disorder detected in physicians during the COVID-19 pandemic. This article describes three cases of professional checking OCD in physicians. Physicians with checking OCD are obsessively concerned about making a mistake that will result in fatal consequences. The most frequent strategies of neutralization include avoiding direct contact with patients by taking sick or vacation leaves; compulsively studying for many hours daily (neglecting other aspects of life); checking the status of their patients by arranging additional follow-up consultations or making phone calls even out of working hours; repeatedly checking the medical history of their patients, and persistently recalling the last appointment. Physicians with check OCD often seek reassurance from their colleagues and consult the scientific literature for information about issues they used to be competent in. These patients may also experience anticipatory anxiety and mental blocks. However, egodystony is milder than in other forms of OCD. The COVID pandemic may have exacerbated these neutralization behaviors, since it has forced physicians to adapt to a new work environment. The recommended treatments (Exposure with Response Prevention therapy or/and SSRI) provide beneficial effects in a short time. KEYWORDS Obsessive-compulsive disorder Mental health of physicians ==== Body pmcIntroduction The demand for mental health care from health professionals has increased, because of the COVID-19 pandemic (Charney et al., 2020; Preti et al., 2020; Zhang et al., 2020). The symptoms most frequently detected in healthcare personnel during the pandemic have been anxiety, depression, and insomnia (Braquehais et al., 2020; Luo et al., 2020; Trumello et al., 2020; Zhang et al., 2020). However, some studies have also reported that physicians presented a higher prevalence of severe OC symptoms than the general population in this period (Ferreira et al., 2021). Contamination, order, repetition or checking are themes very frequent in OCD. Perhaps the best-known topics are contamination and cleanliness. In these cases, patients fear infecting themselves or infecting other people with a germ and to avoid this, they prevent touching the places they consider contaminated and if they do it, they wash their hands compulsively, even with abrasive products. The issue of ordering things is also frequent, by which the patient must order the utensils in a determined way, usually following a symmetrical order, to avoid misfortunes, for example, that a child has an accident and could die. In these cases, the person is aware that there is no logical connection between these two events. In the repetition theme, the person is driven to perform some actions a certain number of times to avoid some misfortune or bad luck, for example, turning the light off and on a certain number of times, or going through the doors forwards and backwards a certain number of times before definitively entering a room, or even touching some objects or furniture a certain number of times. In the case of the checking theme, the person repeatedly checks some things, frequently at home, to prevent a misfortune from happening, for example, repeatedly checks that the door is closed when leaving the house to prevent theft, that the electrical appliances are unplugged to prevent fires, or that the taps are properly closed, for a flood. There are some previous descriptions of obsessions in medical undergrade students in the literature (Al-Shatanawi et al., 2021; Torres et al., 2016), even before the COVID-19 pandemic. Specifically, it has been suggested that OCD symptoms are more frequent in medical students than in the general population, especially during the first year of university studies (Torres et al., 2016), and that this may have increased with the COVID-19 pandemic (Al-Shatanawi et al., 2021; Ferreira et al., 2021). However, we are not aware of diagnosed cases in experienced doctors in which the content of the obsessions is focused on their daily work, and the involved neutralization behaviours. In our unit, we have witnessed an increase in the incidence of a new form of checking obsessive-compulsive disorder (OCD) in physicians, which was hitherto unreported in the scientific literature. We detected a worrying delay in diagnosis, which deprived patients of early treatment. Fortunately, therapies yielded very positive outcomes. This case series study may provide guidance to mental health and primary care professionals to facilitate early diagnosis and treatment. The three cases have been described following the CARE checklist. Case A A 55-year-old woman specialist in Urology. The patient was referred to the outpatient Community Mental Health Unit from the occupational health services of her hospital and had no previous history of mental disorders. Some months before, the patient had been through a very stressful experience where one of her sons was hospitalized in a far-abroad country. She could not help or stay with him due to the pandemic. Although the event solved satisfactorily, she started thinking about how a little mistake can result in fatal consequences or aggressions. Additionally, some months before, she had decided to move to another hospital. She was increasingly concerned about making a mistake with respect to diagnosis or treatment of serious illnesses in her specialty (not relate with COVID-19), which made her anxiety grow. After a sick leave, she consulted the occupational health services of her hospital and requested to be reassigned to an alternative position to relieve her anxiety. The patient was referred to our Mental Health Unit with a diagnosis of adaptive disorder. When she was evaluated in our unit, she had been on a sick leave for several months to avoid the stressful situation of interacting with patients. As she did not have to deal with patients, she was calm at that moment, since her obsession was centered on the possibility of making a fatal mistake that caused the death to a patient or resulted in an aggression. Because she refused to attend patients, she had been advised to take a sick leave. She was prescribed vortioxetine 50 mg, which caused her headache and was suspended a month later. Then, she was prescribed mirtazapine 15 mg –which she did not start– and was referred to our unit. At two months, she was discharged by the medical inspector and started to experience anticipatory anxiety with associated tachycardia before every workday. She was tense and with symptoms of physiological hyperactivation when attending to her patients. The patient repeatedly reviewed the medical histories of the patients of the day, persistently recalled visits, phoned patients to continue or repeat clinical examination, and occasionally scheduled unnecessary appointments. On several occasions, she returned to the hospital out of her working hours to check some medical records. Additional checking is interpreted as compulsive behavior. After every workday, the patient felt exhausted because of the physiological hyperactivation she experienced at work. After a few weeks, she took her annual leave and requested a new sick leave to avoid the anxiogenic situation. At our center, she was prescribed sertraline 50mg and lorazepam 1mg, if needed. The patient was subsequently referred to a psychologist and was diagnosed with OCD. She received psychoeducation about OCD and learned about the maintaining role of avoidance behavior and compulsive checking. She was informed about the psychological approach recommended for this disorder i.e. Exposure and Response Prevention (ERP) Therapy. She was provided guidance about the behaviors that made fear persist, i.e., avoidance of the anxiogenic situation through sick and vacation leaves, persistent checking, calling patients on the phone, and arranging unnecessary visits, to name a few. The patient refused to receive any psychological and/or pharmacological treatment. Six months later, return to work emerged as the only feasible option, and she was recommended to resume sertraline and attend psychotherapy. A month after resuming treatment, she changed her mind and requested discharge against medical advice. She worked four days and took a new vacation leave, although she had agreed to expose to the anxiogenic situation. By this time, ERP therapy sessions focused on response prevention since, during the pandemic, physicians could remotely access the medical histories of their patients from outside the hospital. After a few weeks, she felt tense during visits, but the fear of making a fatal mistake decreased significantly. Now, five months after her return to work, compulsive checking has decreased and only occurs at working hours. Generally, the patient shows a normal work functioning. Case B A 36-year-old woman who works as a pediatrician at a primary care center. She was referred to our Mental Health Unit one a month and a half after the beginning of the lockdown because of COVID-19 pandemic. At that moment, the situation was dramatic, health professionals were not provided with adequate protection equipment and were not familiar with this type of situation. The patient did not have an individual or familial history of mental disorders. She received a diagnosis of generalized anxiety disorder by her family doctor, who prescribed lorazepam and recommended the patient to consult the mental health unit. She experienced mental blocks at work, something that she had never had before, after more than 10 years of professional practice. She attended patients in the morning (at that moment, telephonically) and worked as an on-call pediatrician in a hospital several days monthly. She worked some days at her primary care center, whereas on others she tele-worked from home. In the afternoons, she studied the COVID-19 protocols of different countries and reviewed the medical histories of her patients. She called parents out of her working hours to make an additional check or ask about the status of the child. She neglected other aspects of her life. Most of the time, she studied COVID-19 protocols until she felt asleep. Although her concert began focused on COVID-19, it later became general to any potential serious illnesses in children. On her first visit to our Mental Health Unit, the psychiatrist deprescribed benzodiazepine and replaced it with venlafaxine 75R, and mirtazapine 15 mg mirtazapine to treat insomnia and poor appetite. In parallel, the patient started to see a clinical psychologist, whose sessions were telephonic during the first months. The patient was recommended to keep working at the primary care center but suspend on-call shifts temporarily. During sessions with the psychologist, the patient received psychoeducation about COD and learned about the maintaining role of avoidance, compulsive checking and reassuring behaviors. This way, the psychologist helped her identify the behaviors (functional analysis) that were contributing to maintaining her disorder. She was encouraged to engage in an ERP Therapy. One month after the interventions had started, the patient experienced a moderate improvement. She stopped reviewing the medical histories of her patients out of working hours, and limited to two hours the time she devoted to studying COVID-19 protocols of other countries. Anxiety decreased notably and mental blocks at work or out of home disappeared. Symptoms improved even though the situation of uncertainty at work persisted. Two months later, the patient experienced a significant improvement, which was maintained after the dose of mirtazapine was reduced. Three months later, the patient could work on-call shifts successfully. A month later, mirtazapine was suspended and the patient remained asymptomatic despite the successive waves of the pandemic. Venlafaxine was suspended and the patient has remained asymptomatic since then. According to the CARE checklist, the next table shows the patient`s perspective on the received treatment. Table 1 Table 1 Perspective of B patient on the received treatment. Table 1• In March 2020 the pandemic exploded... and my mind with it. After several weeks of not being able to sleep, continuous changes in protocols, absence of protective material, fear of infecting myself, fear of infecting my loved ones... the situation overwhelmed me and I decided to ask for help. • I had my first telephone consultation with the Mental Health Unit before 24 hours. Initially they focused the situation on an anxious-depressive problem, but after three interviews and with the implicit difficulty to making a diagnose by phone, without a face-to-face assessment, they oriented my diagnosis to OCD. They associated venlafaxine with my treatment, which, together with psychotherapy with a clinical psychologist and a psychiatrist, helped me getting out of that situation. After two months of telephone sessions, we were finally able to make the first face-to-face visit. The treatment was always wonderful, but much more effective when it was provided in a face-to-face approach. • After 15 months of treatment, we agreed to begin the medication withdrawal. What was supposed to take a few weeks was extended to almost 3 months due to a SSRI discontinuation syndrome. • The diagnosis and treatment were perfect and I currently lead a completely normal life, both personally and at work. Case C A 48-year-old man who works as an out-of-hospital emergency physician. Some months before his referral to the mental health unit, the patient had decided to change to another service. Although his work performance and adaptation was good, it was a very demanding position, and the patient decided to move to an ambulatory care center. With the outbreak of the pandemic, the patient started to experience high levels of anxiety. No familial or personal medical history of interest. The patient has supportive family and friends. During the first visit to our unit, the patient reported high levels of anxiety, continuous ruminative thinking; and repeatedly checking medical records at work and at home (where he had remote access) with respect to several severe illnesses (unrelated with COVID-19). He devoted more a more time to studying and consulted his colleagues on issues he was competent in but which he now felt uncertain about. Additionally, he had developed hypochondriac thoughts about the possibility that a relative had a severe disease. He was in a low mood, with mild anhedonia. However, when he was not at work, he could get away and moderately enjoy the activities he used to like in the past. The patient experienced a mild weight loss and did not show sleep problems. Following the initial examination by the psychiatrist, he was recommended to increase the dose of the antidepressant he had been taking in the last month (duloxetine, from 90 to 120mg) and was referred to the clinical psychology service to start ERP therapy. He was discouraged from taking a sick leave because of the risk that avoidance behaviors would be perpetuated. During clinical psychology sessions, the patient received psychoeducation that helped him understand how obsessive thoughts work and the role that repeated checking plays. He received guidance about how to identify compulsive behaviors such as avoidance and escape behavior, which contributed to maintaining his disorder. In a few weeks, the patient experienced a significant improvement; obsessive thoughts, compulsive checking and anxiety disappeared and the patient had a normal mood. Six months after the improvement of symptoms, the dose of duloxetine started to be progressively reduced. At present, the patient is receiving a dose of 30mg and remains asymptomatic. Presumably, the drug will be suspended during the next visit. Discussion In most cases of checking OCD, there is a trigger factor, such as the disease of a loved one, a change of position at work, or the pandemic itself. These situations were stressful for the patients, and generated feelings of uncertainty and fear. However, in none of these three cases sere there COVID-19 infections of their own or from family members that precipitated the disorder, nor was there a significant change in the workload since none of them was in charge of the first line care of COVID-19 patients. According to the trigger factors and previous history of each subject, once the OCD process has started, some situations become anxiogenic. Such is the case of dealing with patients and assuming responsibility for their diagnosis and treatment, especially in the context of significant uncertainty generated by the COVID-19 pandemic. These situations cause anxiety because they are associated with the patient's obsessions and activate them involuntarily and automatically. In the cases reported here, the patients experienced obsessive thoughts about whether they could have missed some important detail and such mistakes could be life-threatening for their patients. Obsessions are characterized by persistent thoughts that subjects recognize as a product of their mind, but do not understand that are the product of their reasoning. Based on our observations, egodystonia may be more moderate, as compared to other forms of OCD. Thought is ego-dystonic when the person feels that he or she is incoherent or is not in tune with his or her own person, and way of being. As a consequence, they sometimes think, that these self-perceived thoughts are illogical. This characteristic makes that the thought is experienced with a certain strangeness, and differently from other thoughts. Otherwise said, the patient may consider these thoughts as illogical or may not have that feeling at all. Then, the subject evaluates his/her obsessions. Here, these thoughts are the product of their own reasoning, such as in: "If I've made a mistake, the patient can die and that would be terrible," "if the patient gets worse, he will attack me...", "Is it safe for my patients that I'm the person responsible for their treatment if I feel so insecure?." The severity of this type of reflection about their obsessions is determined by how concerning or inacceptable this reflection is for the subject. These reflections generate substantial emotional pain and, frequently, guilt. Since this feeling is extremely stressful, subjects develop strategies of neutralization that include avoiding the anxiogenic situations (i.e. they avoid seeing patients by taking sick/vacation leaves, among other strategies); adopting compulsive behaviors or rituals that include compulsive checking (they repeatedly review the medical histories of their patients, repeatedly review medical protocols...); and seeking reassurance (they consult their colleagues on clinical cases they perfectly know how to manage; calling their patients unnecessarily; and by arranging unnecessary appointments, among other strategies). These strategies of neutralization are aimed at seeking refuge from their fears, which make angst drop away and provide temporary relief. Because of this feeling of relief, patients believe that these strategies are effective and beneficial for them and use them whenever the process starts again, until these behaviors occur automatically. The limitation in this process lies in that these strategies are effective in the short term, but make fears persist in the long term. The repeated use of these strategies leads the subject to think that excessive cautions are effective in preventing the dreaded event (i.e. causing the death of a patient), which makes the fear increase in the long term. The OCD process itself prevents the patient from realizing that standard protocols alone safeguard the safety of patients, and that additional checking is not necessary. Table 2 Table 2 Main components of OCD treatment used in this case series Table 2Approach Components What does it consist of? ERP Therapy (In case B, a telephone-based approach was used because a generalised COVID 19 lockdown had been imposed) Assessment: functional analysis During the assessment it is essential to identify the types of neutralisation behaviour presented, in order to intervene appropriately. Psychoeducation on the functioning of OCD and explanation of the therapy ERP is strongly counterintuitive, and therefore it is very important to convey a good understanding of how OCD works and of the logic underlying the tasks included in the therapy. Construction of the exposure hierarchy All the anxiogenic situations identified are listed and the coping difficulty determined is graded from 1 to 10. ERP ERP begins with the easiest items. Exposure means that the situation generating anxiety is not avoided or escaped from (in this case, the physicians are not on sick leave, and continue with their daily clinical activity). Response prevention refers to the fact that the person does not perform neutralising behaviour in response to the outcome of the functional analysis (hence there are no additional checks of medical records, no further reassurance calls to patients, no more questions addressed to colleagues, no additional (unnecessary) appointments, no additions to an already busy study schedule, and no mentally repetitive reconsideration of the medical consultation). In other words, no steps are taken to reduce the anxiety currently being experienced. For this purpose, records of ERP sessions are normally used. ERP monitoring The ERP and its effects on the frequency and intensity of the obsessions are reviewed, with special attention to the duration of the sessions, and to whether particular types of neutralisation behaviour have or have not been replaced by others. Pharmacotherapy SSRI, NSRI Sertraline (SSRI), Venlafaxine, Duloxetine (NSRI) are compatible with and enhance the effect of ERP therapy. Clomipramine Clomipramine should be considered when there is no response to SSRI or NSRI, although in the reported cases this was not necessary. Do not use anxiolytic drugs during ERP Exposure techniques require attentional commitment and are less effective if performed under the influence of anxiolytics or alcohol. In this sense, the COVID-19 pandemic has forced healthcare professionals to adapt to a new environment, which makes them more prone to developing this type of OCD. For example, the digitalization and integration of medical histories make it possible for health professionals to repeatedly review histories remotely. The implantation of telephonic consultations due to the COVID pandemic may have also played a role in the development of this process since it makes it possible for physicians to call their patients anytime to make a check and/or seek reassurance, which was not as frequent in the past. During the pandemic, consultations with colleagues also became more frequent among physicians (possible reassurance seeking). Otherwise said, the flexibilization of the working conditions of physicians due to the COVID-19 pandemic may have favored the development of these disorders. Conclusions In the years before the pandemic, only a case of checking OCD was recorded in our unit. A family doctor who lived near the hospital and went daily to emergency care in her free time to review the medical histories of her patients, until she started to experience mental blocks every morning before going to work. The changes to healthcare practice brought about by COVID-19 have resulted in three cases of OCD among physicians in less than a year, which represents a significant increase. Table 3 Table 3 Strategies of neutralization of the obsession to make a mistake with fatal consequences (not necessarily related with COVID 19) in medical checking OCD Table 3Strategies of neutralization Behaviours Avoiding Sick or vacation leaves. Request a second activity that does not involve direct contact with patients. Compulsive checking Repeatedly review the medical records, also during rest hours. Repeatedly review medical protocols. Mentally review what happened during the visit over and over again. Reassurance Phone their patients unnecessarily, also during rest hours. Arrange unnecessary appointments. Consult their colleagues on clinical cases they perfectly know how to manage. The treatment of choice for OCD is ERP Therapy, in its different modalities according to each case. When the first-line treatment fails, SSRIs are recommended (National Institute for Health and Care Excellence, 2005, 2019). In the three cases described here, the two approaches were used. However, none of the three cases reviewed in this paper (neither the previous case reported before the pandemic) were diagnosed early but were identified as more unspecific disorders such as GAD and adaptive disorder, to name a few. Failure to determine early diagnosis hinders early treatment and management. When a correct diagnosis of OCD is established and the patient receives an appropriate therapy, these patients improve notably, and their working functionality returns to normality. Declaration of interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References Al-Shatanawi T.N. Sakka S.A. Kheirallah K.A. Al-Mistarehi A.H. Al-Tamimi S. Alrabadi N. Alsulaiman J. Al Khader A. Abdallah F. Tawalbeh L.I. Saleh T. Hijazi W. Alnsour A.R. Younes N.A. Self-Reported Obsession Toward COVID-19 Preventive Measures Among Undergraduate Medical Students During the Early Phase of Pandemic in Jordan Frontiers in Public Health 9 2021 10.3389/FPUBH.2021.719668 Braquehais M.D. Vargas-Cáceres S. Gómez-Durán E. Nieva G. Valero S. Casas M. Bruguera E. The impact of the COVID-19 pandemic on the mental health of healthcare professionals QJM : Monthly Journal of the Association of Physicians 113 9 2020 613 617 10.1093/QJMED/HCAA207 Charney A.W. Katz C. Southwick S.M. Charney D.S. A Call to Protect the Health Care Workers Fighting COVID-19 in the United States The American Journal of Psychiatry 177 10 2020 900 901 10.1176/APPI.AJP.2020.20040535 32731814 Ferreira S. Sousa M.M. Moreira P.S. Sousa N. Picó-Pérez M. Morgado P. A Wake-up Call for Burnout in Portuguese Physicians During the COVID-19 Outbreak: National Survey Study JMIR Public Health and Surveillance 7 6 2021 10.2196/24312 Luo M. Guo L. Yu M. Wang H. The psychological and mental impact of coronavirus disease 2019 (COVID-19) on medical staff and general public - A systematic review and meta-analysis Psychiatry Research 291 2020 10.1016/J.PSYCHRES.2020.113190 National Institute for Health and Care Excellence Obsessive-compulsive disorder and body dysmorphic disorder: treatment (NICE guidelines CG31) 2005 NICE National Institute for Health and Care Excellence 2019 surveillance of obsessive-compulsive disorder and body dysmorphic disorder: treatment (NICE guideline CG31) 2019 National Institute for Health and Care Excellence (UK) https://www.ncbi.nlm.nih.gov/books/NBK551808/ Preti E. Di Mattei V. Perego G. Ferrari F. Mazzetti M. Taranto P. Di Pierro R. Madeddu F. Calati R. The Psychological Impact of Epidemic and Pandemic Outbreaks on Healthcare Workers: Rapid Review of the Evidence Current Psychiatry Reports 22 8 2020 10.1007/S11920-020-01166-Z Torres A.R. Cruz B.L. Vicentini H.C. Lima M.C.P. Ramos-Cerqueira A.T.A. Obsessive-Compulsive Symptoms in Medical Students: Prevalence, Severity, and Correlates Academic Psychiatry: The Journal of the American Association of Directors of Psychiatric Residency Training and the Association for Academic Psychiatry 40 1 2016 46 54 10.1007/S40596-015-0357-2 26108391 Trumello C. Bramanti S.M. Ballarotto G. Candelori C. Cerniglia L. Cimino S. Crudele M. Lombardi L. Pignataro S. Viceconti M.L. Babore A. Psychological Adjustment of Healthcare Workers in Italy during the COVID-19 Pandemic: Differences in Stress, Anxiety, Depression, Burnout, Secondary Trauma, and Compassion Satisfaction between Frontline and Non-Frontline Professionals International Journal of Environmental Research and Public Health 17 22 2020 1 13 10.3390/IJERPH17228358 Zhang W.R. Wang K. Yin L. Zhao W.F. Xue Q. Peng M. Min B.Q. Tian Q. Leng H.X. Du J.L. Chang H. Yang Y. Li W. Shangguan F.F. Yan T.Y. Dong H.Q. Han Y. Wang Y.P. Cosci F. Wang H.X. Mental Health and Psychosocial Problems of Medical Health Workers during the COVID-19 Epidemic in China Psychotherapy and Psychosomatics 89 4 2020 242 250 10.1159/000507639 32272480

0

PMC9749375

NO-CC CODE

2022-12-15 23:23:20

no

Psychiatry Res Case Rep. 2022 Dec 14;:100085

utf-8

Psychiatry Res Case Rep

10.1016/j.psycr.2022.100085

oa_other

==== Front Heliyon Heliyon Heliyon 2405-8440 The Author(s). Published by Elsevier Ltd. S2405-8440(22)03559-9 10.1016/j.heliyon.2022.e12271 e12271 Research Article Social factors influence on anxiety, depression level and psychological trauma of obstetrics and gynecology residents during COVID-19 pandemic Harzif Achmad Kemal a∗ Lukman Donny Damiar Santoso a Maidarti Mila a Charilda Fistyanisa Elya a Andyra Azizah Fitriayu a Raharjanti Natalia Widiasih b Levania Monika Kristi b Nora Hilwah c Yeni Cut Meurah c Sauqi Hardyan d Armanza Ferry d Sumapraja Kanadi a a Department of Obstetrics and Gynecology, Faculty of Medicine University of Indonesia, Dr. Cipto Mangunkusumo Hospital, Jakarta, Indonesia b Department of Psychiatry, Faculty of Medicine University of Indonesia, Dr. Cipto Mangunkusumo Hospital, Jakarta, Indonesia c Department of Obstetrics and Gynecology, Faculty of Medicine University of Syiah Kuala, Aceh, Indonesia d Department of Obstetrics and Gynecology, Faculty of Medicine University of Lambung Mangkurat, Kalimantan, Indonesia ∗ Corresponding author. 14 12 2022 12 2022 14 12 2022 8 12 e12271e12271 13 4 2022 17 10 2022 2 12 2022 © 2022 The Author(s) 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Background and aim The novel coronavirus disease 2019 (COVID-19) has enforced obstetrics and gynecology (ObGyn) residency training process to undergo wide changes including lessons modifications, yet their mental health were not evaluated. Hence, this study aimed to evaluate the influence of social factors on anxiety, depression level, and psychological trauma of ObGyn residents during the COVID-19 pandemic as well as the impact of COVID-19 to residency training program. Methods A cross-sectional study was conducted in three institutions in Indonesia: the University of Indonesia, the University of Lambung Mangkurat, and the University of Syiah Kuala. A total of 169 ObGyn residents agreed to participate and were enrolled in this study. Results Total 169 residents were eligible with a mean age of 26–42 years. 76.9% of the residents were exposed to COVID-19 patients during clinical rotation. Approximately half of them (52.6%) thought COVID-19 has brought negative effects. Long-distance learning was considered of good quality by 40.2% of participants. The majority experienced enough resting periods, nearly half of them (45.5%) were concerned about the impact of being a less competent specialist. Conclusion Overall, no significant statistical relationship were revealed between social factors and depression, anxiety and psychological trauma in ObGyn residents during COVID-19 pandemic. COVID-19; Obstetrics and gynecology; Residency; Mental health; Psychology trauma. Keywords COVID-19 Obstetrics and gynecology Residency Mental health Psychology trauma ==== Body pmc1 Introduction The novel coronavirus disease (COVID-19) has dramatically changed the live of healthcare workers. The increasing cases has added a new vulnerable population, healthcare workers who treat COVID-19 patients themselves. The vulnerabilities of healthcare workers at the time of crisis first reported after the death of Dr. Li Wenliang, a doctor in Wuhan who was trying to warn his colleagues about the possibility of a new infectious virus while treating his patients [1]. Moreover, shortage of personal protective equipment (PPE) supplies has augmented the obstacles faced in the beginning of pandemic situation. At the beginning of 2020, Indonesian government directed the health worker to use raincoats as PPE due to the lack of supplies. Medical residents are considered essential within the healthcare system. Since the pandemic has emerged, a proportion of resident physicians were quickly reorganized and redeployed to the frontline. The training process has undergone widely changes, particularly due to the preventive implementation from the government through health advocacy campaigns, lockdown, and restricting public meetings which have forced universities and institutions to modify lessons, training, and work shifts [2, 3]. The current pandemic situation has imposed a significant adjustment especially in the field of surgery [4]. Obstetrics and gynecology (ObGyn) residency training has to adapt well and quickly into the lockdown period. This quarantine period has led to limiting the presence of residents who are on duty, postponing elective procedures, and cancelling lectures and educational conferences. Besides physical health, latest literature showed that healthcare workers are at a high risk of experiencing mental health problems such as stress, anxiety, depression and sleep disorders with considerable degree due to COVID-19 pandemic.[5]. In the general population, COVID-19 pandemic can induce psychiatric symptoms or make the existing COVID-19 symptoms even worse. Worrying about the possibility of getting infected can lead to fear of death, anxiety, a sense of helplessness, and a tendency to blame the sick person. Various mental illnesses can occur in this condition include depression, anxiety, panic attacks, somatic symptoms, posttraumatic stress disorder (PTSD), delirium, psychosis, and suicidal thoughts [6, 7, 8, 9]. Studies have reported that health workers, especially those working in the emergency department, intensive care unit, and infectious disease wards have a higher risk of experiencing adverse psychological effects [10]. However, a contradictive finding stated that they did not find significant differences in stress level between frontline health workers and other groups [5]. As this pandemic continues, residents' well-being, environment of clinical practice and medical education are yet to be evaluated. Therefore, the present study aimed to evaluate the social factors influence on the level of anxiety, depression and psychological trauma of ObGyn residents during the current pandemic as these residents were exposed to COVID-19 patients and are prone to COVID-19 infection due to their placement during clinical rotation. This study also aimed to identify possible differences between various main institutions for ObGyn residencies. This manuscript was reported in line SQUIRE 2.0 criteria [11]. 2 Materials and methods The study protocol was approved by the Ethics Committee of Faculty of Medicine University of Indonesia, number KET-1161/UN2.F1/ETIK/PPM.00.02/2020, and it was developed to conform with the principles of the Helsinki Declaration. The participating residents were informed about the objectives of the study and then invited to participate. Those who agreed signed the informed consent form. This study took place in three different universities in Indonesia that provide ObGyn residency program. The chosen universities are University of Indonesia, University of Lambung Mangkurat, and University of Syiah Kuala. This descriptive cross-sectional survey was conducted during COVID-19 pandemic from September 2020 to March 2021. Participants were from ObGyn resident physicians who were under a five-year standardized residency training in Indonesia, were registered in the ObGyn department in each university, and still actively working during the pandemic situation. Residents were informed that their participation was completely voluntary and all the answers were guaranteed to be confidential. If the interviewee agreed to participate, the questionnaire was forwarded with consecutive-sampling method and shared through virtual learning environments and social networks (e.g. WhatsApp). The questionnaire consisted of four main parts: (1) basic information (age, marital status and children, living status, educational institution, clinical residency years and rotation) and how COVID-19 impacted the residents' residency program (learning process, moral condition, health condition, financial problem), (2) symptoms of anxiety during the pandemic – scored using General Anxiety Disorder-7 (GAD-7), (3) symptoms of depression related to the pandemic – scored using Patient Health Questionnaire (PHQ-9), (4) symptoms of psychological trauma due to the pandemic – scored using PTSD Checklist-Civilian (PCL-C-17). To ensure data integrity, if any missing data occurred in the e-questionnaire, answers wouldn’t be submitted. GAD-7 was used for symptoms of anxiety during the pandemic and was containing 7 questions and calculated by assigning score of 0, 1, 2, and 3. The total range will score between 0 and 21. Cut off points of 0–5, 6–10 and 11–15 were interpreted as mild, moderate and severe anxiety level. PHQ-9 was used for symptoms of depression related to the pandemic and was containing 9 questions and calculated by assigning score of 0, 1, 2, and 3. The total range will score between 0 and 27. Cut off points of 0–4, 5–9, 10–14, 15–19, 20–27 were interpreted as minimal, mild, moderate, moderately severe, and severe depression. PCL-C-17 used for symptoms of psychological trauma due to the pandemic and was containing 17 items that correspond to the key symptoms of PTSD and calculated by assigning score of 1, 2, 3, 4, and 5. The total range will score between 17 and 85. Cut off points of 17–29, 30–44, 45–85 were interpreted as little to no severity, moderate severity and high severity. The samples were accessed through registration of each institution involved. Residents were offered to participate in the study by authors who represents each institution. A total of 169 ObGyn residents agreed to participate and were enrolled in this study. All of the questionnaires were filled therefore there wasn’t any loss and all of the residents included made up the final sample. To assess general characteristics, learning quality, moral, and health condition of the participants, we used the above mentioned questionnaires (Indonesian version) that had been tested for validity and reliability. Pearson-correlation or Spearman test was used to measure the validity while Cronbach Alpha test was used to assess the reliability. Cronbach alpha test results for GAD-7, PHQ-9 and PCL-C-17 are 0.867, 0.885, 0.946. Learning quality was defined as learning effectiveness with adaptation during COVID-19 pandemic. Participants' perception of motivation and mental resilience during COVID-19 pandemic were observed as the moral section in the questionnaire. Health condition is individual perception about physical and mental health in residency training times during pandemic situation. Data were stored and organized on an electronic spreadsheet. Statistical analyses were performed using Statistical Package for Social Sciences software, version 20.0. Afterwards, data were presented in the form of tables and narratives. Descriptive statistics were used to summarize the data. The relationship between social factors and depression, anxiety, and physiological trauma was tested with appropriate statistical tests. The alpha level was set at p < 0.05. Analysis was done using Kruskal-Wallis to assess participants' age and Fisher exact test to assess the other variables. 3 Results There were 169 subjects who were eligible to participate in this study. The mean age was 31 years (range 26–42 years). As shown in Table 1 , the majority were in the 4th and 5th year of residencies, followed by the 2nd year of residency. More than half participants were married 98/169 (57.9%) while one-third of them were living alone 62/169 (36.6%). Majority of the residents 130/169 (76.9%) were exposed to COVID-19 patients during clinical rotation when giving the optimal patient care.Table 1 Participant characteristics. Table 1 N (%)/Mean (SD) Age 31 yrs (26–42) Year of residency training program  First year 26 (15.3)  Second year 33 (19.5)  Third year 28 (16.5)  Fourth year 41 (24.2)  Fifth year 41 (24.2) Educational institution  University of Indonesia 115 (68.1)  University of Lambung Mangkurat 20 (11.8)  University of Syiah Kuala 34 (20.1) Marital status  Married 98 (57.9)  Single 71 (42.0) Living status  Alone 62 (36.6)  With spouse 59 (34.9)  With parents 48 (28.4) Clinical rotation exposed with COVID-19  Exposed with COVID-19 130 (76.9)  Unexposed to COVID-19 39 (23.0) Level of anxiety  Minimal anxiety 144 (85.2)  Mild anxiety 22 (13.0)  Moderate anxiety 2 (1.18)  Severe anxiety 1 (0.59) Level of depression  Not depressed 125 (73.9)  Mild depression 32 (18.9)  Moderate depression 10 (5.91)  Moderate-severe depression 2 (1.18)  Severe depression 0 (0) Psychological trauma  Mild symptoms 135 (79.8)  Moderate symptoms 28 (16.5)  Severe symptoms 6 (3.55) Table 2 reports the impact of COVID-19 to residency training program. Nearly half of the residents 89 (52.6%) thought COVID-19 has brought negative effects on the participants, hence, they prefer to work inside the department. Half of the residents' moral perceptions were unchanged (46.1%) during the pandemic situation. Long distance learning was considered a good quality by 68/169 (40.2%) participants. Participants experienced an enough resting period without fatigue. However, nearly half of them 61/169 (45.5%) concerned about the impact on being a less competent ObGyn specialist.Table 2 Impact of COVID-19 on Residency Education Program. Table 2 N (%) Opinions about COVID-19 effect  Positive 48 (28.4)  Neutral 32 (18.9)  Negative 89 (52.6) Adaptation process during COVID-19 pandemic  Very easy 5 (2.95)  Easy 40 (23.6)  Moderate 99 (58.5)  Difficult 24 (14.2)  Very difficult 1 (0.59) Preference of working place  Working inside the department 126 (74.5)  Working outside the department 21 (12.4)  Not working inside the department 20 (11.8)  Not working outside the department 2 (1.18) Moral perception  Very increasing 1 (0.59)  Increasing 23 (13.6)  Unchanged 78 (46.1)  Decreasing 66 (39.0)  Very decreasing 1 (0.59) Individual moral  Very increasing 1 (0.59)  Increasing 23 (13.6)  Unchanged 92 (54.4)  Decreasing 53 (31.3)  Very decreasing 0 (0) Mentoring quality  Excellent 10 (5.91)  Good 64 (37.8)  Average 65 (38.4)  Poor 27 (15.9)  Bad 3 (1.77) Mentoring quantity  Excellent 8 (4.73)  Good 64 (37.8)  Average 68 (40.2)  Poor 24 (14.2)  Bad 5 (2.95) Long distance learning effectiveness  Excellent 20 (11.8)  Good 52 (30.7)  Average 68 (40.2)  Poor 27 (15.9)  Bad 2 (1.18) Individual work effectiveness  Strongly agree 7 (4.14)  Agree 117 (69.2)  Disagree 43 (25.4)  Strongly disagree 2 (1.18) Colleagues work effectiveness  Strongly agree 7 (4.14)  Agree 113 (66.8)  Disagree 48 (28.4)  Strongly disagree 1 (0.59) Health condition  Strongly agree 17 (10.0)  Agree 125 (73.9)  Disagree 25 (14.7)  Strongly disagree 2 (1.18) Overworked condition  Strongly agree 6 (3.55)  Agree 66 (39.0)  Disagree 86 (50.8)  Strongly disagree 11 (6.50) Enough rest period  Strongly agree 11 (6.50)  Agree 120 (71.0)  Disagree 36 (21.3)  Strongly disagree 2 (1.18) In need of psychological help Strongly agree 6 (3.55)  Agree 34 (20.1)  Disagree 73 (43.1)  Strongly disagree 56 (33.1) Financial problems  Strongly agree 21 (12.4)  Agree 67 (39.6)  Disagree 60 (35.5)  Strongly disagree 21 (12.4) Less-competent to be an ObGyn specialist  Strongly agree 12 (7.10)  Agree 77 (45.5)  Disagree 61 (36.0)  Strongly disagree 19 (11.2) Overall no significant statistical relationship were revealed between social factors (such as age, levels of residency, marital status, living status, and COVID-19 exposure during clinical pandemic (Tables 3, 4, and 5 ). They were most likely to have minimal anxiety, not in a depressed condition, and revealed a psychological trauma with mild symptoms. Less than 5% of all participants experienced moderate to severe anxiety, moderate to severe level of depression, and psychological trauma with severe symptoms.Table 3 Relationship between social factors and anxiety levels. Table 3Social Factors Minimal anxiety Mild anxiety Moderate anxiety Severe anxiety P-value Age 31 (26–40) 30 (27–42) 28.5 (28–29) 0.064 Year of residency training program 0.670  First year 20 (76.9) 5 (19.2) 1 (3.8) 0 (0)  Second year 28 (84.8) 4 (12.1) 1 (3) 0 (0)  Third year 24 (85.7) 3 (10.7) 0 (0) 1 (3.5)  Fourth year 35 (85.3) 6 (14.6) 0 (0) 0 (0)  Fifth year 37 (90.2) 4 (9.7) 0 (0) 0 (0) Educational institution 0.085  University of Indonesia 99 (86) 14 (12.1) 2 (1.7) 0 (0)  University of Lambung Mangkurat 14 (70) 6 (30) 0 (0) 0 (0)  University of Syiah Kuala 31 (91.1) 2 (5.8) 0 (0) 1 (2.9) Marital status 0.780  Married 62 (87.3) 9 (12.6) 0 (0) 0 (0)  Single 82 (83.6) 13 (13.2) 2 (2) 1 (1) Living status 0.592  Alone 52 (83.8) 9 (14.5) 1 (1,6) 0 (0)  With spouse 50 (84.7) 9 (15.2) 0 (0) 0 (0)  With parents 42 (87.5) 4 (8.3) 1 (2) 1 (2) Clinical rotation exposed with COVID-19 0.514  Exposed with COVID-19 32 (82) 6 (15.3) 1 (2.5) 0 (0)  Unexposed to COVID-19 112 (86.1) 16 (12.3) 1 (0.7) 1 (0.7) Table 4 Relationship between social factors and depression levels. Table 4Social Factors Not depressed Mild depression Moderate depression Moderate-severe depression P-value Age 31 (26–42) 31 (26–40) 31 (28–36) 30 (29–31) 0.900 Year of residency training program 0.324  First year 21 (80.7) 2 (7.6) 3 (11.5) 0 (0)  Second year 26 (78.7) 6 (18.1) 0 (0) 1 (3)  Third year 21 (75) 4 (14.2) 3 (10.7) 0 (0)  Fourth year 29 (70.7) 9 (21.9) 3 (7.3) 0 (0)  Fifth year 28 (68.2) 11 (26.8) 1 (2.4) 1 (2.4) Educational institution 0.773  University of Indonesia 82 (71.3) 23 (20) 8 (6.9) 2 (1.7)  University of Lambung Mangkurat 14 (70) 5 (25) 1 (5) 0 (0)  University of Syiah Kuala 29 (85.2) 4 (11.7) 1 (2.9) 0 (0) Marital status 0.461  Single 53 (74.6) 12 (16.9) 4 (5.6) 2 (2.8)  Married 72 (73.4) 20 (20.4) 6 (6.1) 0 (0) Living status 0.837  Alone 44 (70.9) 12 (19.3) 5 (8) 1 (1.6)  With spouse 43 (72.8) 13 (22) 3 (5) 0 (0)  With parents 38 (79.1) 7 (14.5) 2 (4.1) 1 (2) Clinical rotation exposed with COVID-19 0.910  Exposed to COVID-19 31 (79.4) 6 (15.3) 2 (5.1) 0 (0)  Unexposed to COVID-19 94 (72.3) 26 (20) 8 (6.1) 2 (1.5) Table 5 Relationship between social factors and psychological trauma. Table 5Social Factors Mild symptoms Moderate symptoms Severe symptoms P-value Age 31 (26–42) 31 (26–36) 30 (29–32) 0.604 Year of residency training program 0.434  First year 22 (84.6) 4 (15.3) 0 (0)  Second year 28 (84.8) 4 (12.1) 1 (3)  Third year 25 (89.2) 3 (10.7) 0 (0)  Fourth year 30 (73.1) 7 (17) 4 (9.7)  Fifth year 30 (73.1) 10 (24.3) 1 (2.4) Educational institution 0.994  University of Indonesia 92 (80) 19 (16.5) 4 (3.4)  University of Lambung 15 (75) 4 (20) 1 (5) Mangkurat 28 (82.3) 5 (14.7) 1 (2.9)  University of Syiah Kuala Marital status 0.279  Married 58 (81.6) 9 (12.6) 4 (5.6)  Single 77 (78.5) 19 (19.3) 2 (2) Living status 0.896  Alone 48 (77.4) 11 (17.7) 3 (4.8)  With spouse 49 (83) 8 (13.5) 2 (3.3)  With parents 38 (79.1) 9 (18.7) 1 (2) Clinical rotation exposed with COVID-19 0.213  Exposed with COVID-19 35 (89.7) 4 (10.2) 0 (0)  Unexposed to COVID-19 100 (76.9) 24 (18.4) 6 (4.6) 4 Discussion The participants of this research were ObGyn residents from three different universities located in Indonesia. The average age of the participants was 31 (range 26–42) years with the majority of participants in the 4th and 5th year of standardized residency training. Balance distribution of total participants provides an advantage to obtain a good variation of results to describe the population. Despite an approximate 85% of the participants classified as minimal anxiety, 2% experienced moderate to severe anxiety. On the other hand, 7% had moderate to severe depression compared with 18.9% who experienced mild depression. The results in this study showed a lower value than previous studies, particularly in the outcome of anxiety and depression. Meta-analysis by Li et al. shows that the incidence of anxiety is 22% along with depression which accounted for 21% [15]. Another study conducted in Indonesia on health workers showed a high level of anxiety in 33% of all research subjects during pandemic [13]. These discrepancies might be caused by a different population of only ObGyn resident physicians included in our study. Previous research showed that 26% pregnant women did not perform antenatal care during the pandemic which supported the abovementioned factors [14]. Limited exposure to patients with symptoms of COVID-19, decreasing workload, and COVID-19 screening before taking care of patients can produce a secure feeling that reduces the occurrence of anxiety and depression. Those who suffered from psychological trauma, however, revealed a higher percentage (20%) compared to other psychopathologies. This result appeared to be similar with the previous study (21.5%) [15]. ObGyn residents felt that COVID-19 has brought a negative impact to the training program. The negative perception of ObGyn residents can be studied further with the perception of the adaptation, moral, training guidance, performance, health, and financial condition. Fifteen percent experienced learning adaptation difficulties during the pandemic. Adaptation can occur in both teaching and learning processes with long distance learning which has never been done before, as well as in providing patient care services during pandemic. Majorities have preferences in performing services inside the department due to the presence of more optimal guidance by the teachers. In addition, a tertiary hospital would affect the types of patients. In a tertiary hospital, patients surely have undergone meticulous screening and gradual referrals. In this study, it can be shown that the residents' moral declined because of the pandemic. This was also demonstrated by the presence of unchanged or increased moral in 60% participants. Eighty percent of total participants considered an average to excellent quality of training guidance. This good result was strongly influenced by the effectiveness of long-distance learning which was classified as moderate to very good in 85% subjects. The use of internet platforms and new teaching modalities greatly influences resident teaching during pandemic. All educational institutions suspended their face-to-face teaching, providing online lecture to guarantee students' teaching and their right to study do to the lockdown period [3, 16]. The use of several meeting platforms from the internet for the dissemination of teaching material and educational meetings, however, has built bridges, albeit virtual, between resident and teachers [15]. The development of new and more structured teaching modalities is crucial to increase the effectiveness of teaching, compared to direct teaching. However, the drawback of using internet platforms during this pandemic is the lack of clinical practice, especially in the field of surgery, expressed in the residents' concern of being less competent ObGyn specialists. Hence, this issue should be tackled accordingly by dividing working shifts with a smaller number of residents to reduce the spread of disease [17]. Analysis between social factors and psychological trauma showed no significant relationship with anxiety disorders, depression, and psychological trauma. This might be caused by the low prevalence of anxiety and depression found in this study, thus, it could not describe the social factors that exist in the population. The level of anxiety in this study showed a non-significant relationship with age, year of residency training program, and institution. Only one subject experienced severe anxiety whereas two subjects experienced moderate anxiety. Previous studies have shown that anxiety level in younger subjects is lower than in older subjects [18]. The relationship between marital status and living conditions with the level of anxiety showed no significant results due to the limited number of research subjects experiencing anxiety during the pandemic. The increased anxiety levels could be caused by concern about the possibilities of transmitting COVID-19 to the closest family especially to married participants. In depressive disorders, residents' age did not show a significant difference between groups. Previous studies have revealed that the highest risk at being depressed was experienced by age group 21–25 years, followed by the group consists of 26–30 years and 31–35 years, consecutively [17]. The incidence of moderate to severe depression was also resulted higher in residents from University of Indonesia compared to the two others with a 9% prevalence. University of Indonesia is both a teaching hospital and a tertiary hospital which is the main national referral hospital. This reason led to a greater workload and higher teachers' expectations of ObGyn residents. Marital status and living conditions did not show significant results. All participants who felt major depression were residents who were not-yet married. This can be caused by the influence of marriage on the incidence of depression [19]. Spouse support has been identified as a major protective factor against depression in difficult circumstances. Apparently, exposure to COVID-19 during clinical rotation did not reduce the incidence of depression in ObGyn residents. Participants with mild, moderate, and severe symptoms of psychological trauma had an age ranged 30–31 years. Previous research revealed that incidence of psychological trauma generally increases with age [20]. However, the age range between 26 and 42 years of the participants indicated that there was a limitation of age variation. This was also shown by the non-significant relationship between year of residency and psychological trauma. The incidence of psychological trauma with severe symptoms was mostly found at the 4th year of residency because of the workload and exposure to different stressors compared to other levels. Workload and stressors in the workplace can increase the risk of psychological trauma [21]. Psychological trauma with severe symptoms mostly occurs in ObGyn residents from the University of Indonesia for similar reasons. Social support from the closest ones is of utmost importance. Studies have proven a decreased complex psychological trauma in married subjects [12, 19, 21]. Clinical rotations exposed to COVID-19 did not show significant results on the severity of symptoms of psychological trauma. Many of the residents reported undergoing severe psychological trauma related to their concern over contracting the disease after being exposed to COVID-19, fear they would infect their family and having to witness the death of close colleagues, which was considered an emotional loss and a reminder of the risk they themselves were taking. Our study has several limitations. This study was conducted in only three institutions in Indonesia that provide ObGyn residency, and thus, further research including other institutions which provide ObGyn residency is needed to validate and extend these findings in similar situations as well as outbreaks of other infectious diseases. Other factors that could potentially contribute to the complex psychosocial responses of an individual such as personality variables, cognitive appraisal mechanisms, and past trauma were not examined in this study. Our study is a cross-sectional study therefore it is not possible to establish cause-effect relationships. This study use self-referenced questionnaires and it is possible that there are no errors when measuring the variables, also the questionnaire were taken online without random sampling so the results cannot be directly extrapolated to the rest of the population of medical residents. Despite the limitations, our study provide initial information and stepstone regarding social factors influence on anxiety, depression level, and psychological trauma of ObGyn residents during COVID-19 pandemic. This may increase the need to evaluate mental health problems during COVID-19 pandemic on ObGyn residents in other institutions, other residents or healthcare workers in general as they are at a high risk of experiencing mental health problems. 5 Conclusion In conclusion, the rapidly spreading COVID-19 on a global scale has led many educational systems to face tough times due to social distancing recommendation. ObGyn residents who should quickly adapt into the current situation experienced a negative impact due to COVID-19 pandemic. However, our study concluded that social factors did not influence the occurrence of anxiety, depression, and psychological trauma of the ObGyn residents during COVID-19 pandemic. Declarations Author contribution statement Donny Damiar Santoso Lukman: Performed the experiments; Analyzed and interpreted the data; Wrote the paper. Mila Maidarti; Achmad Kemal Harzif; Kanadi Sumapraja: Conceived and designed the experiments; Wrote the paper. Fistyanisa Elya Charilda; Azizah Fitriayu Andyra: Analyzed and interpreted the data; Wrote the paper. Natalia Widiasih Raharjanti; Monika Kristi Levania; Hilwah Nora; Cut Meurah Yeni; Hardyan Sauqi; Ferry Armanza: Analyzed and interpreted the data. Funding statement This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Data availability statement The authors are unable or have chosen not to specify which data has been used. Declaration of interest’s statement The authors declare no competing interests. Additional information No additional information is available for this paper. Acknowledgment Not applicable. ==== Refs References 1 Parrish RK, Stewart MW, Powers SLD. Ophthalmologists Are more than eye doctors—in memoriam Li Wenliang. Am. J. Ophthalmol. Published online 2020: A1-A2. 2 Daroedono E. Siagian F.E. Alfarabi M. The impact of COVID-19 on medical education: our students perception on practice of online education Int. J. Commun. Med. Publ. Health 7 7 2020 2790 2796 3 Sethi B.A. Sethi A. Ali S. Aamir H.S. Impact of Coronavirus disease (COVID-19) pandemic on health professionals Pak. J. Med. Health Sci. 36 COVID19-6-11 2020 4 Balhareth A. AlDuhileb M.A. Aldulaijan F.A. Aldossary M.Y. Impact of COVID-19 pandemic on residency and fellowship training programs in Saudi Arabia: a nationwide cross-sectional study Ann. Med. Surg. 57 June 2020 127 132 5 Spoorthy M.S. Pratapa S.K. Supriya M. Mental health problems faced by healthcare workers due to the COVID-19 pandemic–a review Asian J. Psychiatric January 2020 6 Xiao H. Zhang Y. Kong D. Li S. Yang N. The effects of social support on sleep quality of medical staff treating patients with coronavirus disease 2019 (COVID-19) in January and February 2020 in China Med. Sci. Mon. Int. Med. J. Exp. Clin. Res. 26 2020 1 8 7 Hall R.C.W. Hall R.C.W. Chapman M.J. The 1995 Kikwit Ebola outbreak: lessons hospitals and physicians can apply to future viral epidemics Gen. Hosp. Psychiatr. 30 5 2008 446 452 8 Müller N. Infectious diseases and mental health Key Issues Ment. Health 179 2015 99 9 Sim K. Huak Chan Y. Chong P.N. Chua H.C. Wen Soon S. Psychosocial and coping responses within the community health care setting towards a national outbreak of an infectious disease J. Psychosom. Res. 68 2 2010 195 202 20105703 10 Naushad V.A. Bierens J.J.L.M. Nishan K.P. A systematic review of the impact of disaster on the mental health of medical responders Prehospital Disaster Med. 34 6 2019 632 643 31625487 11 SQUIRE. Revised Standards for Quality Improvement Reporting Excellence SQUIRE 2.0. 12 Mina S. Predictors of marriage in psychiatric illness: a review of literature. Journal of Psychiatr. Psychiatr. Disord. 3 1 2019 [Google Scholar] 13 Setiawati Y. Wahyuhadi J. Joestandari F. Maramis M.M. Atika A. Anxiety and resilience of healthcare workers during COVID-19 pandemic in Indonesia J. Multidiscip. Healthc. 14 2021 1 8 33442258 14 Ariestanti Y. Widayati T. Sulistyowati Y. Determinan Perilaku Ibu Hamil Melakukan Pemeriksaan Kehamilan (antenatal care) Pada Masa Pandemi Covid -19 Jurnal Bidang Ilmu Kesehatan 10 2 2020 203 216 15 Li Y. Scherer N. Felix L. Kuper H. Prevalence of depression, anxiety and posttraumatic stress disorder in health care workers during the COVID-19 pandemic: a systematic review and meta-Analysis PLoS One 16 3 March 2021 1 19 16 Giordano L. Cipollaro L. Migliorini F. Maffulli N. Impact of Covid-19 on undergraduate and residency training Surgeon 2020 1 8 Published online 17 Brenes G.A. Age differences in the presentation of anxiety Aging Ment. Health 10 3 2006 298 302 16777658 18 Talukder U.S. Uddin M.J. Mohammad Khan N. Major depressive disorder in different age groups and quality of life Bangladesh J. Psychiatr. 28 2 2017 58 61 19 Bulloch A.G. Williams J.V. Lavorato D.H. Patten S.B. The relationship between major depression and marital disruption is bidirectional Depress. Anxiety 26 12 2009 1172 1177 19798680 20 Raudenska J. Steinerova V. Javurkova A. Occupational burnout syndrome and post-traumatic stress among healthcare professionals during the novel coronavirus disease 2019 (COVID-19) pandemic Best Pract. Res. Clin. Anaesthesiol. 34 2020 553 560 33004166 21 Nieder C. Kämpe T.A. Does marital status influence levels of anxiety and depression before palliative radiotherapy? In Vivo 32 2 2018 327 330 29475916

0

PMC9749376

NO-CC CODE

2022-12-15 23:23:24

no

Heliyon. 2022 Dec 14; 8(12):e12271

utf-8

Heliyon

10.1016/j.heliyon.2022.e12271

oa_other

==== Front Curr Probl Cardiol Curr Probl Cardiol Current Problems in Cardiology 0146-2806 1535-6280 Mosby-Year Book S0146-2806(22)00450-9 10.1016/j.cpcardiol.2022.101553 101553 Article Impact of COVID-19 on Patients Hospitalized with Deep Vein Thrombosis and/or Pulmonary Embolism: A Nationwide Analysis Hajra Adrija 1 Goel Akshay 2 Malik Aaqib H 2 Isath Ameesh 2 Shrivastav Rishi 3 Gupta Rahul 4 Das Subrat 2 Krittanawong Chayakrit 5 Bandyopadhyay Dhrubajyoti 2⁎ 1 Montefiore Medical Center/Albert Einstein College of Medicine, Bronx, NY. 2 Westchester Medical Center, New York Medical College, Valhalla, NY. 3 Mount Sinai St. Luke's-BronxCare, New York, NY. 4 Lehigh Valley Heart Institute, Lehigh Valley Health Network, Allentown, PA. 5 NYU Langone Health, New York, NY ⁎ Corresponding author: Dhrubajyoti Bandyopadhyay, Department of Cardiology, Westchester Medical Center, New York Medical College, Valhalla, NY 14 12 2022 14 12 2022 101553. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The Coronavirus disease 2019 (COVID-19) infection predisposes patients to develop deep vein thrombosis (DVT) and pulmonary embolism (PE). In this study, we compared the in-hospital outcomes of patients with DVT and/or PE with concurrent COVID-19 infection vs. those with concurrent flu infection. The National Inpatient Sample from 2019 to 2020 was analyzed to identify all adult admissions diagnosed with DVT and PE. These patients were then stratified based on whether they had concomitant COVID-19 or flu. We identified 62,895 hospitalizations with the diagnosis of DVT and/or PE with concomitant COVID-19, and 8,155 hospitalizations with DVT and/or PE with concomitant flu infection. After 1:1 propensity score match, the incidence of cardiac arrest and inpatient mortality were higher in the COVID-19 group. The incidence of cardiogenic shock was higher in the flu group. Increased age, Hispanic race, diabetes, chronic kidney disease, arrhythmia, liver disease, coagulopathy, and rheumatologic diseases were the independent predictors of mortality in patients with DVT and/or PE with concomitant COVID-19. Keywords Coronavirus disease 2019 (COVID-19) deep vein thrombosis pulmonary embolism National Inpatient Sample ==== Body pmcIntroduction The incidence of thrombotic events in patients with coronavirus disease 2019 (COVID-19) is significantly higher than the incidence in patients without COVID-19.1 Influenza or flu infection has also been found to be associated with deep vein thrombosis (DVT).2 Various studies have been done to identify the characteristics of patients affected with COVID-19 or flu infection who developed thrombotic complications, including DVT or pulmonary embolism (PE).3 , 4 Studies have been conducted to show the difference in the occurrence of thrombotic events between patients with COVID-19 and patients with non-COVID-19 viral infections, including flu.5 , 6 But there is no large-scale study comparing the in-hospital outcomes in patients admitted with a diagnosis of DVT and/or PE who have concurrent COVID-19 vs. flu infection. We have analyzed National Inpatient Sample (NIS) database to compare the demographic data, co-morbidities, in-hospital complications, and mortality in patients with DVT and/or PE who had concomitant COVID-19 infection vs. patients admitted with DVT/PE with concomitant flu infection. Also, we demonstrated the clinical predictors of adverse outcomes in DVT/PE patients with COVID-19 infection. Methods Data Source The HCUP NIS database is the largest all-payer in-hospital database in the US and is available publicly. We have used the NIS database from 2019 to 2020 for our study. The NIS represents 95% of US hospitalizations from 44 states participating in HCUP and provides a stratified sample of 20% of discharges, including up to 8 million hospital discharges per year. The NIS database has been previously demonstrated to correlate well with other discharge databases in the US. In addition, it has been validated in various studies to provide reliable estimates of admissions within the US.7 Study Population We included hospitalizations with DVT and/or PE and stratified them based on concurrent COVID-19 and influenza infection diagnosis by International Classification of Diseases 10th Revision clinical modification (ICD-10-CM) codes. Studies have shown overall positive predictive value for COVID-19 diagnosis with ICD-10-CM is 99%.8 Outcomes The primary outcome of interest was in-hospital mortality in patients with DVT and/or PE with COVID-19 compared with those with DVT and/or PE with influenza infection. Secondary outcomes included acute kidney injury (AKI) as well as, AKI requiring hemodialysis, sepsis, stroke, cardiogenic shock, the requirement of vasopressors, acute respiratory failure as well as, respiratory failure requiring intubation, need for mechanical circulatory support such as intra-aortic balloon pump (IABP), extracorporeal membrane oxygenation (ECMO), length of stay (LOS), and hospitalization costs. Statistical Analysis Statistical analyses were performed using Stata 16.0 (StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC). The discharge weights provided by the Agency for Healthcare Research and Quality were applied to obtain weighted numbers to calculate national estimates. A 1:1 propensity score matching (nearest-neighbor matching with a caliper width of 0.1 of the estimated propensity scores) was performed to compare outcomes for patients with DVT and/or PE and concomitant COVID-19 vs. patients with DVT/PE with concurrent influenza infection. Multivariate logistic regression models were generated to identify the independent predictors of mortality and reported as adjusted odds ratio (aOR) with a 95% confidence interval (CI). Categorical variables were expressed as percentages. Continuous variables were expressed as median and interquartile range. Categorical variables were compared using the Pearson chi-square test, while continuous variables were compared using the student's t-test. All reported P values are 2-sided, with a value of < 0.05 considered significant. Results We identified 62,895 hospitalizations with DVT and/or PE with a concurrent diagnosis of COVID-19 infection. The number of hospitalizations with DVT and/or PE with a concurrent diagnosis of influenza was 8155. Propensity score matching was performed to create a more balanced population, with 8110 hospitalizations in each group. Table 1 describes the baseline characteristics of patients admitted with DVT and/or PE with concomitant COVID-19 vs. concomitant influenza infection. The mean age of DVT and/or PE patients with COVID-19 was 66 years [standard deviation (SD): 55-77] vs. 63 years (SD: 51-74) in DVT/PE with flu group. The percentage of female patients (43.4% vs. 41.1%, P <0.001) and the Caucasian population (62% vs. 50.2%, P <0.001) were significantly higher in the flu group compared to the COVID-19 group. The Hispanic population was higher in the COVID-19 group (10.4% vs. 17.1%). The number of patients with a history of congestive heart failure (34% vs. 18%, P <0.001) and valvular heart disease (8% vs. 4%, P <0.001) was higher in the flu group compared to the COVID-19. (Table 1)Table 1 Baseline characteristics of patients with DVT and/or PE with concurrent Flu and COVID-19 before and after propensity match with complications of hospitalized patients. Table 1:Characteristics Before Matching After Matching With Flu With COVID-19 P Value With Flu With COVID-19 P Value Total number of patients 8155 62,895 8110 8110 Age, median IQR, years 63 (51-74) 66 (55-77) <0.001 63 (51-74) 63 (52-73) 0.693 Age groups (%) <0.001 0.873 18-59 41.3 33.8 430 420 60-69 24.8 24.5 85 75 70-79 19.5 23.2 95 115 >79 years 14.4 18.5 70 70 Female (%) 47.4 41.1 <0.001 47.5 49 0.0385 Caucasian race (%) 61.2 50.2 <0.001 61.8 62.9 0.534 African American race (%) 19.9 22.5 0.021 20 19 0.49 Hispanics (%) 10.4 17.1 <0.001 10.4 11 0.588 Atrial fibrillation (%) 17.5 12.9 <0.001 17.3 18.4 0.407 Diabetes mellitus (%) 30.1 37.9 <0.001 30.1 27.8 0.13 Hypertension (%) 64.4 63.6 0.539 64.4 63.6 0.667 Chronic kidney disease (%) 19.8 18.8 0.348 19.9 19.3 0.695 CHF (%) 34 18.3 <0.001 34.1 21 <0.001 Peripheral vascular disease (%) 8.1 5.4 <0.001 8 8.1 0.949 Dementia (%) 7.1 11.2 <0.001 7.2 7.5 0.734 COPD (%) 39.4 21.1 <0.001 39.1 39.4 0.887 Valvular heart disease (%) 8.1 4.1 <0.001 8.1 5.1 <0.001 Arrhythmias (%) 33.3 28.5 <0.001 33.1 31.9 0.456 Liver disease (%) 9.6 7.1 <0.001 9.6 7.6 0.048 Hypothyroidism (%) 10.7 11.6 0.247 10.7 13.8 0.005 Anemia (%) 7.4 5.3 0.001 7.5 7.6 0.895 Cancer (%) 13.1 5.8 <0.001 12.8 13.9 0.372 Rheumatological disorders (%) 5.2 3 <0.001 5.2 4.7 0.563 Weight loss (%) 17 13.1 <0.001 16.9 16.8 0.927 Coagulopathy (%) 19.4 22.7 0.003 19.5 19.1 0.793 Obesity (%) 25.2 27.2 0.096 25.3 24 0.377 Smoking history (%) 16.7 5.3 <0.001 16.3 14.7 0.22 Coronary artery disease (%) 16.5 15.4 0.248 16.4 16.5 0.961 Prior stroke (%) 6.7 7.2 0.552 6.7 7 0.783 Prior PCI (%) 3.3 3.2 0.886 3.3 3.1 0.761 Prior CABG (%) 2.7 2.5 0.619 2.7 2.4 0.578 Alcohol (%) 3.7 1.7 <0.001 3.6 4.1 0.469 Prior MI (%) 4.1 3.3 0.084 4.1 4 0.856 Discharge (%) Routine 38.4 34.9 <0.001 38.5 34.8 <0.001 SNF/NH/IC 28.5 23.3 28.5 24.8 Home healthcare 16.3 14.8 16.2 14 Length of stay, median (IQR), days 8 (4-18) 9 (4-18) 0.83 8 (4-18) 9 (4-18) 0.6222 Hospital location and teaching status (%) Rural 7.1 7.9 0.4911 7 7.5 0.627 Urban non-teaching 16.7 17.3 16.7 17.8 Urban teaching 76.2 74.8 76.3 74.7 Hospital region (%) 0.042 0.888 Northeast 16.2 18.9 16.3 17.1 Midwest 24.7 25.1 24.5 25.3 South 39.2 39.4 39.3 38.1 West 19.9 16.7 19.9 19.5 Insurance (%) <0.001 <0.001 Medicare 51 51 50.9 47.2 Medicaid 17.7 12.6 17.7 14 Private including HMO 24.7 28 24.8 29.7 Self-pay 4.1 3.6 4.1 4.1 Median household income (%) 0.032 0.875 0-25th percentile 32.8 32.8 32.7 31.7 26-50th percentile 24.9 27.5 25 26.1 51-75th percentile 21.2 22 21.2 21.1 76-100th percentile 19 16.3 19 19.3 Total hospital cost USD median IQR 27,491 (11224-66827) 21,304 (10381-49243) <0.001 27,435 (11158-66872) 21,153 (10584-50103) <0.001 Hospital bed size (%) 0.002 0.760 Small 18.9 21 19 19.5 Intermediate 25.4 29.1 25.5 26.4 Large 55.7 50 55.6 54.1 CABG- coronary artery bypass graft, CHF- congestive heart failure, COVID 19- coronavirus disease of 2019, CKD- chronic kidney disease, COPD- chronic obstructive pulmonary disease, HMO- health maintenance organization, MI- myocardial infarction, PCI- percutaneous coronary intervention, SNF/NH/IC- skilled nursing facility/nursing home/ intermediate care Table 2 describes the in-hospital complications and outcomes of the admitted patients. Patients with DVT and/or PE with flu had more incidences of sepsis (43.2% vs. 37.2%), in-hospital stroke (2.3% vs. 1.5%), and cardiogenic shock (3.8% vs. 1.8%). The use of Impella and the incidence of coronary artery bypass graft were higher in the flu group. Even after the propensity matching, the number of patients with the above-mentioned complications was significantly higher in the flu group. Cardiac arrest (3.1% vs. 5.5%) and in-hospital mortality (11.2% vs. 23.1%) were significantly higher in patients with DVT and/or PE with concurrent COVID-19 infection before and after propensity matching (P <0.001). (Table 2) There was no significant difference in the LOS between these two groups, but the cost of hospitalization was significantly higher in the flu group before and after propensity matching (P <0.001). (Table 1)Table 2 Complication of hospitalized patients with DVT and/or PE with concurrent Flu and COVID-19. Table 2:Complications Before Matching After Matching With Flu (%) With COVID-19 (%) P Value With Flu (%) With COVID-19 (%) P Value Total number of patients 8155 62,895 8110 8110 AKI 39.2 40 0.599 39.2 36.7 0.149 AKI leading to HD 5.9 6 0.884 6 5.4 0.495 UTI 14.1 12.6 0.078 14.2 12.5 0.137 Sepsis 43.2 37.2 <0.001 43.2 38 0.003 DVT 56.1 46.6 <0.001 56.2 46.4 <0.001 PE 55.1 63.9 <0.001 55.1 63.6 <0.001 Stroke in-hospital 2.3 1.5 0.011 2.3 1.1 0.013 Cardiogenic shock 3.9 1.8 <0.001 3.8 1.6 <0.001 Cardiac arrest 3.1 5.6 <0.001 3.1 6 <0.001 VT 4.3 3.4 0.064 4.2 2.7 0.025 VF 0.5 0.7 0.439 0.5 0.6 0.635 Bleeding requiring transfusion 10.9 8.8 0.007 11 9.2 0.113 Death 11.2 23.1 <0.001 11.2 21.5 <0.001 Vasopressors 5.6 6.8 0.102 5.6 5.9 0.759 Prolonged intubations >24 hours 26.5 25.3 0.299 26.5 23.2 0.037 Respiratory failure 59.4 59.2 0.882 59.3 58.4 0.626 ECMO utilization 0.4 0.2 0.034 0.4 0.2 0.224 Impella 0.3 0.02 <0.001 0.2 0 0.045 IABP 0.1 0.04 0.689 0.1 0 0.316 CABG 0.25 0.01 <0.001 0.2 0 0.045 PCI 0.4 0.2 0.024 0.4 15 0.205 AKI- acute kidney injury, COVID 19- coronavirus disease of 2019, CABG- coronary artery bypass graft, DVT- deep vein thrombosis, ECMO- extracorporeal membrane oxygenation, HD- hemodialysis, HTN- hypertension, IABP- intra-aortic balloon pump, MI- myocardial infarction, PCI- percutaneous coronary intervention, PE- pulmonary embolism, UTI- urinary tract infection, VT- ventricular tachycardia, VF- ventricular fibrillation Predictors of Mortality On multivariable regression analysis, increased age [adjusted odds ratio (aOR) 1.023, 95% confidence interval (CI) 1.020-1.027, P <0.001], Hispanic race (aOR 1.196, 95% CI 1.019-1.403, P <0.05), presence of diabetes (aOR 1.144, 95% CI 1.045-1.251, P<0.05), chronic kidney disease (aOR 1.309, 95% CI 1.171-1.464, P <0.001), congestive heart failure (CHF) (aOR 1.180, 95% CI 1.057-1.319, P= 0.003), chronic obstructive pulmonary disease (COPD) (aOR1.125, 95% CI 1.006-1.258, P= 0.038), arrhythmia (aOR 1.777, 95% CI 1.618-1.951, P <0.001), liver disease (aOR 2.639, 95% CI 2.282-3.051, P<0.001), coagulopathy (aOR 1.822, 95% CI 1.645-2.019, P<0.001), rheumatologic diseases (aOR 1.398, 95% CI 1.081-1.806, P<0.011), and weight loss (aOR 1.332, 95% CI 1.169-1.518, P <0.001) were independent predictors of mortality in patients with DVT and/or PE and concurrent COVID-19 infection. (Table 3 ).Table 3 Predictors of mortality after multivariate analysis. Table 3:Variable Odds Ratio Lower Limit Upper Limit P Value Age 1.023 1.020 1.027 <0.001 Hispanic race 1.196 1.019 1.403 0.028 Diabetes 1.144 1.045 1.251 0.003 Coagulopathy 1.822 1.645 2.019 <0.001 Weight loss 1.332 1.169 1.518 <0.001 Arrhythmias 1.777 1.618 1.951 <0.001 Liver disease 2.639 2.282 3.051 <0.001 CKD 1.309 1.171 1.464 <0.001 CHF 1.180 1.057 1.319 0.003 COPD 1.125 1.006 1.258 0.038 Rheumatologic diseases 1.398 1.081 1.806 0.011 CKD- chronic kidney disease, CHF- congestive heart failure, COPD- chronic obstructive pulmonary disease Discussion To the best of our knowledge, this is the largest nationwide data to report the characteristics and outcomes of patients with DVT and/or PE and concurrent COVID-19 vs. flu infection. Direct endothelial injury, activation of coagulation factors, cytokine storms, and suppression of fibrinolytic associated with severe COVID-19 infection contribute to hypercoagulability and thrombotic complications.1 Studies have shown up to 30% of patients hospitalized with COVID-19 develop arterial or venous thromboembolism.9 Prior studies have shown older age, CKD, COPD, heart failure, and prior venous thrombotic events predispose venous thromboembolism in COVID-19 patients.9 , 10 As found in our study, the predictors of mortality in patients with DVT/ PE with concomitant COVID-19 also include similar risk factors. A study by Lo Re et al. showed that after an inpatient venous thrombotic event, the risk of 30-day mortality was significantly higher in patients with COVID-19 compared to those with influenza.9 We also found inpatient mortality is higher in COVID-19 patients than in the flu. A cohort study from the US showed that COVID-19 was independently associated with a higher 90-day risk for venous thrombosis but not arterial thrombosis, compared with influenza.11 Another study from Europe showed patients with influenza were more often diagnosed with arterial thrombotic complications than patients with COVID-19 infection.12 Our analysis found that in-hospital stroke was higher in flu patients compared to COVID-19 patients. The finding of increased stroke corroborates with the increased risk of arterial thrombosis in influenza patients. The severity of infection has been linked with the incidence of DVT and PE in COVID-19 cohort.13 Our study found that cardiac arrest and inpatient mortality was higher in the COVID-19-affected individuals. This finding indirectly indicates the possible association of DVT and PE with the severity of COVID-19 infection. The COVID-19 patients admitted with DVT and/or PE were possibly affected with severe COVID-19 disease. The severity of the disease contributed to the increased mortality. COVID-19 has already been identified as a significant economic burden in the US.14 Our data analysis also showed that the hospital cost was significantly higher in the COVID-19-affected patients. Limitations NIS study has its inherent limitations. NIS data is a retrospective database analysis with discharge diagnosis and does not have patient-level information. Unmeasured confounding factors may affect these findings. We cannot get any follow-up information from NIS data analysis. As this data is from 2019 to 2020, the impact of COVID-19 vaccination on these outcomes was not available. Despite these limitations, NIS is a well-validated representation of the US population and with internal and external quality control measures. The large sample size of NIS data also compensates for the residual confounders. Conclusion COVID-19 infection among patients hospitalized with DVT and/or PE is associated with significantly higher in-hospital mortality. In addition, increased age, Hispanic race, presence of diabetes, CKD, COPD, CHF, arrhythmia, liver disease, coagulopathy, and rheumatologic disease were independent predictors of mortality in patients with DVT and/or PE with concurrent COVID-19. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Sources of funding and support No external funding was used in the preparation of this manuscript. Ethical Approval of Studies and Informed Consent Not applicable as it is a retrospective analysis of data Acknowledgement None Author access to data Publicly available National Inpatient Sample of the US ==== Refs References 1 Hajra A. Mathai S.V. Ball S. Management of Thrombotic Complications in COVID-19: An Update Drugs 80 2020 1553 1562 10.1007/s40265-020-01377-x 32803670 2 Ishiguro T Matsuo K Fujii S Takayanagi N. Acute thrombotic vascular events complicating influenza-associated pneumonia Respiratory medicine case reports 28 2019 Jan 1 100884 3 Agarwal G Hajra A Chakraborty S Patel N Biswas S Adler MK Lavie CJ. Predictors and mortality risk of venous thromboembolism in patients with COVID-19: systematic review and meta-analysis of observational studies Therapeutic advances in cardiovascular disease 2022 Jun; 16 17539447221105013 4 Bunce Paul E. High Sasha M. Nadjafi Maral Stanley Katherine Conrad Liles W. Christian Michael D. Pandemic H1N1 Influenza Infection and Vascular Thrombosis Clinical Infectious Diseases Volume 52 Issue 2 15 January 2011 e14 e17 10.1093/cid/ciq125 5 Mai V Tan BK Mainbourg S Potus F Cucherat M Lega JC Provencher S. Venous thromboembolism in COVID-19 compared to non-COVID-19 cohorts: A systematic review with meta-analysis Vascul Pharmacol 139 2021 Aug 106882 10.1016/j.vph.2021.106882 Epub 2021 Jun 2. PMID: 34087481; PMCID: PMC8169236 6 Tufano A Rendina D Abate V Casoria A Marra A Buonanno P Galletti F Di Minno G Servillo G Vargas M. Venous Thromboembolism in COVID-19 Compared to Non-COVID-19 Cohorts: A Systematic Review with Meta-Analysis J Clin Med 10 21 2021 Oct 25 4925 10.3390/jcm10214925 PMID: 34768445; PMCID: PMC8584903 34768445 7 Isath A Malik AH Goel A Gupta R Shrivastav R Bandyopadhyay D. Nationwide Analysis of the Outcomes and Mortality of Hospitalized COVID-19 Patients Curr Probl Cardiol 48 2 2022 Oct 8 101440 10.1016/j.cpcardiol.2022.101440 Epub ahead of printPMID: 36216202; PMCID: PMC9546497 8 Bodilsen J Leth S Nielsen SL Holler JG Benfield T Omland LH. Positive Predictive Value of ICD-10 Diagnosis Codes for COVID-19 Clin Epidemiol 13 2021 May 25 367 372 10.2147/CLEP.S309840 PMID: 34079379; PMCID: PMC8164665 34079379 9 Lo Re V 3rd Dutcher SK Connolly JG Perez-Vilar S Carbonari DM DeFor TA Djibo DA Harrington LB Hou L Hennessy S Hubbard RA Kempner ME Kuntz JL McMahill-Walraven CN Mosley J Pawloski PA Petrone AB Pishko AM Driscoll MR Steiner CA Zhou Y Cocoros NM Association of COVID-19 vs Influenza With Risk of Arterial and Venous Thrombotic Events Among Hospitalized Patients JAMA 328 7 2022 Aug 16 637 651 10.1001/jama.2022.13072 PMID: 35972486; PMCID: PMC9382447 35972486 10 Chang H Rockman CB Jacobowitz GR Speranza G Johnson WS Horowitz JM Garg K Maldonado TS Sadek M Barfield ME. Deep vein thrombosis in hospitalized patients with coronavirus disease 2019 J Vasc Surg Venous Lymphat Disord 9 3 2021 May 597 604 10.1016/j.jvsv.2020.09.010 Epub 2020 Oct 8. PMID: 33039545; PMCID: PMC7543928 33039545 11 Ward A Sarraju A Lee D Bhasin K Gad S Beetel R Chang S Bonafede M Rodriguez F Dash R. COVID-19 is associated with higher risk of venous thrombosis, but not arterial thrombosis, compared with influenza: Insights from a large US cohort PLoS One 17 1 2022 Jan 12 e0261786 10.1371/journal.pone.0261786 PMID: 35020742; PMCID: PMC8754296 12 Stals MAM Grootenboers MJJH van Guldener C Kaptein FHJ Braken SJE Chen Q Chu G van Driel EM Iglesias Del Sol A de Jonge E Kant KM Pals F Toorop MMA Cannegieter SC Klok FA MV Huisman Dutch COVID & Thrombosis Coalition (DCTC). Risk of thrombotic complications in influenza versus COVID-19 hospitalized patients Res Pract Thromb Haemost 5 3 2021 Apr 8 412 420 10.1002/rth2.12496 PMID: 33821230; PMCID: PMC8014477 33821230 13 Rali P O'Corragain O Oresanya L Yu D Sheriff O Weiss R Myers C Desai P Ali N Stack A Bromberg M Lubitz AL Panaro J Bashir R Lakhter V Caricchio R Gupta R Dass C Maruti K Lu X Rao AK Cohen G Criner GJ Choi ET Temple University COVID-19 Research Group Incidence of venous thromboembolism in coronavirus disease 2019: An experience from a single large academic center J Vasc Surg Venous Lymphat Disord 9 3 2021 May 585 591 10.1016/j.jvsv.2020.09.006 e2Epub 2020 Oct 5. Erratum in: J Vasc Surg Venous Lymphat Disord. 2022 May;10(3):799. PMID: 32979557; PMCID: PMC7535542 32979557 14 DeMartino JK Swallow E Goldschmidt D Yang K Viola M Radtke T Kirson N. Direct health care costs associated with COVID-19 in the United States J Manag Care Spec Pharm 28 9 2022 Sep 936 947 10.18553/jmcp.2022.22050 Epub 2022 Jun 18. PMID: 35722829 35722829

0

PMC9749377

NO-CC CODE

2022-12-15 23:22:53

no

Curr Probl Cardiol. 2022 Dec 14;:101553

utf-8

Curr Probl Cardiol

10.1016/j.cpcardiol.2022.101553

oa_other

==== Front Am J Emerg Med Am J Emerg Med The American Journal of Emergency Medicine 0735-6757 1532-8171 Elsevier Inc. S0735-6757(22)00753-7 10.1016/j.ajem.2022.12.012 Article Incidence of unknown COVID-19 infection in a cohort of emergency physicians and advance practice providers Aaron Nathan Barksdale M.D, FACEP a⁎ Wood Macy G. Ph.D., D (ABMM) b Branecki Chad E. DO, FACEP a Zimmerman Brooklin MSN, RN a Lyden Elizabeth MS c Nguyen Thang T. PhD a Hatfield Andrew MD a Koepsell Scott MD, Ph.D d Langenfeld Jason MD a Zeger Wesley G. DO, FACEP a Wadman Michael C. M.D., FACEP a a Department of Emergency Medicine, University of Nebraska Medical Center, Omaha, NE, United States of America b Department of Pathology, Medical College of Wisconsin, Milwaukee, WI, United States of America c Research Design and Analysis, College of Public Health, University of Nebraska Medical Center, Omaha, NE, United States of America d Clinical Operations, Department of Pathology and Microbiology, University of Nebraska Medical Center, Omaha, NE, United States of America ⁎ Corresponding author at: 981150 Nebraska Medical Center, Omaha, NE 68198-1150, United States of America. 14 12 2022 14 12 2022 5 12 2022 12 12 2022 © 2022 Elsevier Inc. All rights reserved. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Introduction In United States, health care workers have been immersed in the COVID-19 pandemic since February 2020. Since availability of COVID-19 vaccines, there is limited literature investigating the incidence of unknown COVID-19 infections in physicians and Advanced Practitioner Providers (APPs) working in emergency departments (EDs). The primary objective is to determine the incidence unknown COVID-19 infection within a cohort of emergency physicians (EPs) and APPs. Methods Prospective observational study at a tertiary academic center with emergency medicine residency and 64,000 annual ED visits. EPs/APPs providing care to ED patients over the prior 12 months were eligible. Serum samples were collected between May 1 and June 30, 2022. Analysis utilized Luminex xMAP® SARS-CoV-2 Multi-Antigen IgG Assay for antibodies to Nucleocapsid, Receptor-binding domain, and Spike subunit 1. Mean Fluorescent Intensity (MFI) ≥ 700 was considered positive. Subjects completed 12 question survey assessing demographics and previously confirmed COVID-19 infection. Fisher's exact test evaluated associations of demographics and clinical characteristics with confirmed COVID-19 status. Analyses performed using SAS, Version 9.4. P < 0.05 considered statistically significant. Results Sixty-nine of 81 eligible subjects (85.2%) participated, 58.0% were male, 97.1% white, with mean age of 37. Eighteen subjects had MFI ≥ 700 strongly suggestive of prior infection, with 17.7% unknown. No statistically significant difference between age, gender, race, children in home, or household member with previously COVID-19 infection. Conclusion Unknown previous COVID-19 infection was less then expected in this cohort of EPs/APPs, and no association with individual characteristics, previously infected household member, or children in the home. Keywords COVID-19 SARS-CoV-2 Antibodies Infection Emergency medicine ==== Body pmc1 Introduction The United States (US) COVID-19 pandemic began in February of 2020, consisting of several waves of increased infection, hospitalization, and mortality rates. Emergency department providers (EDPs) including physicians, residents, and Advanced Practice Providers (APPs) have experienced some of the highest exposure rates with individuals infected with the SARS-CoV-2 virus. Studies following the initial wave of infections suggested confirmed positive rates in physicians and APPs ranging from 5 to 40% [1,2]. Following the availability of COVID-19 vaccines in December of 2020, the incidence of infection in health care workers (HCWs) had significantly decreased. Data from a recent systematic review and meta-analysis reported confirmed rates of 9–11% amongst HCWs in general but did not specify the prevalence in EDPs [3]. Evidence consistently suggests that fully vaccinated individuals experienced good protection from severe disease and death due to the SARS-CoV-2 variants that predominated the first three waves of the pandemic (Alpha, Beta, and Delta) [4]. In December 2021, the SARS-CoV-2 B.1.1.529 (Omicron) variant emerged and overtook Delta as the predominant variant. Studies evaluating serum samples in vaccinated individuals demonstrated lower titers of neutralizing antibodies against Omicron compared to the previous three predominate variants. In addition, those previously infected with COVID-19 had low or even undetectable titers against Omicron [5,6]. This supports the much more transmissible and infectious pattern that Omicron has demonstrated [7]. Fortunately, Omicron and its variants' severity of disease and mortality rates have been less than the previous strains. In fully vaccinated individuals, breakthrough infections are comparable to minor common cold infections with symptoms lasting for 4–6 days, and numerous confirmed cases with minimal or no symptoms [8]. As of March 2022, the incidence of new COVID-19 infections declined dramatically throughout the US. At the initiation of this study, the Emergency Department (ED) had a total of 11 PCR confirmed COVID-19 infections amongst the EDP group (n = 81). Two cases occurred during the initial “Omicron wave”, while the remaining 9 within the prior 12 months. Investigators of this study hypothesize there have been many more unconfirmed COVID-19 infections amongst our EDP group. With developments in laboratory testing, we utilized a SARS-CoV-2 multi-antigen antibody assay to distinguish between serum antibodies acquired from natural infection versus those from vaccination. Specifically, nucleocapsid (N) antibodies are a unique marker for natural infection, while the Receptor-binding domain (RBD) is the target of SARS-CoV-2 vaccines [9]. Therefore, the primary objective of this study is to determine the incidence of unknown COVID-19 infection within our cohort of EDPs by analyzing their blood for natural versus vaccine acquired immunity. 2 Materials and methods This was a prospective observational study, evaluating human serum for the presence of SARS-CoV-2 antibodies and individual characteristics and demographics amongst a cohort of EDPs, and received approval from the local Institutional Review Board. This study was conducted at a level I trauma center with an emergency medicine residency and approximately 64,000 annual ED patient visits. The ED consists of 44 beds with an additional 9 patient care areas in a separate low acuity area. Subjects were considered attending emergency physicians, emergency medicine residents/fellows, and APPs. All study participants provided care to COVID-19 patients on a regular basis over the 12 months prior to study enrollment. Eligible subjects were initially invited to participate via email. Those responding with interest were met on an individual basis and underwent informed written consent. Participants underwent a 3 ml venous blood draw. In addition, subjects were asked to complete a 12-question survey including descriptive variables, vaccination status, prior PCR confirmed COVID-19 infection, known positive household members, amongst other questions displayed in Fig. 1 . Subjects were consecutively enrolled from April 21, 2022, to June 30, 2022. Samples were centrifuged for 15 min at 7000 RPM, then stored per the manufacturer's instructions until the assays were performed in July 2022.Fig. 1 Survey for emergency medicine providers undergoing SARS-CoV-2 serum antibody analysis. Fig. 1 The primary objective was to determine the incidence of unknown COVID-19 infection amongst this cohort of EDPs. Secondary analysis evaluated for significant statistical differences in the combined group with prior natural infection (known and unknown) versus those who did not, and between the known and unknown individuals previously infected with COVID-19. Variables included: age, gender, race, clinical position, vaccination status, confirmed COVID-19 in prior 12 months, presumed but unconfirmed COVID-19, household member with confirmed COVID-19 in prior 12 months, and school-age children in household. 2.1 Luminex xMAP® SARS-CoV-2 Multi-Antigen IgG Assay The Luminex xMAP® SARS-CoV-2 Multi-Antigen IgG Assay (RUO; # 30–00127) is a multiplex, microsphere-based assay that detects the presence of IgG antibodies against three different SARS-CoV-2 antigens- N, RBD, and Spike subunit 1 (S1)) from serum or plasma. Manufacturer-designated threshold values were 700 MFI for the qualitative detection of N and RBD targets and 300 MFI for the background for all lots used. Clinical specimens that are positive for both N and RBD targets are considered positive for infection with COVID-19. A positive RBD antibody result in the absence of nucleocapsid detection is consistent with an immune response to a SARS-CoV-2 vaccine. Antibody levels detected were plotted using GraphPad Prism version 9.4.1. 2.2 Primary data analysis Counts, and percentages were used to summarize the demographic and clinical characteristics of the providers. Fisher's exact test was used to look at the association of demographic and clinical characteristics with assay results dichotomized at 700 MFI and those ≥700 (n = 18). All analyses were done using SAS, Version 9.4. P < 0.05 was considered statistically significant. 3 Results Ultimately, 69 of the 81 eligible subjects (85.2%) agreed to participate. Of these, 58.0% were male, 97.1% white, and mean age was 37 years. Thirty-four (49.3%) of the subjects were emergency medicine residents/fellows, 31 (44.9%) emergency medicine attendings, and 4 (5.8%) APPs (Table 1 ). Eighteen individuals displayed an MFI ≥ 700 for N, strongly suggestive of prior natural COVID-19 infection. Eleven were unknown, providing a cumulative incidence of 17.7% (11 out of 62) in which the provider was unaware of their COVID-19 infection. Of the 11 individuals with confirmed COVID-19 infection in the prior 12 months, only 7 (63.6%) had an MFI ≥ 700 for N, demonstrating their prior infection. Fifteen individuals presumed (not confirmed) they had been infected with the SARS-CoV-2 virus within the prior 12 months, where only 4 (26.7%) displayed an MFI ≥ 700 for N. When comparing those with an MFI dichotomized at 700, there was no significant difference between age, gender, race, position, children in the home, or household member with confirmed COVID-19 infection in the prior 12 months. The only statistically significant association was in those with previously confirmed COVID-19 infection (p = 0.0049; Table 1). In comparing the combined group with an MFI ≥ 700 for N, there were no significantly distinguishing demographic or clinical characteristics (Table 2 ). All individuals displayed an MFI ≥ 700 for RBD regardless of their nucleocapsid response. This suggests that those without evidence of prior COVID-19 infection (MFI < 700 for N) were still retaining a significant level of antibodies from their prior vaccinations, with a mean time of 7 months and up to 12 months since last vaccine. Of note, all subjects had received at least two vaccinations and most a third. Fig. 2 demonstrates the varying degrees of N and RBD antibody levels (MFI) for each of the subjects enrolled. In addition, it is important to note that all subjects answered “yes” to the final question of survey, asking if they felt they had access to appropriate personal protective equipment (PPE) over the prior 12 months (Fig. 1 ). Table 1 Comparisons of patient and clinical characteristics with assay 700 status. Table 1 Nucleocapsid cut point 700 <700 (N = 51) ≥ 700 (N = 18) Total (N = 69) P-value Gender, n (%) 0.78891  Male 29 (56.9%) 11 (61.1%) 40 (58.0%)  Female 22 (43.1%) 7 (38.9%) 29 (42.0%) Race, n (%) 1.00001  White 49 (96.1%) 18 (100.0%) 67 (97.1%)  Latino/Hispanic 1 (2.0%) 0 (0.0%) 1 (1.4%)  Other 1 (2.0%) 0 (0.0%) 1 (1.4%) Position, n (%) 0.32191  PGY1 8 (15.7%) 4 (22.2%) 12 (17.4%)  PGY2 10 (19.6%) 1 (5.6%) 11 (15.9%)  PGY3 8 (15.7%) 3 (16.7%) 11 (15.9%)  Fellow 0 (0.0%) 1 (5.6%) 1 (1.4%)  APP 4 (7.8%) 0 (0.0%) 4 (5.8%)  Attending physician 21 (41.2%) 9 (50.0%) 30 (43.5%) Vaccination Status, n (%) 0.44601  two mRNA-1 1 (2.0%) 1 (5.9%) 2 (3.0%)  two mRNA + mRNA booster 49 (98.0%) 16 (94.1%) 65 (97.0%)  two mRNA + J&J booster 0 (0.0%)  J&J with J&J booster 0 (0.0%)  Missing 1 1 2 Confirmed COVID w/in 12 mths, n (%) 0.00491  Yes 4 (7.8%) 7 (38.9%) 11 (15.9%)  No 47 (92.2%) 11 (61.1%) 58 (84.1%) Presumed COVID w/in 12 mths, n (%) 0.74611  Yes 11 (21.6%) 5 (27.8%) 16 (23.2%)  No 40 (78.4%) 13 (72.2%) 53 (76.8%) Children in home, n (%) 0.78291  Yes 20 (39.2%) 8 (44.4%) 28 (40.6%)  No 31 (60.8%) 10 (55.6%) 41 (59.4%) Household member with confirmed COVID, n (%) 0.51441  Yes 10 (19.6%) 5 (27.8%) 15 (21.7%)  No 41 (80.4%) 13 (72.2%) 54 (78.3%) Appropriate Access to PPE, n (%)  Yes 51 (100.0%) 18 (100.0%) 69 (100.0%) 1 Fisher Exact p-value; Table 2 Comparisons with COVID status, Natural Infection Group. Table 2 Confirmed COVID w/in 12 mths Yes (N = 7) No (N = 11) Total (N = 18) P-value Gender, n (%) 1.00001  Male 4 (57.1%) 7 (63.6%) 11 (61.1%)  Female 3 (42.9%) 4 (36.4%) 7 (38.9%) Race, n (%)  White 7 (100.0%) 11 (100.0%) 18 (100.0%) Position, n (%) 0.20361  PGY1 0 (0.0%) 4 (36.4%) 4 (22.2%)  PGY2 1 (14.3%) 0 (0.0%) 1 (5.6%)  PGY3 2 (28.6%) 1 (9.1%) 3 (16.7%)  Fellow 0 (0.0%) 1 (9.1%) 1 (5.6%)  Attending physician 4 (57.1%) 5 (45.5%) 9 (50.0%) Vaccination Status, n (%) 0.64051  two mRNA-1 1 (14.3%) 0 (0.0%) 1 (5.6%)  two mRNA + mRNA booster 6 (85.7%) 10 (90.9%) 16 (88.9%)  J&J with J&J booster 0 1 (9.1%) 1 (5.6%) Presumed COVID w/in 12 mths, n (%) 0.59561  Yes 1 (14.3%) 4 (36.4%) 5 (27.8%)  No 6 (85.7%) 7 (63.6%) 13 (72.2%) Children in home, n (%) 1.00001  Yes 3 (42.9%) 5 (45.5%) 8 (44.4%)  No 4 (57.1%) 6 (54.5%) 10 (55.6%) Household member with confirmed COVID, n (%) 0.32601  Yes 3 (42.9%) 2 (18.2%) 5 (27.8%)  No 4 (57.1%) 9 (81.8%) 13 (72.2%) Appropriate Access to PPE, n (%)  Yes 7 (100.0%) 11 (100.0%) 18 (100.0%) 1 Fisher Exact p-value; Fig. 2 SARS-CoV-2 antibody levels in 69 Emergency Medicine Providers (EMP). Serum collected from EMPs was analyzed using a multi-antigen assay to determine the Mean Fluorescent Intensity (MFI) of Nucleocapsid (N; blue bar) and Receptor-Binding Domain (RBD; orange bar) levels. Nucleocapsid antibodies are a unique marker for COVID-19 infection, while RBD antibodies can be from COVID-19 infection and post-vaccination for SARS-CoV-2. Nucleocapsid levels (MFI) superimposed on RBD MFI detected; the dotted line (700 MFI) represents the threshold for a positive antibody response (GraphPad Prism version 9.4.1). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 2 4 Discussion The primary purpose of this study was to determine the incidence of unknown COVID-19 infection amongst a cohort of EDPs, who had provided regular care in an ED to patients infected with the SARS-CoV-2 virus. With an 85.2% participation rate, the results of our serum antibody analysis suggested that 26.1% of the subjects had previously been infected, and 17.7% of those individuals had never received a COVID-19 test that resulted positive. In addition, 24.2% (15 of 62) of our study population presumed they had recently been infected, whereas only 26.7% (4 of 15) had MFI levels suggesting they truly had. To our knowledge, this is the first study to investigate this question specifically amongst EDPs providing care to COVID-19 infected individuals post vaccination roll-outs. Early in the COVID-19 pandemic, there were several studies that evaluated for the presence of SAR-CoV-2 antibodies in HCWs, reporting a prevalence ranging from 5 to 42% [[1], [2], [3],10]. Important to note that this was a time many institutions lacked appropriate PPE. Post emergence of COVID-19 vaccinations, there were diminished studies reporting on this topic, possibly due to the lack of assays available to distinguish between antibodies from natural versus acquired immunity. At the initiation of this study, Omicron and its subsequent variants had been the predominant SARS-CoV-2 strain in our region for the prior 4–6 months at the end of enrollment. Numerous recent studies have described the increased infectivity of Omicron subvariants, suggesting three to ten times compared to Delta and its previous variants [11,12]. Additionally, Omicron subvariants have consistently demonstrated their ability to evade neutralizing antibodies in those who had received three prior mRNA vaccines, as well as those with antibodies acquired from natural infection preceding the initial Omicron wave [6,11,[13], [14], [15]]. Fortunately, lower respiratory involvement and severity of disease had decreased, resulting in fewer hospitalizations throughout the country [11,15]. Towards the end of January 2022, our institution had over 500 hospital employees on home quarantine due to acute COVID-19 infection and our system had greater than a 99% full vaccination rate, yet our number of COVID-19 units decreased from six to two within two months. Due to the previously described characteristics of the Omicron subvariants and the high degree of COVID-19 exposure experienced by EDPs, we hypothesized that the unknown incidence of infection would be higher than 17.7%. One explanation is that this institution fortunately never experienced a shortage of PPE, which was supported by our survey data. More intriguing is the numerous individuals who had measurable MFIs for the nucleocapsid but did not reach the threshold of 700 MFI to accurately report they had experienced a previous COVID-19 infection. It has been previously demonstrated that it can take up to two weeks to develop antibodies following acute infection, and >90% of them by end of one month [16,17,18]. Therefore, many of our subjects may have experienced recent unknown infection, but the duration of time between their exposure and blood draw was not sufficient to surmount a significant response. An alternative explanation is that these individuals had previously been infected, but their antibody levels had diminished below the assay threshold at the time of their serum analysis. Finally, all of our subjects had MFI levels RBD ≥ 700 regardless of reaction to the nucleocapsid. Prior studies have suggested that even following a booster, antibody levels may begin to wane as early as three and commonly by six months [15,18,19]. This is encouraging within our cohort of providers considering that the mean time since their last booster was seven months, which may have contributed to our lower-than-expected incidence of prior infection. 4.1 Limitations Although, we did enroll 85% of the eligible subjects, data from the remaining 12 EDPs could have affected our findings. Our subjects were predominantly white with a mean age of 37, therefore our findings may not extrapolate to other cohorts of EDPs. Another limitation is the self-reporting of presumed past COVID-19 and the possibility of asymptomatic infection in a healthy cohort of HCWs. Finally, during enrollment our institution was experiencing a high patient positive rate and some subjects may have a recent COVID-19 infection but had inadequate time to mount an antibody response, therefore underrepresented our reported incidence of unknown infections. 5 Conclusion This study demonstrated a less than expected incidence of unknown COVID-19 infection in a cohort of EDPs working at a large tertiary academic center. There was no significant correlation with subject characteristics, previous confirmed infection in household members, or presence of school age children in the home. However, these data are encouraging and support the utility of adequate access to PPE and SARS-CoV-2 vaccination in the reduction of COVID-19. Author contributions ANB: Study concept/design, acquisition of the data, analysis/interpretation data, drafting and revisions of manuscript. MGW: Study concept/design, analysis/interpretation data, drafting and revisions of manuscript. CEB: Study concept/design, acquisition of the data, analysis/interpretation data, revisions of manuscript. BZ: Study design, acquisitions of data, drafting and revisions of manuscript. EL: Analysis/interpretation data, statistical expertise, revisions of manuscript. TTN: Study concept/design, acquisition of the data, analysis/interpretation data, drafting and revisions of manuscript. AH: Acquisition of the data, analysis/interpretation data, revisions of manuscript. SK: Study concept/design, analysis/interpretation data, revisions of manuscript. JH: Study concept/design, analysis/interpretation data, drafting and revisions of manuscript. WGZ: Study concept/design, analysis/interpretation data, drafting and revisions of manuscript. MCW: Study concept/design, analysis/interpretation data, drafting and revisions of manuscript. Conflicts of interest and funding None. CRediT authorship contribution statement Aaron Nathan Barksdale: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Supervision, Writing – original draft, Writing – review & editing. Macy G. Wood: Formal analysis, Methodology, Resources, Writing – review & editing. Chad E. Branecki: Conceptualization, Data curation, Investigation, Writing – review & editing. Brooklin Zimmerman: Conceptualization, Data curation, Investigation, Writing – review & editing. Elizabeth Lyden: Conceptualization, Formal analysis, Writing – review & editing. Thang T. Nguyen: Conceptualization, Data curation, Methodology, Writing – review & editing. Andrew Hatfield: Data curation, Investigation, Methodology, Writing – review & editing. Scott Koepsell: Conceptualization, Methodology, Writing – review & editing. Jason Langenfeld: Conceptualization, Methodology, Writing – review & editing. Wesley G. Zeger: Conceptualization, Methodology, Writing – review & editing. Michael C. Wadman: Conceptualization, Methodology, Supervision, Writing – review & editing. ==== Refs References 1 Frisch S. Jones S. Willis J. Sinert R. COVID-19 infection and symptoms among emergency medicine residents and fellows in an urban academic hospital setting:cross-sectional questionnaire study JMIRx Med 3 1 2022 1 7 2 Murakami E. Ghatak-Roy A. Popova M. COVID-19 infection among emergency department healthcare providers in a large tertiary academic medicine center following the peak of the pandemic Am J Emerg Med 40 2021 27 31 33340874 3 Dzinamarira T. Murewanhema G. Mhango M. COVID-19 prevalence amoung healthcare workers. A systematic review and meta-analysis Int J Environ Res Public Health 19 146 2022 1 13 4 Johnson A.G. Amin A.B. Ali A.R. COVID-19 incidence and death rates among unvaccinated and fully vaccinated adults with and without booster doses during periods of delta and Omicron variant emergence-25 US jurisdictions, April 4–December 25, 2021 MMWR 71 4 Jan 2022 132 138 35085223 5 Schmidt F. Muecksch F. Weisblum Y. Silva J.D. Bednarski E. Cho A. Plasma neutralization of the SARS-CoV-2 omicron variant N ENGL J Med Feb 2022 1 3 6 Rossler A. Riepler L. Bante D. Von Laer D. Kimpel J. SARS-CoV-2 omicron variant neutralization in serum from vaccinated and convalescent persons N ENGL J Med Feb 2022 1 3 7 Mostafavi E. Dubery A.K. Teodori L. Ramakrishna S. Kauskik A. SARS-CoV-2 omicron variant: a next phase of the COVID-19 pandemic and a call to arms for system sciences and precision medicine SCIMEA Feb 3 2022 1 8 8 Juliano A.D. Brunkard J.M. Boehmer T.K. Peterson E. Adjei S. Binder A.M. Trends in disease severity and health care utilization during the early omicron variant period compared with previous SARS-CoV-2 high transmission periods--United States, December 2020-January 2022 MMWR 71 Jan 2022 1 7 9 Wheeler S.E. Shurin G.V. Yost M. Differential antibody response to mRNA COVID-19 vaccines in healthy subjects Microbiol Spectrum 9 1 2021 e00341–21 10 Sb Doernber Holubar M. Jain V. Incidence and prevalence of coronavirus disease 2019 within a healthcare worker cohort during the first year of the sever acute respiratory syndrome coronavirus 2 pandemic Clin Infect Dis 2022 11 Shretha L.B. Foster C. Rawlinson W. Tedla N. Bull R.A. Evolution of the SARS-CoV-2 omicron variants BA.1 to BA.5: implication for immune escape and transmission Rev Med Virol July 2022 e2381 12 Joung S.Y. Ebinger J.E. Sun N. Awareness of SARS-CoV-2 Omicdron variant infection among adults with recent COVID-19 Seropositivity JAMA Netw Open 5 8 2022 e2227241 13 Yao L. Ahu K.L. Jiang X.L. Omicron subvariants escape antibodies elicited by vaccination and BA.2.2 infection Lancet Infect Dis 22 8 2022 1116 1117 35738299 14 Arora P. Kempf A. Nehlmeier I. Augmented neutralization resistance of emerging omicrons subvariants BA.2.12.1, BA.4, and BA.5 Lancet Infect Dis 22 8 2022 1116 1117 35738299 15 Vietri M.T. D'Elia G. Caliendo G. Antibody levels after BNT162b2 vaccine booster and SARS-CoV-2 omicron infection Vaccine 2022 16 Hamady A. Lee J. Loboda Z.A. Waning antibody response in COVID-19: what can we learn from the analysis of other coronaviruses? Infection 50 2022 11 25 34324165 17 Subissi L. Mulders M. Friede M. Van Kerkhove M. Perkins M. COVID-19 natural immunity: Scientific brief 2021 WHO 18 Cordioli M. Mirandola M. Gios L. COVID-19 seroprevalence amongst healthcare workers: potential biases in estimating infection prevalence Epidemiol Infect 150 2022 19 Lau C.S. Phua S.K. Liang Y.L. Helen M.L. Aw T.C. SARS-CoV-2 spike and neutralizing antibody kinetics 90 days after three doses of BNT162b2 mRNA COVID-19 vaccine in Singapore Vaccines 10 2 2022 331 35214789

0

PMC9749378

NO-CC CODE

2022-12-15 23:23:21

no

Am J Emerg Med. 2022 Dec 14; doi: 10.1016/j.ajem.2022.12.012

utf-8

Am J Emerg Med

10.1016/j.ajem.2022.12.012

oa_other

==== Front Prog Cardiovasc Dis Prog Cardiovasc Dis Progress in Cardiovascular Diseases 0033-0620 1873-1740 Elsevier Inc. S0033-0620(22)00154-2 10.1016/j.pcad.2022.12.002 Article Impact of COVID-19 in patients hospitalized with stress cardiomyopathy: A nationwide analysis Hajra Adrija a Malik Aaqib b Bandyopadhyay Dhrubajyoti b⁎ Goel Akshay b Isath Ameesh b Gupta Rahul c Krishnan Suraj d Rai Devesh e Krittanawong Chayakrit f Virani Salim S. g Fonarow Gregg C. h Lavie Carl J. i a Department of Internal Medicine, Montefiore Medical Center/Albert Einstein College of Medicine, Bronx, NY, USA b Department of Cardiology, Westchester Medical Center, New York Medical College, Valhalla, NY, USA c Department of Cardiology, Lehigh Valley Heart Institute, Lehigh Valley Health Network, Allentown, PA, USA d Department of Internal Medicine, Jacobi Medical Center/Albert Einstein College of Medicine, Bronx, NY, USA e Department of Cardiology, Sands-Constellation Heart Institute, Rochester Regional Health, Rochester, NY, USA f Department of Cardiology, NYU Langone Health, New York, USA g Michael E. DeBakey Veterans Affairs Medical Center, Section of Cardiovascular Research, Baylor College of Medicine, Houston, TX, USA h Ahmanson-UCLA Cardiomyopathy Center, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA., USA i John Ochsner Heart and Vascular Institute, Ochsner Clinical School, The University of Queensland School of Medicine, New Orleans, LA, USA ⁎ Corresponding author. 14 12 2022 14 12 2022 © 2022 Elsevier Inc. All rights reserved. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Stress cardiomyopathy was noted to occur at a higher incidence during coronavirus disease of 2019 (COVID-19) pandemic. This database analysis has been done to compare the in-hospital outcomes in patients with stress cardiomyopathy and concurrent COVID-19 infection with those without COVID-19 infection. The National Inpatient Sample database for the year 2020 was queried to identify all admissions diagnosed with stress cardiomyopathy. These patients were then stratified based on whether they had concomitant COVID-19 infection or not. A 1:1 propensity score matching was performed. Multivariate logistic regression analysis was done to identify predictors of mortality. We identified 41,290 hospitalizations for stress cardiomyopathy, including 1665 patients with concurrent diagnosis of COVID-19. The female preponderance was significantly lower in patients with stress cardiomyopathy and COVID-19. Patients with concomitant COVID-19 were more likely to be African American, diabetic and have chronic kidney disease. After propensity matching, the incidence of complications, including acute kidney injury (AKI), AKI requiring dialysis, coagulopathy, sepsis, cardiogenic shock, cases with prolonged intubation of >24 h, requirement of vasopressor and inpatient mortality, were noted to be significantly higher in patients with COVID-19. Concomitant COVID-19 infection was independently associated with worse outcomes and increased mortality in patients hospitalized with stress cardiomyopathy. Keywords Covid Acute coronary syndrome Stress cardiomyopathy Congestive heart failure ==== Body pmcList of abbreviations Unlabelled TableAKI Acute kidney injury aOR Adjusted odds ratio COVID 19 Coronavirus disease of 2019 CV Cardiovascular CKD Chronic kidney disease COPD Chronic obstructive pulmonary disease CABG Coronary artery bypass graft CI Confidence interval CHF Congestive heart failure DVT Deep vein thrombosis ECMO Extracorporeal membrane oxygenation HCUP Healthcare Cost and Utilization Project HMO Health Maintenance Organization HD Hemodialysis HTN Hypertension ICD-10-CM International Classification of Diseases, Tenth Revision, Clinical Modification IABP Intra-aortic balloon pump IFNγ Interferon-gamma IL1B Interleukin 1B IQR Interquartile range LOS Length of hospital stay MI Myocardial infarction NIS Nationwide Inpatient Sample PE Pulmonary embolism PCI Percutaneous coronary intervention SNF/NH/IC Skilled nursing facility/nursing home/ intermediate care TNFα Tissue necrosis factor alpha USD United States dollar UTI Urinary tract infection VT Ventricular tachycardia VF Ventricular fibrillation Introduction The coronavirus disease of 2019 (COVID-19) pandemic has various cardiovascular (CV) manifestations, including myocardial injury, arrhythmias, cardiac arrests, heart failure, and coagulation abnormality.1 , 2 Cases of stress cardiomyopathy have also been reported in COVID-19 patients. The incidence of stress cardiomyopathy has drawn attention among clinicians for its significant effects on patient management and treatment.3 Since its clinical discovery, the pathogenesis of stress cardiomyopathy remains unclear. Systemic viral illnesses, including influenza, have been noted to be associated with stress cardiomyopathy, and cases have been reported in COVID-19-affected individuals, particularly patients with severe disease.4 But data are sparse regarding the baseline characteristics, risk factors associated with inpatient morbidity, and mortality in stress cardiomyopathy patients affected with COVID-19. We have conducted a population-based analysis using a large nationally representative database to compare the characteristics and outcomes of adult patients hospitalized with stress cardiomyopathy with and without concomitant COVID-19 in the United States (US). We also aimed to determine the clinical predictors of adverse outcomes in stress cardiomyopathy patients with COVID-19. Methods Data source The Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample (NIS) database is the largest all-payer inpatient dataset in the US and is available publicly. The NIS represents 95% of US hospitalizations from 44 states participating in HCUP and provides a stratified sample of 20% of discharges, including up to 8 million hospital discharges per year. The NIS database has been previously demonstrated to correlate well with other discharge databases in the US. In addition, it has been validated in various studies to provide reliable estimates of admissions within the US.5 Study population We included hospitalizations with a diagnosis of stress cardiomyopathy based on the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) code (I51. 81) which has positive predictive value of 98% and identified patients with and without a concurrent diagnosis of COVID-19 based on ICD-10-CM code U07.1.6 Outcomes The primary outcome of interest was in-hospital mortality in patients with stress cardiomyopathy with concurrent COVID-19 infection compared with those with stress cardiomyopathy without concurrent COVID-19 infection. Secondary outcomes included AKI as well as, AKI requiring dialysis, acute respiratory failure as well as, respiratory failure requiring intubation, need for mechanical circulatory support such as intra-aortic balloon pump (IABP) (and/or impella), extracorporeal membrane oxygenation (ECMO), length of stay (LOS) and hospitalization costs. Statistical analysis Statistical analyses were performed using Stata 16.0 (StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC). The discharge weights provided by the Agency for Healthcare Research and Quality were applied to obtain weighted numbers to calculate national estimates. A 1:1 propensity score matching was performed to compare outcomes for patients with concomitant stress cardiomyopathy and COVID-19 and patients with stress cardiomyopathy without concomitant COVID-19 using a propensity score calculated based on a multivariable logistic regression model. Propensity score matching without a replacement was performed in a 1:1 nearest-neighbor fashion with a caliper width of 0.1 of the estimated propensity scores. Multivariate logistic regression models were generated to identify the independent predictors and were reported as adjusted odds ratio (aOR) with 95% confidence interval (CI). Categorical variables were expressed as percentages. Continuous variables were expressed as median and interquartile range. Categorical variables were compared using the Pearson chi-square test, while continuous variables were compared using the student's t-test. All reported P values are 2-sided, with a value of <0.05 considered significant. Results A total of 41,290 hospitalizations for stress cardiomyopathy were identified, of which 1665 patients (4%) had a concurrent diagnosis of COVID-19. Table 1 describes the baseline characteristics of patients admitted with COVID-19. Stress cardiomyopathy patients with COVID-19 had a mean age of 71 years compared to a mean age of 68 years in stress cardiomyopathy patients without COVID-19. Approximately, 12.3% of patients in the COVID-19 group were African American vs. 8.2% of patients without COVID-19. Before propensity matching, stress cardiomyopathy hospitalizations with COVID-19 had higher prevalence of diabetes mellitus (39.0 vs. 24.5%, p-value <0.001), chronic kidney disease (CKD) (23.1% vs. 15%, p-value <0.001), coagulopathy (22.2% vs. 9.5%, p-value <0.001). Around 5.4% of stress cardiomyopathy patients with COVID-19 had a smoking history (18.6% in patients without COVID-19, p-value <0.001). Approximately, 44.2% of stress cardiomyopathy patients with COVID-19 had a history of coronary artery disease (31.2% in patients without COVID-19, p-value <0.001). The cost of hospitalization was higher, and the LOS was longer in patients with COVID-19, with statistical significance. Table 2 shows the complications and outcomes of stress cardiomyopathy patients with and without COVID-19 infection. Inpatient mortality was higher in COVID-19 affected patients than patients without COVID-19 (33.9% vs. 7.3%, p-value <0.001). Incidence of acute kidney injury (AKI) (48.1% vs. 25.2%) and AKI leading to hemodialysis (6.6% vs. 2.2%), myocarditis (5.7% vs. 0.4%), respiratory failure with intubation of >24 h (25.5% vs. 15.8%), cardiogenic shock (16.5% vs. 7.0%), requirement of vasopressors (12.3% vs. 5.6%), sepsis (44.4% vs. 18.7%) is more in patients with COVID-19 compared to patients without COVID-19 with p-value of <0.001. (Table 2).Table 1 Baseline characteristics of patients with COVID-19 and stress cardiomyopathy before and after propensity match with complications of hospitalized patients. Table 1Characteristics Before Matching After Matching Without COVID-19 With COVID-19 P Value Without COVID-19 With COVID-19 P Value 39,625 1665 1620 1620 Age, median IQR, years 68 (58–77) 71 (61–78) 0.0391 72 (61–81) 71 (61–78) 0.1094 Age groups 18–59 10,785 380.00 0.1057 345 370 0.0300 60–69 10,355 405.00 375 390 70–79 10,750 540.00 415 525 >79 years 7735 340.00 485 335 Female 32,040 1035 <0.001 1085 1030 0.3584 Caucasian race 3250 205 0.0113 255 200 0.2017 African American race 2425 325 <0.001 305 295 0.8303 Hispanics 2015 125 0.0552 130 115 0.6599 Atrial fibrillation 6535 330 0.1058 340 315 0.5984 Diabetes mellitus 9690 650 <0.001 660 620 0.4938 Hypertension 25,955 1110 0.6534 1070 1075 0.9297 Chronic kidney disease 5930 385 <0.001 390 375 0.7819 CHF 18,590 830 0.2811 845 800 0.4582 Peripheral vascular disease 3375 110 0.2317 110 110 0.9999 Dementia 2310 190 <0.001 220 185 0.3744 COPD 12,480 525 0.9891 480 515 0.5495 Valvular heart disease 1035 145 0.3796 150 145 0.8847 Arrhythmias 14,645 640 0.5790 675 625 0.4128 Liver disease 3520 190 0.1114 205 180 0.5403 Hypothyroidism 7205 245 0.1059 280 240 0.3767 Anemia 2155 115 0.2572 165 110 0.0947 Cancer 3665 75 0.002 60 75 0.5651 Rheumatological disorders 2195 70 0.2954 55 70 0.5174 Weight loss 5910 250 0.9606 280 230 0.2729 Coagulopathy 3770 370 <0.001 300 340 0.4387 Obesity 5155 345 0.0001 260 335 0.1122 Smoking history 7385 90 <0.001 95 90 0.8610 Coronary artery disease 17,510 520 <0.001 540 520 0.7393 Prior stroke 3870 180 0.5245 205 180 0.5230 Prior PCI 1915 80 0.9810 55 80 0.3277 Prior CABG 600 25 0.9850 15 25 0.4742 Alcohol 2420 20 0.0002 20 20 0.9999 Prior MI 4615 135 0.0537 200 135 0.0868 Discharge Routine 20,380 425 <0.001 655 420 <0.001 SNF/NH/IC 7515 390 380 380 Home healthcare 7375 220 280 210 Length of stay, median (IQR), days 4 (2–8) 8 (4–16) <0.001 5 (2–9) 8 (4–16) <0.001 Weekend admission 9560 470 0.0862 330 460 0.0196 Elective admission 2540 35 0.0049 85 35 0.0468 Hospital location and teaching status Prior CABG 600 25 0.9850 15 25 0.4742 Alcohol 2420 20 0.0002 20 20 0.9999 Prior MI 4615 135 0.0537 200 135 0.0868 Discharge Routine 20,380 425 <0.001 655 420 <0.001 SNF/NH/IC 7515 390 380 380 Home healthcare 7375 220 280 210 Length of stay, median (IQR), days 4 (2–8) 8 (4–16) <0.001 5 (2–9) 8 (4–16) <0.001 Weekend admission 9560 470 0.0862 330 460 0.0196 Elective admission 2540 35 0.0049 85 35 0.0468 Hospital location and teaching status Prior CABG 600 25 0.9850 15 25 0.4742 Alcohol 2420 20 0.0002 20 20 0.9999 Prior MI 4615 135 0.0537 200 135 0.0868 Discharge Routine 20,380 425 <0.001 655 420 <0.001 SNF/NH/IC 7515 390 380 380 Home healthcare 7375 220 280 210 Length of stay, median (IQR), days 4 (2–8) 8 (4–16) <0.001 5 (2–9) 8 (4–16) <0.001 Weekend admission 9560 470 0.0862 330 460 0.0196 Elective admission 2540 35 0.0049 85 35 0.0468 Hospital location and teaching status Rural 2380 105 0.3293 100 100 0.9599 Urban non-teaching 6470 220 230 220 Urban teaching 30,775 1340 1290 1300 Hospital region Northeast 8265 380 0.0235 320 375 0.2712 Midwest 9235 490 570 470 South 13,535 450 385 445 West 8590 345 345 330 Insurance Medicare 24,610 1040 0.1626 1070 1015 0.1769 Medicaid 4745 175 220 175 Private including HMO 8090 340 260 325 Self-pay 1175 40 30 40 Median household income (%) 0-25th percentile 9830 415 0.6795 405 405 0.8133 26-50th percentile 10,510 475 445 450 51-75th percentile 9770 420 445 415 76-100th percentile 8775 325 275 320 Total hospital cost USD median IQR 14,892 (9152–28,180) 25,887 (11438–56,742) <0.001 18,753 (9396–34,557) 25,864 (11206–56,327) 0.0013 Hospital bed size Small 7515 315 0.4478 270 295 0.4633 Intermediate 10,770 400 340 390 Large 21,340 950 1010 935 CABG- coronary artery bypass graft, CHF- congestive heart failure, COVID 19- coronavirus disease 2019, CKD- chronic kidney disease, COPD- chronic obstructive pulmonary disease, HMO- health maintenance organization, MI- myocardial infarction, PCI- percutaneous coronary intervention, SNF/NH/IC- skilled nursing facility/nursing home/ intermediate care. Table 2 Complication of hospitalized patients with Takotsubo cardiomyopathy with or without COVID-19. Table 2Complications Before Matching After Matching Without COVID-19 With COVID-19 P Value Without COVID-19 With COVID-19 P Value AKI 9990 800 <0.001 595 780 0.0042 AKI leading to HD 855 110 <0.001 100 105 0.8667 UTI 4605 235 0.1529 250 225 0.5556 Sepsis 7425 740 <0.001 465 715 <0.001 DVT 910 85 0.0015 65 80 0.5769 PE 735 50 0.1369 45 50 0.8069 Stroke in-hospital 1075 25 0.1786 60 25 0.0678 Cardiogenic shock 2785 275 <0.001 205 265 0.1772 Cardiac arrest 1700 145 0.0002 80 145 0.0364 VT 2340 90 0.6967 110 90 0.5011 VF 765 15 0.1775 25 15 0.4805 Bleeding requiring transfusion 2025 145 0.0029 120 130 0.7442 Death 2895 565 <0.001 220 545 <0.001 Vasopressors 2205 205 <0.001 130 200 0.0572 Prolonged intubations >24 h 5710 625 <0.001 345 600 <0.001 Respiratory failure 13,910 945 <0.001 605 930 <0.001 Resp failure with intubation >24 h 6265 425 <0.001 365 420 <0.001 ECMO utilization 80 5 0.6986 5 5 0.999 Impella 165 10 0.6131 0 10 0.1572 IABP 250 15 0.5480 20 15 0.7055 CABG 100 0 0.3576 NA NA NA PCI 1140 30 0.2459 35 30 0.7639 Tamponade 105 10 0.2577 5 10 0.5621 Acute heart failure 1335 5 0.0020 40 5 0.0185 Myocarditis 155 95 <0.001 5 90 0.0185 HTN crises 1705 25 0.0122 45 25 0.2767 AKI- acute kidney injury, COVID 19- coronavirus disease 2019, CABG- coronary artery bypass graft, DVT- deep vein thrombosis, ECMO- extracorporeal membrane oxygenation, HD- hemodialysis, HTN- hypertension, IABP- intra-aortic balloon pump, MI- myocardial infarction, PCI- percutaneous coronary intervention, PE- pulmonary embolism, UTI- urinary tract infection, VT- ventricular tachycardia, VF- ventricular fibrillation, Propensity score matching was performed to create a more balanced population, with 1620 hospitalizations in each group. In a propensity score-matched population, stress cardiomyopathy patients with COVID-19 had a higher incidence of in-hospital mortality than stress cardiomyopathy patients without COVID-19 (33.6% vs.13.6%, respectively; aOR 3.22, CI 2.19–4.72, p < 0.001). In addition, sepsis, respiratory failure, respiratory failure requiring prolonged intubation for >24 h, cases with prolonged intubation and length of stay were significantly higher in stress cardiomyopathy patients with COVID-19, even after propensity match. There was an increase in the number of stress cardiomyopathy hospitalizations with concomitant COVID-19 throughout the year. The risk of in-hospital mortality was highest for those admitted earlier in the year, and decreased after the initial months, with another peak in the later part of the year (Fig. 1 ).Fig. 1 Incidence by month, mortality by month and percentage died of total admissions with COVID-19. COVID 19- coronavirus disease of 2019. Fig. 1 Predictors of mortality On multivariable regression analysis, COVID-19 was found to be independently associated with mortality in patients admitted with stress cardiomyopathy (aOR 6.10, 95% CI 4.62–8.05, p-value <0.001). Additionally, arrhythmias (aOR 1.56, CI 1.33–1.84, p-value <0.001), coagulopathy (aOR 2.55, 95% CI 2.06–3.15, p-value <0.001), and liver disease (aOR 2.59, 95% CI 2.086–3.23, p-value <0.001) were found to be independently associated with increased odds of mortality in stress cardiomyopathy patients with concurrent COVID-19 infection (Table 3 ).Table 3 Predictors of mortality after multivariate analysis. Table 3Variable Odds Ratio Lower Limit Upper Limit P Value COVID 19 6.102 4.624 8.054 <0.001 Age 0.659 0.545 0.798 <0.001 CKD 1.237 1.000 1.529 0.0500 Coagulopathy 2.548 2.062 3.149 <0.001 Weight loss 1.435 1.157 1.778 0.0010 Arrhythmias 1.566 1.332 1.841 <0.001 Liver disease 2.597 2.086 3.233 <0.001 Cancer 2.117 1.678 2.67 COVID 19- coronavirus disease 2019, CKD- chronic kidney disease. Discussion To the best of our knowledge, this is the first analysis of nationwide data to report the characteristics and outcomes of patients with stress cardiomyopathy and concomitant COVID-19 infection. Stress cardiomyopathy is known to be more common in female patients.7 Interestingly, we found a lower number of female patients with stress cardiomyopathy in the COVID-19 affected group. Stress cardiomyopathy, previously known as Takotsubo cardiomyopathy, is caused by intense emotional or physical stress leading to rapid deterioration of cardiac function.8 , 9 Various possible mechanisms like sympathetic nervous system stimulation, estrogen deficiency, and excess deposition of extracellular matrix have been found to contribute to the pathogenesis of stress cardiomyopathy. COVID-19, complicated with multiorgan failure, shock, and profound hypoxia with adult respiratory distress syndrome, is hypothesized to trigger stress cardiomyopathy due to catecholamine surge. Direct myocardial injury in COVID-19 is also postulated to contribute to the pathogenesis of stress cardiomyopathy.10 Patients with COVID-19 have elevated levels of proinflammatory cytokines such as interleukin 1B (IL1B), interferon-gamma (IFNγ), and tumor necrosis factor alpha (TNFα). The cytokine storm, and physical and chemical stressors with postmenopausal status, could also contribute to the development of stress cardiomyopathy.4 , 11 , 12 A recent study by Zuin et al. showed an increased incidence of stress cardiomyopathy during the pandemic compared to control groups.13 Generalized increases in psychological distress, cytokine storm, increased sympathetic responses in patients with COVID-19, and microvascular dysfunction may result in this increased incidence.13 , 14 In our study, stress cardiomyopathy patients with COVID-19 had an increased incidence of respiratory failure requiring intubation, indicating that physiological stress is associated with worse outcomes in stress cardiomyopathy patients. In this large, national, propensity-matched analysis, we found COVID-19 to be an independent predictor of in-hospital mortality, with a higher rate of adverse clinical outcomes and increased healthcare resource utilization in patients hospitalized with stress cardiomyopathy. In our study, concomitant COVID-19 infection resulted in approximately a two and a half times higher mortality in patients with stress cardiomyopathy. In our study, the inpatient mortality of stress cardiomyopathy patients with concurrent COVID-19 infection was 33.64%. Recent studies have shown that, in general, stress cardiomyopathy has in-hospital mortality of 3.5–10.6%, comparable to that of acute coronary syndromes.15 Undoubtedly, our study has highlighted the significantly worse prognosis of stress cardiomyopathy in the setting of COVID-19 infection. Studies have also shown an association between the severity of COVID-19 infection with various complications, including AKI, sepsis, and organ failure.16 , 17 In our study, the incidence of complications including AKI, AKI requiring dialysis, respiratory failure with intubation of >24 h, cardiogenic shock, the requirement of vasopressors, and sepsis was higher in patients with stress cardiomyopathy and COVID-19 infection. These findings indicate the possible association of stress cardiomyopathy with the severity of COVID-19 infection. Studies have shown that patients with COVID-19 and stress cardiomyopathy have a higher incidence of CV risk factors, including diabetes, and an increased coagulopathy risk, as our study found.18 A study comparing patients with stress cardiomyopathy before the COVID-19 pandemic and during the pandemic showed increased LOS for affected patients with statistical significance, as noted in our study. This increased burden on the healthcare system is a matter of concern, and clinicians should be aware of this.19 Our findings validate that COVID-19 is associated with significantly increased morbidity and mortality in patients with stress cardiomyopathy. The findings of this analysis will help clinicians to be aware of the importance of early suspicion of deterioration of patients with concurrent stress cardiomyopathy and COVID-19. Early detection and aggressive management may change the outcome of patients suffering both an acute CV condition and viral infection. Limitations Our study has its inherent limitations. Firstly, it is a retrospective database analysis based on discharge diagnoses. We also do not have access to patient-level information and, thus, unmeasured confounding may affect these findings. The treatment guidelines and vaccination policy for COVID-19 have emerged with time. We do not have the option to find out if any management guideline or vaccination status would have changed the outcome of patients included in this analysis. Cases are being reported with stress cardiomyopathy after COVID-19 treatment.20 Studies have also shown an association between stress cardiomyopathy with COVID-19 vaccination.21 Also, one case of stress cardiomyopathy with a history of COVID-19 infection has been reported recently.22 More studies are required to understand the disease process better so that preventive measures can be taken in the future. Despite these limitations, NIS is a well-validated representation of the US population and with internal and external quality control measures. The large sample size of NIS data also compensates for residual confounders. Conclusion COVID-19 infection among patients hospitalized with stress cardiomyopathy is associated with significantly higher in-hospital mortality, adverse clinical outcomes, and use of in-hospital resources. In addition, advanced age, arrhythmia, liver disease, and coagulopathy were independent predictors of mortality in patients with stress cardiomyopathy hospitalized with concomitant COVID-19. Author access to data Publicly available National Inpatient Sample of the US. Funding No external funding was used in the preparation of this manuscript. Ethical approval of studies and informed consent Not applicable as it is a retrospective analysis of data. CRediT authorship contribution statement Adrija Hajra: Writing – original draft. Aaqib Malik: Conceptualization, Methodology, Data curation. Dhrubajyoti Bandyopadhyay: Conceptualization, Methodology, Data curation. Akshay Goel: Conceptualization, Methodology, Data curation. Ameesh Isath: Conceptualization, Methodology, Data curation. Rahul Gupta: Writing – original draft. Devesh Rai: Writing – original draft. Salim S. Virani: Writing – review & editing, Supervision. Gregg C. Fonarow: Writing – review & editing, Supervision. Carl J. Lavie: Writing – review & editing, Supervision. Declaration of Competing Interest Dr. Fonarow has served as a consultant for Abbott, Amgen, Bayer, Janssen, Medtronic, and Novartis. Dr. Virani discloses the following relationships: Research support: Department of Veterans Affairs, World Heart Federation, Tahir and Jooma Family Honorarium: American College of Cardiology (Associate Editor for Innovations, acc.org). Others: No conflicts of interest. Acknowledgment None. ==== Refs References 1. Kwenandar F. Japar K.V. Damay V. Coronavirus disease 2019 and cardiovascular system: a narrative review Int J Cardiol Heart Vasc 29 2020 Jun 3 100557 10.1016/j.ijcha.2020.100557 PMID: 32550259; PMCID: PMC7266760 2. Isath A. Malik A.H. Goel A. Gupta R. Srivastav R. Bandyopadhyay D. Nationwide analysis of the outcomes and mortality of hospitalized COVID-19 patients Curr Probl Cardiol 2022 Oct 7 101440 10.1016/j.cpcardiol.2022.101440 [Epub ahead of print. PMID: 36216202; PMCID: PMC9546497] 3. van Osch D. Asselbergs F.W. Teske A.J. Takotsubo cardiomyopathy in COVID-19: a case report. Haemodynamic and therapeutic considerations Eur Heart J Case Rep 4 FI1 2020 Aug 27 1 6 10.1093/ehjcr/ytaa271 [PMID: 33437922; PMCID: PMC7528942] 4. Gomez J.M. Nair G. Nanavaty P. Rao A. Marinescu K. Suboc T. COVID-19-associated takotsubo cardiomyopathy BMJ Case Reports CP 13 12 2020 Dec 1 e236811 5. Bandyopadhyay D. Devanabanda A.R. Hajra A. Impact of pulmonary hypertension in patients undergoing atrial fibrillation ablation: a nationwide study Int J Cardiol Heart Vasc 23 2019 Mar 29 100348 10.1016/j.ijcha.2019.100348 PMID: 30976653; PMCID: PMC6441786 6. Bhat A.G. White K. Gobeil K. Lagu T. Lindenauer P.K. Pack Q.R. Utility of ICD codes for stress cardiomyopathy in hospital administrative databases: what do they signify? J Hosp Med 14 3 2019 Dec 23 E1 E4 10.12788/jhm.3344 [Epub ahead of print. PMID: 31869294; PMCID: PMC7064300] 7. Schneider B. Sechtem U. Influence of age and gender in Takotsubo syndrome Heart Fail Clin 12 4 2016 Oct 521 530 10.1016/j.hfc.2016.06.001 [Epub 2016 Aug 11. PMID: 27638022] 27638022 8. Ramaraj R. Stress cardiomyopathy: aetiology and management Postgrad Med J 83 982 2007 Aug 543 546 10.1136/pgmj.2007.058776 PMID: 17675548; PMCID: PMC2600114 17675548 9. O'Keefe E.L. Torres-Acosta N. O'Keefe J.H. Sturgess J.E. Lavie C.J. Bybee K.A. Takotsubo syndrome: Cardiotoxic stress in the COVID Era Mayo Clin Proc Innov Qual Outcomes 4 6 2020 Dec 775 785 10.1016/j.mayocpiqo.2020.08.008 [Epub 2020 Nov 30. PMID: 33283161; PMCID: PMC7704068] 33283161 10. Touyz R.M. Boyd M.O.E. Guzik T. Cardiovascular and renal risk factors and complications associated with COVID-19 CJC Open 3 10 2021 Oct 1257 1272 10.1016/j.cjco.2021.05.020 [Epub 2021 Jun 16. PMID: 34151246; PMCID: PMC8205551] 34151246 11. Angelini P. Postalian A. Hernandez-Vila E. Uribe C. Costello B. COVID-19 and the heart: could transient Takotsubo cardiomyopathy be related to the pandemic by incidence and mechanisms? Front Cardiovasc Med 2022 9 12. Singh T. Khan H. Gamble D.T. Scally C. Newby D.E. Dawson D. Takotsubo syndrome: pathophysiology, emerging concepts, and clinical implications Circulation 145 13 2022 Mar 29 1002 1019 10.1161/CIRCULATIONAHA.121.055854 [Epub 2022 Mar 28. Erratum in: Circulation. 2022 May 17;145(20):e1053. PMID: 35344411; PMCID: PMC7612566] 35344411 13. Zuin M. Mugnai G. Anselmi M. Takotsubo syndrome during COVID-19 pandemic in the Veneto region, Italy Viruses. 14 9 2022 Sep 6 1971 36146778 14. Dubey S. Biswas P. Ghosh R. Psychosocial impact of COVID-19 Diabetes Metab Syndr 14 5 2020 Sep-Oct 779 788 10.1016/j.dsx.2020.05.035 [Epub 2020 May 27. PMID: 32526627; PMCID: PMC7255207] 32526627 15. Okura H. Update of takotsubo syndrome in the era of COVID-19 J Cardiol 77 4 2021 Apr 1 361 369 33148469 16. Trifi A. Abdellatif S. Masseoudi Y. COVID-19-induced acute kidney injury in critically ill patients: epidemiology, risk factors, and outcome Acute Crit Care 36 4 2021 Nov 308 316 10.4266/acc.2021.00934 [Epub 2021 Nov 22. PMID: 35263826; PMCID: PMC8907460] 35263826 17. Lalueza A. Lora-Tamayo J. de la Calle C. The early use of sepsis scores to predict respiratory failure and mortality in non-ICU patients with COVID-19 Rev Clin Esp (Barc) 222 5 2022 May 293 298 10.1016/j.rceng.2020.10.004 [Epub 2021 Feb 17. PMID: 35512908; PMCID: PMC7888251] 35512908 18. Singh S. Desai R. Gandhi Z. Takotsubo syndrome in patients with COVID-19: a systematic review of published cases SN Compr Clin Med 2 11 2020 2102 2108 10.1007/s42399-020-00557-w [Epub 2020 Oct 6. PMID: 33043251; PMCID: PMC7538054] 33043251 19. Jabri A. Kalra A. Kumar A. Incidence of stress cardiomyopathy during the coronavirus disease 2019 pandemic JAMA Netw Open 3 7 2020 Jul 1 e2014780 10.1001/jamanetworkopen.2020.14780 20. Salehin S. Abu Jazar D. Hasan S.M. Al-Sudani H. Raja M.W. Reverse Takotsubo cardiomyopathy after Casirivimab-Imdevimab therapy in a patient with COVID-19: a case report Am J Case Rep 23 2022 Sep 22 e936886 10.12659/AJCR.936886 [PMID: 36131520] 21. Khalid Ahmed S. Gamal Mohamed M. Abdulrahman Essa R. Abdelaziz Ahmed Rashad Dabou E. Omar Abdulqadir S. Muhammad Omar R. Global reports of takotsubo (stress) cardiomyopathy following COVID-19 vaccination: a systematic review and meta-analysis Int J Cardiol Heart Vasc 43 2022 Dec 101108 10.1016/j.ijcha.2022.101108 [Epub 2022 Aug 17. PMID: 35992364; PMCID: PMC9381427] 22. Clark D.E. Dendy J.M. Li D.L. Cardiovascular magnetic resonance evaluation of soldiers after recovery from symptomatic SARS-CoV-2 infection: a case-control study of cardiovascular post-acute sequelae of SARS-CoV-2 infection (CV PASC) J Cardiovasc Magn Reson 23 1 2021 Oct 7 106 10.1186/s12968-021-00798-1 [PMID: 34620179; PMCID: PMC8495668] 34620179

0

PMC9749379

NO-CC CODE

2022-12-15 23:22:54

no

Prog Cardiovasc Dis. 2022 Dec 14; doi: 10.1016/j.pcad.2022.12.002

utf-8

Prog Cardiovasc Dis

10.1016/j.pcad.2022.12.002

oa_other

==== Front Asia Pac J Oncol Nurs Asia Pac J Oncol Nurs Asia-Pacific Journal of Oncology Nursing 2347-5625 2349-6673 The Author(s). Published by Elsevier Inc. on behalf of Asian Oncology Nursing Society. S2347-5625(22)00237-2 10.1016/j.apjon.2022.100179 100179 Original Article A qualitative study about colorectal cancer patients and spousal caregivers’ experience and needs during COVID-19: implications for self-efficacy intervention Gong Jiali a Chen Meizhen a Cao Qian a Lin Yi a Loke Alice Yuen b Li Qiuping ac∗ a Wuxi School of Medicine, Jiangnan University, Wuxi, Jiangsu Province, China b School of Nursing, The Hong Kong Polytechnic University, Hung Hom, Kowloon Hong Kong, China c Affiliated Hospital, Jiangnan University, Wuxi, Jiangsu Province, China ∗ Corresponding author. Wuxi School of Medicine, Jiangnan University, Wuxi, Jiangsu Province, China.PhD, FAAN. 14 12 2022 14 12 2022 10017913 9 2022 7 12 2022 8 12 2022 © 2022 The Author(s) 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objective This study aims to understand the experiences of colorectal cancer (CRC) patients and their spousal caregivers during the COVID-19 pandemic, and to refine a self-efficacy (SE) intervention for these couples. Methods A descriptive phenomenological approach was used in this study. Data were collected from 11 CRC couples. All interviews were recorded, transcribed, and analysed using the Colaizzi strategy. Results Three themes and eight subthemes emerged: (a) Get and contribute support, (b) Life’s challenges, and (c) The Journey of Reconstruction. The CRC couples encountered escalating challenges in coping with cancer during COVID-19. At the same time, they have received considerable support and developed confidence in rebuilding themselves in the process. Healthcare providers are advised to focus on giving appropriate support to CRC couples, so they can go further. Conclusions This study gave insights to healthcare providers on the experiences of CRC couples and the development of SE intervention programme to support these couples: (a) initially providing caregiving training for spousal caregivers and psychological support for patients, (b) encouraging self-care for CRC couples in the middle stage, (c) guiding them to view life positively in the later stage, and (d) assessing their situation in time to identify their needs and to provide support. Healthcare providers are recommended to increase flexibility in the SE intervention programme delivery format, to reduce the impact of COVID-19 on CRC couples. Keywords Colorectal cancer COVID-19 pandemic qualitative study self-efficacy intervention spousal caregivers ==== Body pmcCredit Author Statement Jiali Gong: Conceptualization, Methodology, Investigation, Writing - Original Draft. Meizhen Chen: Investigation, Writing - Original Draft. Qian Cao: Investigation, Writing - Original Draft. Qiuping Li: Writing - Review & Editing, Supervision. All authors read the final draft of this manuscript and approve its submission for publication. Introduction Since January 2020, the COVID-19 pandemic has spread across the world. The Chinese government has developed a series of measures to control the spread of the virus, such as restricting unnecessary travel, reducing gatherings, and even locking down communities or cities if necessary.1 According to statistics released by the National Health Commission of the People’s Republic of China, there has been a cumulative total of 232,109 confirmed COVID-19 cases nationwide, of which 223,987 have been cured, 5,226 have died, and nearly 3,000 are still under treatment.2 The disease and death rate has dropped from 4.19% in the early days to 2.25% now.3 Although the spread of COVID-19 has been effectively controlled, it still affects people’s lives, particularly hurting vulnerable groups, such as cancer patients and caregivers. Recent evidence suggests that the negative impacts of COVID-19 on cancer patients and caregivers are manifold and may be long-term.4, 5, 6 Colorectal cancer (CRC) is the world’s third most common cancer, and in China has been increasing in incidence and mortality rates with each passing year.7 Research has revealed that the pandemic will increase mortality in CRC patients, who will have an increased mortality rate of 15.3%-16.6% after five years.8 Although the pandemic situation is ameliorating, it has already had an irreversible impact on CRC patients. We must admit that the COVID-19 pandemic is more challenging for cancer patients and cancer survivors, who are not only at higher risk of contracting COVID-19, but are also affected by a lack of timely access to services and treatment.9 Meanwhile, both treatment and rehabilitation also have an impact on family caregivers, the majority of whom are spousal caregivers (SCs).10 , 11 SCs are required to not only meet the daily living, disease care, and emotional support needs of patients, but also to assume more family and social responsibilities. Since the pandemic, individuals have been forced to implement rehabilitation programmes at home, which has again increased the burden on SCs to some extent.12 CRC patients and SCs can be viewed as a whole, suffering from a health crisis at the same time.11 Therefore, we are supposed to adopt new evolutionary interventions for CRC couples in the context of the pandemic. Self-efficacy (SE), first formulated by Bandura, is the idea that people can autonomously modify behavioural, environmental, and personal factors to promote health outcomes.13 , 14 As a personal behavioural feature, SE is inextricably linked to personal traits, environment, and events, which means that this behavioural trait can be modified by external influences.15 , 16 SE acts as a mediator of myriad positive outcomes, such as high self-management, and high quality of life, which play an important role for CRC couples.17, 18, 19 Previous studies have shown that patients are encouraged to engage in rehabilitation programmes at home during the COVID-19 pandemic, which requires a higher level of self-management.12 , 20 Consequently, a commitment to enhancing SE in patients and SCs is appropriate in the context of COVID-19. In addition, high levels of SE imply better symptom control in cancer patients,21 , 22 fewer complications of Peripherally Inserted Central Catheter (PICC) in chemotherapy patients,23 and less emotional distress for SCs.16 Thus, developing SE interventions would be beneficial for CRC couples. Previous studies have been analysed for SE interventions.24, 25, 26, 27, 28 These studies used different approaches to improve the SE of CRC patients and/or SCs, but obtained different outcomes. Studies that failed to produce positive outcomes had the following characteristics: SE was not considered the main intervention goal,24, 25, 26, 27, 28 patient preference was barely considered in advance,25 , 26 , 28 and the intervention approach was single (e.g., online only).26 Therefore, a multi-path intervention that aims to improve SE based on CRC couples’ preferences could be a positive future direction. Our previous study focused on summarising existing intervention studies for CRC patients and caregivers to improve their SE. Based on the findings of this review, a preliminary SE intervention programme was constructed.27 This programme is based on SE theory, with the primary goal of improving the SE of CRC couples. The intervention content targeted the four sources of SE as defined by SE theory, including performance achievement, vicarious experience, verbal persuasion, and emotional arousal, all of which contributes to SE development. For cancer patients and/or caregivers, performance achievement refers to the patients’ and/or caregivers’ successful experience in coping with cancer. The accumulated experience of coping with cancer will help diminish the negative effects of occasional failures in coping with cancer. Vicarious experience means that patients can encourage themselves to develop positive behaviours in coping with cancer by observing the positive behaviours of others in coping with cancer. Verbal persuasion is the process of persuading cancer patients and/or caregivers to be positive in dealing with the cancer challenge. The emotions evoked in each response to the task will also positively affect SE. The intervention content is as follows: Face-to-face sessions on skills training and knowledge strategies are provided, as well as web-based counselling and peer support services. The face-to-face sessions aim to enhance the performance accomplishments of CRC couples. The web-based sessions aim to provide vicarious experiences and verbal persuasion. Health providers attend to the emotional reactions of CRC couples at every session and guide them to positive emotions whenever possible. Skills training includes stoma care, coping with relationship intimacy challenges, and relaxation skills. Knowledge strategies include stoma, cancer symptoms, communication, and psychology. Each face-to-face session lasts less than one hour, while a web-based meeting, like a consultation, lasts about 30 minutes. Evaluation of intervention outcomes included SE, anxiety, depression, and quality of life, which interacted with each other, with SE being the primary outcome. However, the programme details have not been adapted to the preferences and needs of CRC couples. For instance, how could the intervention delivery order be better adapted to a couple’s experience of coping with cancer? What are the components that CRC couples value that we overlooked? There is a way to capture their experiences by conducting in-depth interviews. In this article, we are interested in learning about the experiences of CRC couples coping with cancer during COVID-19, and what needs and preferences developed from these experiences to further modify the SE intervention. Methods Aim The study aims were to understand the experiences of colorectal cancer patients and their spousal caregivers during the COVID-19 pandemic, and to develop an SE intervention for these couples. Study design This study used the descriptive phenomenological approach by Husserl,29 which could gain insight into the essentials of human experiences.30 This approach allowed CRC couples to narrate their own experiences of coping with cancer, to deepen the researchers’ understanding of their lived experiences. To ensure rigour, we used the criteria established by Lincoln and Guba: credibility, transferability, dependability, and confirmability.31 We ensured these four points by describing in detail the research environment, background, recruitment process, participants, and collection of data. Participants We are extending an invitation to eligible CRC patients and SCs. The inclusion criteria were (a) patients diagnosed with pathologically confirmed CRC; (b) his or her spouse was the patient’s principal caregiver; (c) all participants were older than 18 years, and (d) both CRC patients and SCs had comprehension and expression abilities and could clearly describe their experiences. If CRC patients and/or SCs had any mental, cognitive, or language disorders, they would be excluded. Participants were selected using the purposive sampling method, which is suitable for this study to obtain more information relevant to the research question. The first (JG) and second (MC) authors are registered nurses with bachelor’s degrees and are currently pursuing master’s degrees. The third author (QC) is a registered nurse with 10 years of clinical experience. All researchers have been trained in qualitative research. When patients were identified as eligible, JG approached them face-to-face to establish a friendly relationship and provided them with information about the study purpose. Those who were interested in participating in the study were asked to sign a formal agreement that included consent to participate in the interview and be recorded. Ethics consideration This study was conducted in accordance with the Declaration of Helsinki. We confirm that all methods were performed in accordance with the relevant guidelines and regulations. Ethical approval was provided by the Jiangnan University research ethics committee (Approval No. JNU20210918RB08). This study obtained verbal or written informed consent from participants, who were told they may leave the research study at any moment without consequence. Participant names and admission numbers were replaced with code numbers to protect their privacy. Data collection Data were collected through semi-structured interviews from September 2021 to November 2021 in the Affiliated Hospital of Jiangnan University. The participants decided upon the interview time. Interviews were conducted by JG, MC, and QC. The interviews were audio recorded and notes were taken to record the participants’ body language and facial expressions. During each interview, only the participant and researcher were present. All interviews were completed with CRC couples (patient and SC present at the same time). Since there were traffic blockages during the pandemic, we used WeChat video or voice conferencing if face-to-face interviews were not possible. The in-depth interview guide was jointly developed by all authors after discussion. The guidance consists of two sections. The first section includes general information, including age, sex, education level, type of cancer, length of time as a caregiver, etc. The second section consists of a list of open-ended questions used in the semi-structured interview. Two pilot interviews were conducted to test the semi-structured interview questions and train the researchers. The final interview guide was derived from discussions among all authors (one of whom is an expert in oncology nursing) and two pilot interviews, and was found to be feasible through practice (see Table 1 ). The guidance was then applied to all participants. The interview questions included two aspects, namely “impact of CRC during the COVID-19 pandemic” and “reflecting on the meaning of the experience”. At the beginning of the interview, the participants were asked to describe either their experiences of the disease, or their experience caring for a patient, and elaborate on their experience with prompts such as, “Can you explain your answer in detail?”. A total of 21 eligible CRC couples was screened. Two couples refused to participate because of illness and no interest in participating. At the time the 11th couple was included in this study, there were no new additional findings at that time and data were considered to have reached saturation. Eleven in-person interviews and one telephone interview were conducted (no repeat interviews). The reason for conducting the telephone interview was that one SC was unable to care for their spouse (CRC patient) in the hospital. The mean interview length was 43 minutes (range, 15 - 77 minutes). The recorded interview data were transcribed verbatim. To ensure accuracy, transcripts were compared to the audio, and then returned to the participants for review.Table 1 Interview guide for CRC patients and caregivers. Table 1Participants CRC Patients Spousal caregivers Warming up Can you talk about your disease with me? Can you talk about your experience as a caregiver taking care of your spouse? Impact of CRC during the COVID-19 pandemic Question 1 Has the disease impacted you? Has caring for the patient impacted you? Question 2 How do you think the COVID-19 pandemic has affected your treatment? How do you think that the COVID-19 pandemic has affected your caring for your spouse? Question 3 As a CRC patient during this pandemic, what stands out to you as the most important? As a CRC caregiver during this pandemic, what stands out to you as the most important? Question 4 What do you get from coping with cancer? (both) Question 5 What is your mental state usually like when dealing with these effects? Has your mental well-being changed while coping with cancer? If so, what are the reasons for these changes? If not, what motivates you to maintain your mindset? (both) Reflecting on the meaning of the experience Question 6 What do you think is most important to you from these findings? (both) Question 7 What advice do you have for other patients/caregivers or health providers? (both) Data Analysis The study results were concluded using the Consolidated Criteria for Reporting Qualitative Research (COREQ) criteria.32 The authors 1-3 used NVivo version 12 software to manage and analyse the data in accordance with the Colaizzi method.33 This is a common method of data analysis in phenomenological research. The text was first analysed by multiple close readings, which were broken down into units of meaning, and then condensed to make it shorter without sacrificing any important information. These meaning units were given a code. The codes were recombined to find themes and subthemes based on the differences and similarities. After the initial identification of themes and subthemes, the inspection results were returned and redundant parts were subtracted to make the results more concise. In this process, JG and MC worked on the initial codes, then the themes were refined and named by JG and QC. When disagreements in analysis occurred, the authorship team would resolve the matter by discussion. Coding was finalised and returned to the participants to ensure that the results truly reflected their feelings and experiences. Results Overall, 11 CRC couples participated in this study. The patients’ mean age was 61.45±13.27 years, while the SCs’ mean age was 60.82±11.79 years. Participants’ demographic characteristics are reported in Table 2 .Table 2 CRC patient and caregiver characteristics. Table 2 Age (y) Gender Marriage length (y) Education level Work or not Place of residence Types of cancer Stoma Length of Time as a SC P P1 63 Male 42 Middle school No City RC No N/A P2 62 Female 40 Primary school No Village CC Yes P3 74 Male 47 Middle school No City RC No P4 71 Male 45 Primary school No Village CC No P5 33 Female 9 College Yes City RC No P6 52 Male 26 Primary school Yes City RC Yes P7 68 Male 38 College No City CC No P8 75 Male 46 Primary school No City CC Yes P9 64 Male 35 Primary school No City RC Yes P10 44 Female 25 Middle school No City CC No P11 70 Female 47 No education No City CC Yes SC SC1 67 Female 42 Middle school No City N/A N/A 1-2 years SC2 66 Male 40 Middle school No Village 1-2 years SC3 68 Female 47 Middle school No City <1 year SC4 70 Female 45 Primary school Yes Village <1 year SC5 37 Male 9 University Yes City <1 year SC6 45 Female 26 Primary school Yes City <1 year SC7 65 Female 38 High school No City >2 years SC8 71 Female 46 Primary school No City <1 year SC9 58 Female 35 High school Yes City >2 years SC10 50 Male 25 Middle school Yes City 1-2 years SC11 72 Male 47 High school No City <1 year abrAbbreviationsCC, colon cancer; N/A, not available; P, patient; RC, rectal cancer; SC, spousal caregiver. Three main themes emerged from the content analysis: (1) Get and contribute support, (2) Life’s challenges, and (3) The Journey of Reconstruction. The themes and sub-themes are presented in Table 3 .Table 3 Themes and subthemes. Table 3Themes Subthemes Get and contribute support Caregivers become primary agents Emotional support from others on the cancer ward Making a contribution can lead to a sense of value Life’s challenges Bodily discomfort causes discrepancies with others Caregivers into crisis Strained spousal relationship The Journey of Reconstruction Adjustment of reflections Rebuild yourself Get and contribute support The overwhelming majority of participants reported that all kinds of support are important in coping with cancer. As a couple, they usually support each other, even if one member of the couple is ill. Moreover, other types of support will also be sought, such as support from other family members, medical support, and so on. As stated by SC11:When she was diagnosed with cancer, I directed my son to call a doctor and make an appointment for a hospital bed and prepare for surgery. She (patient) didn't need extra care before the operation. Since we came here, I have always been with her. We are getting old, and always take care of each other at home. We would be happy if our son came to visit us. In the early stages of CRC diagnosis, couples tend to actively seek support. They are willing to share their experiences with others in the same situation and give others support in the process. These supports include the three aspects described in Table 4 .Table 4 Interview excerpts of the subtheme of Get and contribute support. Table 4Subtheme Interview excerpts Caregivers become primary agents ◆ From caregivers He (SC) is always there, although he occasionally fights with me. I rely on him a lot, he’s the only one who cares for my stoma. (P10) Usually, I need to help him to expand the anus (every four or five days) and observe his stoma. Sometimes it gets a little red. I would help to disinfect, and apply a little gin paste. (SC9) ◆ From health providers During the disease period, we have kept communicating with a doctor to help my recovering better. (P3) After the operation, he needs to turn his body once every two hours. The nurses did a great job taking care of my husband, particularly during that special period. They kept helping my husband turn his body once every two hours, and observing his condition now and then. I would say they are very responsible and kindly help us a lot. (SC6) ◆ Support from relatives We are from the countryside. Relatives sometimes comfort me. We often talk to and give advice to each other, such as eating well, having a good rest, and not worrying about the disease. (P2) My nephew was very kind to us, and when he found out his uncle was ill, he sent us ten thousand dollars overnight. (SC6) Emotional support from others on the cancer ward ◆ Support from network Sometimes I think online friends give more encouragement. They don't want anything in return. There are many patients on websites who share their lives, and I can chat with them freely. (P5) ◆ Support from wardmates We're in the same ward, we talk to each other about the illness and recovery. After being discharged from the hospital, we will add WeChat and communicate with each other. Due to having the same disease and experience, it is easy to communicate and understand each other, and we have more common language. (SC11) Making a contribution can lead to a sense of value ◆ Couples support each other I feel better now that she (SC) can go play mahjong in the afternoon. As long as we are in a good frame of mind, life will be better. (P8) We both got used to supporting each other and the whole family. Before the disease, we were at work, we have tried our best to support each other and balance work and life, e.g., we cook together, take care of our children and grandchildren, and now it is basically him to do this work. I'm relaxed now, you see I'm so fat, haha. (SC9) ◆ Give back to family They're so nice to me, and I'm still wondering how to pay them back. One is that I take good care of myself, after which I continue to make money to repay them. (P6) ◆ Support companions A 49-year-old friend was very sad. She wonders why I got the disease when I was so young. Then I said to her, you are 49 years old. You see how old I am, I am like this. You do not have to compare yourself with older people. In the end I succeeded in comforting her. (P5) ◆ Contribute to social development We also want to see some miracles. If he recovers well, it will not only be a continuation of his life, but also a contribution to the medical treatment cause. (SC1) Alipay can choose to donate the remains. I want to donate organs, but unfortunately, I can't donate my organs after chemotherapy. (P5) Caregivers become primary agents Some CRC patients reported they were usually with their SC, and living together made them dependent on each other. As one participant said, “The doctors are talking to my wife about my illness, she knows more about it than myself. (P6)” Patients prefer to stay behind their SC during rehabilitation. Indeed, the caregiver role is that of an agent of patient self-management. When SCs have difficulty caring for patients, it is often up to them to proactively seek support from health providers, with most of the support sought being related to disease knowledge and care strategies. Emotional support from others on the cancer ward There are also special moments when patients tend to seek comfort from strangers, such as friends met in the hospital. Since the caregiver has already taken on the caregiving task, the patient might be too embarrassed to again express his or her emotional needs to the SC and would turn to strangers instead. As one participant put it: “Getting support from strangers makes me less burdened in my mind. (P5)” Furthermore, it’s a positive sign that wardmates, who are in the same situation, encourage and comfort each other during hospitalisation. Participants indicated that they kept in touch with their wardmates, and encouraged each other, which gave them the confidence to cope with cancer. SCs also communicate with other caregivers, but the topic is usually related to caring for the patient. Making a contribution can lead to a sense of value Once the CRC patient has improved, they can also be freed up to help their SC and take care of each other in their daily lives. Several participants expressed a desire to contribute to their families after they were feeling better. For instance, some participants indicated they wanted to continue to work and earn money to support their family, while others said that they wished to help their spouse take on household chores. Living with CRC leaves them temporarily incapable of working or even taking care of themselves. To some extent, it blocks the source of SE for CRC patients, which affects the accumulation of performance achievements. They hoped to reintegrate into society if possible. This altruism can give patients a sense of value and perhaps allow SCs to gradually step down from the role of agent. In the beginning, the SCs assumed the role of the patient’s agent, with the patient passively receiving care and working with the SC to promote recovery outcomes. Later, as the patients adjust and regain strength, they take on some of the responsibility for self-care while expressing a willingness to help others, but this takes some time. Life’s challenges The CRC couples mentioned the challenges they have faced in dealing with cancer. Since a cancer diagnosis, CRC couples face a spectrum of challenges that have emerged gradually over time. Even though they have become accustomed to living with cancer, there are still inevitable negative effects, such as maintaining relationships. As a patient who had completed an ileostomy closure said,He didn’t take care of our shared home when he was young, but he wants me to take care of him when he’s old. My heart died a long time ago. He was selfish and never cared about my feelings. Maybe he will only be good to me when he is dead. (SC4) Thus, there are many life challenges in coping with CRC. These impacts include the three aspects as described in Table 5 .Table 5 Interview excerpts of the subtheme of Life’s challenges. Table 5Subtheme Interview excerpts Bodily discomfort causes discrepancies with others ◆ Stress of surgery When I was to sign the consent form for the operation, the doctor told me that I may not be able to protect the anus, then I collapsed. (P5) ◆ Care burden of stoma It is always inconvenient with this stoma, particularly for going out, which limits my social activities indeed. I took care of the stoma for him every time when we went out. I had more say in the matter. (SC9) ◆ Side effect of chemotherapy My tongue becomes painful and red as soon as I eat. (P7); She had just finished chemotherapy and was feeling nauseous a lot at that time, and I couldn't handle the situation. We went to the hospital and the doctor was able to prescribe that antiemetic. I told the doctor that I didn't care how much it cost, as long as it made her feel better. (SC10) Caregivers into crisis ◆ Disease Diabetes and numb legs have caused my poor health. I feel uncomfortable all over and in severe situations, I called 120 for help because of dizziness. (SC4) ◆ Multiple burden Due to the COVID-19 pandemic, we need to come to the hospital for nucleic acid tests one day before the official admission. Only a negative nucleic acid test makes us eligible for hospitalisation. (P7, SC7) In addition to taking care of the sick, I also need to take care of our elderly parents. To be honest, sometimes I feel tired. (P9, SC9) There was no such thing before. Due to our aging legs, it is inconvenient for us to go to the hospital. Moreover, it is depressing that we need to shuttle back and forth to the hospital on a hot day with our luggage, and I think we're going to die. (P4, SC4) ◆ Like being on a circle and not stopping The treatment process seems to be endless. After the operation, I have been receiving chemotherapy, and now we still cannot see the end. (P7, SC7) Strained spousal relationship ◆ Lack of "we" consciousness I often fight with him and he rarely considers my feelings or opinions when it comes to changing the ostomy bag. He could not get it right, he said, then you do yourself. (P10) Before he (the patient) got sick, he liked to play and never took care of our family. Now he is ill and depends on me. Don't we quarrel? (SC4) ◆ Lack of proactive communication She (SC) is so grumpy that she never considers the other person's feelings when she speaks. (P4) To be honest, I'm sadder than him (the patient). To prevent our brothers and sisters from laughing at us, I tend to avoid communicating with them, and I need to be strong. (SC6) Bodily discomfort causes discrepancies with others With cancer treatment, patients were changed physiologically, creating differences compared to the general population. Patients who undergo surgical treatment usually need to live with their stoma for a period of time, or even the rest of their lives. Patients feel more anxious and disturbed if they are different from others. Meanwhile, the SC experiences this sense of difference together with them. Some SCs said they kept track of patient visits and helped them carry tissues to facilitate stoma care. When asked about their spouse's stoma, they would answer along with the patient and even talk more, displaying the same conversational tone and attitude as the patient did. Even as they adjust to such a life, they never regard cancer treatment as a positive event, and feel exhausted and helpless. There are reports of CRC patients whose weight continues to decline. They usually look miserable when surgery, chemotherapy, or stoma issues are discussed. Caregivers in crisis SCs were also plagued by illness, expressing they had no time to take care of their physical needs. Not only that, the COVID-19 pandemic has also increased the complexity of their care work. SCs said that it’s not easy to get into the hospital, they needed to go through layers of formalities, and there was no end in sight. Although the SC initially assumes the role of the patient’s agent, he or she is unable to sustain it all of the time. They may complain or even want to escape from the SC role once they have been in the proxy role for too long, or their physical condition is in crisis. Also, CRC patients were aware of this situation but were helpless to make changes. Strained spousal relationship Some patients and their SCs also lack the desire to communicate, over time, which may make it difficult for couples to develop a mutual understanding. This may lead to the patient lacking knowledge about the SC, resulting in their ignoring the SC’s feelings. For instance, a CRC couple reported that they could not find the right time to communicate: one wants to communicate, but the other refuses to say how they feel (P4, SC4). In an interview, the SC expected the patient to express himself more, but the patient did not have the desire to talk about his thoughts. The SC said that her spouse’s refusal to communicate is a regular part of daily life, so she has no more appetite to express herself (SC7). The Journey of Reconstruction Interviews showed that CRC patients and SCs develop ideas in the course of coping with cancer. For example, some participants expressed their appreciation for time spent with their family and planned to travel together when they were feeling better. They also actively considered the implications of the disease, and then developed some experiences to rebuild themselves. They reflected on their previous life with cancer before adjusting their behaviour, so as to complete the process of self-rebuilding. As a patient said,Over the years, it seems to me that the goal of my life has been to work and make money. However, after being diagnosed with cancer, I felt it was time to make a change. I felt that I could not just focus on working and making money - I also needed to take care of myself and try to enjoy life. (P5) The theme of The Journey of Reconstruction emerged from the data with two subthemes (Table 6 ).Table 6 Interview excerpts of the subtheme of The Journey of Reconstruction. Table 6Subtheme Interview excerpts Adjustment of reflections ◆ Understand each other Nurses often come to change the dressing, and they are very kind and gentle. (SC11) It is good for all of us to do a good job in the prevention and control of the epidemic. (P8) ◆ Remain calm and positive I have actually accepted the truth and do not think about how to live long. No matter what you think, it's already like this. (P5) Basically, he takes care of himself without my care, and apart from being at home, he usually goes fishing to relax. (SC9) ◆ Sum up experiences I perceived that you have to undergo chemotherapy when you have cancer. The capital for having chemotherapy is the quality of one's body. (SC1) At first, I was not used to the stoma in my stomach, but now I am used to it. Compared with the original troublesome ostomy bag, which requires handwashing before each use, there are now convenient disposable ostomy bags that can be thrown away after use. Moreover, this disposable ostomy bag is very cheap. (P9) Rebuild yourself ◆ Change multiple plans Before the cancer metastasised, we could play cards with our friends or travel. Now there is no way, we can only actively cooperate with the treatment. (P7) My daughter was supposed to be a graduate student, but she gave that up after knowing her dad was sick, with an intention to relieve the economic burden of the family. (SC6) ◆ Gradually adapt to epidemic prevention and treatment The doctors in this hospital are very responsible and come to the ward every day to give us nucleic acid tests. We are all at ease. (P8) Before going out, we have to think about what to bring in order to prepare for the stoma that needs our attention. Wherever we planned to go, he (the patient) kept asking me if we had something ready, e.g., a wet wipe…… These preparations are common now. (SC9) ◆ Keep healthy with knowledge and skills Now I've learned how to care for my stoma by myself. I don't need my wife to help me since I learned how to handle it. As a carpenter, I think I do learn things quickly. (P6) When we are in the hospital, we will listen to the doctors. .... They tell us what we should do. (SC2) I follow the instructions of doctors to take care of him (the patient). In some ways, I might be doing better than the doctor asked. As long as the doctor gives me some advice, I will do better. (SC1) Adjustment of reflections Over time, the state of recovery from illness will replace the previous state of life as the new normal. Suffering from CRC becomes a booster for couples to continue to grow. Participants indicated they would look back on their lives when they were alone. They have more time for solitude, thanks to the pandemic. They are relatively calm and do not feel regret about their lives, even though they are sick. They have realised the importance of self-regulation of the mind, so even if they have negative emotions, they will make adjustments to become more positive. In the recovery process, each successful act of care inspires more positive emotions, such as confidence, in both the patient and SC. Participants agreed with the statement that they obtain experience by practising. Some SCs who assist patients with stoma care said that they had tried several types of ostomy products, and then chose the right model for patients, based on their preferences. In addition, CRC patients have found ways to divert their attention away from the negative consequences of cancer treatment, such as postoperative pain or loss of appetite after chemotherapy. The CRC couple has continued to adapt their thoughts and behaviours as they struggle with living with cancer on a daily basis. Rebuild yourself The CRC couple who had rebuilt themselves were able to easily deal with day-to-day events involving cancer, which became a new life. First, they changed their plans, e.g., family plans and social plans. Cancer has affected the whole family. Almost all family members made dealing with cancer together with a priority, believing that only in this way can they better cope with cancer. However, due to the pandemic, the SC assumes the role of the primary agent and is difficult to transfer out. As time went by, CRC patients and SCs adapted both to cancer treatment and COVID-19 pandemic prevention, thanks to multiple supports. Regarding the impact of the COVID-19 pandemic on their lives over time, participants stated that they were accustomed to monitoring their body temperature every day and to wearing masks in public. But staffing constraints limit types of family support. Caregiving tasks continue to be a priority for CRC couples. Keeping healthy has become a major goal for CRC patients and SCs. Participants stated that they put a lot of effort into this, such as by learning self-care methods and practising the rehabilitation advice received from doctors, which also reflected their strong willpower and determination to take care of themselves. Discussion To the best of our knowledge, this is the first in-depth study to explore CRC couples’ experiences during the COVID-19 pandemic. The findings have revealed that CRC couples are adapting to cope with the impact of CRC and the COVID-19 pandemic, but challenges continue to emerge one after another and in rapid succession. The interviews revealed that CRC couples mobilise their own SE in coping with cancer. The SE theory explains which supports CRC couples need in their journey of coping with cancer together. A total of three themes was identified in this study: Get and contribute support, Life’s challenges, and The Journey of Reconstruction. Participants indicated that they would actively seek support to cope with cancer after being diagnosed with CRC (subtheme ‘Get and contribute support’). At first, the patient seems to be hidden behind the SC, with the SC taking on spousal care at the outset. This is a challenge to the SCs’ performance accomplishment. According to the previous SE programmes, it seems appropriate to add caregiving skills such as stoma care, cancer symptom care to SCs’ role at the onset of diagnosis as the first stage of the SE programme. Through the guidance provided by healthcare providers, SCs can better help patients cope with cancer. In addition, other family members may provide support, such as financial support, to CRC couples, while healthcare providers providing psychological counselling (verbal persuasion) to patients to courageously face the challenges of cancer alongside their SCs. During COVID-19, the cancer care content set for SCs should be available in both booklet and online versions for CRC couples. Access to psychological support for patients has also changed to both an online and face-to-face format. Some stated that they hid their feelings from their spouses, and turned to peers (wardmates or friends on a social media network) for comfort. On the one hand, their psychological needs may have been ignored by healthcare providers and spouses,34 possibly due to reduced contact with relatives as a result of the pandemic.35 On the other hand, seeking support from strangers may alleviate the guilt they feel towards their relatives due to their illness.36 This situation reflected the fact that support from peers rather than relatives was more acceptable to patients.37 , 38 Peer support seems to work more easily for patients who do not interact directly with their healthcare providers, as they lack confidence in understanding the instructions of their healthcare providers. Therefore, it seems possible to set patients as the primary recipients of peer support in our previous preliminary SE programme. Peer support enables patient-to-patient transmission of health or illness-related experiences (vicarious experiences). This kind of giving can also make them feel satisfied and become more confident.15 , 39 Corresponding to SE theory, the main goal of the first session is to enhance SC performance achievement and help patients stimulate positive coping emotions. Previous studies have provided psychosexual support for prostate cancer couples with the aim of improving intimacy impaired by cancer side effects (urinary, prostate function),40 while the SE intervention provided peer-given psychological support with content focused on adapting to cancer and restoring health, rather than on improving intimacy. Since surgery, CRC couples have officially begun their struggle with cancer (subtheme ‘Life’s challenges’). In addition to surgery, chemotherapy, and carrying a stoma are the main events that affect life. Moreover, the COVID-19 pandemic has prevented them from alleviating the discomfort of treatment. Recovery challenges will be ongoing for CRC couples. Therefore, the main goal of the second stage may be to support patients in improving skills and knowledge relevant to self-care (performance accomplishment). As coping with cancer has become the normalisation, a better outcome can only be achieved if patients and SCs work together. Otherwise, the SCs may be in a health crisis, and even intimate relationships may suffer as a result.41 , 42 Some participants in our study overlooked the importance of communication, leading to a number of problems in which they communicated less or had difficulty communicating effectively with their spouse. As in other cases, there will always be quarrels between husband and wife, which can lead to misunderstandings or resentments that are difficult to solve effectively. This ultimately leads to a lower quality of intimacy. SCs often put their health aside, even if they also suffer from chronic conditions.43 This sense of crisis seems to be stronger compared to before the pandemic.11 , 44 Prior to the pandemic, it was easier for couples to obtain more support, such as when one partner was tired and the other sought support elsewhere.44 However, SCs now need to care for their spouses alone and it is difficult to receive a break by alternating with others. In the SE programme’s second stage, it might be a suitable time to consider incorporating SC health education to improve their health.45 Moreover, intimacy promotion courses should be considered at this stage, to help CRC couples enhance their happiness and maintain a good relationship.24 , 46 The aim of this SE programme’s intimacy promotion component is to help both partners reach a consensus and rapport when coping with cancer. Previous cancer couple coping interventions have focused more on addressing couples’ cancer-related sexual problems and improving intimacy.40 It is worth noting that by this stage, the patient has been suffering from the disease for a period of time, causing the SC to gradually run out of energy, and CRC couples were in the grip of negative emotions. This is evident from the comments of some participants, who expressed concerns about their health, lives, and future. Health providers and peers remain sensitive to the patient’s withdrawal and the SC’s resistant attitude. Positive guidance should be given to CRC couples to be optimistic about the challenges that arise in their lives (Emotional Arousal). COVID-19 amplifies CRC couples’ feelings of helplessness, and we may need to increase our contact with CRC couples to support them online or by phone. We no longer mandate that all content be provided to CRC couples in one session. Corresponding to the SE theory, the main goal of the second session was to enhance the performance achievement of patients coping with cancer and to help SCs improve their own health to relieve the negative emotions of both parties as a result of coping with cancer. Over time, participants have practical experiences, whether experiences of failure or success, that have allowed them to adjust (subtheme ‘The Journey of Reconstruction’). This can be seen as a cognitive adjustment process.15 They evaluated their past lives. If given a positive self-evaluation, it may bring benefits and encourage participants to face life positively.47 , 48 If the self-evaluation was negative, it could lead them to fall into negative emotions with a low SE level.49 Therefore, in the SE programme’s third stage, health providers are expected to focus on guiding CRC couples by making positive evaluations of their previous lives and directing them to focus on happy, pleasurable moments to motivate themselves to move forward (verbal persuasion). This positive guidance may bring comfort to both patients and SCs. In contrast to palliative psychotherapy for couples with advanced cancer, the SE programme targets cancer patients who are likely to survive in the long term alongside their SCs. Healthcare providers could guide and inspire them to feel confidence in living a longer life together, while interventions for patients with advanced cancer were aimed at soothing emotions, alleviating discomfort, and achieving calmness and tranquility.50 In the final, fourth stage, the healthcare providers review the CRC couple’s self-rebuilding efforts. This review helps us discover what needs CRC couples still have that remain unmet.51 Corresponding to SE theory, the main task of the third and fourth sessions was to verbally persuade the CRC couple, provide counselling services to identify their unmet needs, and awaken positive emotions. To accommodate the impact of COVID-19, the third and fourth course delivery format, which is primarily counselling-based persuasion, can be adapted to telephone or online. In summary, by understanding the experiences of CRC couples coping with cancer during the COVID-19 pandemic, we clarified the specific content and sequence of the previous SE programme. In the first session, SCs are taught CRC caregiving knowledge and skills to help them feel confident in becoming a caregiver. Patients need psychological support at this time. Healthcare providers identify their avoidant attitudes and encourage patients to face reality to gradually work with their SCs to overcome the challenges of cancer. During this period, patients can be encouraged to interact with their peers if they avoid contact with healthcare providers. As CRC couples progress towards recovery, patients should be encouraged to complete self-care to the best of their ability. The core of the second session is to teach both CRC patients and SCs the knowledge and skills of self-care and intimate relationships. The patient no longer hides behind the SC to truly ease the SC’s burden, whether it be psychological or physical. The emotions of couples, especially SCs, need to be noticed during this period. When they feel overwhelmed, it may exacerbate the effects of negative emotions. In the third session, CRC couples are guided to make positive comments about their previous recovery experiences. As of the third session, all of the main content has been provided to CRC couples. In the fourth session, there is no longer set content. Prior to the fourth session, the healthcare providers check the CRC couples’ mastery of the content of the previous sessions and determine the content of the fourth session based on individual couples’ needs. At the same time, researchers cannot ignore the impact of COVID-19, and adapting the delivery format of the above content to be more flexible is in line with the trend. Overall, previous couples coping interventions have been more comprehensive, helping patients and caregivers to reduce distress, improve coping, adapt to cancer, and promote intimacy.52 The present SE programme is guided by SE theory with the primary goal of enhancing couples’ SE. SE is a mediator of multiple factors such as quality of life, anxiety, and depression.53 It appears that SE interventions can be a mediating intervention to help cancer couples achieve additional benefits in other areas. Meanwhile, there are many commonalities between the present and previous studies. The SE programme also focuses on the psychological needs of cancer couples by considering the cancer patient and SC as a dyad with simultaneous interventions. The SE programme may enable couples to successfully resist stressful events as a result of cancer and reduce negative emotions by improving SE levels. It is worth noting that the majority of study participants had received a junior or senior high school education, which may have influenced their attitudes toward cancer. Moreover, most were from urban areas, so future studies could be conducted in rural areas. Strengths and limitations Several strengths and limitations exist in this study. First, we paint a general portrait of what CRC couples feel when coping with cancer. According to their experiences, we have made a specific design for the SE programme. The SE programme will be divided into four sessions to teach a variety of content. It may be able to increase SE programme acceptance to some extent, so that CRC couples benefit from the programme. There are several limitations in this study. The CRC couples were interviewed at the same time, which may have encouraged them to say what they believed the other person wanted to hear. Future studies could consider interviewing patients and SCs separately. In addition, because we performed the interviews and analysed the content in Chinese and then translated it into English, language differences can be considered a study limitation. The interview guide was validated by nursing experts and pre-interviewed participants, but no other studies have been conducted using this interview guide, which may be a limitation of this study. Implications for practice The study findings suggest that COVID-19 measures appear to reduce ease of access to support for couples coping with cancer. SC helplessness may be amplified during a pandemic, which in turn may lead to faster and more pronounced exposure of problems in CRC couples. Therefore, professional guidance may still be needed for CRC patients and SCs. In the future, we would encourage researchers to develop easily accessible interventions, such as web-based interventions, to help cancer couples. In addition, it is suggested to help cancer couples identify problems in their intimate relationships and actively guide them to view setbacks with optimism. Moreover, the situation of cancer couples is assessed on time, to ensure that interventions can match the reality of the situation. Conclusion The CRC patients, with their SCs, described several disruptions caused by cancer and the COVID-19 pandemic. The challenges faced by CRC couples are continuing and even increasing during the pandemic. The study findings will contribute to the improvement of the SE programme, which will offer four sessions to CRC couples. The first session provides counselling to patients and caregiving skills to caregivers, which was neglected before this study. The second course provides self-care knowledge and intimacy-related content. The third session provides positive guidance to help CRC couples discover the positive aspects of their lives. The self-reconstruction of CRC couples was done in the last session. The content of the fourth session is decided according to the actual situation of the individual CRC couple. To accommodate the pandemic, it is considered of great importance to offer booklet copies of course content delivered to CRC couples, and provide timely telephone instruction. Financial support and sponsorship This work was financially supported by the National Natural Science Foundation of China (No. 82172844). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Conflicts of interests The authors declare no conflicts of interest. Acknowledgments The authors gratefully acknowledge the support from the hospital and its nursing staff, and all participants for sharing their experiences in this study. ==== Refs References 1 Prevention BoDCa. Questions and Answers about the Novel Coronavirus Pneumonia Prevention and Control Program (9th Edition). National Health Commission of the People’s Republic of China. Published 2022. Accessed. 2 Office HER. Update on the COVID-19 outbreak as of 24:00 on August 9. National Health Commission of the People’s Republic of China. Published 2022. Accessed. 3 Yin chaonan. Analysis of epidemiological characteristics and disease burden of COVID-19 in mainland China at early stage, Shandong University. 2021 4 Edge R. Mazariego C. Li Z. Canfell K. Miller A. Koczwara B. Psychosocial impact of COVID-19 on cancer patients, survivors, and carers in Australia: a real-time assessment of cancer support services Support Care Cancer 29 9 2021 5463 5473 10.1007/s00520-021-06101-3 33694089 5 Amaniera I. Bach C. Vachani C. Hampshire M. Arnold-Korzeniowski K. Healy M. Psychosocial impact of the COVID-19 pandemic on cancer patients, survivors and caregivers J Psychosoc Oncol 39 3 2021 485 492 10.1080/07347332.2021.1913780 33870877 6 Ng D.W.L. Chan F.H.F. Barry T.J. Lam C. Chong C.Y. Kok H.C.S. Psychological distress during the 2019 Coronavirus Disease (COVID-19) pandemic among cancer survivors and healthy controls Psychooncology 29 9 2020 1380 1383 10.1002/pon.5437 32497353 7 Sung H. Ferlay J. Siegel R.L. Laversanne M. Soerjomataram I. Jemal A. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries CA Cancer J Clin 71 3 2021 209 249 10.3322/caac.21660 33538338 8 Maringe C. Spicer J. Morris M. Purushotham A. Nolte E. Sullivan R. The impact of the COVID-19 pandemic on cancer deaths due to delays in diagnosis in England, UK: a national, population-based, modelling study Lancet Oncol 21 8 2020 1023 1034 10.1016/s1470-2045(20)30388-0 32702310 9 Pellino G. Spinelli A. How Coronavirus Disease 2019 Outbreak Is Impacting Colorectal Cancer Patients in Italy: A Long Shadow Beyond Infection Dis Colon Rectum 63 6 2020 720 722 10.1097/dcr.0000000000001685 32384401 10 Li Q. Loke A.Y. A literature review on the mutual impact of the spousal caregiver-cancer patients dyads: 'communication', 'reciprocal influence', and 'caregiver-patient congruence Eur J Oncol Nurs 18 1 2014 58 65 10.1016/j.ejon.2013.09.003 24100089 11 Li Q. Lin Y. Chen Y. Loke A.Y. Mutual Support and Challenges Among Chinese Couples Living With Colorectal Cancer: A Qualitative Study Cancer Nurs 41 5 2018 E50 e60 10.1097/ncc.0000000000000553 29200000 12 López-Rodríguez-Arias F. Sánchez-Guillén L. Aranaz-Ostáriz V. Triguero-Cánovas D. Lario-Pérez S. Barber-Valles X. Effect of home-based prehabilitation in an enhanced recovery after surgery program for patients undergoing colorectal cancer surgery during the COVID-19 pandemic Support Care Cancer 29 12 2021 7785 7791 10.1007/s00520-021-06343-1 34169328 13 Bandura A. Self-efficacy: Toward a unifying theory of behavioral change Psychological Review 84 2 1977 191 215 10.1037/0033-295X.84.2.191 847061 14 Thomas Hebdon M.C. Coombs L.A. Reed P. Crane T.E. Badger T.A. Self-efficacy in caregivers of adults diagnosed with cancer: An integrative review Eur J Oncol Nurs 52 2021 101933 10.1016/j.ejon.2021.101933 15 Bandura A. Social cognitive theory: An agentic perspective Annual Review of Psychology 52 2001 1 26 10.1146/annurev.psych.52.1.1 16 Levesque B.H.P.J. Bamgboje-Ayodele B.T.M.P.A. Paradiso B.H.J. Descallar B.M.J. Girgis B.P.A. Cancer-related challenges, unmet needs and emotional distress in male caregivers of women with breast cancer: the influence of self-efficacy J Psychosoc Oncol 40 4 2022 519 526 10.1080/07347332.2021.1955317 34392802 17 Ungar N. Wiskemann J. Sieverding M. Physical Activity Enjoyment and Self-Efficacy As Predictors of Cancer Patients' Physical Activity Level Front Psychol 7 2016 898 10.3389/fpsyg.2016.00898 27445882 18 Papadopoulou C. Kotronoulas G. Schneider A. Miller M.I. McBride J. Polly Z. Patient-Reported Self-Efficacy, Anxiety, and Health-Related Quality of Life During Chemotherapy: Results From a Longitudinal Study Oncol Nurs Forum 44 1 2017 127 136 10.1188/17.Onf.127-136 27991612 19 Manne S.L. Ostroff J.S. Norton T.R. Fox K. Grana G. Goldstein L. Cancer-specific self-efficacy and psychosocial and functional adaptation to early stage breast cancer Ann Behav Med 31 2 2006 145 154 10.1207/s15324796abm3102_6 16542129 20 Jammu A.S. Chasen M.R. Lofters A.K. Bhargava R. Systematic rapid living review of the impact of the COVID-19 pandemic on cancer survivors: update to August 27, 2020 Support Care Cancer 29 6 2021 2841 2850 10.1007/s00520-020-05908-w 33242162 21 Kurt S. Altan Sarikaya N. Correlation of self-efficacy and symptom control in cancer patients Support Care Cancer 30 7 2022 5849 5857 10.1007/s00520-022-06972-0 35364732 22 Marie N. Luckett T. Davidson P.M. Lovell M. Lal S. Optimal patient education for cancer pain: a systematic review and theory-based meta-analysis Support Care Cancer 21 12 2013 3529 3537 10.1007/s00520-013-1995-0 24085650 23 Liu X. Tao S. Ji H. Chen S. Gu Y. Jin X. Risk factors for peripherally inserted central catheter (PICC)-associated infections in patients receiving chemotherapy and the preventive effect of a self-efficacy intervention program: a randomized controlled trial Ann Palliat Med 10 9 2021 9398 9405 10.21037/apm-21-1848 34628865 24 Barsky Reese J. Porter L.S. Regan K.R. Keefe F.J. Azad N.S. Diaz L.A. Jr. A randomized pilot trial of a telephone-based couples intervention for physical intimacy and sexual concerns in colorectal cancer Psychooncology 23 9 2014 1005 1013 10.1002/pon.3508 24615831 25 Lim S.H. Chan S.W.C. Chow A. Zhu L. Lai J.H. He H.G. Pilot trial of a STOMA psychosocial intervention programme for colorectal cancer patients with stomas Journal of Advanced Nursing (John Wiley & Sons, Inc) 75 6 2019 1338 1346 10.1111/jan.13973 26 Giesler J.M. Keller B. Repke T. Leonhart R. Weis J. Muckelbauer R. Effect of a Website That Presents Patients' Experiences on Self-Efficacy and Patient Competence of Colorectal Cancer Patients: Web-Based Randomized Controlled Trial J Med Internet Res 19 10 2017 e334 10.2196/jmir.7639 29030329 27 Gong J. Hu C. Chen M. Cao Q. Li Q. Interventions to Improve Self-Efficacy in Colorectal Cancer Patients and/or Caregivers: A Systematic Review and Meta-Analysis Journal of Oncology 2021 2021 4553613 10.1155/2021/4553613 28 Cramer H. Pokhrel B. Fester C. Meier B. Gass F. Lauche R. A randomized controlled bicenter trial of yoga for patients with colorectal cancer Psychooncology 25 4 2016 412 420 10.1002/pon.3927 26228466 29 Husserl E. The Crisis of European Sciences and Transcendental Phenomenology: An Introduction to Phenomenological Philosophy. Evanston, IL: Northwestern University Press; 1954; 1970. 30 Lopez K.A. Willis D.G. Descriptive versus interpretive phenomenology: their contributions to nursing knowledge Qual Health Res 14 5 2004 726 735 10.1177/1049732304263638 15107174 31 Lincoln YS, and Egon G. Guba. . Naturalistic inquiry sage; 1985. 32 Tong A. Sainsbury P. Craig J. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups Int J Qual Health Care 19 6 2007 349 357 10.1093/intqhc/mzm042 17872937 33 Shosha GMA. EMPLOYMENT OF COLAIZZI'S STRATEGY IN DESCRIPTIVE PHENOMENOLOGY: A REFLECTION OF A RESEARCHER. European Scientific Institute. 2012(27). 34 Altschuler A. Liljestrand P. Grant M. Hornbrook M.C. Krouse R.S. McMullen C.K. Caregiving and mutuality among long-term colorectal cancer survivors with ostomies: qualitative study Support Care Cancer 26 2 2018 529 537 10.1007/s00520-017-3862-x 28844086 35 Zomerdijk N. Jongenelis M. Yuen E. Turner J. Huntley K. Smith A. Experiences and needs of people with haematological cancers during the COVID-19 pandemic: A qualitative study Psychooncology 31 3 2022 416 424 10.1002/pon.5819 34500498 36 Perloff T. King J.C. Rigney M. Ostroff J.S. Johnson Shen M. Survivor guilt: The secret burden of lung cancer survivorship J Psychosoc Oncol 37 5 2019 573 585 10.1080/07347332.2019.1569191 30798776 37 Park H.Y. Kim M.J. Kim J.Y. Kim S. Choi J.Y. Kim J.H. Could Peer Support Programs Be a Good Resource for Managing the Unmet Needs of Cancer Patients? J Cancer Educ 34 5 2019 950 957 10.1007/s13187-018-1399-4 30091013 38 Brodar K.E. Carlisle V. Tang P.Y. Fisher E.B. Identification and Characterization of Peer Support for Cancer Prevention and Care: A Practice Review J Cancer Educ 2020 1 10 10.1007/s13187-020-01861-8 39 Mokhtari L. Markani A.K. Khalkhali H.R. Feizi A. The perceived social support by Iranian women with breast cancer: a qualitative study Support Care Cancer 30 1 2022 941 949 0.1007/s00520-021-06478-1 34417884 40 Robertson J.M. Molloy G.J. Bollina P.R. Kelly D.M. McNeill S.A. Forbat L. Exploring the feasibility and acceptability of couple-based psychosexual support following prostate cancer surgery: study protocol for a pilot randomised controlled trial Trials 15 2014 183 10.1186/1745-6215-15-183 24886676 41 Jenkins C. Ho H.T. Nghiem H.P.L. Prue G. Lohfeld L. Donnelly M. A qualitative study on the needs of cancer caregivers in Vietnam Glob Health Action 14 1 2021 1961403 10.1080/16549716.2021.1961403 42 Collaço N. Wagland R. Alexis O. Gavin A. Glaser A. Watson E.K. The experiences and needs of couples affected by prostate cancer aged 65 and under: a qualitative study J Cancer Surviv 15 2 2021 358 366 10.1007/s11764-020-00936-1 32968952 43 Li Q. Chiang V.C. Xu X. Xu Y. Loke A.Y. The Experiences of Chinese Couples Living With Cancer: A Focus Group Study Cancer Nurs 38 5 2015 383 394 10.1097/ncc.0000000000000196 25159079 44 Collaço N. Rivas C. Matheson L. Nayoan J. Wagland R. Alexis O. Prostate cancer and the impact on couples: a qualitative metasynthesis Support Care Cancer 26 6 2018 1703 1713 10.1007/s00520-018-4134-0 29511952 45 Northouse L.L. Katapodi M.C. Song L. Zhang L. Mood D.W. Interventions with family caregivers of cancer patients: meta-analysis of randomized trials CA Cancer J Clin 60 5 2010 317 339 10.3322/caac.20081 20709946 46 Tiete J. Delvaux N. Lienard A. Razavi D. Efficacy of a dyadic intervention to improve communication between patients with cancer and their caregivers: A randomized pilot trial Patient Educ Couns 104 3 2021 563 570 10.1016/j.pec.2020.08.024 33129628 47 Li Q. Loke A.Y. The positive aspects of caregiving for cancer patients: a critical review of the literature and directions for future research Psychooncology 22 11 2013 2399 2407 10.1002/pon.3311 23712938 48 Jansen L. Hoffmeister M. Chang-Claude J. Brenner H. Arndt V. Benefit finding and post-traumatic growth in long-term colorectal cancer survivors: prevalence, determinants, and associations with quality of life Br J Cancer 105 8 2011 1158 1165 10.1038/bjc.2011.335 21878935 49 Hinz A. Friedrich M. Kuhnt S. Zenger M. Schulte T. The influence of self-efficacy and resilient coping on cancer patients' quality of life Eur J Cancer Care (Engl). 28 1 2019 e12952 10.1111/ecc.12952 50 von Heymann-Horan A. Bidstrup P.E. Johansen C. Rottmann N. Andersen E.A.W. Sjøgren P. Dyadic coping in specialized palliative care intervention for patients with advanced cancer and their caregivers: Effects and mediation in a randomized controlled trial Psychooncology 28 2 2019 264 270 10.1002/pon.4932 30353600 51 Hajdarevic S. Fallbjörk U. Fransson P. Åström S. Need of support perceived by patients primarily curatively treated for breast, colorectal, or prostate cancer and close to discharge from hospital-A qualitative study J Clin Nurs 2021 10.1111/jocn.15977 52 Regan T.W. Lambert S.D. Girgis A. Kelly B. Kayser K. Turner J. Do couple-based interventions make a difference for couples affected by cancer? A systematic review BMC Cancer 12 2012 279 10.1186/1471-2407-12-279 22769228 53 Cao Q. Gong J. Chen M. Lin Y. Li Q. The Dyadic Effects of Self-Efficacy on Quality of Life in Advanced Cancer Patient and Family Caregiver Dyads: The Mediating Role of Benefit Finding, Anxiety, and Depression Journal of Oncology 2022 2022 3073358 10.1155/2022/3073358

0

PMC9749380

NO-CC CODE

2022-12-15 23:22:53

no

Asia Pac J Oncol Nurs. 2022 Dec 14;:100179

utf-8

Asia Pac J Oncol Nurs

10.1016/j.apjon.2022.100179

oa_other

==== Front J Theor Biol J Theor Biol Journal of Theoretical Biology 0022-5193 1095-8541 The Author(s). Published by Elsevier Ltd. S0022-5193(22)00375-7 10.1016/j.jtbi.2022.111384 111384 Article Projecting the COVID-19 immune landscape in Japan in the presence of waning immunity and booster vaccination Sasanami Misaki Fujimoto Marie Kayano Taishi Hayashi Katsuma Nishiura Hiroshi ⁎ Kyoto University School of Public Health, Yoshida-Konoe, Sakyo, Kyoto, 606-8601, Japan ⁎ Corresponding author. 14 12 2022 14 12 2022 11138424 5 2022 8 12 2022 11 12 2022 © 2022 The Author(s). Published by Elsevier Ltd. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Coronavirus disease 2019 (COVID-19) booster vaccination has been implemented globally in the midst of surges in infection due to the Delta and Omicron variants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The objective of the present study was to present a framework to estimate the proportion of the population that is immune to symptomatic SARS-CoV-2 infection with the Omicron variant (immune proportion) in Japan, considering the waning of immunity resulting from vaccination and naturally acquired infection. We quantified the decay rate of immunity against symptomatic infection with Omicron conferred by the second and third doses of COVID-19 vaccine. We estimated the current and future vaccination coverage for the second and third vaccine doses from February 17, 2021 to August 1, 2022 and used data on the confirmed COVID-19 incidence from February 17, 2021 to April 10, 2022. From this information, we estimated the age-specific immune proportion over the period from February 17, 2021 to August 1, 2022. Vaccine-induced immunity, conferred by the second vaccine dose in particular, was estimated to rapidly wane considerably. There were substantial variations in the estimated immune proportion by age group because each age cohort experienced different vaccination rollout timing and speed as well as a different infection risk. Such variations collectively contributed to heterogeneous immune landscape trajectories over time and age. The resulting prediction of the proportion of the population that is immune to symptomatic SARS-CoV-2 infection could aid decision-making on when and for whom another round of booster vaccination should be considered. This manuscript was submitted as part of a theme issue on “Modelling COVID-19 and Preparedness for Future Pandemics”. Keywords Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Vaccine effectiveness Waning immunity Mathematical model Statistical model ==== Body pmc1 Introduction At present (October 2022), more than 2 years have passed since the World Health Organization (WHO) declared the global pandemic of coronavirus disease 2019 (COVID-19) (Cucinotta and Vanelli, 2020), and yet COVID-19 is still affecting our lives. Many countries initially implemented public health and social measures, such as lockdown and social distancing, because only non-specific countermeasures were available, and subsequently shifted their intervention strategies to incorporate vaccination from December 2020 (Haas et al., 2021, Hall et al., 2021, Mathieu et al., 2021, Thompson et al., 2021). The worldwide vaccination rollouts brought hope that the local epidemics could be suppressed more efficiently and regional or temporal herd immunity could be reached (Grauer et al., 2020). However, reports of breakthrough infections soon appeared (Bergwerk et al., 2021, Brown et al., 2021, CDC COVID-19 Vaccine Breakthrough Case Investigations Team, 2021, Juthani et al., 2021), and evidence suggests that the available vaccines have substantially lower effectiveness against the newly emerged variants, including the Delta variant (B.1.617) and later the Omicron variant (B.1.1.529), which quickly became the dominant variants worldwide in 2021 and early 2022, respectively (Andrews et al., 2022, Kahn et al., 2022, Loconsole et al., 2022, Lopez Bernal et al., 2021, Madhi et al., 2022, Mohapatra et al., 2022, Seppälä et al., 2021, Tseng et al., 2022). It was also found that vaccine-induced immunity wanes rapidly after the second and even third doses (Andrews et al., 2022, Ferdinands et al., 2022, Tseng et al., 2022). Out of concern for the evident waning immunity, some countries have already started offering a fourth dose of COVID-19 vaccine to increase the level of protection (Bar-On et al., 2022; Centers for Disease Control and Prevention (CDC), 2022; UK Health Security Agency, 2022). Japan launched its COVID-19 vaccination campaign on February 17, 2021, mainly using the messenger (m)RNA vaccines BNT162b2 and mRNA-1273 (Sasanami et al., 2022a). Healthcare workers were prioritized as the first group to receive the vaccines. The eligible population was then expanded on April 12, 2021 to include people aged ≥65 years, and later was further expanded to include younger members of the population in a descending manner. Japan achieved high vaccination coverage; overall, more than 75% of the population received the second dose of COVID-19 vaccine, and approximately 95% of the people aged ≥65 years became fully vaccinated (Figure 1 ) (Digital Agency, 2022, Prime Minister’s Office of Japan, 2022). A third dose of mRNA COVID-19 vaccine became available on December 1, 2021 to boost the immunity among healthcare workers and people who had received the second dose more than 6 or 7 months before. As of May 2022, Japan is considering another round of boosters (i.e., a fourth COVID-19 vaccine dose) using mRNA vaccines for people aged ≥60 years or those who have underlying health conditions and received their third dose more than 5 months previously (Ministry of Health Labour and Welfare, 2022a).Figure 1 The age-specific number of confirmed COVID-19 cases and second vaccine dose coverage in Japan from February 17, 2021 to April 10, 2022. (A–F) The black bars show the confirmed COVID-19 incidence among people aged 20–29 (A), 30–39 (B), 40–49 (C), 50–59 (D), 60–69 (E), and ≥70 (F) years. The blue line represents the estimated second vaccine dose coverage; it should be noted that the age grouping of COVID-19 vaccination is slightly different from that of COVID-19 incidence, i.e., aged 15–24 (A), 25–34 (B), 35–44 (C), 45–54 (D), 55–64 (E), and ≥65 (F) years. The methods applied for estimating the second vaccine dose coverage are described elsewhere (Sasanami et al., 2022b), and we estimated the coverage until March 13, 2022, when the daily proportion of the population that was newly vaccinated in each age group became lower than 0.01% of the age-specific population and the coverage was deemed to have plateaued. Japan has experienced a substantially lower number of COVID-19 cases and deaths compared with many Western countries (Ritchie et al., 2020). By the end of 2021, there was less than 5% of the cumulative risk of confirmed COVID-19 cases, and approximately 18,000 deaths in total had been reported as of May 16, 2022 (Ministry of Health Labour and Welfare, 2021a). However, after the Omicron variant became dominant in Japan, the country experienced its worst hit to date (sixth wave of the pandemic), with a maximum daily incidence reaching approximately 100,000 confirmed cases in early February 2022 (Figure 1) (Ministry of Health Labour and Welfare, 2022b). This surge should have conferred naturally acquired immunity among infected individuals. The impacts of such natural infection-induced immunity in combination with the intrinsic waning effects of the second and third vaccine doses have complicated monitoring the immune landscape, and to our knowledge there is no study thus far that regularly estimates the population-level immunity against symptomatic infection considering these effects. Here, we tackle the issue of reconstructing the fraction of the population that is immune to symptomatic SARS-CoV-2 infection over time and age. The objective of the present study was to present a framework that enables us to monitor the immune landscape, specifically the proportion of people who are immune to symptomatic SARS-CoV-2 infection (immune proportion), and incorporates the abovementioned complexities. We estimated the age-specific immune proportion, which is deemed more informative as opposed to the overall immune proportion for planning public health interventions or effective booster campaigns for additional vaccination rounds in the future, given the evidence of heterogeneities not only in contact patterns but also in risk in different age groups (Antonelli et al., 2022, Jordan et al., 2020, Lovell-Read et al., 2022). This study also accounts for the buildup and waning of vaccine-induced immunity after boosting, which should improve the likelihood of selecting an optimal initiation time, and the initiation timing of a vaccination rollout is key to its success (Gavish et al., 2022). 2 Materials and methods 2.1 Epidemiological data 2.1.1 Vaccination registry data We obtained Vaccination Record System (VRS) data from the Ministry of Health, Labour and Welfare, which recorded the daily number of vaccinees and their ages as aggregated into 5-year cohorts, for the period from February 17, 2021 to April 10, 2022. We also obtained Vaccination System (V-SYS) data from the website of the Prime Minister’s Office of Japan, which recorded the daily total number of distributed COVID-19 vaccine doses (in contrast with the number of doses actually administered) (Prime Minister’s Office of Japan, 2022). 2.1.2 Confirmed COVID-19 incidence data We obtained two different pieces of data from the Ministry of Health, Labour and Welfare (Ministry of Health Labour and Welfare, 2022c): (1) the daily number of newly confirmed COVID-19 cases (with no information on age), and (2) the weekly number of confirmed COVID-19 cases by sex and age. We retrieved subsets of data for the period of interest, which spanned from February 17, 2021 to April 10 2022. We modified dataset (1) to categorize the data into six age groups (na=6): 20–29, 30–39, 40–49, 50–59, 60–69, and ≥70 years by assuming that the age distribution of COVID-19 cases reported in dataset (2) is constant throughout the week. Furthermore, we assumed that the actual COVID-19 incidence (i.e., including unconfirmed cases) was four times higher than the number reported in dataset (1) (Sanada et al., 2022). Henceforth, “age group” refers to the abovementioned six age categories. 2.1.3 Vaccine effectiveness data We used published data on the vaccine effectiveness of BNT162b2 against symptomatic infection with Delta and Omicron variants, estimated in a test-negative case–control study conducted in England (Andrews et al., 2022). The study provides estimates for the second and third vaccine dose over time since vaccination, showing the evidence of waning protection levels. 2.1.4 Statistics for the Japanese population We obtained statistics for the Japanese population, including age-specific and prefecture-specific populations, from the Portal Site of the Official Statistics of Japan (Ministry of Internal Affairs and Communications, 2022). 2.2 Modelling the protected fraction of the population 2.2.1 Quantifying waning vaccine effectivenes To describe the waning immunity against infectious diseases including COVID-19, various decay functions have been employed in published modelling studies (Feng et al., 2022, Hogan et al., 2021, Khoury et al., 2021, Nishiura et al., 2006), but there is no general consensus or evidence that suggests the most plausible model to be used. We therefore employed the simplest, exponential function to model the decaying vaccine effectiveness against symptomatic infection with the Omicron variant induced by the second and third vaccine doses as follows:(1) δτ=me-γτ where δτ represents the vaccine effectiveness at τ days after the second or third vaccine dose; m represents the maximum vaccine effectiveness against symptomatic infection with Omicron, and γ regulates the speed of decay for immunity among vaccinees. Note that the spike in effectiveness shortly after receiving a dose of vaccine is omitted, and thus, δτ is a decreasing function. We estimated the parameters m and γ by fitting the exponential function to the data from a published study that reports vaccine effectiveness estimates against symptomatic infection with the Omicron variant, via the maximum likelihood method assuming Gaussian distributed errors, as a function of time since immunization (Andrews et al., 2022). From the published study, we specifically retrieved estimates for the effectiveness of BNT162b2, which the vast majority of the Japanese population received (Prime Minister’s Office of Japan, 2022). We obtained 95% confidence intervals (CIs) via the parametric bootstrapping method. Furthermore, we assumed that infection-induced immunity wanes in a manner identical to that of the effectiveness of the third vaccine dose, although it was assumed to confer perfect protection immediately after infection (m=1). 2.2.2 Estimating age-specific vaccination coverage for the second and third doses We estimated age-specific second and third vaccine dose coverage, by conducting all the computations shown below for each of the aforementioned age cohorts (i.e., the age-specific vaccination coverage was used for the following calculations). Because more than 98% of people in Japan who were vaccinated with the first dose received the second dose (Prime Minister’s Office of Japan, 2022), here, we considered the effectiveness conferred from the second and third doses only. 2.2.2.1 Estimating second vaccine dose coverage We calculated the uptake of the second vaccine dose, mainly using VRS data, while accounting for the reporting delay in the data. We also employed V-SYS data to complement the missing parts of the VRS data. A detailed explanation of this approach is provided elsewhere (Sasanami et al., 2022b). 2.2.2.2 Estimating third vaccine dose coverage To address the right truncation of the observed vaccination coverage, we fitted a logistic function to the third vaccine dose coverage data to estimate time-dependent coverage until April 10, 2022 and to predict it through August 1, 2022:(2) ω(t)=L1+e-k(t-x0) where ω(t) represents the vaccination coverage at calendar time t, which is the elapsed time since the start of the booster vaccination campaign (i.e., December 1, 2021); k represents the speed of increase in the vaccination coverage; L is the eventual booster vaccination coverage; and x0 represents the duration required for the coverage to reach 50% of its maximum (L). For the age groups 20–29 years, 30–39 years, and 40–49 years, we fixed L to be identical to the vaccination coverage for the second dose in each age group reported by the Prime Minister’s Office of Japan as of April 11, 2022 (Prime Minister’s Office of Japan, 2022), i.e., 80.0%, 79.9%, and 83.0%, respectively. We fixed these values because there were still too few people vaccinated to estimate all three parameters (i.e., k, L, and x0) for these age cohorts. By contrast, for the older age groups of 50–59 years, 60–69 years, and ≥70 years, all the parameters were estimated from empirical data. The abovementioned maximum likelihood estimation was performed on these vaccine rollout data. 2.2.3 Estimating the immune proportion We estimated the age-specific immune proportion, accounting for waning vaccine effectiveness following the second and third vaccine doses and infection-induced immunity. First, we expressed the population who obtained immunity from vaccination or infection as follows:(3) ∂∂t+∂∂sjtypet,s=-δtypesjtypet,sjtypet,0=λtype(t) where type indicates either the second or third vaccine dose or natural infection; λ(t) represents the number of people who newly received a vaccine dose or were infected at calendar time t (as described in 2.2.2.1-2. and 2.1.2); jtypet,s is the number of immune populations at time t who were vaccinated or infected s days ago; and δ∙ is the immune decay function estimated in subsection 2.1. Integration over the characteristic line gives the solution of the McKendrick equation (3), i.e.,(4) jtypet,s=λtypet-sexp-∫0sδtypexdx for t > s. Therefore, the total immune population at time t, J(t) is:Jtype(t)=∫0∞λtypet-yexp-∫0yδtypexdxdy We then computed the immune proportion by summing the number of people who were immune owing to vaccination and infection:(6) Jtotal(t)=Jsecond(t)+Jthird(t)+Jinfection(t)/P where the summation in the right-hand side of the equation computes the number of persons who are immune owing to the second or third vaccine dose or infection, and we converted this value into a fraction by dividing it by the age-specific population size, denoted as P. It should be noted that here we made the assumption that people who had lost immunity from their second vaccine dose regained immunity from booster vaccination or natural infection. In other words, under this assumption, the people who remained immune to symptomatic infection after their second vaccine dose had not yet received a third vaccine dose or been infected. This assumption enabled the simple summation in equation (6) and was deemed reasonable given the rapidity of the waning protection against the Omicron variant provided by the second vaccine dose and Japan’s vaccination scheme in which people generally became eligible for booster vaccination 6 or 7 months after receiving their second vaccine dose. The CIs for the estimated immune proportion were computed using the two sets of 1,000-parameter samples gained from the estimation of waning immunity from the second and third vaccine doses. Furthermore, we conducted sensitivity analyses in which three different scenarios were assumed. Because, to our knowledge, there is no established evidence regarding the waning rate of immunity from natural infection, the first two sensitivity analyses examined the impacts of different decay rates for infection-induced immunity. We assumed that the waning speed was: i) identical to that from the second vaccine dose; or ii) dependent on the dominant SARS-CoV-2 variant in circulation. For assumption ii), we applied the decay rate for the second and third vaccine doses to infection that occurred during the Delta- (until 31 December 2021) and Omicron- (1 January 2022 onward) dominant periods, respectively (Ministry of Health Labour and Welfare, 2021b). The third sensitivity analysis explored the scenario in which neither emergence of the Omicron variant nor a booster vaccination campaign had occurred. For this analysis, we first estimated the time-dependent vaccine effectiveness from the second vaccine dose against symptomatic infection with Delta, based on data from a published study (Andrews et al., 2022), and then computed the immune proportion using the same methods described above. The results of the sensitivity analyses are documented in the Supplementary material (https://github.com/nishiurah/immunelandscape). 2.3 Validation To assess the validity of our estimates, we conducted a real-time analysis, investigating if there is an association between the epidemic dynamics and the immune proportion using prefecture-level data. 2.3.1 COVID-19 incidence trend in each prefecture We collected data on the incidence of SARS-CoV-2 infection in each prefecture from the Japan Broadcasting Corporation (Nippon Hoso Kyokai; NHK) website (NHK, 2022). As a snapshot evaluation for the purposes of this study, we used data from February 7 to April 10, 2022, when the sixth wave of COVID-19 hit Japan, and calculated the prefecture-specific risk of infection per 100,000 people and then fitted it to a simple exponential growth model. If the estimated growth rate was positive, the prefecture was considered to have an increasing risk of infection, whereas if negative, it was deemed that the prefecture had a decreasing risk. The prefectures for which the estimated coefficient was not statistically significantly different from 0 (α=0.05) were removed from the analysis. 2.3.2 Estimating the prefecture-specific immune proportion We applied the aforementioned method of computing the immune landscape to estimate the prefecture-specific immune proportion from symptomatic illness. Because of the limited data availability, this analysis specifically considered the immunity gained from the third vaccine dose alone. We did not account for the second dose because of its small impact on preventing symptomatic infection from Omicron around the time period of interest and also because of the waning effect of infection-acquired immunity, which was assumed to be negligible given the short time period of interest. 2.3.3 Comparison of the mean immune proportions according to the risk trend Subsequently, we conducted the Wilcoxon rank sum test to compare the mean prefecture-specific immune proportion on April 10, 2022 according to the upward or downward risk trend. All analyses were conducted in R (version 4.2.0), and the data and codes are provided in the Supplementary files (https://github.com/nishiurah/immunelandscape). Note that vaccination data used here are provided in a publicly available form (downloaded from the website of the Prime Minister’s Office of Japan [Prime Minister’s Office of Japan, 2022]). 3 Results Figure 1 shows the second vaccine dose coverage and the total number of confirmed SARS-CoV-2 infections during the period from February 17, 2021 to April 10, 2022. The rate at which vaccination was conducted exceeded 1 million doses per day during the second dose program; this was set as the goal for the third vaccine dose campaign as well, and the goal was achieved in early March 2022. The most up-to-date coverage and the predicted booster vaccination coverage by age group are shown in Figure 2 . The coverage started plateauing in the older age groups (specifically, the 60–69 and ≥70-year-old groups) at the time when our analysis was conducted. By contrast, it was estimated that it will take approximately another 4 months for the younger age groups to reach the maximum possible coverage if the vaccination speed remains the same.Figure 2 Estimated waning of vaccine-induced protection against symptomatic infection with the Omicron variant after the second (left) and third (right) vaccine doses. The blue lines show the estimated protected fraction, with the shaded area representing their 95% confidence intervals (CIs) derived by parametric bootstrapping. The points represent the empirically reported vaccine effectiveness against symptomatic infection, with error bars displaying their 95% CIs, according to a published study (Andrews et al., 2022). The estimated waning vaccine effectiveness against symptomatic infection with the Omicron variant over time is shown in Figure 3 . The parameters in the immune decay function m (i.e., the maximum vaccine effectiveness) and γ (i.e., the waning rate of vaccine effectiveness) were estimated to be 0.93 (95% CI: 0.90–0.96) and 0.014 (95% CI: 0.014–0.015) for the second vaccine dose and 0.70 (95% CI: 0.70–0.71) and 0.0050 (95% CI: 0.0050–0.0055) for the third vaccine dose, respectively. The half-life of vaccine effectiveness was 50 days and 139 days for the second and third vaccine doses, respectively.Figure 3 Reported and predicted booster vaccination coverage stratified by age group for December 1, 2021 to August 1, 2022. (A–F) The dark blue lines show the reported booster vaccination coverage until April 10, 2022, and the light blue dash-dotted line represents the projected booster vaccination coverage among people aged from 20–29 (A), 30–39 (B), 40–49 (C), 50–59 (D), 60–69 (E), and ≥70 (F) years. The estimated immune landscape notably varied by age group over the course of time (Figure 4 and Table 1 ). The surge in COVID-19 cases substantially contributed to conferring immunity to the younger subset of the population. Such effects were most evident among those aged 20–29 years; by April 10, 2022, 21.0% (95% CI: 20.6–21.4) of this population was estimated to have acquired immunity solely from vaccination, but the addition of those who gained immunity from natural infection brought this value up to 44.4% (95% CI: 44.0–44.8). Incorporating infection-induced immunity, the immune proportion in this age group was predicted to be 52.3% (95% CI: 51.9–52.7) and 47.7% (95% CI: 47.1–48.4) on June 1 and August 1, 2022, respectively. The 30–39 and 40–49-year-old cohorts followed similar trends. By contrast, the immune proportions of the three older age groups were less impacted by infection. Among people aged ≥70 years, 48.6% (95% CI: 48.1–49.1) were estimated to have acquired immunity solely from vaccination, whereas the overall immune proportion (accounting for the effects of both vaccination and infection) in this group was 53.0% (95% CI: 52.5–53.5). As of April 10, 2022, a reduction in the immune proportion was already apparent, and the immune proportion was estimated to decline to 28.9% (95% CI: 27.9–30.0) by the summer of 2022.Figure 4 Age-specific proportion of the population immune to symptomatic SARS-CoV-2 infection from February 17, 2021 to August 1, 2022. (A–F) The lines show the time-dependent proportion of people immune to symptomatic SARS-CoV-2 infection among those aged from 20–29 (A), 30–39 (B), 40–49 (C), 50–59 (D), 60–69 (E), and ≥70 (F) years, and the shaded areas are their 95% confidence intervals (CIs), computed via the bootstrapping method. The red lines reflect the scenarios in which the immunity from natural infection as well as the immunity from the second and third vaccine doses are considered, whereas the light blue lines reflect the scenarios in which only the immunity from the second and third vaccine doses is taken into account. The projection was performed on April 10, 2022, and the estimates afterward are from a model-based projection. Table 1 Age-specific proportion of the population susceptible to symptomatic SARS-CoV-2 infection on April 10, June 1, and August 1, 2022. The values in V and V + I represent the proportion of the population that is susceptible to symptomatic infection with the Omicron variant, considering immunity solely from vaccination (V) and immunity from both vaccination and natural infection (V+I), respectively. The projection was performed on April 10, 2022, and the estimates afterward are from a model-based projection. Numbers in parentheses show the 95% confidence intervals (CIs) as computed by the parametric bootstrap method. 2022/4/10 2022/6/1 2022/8/1 Age group V (95% CI) V + I (95% CI) V (95% CI) V + I (95% CI) V (95% CI) V + I (95% CI) 20–29 79.0 (78.6–79.4) 55.6 (55.2–56.0) 65.7 (65.3–66.1) 47.7 (47.3–48.1) 65.5 (64.9–66.2) 52.3 (51.6–52.9) 30–39 78.9 (78.5–79.2) 59.7 (59.3–60.1) 65.9 (65.5–66.3) 51.2 (50.8–51.6) 66.0 (65.4–66.7) 55.2 (54.5–55.9) 40–49 73.8 (73.4–74.1) 59.2 (58.8–59.6) 60.5 (60.1–61.0) 49.3 (48.9–49.8) 65.6 (64.9–66.4) 57.4 (56.6–58.2) 50–59 63.7 (63.2–64.1) 53.9 (53.5–54.3) 63.2 (62.7–63.8) 55.7 (55.2–56.3) 73.2 (72.4–74.0) 57.7 (57.3–58.9) 60–69 54.9 (54.5–55.4) 49.4 (48.9–49.9) 64.8 (64.2–65.5) 60.5 (59.9–61.3) 74.9 (74.0–75.8) 68.2 (66.8–68.5) ≥70 51.4 (50.9–51.9) 47.0 (46.5–47.5) 63.3 (62.4–64.1) 59.8 (59.0–60.7) 73.6 (72.5–74.6) 71.1 (70.0–72.1) A snapshot of the predicted immune landscape on April 10 (present time), June 1, and August 1, 2022 is presented in Figure 5 . As indicated in Figure 4, the susceptible proportion in the three younger age groups was estimated to decrease over the 4 months following April 10. However, it was estimated that there will be a considerable increase in the susceptible population among the older age groups. The specific values are listed in Table 1, although it should be noted that the values presented in Table 1 are the immune proportions as opposed to the susceptible proportions shown in Figure 5 (i.e., the values in Table 1 should be deducted from 100% to obtain the susceptible proportion).Figure 5 Predicted proportion of individuals susceptible to symptomatic SARS-CoV-2 infection on 10 April, 1 June 1, and August 1, 2022. The bars represent the proportion of individuals susceptible to symptomatic SARS-CoV-2 infection over time, according to the age groups 20–29, 30–39, 40–49, 50–59, 60–69, and ≥70 years, presented from left to right. The projection was performed on April 10, 2022, and the estimates afterward are from a model-based projection. While abovementioned results dealt nationwide data as a single group, independent estimate of immune fraction was also obtained for each prefecture. Figure 6 shows the prefecture-specific immune proportions grouped by epidemiological dynamics (i.e., by increasing or decreasing trend). The difference in the immune fraction against symptomatic infection between the two groups was 4.12% (95% CI: 1.39–7.64), indicating that the prefectures that had higher estimated immune proportions tended to be in the decreasing risk phase.Figure 6 The prefecture-specific immune proportion according to the trend of infection with SARS-CoV-2. The box-plot shows the estimated immune proportion on April 10, 2022 among the prefectures that had decreasing and increasing risks of infection on the left and right, respectively, from February 7 to April 10, 2022. Each dot represents the estimated fraction immune in a single prefecture, and there are in total 47 prefectures in Japan. Grouping each prefecture as decreasing or increasing in its incidence was judged by the growth rate of incidence as on April 10, 2022 taking negative or positive values, respectively. The lines inside the box represent the median, and lower and upper box boundaries are the first and third quantiles, respectively. The whiskers indicate 1.5× the interquartile range from the first and third quantiles. 4 Discussion The present study presents the COVID-19 immune landscape in Japan, accounting for waning immunity from vaccination and natural infection. We quantified the decay rate of vaccine-induced protection against symptomatic infection, showing the dramatically rapid decrease in immunity against the Omicron variant among those who received only the second vaccine dose. As such, our results show that the estimated immune proportion has substantially declined over time since vaccination coverage plateaued. As expected, the immune proportion was boosted following the third vaccine rollout, although it was estimated that some older age groups have already begun experiencing a decrease in their booster dose-induced immunity. The recent surge in infection, caused by the Omicron variant in particular, contributed to a considerable increase in the immune proportion among young age groups compared with the older population. Our results predict that by August 2022, there will be a substantial increase in the susceptible proportion among the older age groups. Lastly, we showed that the estimated immune proportion was associated with the infection risk trend by comparing the growth trend by prefecture in Japan, as a simple validation of our overall analysis framework. Although previous studies have quantified waning immunity against infectious diseases (Feng et al., 2022, Hogan et al., 2021, Khoury et al., 2021, Nishiura et al., 2006), to our knowledge, this is the first study to report estimates of the immune proportion at the population-level. We provide a framework that allows for estimation of the proportion of people that are immune to symptomatic SARS-CoV-2 infection in real time, accounting for prior immunity waning in the midst of expanding booster vaccination and a surge in infection. The half-life of the effectiveness of the second vaccine dose was estimated to be substantially shorter than that of the third dose and this dropped to <10% in 6 months. This supports the decision made by many countries to encourage people to get a booster dose within no more than 6 months after receiving their second vaccine dose. The estimated rollout rate and goal (plateaued value) of booster vaccination coverage allowed us to estimate the current immune landscape in real time and to project the future trajectory. Such estimations could be particularly useful for countries in which the booster vaccine rollout has just begun. To deal with the great uncertainty regarding the effectiveness of infection-induced immunity, we examined different scenarios regarding the waning rate of infection-induced immunity. The results showed that different immunity decay rates could provide notably different immune landscapes over time. It was suggested that if the infection-induced immunity decays rapidly in the same manner as the second vaccine dose-induced immunity, the past surge in infection will not necessarily contribute to a substantial long-term gain in the immune proportion (see Supplementary material). It was vital to account for the age group when estimating and projecting the trajectory of the immune landscape. Each age cohort experienced different vaccination coverage timings, rollout rates, and goals (plateaued values), as well as different risks of natural infection with SARS-CoV-2, and these factors collectively contributed to substantial variations in the age-specific immune proportion. The older age groups received a second vaccine dose earlier, and the vaccination coverage increased more quickly in these age groups compared with the younger age groups. They experienced many fewer cases of infection and thus had lower proportions of individuals who acquired immunity from infection, perhaps owing to risk awareness and more cautious behaviors among the individuals in these age groups, in response to the finding that older people are at greater risk of complications and death compared with younger individuals. These factors contributed to a more rapid increase in the immune proportion in the older age groups but also to a subsequent dramatic decrease. It was predicted that the immune fraction could decline to a low level of <30% among people aged ≥70 years by the summer of 2022. Given that they were at higher risk of severe outcomes, the older generation were considered to be the priority in the next round of the booster vaccination campaign, and indeed, the government of Japan implemented fourth dose booster vaccinations for the older population and for those with underlying comorbidities. By contrast, although the younger population had lower vaccination coverage, a steep upsurge in COVID-19 incidence was observed, resulting in a higher immune fraction compared with the older population. Understanding the age-specific immune landscape could aid decision-making regarding the commencement timing and target age groups of future vaccination campaigns, while providing valuable insights into the COVID-19 transmission dynamics in a country. There are some limitations in the present study. First, we did not consider the immune protection against severe illness and death, or the potentially different levels of immune response and decay rates of immunity according to age group because of a serious shortage of accumulated evidence regarding the effectiveness of this induced immunity and its duration against the Omicron variant, especially the variants-of-concern that emerged from the Omicron group (e.g., BA.4 and BA.5). Statistical estimation in this regard would be helpful, particularly when the main aim of booster vaccination is to prevent these outcomes (i.e., severe disease or death). Second, we imposed an assumption that the actual COVID-19 incidence is four times higher than the number of reported cases. Although this estimate rests on a published statistical estimate, the figure could be biased because the ascertainment rate can be expected to vary over time owing to the epidemic situation. For example, the ascertainment rate could depend on the testing and contact tracing capacity and the healthcare-seeking behaviors among the public (Sen et al., 2021); thus, the bias could have been elevated by a surge of cases with the Omicron variant. Third, although we estimated the future vaccination coverage and used these estimates to predict the immune proportions over the next 4 months, we did not make such predictions for infection-induced immunity, and thus, the fraction of the population that has immunity from naturally acquired infections should be deemed the minimum bound. The time-dependent proportion of individuals with infection-induced immunity reflects the number of individuals who were infected with SARS-CoV-2 at the time the analysis was conducted; if a substantial number of infections occur in the future, it could provide a very different immune landscape. Lastly, the lack of evidence limited our study as we did not consider those who gained immunity from both previous infections and following vaccinations, which recently appears to confer a greater neutralizing response than those who were immunized by a natural infection alone or vaccination alone. Despite the limitations described above, the present study provides a simple and easily tractable framework that enables an estimation of the proportion of the population that was immune to symptomatic SARS-CoV-2 infection during the time that the booster vaccination coverage was rapidly increasing and the country was experiencing a surge of infection with the Omicron variant. In the future, the estimated decay rate of immunity and the projected age-specific immune proportions will aid our understanding of the possible risk groups and heterogeneous transmission dynamics of COVID-19 as well as our decision-making regarding when a rollout of booster vaccination should commence and the target population. Funding sources H.N. received funding from Health and Labour Sciences Research Grants (20CA2024, 20HA2007, 21HB1002, and 21HA2016); the Japan Agency for Medical Research and Development (JP20fk0108140, JP20fk0108535, and JP21fk0108612); the Japan Society for the Promotion of Science (JSPS) KAKENHI (21H03198 and 22K19670); the Environment Research and Technology Development Fund (JPMEERF20S11804) of the Environmental Restoration and Conservation Agency of Japan; Kao Health Science Research; Daikin GAP fund program of Kyoto University; and the Japan Science and Technology Agency SICORP program (JPMJSC20U3 and JPMJSC2105) and RISTEX program for Science of Science, Technology and Innovation Policy (JPMJRS22B4). T.K. received funding from the JSPS KAKENHI (21K10495), and K.H. received funding from the JSPS KAKENHI (20K18953) and The Health Care Science Institute (IKEN). Declaration of Competing Interest All authors declare no conflicts of interest with regards to this paper. Uncited reference Centers for Disease Control and Prevention (CDC) (2022). CRediT authorship contribution statement Misaki Sasanami: . Marie Fujimoto: Validation. Taishi Kayano: . Katsuma Hayashi: . Hiroshi Nishiura: . Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments We thank the local governments, public health centers, and institutes for surveillance, laboratory testing, epidemiological investigation, and data collection. We thank Katie Oakley, PhD, from Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript. The funders had no role in the study design, data collection and analysis, the decision to publish, or preparation of this manuscript. Ethical approval statement The present study used publicly available data from the Ministry of Health, Labour and Welfare, and all data were de-identified prior to analysis. This study was approved by the Medical Ethics Board of the Graduate School of Medicine, Kyoto University (R2676). ==== Refs References N. Andrews J. Stowe F. Kirsebom S. Toffa T. Rickeard E. Gallagher C. Gower M. Kall N. Groves A.-M. O’Connell D. Simons P.B. Blomquist A. Zaidi S. Nash I.B.A. Aziz N., Thelwall, S., Dabrera, G., Myers, R., Amirthalingam, G., Gharbia, S., Barrett, J.C., Elson, R., Ladhani, S.N., Ferguson, N., Zambon, M., Campbell, C.N.J., Brown, K., Hopkins, S., Chand, M., Ramsay, M., Lopez Bernal, J., Covid-19 Vaccine Effectiveness against the Omicron (B.1.1.529) Variant New England Journal of Medicine 1–15 2022 10.1056/nejmoa2119451 Antonelli M. Penfold R.S. Merino J. Sudre C.H. Molteni E. Berry S. Canas L.S. Graham M.S. Klaser K. Modat M. Murray B. Kerfoot E. Chen L. Deng J. Österdahl M.F. Cheetham N.J. Drew D.A. Nguyen L.H. Pujol J.C. Hu C. Selvachandran S. Polidori L. May A. Wolf J. Chan A.T. Hammers A. Duncan E.L. Spector T.D. Ourselin S. Steves C.J. Risk factors and disease profile of post-vaccination SARS-CoV-2 infection in UK users of the COVID Symptom Study app: a prospective, community-based, nested, case-control study Lancet Infect Dis 22 2022 43 55 10.1016/S1473-3099(21)00460-6 34480857 Bar-On Y.M. Goldberg Y. Mandel M. Bodenheimer O. Amir O. Freedman L. Alroy-Preis S. Ash N. Huppert A. Milo R. Protection by a Fourth Dose of BNT162b2 against Omicron in Israel New England Journal of Medicine 1–9 2022 10.1056/nejmoa2201570 Bergwerk M. Gonen T. Lustig Y. Amit S. Lipsitch M. Cohen C. Mandelboim M. Levin E.G. Rubin C. Indenbaum V. Tal I. Zavitan M. Zuckerman N. Bar-Chaim A. Kreiss Y. Regev-Yochay G. Covid-19 Breakthrough Infections in Vaccinated Health Care Workers New England Journal of Medicine 385 2021 1474 1484 10.1056/nejmoa2109072 34320281 Brown C.M. Vostok J. Johnson H. Burns M. Gharpure R. Sami S. Sabo R.T. Hall N. Foreman A. Schubert P.L. Gallagher G.R. Fink T. Madoff L.C. Gabriel S.B. MacInnis B. Park D.J. Siddle K.J. Harik V. Arvidson D. Brock-Fisher T. Dunn M. Kearns A. Laney A.S. Outbreak of SARS-CoV-2 Infections, Including COVID-19 Vaccine Breakthrough Infections, Associated with Large Public Gatherings — Barnstable County, Massachusetts, July 2021 MMWR Morb Mortal Wkly Rep 70 2021 1059 1062 10.15585/mmwr.mm7031e2 34351882 CDC COVID-19 Vaccine Breakthrough Case Investigations Team COVID-19 Vaccine Breakthrough Infections Reported to CDC — MMWR Morb Mortal Wkly Rep 70 2021 792 793 34043615 Centers for Disease Control and Prevention (CDC), 2022. CDC Recommends Additional Boosters for Certain Individuals [WWW Document]. URL https://www.cdc.gov/media/releases/2022/s0328-covid-19-boosters.html#:∼:text=“Today%2C CDC expanded eligibility for,to increase their protection further. Cucinotta D. Vanelli M. WHO declares COVID-19 a pandemic Acta Biomedica 91 2020 157 160 10.23750/abm.v91i1.9397 32191675 D. Agency COVID-19 vaccination data [WWW Document] 2022 https://info.vrs.digital.go.jp/dashboard/#overview Feng, A., Obolski, U., Stone, L., He, D., 2022. Modelling COVID-19 Vaccine Breakthrough Infections in Highly Vaccinated Israel – the effects of waning immunity and third vaccination dose. medRxiv, https://doi.org/10.1101/2022.01.08.22268950. Ferdinands J.M. Rao S. Dixon B.E. Mitchell P.K. DeSilva M.B. Irving S.A. Lewis N. Natarajan K. Stenehjem E. Grannis S.J. Han J. McEvoy C. Ong T.C. Naleway A.L. Reese S.E. Embi P.J. Dascomb K. Klein N.P. Griggs E.P. Konatham D. Kharbanda A.B. Yang D.-H. Fadel W.F. Grisel N. Goddard K. Patel P. Liao I.-C. Birch R. Valvi N.R. Reynolds S. Arndorfer J. Zerbo O. Dickerson M. Murthy K. Williams J. Bozio C.H. Blanton L. Verani J.R. Schrag S.J. Dalton A.F. Wondimu M.H. Link-Gelles R. Azziz-Baumgartner E. Barron M.A. Gaglani M. Thompson M.G. Fireman B. Waning 2-Dose and 3-Dose Effectiveness of mRNA Vaccines Against COVID-19–Associated Emergency Department and Urgent Care Encounters and Hospitalizations Among Adults During Periods of Delta and Omicron Variant Predominance — VISION Network, 10 States, Aug MMWR Morb Mortal Wkly Rep 71 2022 255 263 10.15585/mmwr.mm7107e2 35176007 Gavish N. Yaari R. Huppert A. Katriel G. Population-level implications of the Israeli booster campaign to curtail COVID-19 resurgence Sci Transl Med 14 2022 10.1126/scitranslmed.abn9836 Grauer J. Löwen H. Liebchen B. Strategic spatiotemporal vaccine distribution increases the survival rate in an infectious disease like Covid-19 Sci Rep 10 2020 1 10 10.1038/s41598-020-78447-3 31913322 Haas E.J. Angulo F.J. McLaughlin J.M. Anis E. Singer S.R. Khan F. Brooks N. Smaja M. Mircus G. Pan K. Southern J. Swerdlow D.L. Jodar L. Levy Y. Alroy-Preis S. Impact and effectiveness of mRNA BNT162b2 vaccine against SARS-CoV-2 infections and COVID-19 cases, hospitalisations, and deaths following a nationwide vaccination campaign in Israel: an observational study using national surveillance data The Lancet 397 2021 1819 1829 10.1016/S0140-6736(21)00947-8 Hall, Victoria Jane, Foulkes, Sarah, Saei, Ayoub, Andrews, Nick, Oguti, Blanche, Charlett, Andre, Wellington, Edgar, Stowe, J., Gillson, Natalie, Atti, Ana, Islam, Jasmin, Karagiannis, Ioannis, Munro, Katie, Khawam, Jameel, Chand, Meera A., Brown, Colin S., Ramsay, Mary, Lopez-Bernal, Jamie, Hopkins, Susan, Andrews, N., Atti, A., Aziz, H., Brooks, T., Brown, C. S., Camero, D., Carr, C., Chand, M. A., Charlett, A., Crawford, H., Cole, M., Conneely, J., D’Arcangelo, S., Ellis, J., Evans, S., Foulkes, S., Gillson, N., Gopal, R., Hall, L., Hall, V. J., Harrington, P., Hopkins, S., Hewson, J., Hoschler, K., Ironmonger, D., Islam, J., Kall, M., Karagiannis, I., Kay, O., Khawam, J., King, E., Kirwan, P., Kyffin, R., Lackenby, A., Lattimore, M., Linley, E., Lopez-Bernal, J., Mabey, L., McGregor, R., Miah, S., Monk, E.J.M., Munro, K., Naheed, Z., Nissr, A., O’Connell, A.M., Oguti, B., Okafor, H., Organ, S., Osbourne, J., Otter, A., Patel, M., Platt, S., Pople, D., Potts, K., Ramsay, M., Robotham, J., Rokadiya, S., Rowe, C., Saei, A., Sebbage, G., Semper, A., Shrotri, M., Simmons, R., Soriano, A., Staves, P., Taylor, S., Taylor, A., Tengbe, A., Tonge, S., Vusirikala, A., Wallace, S., Wellington, E., Zambon, M., Corrigan, D., Sartaj, M., Cromey, L., Campbell, S., Braithwaite, K., Price, L., Haahr, L., Stewart, S., Lacey, E.D., Partridge, L., Stevens, G., Ellis, Y., Hodgson, H., Norman, C., Larru, B., Mcwilliam, S., Roynon, A., Northfield, J., Winchester, S., Cieciwa, P., Pai, A., Bakker, P., Loughrey, C., Watt, A., Adair, F., Hawkins, A., Grant, A., Temple-Purcell, R., Howard, J., Slawson, N., Subudhi, C., Davies, S., Bexley, A., Penn, R., Wong, N., Boyd, G., Rajgopal, A., Arenas-Pinto, A., Matthews, R., Whileman, A., Laugharne, R., Ledger, J., Barnes, T., Jones, C., Osuji, N., Chitalia, N., Bailey, T., Akhtar, S., Harrison, G., Horne, S., Walker, N., Agwuh, K., Maxwell, V., Graves, J., Williams, S., O’Kelly, A., Ridley, P., Cowley, A., Johnstone, H., Swift, P., Democratis, J., Meda, M., Brake, S., Gunn, J., Selassi, A., Hams, S., Irvine, V., Chandrasekaran, B., Forsyth, C., Radmore, J., Thomas, C., Brown, K., Roberts, S., Burns, P., Gajee, K., Lewis, T., Byrne, T.M., Sanderson, F., Knight, S., Macnaughton, E., Burton, B.J.L., Smith, H., Chaudhuri, R., Aeron-Thomas, J., Hollinshead, K., Shorten, R.J., Swan, A., Favager, C., Murira, J., Baillon, S., Hamer, S., Shah, A., Russell, J., Brennan, D., Dave, A., Chawla, A., Westwell, F., Adeboyeku, D., Papineni, P., Pegg, C., Williams, M., Ahmad, S., Horsley, A., Gabriel, C., Pagget, K., Maloney, G., Ashcroft, J., Del Rosario, I., Crosby-Nwaobi, R., Flanagan, D., Dhasmana, D., Fowler, S., Cameron, E., Prentice, L., Sinclair, C., Bateman, V., McLelland-Brooks, K., Ho, A., Murphy, M., Cochrane, A., Gibson, A., Black, K., Tempeton, K., Donaldson, S., Coke, L., Elumogo, N., Elliott, J., Padgett, D., Cross, A., Mirfenderesky, M., Joyce, S., Sinanovic, I., Howard, M., Cowling, P., Brazil, M., Hanna, E., Abdelrazik, A., Brand, S., Sheridan, E.A., Wadams, B., Lloyd, A., Mouland, J., Giles, J., Pottinger, G., Coles, H., Joseph, M., Lee, M., Orr, S., Chenoweth, H., Browne, D., Auckland, C., Lear, R., Mahungu, T., Rodger, A., Warren, S., Brooking, D., Pai, S., Druyeh, R., Smith, E., Stone, S., Meisner, S., Delgado, D., Underhill, E., Keen, L., Aga, M., Domingos, P., Gormley, S., Kerrison, C., Birch, S., DeSilva, T., Allsop, L., Ambalkar, S., Beekes, M., Jose, S., Tomlinson, J., Painter, S., Price, C., Pepperell, J., James, K., Trinick, T., Moore, L., Day, J., Boulos, A., Knox, I., Defever, E., McCracken, D., Gray, K., Houston, A., Planche, T., Pritchard Jones, R., Wycherley, D., Bennett, S., Marrs, J., Nimako, K., Stewart, B., Bain, S.C., Kalakonda, N., Khanduri, S., Ashby, A., Holden, M., Mahabir, N., Harwood, J., Payne, B., Court, K., White, N., Longfellow, R., Hughes, L.E., Green, M.E., Halkes, M., Mercer, P., Roebuck, A., Wilson-Davies, E., Gallego, L., Lazarus, R., Aldridge, N., Berry, L., Game, F., Reynolds, T., Holmes, C., Wiselka, M., Higham, A., Booth, M., Duff, C., Alderton, J., Hilton, D., Powell, J., Jackson, A., Plant, A.J., Ahmed, N., Chin, T., Qazzafi, M.Z., Moody, A.M., Tilley, R.E., Donaghy, T., O’Kane, M., Shipman, K., Sierra, R., Parmar, C., Mills, G., Harvey, D., Huang, Y.W.J., Birch, J., Robinson, L., Board, S., Broadley, A., Laven, C., Todd, N., Eyre, D.W., Jeffery, K., Dunachie, S., Duncan, C., Klenerman, P., Turtle, L., Baxendale, H., Heeney, J.L., 2021. COVID-19 vaccine coverage in health-care workers in England and effectiveness of BNT162b2 mRNA vaccine against infection (SIREN): a prospective, multicentre, cohort study. The Lancet 397, 1725–1735. https://doi.org/10.1016/S0140-6736(21)00790-X Hogan, A.B., Wu, S.L., Doohan, P., Watson, O.J., Winskill, P., Charles, G., Barnsley, G., Riley, E.M., Khoury, D.S., Ferguson, N.M., Ghani, A.C., 2021. Report 48 - The value of vaccine booster doses to mitigate the global impact of the Omicron SARS-CoV-2 variant. Jordan R.E. Adab P. Cheng K.K. Covid-19: Risk factors for severe disease and death The BMJ 368 2020 1 2 10.1136/bmj.m1198 Juthani P.V. Gupta A. Borges K.A. Price C.C. Lee A.I. Won C.H. Chun H.J. Hospitalisation among vaccine breakthrough COVID-19 infections Lancet Infect Dis 21 2021 1485 1486 10.1016/S1473-3099(21)00558-2 34506735 Kahn F. Bonander C. Moghaddassi M. Rasmussen M. Malmqvist U. Inghammar M. Björk J. Risk of severe COVID-19 from the Delta and Omicron variants in relation to vaccination status, sex, age and comorbidities - surveillance results from southern Sweden, July 2021 to January 2022 Euro Surveill 27 2022 1 6 10.2807/1560-7917.ES.2022.27.9.2200121 Khoury D.S. Cromer D. Reynaldi A. Schlub T.E. Wheatley A.K. Juno J.A. Subbarao K. Kent S.J. Triccas J.A. Davenport M.P. Neutralizing antibody levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infection Nat Med 27 2021 1205 1211 10.1038/s41591-021-01377-8 34002089 Loconsole, D., Centrone, F., Sallustio, A., Accogli, M., Casulli, D., Sacco, D., Zagaria, R., Morcavallo, C., Chironna, M., 2022. Characteristics of the First 284 Patients Infected with the SARS-CoV-2 Omicron BA . 2 Subvariant at a Single Center in the Apulia Region of Italy , January – March 2022 4–11. Lopez Bernal J. Andrews N. Gower C. Gallagher E. Simmons R. Thelwall S. Stowe J. Tessier E. Groves N. Dabrera G. Myers R. Campbell C.N.J. Amirthalingam G. Edmunds M. Zambon M. Brown K.E. Hopkins S. Chand M. Ramsay M. Effectiveness of Covid-19 Vaccines against the B.1.617.2 (Delta) Variant New England Journal of Medicine 385 2021 585 594 10.1056/nejmoa2108891 34289274 Lovell-Read F.A. Shen S. Thompson R.N. Estimating local outbreak risks and the effects of non-pharmaceutical interventions in age-structured populations: SARS-CoV-2 as a case study J Theor Biol 535 2022 10.1016/j.jtbi.2021.110983 Madhi S.A. Kwatra G. Myers J.E. Jassat W. Dhar N. Mukendi C.K. Nana A.J. Blumberg L. Welch R. Ngorima-Mabhena N. Mutevedzi P.C. Population Immunity and Covid-19 Severity with Omicron Variant in South Africa New England Journal of Medicine 2022 10.1056/nejmoa2119658 Mathieu E. Ritchie H. Ortiz-Ospina E. Roser M. Hasell J. Appel C. Giattino C. Rodés-Guirao L. A global database of COVID-19 vaccinations Nat Hum Behav 5 2021 947 953 10.1038/s41562-021-01122-8 33972767 Ministry of Health Labour and Welfare, 2022a. Notice about fourth dose vaccine. Ministry of Health Labour and Welfare Open data [WWW Document] 2022 https://www.mhlw.go.jp/stf/covid-19/open-data.html Ministry of Health Labour and Welfare Visualizing the data: information on COVID-19 infections [WWW Document] 2022 https://covid19.mhlw.go.jp/en/ Ministry of Health Labour and Welfare The situation analysis on COVID-19 and response from MHLW (2021/12/31) [WWW Document] 2021 https://www.mhlw.go.jp/stf/newpage_23137.html Ministry of Health Labour and Welfare Advisory board for COVID-19 control, 65th [WWW Document] 2021 https://www.mhlw.go.jp/stf/seisakunitsuite/bunya/0000121431_00294.html Ministry of Internal Affairs and Communications Portal Site of Official Statistics of Japan (e-Stat) [WWW Document] 2022 https://www.e-stat.go.jp/en/about Mohapatra R.K. Tiwari R. Sarangi A.K. Sharma S.K. Khandia R. Saikumar G. Dhama K. Twin combination of Omicron and Delta variants triggering a tsunami wave of ever high surges in COVID-19 cases: A challenging global threat with a special focus on the Indian subcontinent J Med Virol 94 2022 1761 1765 10.1002/jmv.27585 35014038 Nhk SARS-CoV-2 [WWW Document] accessed 9.24.22 https://www3.nhk.or.jp/news/special/coronavirus/ 2022 Nishiura H. Schwehm M. Eichner M. Still protected against smallpox? Estimation of the duration of vaccine-induced immunity against smallpox Epidemiology 17 2006 576 581 10.1097/01.ede.0000229196.41862.c2 16804475 Prime Minister’s Office of Japan About COVID-19 vaccination [WWW Document] 2022 https://www.kantei.go.jp/jp/headline/kansensho/vaccine.html Ritchie, H., Mathieu, E., Rodés-Guirao, L., Appel, C., Giattino, C., Ortiz-Ospina, E., Hasell, J., Macdonald, B., Beltekian, D., Roser, M., 2020. Coronavirus Pandemic (COVID-19) [WWW Document]. URL https://ourworldindata.org/coronavirus#coronavirus-country-profiles (accessed 4.1.22). Sanada T. Honda T. Yasui F. Yamaji K. Munakata T. Yamamoto N. Kurano M. Matsumoto Y. Kohno R. Toyama S. Kishi Y. Horibe T. Kaneko Y. Kakegawa M. Fukui K. Kawamura T. Daming W. Qian C. Xia F. He F. Yamasaki S. Nishida A. Harada T. Higa M. Tokunaga Y. Takagi A. Itokawa M. Kodama T. Kohara M. Serologic Survey of IgG Against SARS-CoV-2 Among Hospital Visitors Without a History of SARS-CoV-2 Infection in Tokyo, 2020–2021 J Epidemiol 32 2022 105 111 10.2188/jea.JE20210324 34776499 Sasanami M. Kayano T. Nishiura H. The number of COVID-19 clusters in healthcare and elderly care facilities averted by vaccination of healthcare workers in Japan, February – June 2021 Mathematical Biosciences and Engineering 19 2022 2762 2773 10.3934/mbe.2022126 35240805 Sasanami M. Kayano T. Nishiura H. Monitoring the COVID-19 immune landscape in Japan International Journal of Infectious Diseases 122 2022 300 306 10.1016/j.ijid.2022.06.005 35688309 Sen P. Yamana T.K. Kandula S. Galanti M. Shaman J. Burden and characteristics of COVID-19 in the United States during 2020 Nature 598 2021 338 341 10.1038/s41586-021-03914-4 34438440 Seppälä E. Veneti L. Starrfelt J. Danielsen A.S. Bragstad K. Hungnes O. Taxt A.M. Watle S.V. Meijerink H. Vaccine effectiveness against infection with the delta (b.1.617.2) variant, norway, april to august 2021 Eurosurveillance 26 2021 1 6 10.2807/1560-7917.ES.2021.26.35.2100793 Thompson M.G. Burgess J.L. Naleway A.L. Tyner H.L. Yoon S.K. Meece J. Olsho L.E.W. Caban-Martinez A.J. Fowlkes A. Lutrick K. Kuntz J.L. Dunnigan K. Odean M.J. Hegmann K.T. Stefanski E. Edwards L.J. Schaefer-Solle N. Grant L. Ellingson K. Groom H.C. Zunie T. Thiese M.S. Ivacic L. Wesley M.G. Lamberte J.M. Sun X. Smith M.E. Phillips A.L. Groover K.D. Yoo Y.M. Gerald J. Brown R.T. Herring M.K. Joseph G. Beitel S. Morrill T.C. Mak J. Rivers P. Harris K.M. Hunt D.R. Arvay M.L. Kutty P. Fry A.M. Gaglani M. Interim Estimates of Vaccine Effectiveness of BNT162b2 and mRNA-1273 COVID-19 Vaccines in Preventing SARS-CoV-2 Infection Among Health Care Personnel, First Responders, and Other Essential and Frontline Workers — Eight U.S. Locations, December 2020–March MMWR Morb Mortal Wkly Rep 70 2021 495 500 10.15585/mmwr.mm7013e3 33793460 Tseng H.F. Ackerson B.K. Luo Y. Sy L.S. Talarico C.A. Tian Y. Bruxvoort K.J. Tubert J.E. Florea A. Ku J.H. Lee G.S. Choi S.K. Takhar H.S. Aragones M. Qian L. Effectiveness of mRNA-1273 against SARS-CoV-2 Omicron and Delta variants Nat Med 19 2022 10.1038/s41591-022-01753-y UK Health Security Agency, 2022. Almost 80% of eligible over-75s receive spring booster [WWW Document].

0

PMC9749381

NO-CC CODE

2022-12-15 23:22:53

no

J Theor Biol. 2023 Feb 21; 559:111384

utf-8

J Theor Biol

10.1016/j.jtbi.2022.111384

oa_other

==== Front Eur Polym J Eur Polym J European Polymer Journal 0014-3057 1873-1945 Published by Elsevier Ltd. S0014-3057(22)00771-6 10.1016/j.eurpolymj.2022.111767 111767 Article Polymer Modification of SARS-CoV-2 Spike Protein Impacts its Ability to Bind Key Receptor Sharfin Rahman Monica a De Alwis Watuthanthrige Nethmi a Chandrarathne Bhagya M. a Page Richard C. a⁎ Konkolewicz Dominik a⁎ a Department of Chemistry and Biochemistry, Miami University, 651 E High St, Oxford, OH 45011, USA ⁎ Corresponding authors. 14 12 2022 14 12 2022 11176720 10 2022 9 12 2022 10 12 2022 © 2022 Published by Elsevier Ltd. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Graphical abstract The global spread of SARS-CoV-2 (severe acute respiratory syndrome coronavirus-2) has caused the loss of many human lives and severe economic losses. SARS-CoV-2 mediates its infection in humans via the spike glycoprotein. The receptor binding domain of the SARS-CoV-2 spike protein binds to its cognate receptor, angiotensin converting enzyme-2 (ACE2) to initiate viral entry. In this study, we examine how polymer modification of the spike protein receptor binding domain impacts binding to ACE2. The horseradish peroxidase conjugated receptor binding domain was modified with a range of polymers including hydrophilic N,N-dimethylacrylamide, hydrophobic N-isopropylacrylamide, cationic 3-(N,N-dimethylamino)propylacrylamide, and anionic 2-acrylamido-2-methylpropane sulfonic acid polymers. The effect of polymer chain length was observed using N,N-dimethylacrylamide polymers with degrees of polymerization of 5, 10 and 25. Polymer conjugation of the receptor binding domain significantly reduced the interaction with ACE2 protein, as determined by an enzyme-linked immunosorbent assay. Stability analysis showed that these conjugates remained highly stable even after seven days incubation at physiological temperature. Hence, this study provides a detailed view of the effect specific type of modification using a library of polymers with different functionalities in interrupting RBD-ACE2 interaction. Keywords Spike protein RAFT Grafting to Protein-polymer conjugation protein-protein interactions ==== Body pmc1 Introduction Novel coronavirus disease 2019 (COVID-19) has become a global pandemic with over 535 million cases and more than 6.3 million deaths as of June 19 , 2022. [1]Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), the causative agent of COVID-19.[2], [3] SARS-CoV-2 is an enveloped positive-sense, single-stranded RNA virus and belongs to the β-coronavirus genus in the coronavirus family.[4] It shares high genetic sequence identity with SARS-CoV and Middle East respiratory syndrome coronavirus (MERS-CoV).[4] The virus is transmitted person to person[5] and includes a diverse clinical manifestation ranging from mild cases to severe cases, with a mortality rate between 9.3% to 19.7% in 2020 of hospitalized patients.[3], [6], [7] SARS-CoV-2 consists of four structural proteins, the spike (S), envelope (E), membrane (M), and nucleocapsid (N) proteins. The S, E, and M proteins are primarily involved in virus assembly and host cell entry while the N protein is needed for RNA synthesis.[8], [9], [10] The S protein on the surface of SARS-COV-2 protein consists of two functional units which includes a receptor binding domain (RBD) containing S1 unit and the membrane fusion domain containing S2 unit.[11], [12] The infection of coronavirus into host cells is initiated by binding of the S1 RBD to the cellular surface receptor angiotensin-converting enzyme 2 (ACE2).[13] The interaction between the RBD of SARS-CoV-2 and ACE2 has been reported to be between 5 and 20 times stronger than for SARS-CoV due to significant increases in salt-bridges and hydrogen bond interactions between these proteins. [14], [15] Subsequently, the S2 unit fuses the host cell and viral membranes to enable transfer of the viral genome into the host cells. Hence, the interaction between spike RBD protein and ACE2 receptor plays a significant role for viral entry into host cells.[16], [17], [18] Therefore, inhibiting the interaction between RBD and ACE2 protein could be used to prevent SARS-CoV-2 infection.[19], [20] Control over polymer structure through reversible deactivation radical polymerization (RDRP)[21] methods such as atom transfer radical polymerization (ATRP) and reversible addition fragmentation chain transfer polymerization (RAFT) has enabled important biochemical applications and biomaterials.[22], [23], [24], [25], [26] The ability to control primary polymer structure across a range of biologically compatible functional groups is a significant strength of RAFT and ATRP. These polymers have been conjugated to various proteins for both biocatalytic and pharmaceutical applications.[27], [28], [29], [30], [31] Polymers can be attached to proteins using several strategies including grafting-to, where preformed polymers are attached to proteins, and grafting-from, where polymers are grown off the protein surface from an initiating or transfer agent.[32] Due to the changes in overall biohybrid structure upon bioconjugation, polymer bioconjugation is an effective way of modulating protein performance. Several molecules have been targeted to the S protein in vitro. The protease inhibitor camostat mesylate and cathepsin L inhibitor E-64d have been shown to block SARS-CoV and SARS-CoV-2 cellular entry.[13] SARS-CoV-2 entry into 293/hACE2 cells are also found to be reduced by a potent inhibitor of phosphatidylinositol 3-phosphate 5-kinase (PIKfyve), Apilimod.[13], [33] Moreover, SARS-CoV-2 S protein mediated cell-cell fusion has been prevented by fusion inhibitors EK1C4,[34] IPB02[35] and nelfinavir mesylate.[36] The SARS-CoV-2 S protein has also shown to be inhibited by several SARS-CoV specific neutralizing antibodies including S309, m396 and CR3022.[4], [37] However, these antiviral agents are associated with disadvantages such as toxicity, short half-life, or acute side effects. These limitations may interrupt their further application in clinical settings, creating an urge to find new and effective therapeutics to treat COVID-19. Recently, bioconjugation has been used for diagnostic, therapeutic as well as vaccine development. In diagnostic applications, one research group showed that SARS-CoV-2-based peptide conjugation with polyacrylamide polymers could be used to detect antibodies.[38] Antibodies specific for SARS-CoV-2 has also been conjugated with gold nanoparticles for SARS-CoV-2 spike protein detection.[39] Bioconjugation approaches for SARS-CoV-2 also have potential therapeutic application. One study has showed that the lipid conjugation of a peptide derived from the C-terminal heptad repeat of SARS-CoV-2 as a promising candidate for SARS-CoV-2 infection.[40] Conjugation of neutralizing antibodies with nanoparticles has also shown potential in inactivating SARS-CoV-2.[41] Additionally, bioconjugation methods can facilitate SARS-CoV-2 vaccine development.[42], [43], [44], [45] However, understanding how synthetic modification of the RBD impacts its ability to interact with the ACE2 protein is important for downstream inhibition of SARS-CoV-2 infection. If small or minor modifications causes significant reductions in RBD binding to ACE2, then developing future inhibitors that target the RBD-ACE2 interaction in-vivo could be a fruitful pathway to new therapeutics against COVID-19 infections.[46] This is because if the binding of RBD to ACE2 is sensitive to changes in the RBD structure or surface features, future therapeutics can target the RBD either through covalent or non-covalent approaches, potentially leading to inhibition of the infection cycle. Polymer modification is an excellent model system to study the impact of modification size and structure, due to the ability to fine tune polymer functionality and molecular weight, through techniques such as RAFT. In this work, the RBD of spike protein was conjugated with synthetic polymers of different hydrophobicity, charge and chain length using a ‘grafting-to’ approach. Conjugation with these polymers significantly reduced the interaction between spike RBD protein and ACE2 protein, suggesting that targeting the interactions between the spike protein and ACE2 or even covalent modification of the S protein could be a fruitful approach. Ongoing efforts to prevent coronavirus impact primarily focused on developing vaccines and therapeutics. However, developing effective therapeutics against SARS-CoV-2 requires a detailed understanding of how modification of SARS-COV-2 protein impacts its interactions with ACE2. Polymers offer an appealing platform from which functionality and size of the attached groups can be systematically varied in a facile manner. From this perspective, our study can provide valuable insight into the development of future inhibitors using downstream modification, by highlighting the sensitivity of the RBD-ACE2 binding to modification and perturbations in the RBD. Initially, a series of hydrophilic and hydrophobic polymers were synthesized to conjugate with spike protein RBD. RAFT was used with the chain transfer agent 2-(((ethylthio)-carbonothioyl)thio propionic acid (PAETC) (Scheme 1 a & 1c).[47], [48] Initially, polymers with a targeted degree of polymerization of 25 units were prepared with differing hydrophilicity i.e., hydrophilic N,N- dimethyl acrylamide (DMAm) and hydrophobic N-isopropyl acrylamide (NIPAm). Generally, hydrophilic polymers tend to form noncovalent hydrogen bond interactions with amino acid residues at the protein surface while hydrophobic polymers tend to reduce the solubility of protein.[47], [48] Additionally, to investigate the impact of charge properties of polymers on bioconjugate performance, cationic N,N-dimethyl aminopropyl acrylamide (DMAPA) and anionic 2-acrylamido-2-methylpropane sulfonic acid (AMPSA) monomers were also incorporated into the DMAm polymers in a 1:1 ratio with a targeted degree of polymerization of 25. The ionizable group containing polymers are predicted to strongly interact with the protein through electrostatics, with a smaller tendency to form hydrogen bonds with the protein.[48] Moreover, to explore the effect of chain length of polymers on the interaction between RBD and ACE2 protein, DMAm polymers were also synthesized in two additional lengths, DP5 and DP10. The synthesized polymers were characterized using 1H NMR and IR spectroscopy (Figure S1 and S2) as well as size exclusion chromatography to observe monomer conversion and molecular weight distribution respectively (Figure S3 and Table S1).Scheme 1 (a) RAFT polymerization of DMAm. (b) Conjugation of pDMAm to the HRP linked spike RBD protein via a “Grafting to” approach. (c) Respective hydrophilic DMAm, hydrophobic NIPAm, cationic DMAPA and anionic AMPSA monomers used in this study. This study used a SARS-CoV-2 Spike RBD-ACE2 blocking antibody detection (enzyme-linked immunosorbent assay (ELISA) kit. The kit provides a solution phase sample of the RBD of spike protein linked with horseradish peroxidase enzyme (RBD-HRP) and a 96-well plate coated with ACE2 protein. Each polymer was conjugated to the HRP linked RBD protein using an in situ 1-ethyl-3-dimethylamino)propyl carbodiimide, hydrochloride/N-hydroxysuccinimide (EDC/NHS) coupling reaction.[48], [49] The spike RBD domain comprises 8 lysine residues with an additional N-terminal amine group.[50] Horseradish peroxidase enzyme comprises 6 lysine residues.[51] The ratio of protein to polymer for conjugation was maintained at 1:20. The conjugation reaction is illustrated in scheme 1b. Accordingly, conjugation was confirmed using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), which showed a complete conjugation of polymer to protein (Figure S4). We also explored the conjugation efficiency to HRP alone using the similar conjugation conditions. There was negligible conjugation of the polymer to the isolated HRP protein (Figure S5), indicating that all polymers will be attached to the RBD section of RBD-HRP. To characterize the conjugates, a centrifugal ultrafiltration protocol was used to determine the conjugation efficiency by taking UV-vis spectra of conjugates before and after filtration. This technique was able to remove excess polymer from the conjugate solution, as seen by the purification of a polymer only solution at the same polymer concentration used in conjugation (Figure S6). The UV-vis spectra of all conjugates before purification were dominated by polymer end group (∼ 306 nm) indicating a substantial amount of free polymer present in the sample. In contrast, the UV-vis spectra of conjugates after purification showed substantial reduction of absorbance at 306 nm indicating removal of free polymers from the sample. Hence, the absorbance ratio, 280:306 nm was calculated to quantify the grafting density of polymers on protein surface. The ratio indicated covalent binding of ∼2 polymers on protein surface (Table S3 & Figure S7). The lysine reactivity of spike RBD protein was also determined using a workflow described in Carmali et al. (2017).[52] Following this workflow, the structure of novel coronavirus spike receptor-binding domain in complex with its receptor ACE2 (Protein Data Bank Identification: 6LZG)[50] was used to calculate the solvent accessible surface area using GETAREA. Additionally, PROPKA[53], [54] was used to estimate pKa values, and Adaptive Poisson-Boltzmann Solver[55] to generate electrostatic map. From this analysis, it was predicted that 2 of the 8 lysine residues, Lys462 and Lys386, present in the RBD are fast reacting. Of these, Lys462 is located about 25 Å from the ACE2 binding site and Lys386 is located about 37 Å from the ACE2 binding site. Therefore, it might be plausible that modification at Lys462 could lead to inhibition of interaction of the spike-RBD with ACE2 (Scheme 2 b).Scheme 2 (a)Spike RBD-ACE2 interaction inhibition via conjugation with polymers and (b)Spike receptor-binding domain complexed with its receptor ACE2 (PDB ID:6LZG). This work used ELISA to evaluate the interaction of the polymer modified RBD with ACE2. For the assay, native SARS-CoV-2 S1 RBD linked to HRP (RBD-HRP) and its conjugates with different polymers i.e., pDMAm25, pDMAm10, pDMAm5, pDMAm12.5-DMAPA12.5, pDMAm12.5-AMPSA12.5, and pNIPAm25 were incubated with sample diluent for 1 hour at 37 °C. The incubated samples were then added to microwell plates coated with ACE2 protein. The polymer modified RBD-HRP will bind with the ACE2 protein to varying degrees depending on the polymer modification of the RBD. Followed by incubation for 1 hour at 37 °C and removal of unbound RBD-HRP in and its conjugates after washing, the colorimetric HRP substrate TMB (3, 3', 5, 5'-tetramethylbenzidine) was added. The addition of TMB, followed by stop solution forms a yellow color and the measured absorbance at 450 nm is proportional to the ability of the native or modified RBD to interact with the ACE2 protein. For example, if polymer conjugation does not block the ACE2 binding surface of the RBD, it will bind with the ACE2 protein coated in the well and yield a high signal. Conversely, if polymer conjugation occludes the ACE2 binding surface, RBD will not be able to bind with ACE2 protein, hence will yield low signal (Scheme S1). Initially, the potential for free polymers in solution to inhibit the interaction of HRP linked RBD protein with ACE2 protein was evaluated. RBD-HRP protein equilibrated with the concentration of polymers that was used in their conjugation was incubated in the wells containing ACE2. The data in Figure 1 indicate that free polymers appear to enhance the interaction of RBD-HRP with ACE2, possibly because of molecular crowding effect.[56], [57] The high concentration of the polymer in solution may occupy a larger proportion of volume in solution, hence reduces the available volume of solvent for RBD-HRP protein. This may increase the effective concentration of RBD-HRP protein, favoring the association of RBD-HRP protein with ACE2 protein. An exception was observed for DMAm DP12.5-AMPSA DP12.5 polymer which may have a denaturing effect on the protein. Previous studies indicate that anionic polymers tend to unfold protein structure, which agrees with this observation.[47], [52] Figure 1 Percent relative ACE2 interaction of HRP linked spike RBD protein in presence of DMAm DP25, DMAm DP10, DMAm DP5, DMAm DP12.5-DMAPA DP12.5, DMAm DP12.5-AMPSA DP12.5 and NIPAm DP25 polymers. The percent relative ACE2 interaction is determined as %relativeACE2interaction=ACE2interactionsignalofsampleHRPactivitysignalofsampleACE2interactionsignalofwildtypeHRPactivitysignalofwildtype×100%. Each column represents the average of triplicates ± SD. The ACE2 interaction assay for the conjugates showed considerable inhibition of the binding between RBD-HRP and ACE2, thereby indicating probable occlusion of the ACE2 binding site within the RBD-HRP protein due to polymer conjugation (Figure 2 a). All systems showed at least an 80% reduction in binding efficiency between HRP linked RBD protein and ACE2, with up to 95% reduction in RBD affinity to ACE2. As a control study, an HRP activity assay was used to evaluate conjugate’s HRP catalytic performance regardless of ACE2 interaction.Figure 2 (a)Percent relative ACE2 interaction of HRP linked spike RBD protein and its conjugates with DMAm DP25, DMAm DP10, DMAm DP5, DMAm DP12.5-DMAPA DP12.5 and NIPAm DP25 polymers. (b) Percent HRP activity of HRP linked spike RBD protein and its conjugates with DMAm DP25, DMAm DP10, DMAm DP5, DMAm DP12.5-DMAPA DP12.5 and NIPAm DP25 polymers. In (a) and (b), the percent relative ACE2 interaction is determined as %relativeACE2interaction=ACE2interactionsignalofsampleHRPactivitysignalofsampleACE2interactionsignalofwildtypeHRPactivitysignalofwildtypex100%and the percent HRP activity was determined as %HRP activity= HRPactivityofsampleHRPactivityofwildtypex100%. Each column in (a) and (b) represents the average of triplicates ± SD. The HRP based control study showed that even though the binding between RBD-HRP and ACE2 was significantly altered, the HRP activity was essentially unaffected in all cases (Figure 2b). This suggests that polymer conjugation did not denature the HRP as the enzyme was still active and able to generate a robust colorimetric response. However, the conjugation with charged polymer, including cationic polymer, DMAm DP12.5-DMAPA DP12.5, and anionic polymer, DMAm DP12.5-AMPSA DP12.5 resulted in almost 40% and 65% loss of HRP activity, respectively (Figure S10). This was consistent with this polymer control assay (Figure 1) where the anionic polymer may induce unfolding of protein structure, reducing activity, and binding profiles. Additionally, chain length effects on HRP activity and RBD-HRP binding to ACE2 were probed using three different lengths pDMAm5, pDMAm10 and pDMAm25. Interestingly, the inhibitory effect on HRP linked RBD protein and ACE2 interaction becomes more prominent with longer chain lengths. However, it is important to note that a significant reduction of RBD-HRP binding to ACE2 occurs even with the smallest length of the polymer, i.e., pDMAm5, which showed greater than 85% reduction in binding of RBD-HRP with ACE 2. Since these conjugations were performed targeting the lysine residues of RBD protein, and the lysine residues predicted to be fast reacting are in close proximity with the ACE2 binding site (Lys462 and Lys386 are 25 Å and 37 Å away from the ACE2 binding site respectively), these conjugations are likely to have a prominent effect on inhibiting ACE2 interaction. To further investigate chain length effects, we wanted to conjugate the chain transfer agent, 2-(((ethylthio)-carbonothioyl)thio) propionic acid (PAETC). Unfortunately, significant protein loss was observed from the sodium dodecyl sulfate-polyacrylamide gel electrophoresis and %HRP activity assay (Figure S11a and S11b). This is most likely due to the PAETC conjugation increasing the hydrophobicity of conjugates and caused protein precipitation (Figure S11a & S11b). The likely mechanism by which the conjugates, with the exception of DMAm DP12.5-AMPSA DP12.5, act is through sterics. As identified by distance estimates, Lys462 and Lys386 are close to the ACE2 binding site of RBD. Attaching synthetic polymers at these Lys462 or Lys386 close to the ACE2 binding site, regardless of functionality, could lead to steric occlusion and reduced binding between RBD and ACE2. It is notable that the results are consistent regardless of whether charged, uncharged hydrophilic or hydrophobic functionality is present in the polymer. The inability for a wide range of free polymers at high concentration to inhibit the RBD/ACE2 interaction further suggests that specific polymer-RBD interactions were not present. Together these data support the steric occlusion mechanism. This also suggests that for downstream applications and inhibitors of RBD, a viable pathway is to attach either relatively small molecules (molecular weight ∼500) through to large antibodies near the binding site of RBD and provide steric occlusions that limit RBD and ACE2 interactions. To determine the stability of RBD-HRP protein and its conjugates with different polymers, a thermal stability approach was performed. The RBD-HRP protein and its conjugates with DMAm DP25, DMAm DP10, DMAm DP5, DMAm DP12.5-DMAPA DP12.5, DMAm DP12.5-AMPSA DP12.5 and NIPAm DP25 polymers were incubated at 37 °C for 1 hour or 7 days. Subsequently, the HRP activity assay as described earlier was performed (Figure 3 ). The RBD-HRP protein retained 97% activity after 1 hour and 94% activity after 7 days incubation. On other hand, HRP activity assay for all conjugates showed at least 93% residual activity after 1 hour incubation except DMAm DP12.5-DMAPA DP12.5 conj. (78% residual activity). Although conjugation with some functional polymers showed increased %HRP activity after 1 hour incubation compared to 0 hour incubation, this is within typical experimental variability. The residual activity of all conjugates however was at least 70% after 7 days incubation. Hence, these data indicate that the conjugates remained stable even after 7 days incubation at a physiological relevant temperature, with no systematic drift in the relative activity of the conjugates during incubation. This is consistent with earlier work, showing that polymer deconjugation is unlikely over the timescale of 7 days, when EDC based coupling is used.[58] Circular dichroism experiments were performed to monitor the secondary helical content of the conjugates (Figure S12). The secondary helical content of the conjugates is essentially the same for the native and for the respective conjugates.Figure 3 Percent HRP activity of spike RBD-HRP linked protein and its conjugates with DMAm DP25, DMAm DP10, DMAm DP5, DMAm DP12.5-DMAPA DP12.5, DMAm DP12.5-AMPSA DP12.5 and NIPAm DP25 polymers followed by incubation at 0 hour, 1 hour and 7 days. The percent relative ACE2 interaction is determined a s%HRPActivity=(HRPactivitysignalofsampleat0hourHRPactivitysignalofsampleat0hror1hror7days) x 100%. Each column represents the average of triplicates ± SD. In summary, a series of bioconjugates were synthesized by attaching a range of well-defined functional polymers i.e., hydrophilic N,N-dimethyl acrylamide, hydrophobic N-isopropyl acrylamide, cationic N,N-dimethyl aminopropyl acrylamide, and anionic 2-acrylamido-2-methylpropane sulfonic acid polymers to the RBD-HRP fusion protein. Free polymers, with exception of anionic 2-acrylamido-2-methylpropane sulfonic acid polymers, increase the binding of RBD to ACE-2 provides a higher stringency environment that may be useful for the search for small molecule compounds that disrupt the RBD-ACE2 interaction. In contrast to their free polymer analogues, bioconjugates of RBD with the water-soluble polymers displayed up to 95% reduction of RBD-HRP interactions with the ACE2 protein. with the greater inhibition observed with longer chain polymers. Additionally, these bioconjugates demonstrated significant thermal stability even after exposure to physiological conditions for extended periods. These results provide promising routes for modification of the RBD, such as those within an inactivated SARS-CoV-2 virion used for vaccines, to prevent interaction between the RBD and ACE2. Hence, our results suggest that simple covalent modification by protein-polymer conjugation can disrupt interactions that are crucial for viral life cycle, thereby showing the potentiality of bioconjugation for altering vital physiologically relevant interaction. This observation may could guide future development of inhibitors targeting RBD-ACE-2 interaction. Funding Sources This work was supported by the National Science Foundation under Grant No. (DMR-2030567). 400 MHz NMR instrumentation is supported through funding from the National Science Foundation under grant number (CHE-1919850). DK acknowledges support from Miami University through the Robert H. and Nancy J, Blayney Professorship. RCP acknowledges support from Miami University through the Ernst H. Volwiler Professorship. Data Availability Statement Raw data for this manuscript will be made available on the Miami University Scholarly Commons. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability Raw Data is Available on Miami University Scholarly Commons Acknowledgments The authors thank Dr Anne Carroll for assistance with NMR experiments and Kate Bradford for assistance with GPC experiments. We also thank Dr C. Scott Hartley for fruitful discussions. ==== Refs References: 1 World Health Organization. WHO Coronavirus (COVID-19) Dashboard. https://covid19.who.int/ (accessed 19 June 2022). 2 Sharma A. Tiwari S. Deb M.K. Marty J.L. Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2): a global pandemic and treatment strategies Int J Antimicrob Agents 56 2 2020 106054 32534188 3 Zhou P. Yang X.L. Wang X.G. Hu B. Zhang L. Zhang W. Si H.R. Zhu Y. Li B. Huang C.L. Chen H.D. Chen J. Luo Y. Guo H. Jiang R.D. Liu M.Q. Chen Y. Shen X.R. Wang X. Zheng X.S. Zhao K. Chen Q.J. Deng F. Liu L.L. Yan B. Zhan F.X. Wang Y.Y. Xiao G.F. Shi Z.L. A pneumonia outbreak associated with a new coronavirus of probable bat origin Nature 579 7798 2020 270 273 32015507 4 Tian X. Li C. Huang A. Xia S. Lu S. Shi Z. Lu L. Jiang S. Yang Z. Wu Y. Ying T. Potent binding of 2019 novel coronavirus spike protein by a SARS coronavirus-specific human monoclonal antibody Emerg Microbes Infect 9 1 2020 382 385 32065055 5 Chan J.-F.-W. Yuan S. Kok K.-H. To K.-K.-W. Chu H. Yang J. Xing F. Liu J. Yip C.-C.-Y. Poon R.-W.-S. Tsoi H.-W. Lo S.-K.-F. Chan K.-H. Poon V.-K.-M. Chan W.-M. Ip J.D. Cai J.-P. Cheng V.-C.-C. Chen H. Hui C.-K.-M. Yuen K.-Y. A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: a study of a family cluster The Lancet 395 10223 2020 514 523 6 Wu F. Zhao S. Yu B. Chen Y.M. Wang W. Song Z.G. Hu Y. Tao Z.W. Tian J.H. Pei Y.Y. Yuan M.L. Zhang Y.L. Dai F.H. Liu Y. Wang Q.M. Zheng J.J. Xu L. Holmes E.C. Zhang Y.Z. A new coronavirus associated with human respiratory disease in China Nature 579 7798 2020 265 269 32015508 7 Finelli L. Gupta V. Petigara T. Yu K. Bauer K.A. Puzniak L.A. Mortality Among US Patients Hospitalized With SARS-CoV-2 Infection in 2020 JAMA Netw Open 4 4 2021 e216556 33830226 8 Li F. Structure, Function, and Evolution of Coronavirus Spike Proteins Annu Rev Virol 3 1 2016 237 261 27578435 9 Ortiz-Prado E. Simbaña-Rivera K. Gómez- Barreno L. Rubio-Neira M. Guaman L.P. Kyriakidis N.C. Muslin C. Jaramillo A.M.G. Barba-Ostria C. Cevallos-Robalino D. Sanches-SanMiguel H. Unigarro L. Zalakeviciute R. Gadian N. López-Cortés A. Clinical, molecular, and epidemiological characterization of the SARS-CoV-2 virus and the Coronavirus Disease 2019 (COVID-19), a comprehensive literature review Diagn Microbiol Infect Dis 98 1 2020 115094 32623267 10 Schoeman D. Fielding B.C. Coronavirus envelope protein: current knowledge Virol J 16 1 2019 69 31133031 11 Rota P.A. Oberste M.S. Monroe S.S. Nix W.A. Campagnoli R. Icenogle J.P. Penaranda S. Bankamp B. Maher K. Chen M.H. Tong S. Tamin A. Lowe L. Frace M. DeRisi J.L. Chen Q. Wang D. Erdman D.D. Peret T.C. Burns C. Ksiazek T.G. Rollin P.E. Sanchez A. Liffick S. Holloway B. Limor J. McCaustland K. Olsen-Rasmussen M. Fouchier R. Gunther S. Osterhaus A.D. Drosten C. Pallansch M.A. Anderson L.J. Bellini W.J. Characterization of a novel coronavirus associated with severe acute respiratory syndrome Science 300 5624 2003 1394 1399 12730500 12 Walls A.C. Park Y.J. Tortorici M.A. Wall A. McGuire A.T. Veesler D. Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein Cell 183 6 2020 1735 33306958 13 Hoffmann M. Kleine-Weber H. Schroeder S. Kruger N. Herrler T. Erichsen S. Schiergens T.S. Herrler G. Wu N.H. Nitsche A. Muller M.A. Drosten C. Pohlmann S. SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor Cell 181 2 2020 271 280 32142651 14 Delgado J.M. Duro N. Rogers D.M. Tkatchenko A. Pandit S.A. Varma S. Molecular basis for higher affinity of SARS-CoV-2 spike RBD for human ACE2 receptor Proteins 89 9 2021 1134 1144 33864655 15 Lan J. Ge J. Yu J. Shan S. Zhou H. Fan S. Zhang Q. Shi X. Wang Q. Zhang L. Wang X. Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor Nature 581 7807 2020 215 220 32225176 16 Wu K. Peng G. Wilken M. Geraghty R.J. Li F. Mechanisms of host receptor adaptation by severe acute respiratory syndrome coronavirus J Biol Chem 287 12 2012 8904 8911 22291007 17 Huo J. Le Bas A. Ruza R.R. Duyvesteyn H.M.E. Mikolajek H. Malinauskas T. Tan T.K. Rijal P. Dumoux M. Ward P.N. Ren J. Zhou D. Harrison P.J. Weckener M. Clare D.K. Vogirala V.K. Radecke J. Moynie L. Zhao Y. Gilbert-Jaramillo J. Knight M.L. Tree J.A. Buttigieg K.R. Coombes N. Elmore M.J. Carroll M.W. Carrique L. Shah P.N.M. James W. Townsend A.R. Stuart D.I. Owens R.J. Naismith J.H. Neutralizing nanobodies bind SARS-CoV-2 spike RBD and block interaction with ACE2 Nat Struct Mol Biol 27 9 2020 846 854 32661423 18 Tai W. He L. Zhang X. Pu J. Voronin D. Jiang S. Zhou Y. Du L. Characterization of the receptor-binding domain (RBD) of 2019 novel coronavirus: implication for development of RBD protein as a viral attachment inhibitor and vaccine Cell Mol Immunol 17 6 2020 613 620 32203189 19 Du L. He Y. Zhou Y. Liu S. Zheng B.J. Jiang S. The spike protein of SARS-CoV–a target for vaccine and therapeutic development Nat Rev Microbiol 7 3 2009 226 236 19198616 20 Huo J. Zhao Y. Ren J. Zhou D. Duyvesteyn H.M.E. Ginn H.M. Carrique L. Malinauskas T. Ruza R.R. Shah P.N.M. Tan T.K. Rijal P. Coombes N. Bewley K.R. Tree J.A. Radecke J. Paterson N.G. Supasa P. Mongkolsapaya J. Screaton G.R. Carroll M. Townsend A. Fry E.E. Owens R.J. Stuart D.I. Neutralization of SARS-CoV-2 by Destruction of the Prefusion Spike Cell Host Microbe 28 3 2020 497 32910920 21 Parkatzidis K. Wang H.S. Truong N.P. Anastasaki A. Recent Developments and Future Challenges in Controlled Radical Polymerization: A 2020 Update Chem 6 7 2020 1575 1588 22 Fairbanks B.D. Gunatillake P.A. Meagher L. Biomedical applications of polymers derived by reversible addition - fragmentation chain-transfer (RAFT) Adv Drug Deliv Rev 91 2015 141 152 26050529 23 Truong N.P. Jones G.R. Bradford K.G.E. Konkolewicz D. Anastasaki A. A comparison of RAFT and ATRP methods for controlled radical polymerization Nature Reviews Chemistry 5 12 2021 859 869 24 Siegwart D.J. Oh J.K. Matyjaszewski K. ATRP in the design of functional materials for biomedical applications Prog Polym Sci 37 1 2012 18 37 23525884 25 Xu F.J. Neoh K.G. Kang E.T. Bioactive surfaces and biomaterials via atom transfer radical polymerization Progress in Polymer Science 34 8 2009 719 761 26 Boyer C. Bulmus V. Davis T.P. Ladmiral V. Liu J. Perrier S. Bioapplications of RAFT Polymerization Chem. Rev. 109 2009 5402 5436 19764725 27 Chen C. Ng D.Y.W. Weil T. Polymer bioconjugates: Modern design concepts toward precision hybrid materials Progress in Polymer Science 105 2020 101241 101281 28 Wu Y. Ng D.Y. Kuan S.L. Weil T. Protein-polymer therapeutics: a macromolecular perspective Biomater Sci 3 2 2015 214 230 26218113 29 Pelegri-O'Day E.M. Lin E.W. Maynard H.D. Therapeutic protein-polymer conjugates: advancing beyond PEGylation J Am Chem Soc 136 41 2014 14323 14332 25216406 30 Wright T.A. Page R.C. Konkolewicz D. Polymer conjugation of proteins as a synthetic post-translational modification to impact their stability and activity Polymer Chemistry 10 4 2019 434 454 31249635 31 Burridge K.M. Page R.C. Konkolewicz D. Bioconjugates – From a specialized past to a diverse future Polymer 211 2020 123062 123079 32 Averick S. Mehl R.A. Das S.R. Matyjaszewski K. Well-defined biohybrids using reversible-deactivation radical polymerization procedures J Control Release 205 2015 45 57 25483427 33 Kang, Y. L.; Chou, Y. Y.; Rothlauf, P. W.; Liu, Z.; Soh, T. K.; Cureton, D.; Case, J. B.; Chen, R. E.; Diamond, M. S.; Whelan, S. P. J.; Kirchhausen, T., Inhibition of PIKfyve kinase prevents infection by Zaire ebolavirus and SARS-CoV-2. bioRxiv 2020. 34 Xia S. Liu M. Wang C. Xu W. Lan Q. Feng S. Qi F. Bao L. Du L. Liu S. Qin C. Sun F. Shi Z. Zhu Y. Jiang S. Lu L. Inhibition of SARS-CoV-2 (previously 2019-nCoV) infection by a highly potent pan-coronavirus fusion inhibitor targeting its spike protein that harbors a high capacity to mediate membrane fusion Cell Res 30 4 2020 343 355 32231345 35 Zhu Y. Yu D. Yan H. Chong H. He Y. Design of Potent Membrane Fusion Inhibitors against SARS-CoV-2, an Emerging Coronavirus with High Fusogenic Activity J Virol 94 14 2020 e00635 e720 32376627 36 Musarrat F. Chouljenko V. Dahal A. Nabi R. Chouljenko T. Jois S.D. Kousoulas K.G. The anti-HIV drug nelfinavir mesylate (Viracept) is a potent inhibitor of cell fusion caused by the SARSCoV-2 spike (S) glycoprotein warranting further evaluation as an antiviral against COVID-19 infections J Med Virol 92 10 2020 2087 2095 32374457 37 Hussain A. Hasan A. Nejadi Babadaei M.M. Bloukh S.H. Chowdhury M.E.H. Sharifi M. Haghighat S. Falahati M. Targeting SARS-CoV2 Spike Protein Receptor Binding Domain by Therapeutic Antibodies Biomed Pharmacother 130 2020 110559 38 Ryzhov I.M. Tuzikov A.B. Nizovtsev A.V. Baidakova L.K. Galanina O.E. Shilova N.V. Ziganshina M.M. Dolgushina N.V. Bayramova G.R. Sukhikh G.T. Williams E.C. Nagappan R. Henry S.M. Bovin N.V. SARS-CoV-2 Peptide Bioconjugates Designed for Antibody Diagnostics Bioconjug Chem 32 8 2021 1606 1616 34181851 39 Silva P.B.D. Silva J.R.D. Rodrigues M.C. Vieira J.A. Andrade I.A. Nagata T. Santos A.S. Silva S.W.D. Rocha M. Bao S.N. Moraes-Vieira P.M. Proenca-Modena J. Angelim M.K.C. de Souza G.F. Muraro S.P. de Barros A.L.B. de Souza Martins G.A. Ribeiro-Dias F. Machado G. Fessel M.R. Chudzinski-Tavassi A.M. Ronconi C.M. Goncalves D. Curi R. Oliveira O.N. Azevedo R.B. Detection of SARS-CoV-2 virus via dynamic light scattering using antibody-gold nanoparticle bioconjugates against viral spike protein Talanta 243 2022 123355 40 Outlaw, V. K.; Bovier, F. T.; Mears, M. C.; Cajimat, M. N.; Zhu, Y.; Lin, M. J.; Addetia, A.; Lieberman, N. A. P.; Peddu, V.; Xie, X.; Shi, P. Y.; Greninger, A. L.; Gellman, S. H.; Bente, D. A.; Moscona, A.; Porotto, M., Inhibition of Coronavirus Entry In Vitro and Ex Vivo by a Lipid-Conjugated Peptide Derived from the SARS-CoV-2 Spike Glycoprotein HRC Domain. mBio 2020, 11 (5), e01935-20. 41 Cai X. Chen M. Prominski A. Lin Y. Ankenbruck N. Rosenberg J. Nguyen M. Shi J. Tomatsidou A. Randall G. Missiakas D. Fung J. Chang E.B. Penaloza-MacMaster P. Tian B. Huang J. A Multifunctional Neutralizing Antibody-Conjugated Nanoparticle Inhibits and Inactivates SARS-CoV-2 Adv Sci (Weinh) 9 2 2022 e2103240 34761549 42 Elia U. Ramishetti S. Rosenfeld R. Dammes N. Bar-Haim E. Naidu G.S. Makdasi E. Yahalom-Ronen Y. Tamir H. Paran N. Cohen O. Peer D. Design of SARS-CoV-2 hFc-Conjugated Receptor-Binding Domain mRNA Vaccine Delivered via Lipid Nanoparticles ACS Nano 15 6 2021 9627 9637 33480671 43 Zhou S.H. Zhang R.Y. Zhang H.W. Liu Y.L. Wen Y. Wang J. Li Y.T. You Z.W. Yin X.G. Qiu H. Gong R. Yang G.F. Guo J. RBD conjugate vaccine with a built-in TLR1/2 agonist is highly immunogenic against SARS-CoV-2 and variants of concern Chem Commun (Camb) 58 13 2022 2120 2123 35040862 44 Long Y. Sun J. Song T.Z. Liu T. Tang F. Zhang X. Ding L. Miao Y. Zhu W. Pan X. An Q. Qin M. Tong X. Peng X. Yu P. Zhu P. Xu J. Zhang X. Zhang Y. Liu D. Chen B. Chen H. Zhang L. Xiao G. Zuo J. Tang W. Zhou J. Li H. Xu Z. Zheng H.Y. Long X.Y. Qin Q. Gan Y. Ren J. Huang W. Zheng Y.T. Jin G. Gong L. CoVac501, a self-adjuvanting peptide vaccine conjugated with TLR7 agonists, against SARS-CoV-2 induces protective immunity Cell Discov 8 1 2022 9 35102138 45 Valdes-Balbin Y. Santana-Mederos D. Quintero L. Fernandez S. Rodriguez L. Sanchez Ramirez B. Perez-Nicado R. Acosta C. Mendez Y. Ricardo M.G. Hernandez T. Bergado G. Pi F. Valdes A. Carmenate T. Ramirez U. Oliva R. Soubal J.P. Garrido R. Cardoso F. Landys M. Gonzalez H. Farinas M. Enriquez J. Noa E. Suarez A. Fang C. Espinosa L.A. Ramos Y. Gonzalez L.J. Climent Y. Rojas G. Relova-Hernandez E. Cabrera Infante Y. Losada S.L. Boggiano T. Ojito E. Leon K. Chiodo F. Paquet F. Chen G.W. Rivera D.G. Garcia-Rivera D. Verez Bencomo V. SARS-CoV-2 RBD-Tetanus Toxoid Conjugate Vaccine Induces a Strong Neutralizing Immunity in Preclinical Studies ACS Chem Biol 16 7 2021 1223 1233 34219448 46 Mostafa-Hedeab G. ACE2 as Drug Target of COVID-19 Virus Treatment, Simplified Updated Review Rep Biochem Mol Biol 9 1 2020 97 105 32821757 47 Rahman M.S. Brown J. Murphy R. Carnes S. Carey B. Averick S. Konkolewicz D. Page R.C. Polymer Modification of Lipases, Substrate Interactions, and Potential Inhibition Biomacromolecules 22 2 2021 309 318 33416313 48 Wright T.A. Lucius Dougherty M. Schmitz B. Burridge K.M. Makaroff K. Stewart J.M. Fischesser H.D. Shepherd J.T. Berberich J.A. Konkolewicz D. Page R.C. Polymer Conjugation to Enhance Cellulase Activity and Preserve Thermal and Functional Stability Bioconjug Chem 28 10 2017 2638 2645 28934551 49 Falatach R. McGlone C. Al-Abdul-Wahid M.S. Averick S. Page R.C. Berberich J.A. Konkolewicz D. The best of both worlds: active enzymes by grafting-to followed by grafting-from a protein Chem Commun (Camb) 51 25 2015 5343 5346 25530198 50 Wang Q. Zhang Y. Wu L. Niu S. Song C. Zhang Z. Lu G. Qiao C. Hu Y. Yuen K.Y. Wang Q. Zhou H. Yan J. Qi J. Structural and Functional Basis of SARS-CoV-2 Entry by Using Human ACE2 Cell 181 4 2020 894 904 32275855 51 Gunnar I. Berglund, G. H. C., Andrew T. Smith, Hanna Szöke, Anette Henriksen & Janos Hajdu, The Catalytic Pathway of Horseradish Peroxidase at High Resolution. Nature 2002, 417, , 463–468 52 Cummings C.S. Campbell A.S. Baker S.L. Carmali S. Murata H. Russell A.J. Design of Stomach Acid-Stable and Mucin-Binding Enzyme Polymer Conjugates Biomacromolecules 18 2 2017 576 586 28081602 53 Sondergaard C.R. Olsson M.H. Rostkowski M. Jensen J.H. Improved Treatment of Ligands and Coupling Effects in Empirical Calculation and Rationalization of pKa Values J Chem Theory Comput 7 7 2011 2284 2295 26606496 54 Olsson M.H. Sondergaard C.R. Rostkowski M. Jensen J.H. PROPKA3: Consistent Treatment of Internal and Surface Residues in Empirical pKa Predictions J Chem Theory Comput 7 2 2011 525 537 26596171 55 Jurrus E. Engel D. Star K. Monson K. Brandi J. Felberg L.E. Brookes D.H. Wilson L. Chen J. Liles K. Chun M. Li P. Gohara D.W. Dolinsky T. Konecny R. Koes D.R. Nielsen J.E. Head-Gordon T. Geng W. Krasny R. Wei G.W. Holst M.J. McCammon J.A. Baker N.A. Improvements to the APBS biomolecular solvation software suite Protein Sci 27 1 2018 112 128 28836357 56 Minton A.P. How can biochemical reactions within cells differ from those in test tubes? J Cell Sci 128 6 2015 1254 25774053 57 Ganji M. Docter M. Le Grice S.F. Abbondanzieri E.A. DNA binding proteins explore multiple local configurations during docking via rapid rebinding Nucleic Acids Res 44 17 2016 8376 8384 27471033 58 Wright T.A. Rahman M.S. Bennett C. Johnson M.R. Fischesser H. Ram N. Tyler A. Page R.C. Konkolewicz D. Hydrolytically Stable Maleimide-End-Functionalized Polymers for Site-Specific Protein Conjugation Bioconjug Chem 32 11 2021 2447 2456 34730954

0

PMC9749382

NO-CC CODE

2022-12-15 23:22:53

no

Eur Polym J. 2022 Dec 14;:111767

utf-8

Eur Polym J

10.1016/j.eurpolymj.2022.111767

oa_other

==== Front Environ Res Environ Res Environmental Research 0013-9351 1096-0953 Elsevier Inc. S0013-9351(21)01285-8 10.1016/j.envres.2021.111990 111990 Article Underestimated impact of the COVID-19 on carbon emission reduction in developing countries – A novel assessment based on scenario analysis Wang Qiang ∗ Li Shuyu Li Rongrong Jiang Feng School of Economics and Management, China University of Petroleum (East China), Qingdao, 266580, People's Republic of China ∗ Corresponding author. 2 9 2021 3 2022 2 9 2021 204 111990111990 24 6 2021 24 8 2021 28 8 2021 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Existing studies on the impact of the COVID-19 pandemic on carbon emissions are mainly based on inter-annual change rate of carbon emissions. This study provided a new way to investigate the impact of the pandemic on carbon emissions by calculating the difference between the pandemic-free carbon emissions and the actual carbon emissions in 2020 based on scenario analysis. In this work, derived from Autoregressive Integrated Moving Average (ARIMA) method and Back Propagation Neural Network (BPNN) method, two combined ARIMA-BPNN and BPNN-ARIMA simulation approaches were developed to simulate the carbon emissions of China, India, U.S. and EU under the pandemic-free scenario. The average relative error of the simulation was about 1%, which could provide reliable simulation results. The scenario simulation of carbon emission reduction in the US and EU were almost the same as the inter-annual change rate of carbon emissions reported by the existing statistics. However, the scenario simulation of carbon emission reduction in China and India is 5% larger than the inter-annual change rate of carbon emissions reported by the existing statistics. In some sense, the impact of the pandemic on carbon emission reduction in developing countries might be underestimated. This work would provide new sight to more comprehensive understanding of the impact of the pandemic on carbon emissions. Keywords COVID-19 Carbon emissions Pandemic-free scene Artificial intelligence Developed and developing economies ==== Body pmcAbbreviations COVID-19 Corona Virus Disease 2019 ARIMA Auto-Regressive Integrated Moving Average BPNN Back Propagation Neural Network AR Auto-Regressive Model MA Moving Average Model BP Back propagation MAPE Mean Absolute Percentage Error RMSE Root Mean Square Error U.S. The United States EU The European Union 1 Introduction The Coronavirus disease 2019 (COVID-19) pandemic on a global scale has caused serious loss of life. The widespread nature of the virus forced economic activities and industrial production in many countries to stagnate in 2020, which has led to a severe decline in carbon emissions (Wang and Wang, 2020). According to a report from the Global Carbon Project (GCP) (Global Carbon Project, 2021), global carbon emissions have dropped by 2.4 GT (close to 7%) compared to 2019. This is the highest rate of decline in the past decade (International Energy Agency, 2020), which also sparked discussions about the true impact of pandemics on carbon emissions. The extent to which COVID-19 curbs carbon emissions varies from country to country (Syed and Ullah, 2021). The measurement around the decline of carbon emissions has become the forefront of environmental governance research. By summarizing the carbon dioxide emissions of different countries around the world, the global carbon dioxide emissions in the first half of 2020 dropped by 8.8% compared with the same period in 2019 (Liu et al., 2020). During the peak period, the average rate of decline in emissions of a single country even reached 26% (Le Quéré et al., 2020). Most of the current studies were based on the inter-annual change rate of carbon emissions, that was, the change in 2020 compared to 2019, which was impossible to simulate the pure effect of the epidemic. To fully measure the reduction in carbon emissions due to the pandemic, this study proposed a new analysis angle in scenario simulation. Relying on historical carbon emissions data, we intend to build an accurate forecasting model to simulate the carbon emissions under pandemic-free scene. Comparing this result with the actual 2020 carbon emissions, the difference between the two was regarded as the carbon emissions reduction caused by the pandemic. This kind of measurement could better reflect the influence of COVID-19 than the rate of annual decline in carbon emissions. China, the United States, the European Union, and India are the four economies with the largest carbon emissions in the world, and the impact of COVID-19 pandemic on them has always been the focus of scholars. The current carbon emission situation and epidemic control policies of these four economies are different, broadly representative of global economies. As typical representatives of developing economies, China and India are still in the growth stage of carbon emissions. The U.S. and EU, typical representatives of developed economies, have entered a stage of declining carbon emissions. In terms of epidemic prevention and control, China was the first country to control the spread of the epidemic. India, U.S. and EU had relatively weak risk awareness and were at different stages of epidemic control. Once the impact of the epidemic on the carbon emissions of these large economies was incorrectly estimated, it might have an adverse impact on the planning of global carbon emission reduction pathways. Therefore, this study would carry out case studies for China, India, U.S. and EU, and apply the proposed forecasting methods to the carbon emission forecasts of these four economies under the pandemic-free scenario. The rest of this paper is as follows. The second section conceived the existing literature and put forward the innovative points of this study. The third section introduced the modeling process of the forecast method. The fourth section gave the predicted process parameters and the accuracy of results. The fifth section focused on the comparison and discussion of the two analysis scenarios. Finally, section 6 summarized the main findings and gives policy recommendations. 2 Literature review 2.1 Literature review on epidemic's impact on carbon emissions China and India are the developing economies with the largest carbon emissions. The acceleration of industrialization and urbanization has led to a rapid increase in carbon dioxide emissions. The impact of the epidemic on carbon emissions in China and India has caused concern. Tollefson et al. pointed out that global carbon emissions declined in the early stage of the epidemic, but the decline was not large, and it quickly picked up. Among them, China emissions experienced a slight decline of 1.4% throughout the year (Tollefson, 2021). Han et al. estimated the real-time change of carbon dioxide based on the change of GDP and found that China's carbon emissions had a reduction of 11.0% over Q1 2019 during the worst period of the pandemic (the first quarter of 2020) (Han et al., 2021). For India, carbon emissions in 2020 was around 8% lower than the same period in 2019 (Sharma et al., 2020). The near real-time monitoring results of carbon emissions pointed out that China and India were the most affected by the epidemic among all developing countries. Their emissions in the first half of 2020 had dropped by 187.2 (3.7%) and 205.2 (15.4%) million tons respectively, compared with the same period in 2019. China's carbon emissions fell on a large scale only in the early stage of COVID-19, and began to rebound rapidly in March 2020, which was closely related to the strictness of its epidemic supervision measures. Other countries such as India only officially adopted blockade measures in March, and carbon emissions began to decline (Liu et al., 2020). Although the carbon emissions of the European Union and the United States had entered a decline stage, they were still the developed economies with the largest emissions. The impact of the epidemic on them also affected the global carbon reduction cause. Carbon emissions in Europe showed a −12.1% emissions change between January and June 2020, compared to the same period of the previous year (Andreoni, 2021). Compared to the start of the lockdown, the average decline was 11% across Europe, with France (42%), Germany (21%), the United Kingdom (13%), Spain (11%) and Italy (8%) among them (Evangeliou et al., 2021). According to different benchmark standards, the results obtained were not the same (Rugani and Caro, 2020). For example, Filimonau et al. pointed out that the carbon footprint of the United Kingdom during the blockade fell by 30% during the April to June 2020 period, by comparing it with the response time period of previous years (Filimonau et al., 2021). Research based on near-real-time activity data pointed out that among the 11 major regions of the world, the carbon emissions of the U.S. and EU were most affected by the epidemic. The U.S. emissions in the first half of 2020 fell by 338.3 million tons (13.3%) compared to the same period in 2019, while Spain, Germany, France and Italy fell by 18.8%, 15.1%, 14.2% and 13.7%, respectively. At present, scholars have conducted extensive discussions on the impact of the epidemic on the carbon emissions of large economies, but the currently collected digital information was the result of comparing a certain period in 2020 with the same period in the past. This measurement method used the past period as a baseline and reflected the current change in carbon emissions. However, assuming it to be equivalent to the true impact of the pandemic on carbon emissions may be unconvincing. 2.2 Literature review on time series forecasting technology In this research idea, how to build a scientific and accurate forecasting model becomes a supporting part. Following the development rules of existing forecasting technologies, innovating high-precision forecasting methods and applying them to predictions under no-pandemic scenarios is the key to answering the true impact of the pandemic on carbon emissions (Wang et al., 2021). Existing forecasting technologies can be divided into two categories: statistical analysis and artificial intelligence (Deb et al., 2017; Yin et al., 2017). The former was dedicated to finding the law of data changes (Jia et al., 2015; Jiang et al., 2017), while the latter took learning and correction into consideration (Raza and Khosravi, 2015). In applied research, autoregressive integrated moving average (ARIMA) model and back propagation neural network (BPNN) model had been broadly implemented in energy fields (Kaytez et al., 2015; Ye et al., 2018). Among them, the ARIMA model reflected the changing characteristics of the sequence itself by constructing a linear equation (Ding et al., 2018). It only needs endogenous variables and does not resort to other exogenous variables. But it only capture linear relationships, not nonlinear fluctuations. Therefore, unless the error was compensated, some type of error propagation must occur. The ARIMA model had excellent linear prediction capabilities. It expanded predictions based on the relationship between past and future values, so it relied heavily on historical data. The predictive effect of the ARIMA model was limited in the presence of outliers that cause data instability. The BPNN model was a multi-layer feedforward network trained by the error back propagation algorithm (Yu and Xu, 2014). The BP neural network model had strong learning ability and data fitting ability, and could handle the instability and fluctuation of the residual sequence well. It approximated nonlinear functions with arbitrary precision. Therefore, the time series forecasting model based on BP neural network could well reflect the nonlinear development trend. Most of the energy predictions made by the previous generations had focused on improving (Arora and Taylor, 2018) and upgrading the original model with some practical principles (Wong et al., 2009). For example, Parag Sen et al. (2016) used the ARIMA (0,1,4) model to predict greenhouse gas emissions for Indian iron and steel manufacturing enterprises. Barak et al. (Barak and Sadegh, 2016) proposed a model that combined ARIMA and ANFIS to predict Iran's energy consumption. With regard to BPNN model, a simulated-based neural network was adopted to predict Iranian monthly electrical consumption (Azadeh et al., 2008). Stamenkovic et al. (Stamenković et al., 2017) proposed an optimized neural network and got better results with input selection based on correlation between input variables. In the process of reviewing the literature (Yang et al., 2017), the performance of the hybrid model could be more advantageous (Qiu et al., 2016). In order to make better use of the advantages of models and make up for their shortcomings, the hybrid model that combined the ARIMA and the BPNN was a way out (Wang and Li, 2021; Wang and Petropoulos, 2016). In short, the innovations of this research include the following three aspects. First, this study measured the impact of the pandemic on carbon emissions from a more systematic perspective, that was, comparing the carbon emissions when there was no pandemic with the actual carbon emissions in 2020. Second, this research integrated advanced ARIMA and BPNN to develop two new combined high-precision prediction technologies, with the principle of “error correction + secondary modeling”. Third, this study applied the constructed new method to the prediction of pandemic-free carbon emissions in typical developing countries (China, India) and developed regions (U.S., EU). By comparing the rate of decline in carbon emissions caused by the pandemic with the rate of inter-annual decline, this study would assess whether there was an underestimation of the decline in carbon emissions in some places. 3 Methods The principles of ARIMA-BPNN and BPNN-ARIMA hybrid models were closely related to the single ARIMA and BP. The formulas for calculating a single model was showed below. Its combination of the two was shown in the flow chart. 3.1 Autoregressive integrated moving average model ARIMA (p, d, q) called differential autoregressive moving average model was the combination of Integration and ARMA model (Sowell, 1992). ARMA model was made of autoregressive model (AR) and moving average model (MA). Assuming that raw sequence is Xt={x10,x20,⋯,xm0}, and prediction sequence is Xt*={x11,x21,⋯,xm1}. On the one hand, AR model was established by the correlation (autocorrelation) between the data in the front part of itself and the data in the latter part (shown in Eqn 1):(1) xt∗=δ+φ1xt−1+φ2xt−2+⋯+φpxt−p+μt On the other hand, MA model described the relationship between current value and historical error term to smooth random fluctuations (shown in Eqn 2):(2) xt*=μ+μt+θ1μt−1+θ2μt−2+⋯+θqμt−q ARMA model was established as Eqn 3 follows.(3) xt*=δ+φ1Xt−1+φ2Xt−2+⋯+φpXt−p+μt+θ1μt−1+θ2μt−2+⋯+θqμt−q The final predicted value was calculated by the following Eqn 4 formula:(4) xt*=(1−B)dxt Where, B=[−(x11+x21)21⋮⋮−(xm−11+xm1)21]. 3.2 Back propagation neural network model The basic idea of the BP algorithm was to first give the network initial weights and thresholds, calculated the output value through forward information transfer between layers, and modified the weights and thresholds of the network according to the error between the actual output and the expected output. Through continuous repeated training and comparison, the error between the actual output and the expected output was minimized (Xiao et al., 2009). The common BP neural network was topology (Fig. 1 ). BP network was a neural network with three or more layers of neurons, including an input layer, an intermediate layer (hidden layer), and an output layer (Ren et al., 2014). The upper and lower layers were completely connected through the weight network, and there was no connection between the same layer.Fig. 1 The structure of BP Neural Network. Fig. 1 When a pair of learning samples provide input neurons (Fig. 2 ), the activation value of the neurons (the output values of the neurons in the layer) was transmitted from the input layer through the hidden layer to the output layer, and then went back through the layers to the input layer, thereby correcting each connection weight (Guo et al., 2011). The algorithm was called “error reverse propagation algorithm”, or BP algorithm.Fig. 2 Neuron topology and sigmoid function. Fig. 2 The learning process of the algorithm was composed of two processes: the forward propagation of the signal and the backward propagation of the error. In the forward propagation, the input samples were passed in from the input layer, and after being processed by the hidden layers, they were passed to the output layer. If the actual output was inconsistent with the expected output, it would enter the error back propagation stage. In back propagation, the output was transmitted back to the input layer through the hidden layer, and the error was apportioned to all units of each layer, so as to obtain the error signal of each layer unit. This error signal was used as the basis for correcting the weight of each unit (Abdi et al., 1996). 3.3 ARIMA-BPNN and BPNN-ARIMA ARIMA-BPNN and BPNN-ARIMA were formed by using error correction principle, which combined linear model-ARIMA and nonlinear model-BPNN. It was also an example of traditional statistical method and new artificial intelligence method. The hybrid models combined the advantages of ARIMA model and BPNN model and made up the deficiency. As for the ARIMA-BPNN model construction, the main idea was to use the ARIMA model for initial prediction, and then performed BPNN prediction on the residual sequence. While avoiding the shortcomings of the single model, the construction process of the ARIMA-BPNN model makes it possible to predict the future and obtain more accurate predictions. The specific operation could be divided into three steps:Step 1: Input the raw data into the ARIMA model to obtain the predicted value and subtract the original data to calculate the residual value. Step 2: Input the residual value of ARIMA model into the BPNN model and correct the residual value multiple times to obtain a new excellent residual value. Step 3: On the one hand, for the data of known years, the sum of the new residual value and the original data is used as the new predicted value. On the other hand, for the data part of the unknown year, that is, the part of the forecast data, the ARIMA prediction value minus the new residual value is used as the new forecast value. At this point, the combination is complete. In terms of the construction principle, both the first prediction and the second revision require the chosen model to be suitable for each step. We use ARIMA to predict and use BPNN to correct, establish the BPNN model of ARIMA residual sequence, rely on the superiority of ARIMA model itself and BPNN self-learning ability to obtain the predicted value of hybrid model. The first step in applying ARIMA model was to smooth the non-stationary sequence and to perform differential operations on it. The ARIMA model had good applicability in dealing with fluctuation residual sequences. In contrast, if BPNN and ARIMA were combined according to the above combination steps, using the unique advantages of the ARIMA model in dealing with fluctuations to process the residual sequence of BPNN may yield unexpected results. And then reverse the experiment. At the same time, it can verify the fitting ability of linear ARIMA model and the modified learning ability of nonlinear BPNN. According to the fitting principle, the BPNN-ARIMA hybrid model was established in this sector. The specific steps of modeling were similar to the above operations and would not be repeated here. The specific modeling process and modeling steps of the hybrid model were as shown in Fig. 3 .Fig. 3 The flowchart of the ARIMA-BPNN and BPNN-ARIMA models. Fig. 3 3.4 Prediction of carbon emissions under the pandemic-free scenario Forecasting the carbon emissions written in the 2020 pandemic-free scenario in China, India, the United States and European Union was carried out in this section. ARIMA-BPNN and BPNN-ARIMA were applied to short term forecasts. In this chapter, the data of the four economies in 1995–2019 was used to build a prediction model, aiming to obtain the carbon emissions free of pandemic in 2020 based on the historical law. The historical data sequence was from the British BP Statistical Yearbook (BP, 2020), as shown in Fig. 4 .Fig. 4 Raw data on carbon emissions in China, India, the U.S. and EU. Fig. 4 3.5 ARIMA-BPNN model parameters The first was to determine the three parameters of the ARIMA (p, d, q) model with the help of the Eviews software. "P" represented the number of lags (lags) of the time series data itself. "d" represented the order that the time series data needed to be stable. "q" represented the number of lags (lags) of the prediction error. Therefore, the first step of the ARIMA model was to determine the difference order through unit root test. ARMA required data to be normally distributed, stable and zero mean. Stationarity generally refered to: the mean was constant, the variance was constant, and the autocovariance was constant. If only the mean was non-zero, then the difference was used to make the series stationary. Here the difference order was "d". If the autocorrelation coefficient was tailed (decayed regularly in exponential or sinusoidal form), and the partial autocorrelation coefficient was p-order truncated (cut off to 0 after a certain value), then the p-order AR model was used. If the autocorrelation coefficient was q-order truncated and the partial autocorrelation coefficient was tailed, the q-order MA model was used. As mentioned before, these two historical sequences were not stable enough. When using the ARIMA model to predict carbon emissions in China, India, the U.S. and EU, the unit root tests in Table 1 indicated that the historical data sequence needed to be differentially processed. Here, Eviews software was used to calculate the correlation coefficient of the sample.Table 1 Unit root test on Eviews 7.2 about China, India, the U.S. and EU. Table 1China Sequence ADF Statistic Critical Value Value of P 1% 5% 10% Q −2.705767 −4.416345 −3.622033 −3.248592 0.2433 Q* −2.778403 −4.467895 −3.644963 −3.261452 0.2193 Q** −4.811472 −4.440739 −3.632896 −3.254671 0.0046 India Sequence ADF Statistic Critical Value Value of P 1% 5% 10% −4.394309 −3.612199 −3.243079 0.5823 Q* −4.537457 −4.416345 −3.622033 −3.248592 0.0077 Q** −9.698250 −4.440739 −3.632896 −3.254671 0.0000 U.S. Sequence ADF Statistic Critical Value Value of P 1% 5% 10% Q −2.444077 −4.394309 −3.612199 −3.243079 0.3499 Q* −4.931456 −4.440739 −3.632896 −3.254671 0.0036 Q** −8.639159 −4.467895 −3.644963 −3.261452 0.0000 EU Sequence ADF Statistic Critical Value Value of P 1% 5% 10% Q −1.731972 −4.394309 −3.612199 −3.243079 0.7052 Q* −5.414444 −4.416345 −3.622033 −3.248592 0.0012 Q** −9.351718 −4.440739 −3.632896 −3.254671 0.0000 Note: Q means zero order difference; Q* means first order difference; Q** means second order difference. After that, the model was fixed. Autocorrelation and partial autocorrelation were the basic methods for judging the trailing truncation and choosing the ARIMA model. The ARMA model required that both the autocorrelation function (ACF) and the partial autocorrelation function (PACF) were tailing (Ervural et al., 2016). Due to the randomness of the sample, the sample correlation coefficient did not exhibit a perfect case of theoretical truncation. From Fig. 5 , it could be seen that the China historical data series, whose autocorrelation coefficient was censored after the third order, and the partial autocorrelation coefficient was censored after the fifth order. The obtained model was ARIMA (3, 2, 5). In the same way, the historical data of India was input into the software to identify its ARIMA (2, 1, 5). The US and EU data series meet the parameter conditions of ARIMA (3, 1, 3) and ARIMA (5, 1, 5).Fig. 5 Correlation diagram of AC and PAC about China, India, the U.S. and EU. Fig. 5 After modeling, ARIMA (3,2,5), ARIMA (2,1,5), ARIMA (3,1,3) and ARIMA (5,1,5) were input to the SPSS software to obtain initial fitted values and predicted values. At this time, the initial prediction part of the ARIMA model was completed. The forecast data of 1995–2018 was used as fitting value to measure the accuracy of the model. The residual error sequence obtained by subtracting the fitting sequence from the original sequence was shown in Fig. 6 . It can be seen that the overall fluctuation was large, which showed that the initial prediction accuracy of the ARIMA model still had room for improvement, and its error could be further reduced. Therefore, this study used the BPNN model to correct its residuals.Fig. 6 The residual of ARIMA in China, India, the U.S. and EU. Fig. 6 Part I: Selection of network structure design and excitation function. The number of hidden layer neurons in the network was directly related to the complexity of the actual problem, the number of neurons in the input and output layers, and the setting of the expected error (Li et al., 2018). In order to control the over-fitting phenomenon, this paper gave priority to the 3-layer network (that is, there is only one hidden layer). Although the network structure shown in Fig. 7 had 4 layers, in reality, the second hidden layer had only one node. The transfer function between it and the output layer was ‘y = x’, so only the first hidden layer was valid. However, the simplified model structure should be based on ensuring the accuracy of the model. In order to reduce the error of the neural network, we appropriately increased the number of hidden layer nodes, which was easier to implement than increasing the number of hidden layers. The maximum number of iterations was set to 1000, and the prediction accuracy of the training set on the network was 10^(-8). After 40 tests, the number of hidden layer nodes was finally determined to be 10. ‘TrainRatio’ ‘ValRatio’ and ‘TestRatio’ was used to divide the sample data, of which 70% was used for training, 15% for verification and 15% for testing. In order to improve the prediction accuracy, the BPNN model added a loop statement in the network design: each predicted value was calculated based on the first four residual data.Fig. 7 The structure and algorithms of BPNN model. Fig. 7 Part II: Model Implementation. The prediction used the neural network toolbox in MATLAB to train the network. The specific implementation and correlation coefficient of the prediction model were shown in Fig. 7. The network completed the training by repeating the learning. After that, the initial residual sequence of the ARIMA model was input to predict the new residual sequence. Fig. 8 showed the calculation results. The black curve represented the original residual value produced by the ARIMA model, while the BPNN model was used to correct the residual sequence, and the resulting new residual sequence was shown as a orange curve. Comparing the original value of ARIMA with the BPNN corrected new residual, it could be seen intuitively that BPNN model could correct the original residual and alleviate the fluctuation.Fig. 8 ARIMA error sequence and BPNN corrected error sequence. Fig. 8 Take China as an example. In terms of the average of absolute error, the average absolute error of China's CO2 emissions predicted by ARIMA model was 116.1 million tons, but the average absolute error after the correction of BPNN model was only 77.4 million tons. In terms of the peak of absolute error, the peak error of China's CO2 emissions predicted by the ARIMA model appeared in 2003, which was 331.5 million tons, but the peak error after the correction of BPNN model was only 293.9 million tons (2012). For the other three economies, after the correction of the BPNN model, the forecast errors had all been converged to varying degrees. The last task was to use existing data to calculate the final predicted value of the hybrid model. Specifically, the actual residual carbon emissions were subtracted from the new residuals to obtain the fitted value of the hybrid model from 1995 to 2019, and the initial predicted value of the ARIMA model plus the residual sequence of the BPNN correction was used to obtain predictive value of 2020–2025. The difference between the final predicted value and the true value of the two models ARIMA, ARIMA-BPNN was shown in Fig. 9 . Based on the principle that the curve was closer; the model predicted better. We could see that the fitting effect of ARIMA and ARIMA-BPNN was good, but in comparison, the ARIMA-BPNN model combining the advantages of the two models had a better fitting effect.Fig. 9 The gap between forecast value and actual value for the three models in China, India, the U.S. and EU. Fig. 9 For the four economies, it could be seen from Fig. 9 that due to the impact of emergencies, the carbon emissions of U.S. and EU had been constantly fluctuating, and the carbon emissions of China and India had been rising all the way with very little fluctuation. Therefore, compared with U.S. and Europe, the overall trend of China and India was easier to identify and the fitting effect was better. Although the smaller scale made the fitting errors of U.S. and EU appear larger, the error analysis in Section 3.3 proved that the forecasting model had good performance in the carbon emission forecasts of the above four economies. 3.6 BPNN-ARIMA model parameters Similarly, the two models were used to process the data, but this time the initial model became BPNN and the residual correction model became the ARIMA model. The first step was to use the BPNN model to train and predict the raw data in China, India, the U.S. and EU. It was found that the above neural network model was still applicable to this data and was not repeated here. After the BPNN training was completed, the raw data was taken into the prediction sequence, and the original residual value obtained by subtracting the BPNN prediction sequence from the raw sequence. After that, ARIMA model was applied to correct the errors of the original residual sequence. In the ARIMA model, ARIMA (3,1,1), ARIMA (1,0,0), ARIMA (1,1,1) and ARIMA (4,1,2) were validated by unit root test, autocorrelation coefficients and partial autocorrelation coefficients. The new residual sequence was obtained in the corresponding model. Compared to the initial error sequence of the BPNN model, the ARIMA model corrected residual sequence was significantly flatter, and the corrective effect is also exerted. Using the initial predictions of BPNN model and the corrected residuals of ARIMA model, the final predictions of the BPNN-ARIMA model could be calculated. Take the original data from 1995 to 2019 minus the corrected residual sequence as the fitted value to determine the accuracy of the mixed model, and use the BPNN data plus the new residual sequence from 2020 to 2025 to obtain the final prediction sequence. To present the advantages of the hybrid model more intuitively, Fig. 10 showed the prediction results of a single model BPNN and a hybrid model BPNN-ARIMA. The correction was based on a single model, and the general trend of the hybrid model was consistent with the single model. The closer the curve was, the better the model predicted. BPNN-ARIMA combined linear and nonlinear models, and used ARIMA to modify BPNN. Among the four economies, Fig. 10 showed that the EU had the worst fitting effect, which was mainly caused by the smaller scale. In terms of the mean value and peak value of the error, the combined model had better prediction performance than the single model.Fig. 10 The gap between forecast value and actual value for the three models in China, India, the U.S. and EU. Fig. 10 3.7 Error analysis, goodness evaluation This section would respond to the above prediction process. Combined with the objectives of this paper, the error and goodness of the model was evaluated to predict the accuracy of the model. The prediction results of these four models were fitted to the actual data of carbon emissions in China, India, the U.S. and EU. The data in 1995–2019 were used as fitting samples to evaluate the proposed hybrid model. To scientifically evaluate and describe the accuracy of multiple models, we introduced the mean absolute percentage error (MAPE) to measure prediction error. As one of the tools widely used to calculate prediction errors, MAPE's measurement principle was that the smaller the value, the higher the accuracy of the model. Fig. 11 showed the prediction results for ARIMA, BPNN and two hybrid ARIMA-BPNN and BPNN-ARIMA in two economies. It could be seen intuitively that the accuracy of the four models was basically above 90%. The MAPE value below 5% proved that the prediction results of this study were convincing. In addition, the hybrid model had a clearer advantage over a single model.Fig. 11 The goodness of four models (ARIMA, ARIMA-BPNN, BPNN and BPNN-ARIMA). Fig. 11 Tables A1, A2, A3 and A4 in the appendix gave the results and deviations of the US (Table A1), China (Table A2), India (Table A3) and EU (Table A4) carbon emissions predicted using these four models in detail. It can be seen from the MAPE results that the prediction performance of the ARIMA model and the BPNN model itself was not bad, but after the residual correction, the prediction accuracy of the hybrid models ARIMA-BPNN and BPNN-ARIMA was better. From the error value, the mean absolute percentage error (MAPE) of BP-ARIMA model was about 1%, which was 0.962%, 0.715%, 0.451% and 1.359% respectively. The MAPE values of ARIMA-BP model were 1.515%, 0.829%, 0.702% and 1.075% respectively. This error result proved that the two combination models established in this study had perfect accuracy. Therefore, the predicted results were also convincing. BPNN-ARIMA had the best data processing accuracy for China, India and U.S.. The ARIMA-BPNN model showed superiority when dealing with EU data. BPNN-ARIMA based on a nonlinear BP neural network model showed better performance when dealing with fluctuation data like China. ARIMA-BPNN was better when dealing with flat data in the European Union. Root mean square error (RMSE) was often used as a measure of the prediction of machine learning model predictions. MAPE and RMSE were given in Table 2 . The measurement results shown by RMSE were consistent with the conclusions of the above MAPE.Table 2 Measurement results of three error measurement coefficients. Table 2 MAPE RMSE BP-ARIMA ARIMA-BP BP-ARIMA ARIMA-BP China 0.962% 1.515% 72 105 India 0.715% 0.829% 17 18 U.S. 0.451% 0.702% 28 57 EU 1.359% 1.075% 71 59 4 Measurement of carbon emissions affected by the pandemic After the previous chapter, the carbon emissions of the four economies under the pandemic-free state were predicted. This chapter gave the amount of carbon emissions affected in two measurement scenarios. The analysis results were discussed. 4.1 Scenario I: measurement of inter-year comparison Scenario I was to measure the amount of change in carbon emissions in the 2020 pandemic year from the perspective of inter-annual comparison. Combined with the latest statistics from authoritative databases, Fig. 12 showed the total carbon emissions of the four major economies in 2019 and 2020. The arrow showed the relative percentage change between years. According to the calculation results, compared with the carbon emissions in 2019, the reduction rate of carbon emissions in various economies in 2020 was different. Among them, China's decline was the smallest among the four economies, at 3.45%. The European Union saw the largest decline at 21.6%. The rates of decline in India and the United States were 14.8% and 16.4%, respectively.Fig. 12 Comparison of carbon emission between 2020 and 2019. Fig. 12 This measurement method gave a statistically intuitive judgment and was often referred to as the rate of decrease in carbon emissions during the pandemic year. If we want to measure the impact of the pandemic on carbon emissions, we need to use the following scenario assumptions. 4.2 Scenario II: measurement of hypothetical scenarios The core setting of the scenario method was to simulate the value of carbon emissions in an pandemic-free state. Specifically, mathematical modeling needed to be used to predict the trend of carbon emissions in 2020. Then compare the carbon emissions in this ideal state with the actual carbon emissions. The difference between the two was regarded as a quantification of the impact of the pandemic on carbon emissions. Fig. 13 showed the carbon emissions of the four economies under the pandemic-free state and the actual state. Among them, green represented the 2020 carbon emissions in an ideal state, and orange represents the actual carbon emissions in 2020. Subtracting the two, the percentage difference was regarded as the reduction in carbon emissions caused by the pandemic. It could be found that there was a significant difference between the extent of carbon emissions affected by the pandemic and the decline in carbon emissions in 2020. Numerically speaking, the pandemic had reduced China's carbon emissions by 8.8%. India, the United States and the European Union were 19.2%, 16.6% and 19.8% respectively.Fig. 13 Comparison of carbon emission between 2020 and expected 2020. Fig. 13 In summary, the first measure gave the degree of reduction in carbon emissions in 2020. The second measurement method gave the extent of changes in carbon emissions caused by the 2020 pandemic. There were obvious differences between the two measurement methods and different economies. This became the focus of our next analysis. 4.3 Discussion and summary The above gave the changes in carbon emissions from two perspectives. From a numerical point of view, there was a difference in the magnitude of change between the two. Further subdivided, this difference had different manifestations in developed and developing economies. This section focused on this difference and aimed to draw interesting conclusions. In the process of result analysis, we divided the four economies into developing and developed two groups for analysis. The first analysis was for China and India. From the data in Fig. 14 A, the relative changes in carbon emissions calculated by the two scenarios were significantly different. The reduction in China's carbon emissions affected by the pandemic was 8.8%, far exceeding the 3.4% decline in 2020. There was a 5.35% gap between the two. Similar to China, India's carbon emissions decline in Scenario II was 19.2%, while the inter-annual decline was 14.8%. The difference between the two scenarios was 4.4%. In other words, the extent of changes in carbon emissions affected by the pandemic in the two economies had been greatly underestimated.Fig. 14 A. The relative changes in carbon emissions calculated by the two scenarios (China and India). Fig. 14B. The relative changes in carbon emissions calculated by the two scenarios (U.S. and EU). Fig. 14 Comparing the results of these two economies, it was easy to find that the impact of the epidemic on India's carbon emissions was far greater than that of China. The reason was closely related to the epidemic control measures of the two countries. China adopted strict measures at the beginning of the epidemic and became the first country to control the spread of the epidemic. China's infrastructure was also more complete, and the living conditions of residents were better. Therefore, the ability to withstand the risks of emergencies such as the epidemic was also stronger. China had restored its normal economic order earlier, so its impact on carbon emissions was therefore relatively small (Wang and Zhang, 2021). India announced its victory over the epidemic too early, and the whole people relaxed their vigilance. The huge population size increased the risk of transmission, and the low vaccination rate made it difficult for people to resist virus infection. Moreover, the living conditions of the bottom people in India were extremely poor, making it difficult to withstand the harm caused by the epidemic. The epidemic had a serious impact on the Indian economy, and as a result, carbon emissions had fallen sharply. Followed by the two regions of the United States and the European Union (as ahown in Fig. 14B). As developed economies that had been more severely affected by the pandemic, the carbon emissions of the United States and the European Union had fallen sharply in 2020 by 16.4% and 21.6%, respectively, lower than in 2019. Among the calculation methods used in this study, the reduction in carbon emissions due to the sudden event of the pandemic was 16.6% and 19.8%, respectively. It could be seen that the results obtained by these two calculation methods did not show obvious differences. Therefore, in measuring the reduction in carbon emissions, the statistical results of the two typical developed economies, the United States and the European Union, were not much different from the actual impact. This confirmed that the extent of changes in carbon emissions affected by the pandemic in developed economies had basically been accurately measured. Comparing the results of these two economies, the impact of the epidemic on the carbon emissions of U.S. and EU was similar. This was related to the relatively similar epidemic control strategies of these two economies. Early risk awareness in EU and U.S. was relatively weak. The U.S. was the country with the largest number of deaths. At present, some regions were still hesitant to vaccinate, and the epidemic had not yet been effectively controlled. Therefore, the economic development caused by the epidemic had been stagnant for a long time, and its impact on carbon emissions was also great, and the decline was much greater than that of China. 5 Conclusion Statistical information from different scholars confirmed the unified conclusion that the pandemic had curbed carbon emissions in many regions of the world (Carbon Brief, 2020). However, the existing methods for measuring the reduction in carbon emissions based on inter-annual comparisons cannot accurately measure the impact of the pandemic on carbon emissions. To fully calculate the decline of carbon emissions caused by the pandemic, this study presented a measurement method based on scenario simulation. Specifically, we predicted the carbon emissions in 2020 without pandemic situation and compared this value with the actual carbon emissions in 2020. The resulting reduction rate was considered as the effect induced by the pandemic. We innovatively developed ARIMA-BP and BP-ARIMA hybrid prediction technology. By comparing with historical data, we found that the MAPE of the two prediction techniques was less than 1%. The high accuracy of the method confirmed the reliability of the carbon emission prediction under the scenario of no pandemic. After comparison, results showed that the specific impact of the pandemic on carbon emissions was greatly underestimated in developing economies. In developed economies, this misappraisal phenomenon was not obvious. In terms of the magnitude of the decline, the decline in carbon emissions in developing economies during the pandemic was underestimated by about 5%. In other words, the reduction in carbon emissions based on inter-year comparisons cannot accurately reflect the specific impact of the pandemic on developing economies. The actual impact was far more serious than reported. Therefore, this study puts forward the following recommendations. (1) This study provides a new perspective to study the impact of the COVID-19 pandemic on carbon emissions. We hope that the public and policy makers will have a more systematic understanding of the impact of emergencies on carbon emissions, rather than being limited to inter-year comparisons. (2) Judging from the results of this study, the epidemic has caused a greater decline in carbon emissions from developing economies such as China and India. Compared with developed economies, developing economies have greater pressure to reduce carbon emissions, so in the post-epidemic period, the changes in carbon emissions in developing economies should be given greater attention. (3) Developed economies are more resilient and the epidemic has less impact on them. Therefore, they should provide more assistance to developing economies, especially in the field of carbon emission reduction. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix Table A1 U.S. actual carbon emissions and ARIMA, ARIMA-BPNN, BPNN and BPNN-ARIMA predictions Table A1Year Actual data ARIMA (3,1,3) ARIMA-BPNN BPNN BPNN-ARIMA Forecasts APE (%) Forecasts APE (%) Forecasts APE (%) Forecasts APE (%) 1995 5228.00 5228.00 0.00 5228.00 0.00 5228.00 0.00 5228.00 0.00 1996 5407.87 5304.26 1.92 5407.87 0.00 5407.87 0.00 5397.74 0.19 1997 5483.18 5456.29 0.49 5483.18 0.00 5483.18 0.00 5468.79 0.26 1998 5524.20 5503.45 0.38 5524.20 0.00 5524.20 0.00 5506.06 0.33 1999 5574.14 5620.89 0.84 5568.69 0.10 5498.73 1.35 5552.72 0.38 2000 5740.77 5642.47 1.71 5764.23 0.41 5689.98 0.88 5700.54 0.70 2001 5650.72 5751.70 1.79 5643.66 0.13 5640.02 0.19 5602.78 0.85 2002 5672.40 5648.46 0.42 5660.58 0.21 5649.15 0.41 5624.75 0.84 2003 5737.87 5784.53 0.81 5795.88 1.01 6033.67 5.16 5688.26 0.86 2004 5838.97 5704.73 2.30 5868.63 0.51 5760.78 1.34 5828.00 0.19 2005 5873.13 5808.42 1.10 5800.07 1.24 5814.98 0.99 5846.88 0.45 2006 5795.10 5849.95 0.95 5682.07 1.95 5784.09 0.19 5763.63 0.54 2007 5884.15 5893.48 0.16 5884.19 0.00 5853.78 0.52 5853.16 0.53 2008 5699.11 5895.21 3.44 5793.21 1.65 5617.33 1.43 5666.75 0.57 2009 5289.14 5617.23 6.20 5424.98 2.57 5251.73 0.71 5252.14 0.70 2010 5485.72 5345.33 2.56 5359.05 2.31 5444.41 0.75 5449.64 0.66 2011 5336.44 5369.56 0.62 5373.39 0.69 5326.09 0.19 5301.05 0.66 2012 5089.97 5088.27 0.03 5043.85 0.91 5214.69 2.45 5058.30 0.62 2013 5249.60 5231.73 0.34 5248.95 0.01 5206.71 0.82 5231.03 0.35 2014 5254.57 5234.04 0.39 5272.63 0.34 5198.19 1.07 5235.67 0.36 2015 5141.41 5068.80 1.41 5132.28 0.18 5109.12 0.63 5124.47 0.33 2016 5042.43 5054.01 0.23 5001.14 0.82 4810.31 4.60 5029.85 0.25 2017 4983.87 5039.68 1.12 4953.58 0.61 4978.63 0.11 4964.02 0.40 2018 5116.79 4940.52 3.44 5201.74 1.66 5108.54 0.16 5106.97 0.19 2019 4964.69 4918.47 0.93 4952.77 0.24 4999.05 0.69 4961.87 0.06 MAPE / / 1.34 / 0.70 / 0.99 / 0.45 Table A2 China actual carbon emissions and ARIMA, ARIMA-BPNN, BPNN and BPNN-ARIMA predictions Table A2Year Actual data ARIMA (3,2,5) ARIMA-BPNN BPNN BPNN-ARIMA Forecasts APE (%) Forecasts APE (%) Forecasts APE (%) Forecasts APE (%) 1995 3028.83 3028.83 0.00 3048.39 0.65 3028.83 0.00 3028.83 0.00 1996 3177.72 3177.72 0.00 3313.75 4.28 3177.72 0.00 3187.86 0.32 1997 3166.80 3339.62 5.46 3200.72 1.07 3166.80 0.00 3181.30 0.46 1998 3163.37 3171.24 0.25 3086.08 2.44 3163.37 0.00 3182.64 0.61 1999 3294.43 3145.54 4.52 3179.11 3.50 3203.23 2.77 3318.42 0.73 2000 3360.87 3498.91 4.11 3251.08 3.27 3601.26 7.15 3371.92 0.33 2001 3523.08 3465.45 1.64 3412.49 3.14 3827.24 8.63 3582.51 1.69 2002 3843.40 3708.47 3.51 3861.16 0.46 3764.82 2.04 3920.25 2.00 2003 4532.15 4200.60 7.32 4480.85 1.13 4464.24 1.50 4596.55 1.42 2004 5334.89 5129.32 3.85 5220.99 2.13 5275.90 1.11 5422.80 1.65 2005 6098.18 6181.62 1.37 6112.63 0.24 6245.40 2.41 6183.67 1.40 2006 6677.29 6743.73 1.00 6592.49 1.27 6666.03 0.17 6768.22 1.36 2007 7239.76 7076.40 2.26 7191.52 0.67 7022.85 3.00 7314.99 1.04 2008 7378.25 7655.32 3.76 7672.11 3.98 7379.35 0.01 7451.88 1.00 2009 7710.06 7486.81 2.90 7785.29 0.98 7674.27 0.46 7797.54 1.13 2010 8143.44 8073.49 0.86 8200.51 0.70 8069.04 0.91 8213.36 0.86 2011 8824.31 8785.85 0.44 9009.92 2.10 8727.32 1.10 8889.55 0.74 2012 9001.26 9283.22 3.13 9085.40 0.93 9913.68 10.14 9065.09 0.71 2013 9244.00 9275.34 0.34 9329.52 0.93 9294.99 0.55 9332.03 0.95 2014 9239.86 9325.68 0.93 9341.08 1.10 9353.35 1.23 9268.45 0.31 2015 9185.99 9283.20 1.06 9228.58 0.46 9277.66 1.00 9274.78 0.97 2016 9137.63 9137.83 0.00 9120.83 0.18 9242.90 1.15 9239.32 1.11 2017 9297.99 9327.71 0.32 9294.68 0.04 9365.84 0.73 9401.50 1.11 2018 9507.11 9509.46 0.02 9449.19 0.61 9582.34 0.79 9610.22 1.08 2019 9825.80 9686.77 1.41 9667.91 1.61 9902.25 0.78 9930.80 1.07 MAPE / / 2.02 / 1.52 / 1.91 / 0.96 Table A3 India actual carbon emissions and ARIMA, ARIMA-BPNN, BPNN and BPNN-ARIMA predictions Table A3Year Actual data ARIMA (2,1,5) ARIMA-BPNN BPNN BPNN-ARIMA Forecasts APE (%) Forecasts APE (%) Forecasts APE (%) Forecasts APE (%) 1995 773.08 773.08 0.00 773.08 0.00 773.08 0.00 776.71 0.47 1996 811.90 808.17 0.46 811.90 0.00 811.90 0.00 814.49 0.32 1997 854.11 849.46 0.54 854.11 0.00 854.11 0.00 857.02 0.34 1998 894.41 896.37 0.22 894.41 0.00 894.41 0.00 897.64 0.36 1999 911.26 940.57 3.22 895.61 1.72 923.30 1.32 914.81 0.39 2000 959.03 961.09 0.21 952.31 0.70 971.13 1.26 968.31 0.97 2001 967.57 997.85 3.13 966.38 0.12 979.45 1.23 977.20 1.00 2002 1019.03 1026.05 0.69 1013.86 0.51 1029.53 1.03 1028.88 0.97 2003 1062.19 1057.24 0.47 1052.12 0.95 1071.71 0.90 1071.75 0.90 2004 1114.38 1115.69 0.12 1106.41 0.72 1107.73 0.60 1123.82 0.85 2005 1203.63 1177.89 2.14 1185.81 1.48 1209.41 0.48 1206.13 0.21 2006 1253.70 1263.51 0.78 1228.86 1.98 1255.34 0.13 1262.11 0.67 2007 1366.44 1338.90 2.02 1350.56 1.16 1362.20 0.31 1373.32 0.50 2008 1466.56 1428.55 2.59 1441.98 1.68 1477.78 0.76 1471.12 0.31 2009 1596.24 1557.24 2.44 1567.85 1.78 1587.53 0.55 1608.06 0.74 2010 1660.65 1690.45 1.79 1656.55 0.25 1656.22 0.27 1663.85 0.19 2011 1735.15 1763.22 1.62 1777.57 2.44 1749.88 0.85 1740.59 0.31 2012 1848.13 1806.41 2.26 1887.81 2.15 1952.36 5.64 1862.50 0.78 2013 1929.35 1930.37 0.05 1928.24 0.06 2040.99 5.79 1984.20 2.84 2014 2083.54 2049.58 1.63 2047.01 1.75 2085.05 0.07 2142.04 2.81 2015 2149.38 2150.44 0.05 2144.92 0.21 2149.97 0.03 2158.77 0.44 2016 2242.89 2259.87 0.76 2246.25 0.15 2245.39 0.11 2252.20 0.41 2017 2329.82 2354.10 1.04 2336.53 0.29 2333.24 0.15 2340.30 0.45 2018 2452.50 2438.21 0.58 2462.20 0.40 2440.34 0.50 2463.72 0.46 2019 2480.35 2542.37 2.50 2474.47 0.24 2484.09 0.15 2484.91 0.18 MAPE / / 1.25 / 0.83 / 0.88 / 0.72 Table A4 EU actual carbon emissions and ARIMA, ARIMA-BPNN, BPNN and BPNN-ARIMA predictions Table A4Year Actual data ARIMA (5,1,5) ARIMA-BPNN BPNN BPNN-ARIMA Forecasts APE (%) Forecasts APE (%) Forecasts APE (%) Forecasts APE (%) 1995 4068.84 4068.84 0.00 4068.84 0.00 4068.84 0.00 4068.84 0.00 1996 4187.33 4093.22 2.25 4187.33 0.00 4187.33 0.00 4170.30 0.41 1997 4113.00 4179.19 1.61 4113.00 0.00 4113.00 0.00 4092.13 0.51 1998 4131.69 4162.53 0.75 4131.69 0.00 4131.69 0.00 4102.03 0.72 1999 4072.67 4127.85 1.36 4138.94 1.63 4013.80 1.45 4041.45 0.77 2000 4081.01 4077.01 0.10 4041.50 0.97 3961.59 2.93 4012.64 1.68 2001 4154.02 4116.41 0.91 4126.61 0.66 3979.78 4.19 4089.81 1.55 2002 4136.62 4111.11 0.62 4109.19 0.66 4155.60 0.46 4083.29 1.29 2003 4240.35 4173.91 1.57 4162.61 1.83 4187.46 1.25 4218.46 0.52 2004 4262.18 4169.68 2.17 4216.80 1.06 4176.38 2.01 4148.44 2.67 2005 4256.20 4261.74 0.13 4271.91 0.37 4326.53 1.65 4172.52 1.97 2006 4293.56 4232.94 1.41 4290.44 0.07 4149.94 3.35 4127.38 3.87 2007 4225.88 4205.20 0.49 4223.95 0.05 3984.35 5.72 4047.45 4.22 2008 4146.62 4226.32 1.92 4197.88 1.24 4058.96 2.11 4106.15 0.98 2009 3830.27 4058.62 5.96 3931.78 2.65 3811.06 0.50 3790.52 1.04 2010 3922.93 3820.67 2.61 3911.07 0.30 3850.07 1.86 3908.78 0.36 2011 3800.36 3835.36 0.92 3851.43 1.34 3807.97 0.20 3774.34 0.68 2012 3737.70 3793.16 1.48 3776.40 1.04 3535.80 5.40 3655.84 2.19 2013 3653.47 3727.33 2.02 3461.12 5.26 3584.55 1.89 3556.99 2.64 2014 3445.59 3505.26 1.73 3415.80 0.86 3508.57 1.83 3496.46 1.48 2015 3486.94 3464.47 0.64 3463.48 0.67 3496.29 0.27 3458.54 0.81 2016 3498.50 3389.05 3.13 3409.84 2.53 3505.11 0.19 3508.34 0.28 2017 3527.15 3493.92 0.94 3530.52 0.10 3490.30 1.04 3513.58 0.38 2018 3466.48 3424.19 1.22 3402.15 1.86 3456.42 0.29 3453.44 0.38 2019 3330.44 3328.82 0.05 3273.55 1.71 3464.38 4.02 3417.38 2.61 MAPE / / 1.44 / 1.07 / 1.70 / 1.36 Acknowledgement This work is supported by 10.13039/501100001809 National Natural Science Foundation of China (Grant No. 71874203). ==== Refs References Abdi H. A widrow–hoff learning rule for a generalization of the linear auto-associator J. Math. Psychol. 40 1996 175 182 Andreoni V. Estimating the European CO2 emissions change due to COVID-19 restrictions Sci. Total Environ. 769 2021 145115 33477052 Arora S. Taylor J.W. Rule-based autoregressive moving average models for forecasting load on special days: a case study for France Eur. J. Oper. Res. 266 2018 259 268 Azadeh A. A simulated-based neural network algorithm for forecasting electrical energy consumption in Iran Energy Pol. 36 2008 2637 2644 Barak S. Sadegh S.S. Forecasting energy consumption using ensemble ARIMA–ANFIS hybrid algorithm Int. J. Electr. Power Energy Syst. 82 2016 92 104 Bp Statistical Review of World Energy 2020 Brief Carbon Temporary Reduction in CO2 Emissions during COVID-19 2020 Deb C. A review on time series forecasting techniques for building energy consumption Renew. Sustain. Energy Rev. 74 2017 902 924 Ding S. Time-aware cloud service recommendation using similarity-enhanced collaborative filtering and ARIMA model Decis. Support Syst. 107 2018 103 115 Ervural B.C. Model estimation of ARMA using genetic algorithms: a case study of forecasting natural gas consumption Proce. Soc. Behav. Sci. 235 2016 537 545 Evangeliou N. Changes in black carbon emissions over Europe due to COVID-19 lockdowns 21 2021 2675 2692 Filimonau V. The carbon footprint of a UK University during the COVID-19 lockdown Sci. Total Environ. 756 2021 143964 33261879 Global Carbon Project Coronavirus Causes ‘record Fall’ in Fossil-Fuel Emissions in 2020 2021 Guo Z.-h. A case study on a hybrid wind speed forecasting method using BP neural network Knowl. Base Syst. 24 2011 1048 1056 Han P. Assessing the recent impact of COVID-19 on carbon emissions from China using domestic economic data Sci. Total Environ. 750 2021 141688 32835964 International Energy Agency Global Energy Review: CO2 Emissions in 2020 2021 Jia S. BP neural network based localization for a front-wheel drive and differential steering mobile robot 2015 IEEE International Conference on Information and Automation. IEEE 2015 2270 2274 Jiang G. A precise positioning method for a puncture robot based on a PSO-optimized BP neural network algorithm Appl. Sci. 7 2017 969 Kaytez F. Forecasting electricity consumption: a comparison of regression analysis, neural networks and least squares support vector machines Int. J. Electr. Power Energy Syst. 67 2015 431 438 Le Quéré C. Temporary reduction in daily global CO2 emissions during the COVID-19 forced confinement Nat. Clim. Change 10 2020 647 653 Li J.-c. A link prediction method for heterogeneous networks based on BP neural network Phys. Stat. Mech. Appl. 495 2018 1 17 Liu Z. Near-real-time monitoring of global CO2 emissions reveals the effects of the COVID-19 pandemic Nat. Commun. 11 2020 5172 33057164 Qiu M. Application of artificial neural network for the prediction of stock market returns: the case of the Japanese stock market Chaos, Solit. Fractals 85 2016 1 7 Raza M.Q. Khosravi A. A review on artificial intelligence based load demand forecasting techniques for smart grid and buildings Renew. Sustain. Energy Rev. 50 2015 1352 1372 Ren C. Optimal parameters selection for BP neural network based on particle swarm optimization: a case study of wind speed forecasting Knowl. Base Syst. 56 2014 226 239 Rugani B. Caro D. Impact of COVID-19 outbreak measures of lockdown on the Italian Carbon Footprint Sci. Total Environ. 737 2020 139806 32492608 Sen P. Application of ARIMA for forecasting energy consumption and GHG emission: a case study of an Indian pig iron manufacturing organization Energy 116 2016 1031 1038 Sharma S. Effect of restricted emissions during COVID-19 on air quality in India Sci. Total Environ. 728 2020 138878 32335409 Sowell F. Modeling long-run behavior with the fractional ARIMA model J. Monetary Econ. 29 1992 277 302 Stamenković L.J. Prediction of nitrogen oxides emissions at the national level based on optimized artificial neural network model Air Q. Atmos. Health 10 2017 15 23 Syed F. Ullah A. Estimation of economic benefits associated with the reduction in the CO2 emission due to COVID-19 Environ. Challeng. 3 2021 100069 Tollefson J.J.N. COVID Curbed Carbon Emissions in 2020-but Not by Much 2021 Wang Q. Li S. Nonlinear Impact of COVID-19 on Pollution – Evidence from Wuhan vol. 65 2021 Sustainable Cities and Society New York, Milan, Madrid 102629 Bandra, London, Tokyo and Mexico City Wang Q. Uncovering the impact of the COVID-19 pandemic on energy consumption: new insight from difference between pandemic-free scenario and actual electricity consumption in China J. Clean. Prod. 2021 127897 Wang Q. Wang S. Preventing carbon emission retaliatory rebound post-COVID-19 requires expanding free trade and improving energy efficiency Sci. Total Environ. 746 2020 141158 32745860 Wang Q. Zhang F. What does the China's economic recovery after COVID-19 pandemic mean for the economic growth and energy consumption of other countries? J. Clean. Prod. 295 2021 126265 33589853 Wang X. Petropoulos F. To select or to combine? The inventory performance of model and expert forecasts Int. J. Prod. Res. 54 2016 5271 5282 Wong W.K. Stitching defect detection and classification using wavelet transform and BP neural network Expert Syst. Appl. 36 2009 3845 3856 Xiao Z. BP neural network with rough set for short term load forecasting Expert Syst. Appl. 36 2009 273 279 Yang S. Distributed formation control of nonholonomic autonomous vehicle via RBF neural network Mech. Syst. Signal Process. 87 2017 81 95 Ye H. Modeling energy-related CO2 emissions from office buildings using general regression neural network Resour. Conserv. Recycl. 129 2018 168 174 Yin L. A calculation method for CO2 emission in utility boilers based on BP neural network and carbon balance Energy Procedia 105 2017 3173 3178 Yu F. Xu X. A short-term load forecasting model of natural gas based on optimized genetic algorithm and improved BP neural network Appl. Energy 134 2014 102 113

34481817

PMC9749383

NO-CC CODE

2022-12-15 23:22:53

no

Environ Res. 2022 Mar 2; 204:111990

utf-8

Environ Res

10.1016/j.envres.2021.111990

oa_other

==== Front Int J Biol Macromol Int J Biol Macromol International Journal of Biological Macromolecules 0141-8130 1879-0003 Elsevier B.V. S0141-8130(22)02992-0 10.1016/j.ijbiomac.2022.12.090 Review Non-anticoagulant heparin derivatives for COVID-19 treatment Cao Min 1 Qiao Meng 1 Sohail Muhammad ⁎ Zhang Xing ⁎ School of Food Science and Pharmaceutical Engineering, Nanjing Normal University, Wenyuan Road 1, Nanjing 210023, China ⁎ Corresponding authors. 1 M.C. and M.Q. contributed equally. 14 12 2022 14 12 2022 25 9 2022 26 11 2022 9 12 2022 © 2022 Elsevier B.V. All rights reserved. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The ongoing pandemic of COVID-19, caused by the infection of SARS-CoV-2, has generated significant harm to the world economy and taken numerous lives. This syndrome is characterized by an acute inflammatory response, mainly in the lungs and kidneys. Accumulated evidence suggests that exogenous heparin might contribute to the alleviation of COVID-19 severity through anticoagulant and various non-anticoagulant mechanisms, including heparanase inhibition, chemokine and cytokine neutralization, leukocyte trafficking interference, viral cellular-entry obstruction, and extracellular cytotoxic histone neutralization. However, the side effects of heparin and potential drawbacks of administering heparin therapy need to be considered. Here, the current heparin therapy drawbacks were covered in great detail: structure-activity relationship (SAR) mystery, potential contamination, and anticoagulant activity. Considering these unfavorable effects, specific non-anticoagulant heparin derivatives with antiviral activity could be promising candidates to treat COVID-19. Furthermore, a structurally diverse library of non-anticoagulant heparin derivatives, constructed by chemical modification and enzymatic depolymerization, would contribute to a deeper understanding of SAR mystery. In short, targeting non-anticoagulant mechanisms may produce better therapeutic effects, overcoming the side effects in patients suffering from COVID-19 and other inflammatory disorders. Keywords Heparin derivatives COVID-19 treatment Non-anticoagulant heparin derivatives ==== Body pmc1 Introduction COVID-19 outbreak, caused by SARS-CoV-2, has been affecting the public health and the global economy since 2019 [1], [2], [3]. Due to the high mortality and spread rate, COVID-19 has become an unprecedented humanitarian crisis in modern history with >600 million infections and 6.47 million deaths reported at the time of writing. Investigators are trying to unleash effective treatment strategies against COVID-19, halting its spread and boosting public health. Vaccinations are undermined by variants and do not give immediate and 100 % protection [4], [5], [6], [7], [8], [9]. Undoubtedly, developing a novel therapeutic is a complicated and time-consuming process. So, repurposing the accustomed and approved therapies with demonstrated safety profiles is an exciting strategy. Heparin has been used for over 100 years as a well-tolerated anticoagulant drug, and biological clinical research began in 1935, it was effective and then widely used. Heparan sulfate (HS) is a co-receptor used by many viruses to invade cells, which allows for a localized increase in viral particle concentration to boost the infection rate [10]. A SARS-CoV strain, obtained from a severely infected patient, was 50 % reduced by 100 μg/mL exogenous heparin treatment in vitro Vero cells test [11]. Also, there is a growing body of literature that explains how heparin works against SARS-CoV-2. However, almost all heparin used in clinical practice is from animal sources at present. Although the natural extraction process is strictly controlled to avoid some side effects, there are still certain problems in clinical application, e.g., potential contamination, and anticoagulant activity. Furthermore, because of the heterogeneity, heparin is quantitated by anticoagulant potency, and structure-activity relationship still is not clear. Hence, these adverse effects should be considered when heparin is a candidate drug for anti-COVID-19. Not surprisingly, non-anticoagulant heparin derivatives, providing direct antiviral activity without anticoagulant side effects, have been attracting considerable interest. Furthermore, chemical or chemoenzymatic approach offers a feasible method to construct a structurally diverse library of non-anticoagulant heparin derivatives, fostering the discovery of functions and structure-activity relationship, as well as the development and application of heparin-based new drugs for COVID-19 treatment. This review aims to critically interpret the literature concerning the demerits of heparin therapy for COVID-19, including structure-activity relationship (SAR) mystery, potential contamination, and anticoagulant activity. Besides, non-anticoagulant heparin derives were discussed, which may avoid the problems of heparin therapy with equivalent therapeutic effects for COVID-19. Moreover, the structurally diverse library of non-anticoagulant heparin derivatives was described, and constructed by chemical modification and enzymatic depolymerization, which may contribute to a deeper understanding of SAR mystery, accelerating the development of precise structures for specific treatments. To date, there is no review about non-anticoagulant heparin derivatives for COVID-19 treatment, which are promising candidates for the said purpose. 2 SARS-CoV-2 and heparin 2.1 Blocking the invasion of SARS-CoV-2 with heparin It would be more beneficial in preventing hospitalization and long-term sequelae of infection by inhibiting viral replication in the initial stages of infection as opposed to treating the symptoms brought on by immune activation and inflammation. For this reason, the cell entry of SARS-CoV-2 has emerged as an appealing and repurposing target for COVID-19, and the development of direct-acting antiviral compounds has been a research priority. The research showed that binding of SARS-CoV-2 spike protein (S glycoprotein) onto cell surface HSPG and angiotensin-converting enzyme 2 (ACE2) is generally the first step required for the initiation of infection, which involves a cascade of interaction [12]. Accordingly, exogenous heparin, or HS, as a competitive inhibitor, competes with the HS chain of HSPG on the cell surface to bind with viral S protein for virus entry inhibition (Fig. 1 ) [13], [14].Fig. 1 Exogenous heparin/HS inhibits COVID-19 infection by competitively blocking the viral entry into the host cell. Fig. 1 2.2 Preclinical evidence for heparin as a SARS-CoV-2 antiviral drug In preclinical studies of viral infections, heparin possesses a direct antiviral activity upon SARS-CoV-2. The early reports indicated that heparin could bind to the proteolytic cleavage site of the S1 and S2 protein based on combined surface plasmon resonance (SPR) and computational study [15], [16]. Zhang et al. found that heparin, HS, and non-anticoagulant low molecular weight heparin at micromolar doses could effectively compete with spike glycoprotein of COVID-19 to block the viral invasion pathway [17]. Mycroft-West et al. demonstrated that heparin (200 μg/mL) inhibited SARS-CoV-2 invasion of Vero cells up to 80 % [18]. Linhardt et al. illustrated that unfractionated heparin (UFH), enoxaparin, 6-O-desulfated UFH, and 6-O-desulfated enoxaparin efficiently neutralized pLV-S particles which pseudotyped SARS-CoV-2 SGP [19]. Boons et al. identified a common octasaccharide sequence (KD = 38 nM) as the most potent in preventing the S protein-heparin interaction [20]. Overall, based on accumulated evidence, heparin and non-anticoagulant heparin derivates may be promising drug candidates for COVID-19 treatment. 2.3 Structure and function of heparin Heparin or HS, is linear sulfated hetero-polysaccharides, consisting of alternating α1-4-linked D-glucosamines (GlcN) and 1-4-linked uronic acids (UA), →GlcN→UA→. The UA residue is either α-L-iduronic acid (IdoA) or β-D-glucuronic acid (GlcA), and these residues could be sulfated at 2-O-position [21]. In addition, one or more modifications on the GlcN residue, include N-sulfation, N-acetylation, 6-O-sulfation, and 3-O-sulfation (Fig. 2 ) [22]. Heparin, a unique form of HS, has a higher level of sulfation and IdoA content than HS due to extensive sulfate modification during its biosynthesis, and displays prominent anticoagulant activity making it a commonly used anticoagulant drug [23], [24]. Furthermore, heparin, a unique class of pharmaceuticals, has effective antidotes available to make its use safer. Heparin or HS, exhibiting a wide range of biological functions, e.g. inflammatory responses, blood coagulation, tumor metastasis, lipid metabolism and so on, is an attractive synthetic target [25]. The clinical implementation of heparin or HS as an anti-Cov-19 drug demands management of various potential problems, including the anticoagulant effect.Fig. 2 The major repeating disaccharide unit in heparin (a) and HS (b). Fig. 2 3 The current drawbacks of heparin therapy 3.1 SAR mystery The chain length and structure of heparin extracted through porcine intestinal mucosa, which is permitted in most countries, are highly heterogeneous. [25]. The structure of the obtained heparin is influenced by the method of extraction and purification, and the type of animal species and organs. Accordingly, due to the uncontrolled degradation site and degree, the compositional structures of different batches of products are not uniform, resulting in the variable activities and inconsistent structure-activity relationship [26] (Table 1 ). Accordingly, there are extensive descriptions about discrepancies in the antiviral potencies of the heparin and heparin analogues in the literature. Tandon et al. reported IC50 values of 5.99, 1.08 × 103, 1.77, and 5.86 × 103 μg/L for UFH, enoxaparin, 6-O-desulfated UFH, and 6-O-desulfated enoxaparin, respectively [19]; Tree et al. reported antiviral IC50 values of 41, 7.8 × 103 μg/mL for UFH and enoxaparin, respectively [27]; Gasbarri et al., found that UFH and enoxaparin could not be effective up to 1 × 106 μg/L [28]. In terms of the absolute values of the dissociation constants, the values reported by the Linhardt, Boons, and Tan research teams vary significantly. They suggested that the discrepancy between the KD values might be due to the method of analysis and/or the experimental materials used in the studies [29]. In a word, it is now well established from a variety of studies that different descriptions exist concerning the antiviral properties of heparin and heparin analogues. As a universal antiviral drug, homogeneous heparin would be desirable.Table 1 Antiviral activity of heparin mimicking compounds against SARS-CoV-2. Table 1Compound IC50a (μg/L) KDb (M) Ref. UFH 5.99 [19] Enoxaparin 1.08 × 103 6-O-desulfated UFH 1.77 6-O-desulfated enoxaparin 5.86 × 103 UFH 4.1 × 104 [27] Enoxaparin 7.8 × 106 UFH >1 × 106 [28] Enoxaparin >1 × 106 Heparin 7.3 × 10−11 [19] Heparin 6.4 × 10−8 [20] Heparin 1.6 × 10−5 [29] Fondaparinux 2.8 × 10−5 a 50 % inhibitory concentration. b Equilibrium dissociation constant (KD = Koff / Kon) of SARS-CoV-2-S-trimer. 3.2 Potential contamination Heparin, isolated from dog liver, was discovered by Jay McLean in 1916 [30], then was approved by the FDA in 1939 and used in Europe and the United States. Until the 1990s, bovine-derived heparin gradually disappeared from the markets due to the outbreak of BSE in Europe [31]. In 2013, FDA issued the “Industrial Guidelines for Quality Control of Crude Heparin Products”, which stipulated that crude heparin is derived from pigs and cannot be mixed with heparin derived from ruminants [32]. However, potential viruses or impurity contaminations still exist for obtaining heparin derived from animals, leading to concerns about the clinical application of heparin [33]. In addition, artificial pollution cannot be ignored. Unfortunately, serious adverse events occurred between 2007 and 2008, as a result of heparin supplies being contaminated with oversulfated polysaccharides [34], [35], [36]. More stringent testing requirements, such as the proton nuclear magnetic resonance spectroscopy for the identification of compound and the implementation of a new potency assay as additional tests for impurities, were introduced by the U.S. Food and Drug Administration and other national authorities to address this problem [37], [38]. Although the issues have been partially addressed by the stringent testing, some contamination cannot be detected by current detection methods because some low-level contamination was below the sensitivity of the assay [39]. Notably, the variability of natural heparin isolated from animals may be affected by environmental factors, animal species, and organ tissue, leading to commercial heparin being complex and heterogeneous mixtures. Accordingly, finding alternatives to heparin derived from animals is clearly needed. 3.3 Non-animal source heparin with homogeneous structure Synthetic heparin can eliminate the adverse effects as noted above, caused by inherently heterogeneous heparin derived from natural sources, and the positive effects of UFH could be further enhanced in HS compounds with optimized oligosaccharide structure, leading to an increase in anti-inflammatory and antiviral properties over animal-derived heparin. Chemical method is still the major way of preparing homogeneous oligosaccharides. Target-oriented, modular, combinatorial, one-pot, and solid-phase syntheses, as well as other chemical synthetic techniques, have all been developed to generate HS/heparin oligosaccharides that range from di- to octasaccharides of various sequences and sulfation patterns. These long and elaborate chemical synthesis processes need for extremely specialized techniques, adding cost and decreasing overall yields [40], [41]. Furthermore, chemoenzymatic syntheses of various HS/heparin derivatives have been realized by the characterization of HS biosynthetic enzymes. It is an alternative technique to catalyze uridine diphosphate N-acetylglucosamine (UDP-GlcNAc) and uridine diphosphate glucuronic acid (UDP-GlcA) using the glycosyltransferase PmHS2 (Pasteurella multocida heparosan synthase 2) to synthesize sugar chains, followed by sugar chain modification (N-deacetylation, N-sulfation, O-sulfation, etc.) to construct a well-defined heparan sulfate [42], [43], [44]. Chemoenzymatic synthesis, which integrates the flexibility of chemical synthesis with the high efficiency of enzymatic catalysis, is still a step-by-step synthesis, accompanied by relatively complicated processes of enzyme separation and purification. Thus, the synthesis is usually not amenable to large-scale commercial heparin preparation. More, heparosan, an E. coli K5 strain fermentation product that is composed of a [→4) β-D-glucuronic acid (GlcA) (1→4) N-acetyl-α-D-glucosamine (GlcNAc) (1→]n repeating disaccharide unit, serves as a starting material, after which the backbone is modified by enzymatic or chemical reactions [43], [45], [46], [47]. The modification involves a series of cascade reactions, mimicking in vivo heparin biosynthesis, including the conversion of N-acetyl heparosan into an N-acetyl, N-sulfo heparosan, C5-epimerization, 2-O-sulfation, 6-O-sulfation, and 3-O-sulfation (Fig. 3 ). The approach is reasonably simple to scale up and can avoid duplicate stages, but complete control over polymer size (chain length), composition, and sequence must be given up.Fig. 3 (a) Chemical synthesis of structurally defined oligosaccharides, where R and R' are protecting groups, and LG is leaving group; (b) chemoenzymatic synthesis assembly, and (c) semisynthetic production of heparin and heparan sulfate polysaccharides starting from heparosan. Fig. 3 3.4 Side-effect: excessive anticoagulant activity The sulfated nature confers heparin with high negative charge density, enabling it to engage strongly and selectively with an immense number of proteins [48]. It interacts most classic with the serine protease inhibitor antithrombin-III (AT3), which presents anticoagulant activity. A pentasaccharide sequence motif determines the interaction of heparin-AT binding, →GlcNS6S→GlcA→GlcNS3S6S→IdoA2S→GlcNS6S→ [49]. In addition, heparin's global anticoagulant activity is dependent on the formation of the ternary heparin-AT-FIIa complex, which is facilitated by a repeating trisulfated disaccharide sequence, →IdoA2S→GlcNS6S→, corresponding to heparin's thrombin or factor IIa, FIIa binding site (Table 2 ) [43]. Interestingly, heparin, providing a continuous anticoagulant effect with a recycling effect, can effortlessly separate from the heparin/AT-III complex and bind to extra AT-III. However, the properties, which may make heparin have a pivotal role in the management of COVID-19, are often hindered by the presence of anticoagulant oligosaccharides, which leads to a significant risk of bleeding [50].Table 2 Anticoagulation structure in heparin, and reported non-anticoagulant heparin derivatives. Table 2Anticoagulation structure in heparin Pentasaccharidea Image 1 Repeating trisulfated disaccharideb Image 2 Compound Structure Anticoagulant activityc Function Ref. Non-anticoagulant version of heparin (NACH) Image 3 Anti-Xa potency: 6 U/mg Antibody immune responses up-regulation [56] 2-O,3-O-desulfated heparin (ODSH) Image 4 Heparin  Anti-Xa: 165–190 U/mg  Anti-IIa: 165–190 U/mg ODSH  Anti-Xa: 1.9 U/mg  Anti-IIa: 1.2 U/mg Anti-inflammatory [57], [58] N-desulfated/acetylated heparin (NAH) Image 5 Heparin  Anti-Xa: 191.8 ± 5.7  Anti-IIa: 183.0 ± 8.9 NAH  Anti-Xa: 20.3 ± 1.8  Anti-IIa: 9.4 ± 0.8 Anti-apoptotic activities [59] 6-O desulfated heparin (6-OdeSH) Image 6 Heparin  Anti-Xa: 191.8 ± 5.7  Anti-IIa: 183.0 ± 8.9 6-OdeSH  Anti-Xa: 20.3 ± 0.9  Anti-IIa: 33.1 ± 1.8 Anti-apoptotic activities [59] 2,6-de-O-sulfated heparin (2,6-OdeSH) Image 7 Heparin  Anti-Xa (IC50): 0.93 μg/mL  Anti-IIa (IC50): 1.03 μg/mL 2,6-OdeSH  Anti-Xa (IC50): NI  Anti-IIa (IC50): NI Anti-metastasis/cancer [60] Ultralow molecular weight glycol-split heparin (Gs-hepULMWH) Mn: 2.828 kDa UFH  Anti-Xa: 163 U/mg  Anti-IIa: 163 U/mg Gs-hepULMWH  Anti-Xa: 0 U/mg  Anti-IIa: 0 U/mg Hearanase inhibition [61] Non-anticoagulant heparin-carrying polystyrene (NAC-HCPS) Image 8 Coagulation time  APTT-Control: 30 s  Heparin: 800 s  NAC-HCPS: 160 s Angiogenesis and tumor invasion inhibition [62] Roneparstat Image 9 Mw: 1.5– 2.5 kDa – Anti-cancer and heparanase inhibition [63] Non-anticoagulant oxidized ultra-LMWH (NA-LMWH) Image 10 Control  Anti-Xa: 0.02 U/mL  Anti-IIa: 0.02 U/mL Enoxaparin (10 mg/kg, SC)  Anti-Xa: 1.2 U/mL  Anti-IIa: 0.6 U/mL NA-LMWH (10 mg/kg, SC)  Anti-Xa: 0.02 U/mL  Anti-IIa: 0.02 U/mL Anti-metastatic [64] NI, no inhibition of coagulation up to 100 μg/mL compounds. a The antithrombin III-binding pentasaccharide. b The thrombin or factor IIa, FIIa binding repeating trisulfated disaccharide. c The anticoagulation tests of relevant studies were different, and we have tried our best to keep the data comparable. 4 Non-anticoagulant heparin derivatives Given that heparin is a treatment for COVID-19, there is merit in conducting clinical studies that assess therapeutic doses of heparin in patients with COVID-19 in terms of the risk of bleeding. Heparin is thought to exert many of its non-anticoagulant actions through binding and modulating proteins, including cytokines, growth factors, adhesion, cytotoxic peptides, and tissue-degrading enzymes, responding to the treatment of clinical events associated with COVID-19, including neutralization of inflammatory chemokines, and cytokines; neutralization of extracellular cytotoxic histones and by interfering with leukocyte trafficking [51]. Since heparin plays versatile biological roles in the life process, the related mechanism is unclear for COVID-19 patients. Therefore, its therapeutic potential must be explored by clinical studies that focus on identifying and obtaining non-anticoagulant heparin derivatives and further elucidating the structure and mechanisms of these non-anticoagulant heparin derivatives. 4.1 The potential antiviral mechanisms of non-anticoagulant heparin derivatives A considerable literature suggests the potential non-anticoagulant mechanisms underlying the treatment of COVID-19 patients with non-anticoagulant heparin derivatives (Fig. 4 ) [52], [53], [54], [55], which includes: 1. Heparanase inhibition (HPSE). The loss of endothelial barrier function in the case of COVID-19, which leads to pulmonary oedema and proteinuria, can be partially attributed to increased HPSE activity-degrading the endothelial glycocalyx. Notably, non-anticoagulant heparin derivatives, proved to be potent HPSE inhibitors, may benefit COVID-19 patients by preventing glycocalyx dysfunction. 2. Chemokine and cytokine neutralization. COVID-19 is related to the generation of elevated levels of pro-inflammatory cytokines. Non-anticoagulant heparin derivatives are able to bind to the majority of chemokines and cytokines to neutralize. 3. Leukocyte trafficking interference. Non-anticoagulant heparin derivatives could attenuate the process of leukocyte adhesion and migration, having a pivotal role in the inflammatory response of COVID-19. 4. Viral cellular-entry obstruction. As outlined, heparin and heparin derivatives could inhibit the binding of SARS-CoV-2 to cells. The relevant mechanism is not directly related to the anticoagulant mechanism, indicating that non-anticoagulant heparin derivatives might have such a function. 5. Extracellular cytotoxic histone neutralization. It has been shown that negative-charged heparin neutralizes the cytotoxic effect of positive-charged histones, potentially reducing organ damage in COVID-19 patients. Hence, non-anticoagulant heparin derivatives may serve the same purpose, circumventing the drawbacks of the excessive anticoagulant activity of heparin.Fig. 4 The potential mechanisms of non-anticoagulant heparin derivatives for the treatment of COVID-19 patients. Fig. 4 4.2 Reported non-anticoagulant heparin derivatives Non-anticoagulant heparin derivatives have attracted the attention of researchers, and recent investigations demonstrated their functions without any significant impact on coagulation (Table 2) [56], [57], [58], [59], [60], [61], [62], [63], [64]. Various non-anticoagulating heparin derivatives, including non-anticoagulant versions of heparin (NACH), 2-O,3-O-desulfated heparin (ODSH), N-desulfated/acetylated heparin (NAH), 6-O-desulfated heparin (6-OdeSH), 2,6-de-O-sulfated heparin (2,6-OdeSH), ultralow molecular weight glycol-split heparin (Gs-hepULMWH), non-anticoagulant heparin-carrying polystyrene (NAC-HCPS), Roneparstat, and non-anticoagulant oxidized ultra-LMWH (NA-LMWH), have been reported which differ from heparin regarding functional domains. For instance, 3-O-sulfation, although it is rare, is essential to form a specific pentasaccharide domain for heparin, being responsible for binding to anti-thrombin with high affinity that is crucial for anti-coagulant activity [65]. The report showed that removing N-sulfoglucosamine (GlcNS3S) C-3 sulfate from pentasaccharide resulted in a 105-fold reduction in binding affinity, and almost maintained its affinity for COVID-19 [19]. Although heparin is known to benefit in COVID-19, the balance between its benefits and risks should be taken into account. Non-anticoagulant heparin derivatives can also be a choice to explore in COVID-19 treatment, circumventing the risk of the bleeding complications. 5 Outlook and conclusion COVID-19 is not only an enormous public health burden, but has significantly impacted civil societies and the global economy. There is a mountain of preclinical evidence demonstrating the advantages of heparin therapy for SARS-CoV-2 infection. In this review, the inherent harms of heparin and the potential challenges in the implementation of heparin were described. 1) SAR mystery: preclinical trials are starting to show some results of discrepancies in the antiviral potencies of heparin and heparin analogues due to the heterogeneous structures. 2) Potential contamination: heparin, sourced from animals, has a chance to be contaminated with bioactive entities, viruses, or prions, presenting safety issues. 3) Side-effect (excessive anticoagulant activity): bleeding complications due to the anticoagulant effect of heparin have been limiting their clinical implementation in treating viral infections. Given the co-existence of clinical benefits and safety concerns, efforts should be focused on maximizing therapeutic effects while minimizing the side effects of heparin. Well-defined, heparin-derived compounds could be beneficial in terms of safe production routes to replace animal-sourced heparin, providing a novel class of compounds for vital therapeutic applications. Non-anticoagulant heparin derivatives can be synthesized for direct antiviral activity by controlling the length, sulfation degree, and sulfation position of heparin. Such efforts could contribute to the development of anti-viral drugs that could be effective against SARS-CoV-2 and other unforeseeable viruses. CRediT authorship contribution statement All authors wrote the manuscript, read and approved the manuscript. Declaration of competing interest We declare that we have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgement This work was supported by grants from the 10.13039/501100012166 National Key Research and Development Program (2021YFC2100100), 10.13039/501100004608 Natural Science Foundation of Jiangsu Province (BK20200728), National Natural Science Foundation, China (22007049), and Postgraduate Research & Practice Innovation Program of Jiangsu Province to Meng Qiao (KYCX21_1402). ==== Refs References 1 Lee D. Moy N. Tritter J. Paolucci F. The COVID-19 pandemic: global health policy and technology responses in the making Health Policy Technol. 9 4 2020 397 398 10.1016/j.hlpt.2020.10.001 33024656 2 Huang C. Wang Y. Li X. Ren L. Zhao J. Hu Y. Zhang L. Fan G. Xu J. Gu X. Cheng Z. Yu T. Xia J. Wei Y. Wu W. Xie X. Yin W. Li H. Liu M. Xiao Y. Gao H. Guo L. Xie J. Wang G. Jiang R. Gao Z. Jin Q. Wang J. Cao B. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China The Lancet 395 10223 2020 497 506 10.1016/S0140-6736(20)30183-5 3 Coronavirus disease (COVID-2019) situation reports. Geneva: World Health Organization https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports/ 2020 4 Han P. Li L. Liu S. Wang Q. Zhang D. Xu Z. Han P. Li X. Peng Q. Su C. Huang B. Li D. Zhang R. Tian M. Fu L. Gao Y. Zhao X. Liu K. Qi J. Gao G.F. Wang P. Receptor binding and complex structures of human ACE2 to spike RBD from omicron and delta SARS-CoV-2 Cell 185 4 2022 630 640 10.1016/j.cell.2022.01.001 e10 35093192 5 Garcia-Beltran W.F. Lam E.C. St Denis K. Nitido A.D. Garcia Z.H. Hauser B.M. Feldman J. Pavlovic M.N. Gregory D.J. Poznansky M.C. Sigal A. Schmidt A.G. Iafrate A.J. Naranbhai V. Balazs A.B. Multiple SARS-CoV-2 variants escape neutralization by vaccine-induced humoral immunity Cell 184 9 2021 2372 2383 10.1016/j.cell.2021.03.013 e9 33743213 6 Rauchman COVID: treatment trials are still urgent Nature 594 7861 2021 26 10.1038/d41586-021-01470-5 7 Karim S. Oliveira T.D. New SARS-CoV-2 variants-clinical, public health, and vaccine implications New. Engl. J. Med. 384 19 2021 1866 1868 10.1056/NEJMc2100362 33761203 8 Ho C.Y. Salimian M. Hegert J. O'Brien J. Choi S.G. Ames H. Morris M. Papadimitriou J.C. Mininni J. Niehaus P. Burke A. Canbeldek L. Jacobs J. LaRocque A. Patel K. Rice K. Li L. Johnson R. LeFevre A. Blanchard T. Shaver C.M. Moyer A. Drachenberg C. Postmortem assessment of olfactory tissue degeneration and microvasculopathy in patients with COVID-19 JAMA Neurol. 79 6 2022 544 553 10.1001/jamaneurol.2022.0154 35404378 9 Huang B. Dai L. Wang H. Hu Z. Yang X. Tan W. Gao G.F. Serum sample neutralisation of BBIBP-CorV and ZF2001 vaccines to SARS-CoV-2 501Y.V2 The Lancet Microbe 2 7 2021 e285 10.1016/S2666-5247(21)00082-3 10 De Pasquale V. Quiccione M.S. Tafuri S. Avallone L. Pavone L.M. Heparan sulfate proteoglycans in viral infection and treatment: a special focus on SARS-CoV-2 Int. J. Mol. Sci. 22 12 2021 6574 10.3390/ijms22126574 34207476 11 Vicenzi E. Canducci F. Pinna D. Mancini N. Carletti S. Lazzarin A. Bordignon C. Poli G. Clementi M. Coronaviridae and SARS-associated coronavirus strain HSR1 Emerg. Infect. Dis. 10 3 2004 413 418 10.3201/eid1003.030683 15109406 12 Clausen T.M. Sandoval D.R. Spliid C.B. Pihl J. Perrett H.R. Painter C.D. Narayanan A. Majowicz S.A. Kwong E.M. McVicar R.N. Thacker B.E. Glass C.A. Yang Z. Torres J.L. Golden G.J. Bartels P.L. Porell R.N. Garretson A.F. Laubach L. Feldman J. Yin X. Pu Y. Hauser B.M. Caradonna T.M. Kellman B.P. Martino C. Gordts P. Chanda S.K. Schmidt A.G. Godula K. Leibel S.L. Jose J. Corbett K.D. Ward A.B. Carlin A.F. Esko J.D. SARS-CoV-2 infection depends on cellular heparan sulfate and ACE2 Cell 183 4 2020 1043 1057 10.1016/j.cell.2020.09.033 e15 32970989 13 Zhang Q. Chen C.Z. Swaroop M. Xu M. Wang L. Lee J. Wang A.Q. Pradhan M. Hagen N. Chen L. Shen M. Luo Z. Xu X. Xu Y. Huang W. Zheng W. Ye Y. Heparan sulfate assists SARS-CoV-2 in cell entry and can be targeted by approved drugs in vitro Cell Discov. 6 1 2020 80 10.1038/s41421-020-00222-5 33298900 14 Kearns F.L. Sandoval D.R. Casalino L. Clausen T.M. Rosenfeld M.A. Spliid C.B. Amaro R.E. Esko J.D. Spike-heparan sulfate interactions in SARS-CoV-2 infection Curr. Opin. Struct. Biol. 76 2022 102439 10.1016/j.sbi.2022.102439 15 Tang T. Bidon M. Jaimes J.A. Whittaker G.R. Daniel S. Coronavirus membrane fusion mechanism offers a potential target for antiviral development Antivir. Res. 178 2020 104792 10.1016/j.antiviral.2020.104792 16 Xu Y. Li Y. You X. Pei C. Wang Z. Jiao S. Zhao X. Lin X. Lu Y. Jin C. Gao G.F. Li J. Wang Q. Du Y. Novel insights into the sulfated glucuronic acid-based anti-SARS-CoV-2 mechanism of exopolysaccharides from halophilic archaeon haloarcula hispanica Front. Chem. 10 2022 871509 10.3389/fchem.2022.871509 17 Kim S.Y. Jin W. Sood A. Montgomery D.W. Grant O.C. Fuster M.M. Fu L. Dordick J.S. Woods R.J. Zhang F. Linhardt R.J. Characterization of heparin and severe acute respiratory syndrome-related coronavirus 2 (SARS-CoV-2) spike glycoprotein binding interactions Antivir. Res. 181 2020 104873 10.1016/j.antiviral.2020.104873 18 Mycroft-West C.J. Su D. Pagani I. Rudd T.R. Elli S. Gandhi N.S. Guimond S.E. Miller G.J. Meneghetti M.C.Z. Nader H.B. Li Y. Nunes Q.M. Procter P. Mancini N. Clementi M. Bisio A. Forsyth N.R. Ferro V. Turnbull J.E. Guerrini M. Fernig D.G. Vicenzi E. Yates E.A. Lima M.A. Skidmore M.A. Heparin inhibits cellular invasion by SARS-CoV-2: structural dependence of the interaction of the spike S1 receptor-binding domain with heparin Thromb. Haemost. 120 12 2020 1700 1715 10.1055/s-0040-1721319 33368089 19 Tandon R. Joshua S.Sharp Zhang F. Vitor H.Pomin Nicole M.Ashpole Mitra D. Martin G.McCandless Jin W. Liu H. Sharma P. Robert J.Linhardt Gallagher T. Effective inhibition of SARS-CoV-2 entry by heparin and enoxaparin derivatives J. Virol. 95 3 2021 e01987-20 10.1128/JVI.01987-20 20 Liu L. Chopra P. Li X. Bouwman K.M. Tompkins S.M. Wolfert M.A. de Vries R.P. Boons G.J. Heparan sulfate proteoglycans as attachment factor for SARS-CoV-2 ACS Cent. Sci. 7 6 2021 1009 1018 10.1021/acscentsci.1c00010 34235261 21 Linhardt R.J. Claude S. Hudson Award address in carbohydrate chemistry. Heparin: structure and activity J. Med. Chem. 46 13 2003 2551 2564 10.1021/jm030176m 12801218 22 Liu J. Linhardt R.J. Chemoenzymatic synthesis of heparan sulfate and heparin Nat. Prod. Rep. 31 12 2014 1676 1685 10.1039/c4np00076e 25197032 23 Baytas S.N. Linhardt R.J. Advances in the preparation and synthesis of heparin and related products Drug Discov. Today 25 12 2020 2095 2109 10.1016/j.drudis.2020.09.011 32947045 24 Liu J. Pedersen L.C. Anticoagulant heparan sulfate: structural specificity and biosynthesis Appl. Microbiol. Biotechnol. 74 2 2007 263 272 10.1007/s00253-006-0722-x 17131147 25 Bernfield M. Götte M. Park P.W. Reizes O. Fitzgerald M.L. Lincecum J. Zako M. Functions of cell surface heparan sulfate proteoglycans Annu. Rev. Biochem. 68 1999 729 777 10.1146/annurev.biochem 10872465 26 Turnbul J.E. Chemistry. Getting the farm out of pharma for heparin production Science 334 6055 2011 462 463 10.1126/science.1211605 22034422 27 Tree J.A. Turnbull J.E. Buttigieg K.R. Elmore M.J. Coombes N. Hogwood J. Mycroft-West C.J. Lima M.A. Skidmore M.A. Karlsson R. Chen Y.H. Yang Z. Spalluto C.M. Staples K.J. Yates E.A. Gray E. Singh D. Wilkinson T. Page C.P. Carroll M.W. Unfractionated heparin inhibits live wild type SARS-CoV-2 cell infectivity at therapeutically relevant concentrations Br. J. Pharmacol. 178 3 2021 626 635 10.1111/bph.15304 33125711 28 Gasbarri M. V'Kovski P. Torriani G. Thiel V. Stellacci F. Tapparel C. Cagno V. SARS-CoV-2 inhibition by sulfonated compounds Microorganisms 8 12 2020 1894 10.3390/microorganisms8121894 33265927 29 Hao W. Ma B. Li Z. Wang X. Gao X. Li Y. Qin B. Shang S. Cui S. Tan Z. Binding of the SARS-CoV-2 spike protein to glycans Sci. Bull. (Beijing) 66 12 2021 1205 1214 10.1016/j.scib.2021.01.010 33495714 30 Mclean J. The thromboplastic action of cephalin Am. J. Phys. 41 2 1916 250 257 10.1152/ajplegacy.1916.41.2.250 31 Shriver Z. Capila I. Venkataraman G. Sasisekharan R. Heparin and heparan sulfate: analyzing structure and microheterogeneity Handb. Exp. Pharmacol. 207 2012 159 176 10.1007/978-3-642-23056-1_8 32 FDA Heparin for drug and medical device use: monitoring crude heparin for quality https://www.fda.gov/regulatory-information/search-fda-guidance-documents/heparin-drug-and-medical-device-use-monitoring-crude-heparin-quality 2013 33 DeAngelis P.L. Liu J. Linhardt R.J. Chemoenzymatic synthesis of glycosaminoglycans: re-creating, re-modeling and re-designing nature’s longest or most complex carbohydrate chains Glycobiology 23 7 2013 764 777 10.1093/glycob/cwt016 23481097 34 Kishimoto T.K. Viswanathan K. Ganguly T. Contaminated heparin associated with adverse clinic events and activation of the contact system N. Engl. J. Med. 358 23 2008 2457 2467 10.1056/NEJMoa0803200 18434646 35 Guerrini M. Beccati D. Shriver Z. Naggi A. Viswanathan K. Bisio A. Capila I. Lansing J.C. Guglieri S. Fraser B. Al-Hakim A. Gunay N.S. Zhang Z. Robinson L. Buhse L. Nasr M. Woodcock J. Langer R. Venkataraman G. Linhardt R.J. Casu B. Torri G. Sasisekharan R. Oversulfated chondroitin sulfate is a contaminant in heparin associated with adverse clinical events Nat. Biotechnol. 26 6 2008 669 675 10.1038/nbt1407 18437154 36 Liu H. Zhang Z. Linhardt R.J. Lessons learned from the contamination of heparin Nat. Prod. Rep. 26 3 2009 313 321 10.1039/b819896a 19240943 37 Guidance for Industry-Heparin for Drug and Medical Device Use: Monitoring Crude Heparin for Quality. Available online: https://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM291390.pdf (accessed on 19 April 2017). 38 Fu L. Li G. Yang B. Onishi A. Li L. Sun P. Zhang F. Linhardt R.J. Structural characterization of pharmaceutical heparins prepared from different animal tissues J. Pharm. Sci. 102 5 2013 1447 1457 10.1002/jps.23501 23526651 39 Crossan C. Scobie L. Godwin J. Hunter J.G. Hawkes T. Dalton H.R. Hepatitis E virus and porcine-derived heparin Emerg. Infect. Dis. 19 4 2013 686 688 10.3201/eid1904.121792 40 Mende M. Bednarek C. Wawryszyn M. Sauter P. Biskup M.B. Schepers U. Brase S. Chemical synthesis of glycosaminoglycans Chem. Rev. 116 14 2016 8193 8255 10.1021/acs.chemrev.6b00010 27410264 41 Pongener I. O'Shea C. Wootton H. Watkinson M. Miller G.J. Developments in the chemical synthesis of heparin and heparan sulfate Chem. Rec. 21 11 2021 3238 3255 10.1002/tcr.202100173 34523797 42 Xu Y. Masuko S. Takieddin M. Xu H. Liu R. Jing J. Mousa S.A. Linhardt R.J. Liu J. Chemoenzymatic synthesis of homogeneous ultralow molecular weight heparins Science 334 6055 2011 498 501 10.1126/science.1207478 22034431 43 Zhang X. Pagadala V. Jester H.M. Lim A.M. Pham T.Q. Goulas A.M.P. Liu J. Linhardt R.J. Chemoenzymatic synthesis of heparan sulfate and heparin oligosaccharides and NMR analysis: paving the way to a diverse library for glycobiologists Chem. Sci. 8 12 2017 7932 7940 10.1039/c7sc03541a 29568440 44 Zhang Z. Mccallum S.A. Jin X. Nieto L. Corzana F. Jimenez-Barbero J. Chen M. Liu J. Linhardt R.J. Solution structures of chemoenzymatically synthesized heparin and its precursors J. Am. Chem. Soc. 130 39 2008 12998 13007 10.1021/ja8026345 18767845 45 Wang Z. Ly M. Zhang F. Zhong W. Suen A. Hickey A.M. Dordick J.S. Linhardt R.J. E. Coli K5 fermentation and the preparation of heparosan, a bioengineered heparin precursor Biotechnol. Bioeng. 107 6 2010 964 973 10.1002/bit.22898 20717972 46 Naggi A. Torri G. Casu B. Oreste P. Zoppetti G. Li J.P. Lindahl U. Toward a biotechnological heparin through combined chemical and enzymatic modification of the Escherichia coli K5 polysaccharide Semin. Thromb. Hemost. 27 5 2001 437 443 10.1055/s-2001-17954 11668413 47 Rusnati M. Vicenzi E. Donalisio M. Oreste P. Landolfo S. Lembo D. Sulfated K5 Escherichia coli polysaccharide derivatives: a novel class of candidate antiviral microbicides Pharmacol. Ther. 123 3 2009 310 322 10.1016/j.pharmthera.2009.05.001 19447134 48 Capila I. Linhardt R.J. Heparin-protein interactions Angew. Chem. Int. Ed. Engl. 41 3 2002 390 412 10.1002/1521-3773(20020201)41:3<390::aid-anie390>3.0.co;2-b 49 Chen Y. Lin L. Agyekum I. Zhang X. St Ange K. Yu Y. Zhang F. Liu J. Amster I.J. Linhardt R.J. Structural analysis of heparin-derived 3-O-sulfated tetrasaccharides: antithrombin binding site variants J. Pharm. Sci. 106 4 2017 973 981 10.1016/j.xphs.2016.11.023 28007564 50 Madhur D.S. Gregory M.P. Niall S. Sukhwinder S.S. Rahul P.P. Non-anticoagulant derivatives of heparin for the management of asthma: distant dream or close reality? Expert Opin. Investig. Drugs 23 3 2014 357 373 10.1517/13543784.2014.866092 51 Buijsers B. Yanginlar C. Maciej-Hulme M.L. de Mast Q. van der Vlag J. Beneficial non-anticoagulant mechanisms underlying heparin treatment of COVID-19 patients EBioMedicine 59 2020 102969 52 Buijsers B. Yanginlar C. de Nooijer A. Grondman I. Maciej-Hulme M.L. Jonkman I. Janssen N.A.F. Rother N. de Graaf M. Pickkers P. Kox M. Joosten L.A.B. Nijenhuis T. Netea M.G. Hilbrands L. van de Veerdonk F.L. Duivenvoorden R. de Mast Q. van der Vlag J. Increased plasma heparanase activity in COVID-19 patients Front. Immunol. 11 2020 575047 10.3389/fimmu.2020.575047 53 Lever R. Page C.P. Non-anticoagulant effects of heparin: an overview Handb. Exp. Pharmacol. 207 2012 281 305 10.1007/978-3-642-23056-1_12 54 Cassinelli G. Naggi A. Old and new applications of non-anticoagulant heparin Int. J. Cardio. 212 2016 S14 S21 10.1016/S0167-5273(16)12004-2 55 Cassinelli G. Torri G. Naggi A. Non-anticoagulant heparins as heparanase inhibitors Adv. Exp. Med. Biol. 1221 2020 493 522 10.1007/978-3-030-34521-1_20 32274724 56 Lin Y.P. Yu Y. Marcinkiewicz A.L. Lederman P. Hart T.M. Zhang F. Linhardt R.J. Non-anticoagulant heparin as a pre-exposure prophylaxis prevents Lyme disease infection ACS Infect. Dis. 6 3 2020 503 514 10.1021/acsinfecdis.9b00425 31961652 57 Rao N.V. Argyle B. Xu X. Reynolds P.R. Walenga J.M. Prechel M. Prestwich G.D. MacArthur R.B. Walters B.B. Hoidal J.R. Kennedy T.P. Low anticoagulant heparin targets multiple sites of inflammation, suppresses heparin-induced thrombocytopenia, and inhibits interaction of RAGE with its ligands Am. J. Physiol. Cell Physiol. 299 1 2010 C97 C110 10.1152/ajpcell.00009.2010 20375277 58 Fryer A. Huang Y.C. Rao G. Jacoby D. Mancilla E. Whorton R. Piantadosi C.A. Kennedy T. Hoidal J. Selective O-desulfation produces non-anticoagulant heparin that retains pharmacological activity in the lung J. Pharmacol. Exp. Ther. 282 1 1997 208 219 10.1021/js970097w 9223556 59 Ji Y. Wang Y. Zeng W. Mei X. Du S. Yan Y. Hao J. Zhang Z. Lu Y. Zhang C. Ge J. Xing X.H. A heparin derivatives library constructed by chemical modification and enzymatic depolymerization for exploitation of non-anticoagulant functions Carbohydr. Polym. 249 2020 116824 10.1016/j.carbpol.2020.116824 60 Duckworth C.A. Guimond S.E. Sindrewicz P. Hughes A.J. French N.S. Lian L.Y. Yates E.A. Pritchard D.M. Rhodes J.M. Turnbull J.E. Yu L.G. Chemically modified, non-anticoagulant heparin derivatives are potent galectin-3 binding inhibitors and inhibit circulating galectin-3-promoted metastasis Oncotarget 6 27 2015 23671 23687 10.18632/oncotarget.4409 26160844 61 Achour O. Poupard N. Bridiau N. Bordenave Juchereau S. Sannier F. Piot J.M. Fruitier Arnaudin I. Maugard T. Anti-heparanase activity of ultra-low-molecular-weight heparin produced by physicochemical depolymerization Carbohydr. Polym. 135 2016 316 323 10.1016/j.carbpol.2015.08.041 26453883 62 Ono K. Ishihara M. Ishikawa K. Ozeki Y. Deguchi H. Sato M. Hashimoto H. Saito Y. Yura H. Kurita A. Maehara T. Periodate-treated, non-anticoagulant heparin-carrying polystyrene (NAC-HCPS) affects angiogenesis and inhibits subcutaneous induced tumour growth and metastasis to the lung Br. J. Cancer 86 11 2002 1803 1812 10.1038/sj.bjc.6600307 12087470 63 Alekseeva A. Mazzini G. Giannini G. Naggi A. Structural features of heparanase-inhibiting non-anticoagulant heparin derivative roneparstat Carbohydr. Polym. 156 2017 470 480 10.1016/j.carbpol.2016.09.032 27842848 64 Mousa S.A. Linhardt R. Francis J.L. Amirkhosravi A. Anti-metastatic effect of a non-anticoagulant low-molecular-weight heparin versus the standard low-molecular-weight heparin, enoxaparin Thromb. Haemost. 96 6 2006 816 821 10.1160/th06-05-0289 17139378 65 Lindahl U. Backstrom G. Thunberg L. Leder I.G. Evidence for a 3-O-sulfated D-glucosamine residue in the antithrombin-binding sequence of heparin Proc. Natl. Acad. Sci. U. S. A. 77 11 1980 6551 6555 10.1073/pnas.77.11.6551 6935668

0

PMC9749384

NO-CC CODE

2022-12-15 23:23:24

no

Int J Biol Macromol. 2023 Jan 31; 226:974-981

utf-8

Int J Biol Macromol

10.1016/j.ijbiomac.2022.12.090

oa_other

==== Front J Allergy Clin Immunol J Allergy Clin Immunol The Journal of Allergy and Clinical Immunology 0091-6749 1097-6825 Published by Elsevier Inc. on behalf of the American Academy of Allergy, Asthma & Immunology. S0091-6749(22)02454-X 10.1016/j.jaci.2022.12.800 Article Treating Asthma in the Time of COVID Carr Tara F. MD a Fajt Merritt L. MD b Kraft Monica MD c Phipatanakul Wanda MD MS d Szefler Stanley J. MD e Zeki Amir A. MD MAS f Peden David B. MD g White Steven R. MD h∗ for the PrecISE Research Groupa a Asthma and Airway Disease Research Center, University of Arizona, Tucson, AZ b Division of Pulmonary, Allergy and Critical Care Medicine, University of Pittsburgh, Pittsburgh, PA c Samuel Bronfman Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY d Division of Allergy and Immunology, Department of Pediatrics, Boston Children’s Hospital, Boston, MA e The University of Colorado School of Medicine and Children’s Hospital Colorado, Department of Pediatrics, The Breathing Institute, Aurora, CO, USA f Department of Internal Medicine; Division of Pulmonary, Critical Care, and Sleep Medicine; University of California, Davis School of Medicine; UC Davis Lung Center; Sacramento, CA g Division of Allergy and Immunology, Department of Pediatrics, University of North Carolina, Chapel Hill, NC h Department of Medicine, the University of Chicago, Chicago, IL ∗ Corresponding author: Steven R. White, MD, University of Chicago, 5841 S. Maryland Ave., MC6076, Chicago, IL 60637, (773) 702-2004 14 12 2022 14 12 2022 17 10 2022 30 11 2022 12 12 2022 © 2022 Published by Elsevier Inc. on behalf of the American Academy of Allergy, Asthma & Immunology. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The Precision Interventions for Severe and/or Exacerbation-Prone Asthma (PrecISE) clinical trials network is actively assessing novel treatments for severe asthma during the COVID-19 pandemic and has needed to adapt to various clinical dilemmas posed by the COVID-19 pandemic. Pharmacologic interactions between established asthma therapies and novel drug interventions for COVID-19 infection, including antivirals, biologics, and vaccines, have emerged as a critical and unanticipated issue in the clinical care of asthma. In particular, impaired metabolism of some long-acting beta-2 agonists by the cytochrome P4503A4 enzyme in the setting of antiviral treatment using ritonavir-boosted nirmatrelvir (NVM/r, brand name Paxlovid) may increase risk for adverse cardiovascular events. While available data have documented the potential for such interactions, these issues are largely unappreciated by clinicians who treat asthma, or those dispensing COVID-19 interventions in patients who happen to have asthma. As these drug-drug interactions have not previously been relevant to patient care, clinicians have had no guidance on management strategies to reduce potentially serious interactions between treatments for asthma and COVID-19. The PrecISE network considered the available literature and product information, and herein share our considerations and plans for treating asthma within the context of these novel COVID-19-related therapies. Keywords Asthma ritonavir COVID-19 salmeterol cytochrome P450 interaction CYP3A4 long-acting beta adrenergic agonists corticosteroids Abbreviations COVID-19, SARS-CoV-2 infection pandemic of 2019-2022 CYP3A4, cytochrome enzyme P4503A4 FDA, Food and Drug Administration GINA, Global Initiative for Asthma HAART, highly active antiretroviral therapy ICS, inhaled corticosteroid IL, interleukin LABA, long-acting beta-adrenergic agonist NVM/r, ritonavir-boosted nirmatrelvir PrecISE, Precision Interventions for severe and exacerbation prone asthma ==== Body pmcConflicts of Interest TFC has served in an advisory role or as a consultant for AstraZeneca, Genentech, GlaxoSmithKline, Novartis and Regeneron and as a writer and editor for Wolters Kluwer UpToDate. MLF has nothing to disclose. MK has received research funding paid to her institution from NIH, ALA, Sanofi, AstraZeneca Synairgen and Janssen; served on an advisory board for AstraZeneca, Sanofi, Chiesi and Synairgen; has received fees for speaking for Chiesi, AstraZeneca and Sanofi; is co-founder and CMO, RaeSedo, Inc, a company evaluating peptidomimetics for treatment of inflammatory lung disease; payments from Elsevier for UptoDate, where she is a section editor for asthma. WP has served in advisory capacity for Genentech, Novartis, Sanofi, Regenron, Teva , Astra Zeneca, GSK. SJS reports consultant fees paid to the university from AstraZeneca, GSK, Moderna, OM Pharma, Propeller Health, Regeneron Pharmaceuticals, Inc., Sanofi. AAZ reports serving as CSO for InStatin, Inc; and on Sanofi/Regeneron consulting and advisory boards. DPB has served as a consultant for TEVA and GSK and receives grant support from NHLBI, NIEHS, NIAID, DOD and EPA. SRW has served as a consultant and speaker for AstraZeneca, Sanofi, and Regeneron. Funding Sources: The PrecISE study is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, grants U24 HL138998, 1UG1HL139054, 1UG1HL139098, 1UG1HL139106, 1UG1HL139117, 1UG1HL139118, 1UG1HL139119, 1UG1HL139123, 1UG1HL139124, 1UG1HL139125, 1UG1HL139126, and 1UG1HL146002. Support for site institutional infrastructure came from National Institute of Health Clinical & Translational Science Award grants UL1TR002451 (Harvard), UL1TR000427 (University of Wisconsin), UL1TR002366 (University of Kansas), UL1TR002389 (University of Chicago), UL1TR002489 (University of North Carolina), UL1TR002548 (Cleveland Clinic), and UL1TR001442 (University of California, San Diego). The study also gratefully acknowledges receiving contributed product from Vitaeris, owned and operated by CSL group (clazakizumab), Vitaflo (MCT), Sun Pharma (imatinib), OM Pharma, a Vifor Pharma Group Company (OM-85, BronchoVaxom), Incyte (itacitinib), Laurel Venture (cavosonstat) and GlaxoSmithKline (Advair Diskus and Ventolin).

0

PMC9749385

NO-CC CODE

2022-12-15 23:23:21

no

J Allergy Clin Immunol. 2022 Dec 14; doi: 10.1016/j.jaci.2022.12.800

utf-8

J Allergy Clin Immunol

10.1016/j.jaci.2022.12.800

oa_other

==== Front Curr Probl Cardiol Curr Probl Cardiol Current Problems in Cardiology 0146-2806 1535-6280 Mosby-Year Book S0146-2806(22)00444-3 10.1016/j.cpcardiol.2022.101547 101547 Article Impact of COVID-19 on Outcomes of Patients Hospitalized with STEMI: A Nationwide Propensity-Matched Analysis Goel Akshay 1⁎ Malik Aaqib H 1 Bandyopadhyay Dhrubajyoti 1 Isath Ameesh 1 Gupta Rahul 2 Hajra Adrija 3 Shrivastav Rishi 4 Virani Salim S 5 Fonarow Gregg C 6 Lavie Carl J 7 Naidu Srihari S 1 1 Department of Cardiology, Westchester Medical Center, New York Medical College, Valhalla, NY, USA 2 Lehigh Valley Heart Institute, Lehigh Valley Health Network, Allentown, PA, USA 3 Montefiore Medical Center/Albert Einstein College of Medicine, New York, NY, USA 4 Icahn School of Medicine at Mount Sinai, New York, NY, USA 5 Baylor College of Medicine, Houston, TX, USA 6 University of California, Los Angeles, CA, USA 7 John Ochsner Heart and Vascular Institute, Ochsner Clinical School, The University of Queensland School of Medicine, New Orleans, LA, USA ⁎ Correspondence: Akshay Goel, MD, Department of Cardiology, Westchester Medical Center, New York Medical College, 100 Woods Road, Macy Pavilion, Suite 100, Valhalla, NY 10595, USA. Ph: 646-785-1084, Fax: 914-493-1854 14 12 2022 14 12 2022 101547. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Introduction Patients with ST-segment elevation myocardial infarction (STEMI) and concurrent coronavirus disease 2019 (COVID-19) have been reported to have poor outcomes. However, previous studies are small and limited. Methods The National Inpatient Sample (NIS) database for the year 2020 was queried to identify all adult hospitalizations with a primary diagnosis of STEMI, with and without concurrent COVID-19. A 1:1 propensity score matching was performed. Results A total of 159,890 hospitalizations with a primary diagnosis of STEMI were identified. Of these, 2,210 (1.38%) had concurrent COVID-19. After propensity matching, STEMI patients with concurrent COVID-19 had a significantly higher mortality (17.8% vs 9.1%, OR 1.96, p <0.001), lower likelihood to receive same-day percutaneous coronary intervention (PCI) (63.6% vs 70.6%, p = 0.019), with a trend towards lower overall PCI (74.9% vs 80.2%, p = 0.057) and significantly lower coronary artery bypass grafting (CABG) (3.0% vs 6.8%, p = 0.008) prior to discharge, compared with STEMI patients without COVID-19. The prevalence of cardiogenic shock, need for mechanical circulatory support, extracorporeal membrane oxygenation (ECMO), cardiac arrest, acute kidney injury (AKI), dialysis, major bleeding and stroke were not significantly different between the groups. COVID-19-positive STEMI patients who received same-day PCI had significantly lower odds of in-hospital mortality (adjusted OR 0.42, 95% CI 0.20-0.85, p = 0.017). Conclusion STEMI patients with concurrent COVID-19 infection had a significantly higher (almost two times) in-hospital mortality, and lower likelihood of receiving same-day PCI, overall (any-day) PCI, and CABG during their admission, compared with STEMI patients without COVID-19. Keywords COVID-19 myocardial infarction STEMI ==== Body pmcINTRODUCTION The coronavirus disease 2019 (COVID-19) pandemic has affected more than 600 million people worldwide and has been a global health challenge. The first case of COVID-19 in the United States (US) was reported in January 2020, and since then the number of confirmed cases has crossed 96 million making it the country with the maximum number of cases.1 In the largest US national series of hospitalized COVID-19 patients till date with information on more than 1.6 million COVID-19-related hospitalizations, the overall in-hospital mortality rate was found to be as high as 13.2%.2 COVID-19 is associated with an increase in the risk of thromboembolic complications, and the risk of myocardial infarction (MI) has been noted to be almost double within a week of a diagnosis of COVID-19.3 ST-segment elevation MI (STEMI) patients with concurrent COVID-19 may not receive revascularization or other treatment in a timely manner due to concerns of infection transmission to healthcare workers and other patients. Patients with STEMI and COVID-19 have been observed to have significantly worse outcomes in some studies.3, 4, 5, 6, 7 However, these studies were limited by their relatively small sample size. Additionally, the exact prevalence of various cardiac and non-cardiac complications, morbidity, and mortality in hospitalized STEMI patients with concurrent COVID-19 infection remain poorly defined, as do the reasons for reported worse outcomes. Hence, it becomes important to investigate this further in a larger nationwide study. The aim of this study was to compare hospitalized STEMI patients with and without COVID-19 using a national database, in an attempt to understand the impact of concurrent COVID-19 infection on the in-hospital outcomes, morbidity and mortality after STEMI. METHODS Data Source We utilized the National Inpatient Sample (NIS) database for the year 2020. The NIS is a part of the Healthcare Cost and Utilization Project (HCUP) sponsored by the Agency for Healthcare Research and Quality (AHRQ), and is available publicly. The NIS is drawn from all states participating in HCUP, covering more than 97 percent of the US population. The NIS comprises approximately 20% of stratified sample discharges from US community hospitals, excluding rehabilitation and long-term acute care hospitals. A discharge weight variable is available to calculate national estimates. With data on up to 8 million hospital discharge records every year, the NIS is considered to be the largest all-payer inpatient database in the US.8 The NIS database is crosschecked annually by the AHRQ to ensure internal validity. Data from the NIS has been validated and shown to correlate well with other discharge databases in the US.9 Data collection methods and administration of NIS are described.8 Study Population The NIS database for the year 2020 was queried to identify all adult hospitalizations with a primary diagnosis of STEMI using the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes. These patients were then classified into two groups based on the presence or absence of concurrent COVID-19 infection (ICD-10-CM code U07.1). Covariables and Comorbidities Definitions A large number of variables were included in our analysis to minimize confounders and allow for a robust analysis. Most of the included variables were supplied by the NIS database including age, sex, race (White, African American, Hispanic, and others), insurance type (Medicare, Medicaid, private insurance, self-pay, and others), median household income (0 to 25th percentile, 26th to 50th percentile, 51st to 75th percentile, and 76th to 100th percentile), day of admission, hospital bed size (large, medium, small), location (urban teaching, urban non-teaching, rural), and region (Northeast, Midwest, South, and West). Other covariables and comorbidities (example, hypertension, diabetes, chronic kidney disease, etc.) were either obtained from the Elixhauser list of comorbidities supplied by the NIS or manually coded using their specific ICD-10-CM codes. Outcomes The primary outcome of interest was in-hospital mortality. Secondary outcomes included same-day percutaneous coronary intervention (PCI), any PCI prior to discharge, coronary artery bypass grafting (CABG), cardiogenic shock, need for mechanical circulatory support with intra-aortic balloon pump (IABP) or Impella percutaneous left ventricular assist device, extracorporeal membrane oxygenation (ECMO), cardiac arrest, acute respiratory failure, respiratory failure requiring prolonged (>24 hours) intubation, acute kidney injury (AKI), AKI requiring dialysis, major bleeding requiring transfusion, sepsis, deep vein thrombosis (DVT), pulmonary embolism (PE), stroke, length of stay (LOS) and hospitalization costs. Statistical Analysis National weighted numbers were estimated using the discharge weight variables provided by NIS. Propensity score matching was performed to compare outcomes for COVID-19-positive STEMI patients with STEMI patients without COVID-19 using a propensity score calculated based on a multivariable logistic regression model. Propensity score matching without a replacement was performed in a 1:1 nearest-neighbor fashion with a caliper width of 20% of the estimated propensity scores.10 To assess the matching performance, mean standardized differences were calculated. All standardized mean differences were within 10%. Using the unmatched cohort, we next performed a multivariable regression analysis to evaluate the association between COVID-19 and in-hospital mortality in admitted STEMI patients. In addition, independent predictors of in-hospital morality in hospitalized STEMI patients were identified based on the multivariable logistic regression analysis, and reverse stepwise regression techniques were used to formalize the final model. All statistical analyses were performed using Stata Statistical Software: Release 16. (StataCorp LLC, College Station, Texas) with a 2-sided p value of <0.05 as cutoff for statistical significance and odds ratio (OR) with 95% confidence interval (CI) as measure of effect size. Continuous variables are expressed as median with interquartile range. Categorical variables are presented as numbers and percentages. Continuous variables were compared using the Student's t test, while categorical variables were compared using the Pearson chi-square test. Ethical approval Given that the NIS is a publicly available de-identified database, this study was deemed exempt from institutional board review. RESULTS Baseline Characteristics – Pre- and Post- Matching Of 159,890 hospitalizations with a primary diagnosis of STEMI, 2,210 (1.38%) had concurrent COVID-19, while the remaining 157,680 (98.62%) did not have COVID-19. The age (median age 62 vs 63 years, p = 0.193) and gender distribution (31.9% vs 30.5% females, p = 0.512) were similar between STEMI patients with and without COVID-19. Compared with STEMI patients without COVID-19, STEMI patients with COVID-19 were more likely to be African American (14.5% vs 8.6%, p <0.001) or Hispanic (21.5% vs 8.1%, p <0.001), and more likely to have diabetes mellitus (44.3% vs 33.4%, p <0.001) and dementia (6.3% vs 3%, p <0.001), but less likely to have valvular heart disease (5.7% vs 8.8%, p = 0.019), prior MI (9.3% vs 12.9%, p = 0.025), prior CABG (1.8% vs 3.8%, p = 0.033), peripheral vascular disease (4.1% vs 7.2%, p = 0.009), smoking history (18.1% vs 32.8%, p <0.001) and hypothyroidism (5.4% vs 9.2%, p = 0.007). The proportion of patients with hypertension, chronic kidney disease, congestive heart failure, atrial fibrillation, prior PCI, prior stroke, obesity, chronic obstructive pulmonary disease (COPD), alcohol use disorder, liver disease, anemia and cancer were not significantly different between the two groups at baseline. The post-matching cohort comprised of 2,195 STEMI hospitalizations with concurrent COVID-19 infection and 2,195 STEMI hospitalizations without COVID-19. Figure 1 illustrates the selection flow chart of hospitalized STEMI patients with and without concurrent COVID-19. Table 1 depicts the baseline characteristics of the unmatched groups and after 1:1 propensity score matching.Figure 1 Selection flow chart of hospitalized STEMI patients with and without concurrent COVID-19 Figure 1: Table 1 Baseline characteristics of hospitalized STEMI patients with and without COVID-19 Table 1 Before propensity matching After propensity matching COVID-19 (n = 2,210) No COVID-19 (n = 157,680) p value COVID-19 (n = 2,195) No COVID-19 (n = 2,195) p value Age (in years), median (IQR) 62 (54-72) 63 (55-72) 0.193 62 (54-72) 63 (55-72) 0.193 Age groups (in years) 0.414 0.989 18-59 42.1% 38.4% 41.9% 42.4% 60-69 27.8% 29.3% 27.8% 27.6% 70-79 18.3% 20.3% 18.5% 17.7% >79 11.8% 12.0% 11.8% 12.3% Female sex 31.9% 30.5% 0.512 32.1% 31.4% 0.826 Race <0.001 0.785 White 49.6% 72.5% 49.9% 50.8% African American 14.5% 8.6% 14.6% 11.9% Hispanic 21.5% 8.1% 21.2% 25.3% Others 14.4% 10.8% 14.3% 12.0% Insurance type <0.001 0.331 Medicare 39.9% 43.2% 39.9% 41.9% Medicaid 15.3% 11.8% 15.3% 12.1% Private 31.3% 36.8% 31.2% 34.9% Self-pay 8.8% 5.7% 8.9% 8.0% Median household income 0.002 0.651 0 to 25th percentile 34.4% 27.6% 34.2% 36.5% 26th to 50th percentile 27.6% 28.1% 27.8% 25.5% 51st to 75th percentile 20.1% 23.0% 20.3% 22.3% 76th to 100th percentile 14.5% 19.6% 14.6% 13.2% Weekend admission 28.9% 26.7% 0.102 28.9% 26.9% 0.492 Hospital bed size 0.055 0.366 Large 53.6% 52.5% 53.3% 53.1% Medium 24.9% 29.5% 25.1% 28.2% Small 21.5% 18.0% 21.6% 18.7% Hospital location 0.013 0.185 Urban teaching 79.8% 75.2% 79.9% 77.2% Urban non-teaching 15.6% 18.1% 15.5% 15.7% Rural 4.6% 6.7% 4.6% 7.1% Hospital region 0.321 0.906 Northeast 16.2% 16.9% 16.2% 17.1% Midwest 23.9% 23.0% 23.9% 22.8% South 39.9% 41.9% 39.9% 41.2% West 20.0% 18.2% 20.0% 18.9% Hypertension 75.3% 75.2% 0.929 75.2% 72.4% 0.340 Diabetes mellitus 44.3% 33.4% <0.001 44.2% 42.4% 0.577 Chronic kidney disease 15.8% 14.5% 0.383 15.7% 15.9% 0.920 Congestive heart failure 38.5% 40.1% 0.498 38.0% 37.4% 0.826 Atrial fibrillation 12.0% 12.3% 0.625 12.1% 11.9% 0.919 Valvular heart disease 5.7% 8.8% 0.019 5.7% 7.1% 0.393 Prior MI 9.3% 12.9% 0.025 9.3% 7.3% 0.285 Prior PCI 12.9% 12.0% 0.545 13.0% 10.7% 0.265 Prior CABG 1.8% 3.8% 0.033 1.8% 2.3% 0.622 Prior stroke 7.2% 6.5% 0.52 7.3% 6.9% 0.346 Peripheral vascular disease 4.1% 7.2% 0.009 4.1% 4.1% 0.999 Obesity 21.5% 20.7% 0.690 21.6% 18.7% 0.265 Smoking history 18.1% 32.8% <0.001 18.2% 18.5% 0.929 COPD 13.6% 15.4% 0.289 13.4% 14.6% 0.626 Alcohol use disorder 2.5% 3.0% 0.501 2.5% 2.9% 0.674 Liver disease 6.3% 5.8% 0.606 6.4% 6.8% 0.779 Coagulopathy 10.2% 5.8% <0.001 10.0% 10.7% 0.744 Dementia 6.3% 3.0% <0.001 6.2% 6.8% 0.675 Hypothyroidism 5.4% 9.2% 0.007 5.5% 6.6% 0.477 Rheumatologic disorders 1.6% 2.2% 0.366 1.6% 0.5% 0.094 Anemia 3.4% 2.4% 0.174 3.4% 2.1% 0.200 Cancer 2.0% 2.4% 0.628 2.0% 1.8% 0.799 IQR: interquartile range, MI: myocardial infarction, PCI: percutaneous coronary intervention, CABG: coronary artery bypass grafting, COPD: chronic obstructive pulmonary disease Pre-matching In-Hospital Mortality Rates of the Two Groups In the pre-matching cohort, a total of 13,330 (8.3%) deaths were noted in all hospitalizations for STEMI: 400/2,210 (18.1%) in those with COVID-19, and 12,930/157,680 (8.2%) in those without concurrent COVID-19 (unadjusted OR 2.21, 95% CI 2.02-2.42, p <0.001). Trend of Hospitalization Numbers and In-Hospital Mortality Rate of Patients with STEMI with Concurrent COVID-19 through the Year 2020 As the year 2020 progressed, there was a significant increase in the number of patients hospitalized for STEMI who had concurrent COVID-19, but their in-hospital mortality rate reduced over this time period (p value for trend <0.05 for both), as shown in Figure 2 .Figure 2 Trend of hospitalizations and in-hospital mortality of patients with STEMI and concurrent COVID-19 by month (in the year 2020) Figure 2: Propensity-Matched Analysis for Mortality and Other Outcomes In the propensity-matched analysis, hospitalized STEMI patients with concurrent COVID-19 had a significantly higher in-hospital mortality (17.8% vs 9.1%, OR 1.96, 95% CI 1.67-2.30, p <0.001), compared with those without concurrent COVID-19. STEMI patients with concurrent COVID-19 infection were less likely to receive same-day PCI (63.6% vs 70.6%, p = 0.019), and a trend was also noted towards these patients receiving lesser (any-day) PCI prior to discharge (74.9% vs 80.2%, p = 0.057), compared with STEMI patients without concurrent COVID-19. The rates of CABG during the same admission were also significantly lower in patients who were COVID-19-positive (3.0% vs 6.8%, p = 0.008). The prevalence of cardiogenic shock, need for mechanical circulatory support with IABP or Impella percutaneous left ventricular assist device, need for ECMO, and cardiac arrest were not significantly different between the groups. Even though STEMI patients with concurrent COVID-19 were more likely to be diagnosed with acute respiratory failure (31.7% vs 18.9%, p <0.001), there was no increased need for prolonged intubation (10.7% vs 9.3%, p = 0.501), compared with hospitalized STEMI patients without COVID-19. The rates of AKI, AKI requiring dialysis, major bleeding needing transfusion, and stroke during hospitalization were similar between the two propensity-matched groups. While the rates of DVT and PE were slightly higher in STEMI patients with concurrent COVID-19, these did not reach statistical significance. As was expected, sepsis complicated the hospital course more commonly in STEMI patients with concurrent COVID-19 infection (8.4% vs 4.1%, p = 0.006) than those without COVID-19. The above results are shown in Figure 3, Figure 4 .Figure 3 Comparison of in-hospital mortality and rates of revascularization in STEMI patients with and without concurrent COVID-19 Figure 3: Figure 4 Comparison of rates of various complications in STEMI patients with and without concurrent COVID-19 Figure 4: While the median LOS was longer by a day in STEMI patients with COVID-19, the increase in median hospitalization cost was not statistically significant compared with STEMI patients without COVID-19. The various in-hospital outcomes of STEMI patients with and without COVID-19 are shown in Table 2 .Table 2 Outcomes of hospitalized STEMI patients with and without COVID-19 Table 2: Before propensity matching After propensity matching COVID-19 (n = 2,210) No COVID-19 (n = 157,680) p value COVID-19 (n = 2,195) No COVID-19 (n = 2,195) p value In-hospital mortality 18.1% 8.2% <0.001 17.8% 9.1% <0.001 Same-day PCI 63.8% 70.8% <0.001 63.6% 70.6% 0.019 Any PCI prior to discharge 75.1% 80.5% 0.006 74.9% 80.2% 0.057 CABG 3.2% 5.2% 0.070 3.0% 6.8% 0.008 Cardiogenic shock 17.7% 14.4% 0.036 17.5% 14.6% 0.225 IABP 10.2% 7.9% 0.121 9.8% 8.0% 0.342 Impella percutaneous left ventricular assist device 2.3% 3.3% 0.208 2.1% 3.0% 0.349 ECMO 0.5% 0.6% 0.647 0.5% 1.3% 0.166 Cardiac arrest 10.4% 6.9% 0.004 10.5% 8.2% 0.214 Acute respiratory failure 31.9% 16.3% <0.001 31.7% 18.9% <0.001 Prolonged intubation >24 hours 10.9 % 6.9% 0.004 10.7% 9.3% 0.501 AKI 26.5% 19.0% <0.001 26.2% 21.2% 0.067 AKI requiring dialysis 2.3% 1.3% 0.076 2.3% 2.3% 0.999 Major bleeding requiring transfusion 4.8% 2.9% 0.014 4.6% 4.1% 0.738 Sepsis 9.0% 2.6% <0.001 8.4% 4.1% 0.006 Deep vein thrombosis 1.4% 0.6% 0.046 1.4% 0.9% 0.483 Pulmonary embolism 1.1% 0.4% 0.011 0.9% 0.2% 0.167 Stroke 0.2% 0.5% 0.348 0.2% 0.6% 0.296 Length of hospital stay (in days), median (IQR) 3 (2-6) 2 (2-4) <0.001 3 (2-6) 2 (2-4) <0.001 Hospitalization costs, median (IQR) $23,868 ($16,771-36,458) $22,231 ($16,167-33,098) 0.029 $23,691 ($16,731-36,100) $22,855 ($16,466-35,114) 0.598 PCI: percutaneous coronary intervention, CABG: coronary artery bypass grafting, IABP: intra-aortic balloon pump, ECMO: extracorporeal membrane oxygenation, AKI: acute kidney injury, IQR: interquartile range Multivariable Regression Analysis to Assess Whether COVID-19 is Independently Associated with Mortality in STEMI Patients On multivariable regression analysis, concurrent COVID-19 infection was confirmed to be independently associated with in-hospital mortality in patients hospitalized with STEMI (adjusted OR [aOR] 2.37, 95% CI 1.76-3.18, p <0.001). Multivariable Regression Analysis to Identify Independent Predictors of Mortality in Patients with STEMI and Concurrent COVID-19 In patients hospitalized for STEMI with concurrent COVID-19 infection, those receiving same-day PCI had a significantly lower in-hospital mortality (aOR 0.42, 95% CI 0.20-0.85, p = 0.017) than those who did not receive PCI on the day of presentation to the hospital (Figure 5 ). In COVID-19-positive hospitalized STEMI patients, age (aOR 1.04, 95% CI 1.01-1.06, p = 0.003), valvular heart disease (aOR 2.43, 95% CI 1.05-5.63, p = 0.039), liver disease (aOR 3.68, 95% CI 1.51-8.96, p = 0.004), and dementia (aOR 2.58, 95% CI 1.05-6.36, p = 0.039) were found to be independently associated with in-hospital mortality on multivariable regression analysis (Figure 5).Figure 5 Forest plot showing independent predictors of in-hospital mortality (with adjusted odds ratio and 95% confidence interval) in STEMI patients with concurrent COVID-19 Figure 5: DISCUSSION Patients with COVID-19 infection have been shown to have a hypercoagulable state, and are at a higher risk of thromboembolic events, including STEMI.11 , 12 The prognosis of STEMI patients with concurrent COVID-19 infection has previously been reported to be highly variable, with in-hospital mortality ranging from 12% to 72% in different studies.3, 4, 5, 6, 7 In this large nationwide propensity-matched analysis, STEMI patients with concurrent COVID-19 infection were found to be less likely to receive same-day PCI, overall (any-day) PCI and CABG prior to discharge, and significantly higher (almost twice) in-hospital mortality rate, compared with STEMI patients without COVID-19. In-hospital complications, such as cardiac arrest, cardiogenic shock, need for mechanical circulatory support, need for ECMO, prolonged intubation, AKI with or without need of dialysis, major bleeding requiring transfusion, stroke, DVT and PE were not significantly different between the two groups. As the year 2020 progressed, despite an increase in the number of patients hospitalized with STEMI and concurrent COVID-19, there was a significant improvement in the in-hospital mortality rate of these patients. Among COVID-19-positive STEMI patients, those who received PCI on the day of their presentation to the hospital were found to have significantly lower in-hospital mortality than patients who did not receive same-day PCI. Additionally, age, valvular heart disease, liver disease and dementia were identified to be independent predictors of mortality in patients hospitalized for STEMI and concurrent COVID-19 infection. To the best of our knowledge, our study is the largest study till date on STEMI patients with concurrent COVID-19 infection (with data on outcomes of 2,210 such patients). Previously, the North American COVID-19 Myocardial Infarction (NACMI) registry was created under the guidance of the Society for Cardiac Angiography and Interventions (SCAI), the Canadian Association of Interventional Cardiology, and the American College of Cardiology Interventional Council, to provide real-time clinical and outcome data on COVID-19-positive STEMI patients.3 , 4 The NACMI registry's initial report included data on 230 STEMI patients with COVID-19. The baseline characteristics of the NACMI registry population (age 56-75 years, 30% females, Hispanics 23%) is identical what we noted in our study. The prevalence of in-hospital cardiac arrest and cardiogenic shock reported by the NACMI registry in these patients was 11% and 18%, respectively, which is identical to what we found in our study (10.4% and 17.7%). Additionally, we observed that 75.1% COVID-19-positive STEMI patients received PCI prior to discharge, which is very similar to the 71% PCI rate that was reported in the NACMI registry. Similar to our study, the NACMI registry had reported that STEMI patients with COVID-19 were less likely to receive revascularization, compared with STEMI patients without COVID-19.4 The in-hospital mortality rate of COVID-19-positive STEMI patients in the initial report of the NACMI registry was 36%, which is much higher than what was observed in our study (18.1%). This could be explained due to the different period of patient enrollment for the initial report of the NACMI registry. Even in our study, the in-hospital mortality for STEMI patients with concurrent COVID-19 was observed to be as high as 30-35% for early months of the year and during a subsequent peak. Indeed, during a follow-up study of the NACMI registry that included 359 COVID-19-positive STEMI patients,3 the in-hospital mortality was noted to be 23% which is closer to what we observed in our current study. The reasons for poor outcomes in hospitalized STEMI patients with concurrent COVID-19 infection are not yet fully understood. It is postulated that COVID-19 patients presenting with STEMI are at a higher risk of associated cardiovascular complications, such as cardiac arrest, cardiogenic shock, need for mechanical circulatory support, AKI and dialysis, major bleeding, and PE. There is lack of robust data to back this hypothesis. In our study, the rates of these complications was only slightly higher, without statistical significance, in COVID-19-positive STEMI patients, compared with STEMI patients without COVID-19. We did find a significantly higher prevalence of acute respiratory failure and sepsis in COVID-19-positive patients, which may have contributed in part to the higher mortality of these patients. COVID-19-positive STEMI patients may have a delayed presentation to the hospital from the time of their symptom-onset, and may have longer door-to-balloon times. Similar to previous studies, we found that STEMI patients with concurrent COVID-19 were less likely to get timely revascularization. This is strongly suspected to contribute to the higher mortality of these patients. In fact, we identified that STEMI patients with COVID-19 who received same-day PCI had almost 60% lower odds of mortality, putting their in-mortality risk at par with STEMI patients without COVID-19 infection. These results indicate that much of the higher inpatient mortality of COVID-19-positive STEMI patients may be attributable to a lack of timely intervention. STEMI patients with COVID-19 have also been shown to have higher thrombus burden.13 , 14 The possibly higher thrombus burden may be a significant contributor to the adverse outcomes, especially in patients not receiving timely revascularization. There may be other interactions of different comorbidities in COVID-19-positive STEMI patients that may contribute to worse outcomes. In our study, we identified, age, valvular heart disease, liver disease and dementia as independent predictors of mortality in STEMI patients with concurrent COVID-19, which is a novel finding. Further studies are required to confirm the contribution of these variables to mortality and adverse outcomes in this patient population. Our study draws strength from the use of the largest inpatient national database available in the US, overcoming the bias seen with single-center and small regional studies. However, some limitations of this study must be acknowledged. Data derived from such national databases is dependent on the accuracy of diagnostic and procedural codes, and is also subject to potential selection and ascertainment bias. However, our large sample size minimizes the risk of serious selection bias. We also performed a propensity score-matched analysis to account for possible confounders. Second, the database did not allow us to investigate some clinically relevant information such as the exact time to revascularization, door to balloon time, time from symptom onset to presentation, echocardiographic or angiographic data, and how that may have affected outcomes in these patients. Since this is a retrospective observational study, the results do not necessarily imply causation. Finally, information on the different COVID-19 variants or the effect of vaccination against COVID-19 was not captured by this study. CONCLUSION Regardless, our study identifies that STEMI patients with concurrent COVID-19 infection are less likely to receive timely revascularization and have a significantly higher in-hospital mortality rate. These results have significant implications in the management and prognosis of patients with STEMI and concurrent COVID-19 infection. Further prospective studies are needed to confirm our findings, and to better elucidate the mechanisms of worse outcomes in STEMI patients with concurrent COVID-19 infection. Disclosures Dr. Fonarow has served as a consultant for Abbott, Amgen, Bayer, Janssen, Medtronic, and Novartis. Dr. Virani discloses the following relationships: Research support: Department of Veterans Affairs, World Heart Federation, Tahir and Jooma Family Honorarium: American College of Cardiology (Associate Editor for Innovations, acc.org). All the other authors have no financial disclosures to make. Funding None Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References 1 WHO COVID-19 Dashboard 2020 World Health Organization Geneva Available online: https://covid19.who.int/ Accessed October 9, 2022 2 Isath A Malik AH Goel A Nationwide analysis of the outcomes and mortality of hospitalized COVID-19 patients Curr Probl Cardiol 2022 10.1016/j.cpcardiol.2022.101440 3 Garcia S Dehghani P Stanberry L Trends in clinical presentation, management, and outcomes of STEMI in patients with COVID-19 J Am Coll Cardiol 79 2022 2236 2244 35390486 4 Garcia S Dehghani P Grines C Initial findings from the North American COVID-19 Myocardial Infarction Registry J Am Coll Cardiol 77 2021 1994 2003 33888249 5 Saad M Kennedy KF Imran H Association between COVID-19 diagnosis and in-hospital mortality in patients hospitalized with ST-segment elevation myocardial infarction JAMA 326 2021 1940 1952 34714327 6 Bangalore S Sharma A Slotwiner A ST-segment elevation in patients with Covid-19 – a case series N Engl J Med 382 2020 2478 2480 32302081 7 Stefanini GG Montorfano M Trabattoni D ST-elevation myocardial infarction in patients with COVID-19: clinical and angiographic outcomes Circulation 141 2020 2113 2116 32352306 8 Healthcare Cost and Utilization Project (HCUP). Overview of the National (Nationwide) Inpatient Sample (NIS). Available at: http://www.hcup-us.ahrq.gov/nisoverview.jsp. Accessed October 9, 2022. 9 Epstein AJ Polsky D Yang F Coronary revascularization trends in the United States, 2001-2008 JAMA 305 2011 1769 1776 21540420 10 Griswold ME Localio AR Mulrow C. Propensity score adjustment with multilevel data: setting your sites on decreasing selection bias Ann Intern Med 152 2010 393 395 20231571 11 Ho FK Man KKC Toshner M Thromboembolic risk in hospitalized and nonhospitalized COVID-19 patients: a self-controlled case series analysis of a nationwide cohort Mayo Clin Proc 96 2021 2587 2597 34607634 12 Kite TA Ludman PF Gale CP International prospective registry of acute coronary syndromes in patients with COVID-19 J Am Coll Cardiol 77 2021 2466 2476 34016259 13 Choudry FA Hamshere SM Rathod KS High thrombus burden in patients with COVID-19 presenting with ST-segment elevation myocardial infarction J Am Coll Cardiol 76 2020 1168 1176 32679155 14 Jain V Gupta K Bhatia K Management of STEMI during the COVID-19 pandemic: Lessons learned in 2020 to prepare for 2021 Trends Cardiovasc Med 31 2021 135 140 33338636

0

PMC9749390

NO-CC CODE

2022-12-15 23:22:53

no

Curr Probl Cardiol. 2022 Dec 14;:101547

utf-8

Curr Probl Cardiol

10.1016/j.cpcardiol.2022.101547

oa_other

==== Front J Clin Epidemiol J Clin Epidemiol Journal of Clinical Epidemiology 0895-4356 1878-5921 Published by Elsevier Inc. S0895-4356(22)00323-7 10.1016/j.jclinepi.2022.12.005 Original Article Minimal reporting improvement after peer review in reports of covid-19 prediction models: systematic review Hudda Mohammed T. 1#∗ Archer Lucinda 23∗ van Smeden Maarten 4 Moons Karel G.M. 45 Collins Gary S. 67 Steyerberg Ewout W. 8 Wahlich Charlotte 1 Reitsma Johannes B. 4 Riley Richard D. 3 Van Calster Ben 89 Wynants Laure 810 1 Population Health Research Institute, St George’s University of London, Cranmer Terrace, London, England, SW17 0RE 2 Centre for Prognosis Research, School of Medicine, Keele University, Staffordshire, UK 3 Institute of Applied Health Research, University of Birmingham, Edgbaston, UK 4 Julius Center for Health Sciences and Primary Care, University Medical Centre Utrecht, Utrecht University, Utrecht, Netherlands 5 Cochrane Netherlands, University Medical Centre Utrecht, Utrecht University, Utrecht, Netherlands 6 Centre for Statistics in Medicine, Nuffield Department of Orthopaedics, Musculoskeletal Sciences, University of Oxford, Oxford, UK 7 NIHR Oxford Biomedical Research Centre, John Radcliffe Hospital, Oxford, UK 8 Department of Biomedical Data Sciences, Leiden University Medical Centre, Leiden, Netherlands 9 Department of Development and Regeneration, KU Leuven, Leuven, Belgium 10 Department of Epidemiology, CAPHRI Care and Public Health Research Institute, Maastricht University, Peter Debyeplein 1, 6229 HA Maastricht, Netherlands # Corresponding author: Dr Mohammed T Hudda, Population Health Research Institute, St George’s University of London, Cranmer Terrace, London, SW17 0RE ∗ Joint first authors 14 12 2022 14 12 2022 13 9 2022 29 11 2022 7 12 2022 © 2022 Published by Elsevier Inc. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Objective To assess improvement in the completeness of reporting COVID-19 prediction models after the peer review process. Study Design and Setting Studies included in a living systematic review of COVID-19 prediction models, with both pre-print and peer-reviewed published versions available, were assessed. The primary outcome was the change in percentage adherence to the TRIPOD reporting guidelines between pre-print and published manuscripts. Results 19 studies were identified including seven (37%) model development studies, two external validations of existing models (11%), and 10 (53%) papers reporting on both development and external validation of the same model. Median percentage adherence amongst pre-print versions was 33% (min-max: 10 to 68%). The percentage adherence of TRIPOD components increased from pre-print to publication in 11/19 studies (58%), with adherence unchanged in the remaining eight studies. The median change in adherence was just 3 percentage points (pp, min-max: 0-14pp) across all studies. No association was observed between the change in percentage adherence and pre-print score, journal impact factor, or time between journal submission and acceptance. Conclusions Pre-print reporting quality of COVID-19 prediction modelling studies is poor and did not improve much after peer review, suggesting peer review had a trivial effect on the completeness of reporting during the pandemic. Key words Peer review reporting guidelines prediction modelling COVID-19 TRIPOD adherence prognosis and diagnosis ==== Body pmc

0

PMC9749392

NO-CC CODE

2022-12-15 23:22:53

no

J Clin Epidemiol. 2022 Dec 14; doi: 10.1016/j.jclinepi.2022.12.005

utf-8

J Clin Epidemiol

10.1016/j.jclinepi.2022.12.005

oa_other

==== Front Int J Biol Macromol Int J Biol Macromol International Journal of Biological Macromolecules 0141-8130 1879-0003 Published by Elsevier B.V. S0141-8130(22)03014-8 10.1016/j.ijbiomac.2022.12.112 Article SARS-CoV-2 RdRp uses NDPs as a substrate and is able to incorporate NHC into RNA from diphosphate form molnupiravir Wang Maofeng Wu Cancan Liu Nan Zhang Fengyu Dong Hongjie Wang Shuai Chen Min Jiang Xiaoqiong Zhang Kundi ⁎ Gu Lichuan ⁎ State Key Laboratory of Microbial Technology, Shandong University, 72 Binhai Road, Qingdao 266237, PR China ⁎ Corresponding authors. 14 12 2022 14 12 2022 23 7 2022 8 12 2022 11 12 2022 © 2022 Published by Elsevier B.V. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The coronavirus disease 2019 has been ravaging throughout the world for three years and has severely impaired both human health and the economy. The causative agent, severe acute respiratory syndrome coronavirus 2 employs the viral RNA dependent RNA polymerase (RdRp) complex for genome replication and transcription, making RdRp an appealing target for antiviral drug development. Systematic characterization of RdRp will undoubtedly aid in the development of antiviral drugs targeting RdRp. Here, our research reveals that RdRp can recognize and utilize nucleoside diphosphates as a substrate to synthesize RNA with an efficiency of about two thirds of using nucleoside triphosphates as a substrate. Nucleoside diphosphates incorporation is also template-specific and has high fidelity. Moreover, RdRp can incorporate β-d-N4-hydroxycytidine into RNA while using diphosphate form molnupiravir as a substrate. This incorporation results in genome mutation and virus death. It is also observed that diphosphate form molnupiravir is a better substrate for RdRp than the triphosphate form molnupiravir, presenting a new strategy for drug design. Keywords Molnupiravir Nucleoside diphosphate RNA dependent RNA polymerase SARS-CoV-2 ==== Body pmc1 Introduction The coronavirus disease 2019 (COVID-19) pandemic has severely impacted global human health and economy since the outbreak in late 2019 [1], [2]. The causative agent, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is homologous to 2002 severe acute respiratory syndrome coronavirus (SARS-CoV) with a genome sequence similarity of 79 % [3], [4], forming a sister clade to SARS-CoV and considered a newly β-coronavirus [5]. As of the date of writing, >491.75 million infections and >6.17 million deaths have been confirmed (https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports). Although a variety of vaccines have been developed to contain the pandemic, the rapidly spreading mutant strains presses need to develop effective medication or treatments to counter the virus [6], [7], [8], [9]. SARS-CoV-2 is a positive-sense single-stranded RNA virus [5] with a large genome of approximately 30 kb organized in 14 open reading frames [4], [10], [11], [12]. Replication of the genome and transcription of genes are all dependent on a protein complex known as RNA dependent RNA polymerase (RdRp). The RdRp of SARS-CoV-2 contains a catalytic subunit (non-structural protein 12, nsp12) and two accessory subunits (nsp7 and nsp8) [13], [14]. Processivity is enhanced by the presence of nsp7, nsp8, and bound to nsp12 in a 1:2:1 stoichiometry [15], [16]. Due to the significant role of RdRp and the lack of homologues in humans, RdRp has become the most appealing target for anti-coronavirus drug development [17], [18]. Traditionally, drug development is a long process that may last for decades. In order to abbreviate this process scientist resort to in-silico techniques. By using this state-of-art methodology, many chemical inhibitors have been designed to target RdRp or other SARS-CoV-2 proteins [19], [20], [21], [22], [23], [24], [25], [26], [27], [28]. To date, the most promising RdRp inhibitors are nucleotide/nucleoside analogs (NAs). NA prodrugs are designed to be metabolized to the active 5′-triphosphate form (5′-TP) once inside cells [29], thus they can compete with endogenous nucleotides for incorporation into newly synthesized viral RNA, resulting in viral genomes lethal mutation or inactivation [30]. Remdesivir is a broad antiviral drug approved for the treatment of SARS-CoV-2, which can be incorporated into RNA to sterically hinder RdRp, thereby blocking viral RNA synthesis [31]. Other NAs (favipiravir, ribavirin) are integrated into nascent viral RNA at high rates but are not recognized as endogenous nucleotides in subsequent rounds of replication, increasing the mutation rate and resulting in an inviable genome, a process known as “lethal mutagenesis” [32], [33], [34]. The recently approved NA antiviral drug molnupiravir significantly increases the frequency of viral RNA mutations and impairs SARS-CoV-2 in both animal models and humans [35], [36]. However, the application of NAs also bring adverse reactions and side effects, including elevated uric acid and hemolytic anemia [37]. Reported cases had also suggest that monupilavir can be incorporated into host DNA with the same mutagenesis activity that affects viral replication, putting the host at risk of mutagenesis [38]. By far none of these RdRp targeting drugs is effective enough to fulfill the need for treatment of COVID-19. To develop a better RdRp targeting drug we need to better understand the properties of RdRp and how these inhibitors function. Previous studies showed that some DNA polymerases [39], [40] and HIV reverse transcriptase [41] can use deoxynucleoside diphosphates (dNDPs) as a substrate to synthesize DNA, and Escherichia coli RNA polymerase [42] use nucleoside diphosphates (NDPs) to synthesize RNA. Although both dNDPs and NDPs showed very weak activity in these experiments, the data prompted us to test if NDPs are an active substrate for RdRp. Our purpose is to find if RdRp has some previously unknown properties which can be taken advantage of. New discoveries about the RdRp function may not only broaden our understanding of SARS-CoV-2 but also yield new strategies for antiviral drug design. This would undoubtedly aid in COVID-19 treatment. In this study, we found that SARS-CoV-2 RdRp can recognize and utilize NDPs as a substrate to synthesize RNA with an efficiency of about two-thirds of using nucleoside triphosphates (NTPs) as a substrate. In addition, we proved that molnupiravir is not only fully functional in its diphosphate form but also more active than in triphosphate form. This could imply a new strategy for designing COVID-19 treatment NA drugs with more efficacy and fewer side effects. 2 Materials and methods 2.1 Protein expression and purification Preparation of SARS-CoV-2 RdRp, composed of nsp12, nsp7 and two copied of the nsp8 subunits, was carried out as described in ref. [43]. The SARS-CoV-2 nsp12 and nsp7 gene were cloned into a modified pET-21b vector with the C-terminus possessing a 6 × His-tag. The nsp8 gene was cloned into the modified pET-32a vector with the N-terminus possessing a trx-His6-tag and PreScission Protease (PPase) site. The nsp12-pET-21b and nsp8-pET-32a plasmids were transformed into E. coli BL21 (DE3) and the transformed cells were cultured at 37 °C in LB with a final concentration of 100 μg/ml ampicillin. The nsp7-pET-21b plasmid was transformed into E. coli Rosetta-gami2 (DE3) and the transformed cells were cultured at 37 °C in LB containing a final concentration of 100 μg/ml ampicillin and 25 μg/ml chloramphenicol. Bacterial cultures were incubated by shaking at 200 rpm, 37 °C to an OD600 of 0.8, then the temperature was lowered to 16 °C and a final concentration of 0.2 mM of isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to induce protein expression for 20 h. Subsequently, the cells were harvested by centrifugation at 5000 ×g for 15 min at 4 °C. The pellet was resuspended in lysis buffer containing 25 mM TrisHCl pH 8.0, 250 mM NaCl, 4 mM MgCl2, 10 % Glycerol and homogenized with a high-pressure cell disrupter at 4 °C. The lysate was centrifuged at 25,200 ×g for 50 min at 4 °C, and the supernatant was then loaded onto Ni-NTA affinity chromatography column for purification. Nsp12, nsp7 were eluted with elution buffer (25 mM TrisHCl pH 8.0, 200 mM NaCl, 4 mM MgCl2, 250 mM Imidazole). Nsp8 was subject to on-column tag cleavage by PPase (purified in our laboratory) and then eluted by the lysis buffer. Next, all proteins were purified by ion exchange chromatography (Source 15Q HR 16/10, GE Healthcare, Boston, MA, USA) and size exclusion chromatography (Superdex 200 10/300GL, GE Healthcare, Boston, MA, USA) with 25 mM TrisHCl pH 8.0, 200 mM NaCl, 4 mM MgCl2. For nsp12-nsp7-nsp8 complex assembly, protein samples were mixed with the molar ratio of nsp7:nsp8:nsp12 = 2:2:1 at 4 °C overnight. The incubated RdRp complex was then concentrated with a 100 kDa molecular weight cut-off centrifugal filter unit (Millipore Corporation, Billerica, MA, USA) and then further purified by size exclusion chromatography using a Superdex 200 10/300 GL column in 25 mM TrisHCl pH 8.0, 200 mM NaCl, 4 mM MgCl2. SARS-CoV RdRp was purified by the same procedure. The non-structural 5 (NS5) RdRp domain of Zika Virus (ZIKV) was purified as previously described [44]. 2.2 Preparation of fluorescently labeled RNA for polymerase activity RNA template-product duplex was designed according to the published SARS-CoV-2 RNA extension assays [43]. FAM labeled 13 nt oligonucleotide with the sequence of FAM - GCUAUGUGAGAUU and the 23 nt complementary RNA strand with the sequence of AGUUAACUUUAAUCUCACAUAGC were synthesized for polymerase activity assay. Quality of the 13 + 23 nt RNA was checked three months later to make sure no degradation occur the throughout the experiment (Supplementary Fig. S3). A 26 nt RNA strand with the sequence UAGCUGGCCUUAAAAUCUCACAUAGC was synthesized for substrate efficiency assay. And a 29 nt RNA strand with the sequence AGUUAACUUUAAAGGGAAUCUCACAUAGC were used for the detection of mutation rates due to the incorporation of β-d-N4-hydroxycytidine (NHC) triphosphate (MTP) and NHC diphosphate (MDP) into RNA. All unmodified and 5’ FAM-labeled RNA oligonucleotides and DNA primers used in other experiments were purchased from Tsingke Biotechnology Co.,Ltd. (Beijing, China). The RNA strands were mixed in equal molar ratio in DEPC water, annealed by heating it to 95 °C for 10 min and gradually cooling to room temperature to make the RNA duplexes. 2.3 RdRp polymerase activity assays DEPC-treated water was used in the preparation of all solutions. RdRp at final concentration of 2 μM was incubated with 200 nM 13 + 23 nt RNA duplex and 50 μM NTPs (or NDPs/nucleoside monophosphates (NMPs)) (All nucleotides are purchased from Sigma-Aldrich, Shanghai, China) in a 20 μl reaction buffer containing 20 mM TrisHCl pH 8.0, 10 mM KCl, 10 mM MgCl2, 0.01 % Triton-X100, 1 mM DTT for 1 min at 37 °C [43], and the reactions were stopped with 2 × stop buffer (10 M urea, 50 mM EDTA). Eventually, glycerol was added to a concentration of 6.5 %. 10 μl RNA product for each reaction was resolved on 20 % denaturing polyacrylamide-urea gels (2.5 g urea, 1.2 ml 5 × TBE, 3 ml 40 % acrylamide, 40 μl 0.1 g/ml ammonium persulfate, 10 μl TEMED) and imaged with a Tanon-5200 Multi Fluorescence Imager. The experiments described were performed in triplet, unless specified otherwise. 2.4 RNA extension assays with NDPs as substrate RdRp of different viruses at final concentration of 2 μM was incubated with 200 nM 13 + 23 nt RNA duplex and 50 μM NTPs (or NDPs/NMPs/adenosine diphosphate (ADP)/guanosine diphosphate (GDP)/cytidine diphosphate (CDP)/uridine diphosphate (UDP)) in a 20 μl reaction buffer containing 20 mM TrisHCl pH 8.0, 10 mM KCl, 10 mM MgCl2, 0.01 % Triton-X100, 1 mM DTT for 1 min at 37 °C. 2.5 Sanger sequencing The RNA product, primer 1 (CCGCTCGAGCGGAGTTAACTT) and deoxy-ribonucleoside triphosphate (dNTPs) were gently mixed at 65 °C for 10 min. The mixture was cooled down to 24 °C then 10 × MMulV buffer and MMulV reverse transcriptase (purified in our laboratory) were added and incubated for 2 h to obtain cDNA. Thereafter, the reverse transcriptase was inactivated at 70 °C for 15 min. Primer 2 (CGCGGATCCGCGGCTATGTGAGATT) was then added, and the product was amplified by PCR catalyzed by the high-fidelity DNA polymerase pfu-Phusion (purified in our laboratory). The PCR product was digested with XhoІ and BamHІ (Purchased from Takara Bio, Tianjin, China) and ligated into pET-15b vector, and transformed into E-coli DH5α to obtain clones. Plasmids from positive clones were then sequenced with T7 universal primers. 2.6 High-throughput sequencing The dsRNA products were mixed with primer 1′ (acgatgcaaagtctcgacaaatggtcgataccaattcaCCGCTCGAGCGGAGTTAACTT) and dNTPs at 65 °C for 10 min and then cooled down to 24 °C for primer annealing. 10 × MMulV buffer, RNase inhibitor, and MMulV reverse transcriptase were added to the above system at 24 °C for 2 h to obtain cDNA, and then the enzymes were inactivated at 70 °C for 15 min. Subsequently, primer 2′ (cagataaactataattcctaatcgcgaggtggcactgcaaCGCGGATCCGCGGCTATGTGAGA) and high-fidelity DNA polymerase pfu-Phusion were added to perform PCR amplification. PCR products were then purified by agarose gel electrophoresis for high-throughput sequencing (Sangon Biotech, Shanghai, China). 2.7 Substrate efficiency assay RNA extension was performed by incubating 2 μM RdRp, 250 nM 13 + 23 nt RNA duplex, 50 μM substrate (which are NTPs, NDPs) in a 20 μl reaction system at 37 °C for 0, 10, 20, 30, 40, 50, 60 s. RdRp catalyzed reactions each containing three NTPs and one NDP (ADP, guanosine triphosphate (GTP), cytidine triphosphate (CTP), uridine triphosphate (UTP); adenosine triphosphate(ATP), GDP, CTP, UTP; ATP, GTP, CDP, UTP and ATP, GTP, CTP, UDP) and 13 + 26 nt RNA duplex was performed for 1 min. The reactions were stopped with 2 × stop buffer. Grayscale analysis was then carried out using ImageJ and illustrated by using OriginPro 8 (OriginLab, Northampton, Massachusetts, USA). 2.8 RNA extension assays with MDP as substrate A 13 + 26 nt RNA duplex was used in this assay (Fig. 2c). The 5′ end of the RNA product strand was labeled with a 6-carboxyfluorescein (FAM) group, which allows us to monitor RNA elongation. RdRp can incorporate NHC monophosphate into RNA at the position where the nucleoside in the template is G or A. MDP and MTP were purchased from MedChemExpress (Shanghai, China). For RNA extension RdRp at final concentration of 2 μM was incubated with 200 nM 13 + 26 nt RNA duplex and 50 μM each of ATP, GTP, UTP or 50 μM each of ATP, GTP, CTP with or without 50 μM MTP/MDP in 20 μl reaction buffer for 1 min at 37 °C. In the experiment to detect the specificity of other viral RdRp using MTP and MDP as substrate, because ZIKV NS5 had a poor polymerization effect on 13 + 26 nt RNA template, 13 + 23 nt RNA was used as the template. The reactions were stopped with 2 × stop buffer. 10 μl RNA product for each reaction was resolved on 20 % denaturing polyacrylamide-urea gels and imaged with a Tanon-5200 Multi Fluorescence Imager to confirm the extension. Grayscale analysis was then carried out using ImageJ and illustrated by using OriginPro 8. 2.9 Incorporation mutation rate detection RdRp at final concentration of 2 μM was incubated with 200 nM 13 + 29 nt RNA duplex and 50 μM each of ATP, GTP, UTP; 0 or 800 nM of CTP; 50 μM of MTP/MDP in 20 μl reaction buffer for 1 min at 37 °C. Mutation rates were determined by High-throughput sequencing as mentioned above. Raw reads were filtered according to three steps: 1) Removing adaptor sequence if reads contains by cutadapt (v 1.2.1); 2) Removing low quality bases from reads 3′ to 5′ (Q < 20) by PRINSEQ-lite (v 0.20.3); 3) Removing chimras sequence by usearch software (v11.0.667) with de novo mode by default parameter. And the remaining clean data were used for further analysis. A frequency of 0.001 was used as a cutoff for variants. The absolute number and type of mutations in three consecutive C bases of 13 + 29 nt RNA are reported. The percentage of the total mutations for each specific mutation type was calculated using these numbers. The difference in percentage for each class of mutation compared with NTPs control is referred to as the relative proportion of these mutations. 2.10 Fluorescence assay for T7 RNA polymerase (RNAP) transcription Transcription of T7 RNAP was measured by a fluorescence assay which employs a pair of nucleic acid probes with fluorescent and quenching groups respectively to detect mRNA production. When the probes pair with the mRNA FAM and quenching group gets closer and the fluorescence produced by FAM will be quenched. The change of the fluorescent signal is proportional to the mRNA production. 3 Results 3.1 RdRp synthesizes RNA by using both NTPs and NDPs as a substrate Since RdRp is by far one of the most attractive targets for anti-coronavirus drug development, a thorough characterization of RdRp polymerase activity will, without doubt, greatly contribute to discovering the effective strategy for drug screening [17], [18]. The RdRp of SARS-CoV-2 was purified by assembling nsp12, nsp7 and nsp8 (Fig. 1a). Based on previous research, SARS-CoV-2 RdRp polymerase activity was measured by conducting a conventional RNA elongation assay [43]. When NTPs was added, elongation catalyzed by RdRp complex occurred on the RNA duplex, resulting in an intact double-stranded RNA product (Fig. 1c).Fig. 1 RdRp uses both NTPs and NDPs as a substrate. a Size-exclusion chromatogram of the SARS-CoV-2 nsp12-nsp7-nsp8 (RdRp) complex. RdRp was also characterized by SDS-PAGE. b The 13 + 23 nt RNA template-product duplex. The direction of RNA extension is shown. The colour of the depicted circles indicates the experimental design: blue, RNA template strand; red, RNA product strand. The 5′ end of the RNA product contains a FAM fluorescent label. c Incubation of the RdRp with RNA duplex and NTPs leads to RNA extension. Nsp12 alone has almost no polymerase activity, and requires nsp7 and nsp8 to form an RdRp complex to have polymerase activity. d RdRp was incubated with the 13 + 23 nt duplex in the presence of NMPs, NDPs or NTPs respectively. These are showing duplicate experiments performed with different batches RdRp. The positions of the template RNA and the full-length extension product are indicated on denaturing gel. e Electrophoretic separation of reaction products of single nucleoside diphosphates as substrates. The positions of the original RNA and extension products are indicated. f Part of the Sanger sequencing chromatogram of the RT-PCR products corresponding to the extended 10 nt RNA on the 13 + 23 nt RNA duplex with NDPs as substrate. The dash line circle represents the 10 nt extension. g Schematic of high-throughput sequencing of RdRp reaction products. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 1 To test if RdRp uses NDPs or NMPs as a substrate, we performed an RNA extension assay by using NTPs, NDPs or NMPs as substrate, respectively. The result revealed that RdRp synthesized RNA products of the same length regardless of NTPs or NDPs as substrate (Fig. 1d). To examine the template specificity of this RNA synthesis, 13 + 23 nt RNA was extended in the presence of ADP, GDP, CDP, and UDP, respectively (Fig. 1b). Only the cognate ADP promoted RNA extension, indicating that NDPs incorporation is template-specific (Fig. 1e). This result also raised the question of whether the RNA made from NDPs has the right sequence. Hence, two experiments were performed: single-clonal sequencing of ligated plasmids, and high-throughput sequencing of PCR products. First, the cDNA was synthesized through reverse transcription of the RNA products, and then amplified by 10 cycles of PCR with primers containing the restriction sites of XhoІ and BamHІ. PCR products were then ligated into the vector pET-15b and used to transform E. coli DH5α strain. Plasmids extracted from the positive clones were sequenced by Sanger sequencing. The results gave a DNA sequence that correlated with the sequence of the nascent RNA, which is the complementary sequence of the RNA template (Fig. 1f). Second, the cDNA amplified by 10 cycles of PCR was sequenced by high-throughput sequencing (Fig. 1g). At least 50,000 sequences were obtained for each product (NTPs is Supplementary File SI.H1, NDPs is Supplementary File SI.H2). The same result was obtained whether NDPs or NTPs was were used as substrate. These two experiments clearly indicated confirm that the RNA synthesized from NDPs has the right sequence. The ADP, GDP, CDP and UDP used in these experiments were characterized by mass spectrometry respectively to make sure these NDPs were not contaminated by NTPs (Supplementary Fig. S1). 3.2 NDPs are nearly two thirds as efficient as NTPs as a substrate of RdRp Previous studies indicated that although some DNA polymerases and RNA polymerases are able to use diphosphate form substrate. The activity of the diphosphate form substrate, however, is much lower than that of the triphosphate form substrate [40], [41], [42]. To determine if this is also the case for RdRp we performed a kinetics study by using NTPs and NDPs as a substrate respectively. The 13 + 23 nt RNA duplex was used as a template and the elongation of product over time was observed (Fig. 2a). Kinetics curves indicated that the relative incorporation efficiency of NDPs is about two thirds (67.94 % ± 2.66 %) of that of NTPs (Fig. 2b). These data suggested that the incorporation rate for NDPs is slower than that of NTPs. Subsequently, we decided to test if all four NDPs equally contribute to the slower incorporation rate. A 13 + 26 nt RNA duplex was synthesized for the assay testing the incorporation efficiency for each NDP (Fig. 2c). Each reaction system contains three NTPs and the NDP of interest (Fig. 2d). Unexpectedly, our data showed that substitution of any single NTP by NDP has no observable effect on the efficiency of RNA synthesis (Fig. 2e). This may imply that the speed of RNA synthesis is only affected by continuous NDP incorporation but not the intermittent insertion.Fig. 2 The activities of NDPs and NTPs are comparable. a Primer-extension reactions on 13 + 23 nt RNA templates sampled at indicated time points. b Quantification of elongation products for (a). The experiment was performed in triplet. c The 13 + 26 nt RNA template-product duplex. The direction of RNA extension is shown. The 5′ end of the RNA product was labeled by FAM. d RdRp catalyzed reactions containing different substrates. Lane 1: no substrate; lane 2: NTPs; lane 3: ATP, GTP, CTP; lane4: ATP, GTP, CTP, UDP; lane5: GTP, CTP, UTP; lane6: GTP, CTP, UTP, ADP; lane7: ATP, CTP, UTP; lane8: ATP, CTP, UTP, GDP; lane9: ATP, GTP, UTP; lane10: ATP, GTP, UTP, CDP. The positions of the 13 + 26 nt RNA and extension products are indicated. e Quantification of elongation products for (d). Data are shown as means ± standard and analyzed using unpaired t-test. There are not statistically significant. Fig. 2 3.3 SARS-CoV-2 RdRp incorporates NHC into RNA by using diphosphate form molnupiravir as a substrate Since RdRp has become one of the most important targets for the development of antiviral drugs, many NA drugs have been constructed to target RdRp for COVID-19 treatment [34], [45], [46], [47]. Among all these chemicals reported by far, molnupiravir has been regarded as a promising drug candidate. The molecular mechanism of molnupiravir was determined recently. Once ingested by patients, molnupiravir undergoes stepwise phosphorylation to yield the active nucleoside triphosphate analogue (MTP). The RdRp then uses MTP as substrate and incorporates NHC monophosphate into RNA at the position where the nucleoside in the template is G or A, resulting in “error catastrophe” and virus death [36], [38], [48] (Fig. 3a).Fig. 3 RdRp incorporates NHC into RNA with MDP as a substrate. a Metabolism of NA molnupiravir in host cell. A two-step model of molnupiravir-induced RNA mutagenesis. b The 13 + 26 nt RNA allows for RNA extension by 13 nucleotides. NHC monophosphate can be incorporated into growing RNA instead of C or U. c RdRp catalyzed reactions containing different substrates. Lane 1: no substrate; lane 2: NTPs; lane 3: ATP, GTP, UTP; lane4: ATP, GTP, UTP, MTP; lane5: ATP, GTP, UTP, MDP; lane6: ATP, GTP, CTP; lane7: ATP, GTP, CTP, MTP; lane8: ATP, GTP, CTP, MDP. The positions of the 13 + 26 nt RNA and extension products are indicated. RNA elongation stalls at the expected positions when the cognate NTP is absent from the reaction. d Quantification of elongation products for (c). Statistical significance is indicated as compared with NTPs as a substrate using a t-test. All of the values shown represent the mean ± standard deviation of the results from three independent experiments (n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001). e The 13 + 29 nt RNA allows for RNA extension by 16 nucleotides. NHC monophosphate can pair with G or A for incorporation into growing RNA. Fig. 3 Although the mechanism of molnupiravir has been well established, our finding of RdRp's ability to utilize NDPs as a substrate still raises a question: is MDP also an active form and recognized by RdRp as a substrate? To address this question, the 13 + 26 nt RNA duplex was used to perform the RNA extension assay with NDPs and MDP as the substrates in comparison with the extension using NTPs and MTP as the substrates. The 13 + 26 nt RNA duplex allows ten nucleotides (nt) of extension. The 13 + 26 nt RNA duplex contains two adjacent G and two adjacent A, allowing NHC incorporation into the RNA (Fig. 3b). In this case, when CTP or UTP is replaced by MDP, the incorporation of NHC does not hinder the incorporation of the next subsequent nucleotide, and RdRp still completes the extension reaction. Similar results were obtained using MTP as the substrate (Fig. 3c). Therefore, it is reasonable to speculate that both MTP and MDP are active forms of monulpiravir in humans. Molnupiravir is an isopropylester prodrug of NHC. We speculate that when the molnupiravir prodrug enters the cells, it is sequentially converted to NHC-monophosphate, MDP and MTP. Since RdRp of SARS-CoV-2 can use both MDP and MTP as a substrate instead of CTP or UTP, mutations in the genomic (−gRNA) and subgenomic RNA will accumulate. Over time, lethal mutations may occur and the viruses inside the cells are eliminated. 3.4 MDP is a better substrate than MTP Since both MTP and MDP are active substrates of RdRp we performed an assay to figure out which one is the better substrate. By observing their efficiency in reaction, we found that when the template is G, the relative incorporation efficiency of MTP is 0.78, while MDP is 0.87 in comparison to CTP. When the template is A, the relative incorporation efficiency of MTP is 0.58, while MDP is 0.77 in comparison to UTP (Fig. 3d). In order to obtain a reliable result, the experiment was repeated six times with two batches of RdRp. The results indicated MDP is the better substrate for RdRp in comparison to MTP. MTP and MDP were characterized by HPLC and mass spectrometry to make sure they are the right pure compounds (Supplementary Fig. S2). Fig. 4 MDP and MTP drives an increase in similar low-frequency C > T transition mutations. a - c, f, g. Numbers (#/10,000 base) of mutations in the presence of condition 1 (50 μM each of ATP, GTP, UTP, CTP) (a), condition 2 (50 μM each of ATP, GTP, UTP, MDP) (b), condition 3 (50 μM each of ATP, GTP, UTP, MTP) (c), condition 4 (50 μM each of ATP, GTP, UTP, MDP and 800 nM of CTP) (f) or condition 5 (50 μM each of ATP, GTP, UTP, MTP and 800 nM of CTP) (g) presented by type. d,e and h,i. Changes in relative proportions of each mutation type after added with condition 2 (d), condition 3 (e), condition 4 (h) or condition 5 (i) compared to NTPs control. The relative proportions of C > T transitions are indicated by green shading. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Fig. 4 Next, we tried to determine if MDP is also more efficient in introducing mutations than MTP. The products of RdRp-catalyzed extension reactions containing MTP or MDP using 13 + 29 nt RNA as a template (Fig. 3e) were reverse-transcribed and PCR-amplified, and the number of mutations was counted by high-throughput sequencing. The statistical results were shown in Fig. 4 and Supplementary File SI.H3. Further analysis of the types of mutations introduced by NHC revealed an increase in the number of C > T mutations when MDP or MTP was added (Fig. 4a-c, f-g). The relative proportions of C > T transitions among all observed mutations were increased by 13 % (MDP) and 12 % (MTP) in the absence of CTP (Fig. 4d and e), by 11 % (MDP) and 12 % (MTP) in the presence of 800 nM CTP compared to the NTPs control (Fig. 4h and i). Together, similar mutation ratios were found when the same concentrations of MTP and MDP were added. 3.5 Comparison of substrate specificity of RNA polymerases from different viruses To test if other virus RNA polymerases also use NDPs as a substrate, we purified SARS-CoV RdRp, NS5 protein of ZIKV and T7 RNAP and measured their activities when NDPs were used as a substrate. Results indicated that of all the RNA polymerases tested SARS-CoV-2 RdRp had the highest activity when using NDPs as substrate. SARS-CoV RdRp could also use NDPs as a substrate, but the product band seam a double RNA strand shorter than expectation implying stalled extension. NS5 of ZIKV did not produce a well-defined product band when using NDPs as a substrate (Fig. 5a). The transcription of T7 RNAP was measured based on fluorescence technology. We found that except for the case in which ATP was replaced by ADP and weak transcription occurred, any other NTP replaced by NDP would inhibit the transcription completely (Fig. 5b).Fig. 5 Comparison of substrate specificity of RNA polymerases from different viruses. a RdRp of SARS-CoV-2 and SARS-CoV and NS5 of ZIKV were incubated with the 13 + 23 nt RNA duplex in the presence of NTPs or NDPs respectively. The positions of the template RNA and the full-length extension product are indicated. b The transcription of T7 RNAP was measured by monitoring the signal of a fluorescent probe. Fluorescent signal continuously changed when transcription occurred, indicating the presence of RNA product. However, when NDPs were used as a substrate, the fluorescent signal remained unchanged, indicating the lack of RNA product. Substitution of ATP by ADP led to weak transcription, any other NTP replaced by NDP as well as UTP replaced by MTP or MDP would inhibit the transcription completely. c RdRp of SARS-CoV-2, SARS-CoV and NS5 of ZIKV were incubated with the 13 + 23 nt RNA duplex in the presence of NTPs; ATP, GTP, CTP; ATP, GTP, CTP, MTP; ATP, GTP, CTP, MDP respectively. The positions of the template RNA and the full-length extension product are indicated. d Quantification of elongation products for (c). Statistical significance is indicated as compared with NTPs as a substrate using a t-test. All of the values shown represented the mean ± standard deviation of the results from three independent experiments (n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001). Fig. 5 Substrate specificity of these enzymes was also tested using MTP or MDP as a substrate. As expected, SARS-CoV RdRp could also effectively use MTP and MDP as a substrate. When UTP is replaced with MDP as a substrate, the relative incorporation efficiency reached 0.77 ± 0.01. Efficiency was even higher when MTP was used as a substrate (0.70 ± 0.02). NS5 of ZIKV weakly incorporates NHC into RNA when MTP or MDP was used to replace UTP, giving the relative incorporation an efficiency of 0.08 ± 0.03 for MTP and 0.06 ± 0.02 for MDP (Fig. 5c,d). As the only DNA dependent RNA polymerase in this experiment, T7 RNAP cannot use MTP and MDP as a substrate (Fig. 5b). 4 Discussion Since its outbreak in late 2019, the COVID-19 pandemic has been plaguing people for three years and still poses a threat to human health [3], [11]. Although many kinds of vaccines have been developed to suppress the pandemic, they are not sufficient to withhold the spread of SARS-CoV-2 [49]. The development of potent antivirals against SARS-CoV-2 is still an urgent need [6], [50]. Since RdRp is critical for viral genome replication and transcription and conserved in RNA virus species, RdRp has become one of the most appealing targets for antivirals development [17], [43], [50], [51]. Thoroughly characterization of RdRp both structurally and biochemically would no doubt be beneficial to drug development. The substrate usage of RdRp is astonishing. To the best of our knowledge, most RNA polymerases, including DNA-dependent and RNA-dependent RNA polymerases, use NTPs as a substrate to synthesize RNA under the guidance of a template strand. We found that SARS-CoV-2 RdRp synthesizes RNA with comparable efficiency using NTPs and NDPs as a substrate, and other RNA viruses that can utilize NDPs as a substrate have also been found. We speculate that the ability to use NDPs as a substrate may be an advantage of RNA virus during infection. During later stages of infection, host cells may not be able to produce enough ATP and other NTPs. By using NDPs as a substrate, SARS-CoV-2 can continue the process of genome replication and assembly of progeny viruses, while most metabolic activities stop. It has long been a general knowledge that to become active against virus NA drugs must undergo stepwise addition of phosphate groups to become the triphosphate form. In order to produce this outcome, this type of antivirals must be recognized by three kinds of kinases [52]. This raises a concern that the triphosphate form of antivirals may be incorporated into host mRNA [48]. Even worse, mutagenic ribonucleoside analogs could be reduced into the 2′-deoxyribonucleotide form by host ribonucleotide reductase and then incorporated into DNA [38], [48]. Our finding that RdRp also uses MDP as a substrate could largely resolve these concerns. Since the diphosphate form is also active, there would be no issue regarding whether the prodrugs can become triphosphate form inside host cells. Furthermore, nucleoside diphosphate analogues, which can evade nucleoside diphosphate kinase (NDPK), would cease to have the possibility of being incorporated into mRNA or DNA. The best strategy is to use the membrane-permeable nucleoside diphosphate analog for COVID-19 treatment [52], [53]. The advantage of this strategy is that the drug is not only able to bypass all the phosphorylation steps and has no risk of incorporation into host mRNA or DNA (Fig. 6 ).Fig. 6 Nucleoside diphosphate analog prodrugs designed by our strategy would have less potential to cause side effects. The terminal phosphate group (β-phosphate) is modified by lipophilic masks to give the prodrug more membrane permeability. Inside cell the masks are removed by enzymatic or chemical reactions. The ribose moiety and the base group should be designed in such a way that the diphosphate form drug can evade NDPK thus eliminating the risk of being incorporated into host mRNA or DNA (Note: NDPs and dNDPs are phosphorylated by the same NDPK). Fig. 6 Previous studies have indicated that the error rate (#mutations/10,000 bases) of viral genes treated with 10 μM NHC increased from 0.015 to 0.09 [54]. The G > A and C > U transition mutations are detectable at low frequencies across the genome [30]. This suggests that incorporation of NHC is discontinuously. The ability of RdRp to incorporate NHC into a short template implies the ability to incorporate NHC into the long genome at any position. Since RdRp is the core of the replication complex, its substrate specificity is unlikely determined by other proteins. However, our results obtained from in vitro data are preliminary. They may be further validated by additional tests in in vivo SARS-CoV-2 infections. The following are the supplementary data related to this article.Supplementary figures Image 1 Supplementary File SI.H1 Supplementary File SI.H1 Supplementary File SI.H2 Supplementary File SI.H2 Supplementary File SI.H3 Supplementary File SI.H3 CRediT authorship contribution statement Maofeng Wang: Methodology, Experimental verification, Investigation, Data curation, Writing-manuscript, Visualization. Cancan Wu: Experimental verification, Writing - Review & Editing. Nan Liu: Experimental verification, Writing - Review & Editing. Fengyu Zhang: Experimental verification. Hongjie Dong: Formal analysis. Shuai Wang: Experimental verification. Min Chen: Experimental verification. Xiaoqiong Jiang: Experimental verification. Kundi Zhang: Visualization, Writing - Review & Editing. Lichuan Gu: Conceptualization, Methodology, Writing - review & editing, Supervision, Funding acquisition. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could influence the work reported in this paper. Data availability All data generated or analyzed during this study are included in this published article (and its Supporting information). Acknowledgments The authors thank Yan Zhang at Zhejiang University for offering SARS-CoV-2 RdRp gene, Peihui Wang at Cheeloo College of Medicine, Shandong University for offering SARS-CoV RdRp gene, Haitao Yang at ShanghaiTech University for offering ZIKV NS5 gene, Wuxi Biortus Biosciences Co. Ltd. for providing SARS-CoV-2 RdRp expressed in insect cells, Carina Muyao Gu at CBe-Learn School for helping with the English expression of the manuscript. This work was supported by Shandong Provincial Key 10.13039/100006190 Research and Development Program (2020CXGC011305). ==== Refs References 1 Chen Y. Liu Q. Guo D. Emerging coronaviruses: genome structure, replication, and pathogenesis J. Med. Virol. 92 2020 418 423 10.1002/jmv.25681 31967327 2 Zhu N. Zhang D. Wang W. Li X. Yang B. Song J. Zhao X. Huang B. Shi W. Lu R. Niu P. Zhan F. Ma X. Wang D. Xu W. Wu G. Gao G.F. Tan W. China Novel Coronavirus Investigating and Research Team, a novel coronavirus from patients with pneumonia in China, 2019 N. Engl. J. Med. 382 2020 727 733 10.1056/NEJMoa2001017 31978945 3 Lu R. Zhao X. Li J. Niu P. Yang B. Wu H. Wang W. Song H. Huang B. Zhu N. Bi Y. Ma X. Zhan F. Wang L. Hu T. Zhou H. Hu Z. Zhou W. Zhao L. Chen J. Meng Y. Wang J. Lin Y. Yuan J. Xie Z. Ma J. Liu W.J. Wang D. Xu W. Holmes E.C. Gao G.F. Wu G. Chen W. Shi W. Tan W. Genomic characterisation and epidemiology of 2019 novel coronavirus: implications for virus origins and receptor binding Lancet 395 2020 565 574 10.1016/S0140-6736(20)30251-8 32007145 4 Chan J.F. Kok K.H. Zhu Z. Chu H. To K.K. Yuan S. Yuen K.Y. Genomic characterization of the 2019 novel human-pathogenic coronavirus isolated from a patient with atypical pneumonia after visiting Wuhan Emerg. Microbes Infect. 9 2020 221 236 10.1080/22221751.2020.1719902 31987001 5 Coronaviridae Study Group of the International Committee on Taxonomy of Viruses The species Severe acute respiratory syndrome-related coronavirus: classifying 2019-nCoV and naming it SARS-CoV-2 Nat. Microbiol. 5 2020 536 544 10.1038/s41564-020-0695-z 32123347 6 Liu C. Zhou Q. Li Y. Garner L.V. Watkins S.P. Carter L.J. Smoot J. Gregg A.C. Daniels A.D. Jervey S. Albaiu D. Research and development on therapeutic agents and vaccines for COVID-19 and related human coronavirus diseases ACS Cent. Sci. 6 2020 315 331 10.1021/acscentsci.0c00272 32226821 7 Harvey W.T. Carabelli A.M. Jackson B. Gupta R.K. Thomson E.C. Harrison E.M. Ludden C. Reeve R. Rambaut A. COVID-19 Genomics UK (COG-UK) ConsortiumPeacock S.J. Robertson D.L. SARS-CoV-2 variants, spike mutations and immune escape Nat. Rev. Microbiol. 19 2021 409 424 10.1038/s41579-021-00573-0 34075212 8 Chen J. Gao K. Wang R. Wei G.W. Prediction and mitigation of mutation threats to COVID-19 vaccines and antibody therapies Chem. Sci. 12 2021 6929 6948 10.1039/d1sc01203g 34123321 9 Chen J. Gao K. Wang R. Wei G.W. Revealing the threat of emerging SARS-CoV-2 mutations to antibody therapies J. Mol. Biol. 433 2021 167155 10.1016/j.jmb.2021.167155 10 Rudd C.E. GSK-3 inhibition as a therapeutic approach against SARs CoV2: dual benefit of inhibiting viral replication while potentiating the immune response Front. Immunol. 11 2020 1638 10.3389/fimmu.2020.01638 32695123 11 Wu F. Zhao S. Yu B. Chen Y.M. Wang W. Song Z.G. Hu Y. Tao Z.W. Tian J.H. Pei Y.Y. Yuan M.L. Zhang Y.L. Dai F.H. Liu Y. Wang Q.M. Zheng J.J. Xu L. Holmes E.C. Zhang Y.Z. A new coronavirus associated with human respiratory disease in China Nature 579 2020 265 269 10.1038/s41586-020-2008-3 32015508 12 Zhang Y.Z. Holmes E.C. A genomic perspective on the origin and emergence of SARS-CoV-2 Cell 181 2020 223 227 10.1016/j.cell.2020.03.035 32220310 13 Wu C. Liu Y. Yang Y. Zhang P. Zhong W. Wang Y. Wang Q. Xu Y. Li M. Li X. Zheng M. Chen L. Li H. Analysis of therapeutic targets for SARS-CoV-2 and discovery of potential drugs by computational methods Acta Pharm. Sin. B 10 2020 766 788 10.1016/j.apsb.2020.02.008 32292689 14 Subissi L. Posthuma C.C. Collet A. Zevenhoven-Dobbe J.C. Gorbalenya A.E. Decroly E. Snijder E.J. Canard B. Imbert I. One severe acute respiratory syndrome coronavirus protein complex integrates processive RNA polymerase and exonuclease activities Proc. Natl. Acad. Sci. U. S. A. 111 2014 E3900 E3909 10.1073/pnas.1323705111 25197083 15 Kirchdoerfer R.N. Ward A.B. Structure of the SARS-CoV nsp12 polymerase bound to nsp7 and nsp8 co-factors Nat. Commun. 10 2019 2342 10.1038/s41467-019-10280-3 31138817 16 Hillen H.S. Kokic G. Farnung L. Dienemann C. Tegunov D. Cramer P. Structure of replicating SARS-CoV-2 polymerase Nature 584 2020 154 156 10.1038/s41586-020-2368-8 32438371 17 McKee D.L. Sternberg A. Stange U. Laufer S. Naujokat C. Candidate drugs against SARS-CoV-2 and COVID-19 Pharmacol. Res. 157 2020 104859 10.1016/j.phrs.2020.104859 18 Chu C.K. Gadthula S. Chen X. Choo H. Olgen S. Barnard D.L. Sidwell R.W. Antiviral activity of nucleoside analogues against SARS-coronavirus (SARS-CoV) Antivir. Chem. Chemother. 17 2006 285 289 10.1177/095632020601700506 17176633 19 Singh R. Bhardwaj V.K. Das P. Bhattacherjee D. Zyryanov G.V. Purohit R. Benchmarking the ability of novel compounds to inhibit SARS-CoV-2 main protease using steered molecular dynamics simulations Comput. Biol. Med. 146 2022 105572 10.1016/j.compbiomed.2022.105572 20 Bhardwaj V.K. Singh R. Sharma J. Rajendran V. Purohit R. Kumar S. Bioactive molecules of tea as potential inhibitors for RNA-dependent RNA polymerase of SARS-CoV-2 Front.Med. 8 2021 684020 10.3389/fmed.2021.684020 21 Singh R. Bhardwaj V.K. Das P. Purohit R. A computational approach for rational discovery of inhibitors for non-structural protein 1 of SARS-CoV-2 Comput. Biol. Med. 135 2021 104555 10.1016/j.compbiomed.2021.104555 22 Bhardwaj V.K. Singh R. Das P. Purohit R. Evaluation of acridinedione analogs as potential SARS-CoV-2 main protease inhibitors and their comparison with repurposed anti-viral drugs Comput. Biol. Med. 128 2021 104117 10.1016/j.compbiomed.2020.104117 23 Bhardwaj V.K. Singh R. Sharma J. Rajendran V. Purohit R. Kumar S. Identification of bioactive molecules from tea plant as SARS-CoV-2 main protease inhibitors J. Biomol. Struct. Dyn. 39 2021 3449 3458 10.1080/07391102.2020.1766572 32397940 24 Singh R. Bhardwaj V.K. Sharma J. Kumar D. Purohit R. Identification of potential plant bioactive as SARS-CoV-2 spike protein and human ACE2 fusion inhibitors Comput. Biol. Med. 136 2021 104631 10.1016/j.compbiomed.2021.104631 25 Singh R. Bhardwaj V.K. Purohit R. Inhibition of nonstructural protein 15 of SARS-CoV-2 by golden spice: a computational insight Cell Biochem. Funct. 2022 10.1002/cbf.3753 26 Singh R. Bhardwaj V.K. Sharma J. Purohit R. Kumar S. In-silico evaluation of bioactive compounds from tea as potential SARS-CoV-2 nonstructural protein 16 inhibitors J. Tradit. Complement. Med. 12 2022 35 43 10.1016/j.jtcme.2021.05.005 34099976 27 Sharma J. Kumar Bhardwaj V. Singh R. Rajendran V. Purohit R. Kumar S. An in-silico evaluation of different bioactive molecules of tea for their inhibition potency against non structural protein-15 of SARS-CoV-2 Food Chem. 346 2021 128933 10.1016/j.foodchem.2020.128933 28 Singh R. Bhardwaj V.K. Purohit R. Potential of turmeric-derived compounds against RNA-dependent RNA polymerase of SARS-CoV-2: an in-silico approach Comput. Biol. Med. 139 2021 104965 10.1016/j.compbiomed.2021.104965 29 Stein D.S. Moore K.H.P. Phosphorylation of nucleoside analog antiretrovirals: a review for clinicians Pharmacotherapy 21 2001 11 34 10.1592/phco.21.1.11.34439 11191730 30 Agostini M.L. Pruijssers A.J. Chappell J.D. Gribble J. Lu X. Andres E.L. Bluemling G.R. Lockwood M.A. Sheahan T.P. Sims A.C. Natchus M.G. Saindane M. Kolykhalov A.A. Painter G.R. Baric R.S. Denison M.R. Small-molecule antiviral β-d-N (4)-hydroxycytidine inhibits a proofreading-intact coronavirus with a high genetic barrier to resistance J. Virol. 93 2019 e01348-19 10.1128/JVI.01348-19 31 Picarazzi F. Vicenti I. Saladini F. Zazzi M. Mori M. Targeting the RdRp of emerging RNA viruses: the structure-based drug design challenge Molecules 25 2020 5695 10.3390/molecules25235695 33287144 32 Crotty S. Maag D. Arnold J.J. Zhong W. Lau J.Y.N. Hong Z. Andino R. Cameron C.E. The broad-spectrum antiviral ribonucleoside ribavirin is an RNA virus mutagen Nat. Med. 6 2000 1375 1379 10.1038/82191 11100123 33 Shannon A. Selisko B. Le N.T. Huchting J. Touret F. Piorkowski G. Fattorini V. Ferron F. Decroly E. Meier C. Coutard B. Peersen O. Canard B. Rapid incorporation of favipiravir by the fast and permissive viral RNA polymerase complex results in SARS-CoV-2 lethal mutagenesis Nat. Commun. 11 2020 4682 10.1038/s41467-020-18463-z 32943628 34 Wang M. Cao R. Zhang L. Yang X. Liu J. Xu M. Shi Z. Hu Z. Zhong W. Xiao G. Remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (2019-nCoV) in vitro Cell Res. 30 2020 269 271 10.1038/s41422-020-0282-0 32020029 35 Sheahan T.P. Sims A.C. Zhou S. Graham R.L. Pruijssers A.J. Agostini M.L. Leist S.R. Schäfer A. Dinnon K.H. III Stevens L.J. Chappell J.D. Lu X. Hughes T.M. George A.S. Hill C.S. Montgomery S.A. Brown A.J. Bluemling G.R. Natchus M.G. Saindane M. Kolykhalov A.A. Painter G. Harcourt J. Tamin A. Thornburg N.J. Swanstrom R. Denison M.R. Baric R.S. An orally bioavailable broad-spectrum antiviral inhibits SARS-CoV-2 in human airway epithelial cell cultures and multiple coronaviruses in mice Sci. Transl. Med. 12 2020 eabb5883 10.1126/scitranslmed.abb5883 36 Kabinger F. Stiller C. Schmitzova J. Dienemann C. Kokic G. Hillen H.S. Hobartner C. Cramer P. Mechanism of molnupiravir-induced SARS-CoV-2 mutagenesis Nat. Struct. Mol. Biol. 28 2021 740 746 10.1038/s41594-021-00651-0 34381216 37 Takahashi T. Luzum J.A. Nicol M.R. Jacobson P.A. Pharmacogenomics of COVID-19 therapies NPJ Genom. Med. 5 2020 35 10.1038/s41525-020-00143-y 32864162 38 Zhou S. Hill C.S. Sarkar S. Tse L.V. Woodburn B.M.D. Schinazi R.F. Sheahan T.P. Baric R.S. Heise M.T. Swanstrom R. Beta-d-N4-hydroxycytidine inhibits SARS-CoV-2 through lethal mutagenesis but is also mutagenic to mammalian cells J. Infect. Dis. 224 2021 415 419 10.1093/infdis/jiab247 33961695 39 Yang G. Franklin M. Li J. Lin T.C. Konigsberg W. Correlation of the kinetics of finger domain mutants in RB69 DNA polymerase with its structure Biochemistry 41 2002 2526 2534 10.1021/bi0119924 11851399 40 Burke C.R. Luptak A. DNA synthesis from diphosphate substrates by DNA polymerases Proc. Natl. Acad. Sci. U. S. A. 115 2018 980 985 10.1073/pnas.1712193115 29339523 41 Garforth S.J. Parniak M.A. Prasad V.R. Utilization of a deoxynucleoside diphosphate substrate by HIV reverse transcriptase PLoS One 3 2008 e2074 10.1371/journal.pone.0002074 42 Gottesman M.E. Mustaev A. Ribonucleoside-5'-diphosphates (NDPs) support RNA polymerase transcription, suggesting NDPs may have been substrates for primordial nucleic acid biosynthesis J. Biol. Chem. 294 2019 11785 11792 10.1074/jbc.RA119.009074 31189650 43 Yin W. Mao C. Luan X. Shen D.-D. Shen Q. Su H. Wang X. Zhou F. Zhao W. Gao M. Chang S. Xie Y.-C. Tian G. Jiang H.-W. Tao S.-C. Shen J. Jiang Y. Jiang H. Xu Y. Zhang S. Zhang Y. Xu H.E. Structural basis for inhibition of the RNA-dependent RNA polymerase from SARS-CoV-2 by remdesivir Science 368 2020 1499 1504 10.1126/science.abc1560 32358203 44 Ferrero D.S. Ruiz-Arroyo V.M. Soler N. Uson I. Guarne A. Verdaguer N. Supramolecular arrangement of the full-length Zika virus NS5 PLoS Pathog. 15 2019 e1007656 10.1371/journal.ppat.1007656 45 Choy K.T. Wong A.Y. Kaewpreedee P. Sia S.F. Chen D. Hui K.P.Y. Chu D.K.W. Chan M.C.W. Cheung P.P. Huang X. Peiris M. Yen H.L. Remdesivir, lopinavir, emetine, and homoharringtonine inhibit SARS-CoV-2 replication in vitro Antivir. Res. 178 2020 104786 10.1016/j.antiviral.2020.104786 46 Williamson B.N. Feldmann F. Schwarz B. Meade-White K. Porter D.P. Schulz J. van Doremalen N. Leighton I. Yinda C.K. Perez-Perez L. Okumura A. Lovaglio J. Hanley P.W. Saturday G. Bosio C.M. Anzick S. Barbian K. Cihlar T. Martens C. Scott D.P. Munster V.J. de Wit E. Clinical benefit of remdesivir in rhesus macaques infected with SARS-CoV-2 Nature 585 2020 273 276 10.1038/s41586-020-2423-5 32516797 47 Xie X. Muruato A.E. Zhang X. Lokugamage K.G. Fontes-Garfias C.R. Zou J. Liu J. Ren P. Balakrishnan M. Cihlar T. Tseng C.K. Makino S. Menachery V.D. Bilello J.P. Shi P.Y. A nanoluciferase SARS-CoV-2 for rapid neutralization testing and screening of anti-infective drugs for COVID-19 Nat. Commun. 11 2020 5214 10.1038/s41467-020-19055-7 33060595 48 Malone B. Campbell E.A. Molnupiravir: coding for catastrophe Nat. Struct. Mol. Biol. 28 2021 706 708 10.1038/s41594-021-00657-8 34518697 49 Noreen S. Maqbool I. Madni A. Dexamethasone: therapeutic potential, risks, and future projection during COVID-19 pandemic Eur. J. Pharmacol. 894 2021 173854 10.1016/j.ejphar.2021.173854 50 Quimque M.T.J. Notarte K.I.R. Fernandez R.A.T. Mendoza M.A.O. Liman R.A.D. Lim J.A.K. Pilapil L.A.E. Ong J.K.H. Pastrana A.M. Khan A. Wei D.Q. Macabeo A.P.G. Virtual screening-driven drug discovery of SARS-CoV2 enzyme inhibitors targeting viral attachment, replication, post-translational modification and host immunity evasion infection mechanisms J. Biomol. Struct. Dyn. 39 2021 4316 4333 10.1080/07391102.2020.1776639 32476574 51 Jiang Y. Yin W. Xu H.E. RNA-dependent RNA polymerase: structure, mechanism, and drug discovery for COVID-19 Biochem. Biophys. Res. Commun. 538 2021 47 53 10.1016/j.bbrc.2020.08.116 32943188 52 Meier C. Nucleoside diphosphate and triphosphate prodrugs - an unsolvable task? Antivir. Chem. Chemother. 25 2017 69 82 10.1177/2040206617738656 29096525 53 Gollnest T. Dinis de Oliveira T. Rath A. Hauber I. Schols D. Balzarini J. Meier C. Membrane-permeable triphosphate prodrugs of nucleoside analogues Angew. Chem. Int. Ed. Engl. 55 2016 5255 5258 10.1002/anie.201511808 27008042 54 Sheahan Timothy P. Sims Amy C. Zhou S. Graham Rachel L. Pruijssers Andrea J. Agostini Maria L. Leist Sarah R. Schäfer A. Dinnon Kenneth H. Stevens Laura J. Chappell James D. Lu X. Hughes Tia M. George Amelia S. Hill Collin S. Montgomery Stephanie A. Brown Ariane J. Bluemling Gregory R. Natchus Michael G. Saindane M. Kolykhalov Alexander A. Painter G. Harcourt J. Tamin A. Thornburg Natalie J. Swanstrom R. Denison Mark R. Baric Ralph S. An orally bioavailable broad-spectrum antiviral inhibits SARS-CoV-2 in human airway epithelial cell cultures and multiple coronaviruses in mice Sci. Transl. Med. 12 2020 eabb5883 10.1126/scitranslmed.abb5883

0

PMC9749393

NO-CC CODE

2022-12-15 23:22:53

no

Int J Biol Macromol. 2023 Jan 31; 226:946-955

utf-8

Int J Biol Macromol

10.1016/j.ijbiomac.2022.12.112

oa_other

==== Front Econ Lett Econ Lett Economics Letters 0165-1765 0165-1765 Elsevier B.V. S0165-1765(22)00437-2 10.1016/j.econlet.2022.110963 110963 Article The impact of the Covid-19 job retention support on employment Meriküll Jaanika ab⁎ Paulus Alari a a Bank of Estonia, Estonia pst 13, 15095 Tallinn, Estonia b University of Tartu, Narva mnt 18, 51009 Tartu, Estonia ⁎ Corresponding author at: Bank of Estonia, Estonia pst 13, 15095 Tallinn, Estonia. 14 12 2022 14 12 2022 1109633 8 2022 9 12 2022 12 12 2022 © 2022 Elsevier B.V. All rights reserved. 2022 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The paper studies the selection into the widespread job retention support that was provided during the Covid-19 pandemic and the employment effects of it, using firm-level administrative data for Estonia that cover the whole population of firms in 2019–2020. The endogeneity of the support is addressed by creating a control group from firms that were as severely hit as those that received the support and by using matching techniques. It is found that there was no selection of firms into the support by productivity once the sample is restricted to firms that experienced similar adverse conditions. The support had a positive effect on employment, as about one job in five that was supported by the scheme was saved and the unemployment rate would have been 2–4 percentage points higher in 2020 without the support. Keywords Covid-19 Job retention scheme Employment ==== Body pmc1 Introduction Job retention schemes quickly became a popular policy tool for combating the adverse economic consequences of the Covid-19 pandemic. The German style of short-time work (Kurzarbeit) had proven an effective policy to counter unemployment during the Great Recession (Boeri and Bruecker, 2011, Cahuc et al., 2021, Kopp and Siegenthaler, 2021), and it inspired similar policies during the Covid-19 crisis. It has been estimated that job retention schemes were used in OECD countries ten times as much in the Covid-19 crisis as in the Great Recession (Scarpetta et al., 2020). However, there is still limited evidence about the employment effects of such support during the pandemic. The aim of this paper is to assess how job retention affected employment during the Covid-19 pandemic. We use rich firm-level administrative data for Estonia in 2019–2020 and matching techniques to identify how the job retention support impacted jobs in the first year of the Covid-19 pandemic. Wider economic shocks tend to hit Estonia, which is a small open economy, faster and more deeply than they hit large European countries, but it also tends to recover more quickly. Before the pandemic, GDP growth and inflation in Estonia were higher and unemployment was lower than the general levels in the EU. The dynamics of the Estonian economy during the pandemic have, however, been less drastic, and have been more similar to those in the EU than was the case in previous crises such as the Great Recession. Estonia also followed several other countries by introducing a job retention scheme and by pursuing sizable fiscal expansion overall. We take a similar approach to that of Kopp and Siegenthaler (2021), who contrasted the firms that applied for the support but did not receive it, with those that received the support. We create the control group from firms that did not take up the support but had a similar decline in sales to those that received the support. By constructing the counterfactual from a similar subgroup of severely hit firms, we expect to identify the causal effect better than we could by constructing the counterfactual from the full sample of firms. Our first contribution is that we extend the literature on the selection of firms to the Covid-19 job retention support. It has been shown that firm productivity is a crucial factor in how effective job retention schemes are. If low-productivity firms are more likely to receive the support, the policy is unlikely to have any long-term effect on jobs and will only postpone job destruction (Giupponi and Landais, 2022). There is inconclusive evidence about which firms were more likely to get the support during the pandemic; some studies show that the firms that did so were less productive (Harasztosi et al., 2022, Kozeniauskas et al., 2022, Morikawa, 2021), but the relationship has also been found to vary across countries (Bighelli et al., 2022). Our second contribution is that we estimate how many jobs were saved by the job retention support in 2020. There is evidence that job retention schemes were effective at containing unemployment during the Great Recession (Boeri and Bruecker, 2011, Cahuc et al., 2021, Giupponi and Landais, 2022, Kopp and Siegenthaler, 2021), but not much is known about how effective such policies were in the Covid-19 pandemic. To the best of our knowledge, Bennedsen et al. (2020) provide the only firm-level estimate of the impact of job support during the pandemic to have been published so far. We find that there was no selection of firms into the support by productivity once we restrict the sample to firms that experienced similar adverse conditions, and that the support reduced lay-offs. We estimate that about one job in five that was supported by the scheme in our private sector sample was saved, and the unemployment rate would otherwise have been 2–4 percentage points higher in 2020. 2 Data and methodology We use Estonian administrative data from the Business Register and the Tax and Customs Board from 2019–2020, and we combine these with the data on the Covid-19 support measures in 2020 from the institutions that administrated those measures. Our sample consists of the whole population of non-financial private sector firms. The job retention scheme accounted for roughly half of all the Covid-19 support in our sample country, and 21% of workers overall received support from the job retention scheme in 2020. We also control for the other forms of Covid-19 support received in 2020, the majority of which was liquidity aid to firms from the state-owned financial intermediation institution Kredex, while a minor share was provided as direct non-refundable aid by Enterprise Estonia. For further details about the data, see Meriküll and Paulus (2022). To be eligible for the support, firms had to meet two of three criteria, which were a decline of at least 30% in turnover, reduced working hours for at least 30% of workers, or a decline of at least 30% in wages for at least 30% of workers, all in comparison to the same month a year earlier. The size of the benefit was initially 70% of the average monthly wage of the employee, capped at 1000 euros, and it was paid out for up to two months from March to May 2020. It was then extended for one month in June 2020 with a tighter turnover criterion of a decline of 50%, and a lower maximum amount of 800 euros. There was also a requirement for employers to contribute as well, as they additionally had to pay each worker a gross salary of at least 150 euros per month. The inclusiveness and generosity of the scheme in our sample country were similar to the average levels in other countries (Müller and Schulten, 2020). The actual eligibility of each firm is not observed, and so we simulate potential eligibility instead by considering firms to be eligible if their turnover dropped by more than 30% in the first or second quarter of 2020 from what it was in the same quarter a year before. These quarterly data are available from the Tax and Customs Board register. After restricting the sample to the potentially eligible firms, we perform a matching exercise to derive the average treatment effect on the treated, using propensity score matching with the three nearest neighbours within the 1 percentage point caliper. There are still some differences between the treatment group and the control group after the sample has been restricted, but the remaining differences decline substantially after the matching; see Meriküll and Paulus (2022), Appendix B. 3 Results We first study how the probability of receiving the job retention support depends on firm productivity prior to the pandemic, using firm-level total factor productivity (TFP) that we estimate using the method of Levinsohn and Petrin (2003). We find that low-productivity firms were more likely to get the support unconditionally, see Appendix B of Meriküll and Paulus (2022), and also after firm size and sector are controlled for, see column 1 of Table 1. The probability of a firm with TFP one unit higher, which corresponds approximately to an increase of one standard deviation in TFP, receiving the support was 0.6 percentage point lower in the whole sample.1 However, when we leave non-eligible firms aside and consider only the firms that were potentially eligible for the support or that received it, the probability of them receiving the support no longer depends on productivity; see column 2 of Table 1. The three policy measures are also tightly related; receiving the liquidity support and direct subsidies increases the probability of receiving the job retention support. In the next step, we estimate the effect of the support on firm employment growth using the matching technique, where the propensity score is obtained from the probit model that is reported in column 2 of Table 1. The remaining difference between the employment growth rates in the treatment group and the control group after the matching is 9.6 percentage points, which is statistically significant and can be attributed to this policy; see Table 2. In other words, job losses in the treatment group would have been 2.2 times larger without the support, as employment growth would have been −0.173 instead of −0.077; the 90% confidence bounds for the estimate range from 1.8 to 2.6.Table 1 Probit model for receiving the support, 2020. Dependent variable: 1 = obtained support in 2020, 0 = did not obtain support in 2020 Whole sample (1) Subgroup of firms that are eligible or that received the support (2) Relative TFP (2019) −0.006∗∗ −0.002 (0.003) (0.004) Regional unemployment growth (2020) 0.160∗∗∗ 0.089∗∗ (0.027) (0.042) Log(employment) (2019) 0.080∗∗∗ 0.172∗∗∗ (0.002) (0.005) Received liquidity support (2020) 0.297∗∗∗ 0.274∗∗∗ (0.040) (0.073) Received direct subsidies (2020) 0.440∗∗∗ 0.409∗∗∗ (0.013) (0.018) Industry FE yes yes N 34,626 18,803 Pseudo R2 0.165 0.218 Note: The table reports marginal effects at mean from the probit model. Authors’ calculations using administrative data. By applying this estimate to the actual employment growth, as shown in column 2 of Table 3, we can finally derive counterfactual employment growth for the treated firms; this is shown in column 3 of Table 3 . This suggests that total employment in the private sector, which was 388,000 jobs in our sample, would have declined by an additional 5.1 percentage points without the policy, meaning roughly 20,000 jobs were saved by the job retention scheme. The 90% confidence bounds of this estimate are 14,000–26,000 jobs. Given that the number of jobs that were supported by the scheme in our private sector sample was 113,000, 17.7% of jobs receiving support were saved. The unemployment rate would have increased from 6.9% to 9.8% in 2020 without the support, and by 2–4 percentage points at the 90% confidence level.2 Table 2 Employment growth rates in the treatment and control groups, 2020. Treated Control Difference Standard error Before matching −0.077 –0.093 0.016∗∗∗ 0.005 After matching −0.077 –0.173 0.096∗∗∗ 0.010 Authors’ calculations using administrative data. Estimates obtained with meticulous identification strategies range from no long-term effect on employment (Giupponi and Landais, 2022) to between one job in six (Cahuc et al., 2021) and one in three (Kopp and Siegenthaler, 2021) saved among all the participants. The effects of the Covid-19 support that were estimated on Danish data found that approximately one job in three was saved at the firms that received some form of support, and half of these can be attributed to the job retention support (Bennedsen et al., 2020). Our estimates are close to those from these studies.Table 3 The effects of the job retention scheme on employment. Group Employment share in 2019–2020 (1) Actual employment growth in 2020 (2) Counterfactual employment growth in 2020 (3) Difference from actual employment growth in 2020 (4) = (3)−(2) All firms 1.000 −0.069 –0.121 –0.051 No eligibility & no receipt 0.501 −0.009 n/a n/a Treated 0.345 −0.119 −0.268 −0.149 Control 0.153 −0.156 n/a n/a Notes: The counterfactual growth rate for the treatment group is derived by multiplying the policy effect from Table 2, which is 2.2, by the observed growth in 2019–2020. It is assumed that non-eligible firms and the control group, which did not participate in the scheme, were unaffected by the policy. Authors’ calculations using administrative data. A back-of-the-envelope calculation suggests that the cost of the job retention scheme in gross terms, which was 211 million euros in our sample, was only slightly higher for the government than the fiscal burden of an additional 20,000 people unemployed would have been in 2020 at about 196 million euros. This cost comes from 95 million euros of foregone revenues from income tax and social security contributions, and extra spending of 101 million euros on unemployment insurance and redundancy benefits. We therefore find, similar to Kopp and Siegenthaler (2021), that the scheme largely paid for itself. The public cost net of additional tax revenues shows unambiguously cost efficiency at 106 million euros rather than 132 million. The scheme was also highly beneficial for firms as their direct costs were 46 million euros because of the requirement to contribute, which was much less than the amount they would otherwise have paid for salaries during the mandatory notice period and redundancy compensation of about 81 million euros. For further details, see the online appendix. 4 Conclusions This paper evaluates the selection into the job retention scheme introduced in Estonia during the Covid-19 crisis and the effects of the scheme. Our results suggest that the support was an effective tool for protecting jobs during the pandemic. We find no adverse selection of low-productivity firms into the support and show that the scheme helped to reduce job destruction during the pandemic. Appendix A Supplementary data The following is the Supplementary material related to this article. MMC S1 Data availability The authors do not have permission to share data. Acknowledgments We thank the editor and an anonymous referee for their useful comments. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The views expressed are those of the authors and do not necessarily represent the official views of Eesti Pank or the Eurosystem. 1 TFP is measured as the difference of firm logarithmic TFP from its NACE 2-digit industry average. The average TFP is close to zero by definition and its standard deviation equaled 0.933 in 2019. 2 Adding the point estimate gives an additional 20,000 workers laid off on top of the 48,400 who were unemployed in 2020, according to Statistics Estonia’s online database (Table TT0151). The estimate would be somewhat higher when extended to workers outside our sample, though presumably not much as these are primarily public sector employees.. Appendix A Supplementary material related to this article can be found online at https://doi.org/10.1016/j.econlet.2022.110963. ==== Refs References Bennedsen M. Larsen B. Schmutte I. Scur D. Preserving job matches during the COVID-19 pandemic: Firm-level evidence on the role of government aid Covid Econ.: Vetted Real-Time Pap. 2020 1 30 Bighelli T. Lalinsky T. Vanhala J. Covid-19 Pandemic, State Aid and Firm Productivity: Discussion Paper No. 1/2022 2022 Bank of Finland Boeri T. Bruecker H. Short-time work benefits revisited: some lessons from the great recession Econ. Policy 26 2011 697 765 10.1111/j.1468-0327.2011.271.x Cahuc P. Kramarz F. Nevoux S. The Heterogeneous Impact of Short-Time Work: From Saved Jobs to Windfall Effects: Discussion Paper No. 16168 2021 CEPR Giupponi G. Landais C. Subsidizing labor hoarding in recessions: The employment & welfare effects of short time work Rev. Econom. Stud. 2022 forthcoming Harasztosi P. Maurin L. Pál R. Revoltella D. van der Wielen W. Firm-level policy support during the crisis: So far, so good? Int. Econ. 171 2022 30 48 10.1016/j.inteco.2022.04.003 Kopp D. Siegenthaler M. Short-time work and unemployment in and after the great recession J. Eur. Econom. Assoc. 19 2021 2283 2321 10.1093/jeea/jvab003 Kozeniauskas N. Moreira P. Santos C. On the cleansing effect of recessions and government policy: Evidence from Covid-19 Eur. Econ. Rev. 2022 144 10.1016/j.euroecorev.2022.104097 Levinsohn J. Petrin A. Estimating production functions using inputs to control for unobservables Rev. Econom. Stud. 70 2003 317 341 Meriküll J. Paulus A. Were Jobs Saved at the Cost of Productivity in the Covid-19 Crisis?: Working Paper No. 5/2022 2022 Bank of Estonia Morikawa M. Productivity of firms using relief policies during the COVID-19 crisis Econom. Lett. 2021 203 10.1016/j.econlet.2021.109869 Müller T. Schulten T. Ensuring Fair Short-Time Work - a European Overview: ETUI Policy Brief No. 7/2020 2020 European Trade Union Institute Scarpetta S. Pearson M. Hijzen A. Salvatori A. Job Retention Schemes During the COVID-19 Lockdown and beyond 2020 OECD

0

PMC9749394

NO-CC CODE

2022-12-15 23:22:54

no

Econ Lett. 2022 Dec 14;:110963

utf-8

Econ Lett

10.1016/j.econlet.2022.110963

oa_other

==== Front Chem Eng J Chem Eng J Chemical Engineering Journal 1385-8947 1385-8947 Elsevier B.V. S1385-8947(22)06443-9 10.1016/j.cej.2022.140963 140963 Article Functionalized biological metal–organic framework with nanosized coronal structure and hierarchical wrapping pattern for enhanced targeting therapy Wang Huafeng a Li Shi b Wang Lei a Liao Zimei a Zhang Hang a Wei Tianxiang b⁎ Dai Zhihui ac⁎ a School of Chemistry and Materials Science, Nanjing Normal University, Nanjing 210023, China b School of Environment, Nanjing Normal University, Nanjing 210023, China c School of Chemistry and Molecular Engineering, Nanjing Tech University, Nanjing 211816, China ⁎ Corresponding authors. 14 12 2022 14 12 2022 14096313 10 2022 25 11 2022 12 12 2022 © 2022 Elsevier B.V. All rights reserved. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Inefficient tumor-targeted delivery and uncontrolled drug release are the major obstacles in cancer chemotherapy. Herein, inspired by the targeting advantage of coronavirus from its size and coronal structure, a coronal biological metal–organic framework nanovehicle (named as corona-BioMOF) is constructed for improving its precise cancer targeting ability. The designed corona-BioMOF is constructed as the carriers-encapsulated carrier model by inner coated with abundant protein-nanocaged doxorubicin particles and external decorated with high-affinity apoferritin proteins to form the spiky surface for constructing the specific coronal structure. The corona-BioMOF shows a higher affinity and enhanced targeting ability towards receptor-positive cancer cells compared to that of MOF-drug composites without spiky surface. It also exhibits the hierarchical wrapping pattern-endowed controlled lysosome-specific drug release and remarkable tumor lethality in vivo. Moreover, water-induced surface defect-based protein handle mechanism is first proposed to shape the coronal-BioMOF. This work will provide better inspiration for nanovehicle construction and be broadly useful for clinical precision nanomedicine. Keywords Metal-organic framework Coronal structure Precise targeting Drug delivery Cancer therapy ==== Body pmc1 Introduction Nanoscale and nanostructure-based therapeutic agents have been playing important roles in cancer therapy [1]. Recently, various nanoagents have been designed as drug delivery platforms for cancer chemotherapy [2], [3], [4], [5]. However, engineering an advanced delivery system capable of both precise targeting and controlled drug release to prevent premature drug leakage and obtain better treatment outcomes remains a tremendous challenge. It has been reported that the coronal structure of virus is beneficial to enhance the targeting ability with host cell [6]. For instance, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) shows high infectivity toward its host cells [7], [8], [9]. The spike proteins on the CoV surface significantly increase the adhesion and infection abilities partly due to the fact that the spiky shape is beneficial to the interaction between species [9], [10]. Recent studies have also confirmed that spiky surfaces enhanced the interaction of nanomaterials with bacteria and cancer cells [11], [12], [13]. Moreover, it has been reported the diameter of the CoV ranges between 50 nm to 140 nm, and the size affects its encounter efficiency with host cells and subsequent infection ability [14], [15]. The large nanocarriers (50–200 nm) can spontaneously accumulate around leaky regions of the tumor vasculature via the “enhanced permeability and retention” (EPR) effect, and the size-related advantages on enhancing targeting ability have emerging in drug delivery and cancer therapy [16]. Therefore, inspired from the CoV, engineering a nanovehicle with an appropriate size and a coronal structrue might be a workable approach to obtain enhanced targeting ability towards cancer cells. Apart from targeting, controlled and efficient drug release is another key point in precise cancer treatment. Many artificial stimuli-responsive nanocarriers have been applied for controlled drug release through the following stimulation ways: light, pH, hypoxia, and multiple stimulations [17], [18], [19], [20], [21], [22]. Thereinto, metal–organic frameworks (MOFs), have been widely applied in the controlled delivery of various drugs, owing to their distinct stimulus responsiveness [23], [24], [25], [26], [27], [28], [29]. For example, zeolitic imidazolate framework-8 (ZIF-8) showed the properties of acid-induced disintegration; Zr-MOF showed high phosphate concentration induced collapse [30], [31], [32], [33]. In recent years, MOFs have been found to show protective effect toward encapsulated biomolecules [34], [35], [36], [37], [38]. Also, the encapsulated biomolecules might increase the functionality of MOFs. For example, some biomolecules, such as silk fibroin and ferritin, have distinct pH-sensitive hydrolytic degradation property different from MOFs [39], [40]. The composite of MOFs and biomolecules might produce a new pH-responsive performance. By considering the pH difference in tumor microenvironment (pH≈6.5) and subcellular organelles (pH≈5), we wonder whether a new biological MOF nanovehicle could be constructed to display a hierarchical pH-responsive property for achieving precise controlled drug release intracellularly. In this work, we designed a coronal biological MOF nanovehicle (named as corona-BioMOF) to achieve both precise targeting and controlled drug release in cancer therapy (Scheme 1 ). Herein, ZIF-8 was introduced as the host material in corona-BioMOF, serving as the nanocarriers and tumor microenvironment (weak acid)-responsive drug nanocontrol. Then, apoferritin (AFt) was chosen as one of the important constituents for constructing corona-BioMOF due to the following three aspects: (1) It can specifically bind to the transferrin receptor 1 (TfR1), a cell membrane receptor up-regulated in malignant proliferating cells like MDA-MB-231 triple-negative breast cancer (TNBC) cells [41]; (2) It can serve as another nanocarrier to load small molecular chemotherapy drugs due to its structure of protein “cage”[42] (Scheme 1A) and can biomineralize MOF simultaneously (Scheme 1B), thus constructing the carriers-encapsulated carrier model in corona-BioMOF and decorating the surface of MOF to form spiky shape; (3) It can be disassembled into subunits at low pH and reassembled at near-neutral pH [43], making the deep design in controlled loading and release of drugs to become possible. Another constituent, glucose oxidase (GOx), was utilized to assist the biomineralization synthesis and famish cancer cells. By the following two-stage process, corona-BioMOF has shown its advantages in targeting ability and hierarchical pH-responsive controlled drug release: First, doxorubicin (DOX)-loaded AFt proteins (DOX@AFt) on the surface of corona-BioMOF endow the nanovehicle strong target activity. Second, weak-acidic (pH≈6.5), as well as the GOx-catalyzed gluconic acid production, will accelerate the degradation of ZIF-8, thus corona-BioMOF can slowly disintegrate accompanied by the exposure of more inclusions (DOX@AFt and GOx) and produce more binding sites towards cancer cells. After interaction, the decrease of pH during the cellular endocytosis drove the further disintegration of corona-BioMOF and the continued release of encapsulated DOX@AFt. Finally, explosive DOX release from DOX@AFt was achieved with the disassembly of AFt at low pH of lysosomes (pH≈5). Thus, the lysosome-specific DOX delivery was accomplished with commendably preventing premature drug leakage, and corona-BioMOF displayed the outstanding antitumor activity. We highly expect that the corona-BioMOF nanovehicle is of benefit to provide inspiration for the development of novel nanoagents in precise medicine.Scheme 1 Schematic illustration of the design and assembly process of the proposed corona-BioMOF nanovehicle and its programmed therapy against breast cancer. (A) Assembly process of DOX@AFt. (B) Preparation steps of the corona-BioMOF nanovehicle. (C) Therapy process of corona-BioMOF in vivo. 2 Materials and methods 2.1 Materials and reagents AFt from equine spleen (0.2 μm filtered), 2-methylimidazole (2-MIM), zinc nitrate hexahydrate, thiazolyl blue tetrazolium bromide (MTT), and GOx were purchased from Sigma-Aldrich. DOX was obtained from Tokyo Chemical Industry Company. MDA-MB-231 and MCF-10A cells were obtained from Chinese National Collection of Authenticated Cell Cultures. MX-1 cells were kindly sponsored by Professor Yuqing Chen. LysoSensor Blue DND-167 was obtained from Thermo Fisher Scientific Incorporated. TfR1 (CD71) mouse monoclonal antibody and CoraLite488-conjugated goat anti-mouse lgG(H+L) were purchased from Proteintech Incorporated. Goat Anti-Mouse lgG H&L (Alexa FluorR 555) preadsorbed was purchased from Abcam. Blocking buffer, antibody dilution buffer for immunofluorescence, antifade mounting medium with DAPI, and PBS were purchased from Beyotime Biotechnology Incorporated. Fetal bovine serum (FBS), Dulbecco’s modified eagle medium (DMEM), RPMI 1640 medium, penicillin/streptomycin, and 0.25% trypsin-EDTA solution were purchased from Gibco. BCA protein assay kit was purchased from CoWin Biotech Co., Ltd. Amicon Ultra 3K filters were purchased from Merck Millipore Company. All the chemicals were of analytical grade and used without further purification. 2.2 Measurements and characterizations TEM images were carried out on JEM-2100 at an accelerating voltage of 200 kV (JEOL, Japan). HAADF-STEM and ColorSTEM were measured with Apreo 2S (Thermo Scientific Apreo 2S, Czech). XRD patterns were recorded on a model D/max-RC X-ray diffractometer (Ragaku, Japan). N2 adsorption−desorption isotherms were performed with ASAP-2050 automated sorption analyzer (Micromeritics, USA). DRIFTS studies were conducted by a VERTEX 70 FT-IR spectrometer (Bruker Ltd., Germany) equipped with a DRIFTS accessory. Ultraviolet-visible absorption spectroscopy was obtained with a Varian Cary 60 spectrophotometer (Agilent, USA). MTT assay was determined with a microplate reader (Thermo Scientific Multiskan GO, USA) at 490 nm. CLSM images were performed on a Nikon A1 confocal microscope (Nikon, Japan). Fluorescence imaging of animals was taken by IVIS spectrum (PerkinElmer, USA). The content of Zn elements in tumors, organs, and blood were examined by ICP-AES (Leeman Labs, USA). 2.3 Synthesis of ZIF-8 Typically, 2-MIM (17.5 mmol) was dissolved in 3.0 mL of methanol at 30 °C. Then, zinc nitrate hexahydrate (0.25 mmol, 0.5 mL) was mixed and stirred for 30 min. The white product was collected by centrifuging at 3500 rpm for 20 min and washed twice. 2.4 Synthesis of DOX@AFt In detail, 2.5 mL of 0.1 mM DOX and 1.0 mL of 1.0 mg mL−1 AFt were mixed and then agitated for 5 min. Then, 1.0 M hydrochloric acid was added to decrease the pH of solution to 2.5, which could disassemble the structure of AFt. The solution was stirred for 15 min. Then the pH of solution was adjusted to 6.5 with 1.0 M sodium hydroxide. The mixture was continuously stirred for 15 min to entrap DOX inside AFt cavity. Water exchange was performed several times through Amicon Ultra 3K filters to remove the non-encapsulated DOX. Finally, DOX@AFt was collected and stored at 4 °C. To investigate the encapsulation efficiency of DOX in DOX@AFt protein cage, the filtrates of reaction mixture in different pH, assembly time, or concentrations of DOX were measured by fluorescence spectrum. 2.5 Synthesis of GOx@ZIF-8, DOX@AFt@ZIF-8, and (DOX@AFt+GOx)@ZIF-8 According to the previous reports with slight changes [34], [37], GOx@ZIF-8, DOX@AFt@ZIF-8, and (DOX@AFt+GOx)@ZIF-8 were prepared with the following procedures. For the synthesis of GOx@ZIF-8, GOx (1.5 mg) was first dispersed in an aqueous solution of 2-MIM (35 mmol, 6.3 mL) and stirred for 10 min at 30 °C. Afterward, 1.0 mL of 0.5 mmol zinc nitrate hexahydrate aqueous solution was added in the above mixture and stirred for another 30 min at 30 °C. The production was collected by centrifuging at 3500 rpm for 20 min, washed twice, and freeze-dried. For the synthesis of DOX@AFt@ZIF-8, DOX@AFt (7.5 mg) was dispersed in an aqueous solution of 2-MIM (35 mmol, 6.3 mL) and stirred for 10 min at 30 °C. Afterward, 1.0 mL of 0.5 mmol zinc nitrate hexahydrate aqueous solution was added in the above mixture and stirred for another 30 min at 30 °C. The production was collected by centrifuging at 3500 rpm for 20 min, washed twice, and freeze-dried. For the synthesis of (DOX@AFt+GOx)@ZIF-8, GOx (1.5 mg) and DOX@AFt (7.5 mg) were dispersed in an aqueous solution of 2-MIM (35 mmol, 6.3 mL) and stirred for 10 min at 30 °C. Afterward, 1.0 mL of 0.5 mmol zinc nitrate hexahydrate aqueous solution was added in the above mixture and stirred for another 30 min at 30 °C. The production was collected by centrifuging at 3500 rpm for 20 min, washed twice, and freeze-dried. 2.6 Synthesis of BioMOF Firstly, GOx (1.5 mg) and DOX@AFt (1.5 mg) were dispersed in an aqueous solution of 2-MIM (35 mmol, 5.8 mL) with stirring for 10 min at 30 °C. Next, an aqueous solution of zinc nitrate hexahydrate (0.5 mmol, 1.0 mL) was added and the mixture turned milky promptly. Then it was stirred for 30 min at 30 °C. The mixture was precipitated by centrifugation at 3500 rpm for 20 min at 4 °C, washed and resuspended in water to obtain BioMOF without coronal surface. 2.7 Synthesis of corona-BioMOF Firstly, GOx (1.5 mg) and DOX@AFt (1.5 mg) were dispersed in an aqueous solution of 2-MIM (35 mmol, 5.8 mL) with stirring for 10 min at 30 °C. Next, an aqueous solution of zinc nitrate hexahydrate (0.5 mmol, 1.0 mL) was added and the mixture turned milky promptly. Then it was stirred for 30 min at 30 °C. The mixture was precipitated by centrifugation at 3500 rpm for 20 min at 4 °C, washed, and resuspended in water to obtain BioMOF, an intermediate product without coronal surface. Immediately, additional DOX@AFt (6.0 mg) was added into the above aqueous solution and stirred for another 30 min at 30 °C. The corona-BioMOF was collected by centrifuging at 3500 rpm for 20 min at 4 °C and freeze-dried. The prepared pink corona-BioMOF was stored at −20 °C for future use. 2.8 Synthesis of BioMOF composites-1 Zinc nitrate hexahydrate (0.15 mmol, 0.3 mL) was added into the solution of BioMOF without coronal surface and stirred for 15 min. Immediately, additional DOX@AFt (6.0 mg) was added into the above aqueous solution and stirred for another 30 min at 30 °C. The BioMOF composites-1 was collected by centrifuging at 3500 rpm for 20 min at 4 °C. 2.9 Synthesis of BioMOF composites-2 2-MIM (1.5 mmol, 0.3 mL) was added into the solution of BioMOF without coronal surface and stirred for 15 min. Immediately, additional DOX@AFt (6.0 mg) was added into the above aqueous solution and stirred for another 30 min at 30 °C. The BioMOF composites-2 was collected by centrifuging at 3500 rpm for 20 min at 4 °C. 2.10 Simulated performance assessment of drug release kinetics To assess the drug release kinetics of DOX from corona-BioMOF in different microenvironment, corona-BioMOF was dispersed in 0.9% NaCl at pH 7.4, 6.5, and 5.0 to simulate the pH values of normal physiological environment, tumor microenvironment, and lysosome, respectively. After incubation for a given time at 37 °C, the mixture was centrifuged at 10000 rpm for 10 min. The fluorescence intensities of the supernatants were measured by fluorescence spectrophotometry at 595 nm with an excitation wavelength of 484 nm. 2.11 Determination of the loading capacity of DOX The loading capacity of DOX in DOX@AFt was determined upon the centrifuged supernatant of reaction mixture. The fluorescence intensity at 595 nm of DOX was measured by fluorescence spectrophotometry. The loading capacity of DOX was calculated as (fluorescence intensity of total added DOX − fluorescence intensity of supernatant DOX) / fluorescence intensity of total added DOX × 100%. 2.12 Cell culture Breast MCF-10A cells and breast adenocarcinoma MX-1 cells were cultured in DMEM medium containing 10% FBS and 1% penicillin-streptomycin. Breast adenocarcinoma MDA-MB-231 cells were cultured in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin-streptomycin. All cell lines were incubated at 37 °C containing 5% CO2 in humidified atmosphere. MCF-10A, MX-1, or MDA-MB-231 cells were digested with trypsin and resuspended in fresh cell culture medium before planting. 2.13 In vitro evaluation of targeting behavior of corona-BioMOF nanovehicle Characterization of TfR1 in MDA-MB-231 was performed by immunofluorescence analysis. Cells were fixed with ice ethanol for 15 min, and then blocked with blocking buffer for 10 min. Primary antibodies were incubated for 1.5 h at RT. The following secondary antibodies were incubated for 1.5 h at RT in the dark. The nuclei were counterstained by incubating with DAPI for 10 min. Targeting behavior of the corona-BioMOF nanovehicle was investigated as follows: MDA-MB-231 or MX-1 cells were seeded on a 35 mm Petri dish with a 10 mm well at the density of 5 × 105 per well. The cells were incubated for 24 h in a water-jacket humidified incubator under 37 °C and 5% CO2 condition. Afterward, 1.0 mL of corona-BioMOF dispersed in culture medium (100 μg mL−1) was added and incubated for 0.5, 2, 6, or 12 h. Subsequently, the cells were washed with PBS and stained with LysoSensor Blue probe for cell imaging. For flow cytometry analysis, MCF-10A, MDA-MB-231, and MX-1 cells were seeded on a 6-well plate at the density of 1 × 106 per well. The cells were incubated for 24 h in a water-jacket humidified incubator under 37 °C and 5% CO2 condition. Afterward, 1.0 mL of corona-BioMOF dispersed in culture medium (100 μg mL−1) was added and incubated for 2 h. Subsequently, the cells were washed with PBS and detached for flow cytometry analysis. To confirm that corona-BioMOF binding to TfR1 of MDA-MB-231 cancer cells plays a role in killing cancer cells, an antibody blocking study was carried out by incubating the MDA-MB-231 cells with TfR1 antibody for 2 h followed by treatment with corona-BioMOF for 0.5, 2, and 6 h. Then CLSM imaging and flow cytometry analysis was done in the same way as described above. 2.14 In vitro cytotoxicity study MCF-10A, MDA-MB-231, and MX-1 cells were planted on 96-well plates at the density of 1 × 104 cells, and were allowed to adhere for 24 h. Next, the cells were incubated with culture medium containing ZIF-8, GOx@ZIF-8, DOX, DOX@AFt@ZIF-8, DOX@AFt, or corona-BioMOF for 12 h. To determine cytotoxicity, 20 μL of 5.0 mg mL−1 MTT was added into each well and incubated sequentially for 4 h. After that, the supernatant was replaced by 150 μL of DMSO. Absorbance values of formazan were measured through microplate reader at 490 nm. 2.15 Antitumor experiment in vivo All animal experiments were conducted in accordance with protocol No. SYXK (Su) 2020-0047 approved by Jiangsu Provincial Department of Science and Technology, and carried out in accordance with the institutional animal use and care regulations approved by the Animal Ethical and Welfare Committee of Nanjing Normal University (Approval Number: IACUC-20200601). Female BALB/c mice models were established by subcutaneously inoculating 9 × 106 MDA-MB-231 cells in 100 μL Matrigel/PBS. When the tumor reached approximately 90 mm3, the MDA-MB-231 tumor-bearing mice were divided into eight groups (n = 5 mice for each group) at randomly and intravenously injected with PBS, ZIF-8, GOx@ZIF-8, DOX, DOX@AFt, DOX@AFt@ZIF-8, or corona-BioMOF (2.5 mg mL−1, 200 μL) every 2 days with a total of five injections per mouse. The medium of injection was PBS. During the therapy, mice body weights and tumor volumes were recorded every other day. The tumor volumes were quantified by formula Length × Width2 / 2. All tumor samples were weighted before fixation. H&E staining of major organs and tumors, immunofluorescence, and TUNEL staining of tumors were performed by the third-party company. 2.16 Statistical analysis All data are expressed as means ± SD. Statistical differences were determined by two-tailed Student’s t test; **P < 0.01, ***P < 0.001, and ****P < 0.0001. 3 Results and discussion 3.1 Synthesis and characterization of the corona-BioMOF nanovehicle The ultraviolet-visible adsorption peaks of DOX@AFt indicated that DOX had been embedded inside the AFt (Fig. S1). Besides, DOX@AFt and corona-BioMOF remained close to DOX in the fluorescence emission spectrum at 595 nm (Fig. S2, Fig. 1 A), implying the encapsulation of DOX in AFt and successful construction of corona-BioMOF. As the assembly of DOX@AFt mainly relied on the pH-dependent disassembly and reassembly of AFt, the pH conditions of AFt reassembly for efficient loading of DOX into DOX@AFt were investigated (Fig. S3). Finally, pH 6.5, 0.10 mM of DOX, and 15 min of the assembly time were selected as the optimized conditions for preparation of DOX@AFt (Fig. S4 and S5). A maximum percentage of drugs (78%) loaded into AFt proteins (Fig. 1B) was obtained, which might attributed to the electrostatic gradient and dynamics of pore residues [44].Fig. 1 Characterization, drug loading yields, and drug release ability of corona-BioMOF. (A) Fluorescence spectra of ZIF-8, DOX@AFt, and corona-BioMOF. (B) The loading yields of DOX in DOX@AFt at different pH values in synthesis process of DOX@AFt. The data are presented as the mean ± SD, n=3. (C) XRD patterns of simulated ZIF-8, GOx@ZIF-8, DOX@AFt@ZIF-8, and corona-BioMOF. TEM images of (D) DOX@AFt and (E) corona-BioMOF. (F) High-magnification TEM image of corona-BioMOF. (G) HAADF-STEM image and (H) Color STEM image of corona-BioMOF. (I) Time-dependent dynamic releasing profiles of DOX in corona-BioMOF at different conditions. The error bars represented the standard deviation of three measurements. The data are presented as the mean ± SD, n=3. The crystal structures of the prepared ZIF-8-based nanomaterials were investigated by the powder X-ray diffraction (XRD) measurements (Fig. 1C), indicating that they retained almost the same crystal structure as the simulated ZIF-8 and the incorporation of GOx and DOX@AFt had little influence on the crystalline form of ZIF-8. Meanwhile, the XRD peaks shifted slightly to the low angle, reflecting the larger lattice constant, which is likely attributed to defects in composite [45]. Besides, the coronal morphology and element distribution of the corona-BioMOF were investigated by transmission electron microscopy (TEM), high angle angular dark field-scanning transmission electron microscopy scanning electron microscopy (HAADF-STEM), and ColorSTEM. As shown in Fig. 1D, DOX@AFt exhibited a spherical morphology with approximately 8 nm in diameter. The corona-BioMOF was uniformly coated with a large number of black spheres protruding from the surface, featuring a distinctive coronal structure (Fig. 1E). Magnified view in Fig. 1F showed more clearly that the coronal morphology and size of corona-BioMOF. HAADF-STEM image and corresponding ColorSTEM image (Fig. 1G and H) demonstrated that the outside small spheres were rich in N element and lacking in Zn element, confirming that the coronal periphery was indeed AFt proteins rather than small-sized MOFs. The above results demonstrated the satisfactory fabrication of nanomaterials with a specific coronal morphology. Furthermore, the Brunauer–Emmett–Teller (BET) surface area of corona-BioMOF (266.4 m2 g−1) was much smaller than ZIF-8 (1099 m2 g−1), DOX@AFt@ZIF-8 (911.1 m2 g−1), and GOx@ZIF-8 (423.8 m2 g−1), confirming the encapsulation of GOx and DOX@AFt in corona-BioMOF (Fig. S6). The loading capacity of total proteins in the corona-BioMOF was determined to be 57.6 μg mg−1 from the BCA protein assay kit (Fig. S7), where the amounts of GOx and AFt were 12.1 and 45.5 μg mg−1, respectively. This loading capacity was superior to previous most reports, which might attribute to the fully utilization of both the internal cavity and external surface of MOF [34], [46]. Besides, for assessing the pH controllability of corona-BioMOF towards DOX release, pH 7.4, 6.5, and 5.0 were chosen to simulate the pH values of normal physiological environment, tumor microenvironment, and lysosome, respectively. The time-dependent dynamic releasing profiles of DOX in corona-BioMOF (Fig. 1I) showed that little drug released upon exposure in pH 7.4 condition within 5 h. While at pH 5.0, a burst release of DOX within 1 h was found, implying a significant pH-responsive property of corona-BioMOF attributed to the combination of the degradation ability of the ZIF-8 structure in weak acid and the disassembly performance of the AFt protein at low pH. Ultimately a maximum DOX release amount (59.8%) was achieved at pH 5.0 with glucose, which was attributed to the fact that GOx could catalase the oxidation of glucose into gluconic acid, which raised the acidity, further accelerating the degradation of ZIF-8, the disassembly of AFt and the drug release. Different from the spiky surface of corona-BioMOF, the TEM images of ZIF-8 (Fig. 2 A), GOx@ZIF-8 (Fig. 2B), and BioMOF (Fig. 2C), the intermediate without spiky surface, showed smooth surfaces without black spheres. Moreover, we found that partial AFt proteins (indicated by arrows and dashed ovals) inside BioMOF began surfacing in pH 6.5 with glucose (Fig. 2D), which confirmed the inclusion of DOX@AFt and acid-induced disintegration of ZIF-8. Furthermore, the statistical comparisons of the amounts of spiky spheres on the surface of corona-BioMOF (Fig. 2E and F) and corona-BioMOF treated with weak acid medium containing glucose (Fig. 2G and H) were performed. A 45.1% increase of exposed DOX@AFt proteins was found in the latter, implying the capability of tumor microenvironment-boosted more favorable targeting behavior of corona-BioMOF.Fig. 2 Performance assessment in simulated tumor microenvironment and possible formation mechanism of corona-BioMOF. TEM images of (A) ZIF-8, (B) GOx@ZIF-8 NPs, (C) BioMOF, (D) BioMOF in pH 6.5 containing glucose, and (E) corona-BioMOF. (F) Corresponding quantitative statistics of DOX@AFt on the surface of corona-BioMOF in Fig. 2E. (G) TEM image of corona-BioMOF in pH 6.5 containing glucose. (H) Corresponding quantitative statistics of DOX@AFt on the surface of materials in Fig. 2F. (I) DRIFTS of ZIF-8 synthesized in methanol, BioMOF, and corona-BioMOF. (J) Water-induced defect-based handle mechanistic illustration for the binding of DOX@AFt on the surface of BioMOF. To elucidate the formation mechanism of the special topography of corona-BioMOF, additional metal ions or ligand molecules were introduced into the BioMOF without coronal surface. We noticed that the later added DOX@AFt was difficult to grow onto the above BioMOF composite surface with the additional introduction of Zn(Ⅱ) or 2-MIM. The additional introduction of zinc ions resulted in visible voids on the surface of the material (Fig. S8), implying that large-scale defects are not conducive to the binding between composites and proteins. Also, extra ligand molecules are not benificial to the formation of coronal structures of corona-BioMOF (Fig. S9). Subsequently, diffuse reflectance infrared Fourier-transform spectra (DRIFTS) were conducted to achieve a deeper insight into the interaction between protein and BioMOF [47]. As shown in Fig. 2I, the adsorption bands at 600–1500 cm–1 were assigned to the bending vibration of the imidazole ring, while the bands at 1620–1730 cm–1 were amide I stretching characteristic of proteins and the band at 409–440 cm–1 were related to the stretch of Zn–N and Zn–O [37], [48], [49], [50]. The DRIFTS of the corona-BioMOF exhibited a red shift of the amide-I stretch (1644 cm−1) compared to BioMOF (1666 cm−1) and ZIF-8 (1703 cm−1), indicating stronger chemical interaction between proteins and MOFs, which was consistent with previous report [37]. The band at 422 cm−1 was assigned to the stretching vibration of Zn–N and the band at 410 cm−1 was assigned to the bond of Zn–O. As shown in partial enlarged view of Fig. 2I, from ZIF to BioMOF, the peak intensity of Zn–N bond gradually weakened at 428 cm−1 and the Zn–O bond increased at 410 cm−1, indicated that the Zn–N bonds were broken to form Zn–O bonds owing to the water-induced defect. When the protein is attached on the surface of the BioMOF, the peak intensity of Zn–O (410 cm−1) weakened and the peak at 422 cm−1 increased as the N of the protein bound to the Zn site of the BioMOF to form a new Zn–N bond. The possible formation mechanism for the binding of DOX@AFt on the surface of BioMOF is illustrated in Fig. 2J. The involvement of water cleaved the Zn–N bond of ZIF-8 and induced a point defect to expose the Zn active site [51], [52], [53]. This facilitates coordination between N-terminal of the later added protein and Zn active site, driving the self-assembly of the protein on the BioMOF surface. These results show that the protein binding occurred via water-induced linker defect mechanism on the surface of BioMOF, where the binding could directly cooperate with the active site of the metal ion. 3.2 Targeting investigation of the corona-BioMOF To confirm the TfR1 targeting in cancer cells, the characterization result of TfR1 in TfR1-positive MDA-MB-231 cancer cells by immunofluorescence staining was shown in Fig. S10. By antibody blocking assay towards TfR1, the cellular interaction mediated by AFt and TfR1 was further proven (Fig. S11 and S12). The CLSM imaging and flow cytometry analysis all showed that the increase of fluorescence intensity is not obvious after incubated with corona-BioMOF, indicating that the interaction was mediated by TfR1. Then the precise targeting ability of corona-BioMOF was investigated (Fig. 3 A). The TfR1-negative MX-1 cancer cells and normal breast MCF-10A cells were utilized as the control [54]. After incubation with corona-BioMOF, MDA-MB-231 cells presented an 8-fold higher fluorescence intensity over MX-1 and MCF-10A cells. Meanwhile, after co-culturing with corona-BioMOF, the fluorescence intensity of the MDA-MB-231 cells was 5-fold higher than that of (DOX@AFt+GOx)@ZIF-8 without spiky surface. The results all indicated the higher cellular uptake ability of corona-BioMOF compared to (DOX@AFt+GOx)@ZIF-8, revealing the superiority of the DOX@AFt-based spiky surface on corona-BioMOF for improving the targeting ability. Also, the dynamic cellular uptake behaviors of corona-BioMOF were investigated. As shown in Fig. 3B, the fluorescence signals of corona-BioMOF and LysoSensor blue (a blue fluorescent marker of lysosome) in MDA-MB-231 cells overlapped gradually from 0.5 h to 6 h, confirming that corona-BioMOF was uptaken by lysosome. Also, it can be seen that some red fluorescence signals appeared in the nucleus at 6 h, implying that chemotherapeutic drug, DOX, achieved lysosomal escape and entered the nucleus due to the imidazole protonation resulted lysosomal rupture [46]. After 6 h, the intensity of blue fluorescence decreased obviously, verifying that some lysosomes have been degraded due to the drug-induced cell apoptosis (Fig. S13). Correspondingly, after 2 h of co-culturing, Confocal laser scanning microscope (CLSM) images showed that MDA-MB-231 cells emitted bright red fluorescence (DOX fluorescence) while MX-1 cells had poor fluorescence, indicating that the corona-BioMOF was rapidly internalized into the TfR1-positive targeted cells within 2 h. Furthermore, the content of Zn element in MDA-MB-231 cells after incubation with corona-BioMOF was 2-fold more than (DOX@AFt+GOx)@ZIF-8 (Fig. 3C). The above results showed that the spiky DOX@AFt coating promoted the cellular uptake, and corona-BioMOF possessed distinct precise targeting and intracellular lysosome-specific drug delivery abilities.Fig. 3 Evaluation on targeting behavior and killing efficacy of the corona-BioMOF in vitro. (A) Flow cytometry analysis of corona-BioMOF and (DOX@AFt+GOx)@ZIF-8 after incubation in TfR1-positive MDA-MB-231 cancer cells, TfR1-negative MX-1 cancer cells, and normal breast MCF-10A cells. (B) CLSM imaging and quantitative fluorescence analysis incubated with 100 μg mL−1 corona-BioMOF in MDA-MB-231 cells for 0.5, 2, 6, 12 h, and in MX-1 cells for 2 h. The red fluorescence was associated with DOX in corona-BioMOF, and the blue fluorescence was expressed by lysosome localized LysoSensor Blue probe. The locations in yellow shadow represent the nucleus. Scale bar represents 40 μm. (C) Quantitative analysis of Zn element by ICP-AES in MDA-MB-231 cells after incubated with (a) (DOX@AFt+GOx)@ZIF-8 and (b) corona-BioMOF. The data are presented as the mean ± SD, n=3. **P < 0.01, analyzed by Student’s t test. 3.3 Therapeutic efficacy in vitro Through standard methyl thiazolyl tetrazolium (MTT) assay, corona-BioMOF showed strong cytotoxicity towards MDA-MB-231 (IC50 = 2.49 μg mL−1) and good security for the normal breast cell line MCF-10A (IC50 = 64.1 μg mL−1), which was ascribed to the difference of cellular microenvironment and the receptor interactions (Fig. 4 A and Fig. S14). In contrast to this, the BioMOF displayed weaker cytotoxicity towards MDA-MB-231 (IC50 = 5.28 μg mL−1) and better security for MCF-10A (IC50 = 98.3 μg mL−1) (Fig. S15), which further proved the therapeutic superiority of the corona structure. Fig. 4B showed that corona-BioMOF exhibited excellent cell killing efficiency towards MDA-MB-231 cells. Moreover, to inspect the possible apoptosis mechanism induced by corona-BioMOF, FITC-Annexin V/PI flow cytometry analysis was conducted. The cell apoptotic rate of the corona-BioMOF treated group (31.5%) was much higher than that of groups treated by ZIF-8 (10.8%), GOx@ZIF-8 (16.8%), DOX (18.3%), DOX@AFt@ZIF-8 (18.6%), and DOX@AFt (26.0%), suggesting a most efficient therapy performance with corona-BioMOF (Fig. 4C). Meanwhile, no severe necrosis was exhibited for the MCF-10A cells treated with corona-BioMOF, which meant its minimized inflammation and damage to normal cells (Fig. S16). Therefore, these results demonstrated the therapeutic superiority of the corona-BioMOF.Fig. 4 (A) Cell viability on MCF-10A and MDA-MB-231 cells in the presence of corona-BioMOF for 12 h (n=3). (B) Cell viability after various treatment in MCF-10A and MDA-MB-231 cells (n=3). ***P < 0.001, analyzed by Student’s t test. (C) Flow-cytogram representing apoptosis assay based on Annexin V-FITC and PI staining of MDA-MB-231 cells after various treatment for 8 h. 3.4 Antitumor activity of corona-BioMOF in vivo We further evaluated its antitumor activity in subcutaneous MDA-MB-231 tumor-bearing nude mice. According to the treatment schedule (Fig. 5 A), after the tumor size reached approximately 90 mm3, MDA-MB-231 tumor-bearing nude mice were intravenously injected with PBS, ZIF-8, GOx@ZIF-8, DOX, DOX@AFt, DOX@AFt@ZIF-8, or corona-BioMOF. Immunofluorescence staining assay of tissue section was performed and verified the TfR1 expression on the tumor cells at first (Fig. 5B). Before injection, the stability of corona-BioMOF nanovehicle in serum was characterized by TEM (Fig. S17 and S18). The corona-BioMOF remained its original coronal morphology in serum after 1 h incubation and the MOF backbone showed little change, while BioMOF has undergone an obvious degradation. It might be attributed to the protection from the cage structure of spiky proteins on the external surface. After injected with corona-BioMOF, the fluorescence signal of DOX from corona-BioMOF accumulated at the tumor sites and peaked at 8 h, and a strong signal was found in the bladder at 24 h (Fig. 5C), indicating the gradual accumulation pathway of corona-BioMOF in tumor site and its final excretion through kidney metabolism. The biodistribution profiles of drugs were also investigated through ex vivo fluorescence images (Fig. 5D). With prolonged metabolism in vivo, similar to most materials, corona-BioMOF might aggregate in the liver and kidneys in the short time. However, the fluorescence intensities in the liver and kidneys decreased obviously at 24 h, demonstrating the good biodegradability and biosafety of the corona-BioMOF nanovehicle, which might be related to the reassuring composition including biological proteins, Zn2+, and 2-MIM. In stark contrast, the fluorescence intensity was strong in the tumor but much weaker in other major organs, confirming the good targeting effect of the corona-BioMOF in tumor site. The most significant inhibition of tumor growth was achieved with the corona-BioMOF treated MDA-MB-231 tumor-bearing mice (Fig. S19). The therapy performance of corona-BioMOF was further evaluated by tumor sizes, tumor weights, relative tumor volumes, and tumor growth inhibition (TGI) rate (Fig. 5E–G and Fig. S20). Significantly, the TGI rate in the corona-BioMOF group was increased to 86.5% at 22 days, displaying the most favorable treatment results. It is worth to be highlighted that the therapeutic effect of corona-BioMOF was significantly better than DOX@AFt and slightly higher than DOX@AFt@ZIF-8, which is in contrast to the minor difference in the cell cytotoxicity assay between them, further reflecting the unique size-induced enhanced permeability and retention (EPR) effect and practical application potential of corona-BioMOF. Besides, the body weights of mice also showed negligible changes affected by various treatments compared to PBS group, which further verified their prominent therapeutic outcomes (Fig. 5H).Fig. 5 In vivo antitumor activity of the corona-BioMOF. (A) Schematic illustration of the therapeutic schedule of the corona-BioMOF. (B) Immunofluorescence image of tumor section from MDA-MB-231 tumor-bearing mice. Scale bar represents 20 μm. (C) In vivo fluorescence imaging of MDA-MB-231 tumor-bearing mice after intravenous injection of corona-BioMOF. (D) Distribution of corona-BioMOF in tumor (Tu) and major viscera (He, heart; Li, liver; Sp, spleen; Lu, lung; and Ki, kidney) of MDA-MB-231 tumor-bearing mice at (Ⅰ) 8 h and (Ⅱ) 24 h post intravenous injection with corona-BioMOF. (E) Photographs of tumor dissection. (F) Analysis of tumor weights after 22 days of different treatments at a dosing frequency via intravenous injection with (a) PBS, (b) ZIF-8, (c) GOx@ZIF-8, (d) DOX, (e) DOX@AFt, (f) DOX@AFt@ZIF-8, and (g) corona-BioMOF. (G) Analysis of relative tumor volumes. (H) Body weights of mice. (I) Histological H&E analysis and TUNEL staining of tumors collected from MDA-MB-231 tumor-bearing mice after 22 days with various treatments. Scale bars represent 100 μm. Data above are presented as the mean ± SD, n=5. **P < 0.01, ***P < 0.001, and ****P < 0.0001, analyzed by Student’s t test. Moreover, we subsequently investigated the histological damage and apoptosis levels of tumors by using anatomical and histological observation, hematoxylin and eosin (H&E) staining, and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay, respectively (Fig. 5I, Fig. S21 and S22, Supporting Information). The results demonstrated no obvious lesions or abnormalities occurred in major organs compared with PBS group during treatment periods, confirming high biocompatibility of corona-BioMOF. The results showed that severe apoptosis happened in tumor cells after the treatments with corona-BioMOF, whereas only moderate damage was examined for those tumors treated with GOx@ZIF-8, DOX, DOX@AFt, or DOX@AFt@ZIF-8, further demonstrating the effectiveness of the proposed corona-BioMOF nanovehicle. Finally, quantitative results obtained from inductively coupled plasma-atomic emission spectrometry (ICP-AES) showed that Zn2+, the degradation product of the corona-BioMOF, could be accumulated in liver through the blood circulation (Fig. S23 and Table S1), which was consistent with previous reports [55]. The above results indicated that corona-BioMOF exhibited decent tumor-targeting ability, effective therapeutic effect, and biosecurity, fully indicating its application performance in cancer therapy. 4 Conclusion In summary, through mimicking the coronal structure and size of CoV, we designed a corona-BioMOF nanovehicle fabricated with a recognition proteins-based spiky surface and a hierarchical wrapping pattern for precise targeting and controlled drug release in cancer therapy. The formation mechanism of the special topography was also studied. The corona-BioMOF displayed remarkable specificity and lethality towards receptor-positive cancer cells in vitro and in vivo. These findings remind us of an interesting news, which reported that SARS-CoV-2 can induce remission of Hodgkin lymphoma [56]. We believe that although viruses are vicious, we can learn from their slyness and take advantages of it for human health. Therefore, the unique design of corona-BioMOF can provide an inspirational biomimetic strategy to offer an efficient tool for precise cargo delivery in diagnosis and treatment, and open the door for application potentials in precision medicine, especially in vivo tracing, individualized interaction diagnostics, and tailored target therapies. Appendix A. Supplementary data Supplementary data to this article can be found online. CRediT authorship contribution statement Huafeng Wang: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Data curation, Writing – original draft, Visualization. Shi Li: Methodology, Investigation, Data curation. Lei Wang: Investigation, Data curation. Zimei Liao: Methodology, Investigation. Hang Zhang: Investigation, Supervision. Tianxiang Wei: Conceptualization, Methodology, Formal analysis, Writing – original draft, Writing – review & editing, Data curation, Supervision, Funding acquisition. Zhihui Dai: Writing – review & editing, Supervision, Project administration, Funding acquisition. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability Data will be made available on request. Acknowledgements This work was financially supported by the National Natural Science Foundation of China for the project (22234005, 21974070 and 22074064), and the Natural Science Foundation of Jiangsu Province (BK20192008). ==== Refs References 1 Mitchell M.J. Billingsley M.M. Haley R.M. Wechsler M.E. Peppas N.A. Langer R. Engineering precision nanoparticles for drug delivery Nat. Rev. Drug Discov. 20 2021 101 124 10.1038/s41573-020-0090-8 33277608 2 Li S.P. Zhang Y.L. Ho S.H. Li B.Z. Wang M.F. Deng X.W. Yang N. Liu G.N. Lu Z.F. Xu J.C. Shi Q.W. Han J.Y. Zhang L.R. Wu Y. Zhao Y.L. Nie G.J. Combination of tumour-infarction therapy and chemotherapy via the co-delivery of doxorubicin and thrombin encapsulated in tumour-targeted nanoparticles Nat. Biomed. Eng. 4 2020 732 742 10.1038/s41551-020-0573-2 32572197 3 Wang H. Wu H. Yi Y. Xue K.-F. Xu J.-F. Wang H. Zhao Y. Zhang X. Self-motivated supramolecular combination chemotherapy for overcoming drug resistance based on acid-activated competition of host–guest interactions CCS Chem. 3 2021 1413 1425 10.31635/ccschem.021.202100964 4 Sunwoo K. Won M. Ko K.P. Choi M. Arambula J.F. Chi S.G. Sessler J.L. Verwilst P. Kim J.S. Mitochondrial relocation of a common synthetic antibiotic: a non-genotoxic approach to cancer therapy Chem 6 2020 1408 1419 10.1016/j.chempr.2020.03.004 32864504 5 Sun L.L. Xiong Z.J. Shen F.Y. Wang Z.H. Liu Z. Biological membrane derived nanomedicines for cancer therapy Sci. China Chem. 64 2021 719 733 10.1007/s11426-020-9943-9 6 Schwarzer D. Buettner F.F. Browning C. Nazarov S. Rabsch W. Bethe A. Oberbeck A. Bowman V.D. Stummeyer K. Mühlenhoff M. Leiman P.G. Gerardy-Schahn R. A multivalent adsorption apparatus explains the broad host range of phage phi92: a comprehensive genomic and structural analysis J. Virol. 86 2012 10384 10398 10.1128/JVI.00801-12 22787233 7 Shirato K. Kawase M. Matsuyama S. Middle East respiratory syndrome coronavirus infection mediated by the transmembrane serine protease TMPRSS2 J. Virol. 87 23 2013 12552 12561 24027332 8 Zhou P. Yang X.L. Wang X.G. Hu B. Zhang L. Zhang W. Si H.R. Zhu Y. Li B. Huang C.L. Chen H.D. Chen J. Luo Y. Guo H. Jiang R.D. Liu M.Q. Chen Y. Shen X.R. Wang X. Zheng X.S. Zhao K. Chen Q.J. Deng F. Liu L.L. Yan B. Zhan F.X. Wang Y.Y. Xiao G.F. Shi Z.L. A pneumonia outbreak associated with a new coronavirus of probable bat origin Nature 579 2020 270 273 10.1038/s41586-020-2012-7 32015507 9 Li F. Li W.H. Farzan M. Harrison S.C. Structure of SARS coronavirus spike receptor-binding domain complexed with receptor Science 309 2005 1864 1868 10.1126/science.1116480 16166518 10 Walls A.C. Park Y.J. Tortorici M.A. Wall A. McGuire A.T. Veesler D. Structure, function, and antigenicity of the SARS-CoV-2 spike glycoprotein Cell 181 2020 281 292 10.1016/j.cell.2020.02.058 32155444 11 Nie C.X. Stadtmüller M. Yang H. Xia Y. Wolff T. Cheng C. Haag R. Spiky nanostructures with geometry-matching topography for virus inhibition Nano Lett. 20 2020 5367 5375 10.1021/acs.nanolett.0c01723 32515974 12 Zhao T.C. Chen L. Wang P.Y. Li B.H. Lin R.F. Al-Khalaf A.A. Hozzein W.N. Zhang F. Li X.M. Zhao D.Y. Surface-kinetics mediated mesoporous multipods for enhanced bacterial adhesion and inhibition Nat. Commun. 10 2019 4387 10.1038/s41467-019-12378-0 31558724 13 Song H. Ahmad Nor Y. Yu M.H. Yang Y. Zhang J. Zhang H.W. Xu C. Mitter N. Yu C.Z. Silica nanopollens enhance adhesion for long-term bacterial inhibition J. Am. Chem. Soc. 138 2016 6455 6462 10.1021/jacs.6b00243 27139159 14 Edwards K.F. Steward G.F. Schvarcz C.R. Ostling A. Making sense of virus size and the tradeoffs shaping viral fitness Ecol. Lett. 24 2 2021 363 373 33146939 15 Rodrigues R.A.L. Abrahão J.S. Drumond B.P. Kroon E.G. Giants among larges: how gigantism impacts giant virus entry into amoebae Curr. Opin. Microbiol. 31 2016 88 93 10.1016/j.mib.2016.03.009 27039270 16 Li L. Sun W. Zhong J. Yang Q. Zhu X. Zhou Z. Zhang Z. Huang Y. Multistage nanovehicle delivery system based on stepwise size rreduction and charge reversal for programmed nuclear targeting of systemically administered anticancer drugs Adv. Funct. Mater. 25 2015 4101 4113 10.1002/adfm.201501248 17 Zhang M. Liu X. Luo Q. Wang Q. Zhao L. Deng G. Ge R. Zhang L. Hu J. Lu J. Tumor environment responsive degradable CuS@mSiO2@MnO2/DOX for MRI guided synergistic chemo-photothermal therapy and chemodynamic therapy Chem. Eng. J. 389 2020 124450 10.1016/j.cej.2020.124450 18 Shao Y.L. Liu B. Di Z.H. Zhang G. Sun L.D. Li L.L. Yan C.H. Engineering of upconverted metal-organic frameworks for near-infrared light-triggered combinational photodynamic/chemo-/immunotherapy against hypoxic tumors J. Am. Chem. Soc. 142 2020 3939 3946 10.1021/jacs.9b12788 31968933 19 Chen W.H. Luo G.F. Vázquez-González M. Cazelles R. Sohn Y.S. Nechushtai R. Mandel Y. Willner I. Glucose-responsive metal–organic-framework nanoparticles act as “smart” sense-and-treat carriers ACS Nano 12 2018 7538 7545 10.1021/acsnano.8b03417 29969227 20 Jia Q. Li Z. Guo C. Huang X. Kang M. Song Y. He L. Zhou N. Wang M. Zhang Z. Fu G. Du M. PEGMA-modified bimetallic NiCo Prussian blue analogue doped with Tb(III) ions: Efficiently pH-responsive and controlled release system for anticancer drug Chem. Eng. J. 389 2020 124468 10.1016/j.cej.2020.124468 21 Zhang K. Meng X.D. Yang Z. Dong H.F. Zhang X.J. Enhanced cancer therapy by hypoxia-responsive copper metal-organic frameworks nanosystem Biomaterials 258 2020 120278 10.1016/j.biomaterials.2020.120278 22 Geng P. Yu N. Macharia D.K. Meng R. Qiu P. Tao C. Li M. Zhang H. Chen Z. Lian W. MOF-derived CuS@Cu-MOF nanocomposites for synergistic photothermal-chemodynamic-chemo therapy Chem. Eng. J. 441 2022 135964 10.1016/j.cej.2022.135964 23 Ji Z. Li T. Yaghi O.M. Sequencing of metals in multivariate metal-organic frameworks Science 369 2020 674 780 10.1126/science.aaz4304 32764067 24 Cai G.R. Yan P. Zhang L.L. Zhou H.C. Jiang H.L. Metal–organic framework-based hierarchically porous materials: synthesis and applications Chem. Rev. 121 2021 12278 12326 10.1021/acs.chemrev.1c00243 34280313 25 Ge X.Y. Wong R. Anisa A. Ma S.Q. Recent development of metal-organic framework nanocomposites for biomedical applications Biomaterials 281 2022 121322 10.1016/j.biomaterials.2021.121322 26 Deng X.R. Liang S. Cai X.C. Huang S.S. Cheng Z.Y. Shi Y.S. Pang M.L. Ma P.A. Lin J. Yolk-shell structured Au nanostar@metal–organic framework for synergistic chemo-photothermal therapy in the second near-infrared window Nano Lett. 19 2019 6772 6780 10.1021/acs.nanolett.9b01716 31496257 27 Ni K.Y. Luo T.K. Culbert A. Kaufmann M. Jiang X.M. Lin W.B. Nanoscale metal-organic framework co-delivers TLR-7 agonists and anti-CD47 antibodies to modulate macrophages and orchestrate cancer immunotherapy J. Am. Chem. Soc. 142 2020 12579 12584 10.1021/jacs.0c05039 32658476 28 Lian X.Z. Huang Y.Y. Zhu Y.Y. Fang Y. Zhao R. Joseph E. Li J.L. Pellois J.P. Zhou H.C. Enzyme-MOF nanoreactor activates nontoxic paracetamol for cancer therapy Angew. Chem. Int. Ed. 57 2018 5725 5730 10.1002/anie.201801378 29 Liu W. Pan Y. Zhong Y. Li B. Ding Q. Xu H. Qiu Y. Ren F. Li B. Muddassir M. Liu J. A multifunctional aminated UiO-67 metal-organic framework for enhancing antitumor cytotoxicity through bimodal drug delivery Chem. Eng. J. 412 2021 127899 10.1016/j.cej.2020.127899 30 Park K.S. Ni Z. Côté A.P. Choi J.Y. Huang R. Uribe-Romo F.J. Chae H.K. O'Keeffe M. Yaghi O.M. Exceptional chemical and thermal stability of zeolitic imidazolate frameworks Proc. Natl. Acad. Sci. U.S.A. 103 2006 10186 10191 10.1073/pnas.0602439103 16798880 31 He Z.M. Huang X.L. Wang C. Li X.L. Liu Y.J. Zhou Z.J. Wang S. Zhang F.W. Wang Z.T. Jacobson O. Zhu J.J. Yu G.C. Dai Y.L. Chen X.Y. A catalase-like metal-organic framework nanohybrid for O2-evolving synergistic chemoradiotherapy Angew. Chem. Int. Ed. 58 2019 8752 8756 10.1002/anie.201902612 32 Bůžek D. Adamec S. Lang K. Demel J. Metal–organic frameworks vs. buffers: case study of UiO-66 stability Inorg. Chem. Front. 8 2021 720 734 10.1039/D0QI00973C 33 Cunha D. Yahia M.B. Hall S. Miller S.R. Chevreau H. Elkaïm E. Maurin G. Horcajada P. Serre C. Chem. Mater. 25 2013 2767 2776 10.1021/cm400798p 34 Chen T.T. Yi J.T. Zhao Y.Y. Chu X. Biomineralized metal–organic framework nanoparticles enable intracellular delivery and endo-lysosomal release of native active proteins J. Am. Chem. Soc. 140 2018 9912 9920 10.1021/jacs.8b04457 30008215 35 Wang Z.J. Fu Y. Kang Z.Z. Liu X.G. Chen N. Wang Q. Tu Y.Q. Wang L.H. Song S.P. Ling D.S. Song H.Y. Kong X.Q. Fan C.H. Organelle-specific triggered release of immunostimulatory oligonucleotides from intrinsically coordinated DNA-metal-organic frameworks with soluble exoskeleton J. Am. Chem. Soc. 139 2017 15784 15791 10.1021/jacs.7b07895 29024595 36 Zheng H.Q. Zhang Y.N. Liu L.F. Wan W. Guo P. Nyström A.M. Zou X.D. One-pot synthesis of metal–organic frameworks with encapsulated target molecules and their applications for controlled drug delivery J. Am. Chem. Soc. 138 2016 962 968 10.1021/jacs.5b11720 26710234 37 Liang K. Ricco R. Doherty C.M. Styles M.J. Bell S. Kirby N. Mudie S. Haylock D. Hill A.J. Doonan C.J. Falcaro P. Biomimetic mineralization of metal-organic frameworks as protective coatings for biomacromolecules Nat. Commun. 6 2015 7240 10.1038/ncomms8240 26041070 38 Liang W. Wied P. Carraro F. Sumby C.J. Nidetzky B. Tsung C.-K. Falcaro P. Doonan C.J. Metal−organic framework-based enzyme biocomposites Chem. Rev. 121 2021 1077 1129 10.1021/acs.chemrev.0c01029 33439632 39 Gou S.Q. Huang Y.M. Wan Y. Ma Y. Zhou X. Tong X.L. Huang J. Kang Y.J. Pan G.Q. Dai F.Y. Xiao B. Multi-bioresponsive silk fibroin-based nanoparticles with on-demand cytoplasmic drug release capacity for CD44-targeted alleviation of ulcerative colitis Biomaterials 212 2019 39 54 10.1016/j.biomaterials.2019.05.012 31103945 40 Liang M.M. Fan K.L. Zhou M. Duan D.M. Zheng J.Y. Yang D.L. Feng J. Yan X.Y. H-ferritin–nanocaged doxorubicin nanoparticles specifically target and kill tumors with a single-dose injection Proc. Natl. Acad. Sci. U.S.A. 111 2014 14900 14905 10.1073/pnas.1407808111 25267615 41 Li L. Fang C.J. Ryan J.C. Niemi E.C. Lebrón J.A. Björkman P.J. Arase H. Torti F.M. Torti S.V. Nakamura M.C. Seaman W.E. Binding and uptake of H-ferritin are mediated by human transferrin receptor-1 Proc. Natl. Acad. Sci. U.S.A. 107 2010 3505 3510 10.1073/pnas.0913192107 20133674 42 Crichton R.R. Structure and function of ferritin Angew. Chem. Int. Ed. 12 1973 57 65 10.1002/anie.197300571 43 Zhen Z.P. Tang W. Chen H.M. Lin X. Todd T. Wang G. Cowger T. Chen X.Y. Xie J. RGD-modified apoferritin nanoparticles for efficient drug delivery to tumors ACS Nano 6 2013 4830 4837 10.1021/nn305791q 44 Mohanty A. Parida A. Raut R.K. Behera R.K. Ferritin: a promising nanoreactor and nanocarrier for bionanotechnology ACS Bio. Med. Chem. Au 2 2022 258 281 10.1021/acsbiomedchemau.2c00003 45 Hu C. Bai Y.X. Hou M. Wang Y. Wang L.C. Cao X. Chan C.W. Sun H. Li W.B. Ge J. Ren K.N. Defect-induced activity enhancement of enzyme-encapsulated metal-organic frameworks revealed in microfluidic gradient mixing synthesis Sci. Adv. 6 2020 eaax5785 10.1126/sciadv.aax5785 32064336 46 Alsaiari S.K. Patil S. Alyami M. Alamoudi K.O. Aleisa F.A. Merzaban J.S. Li M. Khashab N.M. Endosomal escape and delivery of CRISPR/Cas9 genome editing machinery enabled by nanoscale zeolitic imidazolate framework J. Am. Chem. Soc. 140 2018 143 146 10.1021/jacs.7b11754 29272114 47 Wang X.J. Zhang X. Pandharkar R. Lyu J.F. Ray D. Yang Y. Kato S. Liu J. Wasson M.C. Islamoglu T. Li Z. Hupp J.T. Cramer C.J. Gagliardi L. Farha O.K. Insights into the structure–activity relationships in metal–organic framework-supported nickel catalysts for ethylene hydrogenation ACS Catal. 10 2020 8995 9005 10.1021/acscatal.0c01844 48 Kayani Z.N. Iqbal M. Riaz S. Zia R. Naseem S. Fabrication and properties of zinc oxide thin film prepared by sol-gel dip coating method Mater. Sci.-Poland 33 2015 515 520 10.1515/msp-2015-0085 49 Hu Y. Kazemian H. Rohani S. Huang Y.N. Song Y. In situ high pressure study of ZIF-8 by FTIR spectroscopy Chem. Commun. 47 2011 12694 12696 10.1039/c1cc15525c 50 Zhang M. Shang Q.G. Wan Y.Q. Cheng Q.R. Liao G.Y. Pan Z.Q. Self-template synthesis of double-shell TiO2@ZIF-8 hollow nanospheres via sonocrystallization with enhanced photocatalytic activities in hydrogen generation Appl. Catal. B: Environ. 241 2019 149 158 10.1016/j.apcatb.2018.09.036 51 Han R. Tymińska N. Schmidt J.R. Sholl D.S. Propagation of degradation-induced defects in zeolitic imidazolate frameworks J. Phys. Chem. C 123 2019 6655 6666 10.1021/acs.jpcc.9b00304 52 Zhang C.Y. Han C. Sholl D.S. Schmidt J.R. Computational characterization of defects in metal–organic frameworks: spontaneous and water-induced point defects in ZIF-8 J. Phys. Chem. Lett. 7 2016 459 464 10.1021/acs.jpclett.5b02683 26771275 53 Cheng P.F. Hu Y.H. H2O-functionalized zeolitic Zn(2-methylimidazole)2 framework (ZIF-8) for H2 storage J. Phys. Chem. C 118 2014 21866 21872 10.1021/jp507030g 54 Cao C.Q. Wang X.X. Cai Y. Sun L. Tian L.X. Wu H. He X.Q. Lei H. Liu W.F. Chen G.J. Zhu R.X. Pan Y.X. Targeted in vivo imaging of microscopic tumors with ferritin-based nanoprobes across biological barriers Adv. Mater. 26 2014 2566 2571 10.1002/adma.201304544 24532221 55 Guo Z. Luo Y. Zhang P. Chetwynd A.J. Qunhui Xie H. Abdolahpur Monikh F. Tao W. Xie C. Liu Y. Xu L. Zhang Z. Valsami-Jones E. Lynch I. Zhao B. Deciphering the particle specific effects on metabolism in rat liver and plasma from ZnO nanoparticles versus ionic Zn exposure Environ. Int. 136 2020 105437 10.1016/j.envint.2019.105437 56 S. Challenor, D. Tucker, SARS-CoV-2-induced remission of Hodgkin lymphoma, Br. J. Haematol. 192 (2021) 415. doi: 10.1111/bjh.17116.

0

PMC9749395

NO-CC CODE

2022-12-15 23:22:54

no

Chem Eng J. 2022 Dec 14;:140963

utf-8

Chem Eng J

10.1016/j.cej.2022.140963

oa_other

==== Front J Hous Econ J Hous Econ Journal of Housing Economics 1051-1377 1096-0791 The Authors. Published by Elsevier Inc. S1051-1377(22)00076-6 10.1016/j.jhe.2022.101904 101904 Article Landlords’ Rental Businesses Before and After the COVID-19 Pandemic: Evidence from a National Cross-Site Survey de la Campa Elijah A. a⁎ Reina Vincent J. b1 a Harvard Kennedy School Bloomberg Harvard City Leadership Initiative, 79 JFK St., Mailbox #74, Cambridge, MA 02138 b University of Pennsylvania Weitzman School of Design, 210 S 34th St, Philadelphia, PA 19104 ⁎ Corresponding author. 1 Vincent J. Reina's contribution to this article occurred prior to him taking a leave of absence from the University of Pennsylvania to join the Biden-Harris administration and reflect his personal views only. 14 12 2022 14 12 2022 1019041 3 2022 24 8 2022 8 12 2022 © 2022 The Authors. Published by Elsevier Inc. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. This paper uses a survey of over 2,500 rental property owners in ten cities across the United States to determine the impact of the COVID-19 pandemic on landlords’ rent collection and business behavior. Our findings show that yearly rent collection was down significantly in 2020 relative to 2019—both within and across rental markets—and that an increasing number of owners have a large share of their portfolio behind on rent. Small owners and owners of color faced the highest exposure to deep tenant arrears in 2020, challenges they were also more likely to face prior to pandemic. Our findings show that owner business practices changed dramatically in 2020, with a higher share of landlords granting tenants rent extensions or forgiving back rent during the pandemic relative to prior. However, many owners also disinvested in their rental properties through deferred maintenance, missed mortgage payments, and property sale listings. Landlords of color pursued disinvestment strategies during the pandemic at an elevated rate compared to white landlords. Owners of properties in neighborhoods with more non-white residents were both more likely to experience decreased rent collection and more likely to pursue evictions and rental late fees holding constant rental payment rates, implying the pandemic has disproportionately affected renters in communities of color. Overall, our findings highlight the strain the pandemic has placed on the housing stock, which has implications for the long-term viability and affordability of many of these units. More concerningly, our results show that households of color—which have been disproportionately affected by the pandemic in other domains—were more likely to face punitive measures from landlords in both 2019 and 2020, suggesting the pandemic has exacerbated existing racial inequality in housing markets. Keywords Landlords Renters Housing affordability Racial discrimination COVID-19 ==== Body pmc1 Introduction The COVID-19 pandemic has had a profound impact on the ability of US renters to make rent. By December 2020, it was estimated that nearly one in five renter households were behind on monthly payments (Airgood-Obrycki et al. 2021), fueling a rent arrears crisis estimated to be as high as $57 billion (Parrot & Zandi 2021). While numerous studies have explored the consequences of this crisis for tenants, comparatively less is known about how rental property owners have absorbed and responded to this financial strain.1 This paper aims to fill a critical gap in our understanding of property owners’ businesses and behaviors during the pandemic, and in doing so, offers new insights into these topics under more typical circumstances. To accomplish this, we report results from an original and comprehensive survey of 2,500 owners across ten US cities. From February to April 2021, we asked these owners to assess the financial health of their pre- and post-COVID rental business.2 We also asked about the tools they have relied on to manage their rental properties and collected information on their demographic, business, and property characteristics. Using these detailed data, we explore heterogeneity in landlords’ rent collection and business responses, both before and after the pandemic, by race and portfolio size, rental market, and property-level neighborhood composition. Overall, we find that landlords’ rental properties generated a significantly lower share of their potential rental revenue in 2020 relative to 2019; landlords modified their business practices accordingly; and punitive actions, such as evictions and rental late fees, were more likely to be reported at properties located in neighborhoods with more non-white residents. Our work offers three key contributions. First, we add to an emerging though somewhat disjointed literature that explores COVID's impact on renters, property owners, and markets. Leveraging data gathered from property owners across ten cities, we corroborate evidence from renters that shows rental payments were down considerably during the first year of the pandemic (e.g., Engelhardt & Eriksen 2021). However, because our survey also asks about owners’ pre-pandemic rent collection, we extend this analysis to show that the pandemic is associated with proportionate year-over-year declines in payment across markets.3 Additionally, our survey asks landlords about their business responses to the crisis, building off several notable studies that have focused on single cities to provide important context to owners’ exposure to losses and practices throughout the pandemic (Reina et al. 2020a; Reina & Goldstein 2021). Critically, we put these disparate strands of information together to show that changes in landlords’ pandemic-era business practices cannot be explained by revenue loss alone, and that this holds true both within and across markets. Second, this paper provides rigorous insight into the conditions under which various segments of the landlord population operate. For example, it is well-documented that there are differences in the business practices and characteristics of small vs. large landlords (e.g., Immergluck & Law 2014; Decker 2021a) and owners of color vs. white owners (e.g., Choi & Young 2020).4 At the same time, few studies even prior to the pandemic have been able to examine the relative differences in the business practices of these populations in a single, unified context.5 Research during the pandemic has been similarly disjointed, with data from the National Multifamily Housing Council (NMHC) (2020) suggesting that rental payments have been down only slightly for large, institutional investors, and national survey data focused on smaller-scale landlords suggesting this population has struggled significantly with rent collection (Decker 2021b; Choi & Goodman 2020). Our survey, offered to a diverse group of thousands of landlords, unifies these disparate strands of research and highlights small landlords and landlords of color as populations that were more likely to struggle with rental collection prior to and during the pandemic relative to their peers. We also find that these populations heavily disinvested in their properties during the pandemic, conditional on rent collection, which likely has the unintended consequence of perpetuating financial and housing instability for vulnerable tenants and property owners alike. Finally, this paper provides additional evidence of the pandemic's outsized impact on Americans of color. We show that, during the pandemic, rental properties located in neighborhoods with a higher share of residents of color were significantly less likely to have tenants experiencing rental forgiveness and significantly more likely to have tenants facing rental late fees or eviction. These results hold after controlling for differences in rental collection; cannot be fully explained by landlord sorting on demographic and business characteristics, nor by residential sorting on economic characteristics; and were mostly observed in these communities prior to the pandemic Overall, we conclude that the disproportionate financial strain experienced by renters in communities of color has likely been exacerbated by landlords’ tendency to pursue business practices in these communities that increase housing instability, and that racialized business practices have persisted and in some cases taken on new forms during the pandemic. These findings add to a literature that documents a long history of discrimination in the rental housing market for Black and Hispanic Americans (Hanson & Hawley 2011; Reina, Pritchett, & Wachter 2020b; Hepburn, Louis, & Desmond 2020), and highlight the need for current and future housing responses to be centered around and actively promote racial equity (Ellen et al. 2021). The remainder of this paper proceeds as follows. Section 2 describes the survey implementation and methodology, Section 3 reviews key findings for landlords’ rent collection, Section 4 reviews findings for landlords’ business practices, and Section 5 concludes. 2 Survey Implementation 2.1 Design and Setting The COVID-19 Landlord Survey is an extension of two prior survey efforts designed by members of the research team: one targeted owners of three or fewer rental properties in Albany and Rochester, New York and was distributed in June and October 2020 (de la Campa 2021), while the other was offered to landlords in Philadelphia (September 2020) and Los Angeles (December 2020) who had at least one tenant apply for pandemic-related emergency rental assistance (Reina et al. 2020a; Reina & Goldstein 2021). Both efforts offered insight into the pandemic's impact on landlords’ rental business, but they were also limited in scope. Accordingly, in December 2020, the research team began reaching out to cities and counties across the US to participate in a larger survey designed to explore the pandemic's impact across different types of rental markets, landlords, and properties. The survey was designed to collect information at two levels: for the landlord's entire city-specific portfolio, and for an individual property representative of the landlord's portfolio.6 For each level, we asked landlords about their pre- and post-COVID rental income, as well as the various actions they have taken to manage their rental business. The survey also asked for basic demographics on the landlord, including race, age, and gender. We also asked landlords general questions about their rental business, such as whether they rely on a property manager or have tenants who use Housing Choice Vouchers (HCVs).7 Municipalities were recruited through the Bloomberg Harvard City Leadership Initiative network, as well as through ongoing rent-relief evaluations being conducted by the Housing Initiative at Penn, and were asked to partner with the research team by sharing landlords’ contact information and facilitating outreach. Conversations with municipalities that maintained significant contact information for landlords—specifically, mobile phone number or email—were prioritized. Overall, the research team had conversations with nearly forty US cities and counties and partnered with ten cities to implement the COVID-19 Landlord Survey: Akron, Ohio; Albany and Rochester, New York; Indianapolis, Indiana; Los Angeles, California; Minneapolis, Minnesota; Philadelphia, Pennsylvania; Racine, Wisconsin; San Jose, California; and Trenton, New Jersey.8 While these municipalities were chosen with an eye towards achieving geographic spread, we caution that our sample is not necessarily representative of all cities in the US.9 Nonetheless, our sample of survey cities resembles the universe of US cities along a few dimensions. Table 1 reports descriptive statistics for residents and renter households of the pooled survey sample cities as well as the population of all US cities.10 Data come from the 2018 ACS 5-year sample, with means and medians calculated from pooled population totals (across all cities within each sample).Table 1 COVID-19 Landlord Survey Cities in Comparison with US Cities Table 1: Survey Cities US Metro and Micropolitan Principal Cities Mean SE Mean SE Panel A: Resident Characteristics White 34.7 0.05 48.2 0.02 Black 18.9 0.05 18.1 0.01 Hispanic 31.3 0.05 22.7 0.02 Asian 11.9 0.04 7.5 0.01 Other race 3.1 0.91 3.5 0.02 Median age (y) 34.0 0.39 36.1 1.4 N Residents 8,500,786 115,799,553 Panel B: Renter Household Characteristics Renter-occupied (among all households) 55.0 0.10 49.9 0.03 Reside in 1-unit property 27.7 0.16 27.9 0.04 Reside in 2-4 unit property 16.7 0.12 19.0 0.04 Reside in 5-9 unit property 11.6 0.11 12.5 0.03 Reside in 10-19 unit property 11.3 0.10 12.2 0.03 Reside in 20+ unit property 32.3 0.12 27.2 0.04 Median income ($) 34,967 1,195 34,041 3,913 Cost-burdened 53.8 0.20 48.7 0.06 Median gross rent ($) 1,057 12.6 908 41 Median age of housing structure (y) 65 1.6 54 3.4 N Renter Households 1,688,205 21,916,506 Notes: This table reports descriptive characteristics of residents and renter households for the ten, pooled COVID-19 Landlord Survey cities as well as for the universe of all US metro and micropolitan principal cities (N=1253). Data come from the 2018 ACS 5-year sample. Medians calculated as weighted averages of city-specific median estimates. Standard errors calculated using the margin of error provided by the Census Bureau for the American Community Survey. Unless otherwise indicated, the means and standard errors above are expressed as percentages. Categorical variables may not sum to 100 due to rounding. Cost-burdened renters are defined as those who spend 30 percent or more of their yearly income on yearly rent. The median age of residents across the cities in our sample is identical to that of residents in US cities as a whole (34.9). Just over half of all households in both survey and US cities are renter-occupied. The distribution of rental properties is also similar across the two groups, though survey cities have a higher share of large apartment buildings (32.3 percent of rental units are located in 20+ unit buildings in survey cities compared to 27.2 percent in U.S. cities as whole). The median income of renter households is also similar across the two groups ($38,577 vs. $36,691). There are also some key differences. Relative to US cities, survey cities are, on average, less white (34.7 vs. 48.1 percent) and more Hispanic (31.9 vs. 23.2 percent). The rental housing stock in survey cities is slightly older than that of US cities overall (built 65 vs. 54 years ago), and median rents are slightly higher ($1,186 vs. $1,027). Along these lines, the share of cost-burdened renters, defined as those who spend 30 percent or more of their income on rent, is slightly higher in survey cities compared to U.S. cities as a whole (53.8 vs. 48.7 percent). Overall, 1.7 million of the nation's 21.8 million city-based rental units are located in the ten cities in our survey sample. 2.2 Outreach and Response We distributed the COVID-19 Landlord Survey on a rolling basis from early February through mid-April 2021. In each city, every landlord for which contact information was obtained was invited—either via email or text message—to participate in the online survey.11 Table 2 shows response rates for each city. Overall, we sent out nearly 58,000 survey invites and received 2,850 partial or complete usable responses, for an overall response rate of 4.9 percent.12 This response rate ranged from a low of 1.3 percent in Los Angeles to a high of 8.8 percent in San Jose. In nearly all analyses, we focus on the sample of 2,547 landlords who reported owning at least one overlapping rental property (in their respective city) in 2019 and 2020.Table 2 COVID-19 Landlord Survey Response Rates Table 2: Overall Akron Albany Indianapolis Los Angeles Minneapolis Philadelphia Racine Rochester San Jose Trenton N Survey Invites 57,994 3,440 1,971 7,615 18,810 10,540 6,156 2,294 2,190 3,476 1,502 N Usable Responses 2,850 258 114 449 248 676 312 172 178 307 136 Response Rate 4.9 7.5 5.8 5.9 1.3 6.4 5.1 7.5 8.1 8.8 9.1 N Analysis Sample 2,547 236 107 413 219 593 277 158 163 267 114 Notes: This table reports, both overall and separately for each participating city, the number of survey invites (less outbound bounces), number of usable survey responses, survey response rate, and analysis sample size for the COVID-19 Landlord Survey. An additional 80 property managers responded to the survey but were routed to the end of the survey when they indicated they were not specifically property owners; these individuals are not included among the number of usable survey responses. Respondents were not asked questions about their rental business profitability and management if they did not report owning at least one overlapping rental property in 2019; the analysis sample excludes these individuals and is comprised solely of respondents who owned at least one overlapping rental property in 2019 and 2020. In Albany, Indianapolis, Los Angeles, Minneapolis, Racine, Philadelphia, San Jose, and Trenton, participants were invited to participate in the survey via email. In Akron and Rochester, participants were invited via text message (SMS). Data come from the COVID-19 Landlord Survey. In eight of the ten sample cities, we obtained landlord contact information from rental dwelling registries. In general, these registries exist to ensure safe living conditions for renters, and they typically require owners of residential properties with rental dwelling units to obtain a permit and pass an interior inspection before units can be legally leased to tenants.13 In San Jose, only owners of properties built before 1979 that contain three or more rental units are required to register.14 These older and larger rental buildings tend to be located in lower-income areas of the city, leading to a San Jose sample that has a disproportionate number of landlords who operate at the lower end of the rental market (though these landlords may also own properties in higher-income areas of the City). Compliance rates on rental registries vary from a low of around 10 percent in Indianapolis, to upwards of 70 percent in Trenton, to nearly 95 percent for San Jose's more limited registry.15 Landlord contact information for the remaining two cities—Los Angeles and Philadelphia—was obtained from emergency rental assistance (ERA) applications. In each city, it was incumbent upon tenants to apply for ERA, meaning the owners represented in this sample did not actively select into the process for receiving funds. Previous research finds that these properties include many landlords who are not traditionally engaged in ownership or trade organizations and/or any federal or local housing assistance programs (Reina & Goldstein 2021; Reina et al. 2020a). Our sample of landlords, therefore, is selected in various ways. First, they are landlords in core US urban centers. Second, they are either landlords who have registered with their cities’ rental registry, or landlords with tenants who applied for emergency rental assistance.16 And third, they are landlords who chose to answer our survey. Consequently, there are many reasons to be cautious when extrapolating to the entire universe of US landlords. We discuss the representativeness of this sample, based on observable characteristics, in the following section. 2.3 Respondent Characteristics and Representativeness Table 3 presents descriptive statistics for survey respondents and explores the representativeness of our sample using two additional data sources from the US Census. The first is the 2020 Current Population Survey Annual Social and Economic Supplement (CPS-ASEC). The CPS is the primary source for the nation's employment statistics and is offered monthly to a random sample of nearly 60,000 households in which there is at least one individual of working age (15 or older) who is not in the Armed Forces. The CPS-ASEC is the annual supplement to this survey effort that provides further detail on work activity and income sources, among other topics. We use the CPS-ASEC to identify all individuals who reported negative or positive rental income in 2019 (i.e., the year prior to the pandemic) and restrict our analysis to this subset.Table 3 Descriptive Statistics of Survey Respondents and Representativeness of Sample Table 3: Survey CPS-ASEC RHFS N Mean SD N Mean SD N Mean SD Male 2255 61.4 48.7 6254 53.5 50.0 Missing gender 2850 20.9 40.7 6254 0 0.0 White 2338 66.3 47.3 6254 75.5 41.6 Black 2338 11.5 31.9 6254 6.3 24.4 Hispanic 2338 6.3 24.3 6254 8.9 25.5 Asian 2338 8.6 28.0 6254 7.6 26.6 Missing race 2850 18.0 38.4 6254 0 0 20-29 years old 2380 0 14.9 6254 3.1 16.8 30–39 years old 2380 14.7 35.5 6254 12.5 31.0 40–49 years old 2380 17.8 38.3 6254 18.8 36.0 50–59 years old 2380 25.6 43.6 6254 21.3 40.5 60+ years old 2380 39.6 48.9 6254 44.1 50.0 Missing age 2850 16.5 37.1 6254 0 0 Individual owner 2255 87.6 32.9 4028 77.9 41.5 Missing ownership structure 2850 20.9 40.7 4330 3.5 18.3 Self-manages rental units 2703 72.3 44.8 4267 72.8 44.5 Missing property manager 2850 5.2 22.1 4330 1.7 12.9 Accepts HCVs 2709 20.8 40.6 3945 5.8 23.3 Missing HCV 2850 4.9 21.7 4330 6.1 23.8 Owns single-family rental(s) (SFRs) 2536 50.7 36.3 Missing home type 2850 1.3 31.2 Small landlord 2803 65.6 47.5 Mid-sized landlord 2803 17.2 37.7 Missing portfolio size 2850 1.6 12.7 Notes: This table reports descriptive statistics for the COVID-19 Landlord Survey respondents, for all individuals in the CPS-ASEC who reported rental income in 2019, and for all U.S. rental properties in 2018. The variables above are expressed as percentages. The omitted category for race is “Other Race.” The omitted category for age is “Under 20 years old.” No respondents in the survey reported being under 20. “Individual owner” indicates an owner who is not incorporated as an LLC or LLP. “Single-family rental” indicates a one- to four-unit rental property. “Small landlord” indicates a landlord who owns between 1 and 5 rental units. “Mid-sized landlord” indicates a landlord who owns between 6 and 19 rental units. The omitted category for landlord size is “large” (owns 20+ rental units). Categorical variables may not sum to 100 due to rounding. Survey data come from city administrative records and the COVID-19 Landlord Survey. CPS-ASEC data come from the 2020 Current Population Survey Additional. RFHS data come from the 2018 Rental Housing Finance Survey. The second is the 2018 Rental Housing Finance Survey (RHFS). The RHFS is offered on a triennial basis to a randomly drawn subset of the owners and/or managers of all US properties with at least one unit that is rented or vacant for-rent (as determined by the American Housing Survey). The RHFS provides insight into the financial, managerial, and physical characteristics of US rental properties over the prior 12-month period. Though the unit of observation for these data is the rental property, with a dearth of information on landlords, we use these data to roughly approximate the business characteristics of our sample. About 60 percent of survey respondents are male, compared to 53.5 percent of all landlords in the CPS-ASEC (53.5 percent). Two-thirds of survey respondents are white, 11.5 percent are Black, 6.3 percent are Hispanic, and 8.6 percent are Asian. A larger share of landlords are white (75.5 percent) and Hispanic (8.9 percent) in the CPS-ASEC, while a lower share are Black (6.3 percent). The over-representation of Black landlords in our sample likely reflects the fact that our survey was offered solely in large urban centers. Nearly 40 percent of our respondents are over the age of 60, the most common age range represented in the survey and a near exact match to landlords in the CPS-ASEC. Nearly 88 percent of owners are individual investors as opposed to owners incorporated as an LLC or LLP. Though an imperfect comparison, a similar share (77.9 percent) of all rental properties in the RHFS are owned by individual investors. The share of landlords who report managing their properties themselves (as opposed to through a manager) is nearly identical to the share of rental properties in the RHFS managed by the owner (72.3 vs. 72.8 percent). Around 20 percent of survey respondents accept HCVs, which is substantially larger than the share of properties in the RHFS that have at least one tenant using HCVs (5.8 percent).17 Just over 50 percent of landlords in our sample own at least one rental property with 1-4 units—commonly referred to as single-family rentals (SFRs)—which are most likely to be owned by small, individual investors (Freddie Mac 2018). Accordingly, nearly two-thirds of landlord respondents own a total of 1-5 rental units. Equal shares of the remainder own 6-19 or 20+ units.18 Overall, while our sample of survey respondents may be an imperfect snapshot for the universe of US landlords, it allows for important insight into a variety of landlords with a variety of property holdings.19 3 Landlords’ Rental Collection Before and After the Pandemic 3.1 Landlords’ rent collection decreased significantly in 2020 Figure 1 reports the landlords’ rental collection rates before (2019) and after (2020) the pandemic. Rent collected is expressed as a percentage of total rent charged across the portfolio and separated into four categories: 100, 90 to 99, 50 to 89, and less than 50 percent of yearly rent received.Figure 1 Landlords’ Rental Collection Prior to and During the Pandemic. Notes: This figure plots landlords’ rental collection rates in 2019 and 2020. Rental payment is expressed as a percentage of total rent charged, in a given year, for a landlord's rental portfolio. The number of survey respondents in the sample is 2,548. Data come from the COVID-19 Landlord Survey. Figure 1: In 2019, the vast majority (88.9 percent) of landlords reported collecting 90 percent or more of their charged yearly rent. In 2020, this share fell by nearly a third, to just over 60 percent, while the share reporting collection of 50 to 89 percent of rent rose from 8.2 percent in 2019 to 28.6 percent in 2020. We also see a substantial share of landlords experiencing serious financial strain during the pandemic, with the share of landlords collecting less than 50 percent of charged rent by year's end increasing from 2.9 percent in 2019 to 9.1 percent in 2020. A lack of data on landlords’ pre-pandemic rental collection makes it difficult to contextualize our 2019 baseline rates. However, our results generally align with those from two national rent tracking systems. First, data from the NMHC (2020) show that, in the two months prior to the pandemic, around 95 percent of units owned by large, professionally managed landlord organizations paid rent in full by the end of the month. Data from a partnership between Avail and the Urban Institute (2021) report a corresponding figure of 87 percent for small, mom-and-pop landlords. Additionally, the pre-pandemic share of landlords reporting 90 percent of more of rent received in our study is nearly identical to that reported by a large survey of landlords in Los Angeles (Reina & Goldstein 2021). There is comparatively more research that seeks to understand the extent of rental delinquency during the pandemic. For example, both the NMHC (2020) and Urban Institute (2021) data show modest, single-digit percentage point declines in the share of units paying rent in full during the pandemic. In a national survey of landlords, Decker (2021b) reports nearly a third of owners collected less than 90 percent of their 2019 rent in 2020. And finally, using consumer banking data from core US cities, Greig et al. (2021) show that landlords’ year-over-year rental revenues fell by as much as 20 percent during the early months of the pandemic. In each case, there are important differences in the instruments, methods, and populations studied,20 but given the lack of data on this topic, these results provide important context for our findings. Given that the Los Angeles and Philadelphia survey participants had at least one tenant who applied for local ERA, we may be concerned that this selection is mechanically biasing downwards our results for rent collection (in 2020 in particular). We offer several pieces of evidence to suggest this is not the case. First, Appendix Figure 1 presents a version of Figure 1 that excludes both Los Angeles and Philadelphia from the sample; rental payment rates for both 2019 and 2020 are nearly identical when including or excluding these cities from the analysis. Second, despite higher rates of tenant ERA participation in these cities (roughly 60 percent), nearly one-quarter of landlords in the other cities sampled based on rental registries also indicated they had tenants who participated in ERA during the pandemic.21 Finally, in Appendix Figure 2, we present rental collection results solely among landlords with at least one tenant participating in emergency rental assistance, separately for the ERA cities of Los Angeles and Philadelphia (Panel A) and the rental registry cities (Panel B). While we observe modest pre-pandemic variation in the share of landlords collecting 100 vs. 90-99 percent of rental revenue, findings are qualitatively similar across the two samples. Moreover, in 2020, landlords’ mean collection rates by rental revenue category are virtually identical in ERA and rental registry cities. Thus, we conclude that differences among the ERA and rental registry samples are not substantially biasing our results for landlords’ rental collection. 3.2 Non-white and small landlords collect less in rent relative to their white and larger landlord counterparts while HCVs generally insulate landlords against rental non-payment It is widely understood that different types of landlords own different portions of the rental housing stock. For example, one- to four-unit rental properties (SFRs), which tend to have lower median rents relative to multifamily properties, are owned primarily by small, mom-and-pop landlords (Freddie Mac 2018). Institutional investors, on the other hand, own most of the nation's multifamily apartment buildings (DeSilver 2021). With the newer, higher-end units of multifamily buildings naturally attracting higher-income individuals—who are disproportionately white—this implies that different types of landlords cater to different segments of the rental market. What is less understood is the degree to which these landlord characteristics (and others) are correlated not only with rent charged, but also with rent collected. To shed light on this issue, we estimate the following OLS regression, separately for both 2019 and 2020:(1) Renticr,y=β1Demosic+β2Businessic+β3Portfolioic+γc+εic. Renticr,y is a binary indicator for whether landlord i in city c collected less than r percent of their charged yearly rent in year y, where r∈{90,50}. We run separate regressions for each combination of r and y. Demosic is a vector of baseline demographic characteristics, Businessic is a vector baseline business characteristics, and Portfolioic is a vector of mutually exclusive and exhaustive indicators for landlord portfolio size. We also include missing indicators for landlord characteristics and city fixed effects (γc) to control for the time-invariant characteristics of sample cities. We report coefficients from (1) in Figure 2 . Panel A presents coefficients when the dependent variable is either Rent90,2019 (dark gray) or Rent90,2020 (light gray) and describes the relationship between landlord characteristics and their exposure to general rental non-payment, defined as collecting less than 90 percent of their charged yearly rent.22 Panel B instead explores the relationship between these characteristics and landlords’ exposure to severe rental non-payment, defined as collecting less than 50 percent of their charged yearly rent.23 Figure 2 Predictors of Landlords’ Rental Collection.Notes: This figure plots coefficients from regressions of rental collection indicators on key landlord baseline variables, missing indicators, and city fixed effects, separately for 2019 and 2020. In Panel (A), the dependent variable is an indicator for receiving less than 90 percent of charged yearly rent. In Panel (B), the dependent variable is an indicator for receiving less than 50 percent of charged yearly rent. Each regression additionally controls for missing indicators for landlord characteristics and city fixed effects. The number of survey respondents in the sample is 2,548. Heteroskedastic-robust confidence intervals are reported. Data come from the COVID-19 Landlord Survey. Figure 2: Prior to the pandemic, male landlords were no more or less likely than female landlords to collect less than 90 percent of charged rent (conditional on all other covariates). However, landlords over the age of 60 and non-white landlords were around 3 and 7 percentage points more likely, respectively, to face exposure to non-payment compared to younger and white landlords.24 These findings—which represent 27 and 64 percent increases from the unconditional Rent90,2019 mean, respectively—are consistent with the notion that a disproportionate number of older landlords and landlords of color house the nation's most economically vulnerable renters (e.g., Choi & Young 2020). Along these lines, Panel B shows that landlords of color were also 5 percentage points more likely to collect less than 50 percent of charged yearly rent prior to the pandemic—nearly double the unconditional Rent50,2019 mean, and the only significant predictor of severe rental non-payment among the group of demographic variables. In 2020, the relationships between landlords’ baseline demographics and rent collection were largely maintained. Notably, landlords of color continued to experience both general and severe rental non-payment compared to their white counterparts. Interestingly, we find no evidence that landlords who manage their properties themselves or own SFRs—two typical markers of mom-and-pop landlords—had different pre- or post-pandemic rental collection rates relative to landlords who rely on property managers or own multifamily buildings after conditioning on additional demographic and business characteristics. This was not the case for individual owners, though, who faced higher pre-pandemic exposure to severe rental non-payment, in particular, relative to institutional ones. Further, the relationship between individual ownership and general rental non-payment was greatly attenuated in 2020, with individual landlords 13 percentage points less likely to experience rental non-payment relative to institutional ones. Next, we explore the relationship between landlords’ acceptance of housing choice vouchers and yearly rent collection. An estimated 2.3 million low-income, primarily person-of-color-headed US renter households rely on HCVs each month to make rent (Reina, Aiken, & Epstein 2021). On the one hand, vouchers guarantee that landlords will receive at least a portion of their monthly rental revenue. On the other, the economic vulnerability of voucher users makes them particularly susceptible to economic downturns, which may hinder their ability to cover the remainder of rent. We find no evidence that landlords who accept HCVs were more likely to face pre-pandemic exposure to general or severe rental non-payment relative to non-voucher landlords. In 2020, however, HCV landlords were 8 percentage points more likely to report collecting less than 90 percent of charged rent, which is perhaps not surprising given the pandemic's disproportionate impact on individuals with similar characteristics to voucher users (Greene and McCargo 2020). At the same time, Panel B shows that these landlords were no more likely to have tenants fall into deep arrears in 2020. Thus, we find that HCVs not only appear to stave off non-payment of rent under typical conditions, they also appear to have insulated landlords from significant arrearage during the COVID crisis.25 Finally, we explore the relationship between landlords’ portfolio size and rental collection. As mentioned above, a potential driver of observed differences in pandemic rental collection rates across studies is the populations studied. The NMHC (2020) data, which show modest declines in landlords’ pandemic-era rental revenue, cover large, professionally managed organizations. Others who have studied exclusively smaller, mom-and-pop landlords (e.g., de la Campa 2021) have reported much larger declines relative to these data. Though two notable studies have focused on individual markets to provide important context to owners’ exposure to losses (Reina & Goldstein 2021; Reina et al. 2020a), there are no studies we are aware of that have deployed a common survey instrument across multiple cities and multiple types of property owners to understand the issue of rental non-payment. Prior to the pandemic, small (1-5 units owned) and mid-sized (6-19 units owned) landlords were about 5 percentage points more likely to collect less than 90 percent of yearly rent compared to large ones (20+ units owned). Smaller landlords were also significantly more likely (3 percentage points) to face exposure to severe rental non-payment prior to the pandemic. As mentioned above, these pre-pandemic differences in rental collection rates likely reflect the different segments of the market these landlords serve. In contrast, small landlords were significantly less likely to face exposure to general rental delinquency in 2020 relative to large ones (10 percentage points). A relatively larger share of mid-sized landlords also reported collecting less than 90 percent of rent during the pandemic, but this proportion is not significantly different from that of small or large owners. These findings likely reflect the fact that, as the number of rental units in one's portfolio increases, so too does the chance of at least one unit falling behind on rent. At the same time, Panel B shows that larger landlords were the group least likely to struggle with severe rental non-payment during the pandemic. The finding for small landlords is particularly striking, as the 7 percentage point increase in the share of landlords collecting less than 50 percent of yearly rent represents a nearly 80 percent increase of the unconditional mean. 3.3 Landlords’ collection rates were down across rental markets While no region of the US has been spared by the COVID-19 pandemic, there has been significant variation in the timing and intensity of the crisis (e.g., Shrawder & Aguilar 2020). Accordingly, in Figure 3 , we present the share of landlords with tenants in rent arrears separately for each city in our study, for both 2019 and 2020.Figure 3 Landlords’ Rental Collection, by City.Notes: This figure plots the raw share of landlords reporting less than 90 percent (panel A) and 50 percent (panel B) of total rent received in 2019 and 2020, by city. Rent received is expressed as a percentage of total rent charged, in a given year, for a landlord's rental portfolio. 10.5 percent of respondents are from San Jose, 8.6 from Los Angeles, 23.3 from Minneapolis, 6.2 from Racine, 16.2 from Indianapolis, 9.3 from Akron, 6.4 from Rochester, 4.2 from Albany, 10.9 from Philadelphia, and 4.5 from Trenton. The total number of survey respondents in the sample is 2,548. Data come from the COVID-19 Landlord Survey. Figure 3: Panel A shows considerable heterogeneity across cities in the share of landlords who were owed 10 percent or more of charged rent by the end of 2019—from a low of 6 percent of Minneapolis to a high of 18 percent in Rochester and Trenton. In general, we find that landlords in the Upper Midwestern cities of Minneapolis and Racine collected the most rent pre-COVID; those in the Industrial Midwestern cities of Indianapolis and Akron as well as the West Coast cities of San Jose and Los Angeles collected slightly less; and those in the East Coast cities of Rochester, Albany, Philadelphia, and Trenton collected the least. In each city, however, we observe a consistent three- to fourfold increase from 2019 to 2020 in the share of landlords owed 10 percent or more of charged rent. These findings support the notion that the pandemic has had a significant impact on the rental business of landlords across a variety of rental markets and political contexts and underscore the importance of looking at relative changes when examining the impact of COVID-19 on rental markets.26 Results are generally consistent when examining year-over-year changes in the percent of landlords reporting less than 50 percent of rental revenue received (Panel B). Once again, a higher share of landlords in the coastal cities of our study reported facing financial difficulty with their rental properties prior to the pandemic, and the share of landlords collecting less than 50 percent of charged rent in 2020 was up significantly across all rental markets. In contrast to Panel A, however, the year-over-year increase in severe rental non-payment was steeper in the coastal cities. Two potential explanations for this finding may be because pandemic unemployment rates were higher in the coastal cities of our sample relative to the Midwestern ones (Chetty et al. 2020a), and renters were more likely to be cost-burdened in these regions prior to the pandemic (JCHS 2019). 3.4 Renters in socially and economically vulnerable communities are further behind on rent Emerging research has shown that, across a variety of domains, Black, Hispanic, and low-income Americans have disproportionately borne the impact of the COVID-19 pandemic. This has been true not only in terms of exposure to the virus (Reitsma et al. 2021; Zelner et al. 2021) and job loss (Lee et al. 2021), but also in other less obvious contexts, such as access to remote education (Bacher-Hicks et al. 2021). Studies have also found that these more socially and economically vulnerable groups were further behind on rent in 2020 compared to higher-income and white Americans (Airgood-Obrycki et al. 2021). While our survey did not ask landlords to report the race of their tenants, because we asked landlords about their experiences with a single property in their portfolio, we can explore heterogeneity in property-level collection rates according to the neighborhood demographics associated with a property's address. Accordingly, in the following section, we change the unit of analysis from the landlord to the landlord-owned rental property and explore whether certain types of properties and communities, if any, were more likely to fall behind on rent.27 In Figure 4 , we explore yearly changes in property-level rent collection rates separately for properties in neighborhoods with a majority of non-white residents with those where white residents are the majority.28 To construct this figure, we first demean the rental payment and majority non-white-neighborhood indicators by city, and then add back the mean of each variable to its demeaned value to aid in interpretability. In so doing, we control for inter-city differences in rental payment and neighborhood racial composition which might affect the pooled analysis of the relationship between these two variables.Figure 4 Landlords’ Property-Level Rental Collection Rates, by Neighborhood Share of Residents of Color.Notes: This figure plots, for 2019 and 2020, the share of landlords reporting less than 90 percent of total rent received at an individual rental property (Panel A) and less than 50 percent of total rent received at an individual rental property (Panel B), according to the neighborhood share of residents of color for that property. A neighborhood's share of residents of color is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. Properties are classified as “Majority ROC” if they are located in a neighborhood with a majority (over 50 percent) of residents of color. Neighborhoods are classified according to census block groups (CBGs). 53.0 percent of properties are located in a neighborhood with a majority of residents of color. See Appendix Table 1 for each city's racial and ethnic composition. Models include city fixed effects. The number of rental properties in the sample is 2,513. Heteroskedastic-robust confidence intervals are reported. Data come from the COVID-19 Landlord Survey and 2018 ACS. Figure 4: Figure 5 Landlords’ Rental Business Practices Prior to and During the Pandemic.Notes: This figure plots landlords’ rental business practices in 2019 and 2020. “Grant Rent Exten.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fees” indicates charging fees for late rent. “Increase Rents” indicates increases to monthly rents. “Decrease Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Props. For Sale” indicates one or more properties were listed for sale. Responses do not sum to 1 because landlords could choose multiple actions. The number of survey respondents in the sample is 2,525. Data come from the COVID-19 Landlord Survey Figure 5: In both neighborhoods with a majority of non-white residents and those with mostly white residents, the share of rental properties behind 10 percent or more on rent roughly tripled from 2019 to 2020 (Panel A). However, the proportion of properties behind on rent in 2020 was significantly larger in neighborhoods of color neighborhoods compared to those with relatively more white residents (38 vs. 28 percent). Correspondingly, Panel B shows that these properties were also more likely to be behind 50 percent or more on rent by the end of 2020 compared to properties in communities with a majority of white residents (14 vs. 8 percent). This basic pattern is also observed when examining changes in rental payment rates according to median neighborhood income (Appendix Figure X). 29 4 Landlords’ Business Practices Before and After the Pandemic 4.1 Landlords have changed their business practices during the pandemic In Figure 6 , we explore year-over-year changes in landlords’ portfolio-level rent collection, tenant, and ownership policies. Landlord actions are shown on the x-axis, while the percent of landlords who reported taking these actions is displayed on the y-axis. Results will not sum to 1 because landlords could report taking multiple steps to manage their rental property portfolio.Figure 6 Landlords’ 2020 Property-Level Business Practices (y-axis) and Neighborhood Share of Non-White Residents (x-axis).Notes: This figure presents binned scatter plots of landlords’ 2020 property-level rental business practices versus the neighborhood share of non-white residents (for the property). For each plot, the share of landlords reporting the practice is reported on the y-axis while the neighborhood share of non-white residents is presented on the x-axis. A neighborhood's share of non-white residents is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. To construct these plots, we first demean both landlords’ rental business practices and neighborhood share of non-white residents by city, average rent collection, and neighborhood population. We then divide the observations into 20 equal-sized groups (vigintiles) based on neighborhood racial composition and plot the share of landlords pursuing the indicated rental practice within each bin. The solid lines show the best linear fit estimated on the underlying micro data using OLS regression, and Table 8 presents these regression estimates. “Grant Rent Exten.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fees” indicates charging fees for late rent. “Increase Rents” indicates increases to monthly rents. “Decrease Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. The number of rental properties for each plot is 2,402. Data come from the COVID-19 Landlord Survey and 2018 ACS Figure 6: Overall, the pandemic is associated with a sharp increase in certain types of actions and a decrease in others.30 For example, 15 percent of landlords reported granting rental extensions to at least one of their tenants in 2019 compared to nearly half the following year. These findings are particularly noteworthy for two reasons. First, while previous qualitative work has indicated that, under typical conditions, landlords rely on rental payment plans to avoid costly evictions (Balzarini & Boyd 2020; Decker 2021a), our study is the first of which we are aware to quantify landlords’ pursuance of this strategy. Second, the marked increase in this tactic in 2020 aligns with numerous anecdotal accounts of landlords—particularly small, mom-and-pop owners—working with tenants to find solutions to rental non-payment during the pandemic (e.g., Arnold 2020). Though charging tenants late rent fees and increasing rents were the two most common actions reported by landlords prior to the pandemic, in 2020, the prevalence of these actions fell by 12 and 9 percentage points, respectively. While these statistics are striking given the lower amount of rent, on average, landlords collected during the pandemic, part of these declines are likely due to pandemic-induced, intermittent prohibitions on these practices in our study cities (Raifman et al. 2020). Perhaps more salient to landlords than the above restrictions were the eviction moratoria put in place at the local, state, and federal level. Indeed, recent research has estimated 1.5 million evictions were prevented during the pandemic due to eviction bans (Hepburn et al. 2021). Nonetheless, the share of landlords who brought eviction proceedings against at least one tenant is nearly identical for both 2019 and 2020 (15 percent). While this implies that the eviction rate conditional on not receiving rent in full was lower in 2020 than in 2019 (23.2 versus 29.4 percent), it may still be surprising that an equivalent share of landlords in 2019 and 2020 indicated that they had brought eviction proceedings against at least one tenant. However, we offer two potential reasons this may be the case. First, our survey asked landlords about the initiation of eviction proceedings rather than their conclusion. Second, despite the aforementioned reduction in evictions, an estimated 1.1 million tenants were evicted in 2020, and it may be the case that landlords who moved forward with evictions during the pandemic—which were relatively more difficult to execute—are those more familiar with the eviction system (e.g., Rutan and Desmond 2021). While our study cannot speak to this phenomenon more broadly, 45 percent of the landlords in our study who brought eviction proceedings against at least one tenant in 2020 did so in 2019 as well.31 On the one hand, this reinforces the growing body of literature which shows that a large share of evictions are often concentrated among a small pool of owners (e.g., McCabe & Rosen 2020), but it also demonstrates how the pandemic may have temporarily expanded the pool of owners looking to use this practice. We also find that some landlord practices that were relatively uncommon prior to the pandemic became widespread in 2020. Around one-fifth of landlords reported forgiving outstanding rent; decreasing rents; and missing at least one mortgage, utility, and/or property tax payment in 2020—in each case, a roughly 15 percentage point increase from the prior year. The share of landlords missing mortgage payments in 2020—nearly one in ten (not shown)—is particularly troubling as it calls into question the future financial viability of these properties.32 Adding additional strain, the share of landlords who reported delaying repairs increased from 5 percent in 2019 to 31 percent in 2020. Ultimately, 13 percent of landlords took steps to sell one or more rental properties in 2020 compared to only 3 percent the prior year. The steps landlords have taken to disinvest from their properties—via delayed payments, deferred maintenance, and property listings—in many ways mirror the actions of other small business owners during the pandemic. Bartik et al. (2020) found that three-quarters of small business owners surveyed at the beginning of 2020 had two months or fewer of cash on hand, and as a result, reported pursuing cost-cutting measures such as reducing staffing and temporarily closing. Along these lines, despite the gradual reopening of the economy in the second half of 2020, Crane et al. (2021) report a 3 percentage point year-over-year increase in the small businesses closure rate for retail and service establishments. While this latter finding is markedly lower than the 10 percentage point year-over-year increase in property listing rates observed in our study, there are crucial differences between these findings. First, we asked landlords solely whether any properties had been listed for sale, not necessarily sold. Second, property sales may generate a financial return, which likely makes them a more attractive option relative to small business closures. However, we note that the high rates of deferred maintenance and missed mortgage payments in our sample likely affects the viability of these potential sales, and in either scenario, there are direct and indirect implications for the renters in and future affordability of those properties. 4.2 Decreased rent collection cannot fully explain landlords’ changing rental business practices It may be the case that changing rental business practices during the pandemic are a reflection of landlords’ decreased rental collection, as observed in Figure 1. To further explore this possibility, we estimate the following OLS regression:(2) Practiceicyp=β1RentLT90icy+β22020y+β3RentLT90icy*2020y+γc+εicy. Practiceicyp is an indicator for whether landlord i in city c and year y implemented rental business practice p. We estimate Equation (2) separately for each of the nine rental business practices p reported in Figure 2. RentLT90icy indicates whether landlord i in city c collected at most 90 percent of their rental revenue in year y, and 2020y is an indicator for the 2020 (i.e., post-COVID) time period. As in (1), we include city fixed effects (γc) to control for the time-invariant characteristics of the cities in our sample. Table 4 presents results from Equation (2), with heteroskedastic-robust standard errors reported in parentheses. Coefficient β1 captures the relationship, in 2019, between rental non-payment and business practice p. Apart from listing one's properties for sale, prior to the pandemic, the intensity with which landlords pursued their rental business practices was highly correlated with yearly rental collection. For example, column (4) shows that collecting at most 90 percent of 2019 rent was associated with a 12.3 percentage point decrease in the share of landlords’ increasing tenants’ rents (in that year). Conversely, relative to collecting 90 percent or more of yearly rental revenue, partial collection is associated with a 13.7 percentage point increase in landlords’ eviction initiation rate (column 6).Table 4 Relationship between Rental Collection and Business Practices Table 4: Grant Rent Ext. Forgive Rent Charge Rent Fee Inc. Rents Dec. Rents Evict Tenants Miss Payments Defer Maint. List Props. for Sale (1) (2) (3) (4) (5) (6) (7) (8) (9) < 90% Rent Received 0.083*** 0.037** 0.064** -0.123*** 0.052*** 0.137*** 0.066*** 0.088*** 0.019 (0.028) (0.016) (0.030) (0.023) (0.015) (0.028) (0.020) (0.022) (0.014) 2020 0.226*** 0.118*** -0.144*** -0.215*** 0.115*** -0.065*** 0.053*** 0.164*** 0.058*** (0.014) (0.010) (0.011) (0.011) (0.009) (0.009) (0.008) (0.011) (0.008) < 90% Rent Received*2020 0.234*** 0.138*** 0.029 0.102*** 0.057** 0.099*** 0.209*** 0.181*** 0.106*** (0.034) (0.023) (0.033) (0.025) (0.022) (0.032) (0.026) (0.029) (0.021) N Landlord-Years 4,808 4,808 4,808 4,808 4,808 4,808 4,808 4,808 4,808 Notes: This table reports OLS estimates of the relationship between landlords’ business practices and rental collection, prior to and during the pandemic. Each column presents results from a separate OLS regression, where the indicated business practice is the dependent variable. “Grant Rent Ext.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fee” indicates charging fees for late rent. “Inc. Rents” indicates increases to monthly rents. “Dec. Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maint.” indicates delayed property repairs or maintenance. “List Props. For Sale” indicates one or more properties were listed for sale. Landlords could choose multiple actions. Models include city fixed effects. Heteroskedastic-robust standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Data come from the COVID-19 Landlord Survey. The coefficient β2 instead reports the effect of the pandemic on landlords’ rental business practices solely among those who received 90 percent or more of their rental revenue. Even for this group of landlords, the pandemic has had an impact on nearly every rental business practice, with particularly large increases (relative to 2019) in the share of landlords granting rent extensions (22.6 percentage points), forgiving rent (11.8 percentage points), and deferring property maintenance (16.4 percentage points). There have also been steep decreases in the share of landlords charging late rent fees (14.4 percentage points), increasing rents (21.5 percentage points), and evicting tenants (6.5 percentage points). Taken together, these findings imply that the observed changes from 2019 to 2020 in landlords’ business practices were not driven exclusively by decreased rent collection from the pandemic and likely reflect a variety of other factors including local policies and restrictions (e.g., eviction moratoria), weakened demand in the rental market, COVID-related limitations on building access, and supply-side challenges for maintenance and repair. Finally, the coefficient on the interaction term (β3) sheds light on how the relationship between rental payment and business practices has changed, if at all, in 2020 (post-COVID) compared to 2019 (pre-COVID). In addition to shifting the levels of nearly all business practices in 2020, the pandemic has also intensified the rate at which landlords have taken certain actions conditional on partial rent payment. This is particularly evident for the implementation of rental payment plans. In 2019, collecting at most 90 percent of rental revenue was associated with an 8.3 percentage point increase in landlords’ implementation of rental payment plans; during the pandemic, the strength of that relationship roughly tripled, such that partial payment was associated with a 31.7 percentage point increase in this business practice (β1+β3). The amplification of this relationship may be a result of the restrictions placed on landlords’ traditional responses to rental non-payment—such as late fees and evictions—during the pandemic (e.g., Raifman et al. 2020).33 Indeed, in 2020, there was no significant relationship between partial rental payment and the implementation of late rent fees, and that between rental payment and evictions was significantly weakened. Other actions that have been significantly altered during the pandemic are those related to property ownership, such as missing financial payment obligations, deferring property maintenance, and listing one's properties for sale. For example, while there was no statistically significant relationship between rental non-payment and property listings in 2019, collecting at most 90 percent of rental revenue in 2020 was associated with a 12.5 percentage point increase in the probability of trying to sell one's property. Combined, the results in column (9) show that, while all landlords became interested in selling their rental properties during the pandemic, this was particularly true for those who collected at most 90 percent of their rental revenue.34 4.3 Landlords’ business practices vary according to their portfolio and demographic characteristics even after accounting for group-level differences in rent collection The results in Table 4 suggest that landlords’ business practices are highly correlated with rent collection, both prior to and during the pandemic. At the same time, Figure 2 shows that landlords’ rental collection rates vary according to salient business and demographic characteristics. Accordingly, in Table 5 we examine yearly variation in six key business practices according to the size of landlords’ rental property portfolios (Panel A) and self-reported race (Panel B), conditional on rent collection. Specifically, we present results from OLS regressions of landlords’ business practices on indicators for landlord characteristics, the interaction of these characteristics with the 2020 indicator, and a categorical variable of rental payment. For reference, in each panel, we report the unconditional mean for the omitted group.35 Table 5 Relationship between Landlord Characteristics and Six Key Business Practices Table 5: Grant Rent Ext. Charge Rent Fee Evict Tenants Miss Payments Defer Maint. List Props. for Sale (1) (2) (3) (4) (5) (6) Panel A: Portfolio Size Small Landlord -0.104*** -0.298*** -0.325*** -0.001 -0.004 -0.051*** (0.028) (0.033) (0.031) (0.013) (0.014) (0.016) Mid-Sized Landlord -0.068** -0.136*** -0.166*** -0.007 0.009 -0.047*** (0.032) (0.038) (0.035) (0.015) (0.018) (0.017) 2020 0.462*** -0.311*** -0.096** 0.119*** 0.290*** 0.173*** (0.037) (0.040) (0.041) (0.027) (0.033) (0.031) Small*2020 -0.241*** 0.223*** 0.071* -0.042 -0.120*** -0.111*** (0.039) (0.041) (0.042) (0.028) (0.035) (0.032) Mid-Sized*2020 -0.097** 0.092* 0.014 0.026 -0.032 -0.078** (0.046) (0.048) (0.048) (0.034) (0.041) (0.036) 50-89% Rent Received 0.249*** 0.089*** 0.182*** 0.161*** 0.198*** 0.084*** (0.018) (0.016) (0.016) (0.015) (0.018) (0.013) <50% Rent Received 0.156*** 0.077*** 0.245*** 0.371*** 0.244*** 0.092*** (0.029) (0.024) (0.028) (0.030) (0.029) (0.022) Omitted Group Mean 0.24 0.46 0.38 0.03 0.05 0.07 N Landlord-Years 4,764 4,764 4,764 4,764 4,764 4,764 Panel B: Race Landlord of Color -0.018 0.027 -0.004 0.013 -0.010 -0.030*** (0.018) (0.021) (0.018) (0.011) (0.012) (0.008) 2020 0.223*** -0.152*** -0.067*** 0.056*** 0.166*** 0.075*** (0.016) (0.014) (0.012) (0.009) (0.013) (0.010) Landlord of Color*2020 0.068** -0.033 0.011 0.084*** 0.059** 0.001 (0.028) (0.025) (0.023) (0.021) (0.024) (0.017) 50-89% Rent Received 0.292*** 0.101*** 0.200*** 0.171*** 0.211*** 0.108*** (0.020) (0.017) (0.017) (0.016) (0.019) (0.015) <50% Rent Received 0.141*** 0.072*** 0.229*** 0.371*** 0.240*** 0.102*** (0.033) (0.027) (0.031) (0.033) (0.032) (0.025) Omitted Group Mean 0.15 0.24 0.14 0.03 0.05 0.04 N Landlord-Years 4,180 4,180 4,180 4,180 4,180 4,180 Notes: This table reports OLS estimates of the relationship between landlords’ business practices and demographic/business characteristics, prior to and during the pandemic. Panel A presents results according to landlords’ portfolio size, while Panel B presents results according to landlords’ race/ethnicity. Each column presents results from a separate OLS regression, where the indicated business practice is the dependent variable. “Grant Rent Ext.” indicates rental extensions and/or putting tenants on repayment plans. “Charge Rent Fee” indicates charging fees for late rent. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maint.” indicates delayed property repairs or maintenance. “List Props. For Sale” indicates one or more properties were listed for sale. Landlords could choose multiple actions. 70.6 percent of respondents are small landlords, and 18.3 percent are mid-sized landlords. 32.3 percent of respondents are landlords of color. Models additionally control for city fixed effects. Heteroskedastic-robust standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Data come from the COVID-19 Landlord Survey. In Panel A, the coefficients on the Small Landlord and Mid-Sized Landlord variables report the relationship between landlord size and business practices for those who collected 90 percent or more of their 2019 yearly rent (relative to larger landlords with comparable rental collection rates). In general, larger landlords, conditional on rental payment, were more likely to report pursuing each of the listed business practices compared to small and mid-sized ones. For example, landlords who collected 90 percent or more of their yearly rent and who have five or fewer units in their portfolio were 29.8 percentage points less likely to charge late rent fees, 32.5 percentage points less likely to evict tenants, and 5.1 percentage points less likely to list a property for sale compared to landlords with the same rental collection rates and 20 or more units in their portfolio. The coefficient on the 2020 variable captures the level shift across years in large landlords’ business practices, conditional on rent collection. Large landlords exhibited considerable adaptability of their business practices—both upwards and downwards—during the pandemic. In fact, when compared to the impact of rent collection on 2019 business practices, large landlords’ behavioral change during the pandemic was significant: the 46.2 percentage point increase in rental extension rate associated with the pandemic was nearly triple that associated with collecting less than 50 percent of yearly rent (for the same business practice). Large landlords were also about twice as likely to list properties for sale in 2020 relative to pursuing this action in response to rental non-payment prior to the pandemic, which may imply that the pandemic has sent landlords’ negative signals about the rental market above and beyond those sent by rental non-payment. The coefficients on the interaction terms show that small and mid-sized landlords demonstrated relatively less behavioral change relative to large ones, though on balance, these groups still adjusted their business practices during the pandemic. In certain instances, these behavioral changes are even larger than large landlords’ pre-pandemic responses to rental non-payment—a finding that is notable given the generally more active role large landlords take in managing their rental properties. For example, column (5) shows that property sale listings increased by 17.3 percentage points in 2020 for large landlords, leading to net increases of 6.1 and 9.5 percentage points for small and mid-sized landlords, respectively—the latter being larger than the coefficients on both 50-89% Rent Received and <50% Rent Received. Panel B explores the relationship between landlord race, business practices, and rental collection. Apart from property listings, there were few meaningful differences in rental practices between landlords of color and white landlords prior to the pandemic. However, the fact that landlords of color were significantly more likely than white landlords to hold onto their rental properties is notable in light of emerging research documenting discrimination faced by Black and Latino sellers in the home sale market (Freddie Mac 2021). Once again, we observe a sharp change in business practices across years for all landlords. Contrary to our findings in Panel A, however, the change in business practices associated with the pandemic are generally of a smaller magnitude than the changes associated with rental non-payment prior to the pandemic for white and non-white landlords alike. However, landlords of color were significantly more likely than their white counterparts to grant rental extensions (6.8 percentage points), defer maintenance (5.9 percentage points), and miss property payments (8.4 percentage points) during the pandemic, holding constant rental collection. While there are various potential explanations for these findings, the latter, in particular, may reflect the fact that Black and Hispanic owners were significantly less likely than their white counterparts to request mortgage forbearance during the pandemic (Gerardi, Lambie-Hanson, & Willen 2021). In sum, we find notable differences in the pre- and post-pandemic business practices of small vs. large and non-white vs. white landlords conditional on rent collection differences across groups. Both small and non-white landlords took significant steps to disinvest from their properties during the pandemic—in the case of non-white landlords, disproportionately so relative to their white counterparts—by deferring maintenance and listing rental properties for sale. Given the critical role both groups of landlords play in housing lower-income renters and renters of color, their tendency to pursue these practices likely has the unintended consequence of perpetuating financial and housing instability for vulnerable tenants and property owners alike. 4.4 Regional variation in pandemic policies contributed to observed cross-city variation in landlord practices Table 6 explores cross-city, cross-year heterogeneity among the six key landlord business practices. Each cell of this table reports the share of landlords in city c (row) pursuing business practice p (column), for year y (sub-column). Column (1) shows that, prior to the pandemic, a substantial share of landlords in each city reported granting rental extensions—from a low of around 10 percent in Los Angeles to a high of nearly 20 percent in Albany. In 2020, these proportions increased by 20 to 50 percentage points (column 2), with the most significant increases concentrated among cities where landlords collected less rent during the pandemic.36 Columns (3) and (4) show an opposite trend for rental fees: though they were common in all cities prior to the pandemic, landlords in each city reported significantly lower rates of this rental business practice in 2020.Table 6 Changes in Six Key Landlord Rental Business Practices, by City Table 6: Grant Rent Extensions Charge Rent Fees Evict Tenants Miss Payments Defer Maintenance List Props. For Sale 2019 2020 2019 2020 2019 2020 2019 2020 2019 2020 2019 2020 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Akron 17.5 49.3 24.5 20.1 19.2 17.9 7.4 21 3.9 29.7 2.2 16.6 Albany 19.6 45.1 27.5 9.8 17.6 21.6 3.9 26.5 9.8 46.1 3.9 22.5 Indianapolis 18.1 53.1 26.0 17.9 17.3 19.1 1.8 9.7 3.6 25.8 4.1 15.3 Los Angeles 9.5 59.5 26.5 4.5 8.5 6.5 2.5 20.5 6.5 35.0 1.0 11.5 Minneapolis 10.2 29.2 14.5 6.6 5.5 3.0 1.6 10.6 2.5 24.2 2.7 8.2 Philadelphia 19.0 68.6 29.1 15.9 17.8 25.6 5.8 32.2 5.4 35.7 3.5 20.9 Racine 13.9 43.8 20.1 9.0 11.1 9.7 2.8 17.4 9.7 24.3 2.1 5.6 Rochester 14.6 53.2 22.8 12.7 25.3 31.6 6.3 33.5 5.1 43.0 3.8 17.7 San Jose 11.4 50.4 26.0 3.5 11.4 5.5 3.1 12.2 5.9 32.7 1.2 6.7 Trenton 22.2 57.4 24.1 17.6 20.4 38.9 5.6 38 9.3 36.1 2.8 15.7 Notes: This table reports the share of landlords’ pursuing five key rental business practices in 2019 and 2020, for each city in the study. “Grant Rent Extensions” indicates rental extensions and/or putting tenants on repayment plans. “Charge Rent Fees” indicates charging fees for late rent. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Props. For Sale” indicates one or more properties were listed for sale. Responses do not sum to 100 (within a city-year) because landlords could choose multiple actions. 10.5 percent of respondents are from San Jose, 8.6 from Los Angeles, 23.3 from Minneapolis, 6.2 from Racine, 16.2 from Indianapolis, 9.3 from Akron, 6.4 from Rochester, 4.2 from Albany, 10.9 from Philadelphia, and 4.5 from Trenton. The total number of survey respondents in the sample is 2,525. Data come from the COVID-19 Landlord Survey. Results are more mixed for evictions (columns 5 and 6). The proportion of landlords initiating eviction proceedings fell by several percentage points for half the cities in our sample—specifically, the West Coast cities of Los Angeles and San Jose and the Midwestern cities of Minneapolis, Racine, and Akron. At the same time, evictions were up slightly in all East Coast cities and the Midwestern city of Indianapolis. While this finding, like that for the granting of rental extensions, may in part be explained by cross-city variation in landlords’ rental collection rates, it may also be a function of the differing intensities and duration of renter protections for the cities in our sample.37 Fewer than 10 percent of landlords in any of our sample's cities reported missing mortgage, utility, and/or property tax payments (column 7) or deferring maintenance (column 9) at one or more of their rental properties prior to the pandemic. The share of landlords pursuing each of these actions increased dramatically in 2020. In each city, at least 10 percent of landlords reported missing financial obligations in 2020, with rates particularly high in the East Coast cities of Rochester, Philadelphia, and Trenton (column 8). Further, roughly one-quarter of Midwestern, one-third of West Coast, and two-fifths of East Coast landlords indicated they had delayed necessary property upkeep for at least one of their rental properties (column 10). Property sales were even less common prior to the pandemic (column 11), but this action increased dramatically in 2020 (column 12). During the pandemic, over 15 percent of landlords in Akron, Albany, Indianapolis, Philadelphia, Rochester, and Trenton reported listing at least one rental property for sale. As mentioned above, several of the year-over-year changes to landlords’ business practices are most striking in cities where declines to rent collection were most severe. At the same time, variation in local rules, regulations, and politicians’ response to the pandemic (e.g., Raifman et al. 2020) might lead to an independent impact on landlords’ rental businesses, irrespective of rental payment. To better explore this issue, in Table 6, we present weighted OLS estimates from a version of Equation (1) collapsed to the city-year level.38 Estimates from this regression shed light on: 1) the average, pre-pandemic relationship between city-level rent collection and business practices (β1), 2) the average impact of the pandemic on landlords’ rental business practices, conditional on city-level collection rates (β2), and 3) whether the pandemic has, on average, altered the relationship between rental collection and business practice implementation across cities (β3). Column (1) of Table 7 shows that, in 2019, a 1-unit increase in the share of landlords collecting at most 90 percent of rental revenue was associated with a statistically significant 0.58 unit increase, on average, in the share of landlords granting rental extensions. If we assume effects are linear throughout the distribution, this implies a 5.8 percentage point increase in the city-level rental extension rate for a 10 percentage point increase in city-level partial rental revenue collection rate. With coefficients on the pandemic (i.e., 2020) indicator and interaction term statistically indistinguishable from 0, we thus conclude that the primary driver of cross-city differences in landlords’ rental extension rates is indeed cross-city variation in their rental collection rates.Table 7 Relationship between City-Level Rental Collection and Business Practices Table 7: Grant Rent Ext. Charge Rent Fees Evict Tenants Miss Payments Defer Maint. List Props. for Sale (1) (2) (3) (4) (5) (6) Share < 90% Rent Received 0.583** 0.756* 1.173*** 0.421*** 0.236 0.041 (0.208) (0.395) (0.204) (0.065) (0.152) (0.051) 2020 0.072 -0.108* -0.126** -0.037 0.117*** -0.004 (0.068) (0.059) (0.045) (0.058) (0.037) (0.024) Share < 90% Rent Received*2020 0.280 -0.561 -0.483* 0.191 0.198 0.253*** (0.272) (0.400) (0.228) (0.150) (0.174) (0.084) N City-Years 20 20 20 20 20 20 Notes: This table reports weighted OLS estimates of the relationship between landlords’ rental collection and business practice implementation rates, at the city-level, prior to and during the pandemic. To generate these estimates, data are first collapsed on means to the city-year level, and all regressions are weighted by the number of survey respondents within the city. “Grant Rent Ext.” indicates rental extensions and/or putting tenants on repayment plans. “Charge Rent Fees” indicates charging fees for late rent. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maint.” indicates delayed property repairs or maintenance. “List Props. For Sale” indicates one or more properties were listed for sale. Landlords could choose multiple actions. Heteroskedastic-robust standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Data come from the COVID-19 Landlord Survey. This is not the case when examining the relationship between rent collection and the incidence of late rent fees (column 2) and tenant evictions (column 3). While we once again observe a strong, positive relationship in 2019 between a city's share of landlords who collected at most 90 percent of their rental revenue and pursued tenant late rent fees and/or evictions, these proportions fell by 10.8 and 12.6 percentage points, respectively, during the pandemic (holding constant rental collection). We also find suggestive evidence that the pandemic has attenuated the relationship between rent collection and the pursuance of evictions. Taken together, these results imply that cross-city variation beyond that observed in rental collection rates—perhaps arising from different pandemic rules, regulations, and responses across cities, among other factors—contributed to the observed variation in the issuance of rental fees and tenant evictions in 2020. This result is notable in that it indicates local policies and regulations directed towards landlords are likely able to affect behavioral change among this population. Finally, in columns (5) and (6) we see that the pandemic has altered city-level rates of deferred property maintenance and rental property sale listings, albeit in slightly different ways. Prior to the pandemic, there was no robust relationship between landlords’ rental collection and deferred maintenance rates. Though the share of landlords reporting the latter practice was up significantly during the pandemic, there remains no significant relationship between citywide rent collection and deferred maintenance in 2020. For property sales, listings were up more dramatically in 2020 in cities with lower rental collection rates, mirroring heterogeneity in small business closures across regions (Bartik et al. 2020). These responses in particular raise concern about the potential impact of the pandemic on both long-term housing stock quality and affordability. 4.5 Landlords’ responses to the pandemic may be increasing housing instability in communities of color Prior to the pandemic, Black and Hispanic Americans have faced discrimination in the rental housing market in numerous ways—from housing search (Hanson & Hawley 2011; Fang et al. 2019), to securing affordable housing via Section 8 (Cunningham et al. 2018), to evictions (Hepburn, Louis, & Desmond 2020). Given this history, a natural question to pose is: how have landlords managed their rental properties during the pandemic in communities of color, particularly given the relatively higher rates of rental non-payment observed in these communities? In Figure 7 , we again switch our unit of analysis to the individual rental property and explore variation in landlords’ 2020 rental business practices according to a neighborhood's racial composition. Specifically, we present nine binned scatter plots of landlords’ rental property business practices (y-axis) versus the neighborhood share of non-white residents (x-axis). To construct these plots, we first demean both landlords’ rental business practices and the neighborhood share of non-white residents by city, 2020 rent collection, and neighborhood population. We then divide the observations into 20 equal-sized groups (vigintiles) based on the racial distribution and plot the share of landlords pursuing the indicated rental practice within each bin. The solid lines show the best linear fit estimated on the underlying micro data using OLS regression, and Panel A of Table 8 presents these regression estimates.Figure 7 Landlords’ 2019 Property-Level Business Practices (y-axis) and Neighborhood Share of Non-White Residents (x-axis).Notes: This figure presents binned scatter plots of landlords’ 2019 property-level rental business practices versus the neighborhood share of non-white residents (for the property). For each plot, the share of landlords reporting the practice is reported on the y-axis while the neighborhood share of non-white residents is presented on the x-axis. A neighborhood's share of non-white residents is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. To construct these plots, we first demean both landlords’ rental business practices and neighborhood share of non-white residents by city, 2019 rent collection, and neighborhood population. We then divide the observations into 20 equal-sized groups (vigintiles) based on neighborhood racial composition and plot the share of landlords pursuing the indicated rental practice within each bin. The solid lines show the best linear fit estimated on the underlying micro data using OLS regression, and Appendix Table 9 presents these regression estimates. “Grant Rent Exten.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fees” indicates charging fees for late rent. “Increase Rents” indicates increases to monthly rents. “Decrease Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. The number of rental properties for each plot is 2,402. Data come from the COVID-19 Landlord Survey and 2018 ACS. Figure 7: Table 8 Relationship Between Landlords’ 2020 Property-Level Rental Business Practices and Neighborhood Share of Non-White Residents Table 8: Grant Rent Ext. Forgive Rent Charge Rent Fee Inc. Rents Dec. Rents Evict Tenants Miss Payments Defer Maint. List Prop. for Sale (1) (2) (3) (4) (5) (6) (7) (8) (9) Unconditional 2020 Mean 0.368 0.148 0.083 0.037 0.123 0.085 0.137 0.257 0.069 Panel A: Baseline Model Share Non-White Residents 0.069** -0.083*** 0.057*** -0.003 -0.090*** 0.076*** 0.104*** 0.022 -0.005 (0.035) (0.028) (0.021) (0.013) (0.026) (0.020) (0.026) (0.033) (0.021) Panel B: Landlord Controls Share Non-White Residents 0.061* -0.064** 0.063*** 0.009 -0.063** 0.076*** 0.077*** 0.021 0.000 (0.036) (0.029) (0.023) (0.013) (0.027) (0.021) (0.026) (0.034) (0.023) Panel C: Neighborhood Controls Share Non-White Residents 0.073* -0.127*** 0.076*** 0.003 -0.103*** 0.094*** 0.097*** -0.002 -0.023 (0.043) (0.035) (0.028) (0.017) (0.031) (0.026) (0.032) (0.041) (0.029) Panel D: Landlord and Neighborhood Controls Share Non-White Residents 0.070 -0.109*** 0.088*** 0.015 -0.076** 0.099*** 0.072** 0.007 -0.019 (0.044) (0.036) (0.030) (0.018) (0.032) (0.026) (0.033) (0.042) (0.030) N Rental Properties 2,402 2,402 2,402 2,402 2,402 2,402 2,402 2,402 2,402 Notes: This table reports OLS regression estimates of the relationship between landlords’ 2020 property-level rental business practices and the neighborhood share of non-white residents. Each column presents results from a separate OLS regression of a residualized version of the indicated business practice on a residualized version of the neighborhood share of non-white residents. In Panel A, we residualize on 2020 rent collection, neighborhood population, and city fixed effects. In Panel B, we residualize on the controls from Panel A, as well as the set of indicators for landlords’ demographics, business details, and portfolio size reported in Figure 2. In Panel C, we residualize on the controls from Panel A, as well as the neighborhood median income and age distribution. In Panel D, we residualize on all controls from Panels A, B, and C. A neighborhood's share of non-white residents is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. “Grant Rent Ext.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fee” indicates charging fees for late rent. “Inc. Rents” indicates increases to monthly rents. “Dec. Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maint.” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. Landlords could choose multiple actions. Heteroskedastic-robust standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Data come from the COVID-19 Landlord Survey and 2018 ACS. Nearly all of landlords’ practices are statistically significantly related to the neighborhood racial and ethnic composition of their rental properties. Typically, properties in communities with more residents of color are more susceptible to business actions that likely contribute to housing instability. For example, a 1-unit increase in a neighborhood's share of residents of color (i.e., moving from a neighborhood with no residents of color to one with exclusively residents of color) was associated with an 8.3 percentage point reduction in the share of landlords offering rental forgiveness, 9.0 percentage point reduction in the share decreasing monthly rents, and a 5.7 percentage point increase in the share charging late rent fees. Put differently, moving from a neighborhood at the 25th percentile of the city-demeaned resident of color distribution to one at the 75th percentile is associated with a 3.4 percentage point decrease in landlords’ rent forgiveness rate, 3.7 percentage point decrease in their monthly rent decrease rate, and 2.3 percentage point increase in their late fee implementation rate.39 These findings, which indicate relatively greater landlord-induced financial strain for renters of color, are particularly relevant given the pandemic's outsized financial impact in these communities (e.g., Lee et al. 2021). We also observe a disproportionate share of landlords reporting tenant evictions at properties in communities of color. In this case, moving from the 25th to 75th percentile of the city-demeaned racial distribution is associated with a 3.1 percentage point increase in the tenant eviction rate. For context, this is over 35 percent of 2020 unconditional mean eviction rate (8.5 percent). The higher rate of displacement in these communities aligns with emerging research on the unequal rate at which Black and Hispanic renters have been evicted during the pandemic (Hepburn et al. 2021; Stein et al. 2021). While landlords have been more likely to miss financial payments in neighborhoods with more residents of color, there once again is no meaningful relationship between neighborhood racial composition and deferred maintenance or property sales. There are many potential explanations for these results. One is that landlord sorting across neighborhoods according to demographic or business characteristics that are known to be correlated with management tactics—such as landlord race—may be driving observed differences.40 Accordingly, in Appendix Figure 5, we explore the degree to which cross-neighborhood differences in landlords’ characteristics may be contributing to landlords’ business practices in communities of color. Even conditional on the full set of landlord characteristics presented in Figure 2, the relationships between landlords’ 2020 business practices and neighborhood racial composition are qualitatively similar to those reported in Figure 8. More specifically, Panel B of Table 8 shows that the coefficients on the Share Non-White Residents variable are remarkably consistent, albeit slightly attenuated, relative to the baseline model. Thus, we conclude that differences in landlord characteristics across neighborhoods alone cannot explain landlords’ tendency to disproportionately pursue punitive actions in communities of color. Together with the results from Table 5, this implies that, while landlords modified their business practices during the pandemic according to their own demographic and business characteristics, the strength of those relationships did not additionally vary according to a neighborhood's racial composition.41 We also investigate whether residential sorting across neighborhoods, particularly on economic characteristics, may influence landlords’ business practices (Appendix Figure 8). This could occur, for example, if landlords’ expectations about tenants’ ability to pay rent in the future vary according to characteristics that are correlated with tenants’ race. To test this hypothesis, we add controls for neighborhood median income and age to our baseline model.42 We choose these controls for two primary reasons: 1) non-white Americans are younger and have lower median incomes relative to white Americans (Schaeffer 2019, Chetty et al. 2020b), and 2) emerging research has shown a disproportionate economic impact of the pandemic on these groups (e.g., Lee et al. 2021). Panel C of Table 8 shows that, conditional on rental collection and neighborhood characteristics that could be correlated with perceived economic ability, landlords were still significantly more likely to pursue actions such as evictions and significantly less likely to offer rent decreases in communities with a higher share of non-white residents. Including landlord controls in this model (Panel D) again does not meaningfully alter our interpretation of the results. Indeed, the qualitative conclusions across all four models are identical: landlords were more likely to report taking punitive actions and less likely to report offering concessions in communities of color during the pandemic. While we cannot identify the precise cause for these differential behaviors, our robustness checks allow us to rule out the possibility that they are fully explained by the sorting of landlords and/or tenants across neighborhoods.43 Finally, to better understand the degree to which the observed relationships between business practices and neighborhood racial composition are unique to the pandemic, Figure 9 presents binned scatter plots of landlords’ 2019 rental property business practices against neighborhood share of non-white residents. Even prior to the pandemic, landlords’ pursual of punitive business practices was increasing in a neighborhood's share of non-white residents. Moreover, the strength of these relationships was stronger in 2019 relative to 2020: moving from a neighborhood at the 25th percentile of the city-demeaned resident of color distribution to one at the 75th percentile was associated with a 3.9 and 5.1 percentage point increase in rental fees and evictions, respectively (compared to 2.3 and 3.1 percentage points in 2020). Conversely, we find no relationship in 2019 between neighborhood racial composition and concessionary business practices, such as the granting of rental extensions, which were decreasing in the share of non-white residents during the pandemic. Thus, while the relationship between punitive actions and neighborhood racial composition was attenuated in 2020—perhaps due to intermittent prohibitions put on these practices (Raifman et al. 2020)—the racialized nature of landlords’ business practices persisted and even took on new forms during the pandemic. We conclude that landlords’ behavioral change in 2020 was moderate compared to 2019, but crucially, their business responses were racialized in both years.44 In sum, landlords’ tendency to pursue business practices differentially according to neighborhood racial composition—even conditional on rental collection rates, landlord characteristics, and other neighborhood characteristics—has resulted in tenants in more marginalized communities disproportionately bearing the consequences of pandemic-induced rental non-payment. While these racialized business responses are not new, they likely have increased housing instability in communities in which tenants were more already likely to be adversely affected by the pandemic in other ways (Bacher-Hicks et al. 2021; Bambra et al. 2020; Lee et al. 2021). We conclude that landlords’ pre- and post-pandemic business practices ultimately serve to exacerbate and reinforce the many historical rental market discriminations facing renters of color (Hanson & Hawley 2011; Cunningham et al. 2018; Hepburn, Louis, & Desmond 2020). 5 Conclusion In this paper, we explore the impact of the COVID-19 pandemic on the rental business of landlords in ten cities across the US. We find that landlords’ rental properties generated a significantly lower share of their potential rental revenue in 2020 relative to 2019. We observe proportionate three- to fourfold increases in rental non-payment during the pandemic for all cities in our sample with 9 percent of landlords receiving less than half of their yearly rent in 2020. Small owners and landlords of color faced high exposure to rental non-payment prior to the pandemic and continued to struggle with rent collection in 2020. While changes to landlord business practices—such as the granting of rental extensions—were strongly correlated with rental revenue decreases, business impacts alone cannot explain landlords’ behavioral responses. The pandemic also amplified the relationship between rental collection and actions such as rent forgiveness and deferred property maintenance, perhaps due to pandemic-induced constraints on landlords’ traditional responses to rental non-payment. This suggests that many owners modified their practices to recover funds and attempted to cut costs by reducing investments in their properties, much like small business owners more generally (e.g., Bartik et al. 2020). This was particularly the case for landlords of color, who delayed property payments and maintenance during the pandemic at a significantly higher rate than white landlords. There are two distinct negative effects of landlords’ rental property disinvestment. First, in the short term, it may imply that renters are residing in units of substandard quality, thus affecting their health and well-being. Second, it may indicate that many properties need further investment post-pandemic to remain viable. This latter implication is particularly worrisome as small and mid-sized typically struggle to access capital to invest in their properties (Local Housing Solutions 2021). Absent concerted efforts to bridge these credit gaps, property owners will have difficulty restructuring their financing to ensure their properties are viable. This may result in rental units leaving the housing stock earlier than they previously would have, thus exacerbating preexisting issues around housing affordability. The rental business challenges associated with the pandemic are clearly affecting owner behavior, with city-level rental non-payment rates positively correlated with property sale listings. Such sales could place further strain on the overall stock of affordable housing, although they also present an opportunity for localities to actively broker the sale and purchase of these properties to ensure their long-term viability. This approach could also serve as an opportunity for localities to provide subsidy support, with coterminous affordability restrictions, to increase the affordability of these units. Cities may be well positioned to pursue such a strategy given the unprecedented level of federal funds currently being deployed to stabilize local housing markets. Among the many concerning findings in our study is the disproportionate impact of the pandemic on the renters in and housing stock of communities of color. By showing that owners are more likely to exercise punitive actions on renters in markets with a majority of residents of color—both prior to and during the pandemic—we demonstrate that the ways in which owners are engaging with challenges around rental collection are racialized. Numerous notable works have documented persistent and pernicious racial discrimination in rental housing markets and investments (e.g., Taylor 2019; Reina et al. 2020), and our findings suggest that these discriminatory actions have persisted and even taken on new forms during the pandemic. As localities continue to build out and sustain renter assistance programs, they may want to offer additional supports and protections for tenants in these communities. One limitation of the study is its relatively limited sample size and owner response rate. It is also worth cautioning that, due to the population surveyed, our results likely reflect the experiences of those landlords who serve at least a portion of lower-income tenants. However, a dearth of information on property owners and their tenants at both the national and local level makes it difficult to assess this claim for certain. Of the pandemic's many important lessons, one is that we still know little about who owns rental properties and how these owners behave. The results of our report are critical to filling this gap, though we encourage readers to consider our findings in concert with other local and national efforts to understand this population. Funding Funding for this work was provided by Bloomberg Philanthropies through the Bloomberg Harvard City Leadership; the Housing Crisis Research Collaborative, supported by JPMorgan Chase & Co. and the Wells Fargo Foundation, managed by the Urban Institute, and made available to the authors through the Joint Center for Housing Studies of Harvard University; and the Annie E. Casey Foundation through the Housing Initiative at Penn. We thank all sources for their generous support and acknowledge that the findings and conclusions presented in this report are those of the authors alone. Uncited References Crane et al., 2020, Fang, Guess and Humphreys, 2019, Greig, Zhao and Lefevre, 2021, Bacher-Hicks, Goodman and Mulhern, 2021, JCHS (The Joint Center for Housing Studies of Harvard University), 2019, Kneebone, O'Regan, Raetz and Underriner, 2021, Lee, Park and Shin, 2021, Lundberg et al., 2021, Raymond et al., 2016 Declaration of Competing Interest None. Appendix A. Call to Participate Dear {CITY} rental property owner, We would like to invite you to participate in a brief online survey about how the COVID-19 pandemic is impacting your rental business. We want to hear your voice! When you are ready, please proceed with the following link: {SURVEY_LINK} The survey should take 10-15 minutes to complete. Participation is entirely voluntary and your individual responses will never be shared with the City of {CITY}. Choosing not to participate will in no way affect your relationship with the City, and City officials will not know which owners participated or those who opted not to. For more information about the study, please visit ash.harvard.edu/covid-19-landlord-survey. After you finish, you will be able to enter for a chance to win a $100 Amazon gift card. Multiple winners will be chosen! We hope you will participate! It won't take long and it will help the City better serve you and your tenants. Sincerely, Dr. Elijah de la Campa, Harvard Kennedy School Bloomberg Harvard City Leadership Initiative on behalf of The City of {CITY} Follow the link to opt out of future emails: ${OPT_OUT_LINK} B. Tables and Figures Appendix Figures 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 Appendix Figure 1 : Landlords’ Rental Collection Prior to and During the Pandemic, Excluding Los Angeles and Philadelphia Landlords.Notes: This figure plots landlords’ rental collection rates in 2019 and 2020, excluding Los Angeles and Philadelphia landlords. Rental payment is expressed as a percentage of total rent charged, in a given year, for a landlord's rental portfolio. The number of survey respondents in the sample is 2,051. Data come from the COVID-19 Landlord Survey. Appendix Figure 1 Appendix Figure 2 : Landlords’ Rental Collection Prior to and During the Pandemic, Among Landlords with at Least One Tenant Participating in ERA.Notes: This figure plots landlords’ rental collection rates in 2019 and 2020 among the sample of landlords who had at least one tenant receiving emergency rental assistance (ERA) in 2020. Panel A presents results for Los Angeles and Philadelphia landlords (N=299). Panel B presents results for Akron, Albany, Indianapolis, Minneapolis, Racine, Rochester, San Jose, and Trenton landlords (N=479). Rental payment is expressed as a percentage of total rent charged, in a given year, for a landlord's rental portfolio. Data come from the COVID-19 Landlord Survey. Appendix Figure 2 Appendix Figure 3 : Landlords’ Rental Collection Prior to and During the Pandemic, for Landlords’ Individual Rental Properties.Notes: This figure plots landlords’ rental collection rates in 2019 and 2020 at the individual rental properties reported in the survey. Rental payment is expressed as a percentage of total rent charged at that property in a given year. The number of rental properties in the sample is 2,513. Data come from the COVID-19 Landlord Survey. Appendix Figure 3 Appendix Figure 4 : Landlords’ Property-Level Rental Collection Rates, by Neighborhood Median Income.Notes: This figure plots, for 2019 and 2020, the share of landlords reporting less than 90 percent of total rent received at an individual rental property (Panel A) and less than 50 percent of total rent received at an individual rental property (Panel B), according to the neighborhood median income for that property. Properties are classified as “Below Median” if they are located in a neighborhood whose median income falls below the median for their city. Neighborhoods are classified according to census block groups (CBGs). 46.5 percent of properties are located in a neighborhood with an above-median household income. See Appendix Table 1 for each city's median household income. Models include city fixed effects. The number of rental properties in the sample is 2,428. Heteroskedastic-robust confidence intervals are reported. Data come from the COVID-19 Landlord Survey and 2018 ACS. Appendix Figure 4 Appendix Figure 5 : Landlords’ 2020 Property-Level Business Practices (y-axis) and Neighborhood Share of Non-White Residents (x-axis), Conditional on Landlords’ Demographic and Business Characteristics.Notes: This figure presents binned scatter plots of landlords’ 2020 property-level rental business practices versus the neighborhood share of non-white residents (for the property), conditional on landlords’ demographics and business characteristics. For each plot, the share of landlords reporting the practice is reported on the y-axis while the neighborhood share of non-white residents is presented on the x-axis. A neighborhood's share of non-white residents is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. To construct these plots, we first demean both landlords’ rental business practices and neighborhood share of non-white residents by city; 2020 rent collection; and the set of indicators for landlords’ demographics, business details, and portfolio size reported in Figure 2. We then divide the observations into 20 equal-sized groups (vigintiles) based on neighborhood racial composition and plot the share of landlords pursuing the indicated rental practice within each bin. The solid lines show the best linear fit estimated on the underlying micro data using OLS regression, and Table 8 presents these regression estimates. “Grant Rent Exten.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fees” indicates charging fees for late rent. “Increase Rents” indicates increases to monthly rents. “Decrease Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. The number of rental properties for each plot is 2,402. Data come from the COVID-19 Landlord Survey and 2018 ACS. Appendix Figure 5 Appendix Figure 6 : Landlords’ 2020 Property-Level Business Practices (y-axis) and Neighborhood Share of Non-White Residents (x-axis),By Landlord Race.Notes: This figure presents binned scatter plots of landlords’ 2020 property-level rental business practices versus the neighborhood share of non-white residents (for the property), separately for white (blue) and non-white (red) landlords. For each plot, the share of landlords reporting the practice is reported on the y-axis while the neighborhood share of non-white residents is presented on the x-axis. A neighborhood's share of non-white residents is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. To construct these plots, we first demean both white and non-white landlords’ rental business practices and neighborhood share of non-white residents by city; 2020 rent collection; and the set of indicators for landlords’ demographics, business details, and portfolio size reported in Figure 2 (with the exception of landlord race), separately for each race. We then divide the observations into 20 equal-sized groups (vigintiles) based on neighborhood racial composition and plot the share of landlords pursuing the indicated rental practice within each bin. The solid lines show the best linear fit estimated on the underlying micro data using OLS regression. “Grant Rent Exten.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fees” indicates charging fees for late rent. “Increase Rents” indicates increases to monthly rents. “Decrease Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. The number of rental properties for each plot is 2,402. Data come from the COVID-19 Landlord Survey and 2018 ACS. Appendix Figure 6 Appendix Figure 7 : Landlords’ 2020 Property-Level Business Practices (y-axis) and Neighborhood Share of Non-White Residents (x-axis), By Landlord Portfolio Size.Notes: This figure presents binned scatter plots of landlords’ 2020 property-level rental business practices versus the neighborhood share of non-white residents (for the property), separately for small (blue), mid-sized (red), and large (orange) landlords. For each plot, the share of landlords reporting the practice is reported on the y-axis while the neighborhood share of non-white residents is presented on the x-axis. A neighborhood's share of non-white residents is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. To construct these plots, we first demean both small, mid-sized, and large landlords’ rental business practices and neighborhood share of non-white residents by city; 2020 rent collection; and the set of indicators for landlords’ demographics and business details reported in Figure 2 (with the exception of portfolio size), separately for each portfolio size grouping. We then divide the observations into 20 equal-sized groups (vigintiles) based on neighborhood racial composition and plot the share of landlords pursuing the indicated rental practice within each bin. The solid lines show the best linear fit estimated on the underlying micro data using OLS regression. “Grant Rent Exten.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fees” indicates charging fees for late rent. “Increase Rents” indicates increases to monthly rents. “Decrease Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. The number of rental properties for each plot is 2,380. Data come from the COVID-19 Landlord Survey and 2018 ACS. Appendix Figure 7 Appendix Figure 8 : Landlords’ 2020 Property-Level Business Practices (y-axis) and Neighborhood Share of Non-White Residents (x-axis), Conditional on Additional Neighborhood Attributes.Notes: This figure presents binned scatter plots of landlords’ 2020 property-level rental business practices versus the neighborhood share of non-white residents (for the property), conditional on the neighborhood median income and age distribution. For each plot, the share of landlords reporting the practice is reported on the y-axis while the neighborhood share of non-white residents is presented on the x-axis. A neighborhood's share of non-white residents is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. To construct these plots, we first demean both landlords’ rental business practices and neighborhood share of non-white residents by city, 2020 rent collection, neighborhood population, neighborhood median income, the share of residents under the age of 25, and the share of residents aged 25 to 54. We then divide the observations into 20 equal-sized groups (vigintiles) based on neighborhood racial composition and plot the share of landlords pursuing the indicated rental practice within each bin. The solid lines show the best linear fit estimated on the underlying micro data using OLS regression, and Table 8 presents these regression estimates. “Grant Rent Exten.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fees” indicates charging fees for late rent. “Increase Rents” indicates increases to monthly rents. “Decrease Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. The number of rental properties for each plot is 2,402. Data come from the COVID-19 Landlord Survey and 2018 ACS. Appendix Figure 8 Appendix Figure 9 : Landlords’ 2020 Property-Level Business Practices (y-axis) and Neighborhood Income (x-axis).Notes: This figure presents nine binned scatter plots of landlords’ 2020 property-level rental business practices versus the neighborhood median income (for the property). For each plot, the share of landlords reporting the practice is reported on the y-axis while the natural log of neighborhood median income is presented on the x-axis. To construct these plots, we first demean both landlords’ rental business practices and neighborhood median income by city, average rent collection, and neighborhood population. We then divide the observations into 20 equal-sized groups (vigintiles) based on the natural log of neighborhood median income and plot the share of landlords pursuing the indicated rental practice within each bin. The solid lines show the best linear fit estimated on the underlying micro data using OLS regression, and Appendix Table 7 presents these regression estimates. “Grant Rent Exten.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fees” indicates charging fees for late rent. “Increase Rents” indicates increases to monthly rents. “Decrease Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. The number of rental properties for each plot is 2,322. Data come from the COVID-19 Landlord Survey and 2018 ACS. Appendix Figure 9 Appendix Figure 10 : Landlords’ 2020 Property-Level Business Practices (y-axis) and Yearly Rent Appreciation (x-axis). Notes: This figure presents nine binned scatter plots of landlords’ 2020 property-level rental business practices versus the three-year average, from 2015 to 2018, in the year-over-year change in neighborhood median gross rent (for the property). For each plot, the share of landlords reporting the practice is reported on the y-axis while the neighborhood average gross rent appreciation, measured in $100s, is presented on the x-axis. To construct these plots, we first demean both landlords’ rental business practices and neighborhood average gross rent appreciation by city, average rent collection, and neighborhood population. We then divide the observations into 20 equal-sized groups (vigintiles) based on the average gross rent appreciation and plot the share of landlords pursuing the indicated rental practice within each bin. The solid lines show the best linear fit estimated on the underlying micro data using OLS regression, and Appendix Table 8 presents these regression estimates. “Grant Rent Exten.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fees” indicates charging fees for late rent. “Increase Rents” indicates increases to monthly rents. “Decrease Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maintenance” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. The number of rental properties for each plot is 2,322. Data come from the COVID-19 Landlord Survey and 2015-2018 ACS. Appendix Figure 10 Appendix Tables 1 , 2 , 3 , 4 , 5 , 6, 7 , 8 , 9 Appendix Table 1 : Descriptive Statistics of Residents and Renter Households in Survey Cities Appendix Table 1 Akron Albany Indianapolis Los Angeles Minneapolis Philadelphia Racine Rochester San Jose Trenton Panel A: Resident Characteristics White 58.5 49.9 55.2 28.5 59.8 34.6 49.9 36.6 26 12.9 Black 29.9 27.9 28.1 8.6 19.1 41.0 22.2 38.2 2.8 48.4 Hispanic 2.5 10.2 10.2 48.6 9.6 14.5 23.1 18.3 32 36.4 Asian 4.6 6.9 3.2 11.7 6.1 7.1 0.9 3.2 35.6 1.1 Other race 4.5 5.0 3.3 2.8 5.5 2.8 3.9 3.8 3.6 1.2 Median age (y) 36.7 31.2 34.2 35.4 32.1 34.3 34 31.9 36.5 33.9 N Residents (100,000s) 2.0 1.0 8.6 39.7 4.2 15.8 0.8 2.1 10.3 0.8 Panel B: Renter Household Characteristics Renter-occupied (among all households) 49.4 63.2 46.7 63.2 52.7 47.0 49.0 63.7 42.8 63.4 Reside in 1-unit property 44.7 9.0 38.2 21.1 15.3 40.8 33.6 30.2 32.9 45.8 Reside in 2-4 unit property 20.5 57 13.6 12 18.1 25 33.1 34.6 13.3 20.6 Reside in 5-9 unit property 8.9 12.3 19.0 12.9 6.3 7.4 7.9 9.2 9.3 7.2 Reside in 10-19 unit property 8.6 6.1 12.8 14.2 12.8 4.5 7.2 5.0 10.1 4.5 Reside in 20+ unit property 17.0 15.5 15.4 39.4 47.3 21.9 17.7 20.6 33.4 22 Median income ($) 25,598 30,972 31,299 43,015 37,155 31,508 28,900 24,043 72,825 24,355 Cost-burdened 47.7 53.5 49.0 57.3 46.3 50.2 50.8 57.0 50.2 56.3 Median gross rent ($) 735 951 865 1,376 985 1,007 824 831 1,970 1,029 Median age of housing structure (y) 63 80 47 55 59 70 68 77 43 - N Renter Households (10,000s) 4.2 2.6 15.6 86.8 9.2 28.0 1.5 5.5 13.8 1.7 Notes: This table reports descriptive characteristics of residents and renter households separately for the ten COVID-19 Landlord Survey cities. Data come from the ACS 2018 five-year sample. Unless otherwise indicated, the variables above are expressed as percentages. Categorical variables may not sum to 100 due to rounding. Cost-burdened renters are defined as those who spend 30 percent or more of their yearly income on yearly rent. Appendix Table 2 : Descriptive Statistics of Survey City Rental Properties, by Rental Registry Compliance Appendix Table 2 Not on Registry On Registry Panel A: Property Characteristics Property units (n) 1.5 2.8 Property age (y) 78.2 92.9 Missing property age 2.7 1.4 LLC or LLP/LP owner 24.8 30.6 Per-unit assessed property value ($) 119,301 97,402 Missing per-unit assessed property value 1.0 0.4 Per-unit residential area (sq. ft.) 1,614 1,366 Missing per-unit residential area 14.0 28.8 Panel B: Neighborhood Characteristics Residents of color 48.1 46.7 Median household income ($) 47,541 44,0230 Median gross rent ($) 870 926 Notes: This table reports descriptive means for all rental registry eligible rental properties in Akron, Albany, Indianapolis, Minneapolis, Racine, Rochester, San Jose, and Trenton. In Rochester, owner-occupied two-family rental properties are exempt from the rental registry, though because we cannot identify these properties, they are included in the eligible sample. In San Jose, only properties built before 1979 with three or more rental units are required to register with the city and thus included in the eligible sample. Unless stated otherwise, the variables above are expressed as percentages. Data on property characteristics come from city administrative records. Data on neighborhood characteristics come from the ACS 2018 five-year sample. Appendix Table 3 : Descriptive Statistics of Survey Respondents, by City Appendix Table 3 Akron Albany Indianapolis Los Angeles Minneapolis Philadelphia Racine Rochester San Jose Trenton Male 63.9 75.8 65.5 51.6 60.4 56.7 54.3 63.8 61.5 68.7 Missing gender 12.0 16.7 23.2 25.8 16.4 23.1 18.6 14.6 32.2 27.2 White 78.4 76.8 74.9 50.8 80.2 42.9 83.4 57.4 47.3 39.6 Black 12.5 8.1 9.8 14.1 3.7 27.2 8.3 20.6 0.9 31.1 Hispanic 2.2 5.1 3.1 18.3 3.1 11.0 3.4 4.5 11.5 6.6 Asian 1.3 5.1 4.7 8.9 6.5 8.7 2.1 6.5 31.4 15.1 Missing race 10.1 13.2 20.3 23.0 15.4 18.6 15.7 12.9 26.4 22.1 20-29 Years Old 1.3 2.0 3.3 0.0 3.7 2.7 1.4 1.3 0.9 1.9 30-39 Years Old 11.5 22.0 15.2 7.2 20.4 18.4 11.6 14.5 3.9 15.7 40-49 Years Old 12.8 18.0 19.8 12.4 19.8 24.2 14.3 18.9 11.2 23.1 50-59 Years Old 34.6 19.0 24.8 23.2 21.6 25.4 32.7 25.2 23.3 37.0 60+ Years Old 39.7 39.0 36.9 57.2 34.4 29.3 40.1 40.3 60.8 22.2 Missing age 9.3 12.3 19.2 21.8 13.2 17.9 14.5 10.7 24.4 20.6 Individual owner 80.2 79.8 92.5 - 95.9 - 85.8 76.8 88.2 66.2 Missing ownership structure 0.4 0.0 1.6 100.0 2.1 100.0 1.7 0.6 1.0 4.4 Self-manages rental units 76.5 74.3 60.8 63.9 75.7 77.6 83.5 83.2 63.4 77.5 Missing management structure 4.3 4.4 5.1 6.0 5.2 5.4 8.1 2.8 4.9 5.1 Accepts HCVs 31.6 27.3 13.1 20.4 10.5 23.5 19.6 32.2 33.7 22.5 Missing HCV status 4.3 3.5 4.9 5.2 5.3 4.5 8.1 2.2 5.2 5.1 Owns single-family rentals(s) (SFRs) 96.7 92.5 93.8 53.3 82.8 89.8 95.4 95.2 63.8 92.1 Missing home type 7.4 7.0 10.7 9.3 13.2 9.0 11.6 7.3 13.7 16.2 Small landlord 72.3 67.5 72.2 40.7 73.3 60.9 78.9 68.0 45.7 69.4 Mid-sized landlord 14.5 20.2 12.0 31.1 12.3 17.9 9.4 14.0 34.1 11.2 Missing portfolio size 0.8 0.0 1.6 2.8 1.3 1.6 0.6 0.0 4.6 1.5 N Respondents 258 114 449 248 676 312 172 178 307 136 Notes: This table reports descriptive statistics for the COVID-19 Landlord Survey respondents, by city. The variables above are expressed as percentages. The omitted category for race is “Other Race.” The omitted category for age is “Under 20 years old.” No respondents in the survey reported being under 20. “Individual owner” indicates an owner who is not incorporated as an LLC or LLP. “Single-family rental” indicates a one- to four-unit rental property. “Small landlord” indicates a landlord who owns between 1 and 5 rental units. “Mid-sized landlord” indicates a landlord who owns between 6 and 19 rental units. The omitted category for landlord size is “large” (owns 20+ rental units). Categorical variables may not sum to 100 due to rounding. Data come from city administrative records and the COVID-19 Landlord Survey. Appendix Table 4 : Descriptive Statistics of Survey Respondents, by Eviction History Appendix Table 4 No Past Evictions 2019 Evictions Only 2020 Evictions Only Evictions 2019 and 2020 Male 58.3 72.6 73.9 70.6 Missing gender 16.1 9.2 27.6 9.5 White 69.7 64.8 54.4 61.7 Black 9.9 7.9 19.7 17.4 Hispanic 5.8 6.7 4.1 4.7 Asian 8.1 8.5 11.6 7.4 Missing race 13.2 4.6 25.0 5.7 20-29 years old 1.8 1.2 0.7 0.7 30–39 years old 13.3 12.4 22.1 17.0 40–49 years old 18.4 14.8 18.8 20.9 50–59 years old 24.7 33.7 26.2 34.0 60+ years old 41.7 37.9 32.2 27.5 Missing age 11.8 2.3 24.0 3.2 Individual owner 90.7 80.4 76.4 67.2 Missing ownership structure 19.9 17.3 24.5 22.8 Self-manages rental units 72.9 59.3 58.5 63.1 Missing property manager 0.8 0.6 0.5 0.6 Accepts HCVs 17.4 42.2 35.9 39.2 Missing HCV 0.6 0.0 0.5 0.0 Owns single-family rental(s) (SFRs) 84.0 80.2 87.6 91.4 Missing home type 3.1 3.5 1.0 4.4 Small landlord 77.1 47.7 59.2 32.7 Mid-sized landlord 16.3 27.9 20.9 28.8 Missing portfolio size 1.0 0.6 0.0 1.3 N Respondents 1,877 173 196 158 Notes: This table reports descriptive statistics for the COVID-19 Landlord Survey respondents according to the number of years in which landlords have reported evicting one or more tenants. The variables above are expressed as percentages. The omitted category for race is “Other Race.” The omitted category for age is “Under 20 years old.” No respondents in the survey reported being under 20. “Individual owner” indicates an owner who is not incorporated as an LLC or LLP. “Single-family rental” indicates a one- to four-unit rental property. “Small landlord” indicates a landlord who owns between 1 and 5 rental units. “Mid-sized landlord” indicates a landlord who owns between 6 and 19 rental units. The omitted category for landlord size is “large” (owns 20+ rental units). Categorical variables may not sum to 100 due to rounding. Survey data come from city administrative records and the COVID-19 Landlord Survey. Appendix Table 5 : Relationship between Rental Collection and Business Practices, Excluding Los Angeles and Philadelphia Landlords Appendix Table 5 Grant Rent Ext. Forgive Rent Charge Rent Fee Inc. Rents Dec. Rents Evict Tenants Miss Payments Defer Maint. List Props. for Sale (1) (2) (3) (4) (5) (6) (7) (8) (9) < 90% Rent Received 0.090*** 0.030* 0.074** -0.112*** 0.058*** 0.135*** 0.071*** 0.091*** 0.025 (0.032) (0.017) (0.034) (0.026) (0.018) (0.032) (0.023) (0.025) (0.017) 2020 0.190*** 0.113*** -0.132*** -0.193*** 0.103*** -0.071*** 0.042*** 0.154*** 0.048*** (0.015) (0.010) (0.012) (0.013) (0.009) (0.010) (0.008) (0.012) (0.008) < 90% Rent Received*2020 0.245*** 0.143*** 0.030 0.093*** 0.054** 0.113*** 0.203*** 0.198*** 0.110*** (0.038) (0.026) (0.038) (0.029) (0.025) (0.037) (0.030) (0.033) (0.024) N Landlord-Years 3,892 3,892 3,892 3,892 3,892 3,892 3,892 3,892 3,892 Notes: This table reports OLS estimates of the relationship between landlords’ business practices and rental collection, prior to and during the pandemic, excluding Los Angeles and Philadelphia landlords, where respondents were invited to participate according to whether they had at least one tenant participate in emergency rental assistance (ERA). Each column presents results from a separate OLS regression, where the indicated business practice is the dependent variable. “Grant Rent Ext.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fee” indicates charging fees for late rent. “Inc. Rents” indicates increases to monthly rents. “Dec. Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maint.” indicates delayed property repairs or maintenance. “List Props. For Sale” indicates one or more properties were listed for sale. Landlords could choose multiple actions. Models include city fixed effects. Heteroskedastic-robust standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Data come from the COVID-19 Landlord Survey. Appendix Table 6 : Descriptive Statistics of Survey Respondents According to Whether They Own at Least One Property in a Majority Non-White Neighborhood Appendix Table 6 Owns at Least One Property in Majority Non-White N'hood Does not Own Any Properties in Majority Non-White N'hood N Mean N Mean Male 1190 59.9 1072 62.3 Missing gender 1331 10.6 1182 9.3 White 1222 54.8 1110 83.8 Black 1222 19.4 1110 2.6 Hispanic 1222 7.9 1110 3.1 Asian 1222 10.5 1110 4.7 Missing race 1331 8.2 1182 62.3 20-29 years old 1245 1.5 1127 1.5 30–39 years old 1245 14.1 1127 14.1 40–49 years old 1245 16.8 1127 18.6 50–59 years old 1245 25.7 1127 25.7 60+ years old 1245 41.9 1127 40.0 Missing age 1331 6.5 1182 4.7 Individual owner 965 80.2 1046 83.7 Missing ownership structure 1331 27.5 1182 11.5 Self-manages rental units 1302 69.7 1156 71.7 Missing property manager 1331 2.2 1182 2.2 Accepts HCVs 1309 29.4 1157 12.5 Missing HCV 1331 1.7 1182 2.1 Owns single-family rental(s) (SFRs) 1295 84.1 1150 85.6 Missing home type 1331 2.7 1182 2.7 Small landlord 1319 70.6 1170 80.1 Mid-sized landlord 1319 18.4 1170 12.7 Missing portfolio size 1331 0.9 1182 1.0 Notes: This table reports descriptive statistics for the COVID-19 Landlord Survey respondents according to whether they own at least one property in a majority non-white neighborhood. A neighborhood's share of non-white residents is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. The variables above are expressed as percentages. The omitted category for race is “Other Race.” The omitted category for age is “Under 20 years old.” No respondents in the survey reported being under 20. “Individual owner” indicates an owner who is not incorporated as an LLC or LLP. “Single-family rental” indicates a one- to four-unit rental property. “Small landlord” indicates a landlord who owns between 1 and 5 rental units. “Mid-sized landlord” indicates a landlord who owns between 6 and 19 rental units. The omitted category for landlord size is “large” (owns 20+ rental units). Categorical variables may not sum to 100 due to rounding. Survey data come from city administrative records and the COVID-19 Landlord Survey and 2018 ACS. Appendix Table 7 : Relationship Between Landlords’ 2020 Property-Level Rental Business Practices and Neighborhood Median Income Appendix Table 7 Grant Rent Ext. Forgive Rent Charge Rent Fee Inc. Rents Dec. Rents Evict Tenants Miss Payments Defer Maint. List Prop. for Sale (1) (2) (3) (4) (5) (6) (7) (8) (9) Unconditional 2020 Mean 0.37 0.15 0.08 0.04 0.12 0.09 0.14 0.26 0.07 Panel A: Baseline Model Log Median Income -0.027 -0.010 -0.016 0.004 0.005 -0.014 -0.027** -0.016 -0.004 (0.020) (0.015) (0.011) (0.008) (0.015) (0.010) (0.013) (0.019) (0.011) Panel B: Landlord Controls Log Median Income -0.019 -0.015 -0.013 0.001 0.001 -0.009 -0.019 -0.011 -0.006 -0.02 -0.016 -0.011 -0.008 -0.015 -0.01 -0.013 -0.019 -0.011 Panel C: Neighborhood Controls Log Median Income -0.019 -0.051*** -0.004 0.005 -0.035** 0.01 0.003 -0.01 -0.005 -0.023 -0.019 -0.014 -0.01 -0.017 -0.013 -0.017 -0.023 -0.014 Panel D: Landlord and Neighborhood Controls Log Median Income -0.009 -0.047** 0.001 0.006 -0.027 0.014 0.003 -0.005 -0.005 -0.023 -0.019 -0.014 -0.01 -0.017 -0.013 -0.017 -0.023 -0.014 N Rental Properties 2,322 2,322 2,322 2,322 2,322 2,322 2,322 2,322 2,322 Notes: This table reports OLS regression estimates of the relationship between landlords’ 2020 property-level rental business practices and the neighborhood median income. Each column presents results from a separate OLS regression of a residualized version of the indicated business practice on a residualized version of neighborhood median income. In Panel A, we residualize on 2020 rent collection, neighborhood population, and city fixed effects. In Panel B, we residualize on the controls from Panel A, as well as the set of indicators for landlords’ demographics, business details, and portfolio size reported in Figure 2. In Panel C, we residualize on the controls from Panel A, as well as the neighborhood median income and age distribution. In Panel D, we residualize on all controls from Panels A, B, and C. “Grant Rent Ext.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fee” indicates charging fees for late rent. “Inc. Rents” indicates increases to monthly rents. “Dec. Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maint.” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. Landlords could choose multiple actions. Heteroskedastic-robust standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Data come from the COVID-19 Landlord Survey and 2018 ACS. Appendix Table 8 : Relationship Between Landlords’ 2020 Property-Level Rental Business Practices and Yearly Rent Appreciation Appendix Table 8 Grant Rent Ext. Forgive Rent Charge Rent Fee Inc. Rents Dec. Rents Evict Tenants Miss Payments Defer Maint. List Prop. for Sale (1) (2) (3) (4) (5) (6) (7) (8) (9) Unconditional 2020 Mean 0.37 0.15 0.08 0.04 0.12 0.09 0.14 0.26 0.07 Panel A: Baseline Model 3-Yr Avg. Rent Appreciation -0.006 0.002 -0.010 0.004 0.018 -0.002 -0.001 0.000 0.014 (0.017) (0.014) (0.010) (0.008) (0.013) (0.009) (0.013) (0.016) (0.010) Panel B: Landlord Controls 3-Yr Avg. Rent Appreciation -0.009 -0.001 -0.010 0.003 0.015 -0.002 0.001 0.003 0.013 (0.017) (0.014) (0.010) (0.008) (0.012) (0.009) (0.012) (0.016) (0.010) Panel C: Neighborhood Controls 3-Yr Avg. Rent Appreciation 0.002 0.001 -0.009 0.005 0.015 -0.002 -0.002 -0.002 0.016* (0.017) (0.014) (0.010) (0.008) (0.013) (0.010) (0.013) (0.017) (0.010) Panel D: Landlord and Neighborhood Controls 3-Yr Avg. Rent Appreciation -0.002 -0.001 -0.008 0.005 0.013 -0.002 -0.000 0.001 0.016 (0.017) (0.014) (0.010) (0.008) (0.012) (0.010) (0.013) (0.016) (0.010) N Rental Properties 2,203 2,203 2,203 2,203 2,203 2,203 2,203 2,203 2,203 Notes: This table reports OLS regression estimates of the relationship between landlords’ 2020 property-level rental business practices and the three-year average, from 2015 to 2018, in the year-over-year change in neighborhood median gross rent (measured in $100s). Each column presents results from a separate OLS regression of a residualized version of the indicated business practice on a residualized version of median gross rent appreciation. In Panel A, we residualize on 2020 rent collection, neighborhood population, and city fixed effects. In Panel B, we residualize on the controls from Panel A, as well as the set of indicators for landlords’ demographics, business details, and portfolio size reported in Figure 2. In Panel C, we residualize on the controls from Panel A, as well as the neighborhood median income and age distribution. In Panel D, we residualize on all controls from Panels A, B, and C. “Grant Rent Ext.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fee” indicates charging fees for late rent. “Inc. Rents” indicates increases to monthly rents. “Dec. Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maint.” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. Landlords could choose multiple actions. Heteroskedastic-robust standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Data come from the COVID-19 Landlord Survey and 2018 ACS. Appendix Table 9 : Relationship Between Landlords’ 2019 Property-Level Rental Business Practices and Neighborhood Share of Non-White Residents Appendix Table 9 Grant Rent Ext. Forgive Rent Charge Rent Fee Inc. Rents Dec. Rents Evict Tenants Miss Payments Defer Maint. List Prop. for Sale (1) (2) (3) (4) (5) (6) (7) (8) (9) Unconditional 2019 Mean 0.166 0.032 0.247 0.286 0.022 0.145 0.040 0.054 0.030 Panel A: Baseline Model Share Non-White Residents 0.059** 0.002 0.095*** -0.118*** -0.014 0.125*** 0.044*** -0.008 -0.016 (0.029) (0.016) (0.034) (0.035) (0.011) (0.028) (0.016) (0.019) (0.015) Panel B: Landlord Controls Share Non-White Residents 0.033 0.009 0.076** -0.108*** -0.009 0.114*** 0.033** -0.001 -0.014 (0.030) (0.016) (0.034) (0.035) (0.011) (0.027) (0.016) (0.019) (0.015) Panel C: Neighborhood Controls Share Non-White Residents 0.011 -0.014 0.097** -0.096** -0.014 0.103*** 0.044** -0.009 -0.010 (0.038) (0.021) (0.042) (0.043) (0.014) (0.036) (0.020) (0.023) (0.021) Panel D: Landlord and Neighborhood Controls Share Non-White Residents -0.007 -0.008 0.098** -0.065 -0.009 0.113*** 0.032* -0.001 -0.008 (0.038) (0.021) (0.042) (0.042) (0.013) (0.034) (0.019) (0.023) (0.021) N Rental Properties 2,402 2,402 2,402 2,402 2,402 2,402 2,402 2,402 2,402 Notes: This table reports OLS regression estimates of the relationship between landlords’ 2019 property-level rental business practices and the neighborhood share of non-white residents. Each column presents results from a separate OLS regression of a residualized version of the indicated business practice on a residualized version of the neighborhood share of non-white residents. In Panel A, we residualize on 2019 rent collection, neighborhood population, and city fixed effects. In Panel B, we residualize on the controls from Panel A, as well as the set of indicators for landlords’ demographics, business details, and portfolio size reported in Figure 2. In Panel C, we residualize on the controls from Panel A, as well as the neighborhood median income and age distribution. In Panel D, we residualize on all controls from Panels A, B, and C. A neighborhood's share of non-white residents is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. “Grant Rent Ext.” indicates rental extensions and/or putting tenants on repayment plans. “Forgive Rent” indicates rental forgiveness (either in full or a portion). “Charge Rent Fee” indicates charging fees for late rent. “Inc. Rents” indicates increases to monthly rents. “Dec. Rents” indicates decreases to monthly rents. “Evict Tenants” indicates the commencement of eviction procedures (and potentially, the conclusion). “Miss Payments” indicates missed mortgage, property tax, and/or utility payments. “Defer Maint.” indicates delayed property repairs or maintenance. “List Prop. For Sale” indicates the property was listed for sale. Landlords could choose multiple actions. Heteroskedastic-robust standard errors are reported in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Data come from the COVID-19 Landlord Survey and 2018 ACS. Data Availability The authors do not have permission to share data. We are grateful to our partners in the City of Akron Mayor's Office and Office of Integrated Development; City of Albany Department of Buildings and Regulatory Compliance; City of Indianapolis Mayor's Office; City of Los Angeles Mayor's Office and Department of Housing and Community Investment Department; City of Minneapolis Department of Regulatory Services; City of Philadelphia Department of Planning and Development; City of Racine Mayor's Office and Department of Management Information Systems; City of Rochester Department of Neighborhood and Business Development; and City of Trenton Mayor's Office and Department of Housing and Economic Development for their support and promotion of this work. We are also grateful to Kate Bischoff, Andrew Kieve, Bolek Kurowski, and Glen Nuckols from Tolemi for providing feedback on early iterations of the survey, assisting with data access, and assisting with survey coding. Jessica Creighton, Tanya Dall, and Sydney Goldstein provided excellent project management. Eleanor Dickens, Raheem Hanifa, and Ashley Marcoux provided exceptional research assistance. Chris Herbert provided invaluable support and feedback throughout all phases of this work. Andrew Bacher-Hicks, Jorrit de Jong, Nat Decker, Ingrid Gould Ellen, Emma Foley, Austin Harrison, Reed Jordan, Monique King-Viehland, Elizabeth Kneebone, David Luberoff, Alan Mallach, Kathy O'Regan, Snapper Poche, Rebecca Yae and two anonymous reviewers provided many useful comments, as did members of the Rental Research Community of Practice. 1 For an excellent overview of the renter-focused research, see Airgood-Obryicki et al. (2021). See Hepburn et al. (2021) for a discussion of tenant evictions during the pandemic. 2 Throughout this paper, we will use the term “property owner” and “landlord” interchangeably. We will also use the term “pre-COVID” to refer to the 2019 calendar year, while “post-COVID” will refer to the 2020 calendar year. Similarly, “pre-pandemic” will refer to 2019, while “during the pandemic” will refer to 2020. 3 Other studies have more explicitly estimated the value of tenant and landlord rental assistance need in specific markets, but these studies have relied on secondary data sources to approximate these findings (Kneebone & Murray 2020; Kneebone & Reid 2020). 4 Most notably, small landlords and landlords of color are typically seen as the primary providers of naturally occurring (i.e., market-provided) affordable housing, and thus, house relatively more financially vulnerable tenants. 5 A notable exception is the work of Raymond et al. (2017) in Fulton County, Georgia. Using parcel-level eviction records, the authors show that corporate landlords are more likely than small landlords to file for tenant eviction, conditional on property and neighborhood characteristics. 6 Specifically, landlords were instructed to choose a property whose profitability prior to the pandemic was typical of their portfolio's pre-pandemic profitability. Asking questions at the rental property level allows us to explore variation according to property and neighborhood characteristics. 7 A potential concern with our survey methodology—particularly with respect to landlords’ pre-pandemic business performance and management—is recall bias (e.g., Sudman & Bradburn 1974). However, our survey was designed to minimize this bias in several ways. First, the survey was distributed close to tax season with the expectation that yearly profitability and business management would be top of mind. Second, we asked about business practices that were hypothesized to occur relatively infrequently, but which also have economic and social consequences—two factors that are likely to affect an event's salience and thus respondents’ recall decay (Bradburn, Sudman, & Wansink 2004). Along these lines, we asked respondents to recall simply whether they ever took the indicated action rather than how frequently as an additional strategy for minimizing this bias (Biemer et al. 2013). 8 The two most common reasons cities did not participate are that they did not maintain sufficient landlord contact data and/or did not have internal capacity to collaborate. 9 Notably, we were not able to secure the participation of any Southern US cities. 10 Appendix Table 1 presents these descriptive statistics separately for each city in the survey sample. 11 See Appendix A for the Call to Participate. Interested parties may contact the authors directly for the survey instrument itself. 12 There were technically 2,930 respondents to the COVID-19 Landlord Survey. We exclude 80 individuals who reported managing as opposed to owning property from our usable response sample, as these individuals were automatically routed to the end of the survey. 13 Typically, owners must pay a small fee to register their rental properties with their city, which covers the cost of a housing habitability inspection. For example, the rental inspection fee in Albany, New York is $50 per rental property unit. Examples of common inspection criteria include working smoke and carbon monoxide detectors; open means of egress; clean, running water; and basic unit security. Owners who fail their initial inspection must remedy any habitability issues and then pass a re-inspection. In most municipalities, though owners are subject to monetary penalties for lapsed rental registrations, they are often given the opportunity to rectify the situation prior to the issuance of fees. 14 These types of properties represent 8.2 percent of all San Jose rental properties, while the units therein represent 35.4 percent of all San Jose rental units. 15 While we cannot adequately explore landlord characteristics for registry compliers and non-compliers, we can explore the characteristics of rental properties by registry compliance status (Appendix Table 2). Properties in compliance with the rental registry tend to be older, have more units, and are more likely to be owned by a landlord registered as a limited liability corporation or partnership. They also tend to have lower per-unit property values and have slightly less residential area. Rental registry properties tend to be located in neighborhoods with a higher share of residents of color, though we do not observe any meaningful differences across compliance status in neighborhood median household income or gross rent. Note that landlords with properties in compliance with their city's rental registry may also own properties not in compliance. Unfortunately, our survey is not equipped to explore this issue. 16 And as discussed above, rental registry requirements may also vary across cities. 17 This discrepancy is perhaps not surprising as voucher use tends to be concentrated in the urban core (McClure, Schwartz, & Taghavi 2015), and past research has shown non-white landlords—which are over-represented in our sample—are more likely than white ones to accept HCVs (Choi & Goodman 2021). 18 A lack of data on landlords’ portfolios makes it difficult to contextualize these figures. 10.3 million individual investors own an average of 1.7 rental properties, roughly 99 percent of which are SFRs. An estimated 1 million institutional landlords own around 45 percent of all rental units (DeSilver 2021). Thus, while most US rental property owners are small- to mid-sized landlords who own at least one SFR, a small share of owners manage nearly half of the rental stock. 19 Appendix Table 3 shows there is considerable variation in these demographics across cities, though most of the landlords that responded to our survey tend to be male, over the age of 50, disproportionately white compared to the racial composition of their city, and own fewer than 20 rental units. In each city, with the exception of Los Angeles and San Jose, over 60 percent of landlords own five or fewer rental units. 20 For example, Decker (2021b) obtained owner information from a third-party company and recruited study participants through mailed letters. This mode of outreach limited the response rate, ultimately affecting the observed and unobserved composition of those owners that responded. 21 Note that even though all Los Angeles and Philadelphia landlords had at least one tenant apply for local ERA, this need not imply that the tenant participated in the program and/or received funds. 22 We treat collecting 90 percent or less of charged yearly rent as exposure to general rental non-payment for two related reasons: 1) existing data show that, even under typical conditions and among the highest-end units, landlords do not necessarily collect rent in full each month (NMHC 2020), and 2) the best way to approximate this fact given our rental collection categories is to aggregate the 50 to 89 and less than 50 percent of yearly rent received buckets. 23 In this case, the dependent variables are Rent90,2020 and Rent50,2020. 24 We focus on landlords over the age of 60 because this is the oldest age grouping in our survey, and thus, the grouping of landlords closest to retirement age. Non-white landlords are defined as those who identify as Black, Hispanic, Asian/Pacific Islander, Native American, or other races. 25 In both cases, our results are supported by prior research on HCVs. Lundberg et al. (2020) show that vouchers significantly reduce non-payment of rent. Kneebone et al. (2021) show that voucher and non-voucher users in California and New York City both struggled to make rent during the pandemic, but subsidized households accrued lower levels of rent arrears. 26 For instance, Parrot and Zandi (2021) use the Census Bureau Household Pulse Survey to demonstrate that renters are further behind on rent in West and East Coast urban areas; our results show that, proportionally, renters are behind on rent at relatively consistent rates across the geographic regions in our study. 27 Recall, landlords were instructed to choose a property whose profitability prior to the pandemic was typical of their portfolio's pre-pandemic profitability. Accordingly, Appendix Figure 3 presents landlords’ 2019 and 2020 rental collection rates at the individual rental property they reported on in the survey. While there is some variation in the magnitude of certain rental payment categories, the conclusions from this figure are qualitatively similar to those of Figure 1 for landlords’ portfolio-level rental collection. 28 A neighborhood's share of residents of color is defined as the sum of individuals who identify as Black, Hispanic, Asian, Native American, multiracial and/or other races. In practice, due to the cities in our study, communities of color are primarily comprised of Black and Hispanic residents. To construct neighborhood racial and ethnic composition classifications, we first match each property in our rental property sample to its census block group (CBG). We then use the 2018 ACS to obtain the mean share of residents of color for that CBG and classify the CBG according to whether this share is above or below 50 percent. We perform this exercise separately for the 10 cities in our sample. 29 To construct neighborhood income classifications, we first match each property in our rental property sample to its census block group (CBG). We then use the 2018 ACS to obtain the median household income for that CBG and classify the CBG according to whether its median household income falls above or below the citywide median. We perform this exercise separately for the 10 cities in our sample. 30 Of course, these actions may be changing precisely because rental collection was down in 2020 relative to 2019. We explore this possibility in further detail in Table 4 below. 31 Compared to the overall pool of landlords, landlords who reported evicting tenants in both 2019 and 2020 are disproportionately male (70.6 percent), Black (17.4 percent), younger (72.5 percent under the age of 60), non-individual owners (32.8 percent structured as an LLC or LLP), accept HCVs (39.2 percent), and own 6+ properties (67.3 percent). Appendix Table 4 provides more detail on this population, as well as those who never reported evicting tenants, those who reported evicting tenants only in 2019, and those who reported evicting tenants only in 2020. 32 Note that our survey specifically asked landlords if they had “missed mortgage payments” at any point during the aforementioned time periods. Though not intended to be inclusive of missed payments due to forbearance, some landlords may have interpreted the payment deferrals offered by these programs to be “missed payments.” Estimates of forbearance enrollment during the pandemic range from around 10 percent (e.g., Grief et al. 2021) to 20 percent (Engelhardt & Eriksen 2021). Our estimate for missed mortgage payments is nearly identical to the rates of non-forbearance-induced missed payments reported in the aforementioned studies. 33 Of course, there may be other reasons this is the case. For example, the pandemic may have caused landlords to develop an increased desire to assist tenants through their financial hardships, thus making rental payment plans a preferred response to rental non-payment. 34 As was the case with rental collection, we may be concerned that landlords in Los Angeles and Philadelphia were predisposed to pursuing certain types of management practices during the pandemic—such as granting rental extensions—since they, by definition, had at least one tenant behind on rent. Accordingly, Appendix Table 5 reproduces the results from Table 4 excluding these cities from the sample. Results are nearly identical across the two tables, indicating that differences among the ERA and rental registry samples are not substantially biasing our results for landlords’ business practices. 35 For Panel A, this is the share of large landlords who collected 90 percent or more of their 2019 rent. For Panel B, this is the share of white landlords who collected 90 percent or more of their 2019 rent. 36 Indeed, it may be the case that decreased rent collection within each city is driving the changes to landlords’ rental extension rates. We explore this possibility in further detail in Table 7 below. 37 In general, the cities of our study with stronger eviction moratoria experienced greater reductions in late fees and landlord eviction filing rates, the latter of which aligns with the findings of Hepburn et al. (2021). For example, in Minneapolis, a ban on all phases of the eviction process has been in place since March 2020, whereas in Rochester and Albany, landlords could serve tenants eviction notices from July through December 2020 (Raifman et al. 2020). 38 Specifically, we estimate Practic¯ecyp= β0+β1RentLT90cy¯+β22020y+β3RentLT90cy¯*2020y+εcy for the five key business practices p reported in Table 5. Practic¯ecyp represents the mean share of landlords pursuing business practice p in city c in year y. RentLT90cy¯ represents the mean share of landlords collecting at most 90 percent of rental revenue in city c in year y. 2020y is a binary indicator for the post-COVID time period. Results are weighted by the number of respondents in each city. 39 Moving from the 25th to 75th percentile of the city-demeaned resident of color distribution is associated with a 41percentage point change in a neighborhood's share of residents of color. 40 Appendix Table 6 provides insight into the different demographics of landlords who own at least one property in a majority non-white neighborhood versus those who do not. In general, the landlords in our sample who own in communities of color are more likely to be people of color themselves (45.2 vs. 16.2 percent), are more likely to accept HCVs (29.4 vs. 12.5 percent), and are less likely to be small landlords (70.6 vs. 80.1 percent). 41 Said differently, the slopes of the relationship between business practices and neighborhood racial composition are roughly constant for landlords with different demographic and business characteristics (conditional on 2020 rent collection, city fixed effects, neighborhood population, and the remaining set of demographic covariates) despite differing intercepts. Appendix Figure 6 shows this visually for non-white versus white landlords: for all business practices apart from the granting of rental extensions and missed payments, the best-fit regression lines are similarly sloped across the two groups. While there is some variation among large vs. small- and mid-sized landlords’ business practice intensity in communities of color, Appendix Figure 7 shows best-fit regression lines also tend to be similarly sloped across these three groups. 42 We control for the neighborhood residential age distribution through two variables: 1) share of residents under the age of 25, and 2) share of residents aged 25 to 54. Data on age come from the 2018 ACS. 43 We also directly explore the relationship between landlords’ business practices and neighborhoods’ economic characteristics. Appendix Figure 9 presents results according to neighborhood median income. We generally observe no meaningful relationship between landlords’ business practices and the neighborhood median income of their rental properties; the one exception is for the rate at which landlords have missed at least one mortgage, property tax, and/or utility payment, which is strongly and statistically significantly decreasing in neighborhood median income. Appendix Table 7 shows that these relationships are slightly more sensitive to the inclusion of controls, though the qualitative conclusion of no real economically meaningful relationship between landlords’ 2020 business practices and neighborhood median income is maintained. Appendix Figure 10 presents results according to neighborhood rent appreciation, which we define as the three-year average (from 2015 to 2018) in the year-over-year change in median gross rent (measured in $100s, data from ACS).We observe no statistically significant or economically meaningful relationship between any of landlords’ business practices and neighborhood rent appreciation, and these results are robust to all specifications (Appendix Table 8). 44 As was the case in 2020, these results largely hold after accounting for basic landlord demographics and additional neighborhood characteristics, indicating that landlord and/or residential sorting also cannot fully explain the 2019 results (Appendix Table 8). ==== Refs References Arnold Chris. Economic Fallout from COVID-19 is Hard on Landlords Too 2020 NPR. May 27 https://www.npr.org/2020/05/27/862904350/economic-fallout-from-covid-19-is-hard-on-landlords-too Airgood-Obrycki Whitney Demers Ben Greene Solomon Herbert Chris Hermann Alexander Luberoff David Wedeen Sophia Renters’ Responses to Financial Stress During the Pandemic The Joint Center for Housing Studies of Harvard University 2021 https://www.jchs.harvard.edu/sites/default/files/research/files/harvard_jchs_renter_responses_covid_airgood-obrycki_etal_2021.pdf Bacher-Hicks Andrew Goodman Josh Mulhern Christine Inequality in Household Adaptation to Schooling Shocks: COVID-Induced Online Learning Engagement in Real Time Journal of Public Economics 193 104345 2021 (January) Bambra Clare Riordan Ryan Ford John Matthews Fiona The COVID-19 Pandemic and Health Inequalities Journal of Epidemiology and Community Health 74 11 2020 964 968 32535550 Bartik Alexander W. Bertrand Marianne Cullen Zoe Glaeser Edward L. Luca Michael Stanton Christopher The impact of COVID-19 on small business outcomes and expectations Proceedings of the national academy of sciences 117 30 2020 17656 17666 Biemer Paul P. Groves Robert M. Lyberg Lars E. Mathiowetz Nancy A. Sudman Seymour Measurement errors in surveys. Vol. 548 2013 John Wiley & Sons Bradburn Norman Sudman Seymour Wansink Brian Asking questions: The definitive guide to questionnaire design 2004 Jossey-Bass San Francisco Chetty Raj Friedman John N. Hendren Nathaniel Stepner Michael The Opportunity Insights Team The Economic Impacts of COVID-19: Evidence from a New Public Database Built Using Private Sector Data 2020 National Bureau of Economic Research Working Paper 27431 Chetty Raj Hendren Nathaniel Jones Maggie R. Porter Sonya R. Race and economic opportunity in the United States: An intergenerational perspective The Quarterly Journal of Economics 135 2 2020 711 783 Choi Jung Hyun Goodman Laurie Mounting Pressures on Mom and Pop Landlords Could Spell Trouble for the Affordable Rental Market Urban Wire: Housing and Housing Finance (blog) 2020 Urban Institute November 10 https://www.urban.org/urban-wire/mounting-pressures-mom-and-pop-landlords-could-spell-trouble-affordable-rental-market Choi Jung Hyun Goodman Laurie Housing Vouchers Have Helped Tenants and Landlords Weather the Pandemic Urban Wire: Housing and Housing Finance (blog) 2021 Urban Institute March 23 https://www.urban.org/urban-wire/housing-vouchers-have-helped-tenants-and-landlords-weather-pandemic Choi Jung Hyun Young Caitlin Owners and Renters of 6.2 Million Units in Small Buildings Are Particularly Vulnerable during the Pandemic Urban Wire: Housing and Housing Finance (blog) 2020 Urban Institute August 10 https://www.urban.org/urban-wire/owners-and-renters-62-million-units-small-buildings-are-particularly-vulnerable-during-pandemic Crane Leland Dod Decker Ryan Flaaen Aaron Hamins-Puertolas Adrian Kurz Christopher Johann Business exit during the COVID-19 pandemic: Non-traditional measures in historical context Finance and Economics Discussion Series 2020-089, Board of Governors of the Federal Reserve System 2020 Cunningham, Mary, Martha M. Galvez, Claudia Aranda, Robert Santos, Douglas A. Wissoker, Alyse D. Oneto, Rob Pitingolo, and James Crawford. 2018. “A Pilot Study of Landlord Acceptance of Housing Choice Vouchers.” Washington, DC: US Department of Housing and Urban Development, Office of Policy Development and Research. Decker Nathaniel. Affordable Housing Without Public Subsidies: Rent-Setting Practices in Small Rental Properties Journal of the American Planning Association 87 1 2021 62 72 Decker Nathaniel. The Uneven Impact of the Pandemic on the Tenants and Owners of Small Rental Properties Terner Center for Housing Innovation, UC Berkeley 2021 July 13 https://ternercenter.berkeley.edu/wp-content/uploads/2021/07/Small-Rental-Properties-Decker-July-2021.pdf DeSilver Drew. As National Eviction Ban Expires, a Look at Who Rents and Who Owns in the US 2021 Pew Research Center August 2 https://www.pewresearch.org/fact-tank/2021/08/02/as-national-eviction-ban-expires-a-look-at-who-rents-and-who-owns-in-the-u-s/ de la Campa Elijah. The Impact of COVID-19 on Small Landlords: Survey Evidence from Albany and Rochester, New York The Joint Center for Housing Studies of Harvard University 2021 https://www.jchs.harvard.edu/sites/default/files/research/files/harvard_jchs_small_landlord_survey_de_la_campa_2021_0.pdf Ellen Ingrid Gould Muscato Brittany Mazzurco Aiken Claudia Reina Vincent Aurand Andrew Yae Rebecca Advancing Racial Equity in Emergency Rental Assistance Programs 2021 NYU Furman Center Housing Initiative at Penn, and the National Low Income Housing Coalition https://furmancenter.org/files/Advancing_Racial_Equity_in_Emergency_Rental_Assistance_Programs_-_Final.pdf Engelhardt Gary Eriksen Michael Housing-Related Financial Distress During the Pandemic Research Institute for Housing America Special Report 2020 2021 Available at SSRN: https://ssrn.com/abstract=3694767 Fang Albert Guess Andrew Humphreys Macartan Can the Government Deter Discrimination? Evidence from a Randomized Intervention in New York City The Journal of Politics 81 1 2019 127 141 Freddie Mac Single Family Rental: An Evolving Market 2018 Freddie Mac Multifamily McLean, VA Freddie Mac Racial and Ethnic Valuation Gaps in Home Purchase Appraisals Freddie Mac Economic & Housing Research Note (blog) 2021 Freddie Mac McLean, VA September Gerardi Kristopher S. Lambie-Hanson Lauren Willen Paul S. Racial Differences in Mortgage Refinancing, Distress, and Housing Wealth Accumulation during COVID-19 Federal Reserve Bank of Atlanta Policy Hub 2021 No. 2021-6 Greene Solomon McCargo Alanna New Data Suggest COVID-19 is Widening Housing Disparities by Race and Income Urban Wire: Housing and Housing Finance (blog) 2020 Urban Institute May 29 https://www.urban.org/urban-wire/new-data-suggest-covid-19-widening-housing-disparities-race-and-income Greig Fiona Zhao Chen Lefevre Alexandra How Did Landlords Fare During COVID? JPMorgan Chase Institute 2021 https://www.jpmorganchase.com/institute/research/household-debt/how-did-landlords-fare-during-covid Hanson Andrew Hawley Zackary Do Landlords Discriminate in the Rental Housing Market? Evidence from an Internet Field Experiment in US Cities Journal of Urban Economics 70 2-3 2011 99 114 Hepburn Peter Louis Renee Desmond Matthew Racial and Gender Disparities among Evicted Americans Sociological Science 7 2020 649 662 Hepburn Peter Louis Renee Fish Joe Lemmerman Emily Alexander Anne Kat Thomas Timothy A. Koehler Robert Benfer Emily Desmond Matthew US Eviction Filing Patterns in 2020 Socius 7 2021 1 18 Immergluck Dan Law Jonathan Investing in Crisis: The Methods, Strategies, and Expectations of Investors in Single-Family Foreclosed Homes in Distressed Neighborhoods Housing Policy Debate 24 3 2014 568 593 JCHS (The Joint Center for Housing Studies of Harvard University) The State of the Nation's Housing 2019 Harvard University https://www.jchs.harvard.edu/sites/default/files/reports/files/Harvard_JCHS_State_of_the_Nations_Housing_2019%20%281%29.pdf Kneebone Elizabeth Murray Cecile Estimating COVID-19’s Near-Term Impact on Renters Terner Center for Housing Innovation, UC Berkeley 2020 April 24 https://ternercenter.berkeley.edu/blog/estimating-covid-19-impact-renters Kneebone Elizabeth Reid Caroline COVID-19 and California's Vulnerable Renters Terner Center for Housing Innovation, UC Berkeley. 2020 2020 https://ternercenter.berkeley.edu/research-and-policy/covid-19-and-californias-vulnerable-renters/ Kneebone Elizabeth O'Regan Katherine Raetz Hayley Underriner Quinn Rent Payments in Affordable Housing During the Pandemic: The Role of Rental Subsidies and the Safety Net 2021 NYU Furman Center https://furmancenter.org/files/publications/Rent_Payments_in_Affordable_Housing_During_the_Pandemic_Terner_Center__Final.pdf Lee Sang Yoon Tim Park Minsung Shin Yongseok Hit Harder, Recover Slower? Unequal Employment Effects of the Covid-19 Shock 2021 National Bureau of Economic Research Working Paper 28354 Local Housing Solutions Expanded Access to Capital for Owners of Unsubsidized Affordable Rental Properties LHS Policy Brief 2021 https://localhousingsolutions.org/housing-policy-library/expanded-access-to-capital-for-owners-of-unsubsidized-affordable-rental-properties/ Lundberg Ian Gold Sarah L. Donnelly Louis Brooks-Gunn Jeanne McLanahan Sara S. Government assistance protects low-income families from eviction Journal of Policy Analysis and Management 40 1 2021 107 127 33814669 McClure Kirk Schwartz Alex F Taghavi Lydia Housing Choice Voucher Location Patterns a Decade Later Housing Policy Debate 25 2 2015 215 233 NMHC (The National Multifamily Housing Council) NMHC Rent Payment Tracker 2020 National Multifamily Housing Council https://www.nmhc.org/research-insight/nmhc-rent-payment-tracker/ Parrot Jim Zandi Mark Averting an Eviction Crisis 2021 Urban Institute https://www.urban.org/sites/default/files/publication/103532/averting-an-eviction-crisis.pdf Raifman Julia Nocka Kristen Jones David Bor Jacob Lipson Sarah Ketchen Jay Jonathan Chan Philip COVID-19 US State Policy Database 2020 Available at www.tinyurl.com/statepolicies Raymond Elora Duckworth Richard Miller Ben Lucas Michael Pokharel Shiraj Corporate Landlords, Institutional Investors, and Displacement: Eviction Rates in Single-Family Rentals.” Federal Reserve Bank of Atlanta Community and Economic Development Discussion Paper 2016 04 16 Reina Vincent Aiken Claudia Epstein Jenna Exploring a Universal Housing Voucher 2021 Housing Initiative at Penn (blog). September 21 https://www.housinginitiative.org/universal-voucher.html Reina Vincent Goldstein Sydney Housing Initiative at Penn Ongoing Challenges for Rental Business Owners in the City of Los Angeles during the COVID-19 Pandemic 2021 Reina, Vincent, Sydney Goldstein, Emily Dowdall, Ira Goldstein, and Elizabeth De Young. 2020. “Residential Rental Property Owner Perspectives and Practices in Philadelphia: Evaluating Challenges during the COVID-19 Pandemic.” Housing Initiative at Penn. Reina Vincent Pritchett Wendell E. Wachter Susan M. Perspectives on Fair Housing 2020 University of Pennsylvania Press Philadelphia Reitsma Marissa Racial/Ethnic Disparities in COVID-19 Exposure Risk, Testing, and Cases at the Subcounty Level in California Health Affairs 40 6 2021 870 878 33979192 Rutan Devin Q. Desmond Matthew The Concentrated Geography of Eviction Annals of the American Academy of Political and Social Science 693 1 2021 64 81 34305147 Schaeffer Katherine. The Most Common Age Among Whites in U.S. is 58 – More than Double That of Racial and Ethnic Minorities 2019 Pew Research Center July 30 https://www.pewresearch.org/fact-tank/2019/07/30/most-common-age-among-us-racial-ethnic-groups/ Shrawder Kevin Aguilar Ricardo Local Area Unemployment Statistics: An Economic Analysis of COVID-19 National Association of Counties 2020 https://www.naco.org/resources/featured/local-area-unemployment-statistics-economic-analysis-covid-19 Stein Sarah Haley Victor Jr. Raymond Elora Lee Woodworth Erik Dill Jessica Special Briefing: Despite CDC Eviction Moratorium, Atlanta-Area Eviction Filings Hit Low-Income, Minority Neighborhoods 2021 Federal Reserve Bank of Atlanta Sudman Seymour Bradburn Norman Response effects in surveys: A review and synthesis 1974 Adline, Chicago, IL Taylor Keeanga-Yamahatta. Race for profit: How banks and the real estate industry undermined black homeownership 2019 UNC Press Books Institute Urban Tracking Rent Payments to Mom-and-Pop Landlords 2021 Urban Institute. https://www.urban.org/features/tracking-rent-payments-mom-and-pop-landlords Zelner Jon Trangucci Rob Naraharisetti Ramya Cao Alex Malosh Ryan Broen Kelly Masters Nina Delamater Paul Racial Disparities in Coronavirus Disease 2019 (COVID-19) Mortality are Driven by Unequal Infection Risks Clinical Infectious Diseases 72 5 2021 e88 e95 33221832

0

PMC9749398

NO-CC CODE

2022-12-16 23:21:37

no

J Hous Econ. 2022 Dec 14;:101904

utf-8

J Hous Econ

10.1016/j.jhe.2022.101904

oa_other

==== Front OpenNano 2352-9520 2352-9520 The Author(s). Published by Elsevier Inc. S2352-9520(22)00080-9 10.1016/j.onano.2022.100118 100118 Review Article Recent Advances of Nanotechnology in COVID 19: A critical review and future perspective Chaudhary Kabi Raj ac⁎ Kujur Sima a Singh Karanvir bc a Department of Pharmaceutics, ISF College of Pharmacy, Ghal Kalan, Ferozpur G.T Road, MOGA, 142001, Punjab, India b Department of Pharmaceutical Chemistry, ISF College of Pharmacy, Ghal Kalan, Ferozpur G.T Road, MOGA, 142001, Punjab, India c Department of Research and Development, United Biotech (P) Ltd. Bagbania, Nalagarh, Solan, Himachal Pradesh, India ⁎ Corresponding author: Kabi Raj Chaudhary, Department of Pharmaceutics, ISF College of Pharmacy, Ghal Kalan, Ferozpur G.T Road, MOGA, 142001, Punjab, India 14 12 2022 14 12 2022 10011824 7 2022 24 11 2022 13 12 2022 © 2022 The Author(s). Published by Elsevier Inc. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. The global anxiety and economic crisis causes the deadly pandemic coronavirus disease of 2019 (COVID 19) affect millions of people right now. Subsequently, this life threatened viral disease is caused due to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). However, morbidity and mortality of infected patients are due to cytokines storm syndrome associated with lung injury and multiorgan failure caused by COVID 19. Thereafter, several methodological advances have been approved by WHO and US-FDA for the detection, diagnosis and control of this wide spreadable communicable disease but still facing multi-challenges to control. Herein, we majorly emphasize the current trends and future perspectives of nano-medicinal based approaches for the delivery of anti-COVID 19 therapeutic moieties. Interestingly, Nanoparticles (NPs) loaded with drug molecules or vaccines resemble morphological features of SARS-CoV-2 in their size (60-140 nm) and shape (circular or spherical) that particularly mimics the virus facilitating strong interaction between them. Indeed, the delivery of anti-COVID 19 cargos via a nanoparticle such as Lipidic nanoparticles, Polymeric nanoparticles, Metallic nanoparticles, and Multi-functionalized nanoparticles to overcome the drawbacks of conventional approaches, specifying the site-specific targeting with reduced drug loading and toxicities, exhibit their immense potential. Additionally, nano-technological based drug delivery with their peculiar characteristics of having low immunogenicity, tunable drug release, multidrug delivery, higher selectivity and specificity, higher efficacy and tolerability switch on the novel pathway for the prevention and treatment of COVID 19. Keywords COVID 19 SARS-CoV-2 Nanotechnology Nano-medicine Nanoparticles therapeutic delivery vaccine delivery ==== Body pmc1 Introduction 1.1 Epidemiology The health condition of the entire globe is in danger right now due to the outbreak of coronavirus disease 2019 (COVID 19) which has already been declared as the Public Health Emergency of International Concern (PHEIC) by the world health organization (WHO) in 30th January 2020 [1,2]. As the 2019-nCoV has been renamed as pandemic coronavirus disease (COVID 19) by WHO on 11th February 2020, which further been renamed as a severe acute respiratory syndrome (SARS-CoV-2) by the international committee on taxonomy of virus (ICTV)[3]. However, the transmission rate of COVID 19 is severe with respect to severe acute respiratory syndrome (SARS-CoV) and middle east respiratory syndrome (MERS-CoV) coronavirus as per many recent studies published in August 2020 [4,5]. The origin place and outbreak year for SARS-CoV was Guangdong province of China in 2002, for MERS-CoV was Arabian Peninsula in 2012 and for COVID 19 was Wuhan city of China in 2019 [6,7] as illustrated in Figure 1 . Unfortunately, patients with COVID 19 are increasing unexpectedly day by day causing an unbearable threat to public health. According to the world health organization (WHO), until March 30th, 2020, 82447 people were infected with the loss of 3310 patients' lives in China with an estimated mortality rate of 4%. Additionally, WHO confirmed that till 10th June 2022, total confirmed case of COVID 19 was estimated as 532,201,219 with 6,305,358 deaths, where Europe (222,417,177) become the leading continent in case of confirmed cases followed by America (158,983,746) western pacific (61,735,224), south east Asia (58,217,287), eastern Mediterranean (21,807,376) and Africa (9,039,645) respectively [8] as shown in Figure 2 . Furthermore, as per the WHO report by 11th June 2022, among the top 8 countries or territories, the United States of America (87,246,309) become the prime area for infection followed by India (43,213,435), Brazil (31,417,341), France (29,753,370), Germany (26,802,782), United Kingdom (22,382,352), Russia (18,369,557) and South Korea (18,218,078) as ascribed in Figure 2. As per the report, 216 countries were infected till 21st September 2020, in addition to this 230 countries of the entire globe have been infected with pandemic COVID 19 till 11th June 2022 [9,10]. The new variant of the corona virus i.e., omicron was first spotted in South Africa and then spread vigorously over other countries like Europe, America, India and 120 other countries till January 2022, respectively [11]. It was reported that this variant consisted of more than 50 mutations when compared to SARS-CoV-2. Among them 30 mutation are responsible for the change in the amino acid spike protein with which it attach and fuse with the cells (ACE 2 receptors). It was also reported that the previous version of the virus have only such 10 mutations with which it can change in the amino acid spikes. From the genomic point of view it has three distinct sub-lineages known as BA.1, BA.2 and BA.3. Although Europe, America and India were severely affected with BA.4, BA.5 and BA2.75 sub variants, respectively [12,13].Figure 1 Three different types of corona virus, showing its origin place including primary and secondary host, outbreak year, affected countries, infected people with total death cases, and fatality rate [6,7]. Figure 1: Figure 2 (a) Global COVID 19 cases showing as continent wise. Europe is the leading continent of infection followed by America, the Western Pacific, South East Asia, Eastern Mediterranean and Africa.s, (b) COVID 19 infected eight highly burdened countries. The United States of America is the leading country subsequently followed by India, Brazil, France, Germany, the United Kingdom, Russia and South Korea. Figure 2: 1.2 Mode of transmission and clinical manifestation The highly contagious and exponentially spreadable COVID 19 with zoonotic origin has single-stranded RNA and has the potential to transmit from one person to another. Many studies suggested that birds or mammals are regarded to be reservoirs of COV, with an identity of exhibiting 88% birds-derived SARS-like CoV as a genomic sequence of COVID 19 [14,15]. Sneezing or coughing via a human from which suspended respiratory droplets are generated is regarded as the primary well-established mode of COVID 19 transmission. Droplet nuclei with diameters of < 5-10 µm are more effectively transmitted in which virus can persist for 2 hours to a few days [16,17]. Besides this, direct (hand-to-hand shake) or indirect (through medical accessories, personal utensils) contact of virally infected hands into the mouth or nose or even eyes further lead to infection. Moreover, oral-fecal contamination of infected patients further leads to its transmission. Henceforth, isolation of an infected individual from the population, personal hygienic behavior, applying well safety measures, wearing masks and gloves, keep quarantine of infected persons are some preventive measures that ultimately lead to control [18,19]. Once the person gets infected with COVID 19, most of the cases showed manifestation within 2-14 days while some cases prevail after 27 days. Although, approximately 5.6 days of incubation period has been identified by a group of many researchers [20,21]. Various research groups revealed that older people (especially males) age of above 60 and children are more susceptible to SARS-CoV-2 [22,23]. As per the basis of observing symptoms in the infected patient, the most common symptoms like fever, dry cough and fatigue, sputum production, and shortness of breath are predominant over another type of manifestation. Similarly, less common symptoms like prevail the aches and pains, conjunctivitis, sore throat, diarrhea, headache, loss of taste or smell and rashes on the skin. However, long lasting serious manifestations like respiratory failure, multiorgan failure, cardiac failure, and renal failure are identified as the primary cause of mortality and morbidity [24], [25], [26], [27], [28], [29]. More importantly, increased levels of pro-inflammatory cytokines and chemokines including granulocyte-colony stimulating factor, monocyte chemoattractant protein-1, and chemokine ligand-3 are observed in COVID 19 patients promoting viral survivability and hence exacerbating viral disease [30], [31], [32], [33]. 1.3 Virology and Pathogenesis of SARS-CoV-2 Coronaviruses are crown-shaped single-strand RNA virus with a size range of 80-160 nm that falls under the order-nidovirus, family-coronaviridae and subfamily-coronavineae. Moreover, they are classified as α, β, γ and δ genus, among which α genus (that contains HCoV-22E and NL63) and β genus (that contains HKU1, 229E, OC43, MERS-CoV, SARS-CoV and the latest outbreak SARS-CoV-2) majorly affects to mankind [34], [35], [36], [37]. Viral components like single-strand RNA, envelop protein, nucleocapsid protein, spike glycoprotein and membrane protein gives the characteristics of shape, function, rigidity and survivability to coronavirus. Spike (S) glycoprotein forms the crown-like structure at the outer surface of the virus that consists of two subunits, S1 which facilitates the binding of human angiotensin-converting enzyme-2 (hACE2) receptors and S2 which facilitates fusion between the host and the viral cell membrane. Interestingly, mutational behavior of receptor-binding domain (RBD) of S protein and the presence of polybasic furin cleavage sites (that facilitates the strong binding of S-glycoprotein to hACE2) and O-linked glycans are genomic features of SARS-CoV-2 that play an important role for binding efficacy to the host [38], [39], [40], [41], [42]. In addition, entry of SARS-CoV-2 virus into human lungs where ACE 2 receptors exist, gets bind with viral RBD of S1 subunit of S-glycoprotein that ultimately leads to downregulation of ACE2 receptors. This subsequent down regulation of ACE2 receptors further leads to increased production of angiotensin-2 (AT2) induces pulmonary vascular permeability and causes lung injury [43,44]. Further, the interaction between antigen presenting cells (APC) of SARS-CoV-2 and the dendritic cell of the host causes Macrophagic stimulation that subsequently leads to severe immunological reaction. After all, these immunogenic reactions further cause the production of excess pro- inflammatory cytokines (IFN-α, IFN-γ, IL-1β, IL-6, IL-12, IL-18, IL-33, TNF-α,) and chemokines (CCL2, CCL3, CCL5, CCL8, CCL9, CCL10, etc.) called cytokines storm which results for epithelial cell lining damage than enters into the bloodstream and finally exerts multiorgan damage as shown in Figure 3 [45], [46], [47], [48].Figure 3 Pathogenesis and mechanism of SASR-CoV 2 infection exhibiting six different steps; 1). Entry of Corona virus and infects lung cells, 2). Production of cytokines after identification of virus via immune cells including macrophages 3). Production of cytokine storm due attraction of more immune cells like white blood cells and causes inflammation. 4). Formation of fibrin, causing lung cell damage, 5) Entry of fluid into lungs cavities through weakened blood vessels causing respiratory failure, 6). Circulation of a virus into various parts of the body through systemic circulation causes multiorgan damage. Figure 3: 1.4 Conventional Diagnostic approach for COVID 19 Early diagnosis of COVID 19 is equally important to save the patient's life. Luckily, various diagnostic approaches have been started to adopt which protect the patient's life from worsening. Screening of this pandemic COVID 19 can be done by below mentioned recommended diagnostic tools. 1.4.1 Infrared scanner and thermal camera An infrared scanner is used to scan individuals, while thermal cameras are used to detect incremental body temperature by which COVID 19 affected population can be isolated from a large population. These scanners and cameras are mainly used in crowded areas like hospitals, airports, academic institutions, railway stations and research centres through which visual images can be observed by converting and detecting infrared energy in terms of heat [49], [50], [51]. 1.4.2 Nucleic acid amplification test (NAAT) NAAT is WHO established based diagnostic approach in which a nasal swab or blood sample can be used to confirm COVID 19 via real-time fluorescence polymerase chain reaction (RT-PCR). Likewise, several diagnostic kits have been launched by the US-FDA and other various regulatory bodies as an emergency kit for the detection of COVID 19 that works based on RT-PCR technology. The first commercial diagnostic kit was CobasR SARS-CoV-2, which was been launched under the clinical laboratory improvement amendments of 1988 (CLIA) specifying the fulfilment of required medical emergency needs [52], [53], [54], [55], [56]. However, numerous complications have been identified associated with false identification, complex sample preparations, less sensitivity and selectivity for low viral load cases, and stability issues along with its time-consuming procedure that limits its use [57]. 1.4.3 Computerized tomography (CT) scan Due to several drawbacks of NAAT associated with false reading and detection for COVID 19, Chinese researchers recommended computerized tomography imaging as a primary diagnostic approach. Huang et al., reported that after examination of the suspected patient's chest CT imaging showed multiple peripheral ground glass opacities in both lungs leading to hospitalization after three days which initially was observed with negative results via RT-PCR. Thereafter, this technique is being highly applied for the detection of COVID 19-associated viral pneumonia [58], [59], [60]. 1.4.4 Immunoassay-based diagnostic approach IgM-IgG combined antibodies detection-based approach against SARS-CoV-2 in COVID 19 infected patients was reported by Li et al. Hopefully, the group has developed an immunoassay-based diagnostic test kit with rapid screening (within 15 mins) and a simple point of view [61], [62], [63]. 1.4.5 BioFire COVID 19 Test 2 This approach has been adopted by FDA under Emergency Use Authorization (EUA) since March 2020, which is based on a molecular diagnostic test. A swab sample taken from the nasopharyngeal region is being detected to confirm the presence of SARS-CoV-2 [64,65]. 2 Current inline treatment After the emergence of the deadly pandemic COVID 19, today's whole world's health organizations, medicinal industries, health institutes and scientists are eagerly searching for potent anti-COVID medicines or vaccines. Numerous preclinical and clinical trials are on-going for both newly developed and existing drug molecules. Most antiviral drugs including other disease-specified drugs are continuously been tested for their anti-COVID activity. Solidarity clinical trials launched by WHO and partners have specified wide varieties of on-going clinically trial therapeutic candidates to determine the efficacy against coronavirus along with its safety for human use [66,67]. Additionally, US-FDA and other various regulatory bodies have equally engaged in research work on anti-COVID moieties and medical countermeasures (MCMs). Numerous anti-COVID 19 drug molecules have been approved by FDA under Emergency Use Authorization (EUA) and recommended by WHO for both hospitalized (inpatients) as well as non-hospitalized patients (outpatients)[68,69]. Hopefully, the number of therapeutic moieties is aggressively being investigated to determine their effectiveness against COVID 19. Preclinical and Clinical trials are progressively being conducted by many regulatory bodies including health institutions, government health agencies and other various researchers of antivirals, monoclonal antibodies, corticosteroids, steroids, anti-inflammatory and other molecules against coronavirus. Interestingly, the solidarity trial committee has been organized by WHO and partners to expedite potential treatment for COVID 19 by conducting clinical trials. Although, the Solidarity trial committee has withdrawn hydroxychloroquine and lopinavir/ritonavir from their trial phase on 4th July 2020 due to their greater adverse effects (like arrhythmia)[66,[70], [71], [72]]. According to a report published by the U.S Department of Human Health Services (HHS) on 24th January 2022, REGEN-COV (Bamlanivimab and etesevimab, casirivimab and imdevimab combinations) was not been found as an effective drug against Omicron variant of COVID 19. Similarly, as per a CDC report published in December 2021, Janssen/ Johnson & Johnson vaccine was less likely to be preferred over Pfizer and Moderna vaccines due to a greater risk of developing severe (but rare) blood clot called thrombosis with thrombocytopenia syndrome (TTS)[73,74]. However, adverse effect of both Pfizer and Moderna's vaccines with the rare occurrence of myocarditis has also been established as per the report suggested by CDC but remains preferable over other due to its higher benefits that outweigh their risks [75,76]. Thereafter, focusing on all these serious toxicities along with intolerable adverse effects of recommended drugs and vaccines for COVID 19, full attention has been gained towards the development of novel Nano-technological based delivery. Nanotechnologies with a great significance specifying the characteristics of rapid and accurate diagnosis, target-specific delivery with reduced dosing frequency and hence toxicities, combinatorial therapy, tunable drug release, biocompatibility, low immunogenicity, multidrug delivery with high selectivity and specificity make the nano therapeutic strategy more significant over conventional delivery as shown in Figure 4 [77], [78], [79]. In this review, we broadly investigated the current approach and future possibilities of a nano-technological based approach for diagnosis, delivery of potential Anti-COVID 19 therapeutic moieties and vaccine delivery as well.Figure 4 Nano-technological based strategy for the Diagnose, treatment and prevention of pandemic SARS-CoV-2 infections. Figure 4: 3 Nanoparticle-based diagnostic approach against COVID 19 Early detection of coronavirus in an infected patient emphasis more important rather to treated them later. To obtain accurate and positive diagnostic results from every patient via a conventional approach is somewhat critical that ultimately limit its use. Besides this, other limitations like the time-consuming process and low efficiency for viral detection through the most popular and widely used RT-PCR method greatly hinder its worldwide acceptance [80,81]. Thereafter multipronged scientific strategy with nanoparticle-based technique seems to be highlighted and needs to be addressed for effective and accurate diagnostic examination. In addition to this, the use of nanoparticles (like metal NP, magnetic NP, quantum dots, etc.) in RT-PCR further creates the track for use in other viral detection methods as well like Enzyme-Linked Immunosorbent Assay (ELISA), Reverse Transcription Loop-mediated isothermal Amplification (RT-LAMP) with a hope to brings the array of greater accuracy, sensitivity and efficiency due their peculiar features of high surface area and ultra-small size. Moreover, unique optical and electrochemical properties including its biocompatible and stable feature enforce the nano-technological field to be investigated in depth for biological application [82], [83], [84], [85], [86]. Various researchers have attempted the use of nano-technological tools to diagnose the COVID 19 infected patients as ascribed in Table 1 .Table 1 Metallic nanoparticles to diagnose COVID-19 Table 1:Studied by Diagnostic agent Method Purpose of study Finding/outcomes References Li H and Rotherber et al., Citrate ion fabricated AuNPs Colorimetric method To measure the efficiency of interaction for AuNPs with viral nucleic acid Change in colour of colloidal solution for dsDNA but not for ssDNA/ssRNA. [87] Moitra et al., Thiol modified Antisense oligonucleotide impregnated AuNPs (TASO-AuNPs Colorimetric assay To determine the propensity of interaction between TASO-AuNPs with viral RNA sequence Change in colour of colloidal solution due aggregation of AuNPs. [88] Aithal et al., Aptamer functionalized AuNPs (Nanoprobe ELISA method To find out the degree of aptamer binding with S-protein of SARS-CoV 2 Unconjugated nanoprobes agglomerate with colloidal solution's colour change and vice versa. [89] Kim et al., Thiolated ssDNA functionalized AuNPs (t-ssDNA-AuNPs Colorimetric assay Aggregatory capability of t-ssDNA-AuNPs with target DNA Formed dsDNA-AuNPs networks cause agglomeration with colloidal solution's colour changes. [90] Fuet al., Polyclonal antibody functionalized Au@Pt NPs Colorimetric assay Exploration of viral enzyme catalytic pathway due to interaction between S1-protein and Au@Pt NPs Enhancement of sensitivity and selectivity for S1 viral protein due to increment of porous core shell nanostructures on the surface of AuNPs. [91] Karakuset al., Recombinant SARS-CoV 2 spike antibody fabricated AuNPs (AuNPs-mAb) Dual colorimetric and electrochemical method To estimate the SARS-CoV 2 spike Antigens Colloidal solution's colour changes due aggregation of AuNPs. Both techniques were highly sensitive against SARS-CoV 2 Antigen compared to H1N1 and MERS-CoV. [92] Teengamet al., Pyrrolidinyl peptide nucleic acid incorporated AgNPs (acpcPNA-AgNPs Colorimetric assay Agglomeration efficiency of AgNPs with acpcPNA DNA-acpcPNA-AgNPs aggregates cause change in colour of colloidal solution leading to ease detection of target Oligonucleotide. [93] Zhang Z et al., Bromine ions and acetonitrile fabricated AgNPs Surface Enhanced Raman Spectroscopy (SERS technique To measures the detectable and quantifiable capability of modified AgNPs for SARS-CoV 2 Modified AgNPs showed improved quantifiable efficiency for SARS-CoV 2 with lower detection limit of 100 copies/test within 1-2 mins. [94] Gong et al., Silica coated superparamagnetic NPs (si@SPMNPs Si@PMNPs-PCR assay To explore the effectiveness of developed si@SPMNPs against viral target cDNA of SARS-CoV Observed to be excellent selectivity and sensitivity for cDNA with detection limit of 2.0 × 103 copies within 6 hrs. [95] Zhoe et al., Polycarboxyl fabricated Magnetic NPs (pcMNPs Colorimetric assay For viral RNA extraction potentiality of SARS-CoV 2 pcMNPs showed rapid (within 30 mins) and improved RNA extraction capability (10-copies sensitivity with 5-log enhanced linearity). [96] Somvanshi et al., Multifunctionalized magnetic NPs (MNPs Colorimetric assay Viral RNA extraction efficiency of Multifunctionalized MNPs Shortens the time period and necessities for viral detection compared to conventional diagnostic tool. [97] Roh et al., RNA aptamer conjugated Quantum Dots (apt-QD Confocal laser scanning microscopic (CLSM)- Biosensor For the detection of N-protein of SARS-CoV Detection limit for SARS-CoV with 0.1 pg/ml paves the effective pathway for diagnosis of SARS-CoV 2 as well. [98] 3.1 Metallic nanoparticle-based diagnostic approach More specifically, localized surface plasmon resonance (LSPR), an optical property developed by gold (Au), silver (Ag) and copper (Cu) NP that represents a color with maximal absorbance wavelength is particularly associated with bio sensing application [99]. In particular, the degree of NP aggregation causes a change in tunable light absorption and scattering of wavelength in the visible region with a change in LSPR extinction maxima of metal NP that results in an obvious change of colloidal solution color as detected by the necked eye [100]. In the current situation, both the gold nanoparticle and silver nanoparticles are vigorously been investigated for their biomedical application. However, several studies suggested that the optical properties of AgNPs were found to be improved due to their higher extinction coefficient compared to AuNPs. In addition, the shape and size of different metallic NPs further determine their optical and electrochemical propensities [85]. The colorimetric assay can be performed for target viral DNA or RNA detection for clinical diagnosis and can be accomplished by the aggregation of AgNPs with subsequent color changes of colloidal solution [101,102]. 3.1.1 Gold nanoparticles In this regard, Li H and Rothber L, suggested efficiency and degree of aggregation for gold NP (AuNPs) after interaction with viral single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA) are completely different due to variations in both viral electrostatic propensities. The study was performed to determine SARS-CoV via a colorimetric hybridization assay with found a lack of AuNPs aggregation after treatment with ssDNA or ssRNA even after the addition of salt. This is happened due interaction of ssDNA or ssRNA with citrate ions fabricated AuNPs that results in the stabilization of NP and hence prevents change in the color of colloidal solution. However, a change in color of colloidal solution was observed with dsDNA, indicating aggregation of NPs. The phenomenon for detection was simple and less time-consuming (within 10 mins) which is a more suitable diagnostic approach against existing and upcoming variants of COVID 19 [87]. In a similar study, Moitra et al., developed thiol modified Antisense Oligonucleotide (T-ASO) fabricated gold NPs (T-ASO-AuNPs). The technique was based on the assessment of the N-gene (nucleocapsid phosphoprotein) of SARS-CoV-2 via a colorimetric assay in which the target viral RNA sequence was treated with T-ASO-AuNPs that resulting in the agglomeration of treated gold NP due to change in LSPR. In particular, this agglomeration cause change in the color of the colloidal solution and the precipitate of these agglomerates can be visually seen by necked eyes after the addition of ribonuclease H (RNase H). Henceforth, this short time-consuming technique (within 10 min after the RNA isolation process) with a detection limit value of 0.18ng/µl may create a novel pathway for an accurate, selective, sensitive and more reproducible diagnostic approach [88]. Recently, another study was performed by Aithal et al., based on aptamer functionalized AuNPs (nanoprobes) for the detection of SARS-CoV-2. Herein, the degree of aptamer binding potency with spike (S) protein of SARS-CoV-2 was estimated by the ELISA technique. Interestingly, nano probes without conjugated with S-protein agglomerates while conjugated nano probes do not due to the stabilization of nano probes after the addition of coagulants (MgCl2 salt solution). These agglomerated nano probes result in a change in the colloidal solution's color with a change in plasmon resonance. Collectively, this short time-consuming diagnostic method with a detection capability of nano-sensors was observed to be 3540 genome copies/µl and higher concentrations of inactivated SARS-CoV-2 at the absorbance of 540 nm, indicating fruitful viral screening and detectable tool for upcoming days [89]. In the same manner, Kim et al., previously proposed a colorimetric assay for the detection of MERS-CoV via hybridization of thiolated ssDNA functionalized gold NP(t-ssDNA-AuNPs) with target DNA (that act as a linker between two ssDNA functionalized AuNPs). In the study, formed dsDNA-AuNPs networks causing aggregation of AuNPs resulted in a dramatic color change of colloidal solution indicating that the use of gold nanoparticles could be effective for analysis of SARS-CoV-2 as well, without using any complicated and sophisticated instruments [90]. The next study performed by Fu et al., demonstrated the colorimetric detection of target viral S1 protein of SARS-CoV-2 by the introduction of polyclonal antibody functionalized Au@Pt NPs based on peroxidase catalysis activity. More deeply, the initial fabrication of reduced Pt4+ions onto the surface of AuNP further creates porous core-shell nanostructures and leads to a remarkable change in enzyme catalysis pathway i.e., from ROS generation to fast electron transfer (FET) process which enables sensitive and more selective identification of viral S1 protein. Notably, the linear detection range for viral spike protein was ascribed to 10-100 ng ml−1 with a limit of detection (LOD) of 11 ng ml−1 within 15 mins. After all, the study is straightforward towards the development of a metallic NP fabricated colorimetric biosensor based on nano-enzyme catalysis for practical applications [91]. Similarly, Karakus et al., developed a monoclonal antibody (recombinant SARS-CoV-2 spike antibody) fabricated gold NPs (AuNPs-mAb) for the estimation of SARS-CoV-2 spike antigen via dual colorimetric and electrochemical analytical technique. Moreover, the study anticipated that after optical visualization, changes in the color of the colloidal solution from red to purple were due to irreversible aggregation of AuNPs in the presence of SARS-CoV-2 antigens. Interestingly, the author developed an electrochemical method for the first time in such a way that SARS-CoV-2 spike antigen can be detected without the requirement of sensor preparation and its modification that subsequently reduces the time-consuming process (within 10 mins). Additionally, both methods were found to be highly sensitive against SARS-CoV-2 antigen irrespective of another infectious viral antigen like H1NI, MERS-CoV, and Streptococcus pneumonia with a detection limit of 48ng/ml and 1pg/ml for colorimetric and electrochemical method respectively [92]. In addition to this, Yano et al., invented large AuNPs-enhanced surface plasmon resonance (AuNPs-SPR) for ultra-sensitive detection of SARS-CoV-2 associated N-protein. Herein, SPR sensitivity of two different sizes of AuNPs of 150nm and 40nm were monitored. The study revealed that AuNPs with large size of 150nm displayed significantly greater sensitivity against viral N-protein with a detection limit 4pg/ml compared to small size particle of 40nm as well as with typical RT-PCR (detection limit of 4.5pg/ml). In this regard, novel SPR-based nanoparticles could have a remarkable capability to diagnose pandemic viruses [93]. Additionally, Silva et al., explored the efficiency of bio-conjugation of Antibody functionalized AuNPs (pAbS1N@AuNPs) with SARS-CoV-2 spike glycoprotein. However, successful development and bio-conjugation of spike glycoprotein with pAbS1N@AuNPs were confirmed through physicochemical characterizations (like FTIR, TEM, UV-VIS spectroscopy and SERS). Most interestingly, a study demonstrated that interaction of pAbS1N@AuNPs with influenza viruses showed no increment in the size of bio-conjugates while with SARS-CoV-2 spike glycoprotein exhibited size enhancement after dynamic light scattering (DLS) measurement, indicating better selectivity for SARS-CoV-2 [103]. In the next investigation, Diaz et al., developed colorimetric sensor-based AuNPs for the detection of SARS-CoV-2 coding sequences (RdRp, E and S protein). More specifically, developing colorimetric sensor-based AuNPs took only 2.5 hrs for complete amplification and detection with a viral load of ≥103-104 viral RNA copies/µl in the patient sample. Forwards, rapid visual detection of SARS-CoV-2 sequences (within 15 mins) through colorimetric sensor-based AuNPs with better stability up to several months and its simplified version further restrict its use for diagnostic purposes against COVID 19 [104]. 3.1.2 Silver nanoparticles Recently, Teengam et al., synthesized the multiplex paper-based colorimetric DNA sensor using pyrrolidinyl peptide Nucleic acid (acpcPNA) conjugated AgNPs (acpcPNA-AgNPs) for the quantitative detection of MERS-CoV along with MTB and HPV oligonucleotides. Herein, aggregation of citrate ion stabilized AgNPs was observed as a result of acpcPNA probe in the absence of complementary DNA. Moreover, DNA-acpcPNA duplex mediated AgNPs dispersion due to electrostatic repulsion in the presence of target DNA causes a significant color change through which target oligonucleotide can be detected easily. Overall, the study with finding a detection limit of 1.53 for MERS-CoV along with improved hybridization efficiency, rapid intense color change, simplified assay estimation and superior sensitivity and selectivity of target DNA detection showing a remarkable and efficient approach for diagnosis of SARS-CoV-2 as well [105]. Going forwards, Zhang et al., engineered bromine ions and acetonitrile fabricated AgNPs based on surface-enhanced Raman spectroscopy for the rapid detection and quantification of SARS-CoV-2 including Human Adenovirus 3 and H1N1 virus. Herein, an author suggested that the viral quantification capability of such modified AgNPs was significantly improved with a lower detection limit of 100 copies /test, with a principal component analytical duration of 1-2 mins for those three viruses. Interestingly, the impregnation of acetonitrile on the surface of AgNPs greatly intensified and amplified the SERS peak signals of those viruses. Additionally, stability and fingerprint for those developed nanoparticles further showed an improved version with preferable cost effectivity as compared to other detection methods, enabling the tractable solution for accurate and immense diagnosis of COVID 19 [106]. 3.1.3 Magnetic nanoparticle Beyond the use of metallic nanoparticles, the potential use of magnetic nanoparticles for selective and sensitive diagnosis of the SARS-CoV-2 virus has further drawn attention in the biopharmaceutical field. Iron oxide nanoparticles among various MNP are immensely investigated for their use in diagnostic purposes against deadly infectious viral diseases due to their higher magnetic efficiency with simple synthesis approaches [94,95]. Various studies ascribed that the use of iron oxide NPs for extraction of target DNA and RNA showed an efficient strategy for diagnostic purposes. Lee et al., in 2018 anticipated the zika viral RNA extraction using iron oxide silica NPs [96]. Additionally, Wang et al., 2017, further performed the study by using silica-coated MNP for the simultaneous extraction of DNA and RNA from hepatocellular carcinoma cells [97]. Previously, Gong et al., demonstrated the study for the detection of SARS-CoV using silica coated-superparamagnetic NPs (si@SMNPs) fabricated PCR-based assay. Herein, magnetic-conjugated dsDNA complex was developed by the introduction of an oligonucleotide probe fabricated silica-coated SMNPs with viral target cDNA, which then subsequently allowed to form enriched cDNA through simple magnetic separation phenomenon followed by amplification of that enriched cDNA through PCR using another magnetic separation step. After all, by the application of silica-coated florescent NPs (SFNPs), amplified viral target cDNA was been detected through sandwich hybridization assay. At last, the study revealed that selectivity and sensitivity for viral target cDNA of SARS-CoV were observed to be excellent with a detection limit of 2.0 × 103 copies within 6 hrs establishing a suitable track for clinically diagnostic track [107]. In an investigation, Zhao et al., synthesized poly-carboxyl fabricated magnetic nanoparticles (pcMNPs) for the effective and sensitive extraction of viral RNA using SARS-CoV-2 pseudo-virus as a model. However, the developed pcMNPs dramatically simplified RNA extraction phenomenon over conventional MNPs and normal RT-PCR by combining lysis and binding steps into one with subsequent introduction of pcMNPs-RNA complex into RT-PCR reactions that ultimately reduce time-consuming as well (completed within 30 mins). Importantly, applied external magnetic force subsequently causes rapid and easier extraction of pcMNPs-RNA complex from the solution as pcMNPs to possess magnetic property. Moreover, the pcMNPs-based approach showed 10-copy sensitivity with immense improved linearity over 5 logs of gradients, indicating superior viral RNA binding performance [108]. Furthermore, Somvanshi et al., highlighted the applicability of multifunctional nano-magnetic particles (MNPs) for the detection of COVID 19 via assessing the viral RNA extraction potentiality. Herewith, silica-coated zinc ferrite magnetic nanoparticles (ZiF@Si@ MNPs) developed by sol-gel auto combustion technique were simultaneously fabricated with carboxyl modified polyvinyl alcohol followed by accumulation of magnet (ZiF@Si@NH2@Cpoly MNPs). The successful synthesis of surface-functionalized MNPs was confirmed after physicochemical characterization. The developed MNPs impressively shorten the operation period and necessities as compared to normal diagnostic tools for the detection of COVID 19 [98]. 3.1.4 Quantum dots Currently, the use of QDs for the detection of verities of pathogenic bacteria and viruses in humans become a keen subject for many researchers. Some unique features for optical and electrical properties including extraordinary plasmonic properties of QDs have attracted attention for their use in biomedical fields. As per several studies, effective diagnosis, detection and control of the deadliest viral infections including pandemic SARS-CoV-2 can be accomplished by the use of QDs as screening agents [109,110]. In one study, Roh et al., developed a QDs-conjugated RNA aptamer for the detection of nucleocapsid (N) protein of SARS-COV on an immobilized protein chip based on the optical signal variation. The study with finding a detection limit of 0.1 pg/ml by detecting fluorescent emission intensity through confocal laser scanning microscopy concluded that this developed biosensor was very sensitive, highly selective and easy to use against SARS-COV protein which further directs the study one step forward for effective clinical diagnosis of novel SARS-CoV-2 antigen as well [111]. 3.2 Carbon-based Nano-materials for diagnostic approach Similar to the applicability of metallic NPs, carbon-based nanomaterials (CBNMs) such as graphene, carbon nanotubes (CNTs), and their derivatives are equally being investigated for the diagnostic and drug or vaccine delivery purposes, against broad spectrum single envelop RNA viruses including SARS-CoV-2 as depicted in figure 5 . Importantly, CBNMs constitute promising alternatives due to exhibiting their peculiar characteristics of biodegradability, biocompatibility, low to no toxicity and tissue regeneration inducement capability [112,113]. Besides, CBNMs with excellent thermoelectrical and mechanical conductivity, higher surface area, and better functionalization potentiality with targeted molecules, are equally enforced for their potential use in bio-medicinal fields [114].Figure 5 Carbon Based Nano material for diagnosis, drug and vaccine delivery, treatment and prevention of COVID 19 Figure 5: 3.2.1 Graphene-based materials Graphenes are two-dimensional hexagonally bonded CBNMs, exhibiting outstanding characteristics of large specific surface area, greater electromechanical and thermal conductivity with improved optical and catalytic properties [115]. Moreover, graphene and its derivatives have immense applicability in the biomedical and biotechnological fields for bio-sensing and bio-imaging, early diagnosis, drug screening and conventional textile utilities [116]. Recently, Seo et al., demonstrated the spike protein antibody fabricated graphene-based field effect transistors (FET) biosensor for the efficient detection of COVID 19. Herein, sensitivity and selectivity for detection of SARS-CoV-2 spike protein were found to be superior in COVID 19 patients with a limit of detection (LOD) of 1fg ml−1, irrespective to control subjects [117]. Similarly, the next study performed by Li et al., further identified the viral detection potentiality against influenza and SARS-CoV-2 viruses through a developed MXene- graphene FET sensor. Interestingly, the study suggested that the developed sensor showed ultra-sensitivity and a short time-consuming phenomenon (~50 ms), towards the determination of viruses with LOD of 125 copies ml−1 and of 1fg ml−1, for influenza virus and recombinant 2019-nCoV spike protein respectively [118]. Additionally, El-Said et al., synthesized AuNPs incorporated reduced porous graphene oxide (rGO) to enhance the biosensing efficiency of ITO electrode, against COVID 19 S- protein. However, the experiment suggested that the developed Spectro-electrochemical biosensor exhibited potential selectivity and sensitivity against viral protein with LOD of 39.5 f mol−1, indicating an efficient alternative for early diagnosis of COVID 19 [119]. Similarly, Shahdeo et al., experimented with the ultrasensitive immuno-sensing of spike S1 antigen (S1-Ag) of SARS-CoV-2 through a developed graphene-based FET sensor (G-FET). Notably, immobilization of SARS-CoV-2 spike S1 antibody (S1-Ab) as a sensing element onto the surface of the G-FET sensor has superior S1-Ag binding potentiality. Furthermore, sensitivity and selectivity of the G-FET sensor against SARS-CoV 2 S1-Ag were found to be superior with LOD of 10 fM [120]. In the next study, Payandehpeyman et al., developed SARS-CoV 2 spike S1 antibody (S1-Ab) fabricated graphene-based nanoresonator sensor for rapid and effective detection of viral S1-Ag. Importantly, the study suggested that immobilized S1-Ab over a single-layer graphene sheet (SLGS) surface showed strong viral S1-Ag capturing capability. However, the sensitivity and detection ability of such SLGS modified G-FET sensor further depends upon the geometry of SLGS. The author showed that the detection capacity of the developed sensor against SARS-CoV 2 was ranging from 10 to 1000 viruses per test, with an LOD of 10 viruses per test [121]. In addition to this, the next study performed by Ang et al., developed Graphene Oxide Nanocolloids (GONC) for the detection of 2019 nCoV targeted sequences. In the study, an author demonstrates the quantitative correlation between the inherent electro-activity of GONC immobilization platform and 2019 nCoV concentrations, showing analyte detection capability of 10−10 to 10−5 M. Moreover, the developed geno-sensor showed the potentiality of rapid and effective viral genome sensing affinity and could be integrated into a DNA amplifier for early SARS-CoV 2 diagnosis [122]. In another investigation, Li et al., synthesized AuNPs decorated G-FET sensor (AuNPs@G-FET) for the detection of SARS-CoV-2 RNA in human throat swab sample. Importantly, the surface of AuNPs@G-FET sensor was initially immobilized with phosphorodiamidate morpholino oligos (PMO) probes. The PMO modified-AuNPs@G-FET sensor showed higher sensitivity and selectivity against SARS-CoV-2 RdRp, with LOD of 2.29 fM in the throat swab sample. More importantly, the developed sensor has amplification-free direct SARS-CoV-2 RdRp detection capability within a very short time (2 min), with ultrasensitive potential to distinguish among another SARS-CoV RdRp sequence [123]. In the next research, Zhao et al., synthesized a calixarene functionalized ultrasensitive super sandwich-type electrochemical sensor to detect the RNA of SARS-CoV-2. Interestingly, an ongoing study revealed that the developed sensor was highly sensitive and specific towards the detection of viral RNA, specifying an LOD of 200 copies/ ml for a clinical specimen. Moreover, the viral RNA detectable ratio of that developed sensor was observed to be higher (85.5% and 46.2%) as compared to RT-qPCR (56.5% and 7.7%), signifying the novel rapid, accurate, and ultrasensitive approach to detecting pandemic SARS-CoV-2 [124]. 3.2.2 Carbon nanotube-based materials Carbon nanotubes (CNTs) are other promising carbonaceous materials with immense utility in the biochemical field. Various peculiar properties such as biocompatible, non-cytotoxic, greater surface area, ROS-producing capability, and mechanochemical resistance, enhance the significant attribution in Nano-technological aspects [125,126]. Recently, Thanihaichelvan et al., proposed CNT- FET biosensor by immobilizing the RNA-dependent RNA polymerase gene of SARS-CoV-2 onto the CNT channel. The synthesized CNT- FET biosensor showed the LOD of 10 fM with selective sensing response, against a positive target viral sequence, indicating a reliable, fast and easy approach for diagnosis of SARS-CoV-2 [127]. Similarly, the next study carried out by Zamzami et al., further demonstrate the CNT-FET electrochemical sensor to detect the S1 antigen of SARS-CoV-2, via immobilization of SARS-CoV-2 S1 antibody onto the surface of CNT channels, using 1-pyrenebutanoic acid succinimidyl ester (PBASE) as a linker. Interestingly, CNT-FET electrochemical sensor exhibited strong selectivity, sensitivity and accuracy against SARS-CoV-2 S1 antigen with LOD of 4.12 fg/ml, while, no response was observed against SARS-CoV-1 S1 and MERS-CoV S1 antigen [128]. Likewise, Pinals et al., highlighted the rapid and sensitive detection capability for SARS-CoV-2 spike protein through developed ACE-2 functionalized single-walled CNTs (ACE2-SWCNTs) optical nanosensor. Interestingly, ACE2-SWCNTs nanosensor showed 2-fold enhancement of fluorescence within 90 min, after exposing SARS-CoV-2 spike protein. Moreover, ACE2-SWCNTs nanosensor further exhibited good colloidal stability with improved retained sensing capacity, showing 73% fluorescence for 35 mg/L SARS-CoV-2 virus-like particles, within 5 sec, indicating an effective novel toolkit to combat COVID-19 [129]. Concurrently, Kim et al., performed optimization of carbon nanotube Thin-Film Immunosensor (CNT-TFI) by a computational method for rapid and effective diagnosis of SARS-CoV-2 virus. Herein, the optimized immunosensor was found to be higher sensitive against SARS-CoV-2, with LOD of 0.024 [fg/ ml]−1 in buffer solution and 0.048 [copies/ ml]−1 in the lysed virus. Thus, an author concludes that this CNT-TFI strategy could be a very effective, accurate, and sensitive technique for SARS-CoV-2 diagnosis, without the need for molecular amplification [130]. In next similar study, Shao et al., revealed antigen detection efficiency of SARS-CoV-2 (both S-antigen and N-antigen) in clinical nasopharyngeal samples through developed high purity semiconducting SWCNTs-FET sensor. Specifically, the developed SWCNTs-FET sensor was well decorated with anti-SARS-CoV-2 spike protein antibody (SAb) and anti- nucleocapsid antibody. Importantly, antibody functionalized SWCNTs-FET sensor showed greater sensitivity against both S-antigen and N-antigen with LOD of 0.55 fg/ ml and 0.016 fg/ ml, respectively. Besides, the sensing capability of SAb- functionalized SWCNTs-FET sensor against both positive and negative clinical samples was also found to be effective, indicating an effective novel diagnostic tool against COVID-19 [131]. In similar next research, Jeong et al., prepared ‘capture’ ssDNA functionalized SWCNTs (ssDNA-SWCNTs) to determine the SARS-CoV-2 viral RNA detection efficiency and sensitivity. The study revealed that viral extraction potentiality of developed ssDNA-SWCNTs from phosphate buffered saline was found to be 100% efficient as compared to the commercial silica-column kit, whose extraction efficiency was observed to be ~20%. Furthermore, viral nucleic acid extraction propensity of ssDNA-SWCNTs was almost 50% from human saliva, which was similar to commercial DNA/RNA extraction kits. Therefore, this CNTs-based viral nucleic acid extraction strategy could be a highly sensitive and high-yield identification technique for future perspective [132]. 3.3 Miscellaneous In the next exploration, Rajil et al., developed streptavidin-coated up conversion NPs conjugated with SARS-CoV-2 RBD as phantom virion (SUPPV NPs) to determine the strength of viral detection potentiality through neutralizing antibodies. In the study, neutralizing antibodies (IgG) containing sample showed no attachment of nanoparticles with ACE2 coated substrate while nanoparticles exhibited a strong affinity for binding with ACE2 coated substrates which lacks neutralizing antibodies. More importantly, developed SUPPV NPs possessed the highest sensitivity against SARS-CoV-2 with a detection limit of 4 ng/ml which was comparatively lower than the commercially available diagnostic kit with a detection limit of 19ng/ml [133]. 4 Nanoparticles for therapeutic drug delivery The emergence and outbreak of this novel SARS-CoV-2 viral infection create dramatic health and economic crisis throughout the world the current situation is due to which the whole world's medical and pharmaceutical researchers became engaged in the development of effective and potential anti-COVID 19 therapeutic candidates. Although selective countable drugs have been approved by regulatory bodies like FDA under EUA and WHO for their treatment but still face challenges due to their high dose regimen therapy and hence toxicities, non-specific site delivery, low therapeutic potentiality, low dose availability at a target site, poor biopharmaceutical attributes and mutational behavior of viral gene [134,135]. Thereafter, keeping into consideration all aforementioned problems associated with conventional therapy, the nano-medicinal based approach has been identified for effective and selective therapeutic drug delivery, particularly specifying the site-specific and viral targeted drug delivery accompanying reduced dosing frequency and hence toxicities [136,137]. In this section, we majorly investigated the various attempts and possible future perspectives of Nano-medicinal based drug delivery system that helps to improve the pharmacological profile of the ingested drug and could be represented as novel valid pharmacotherapeutic options to treat COVID 19 as depicted in table 2 .Table 2 Nanoparticles for the delivery of therapeutical moieties against COVID 19 Table 2:Studied by Delivery system /therapeutic agents Purpose of study Outcomes References Pooladanda et al., iRGD peptide fabricated Nimbolide liposome (iRGD-NIMLIP To determine the anti-inflammatory activity of iRGD-NIMLIP iRGD-NIMLIP possess superior anti-inflammatory property compared to free Nimbolide, NIMLIP and Dexamethasone enabling effective approach against SARS-CoV 2 associated ARDS as well. [138] Lima et al., chloroquine encapsulated Polylactic Acid NPs (CH-PLA NPs To explore the affinity of CH-PLA NPs to inhibit the entry of virus into host Developed NPs significantly blocks the SARS-CoV 2 entry into human host due to morphologically resemble. [139] Abouaitah et al., Ellagic acid impregnated ZnONPs functionalized with triptycene organic molecule (TRP-ELG-ZnONPs Antiviral activity of TRP-ELG-ZnONPs TRP-ELG-ZnONPs with virucidal efficiency of >60% was comparatively less cytotoxic with respect to free ELG and ZnONPs. [140] Molinaro et al., Dexamethasone incorporated NPs (Dex-NPs to estimate anti-inflammatory activity of Dex-NPs Dex-NPs cause inducement of immune response against pro-inflammatory cytokines in LPS- injected murine model enabling their survivability rate. [141] He et al., Surface modified AgNPs To determine virucidal activity modified AgNPs Branched polyethyleneimine (BPEI AgNPs showed minimal cytotoxic effect against vero-E6 cells with superior virucidal potentiality against SARS-CoV 2 as compared to citrate and PVP modified AgNPs and free AgNPs. [142] Almanza et al., AgNO3 solution Therapeutic exploration of AgNO3 solution against SARS-CoV 2 infection Participants allowed to do mouthwash and nose rinse with AgNO3 solution showed minimal (only 1.8% incidence of infection compared to control group (showed 28.2%). In-vitro study performed in culture cells further revealed greater inhibitory activity. [143] Jin et al., Hesperidin conjugated chitosan NPs (HPD-CHNPs Anti-inflammatory activity against acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) associated cytokine storm syndrome (CSS). Exhibited significant reduction of ALI and ARDS induced inflammatory cytokines along with improved cellular uptake in inflammatory microenvironment irrespective to free HPD. [144] Ozturk et al., Cefaclor loaded Eudragit 100 NPs (CFC-ENPs, Cefaclor loaded Eudragit 100+PLGA NPs (CFC-EPNPs) and Cefaclor loaded PLGA NPs (CFC-PNPs) Antimycobacterial property and therapeutic efficacy against SARS-CoV 2 accompanied coinfections and abdominal discomfort Antimycobacterial activity for CFC-ENPs and CFC-EPNPs was observed to be 16-folds and for CFC-PNPs was 2-folds superior against S. aureus and E. coli, compared to free CFC. [145] Mehranfar et al., Functionalized AuNPs (AuNPs-pep To determine the binding affinity of AuNPs-pep for RBD of SARS-CoV 2 AuNPs-pep exhibited greater interaction with RBD forming stable complex with it and hence inhibit viral infection. [146] Dormont et al., Squalene based multidrug NPs (SQ-Ad\Vit.E NPs Anti-inflammatory activity against covid 19 infection SQ-Ad\Vit.E NPs at concentration of 10µg/ml showed enhanced inhibitory potentiality of pro-inflammatory cytokines and reactive nitrogen species in LPS induced in-vitro model of RAW 264.7 macrophage. In-vivo endotoxemia mouse model showed significant reduction of TNF-α, MLP-1, IL-6 with improved survivability rate. [147] Hanafy et al., Dual silymarin/curcumin loaded chitosan coated BSA NPs (CH-Curc/Sily BSA NPs Determination of anti-inflammatory and anti-viral activity against covid 19 showed potential anti-viral activity against covid 19 at conc. of 25µg/ml. Additionally, developed NPs exhibited strong anti-inflammatory activity against oleic acid model at minimal concentration. [148] Ding et al., RBC incorporated methylprednisolone sodium succinate loaded chitosan NPs (RBC-MPSS-CH NPs To find out the potentiality and effectivity of developed NPs for lung targeting where SARS-CoV 2 primarily exist. RBC-MPSS-CH NPs showed improved pharmacokinetic profile with greater drug accumulation throughout lungs. RBC-MPSS-CH NPs further displayed superior anti-inflammatory activity against LPS induced mouse model. [149] Khater et al., Fluoxetine HCL loaded lipid polymer hybrid NPs (FH-LPH NPs To investigate therapeutic efficacy against SARS-CoV 2 Showed highest possible interaction with viral protease. Displayed enhanced cellular internalization with good biocompatibility. [150] Idris et al., siRNA incorporated stealth lipid NPs (siRNA-SLNPs Examined the pharmacotherapeutic activity against SARS-CoV 2. In-vivo study demonstrated the enhanced viral suppressive efficiency with improved survival rate in mice model. [151] Martins et al., Meso 2,3-dimercaptosuccinic acid fabricated iron oxide NPs (DMSA-Fe3O4 NPs To investigate the possibility for drug delivery and treatment of SARS-CoV 2. Successfully developed and physiochemically characterized NPs showed good biocompatibility against normal cell lines. [152] Hamoudaet al., AgNPs fabricated wearing mask Antiviral activity and breathability performance AgNPs exhibited less toxicity with immense viral inhibitory potentiality. AgNPs fabricated cotton mask showed better air permeability and breathability compared to surgical mask. [153] Archanaet al., Flower extract incorporated copper iodide NPs (FE-CuI NPs) Antimicrobial activity against SARS-CoV 2 In-vitro study suggested less toxic against normal cells. Showed highest possible interaction with vial protease after molecular docking. FE-CuI NPs engineered cotton mask provides better viral entry inhibition capacity. [154] Denget al., mRNA incorporated lipid NPs (mRNA- HB27-LNPs Antiviral efficacy against corona virus In- vivo study speculated the improved circulating half-life with greater protection against virus after i.v administration of mRNA-HB27-LNPs into mice model. [155] Sanna et al., Remdesivir loaded targeted NPs (RDV-TNP-1 Antiviral activity against COVID 19 RDV-TNP-1 showed highest binding propensity to ACE2 receptor with enhanced antiviral efficacy and better biocompatibility. [156] Zheng et al., RBC hitchhiked ivermectin loaded PLGA NPs (RBC-IVM-PNPs and chitosan coated PLGA NPs (RBC-IVM-CSPNPs) Anti-inflammatory activity of ivermectin associated with SARS-CoV 2 virus. RBC-IVM-CSPNPs exhibited prolonged circulating time profile with superior anti-inflammatory activity as compared to RBC-IVM-PNPs. [157] In this context, Pooladanda et al., developed an iRGD peptide fabricated Nimbolide liposome (iRGD - NIMLIP) to determine its anti-inflammatory activity. Interestingly, the study showed that iRGD - NIMLIP possessed superior potentiality to inhibit STAT3-DNMT-induced oxidative stress and cytokine storm as compared to free NIM, NIMLIP and dexamethasone. Additionally, iRGD - NIMLIP further demonstrated the downregulation of bacterial endotoxin lipopolysaccharide (LPS) induced lipid peroxidation, pro-inflammatory cytokines, TNF-α mediated P65NF-κB signalling with additional antioxidant property in in-vitro and in-vivo mouse model. Therefore, an author concluded that iRGD peptide fabricated Nimbolide liposome could be an effective approach for the treatment of SARS-CoV-2 induced acute respiratory distress syndrome (ARDS), a prime cause of mortality [138]. In one study, Lima et al., accessed the improved efficiency of poly lactic acid (PLA) encapsulated chloroquine to inhibit endocytosis of NPs. The study suggested that targeted delivery of nanoparticle fabricated chloroquine that morphologically resembles with SARS-CoV-2 virus may greatly block the host's cellular entry and hence could be one strong Nano-therapeutic approach for COVID 19 treatment as well [139]. In another investigation, Abouaitah et al., demonstrated the antiviral activity of ellagic acid (ELG) impregnated zinc oxide NP (ZnONPs) functionalized with triptycene organic molecule (TRP), a hybrid Nano formulation. In the study, the cytotoxic activity of free ELG against host cells was observed to be significantly higher compared to both ZnONPs and hybrid Nano formulations. Moreover, the antiviral potentiality of the developed Nano formulation was found to be superior irrespective of ZnONPs and ELG alone, with a therapeutic index of 77.3 for H1N1, 75.7 for HCoV-229E (RNA virus), 57.5 for HSV-2 and 51.7 for Ad-7 (DNA virus). Hence, a hybrid Nano formulation exhibiting direct viricidal efficiency of > 60 % could be a new alternative therapeutic strategy to treat COVID 19 [140]. Similar to another study, Molinaro et al., assessed the anti-inflammatory property of dexamethasone using leukocyte-derived nano-vesicles called leukosomes. The study revealed that dexamethasone fabricated biomimetic nanoparticles could significantly reduce pro-inflammatory cytokines in lipopolysaccharide (LPS) injected murine model by enhancing their immune response and hence survivability rate as compared to free drug. Eventually, an author speculated that cytokine storm syndrome associated with COVID 19 can be mitigated and treated well through nanoparticle-incorporated corticosteroids including other anti-COVID 19 pharmacotherapeutic moieties [141]. In addition to this study, He et al., emphasized the virucidal potentiality of three different surfaces impregnated AgNPs (citrate modified, polyvinyl pyrrolidone modified-PVP and branched polyethyleneimine modified-BPEI) against SARS-CoV-2. In the observation, all three developed surfaces modified AgNPs were found to be less toxic against Vero E6 cells exhibiting significant cell viability concerning AgNO3. Moreover, BPEI-modified AgNPs with 50 nm particle size showed greater antiviral activity than citrate-modified and PVP-modified AgNPs providing sophisticated insight for the development of such battle-winner tools against COVID 19 [142]. On other hand, Almanza et al., presented the potential therapeutic efficiency of AgNO3 solution against SARS-CoV-2 in healthcare personnel. In the investigation, in vitro study performed in cultured cells showed higher inhibitory activity. Moreover, a clinical study conducted on 231 participants over 9 weeks further demonstrated that the experimental group who were instructed to do mouthwash and nose rinse through AgNO3 solution showed a significantly lower incidence of infection (only 2 out of 114, i.e., 1.8%) compared to control group (got an infection in 33 out of 117, i.e.,28.2%) who were instructed to use in a conventional way indicating that nanoparticle-based treatment could be worth noting [143]. Additionally, to get insight into the treatment of COVID 19, Jin et al., investigated the anti-inflammatory activity of hesperidin conjugated chitosan NPs (HPD-CHNPs) against acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) associated cytokine storm syndrome (CSS) through nasal delivery in a mouse model. However, in vivo study suggested that compared to free HPD, HPD-CHNPs exhibited a significant reduction of inflammatory cytokines and vasculature permeability that ultimately results in marked inhibition of ALI and ARDS, in addition to higher cellular uptake of drug in an inflammatory microenvironment. Overall, nanoparticle-based drug delivery could be relevant to attenuating CSS and ARDS specified for COVID 19 patients [144]. In another context of an investigation, Ozturk et al., performed the study of Cefaclor monohydrate (CFC) for its antimycobacterial activity and therapeutic efficacy against COVID 19 accompanied coinfections and intestinal symptoms through the delivery of either CFC-loaded Eudragit S100 NPs (CFC-ENPs) or CFC loaded Eudragit S100+PLGA NPs (CFC-EPNPs) or through CFC loaded PLGA NPs (CFC-PNPs). As per the study, the antibacterial activity for all three different nanoparticle-based formulations exhibiting the size range of 171.4-198.8 nm and encapsulation efficiency of 58.4%- 81.2% were found to be superior by 16-folds for both CFC-ENPs and CFC-EPNPs and by 2-folds for CFC-PNPs as compared to free CFS against S. aureus and E. coli. Additionally, Eudragit fabricated both nano formulations were found to be pH -sensitive, a colon targetable while CFC-PNPs showed prolonged release profile after invitro dissolution study [145]. Likewise, Mehranfar et al., synthesized the functionalized AuNPs to determine their ability to bind RBD of SARS-CoV-2 via the molecular (MD) dynamic simulation method. Interestingly, AuNPs-pep exhibited minimum distance with RBD indicating greater interaction between them compared to other functionalized nanoparticles. Additionally, AuNPs-pep further causes a significant reduction of average solvent accessible surface area value (SASA) of RBD with 8970 A2, which results in coverage of the whole binding surface of RBD with forming most stable complex as similar to ACE2 whose SASA value of RBD was 8941 A2, providing new insight for subsequent inhibition of COVID 19 [146]. In the search, Dormont et al., devised squalene-based multidrug NPs to determine their potentiality against COVID 19 induced hyper inflammation. Indeed, successful delivery of adenosine (endogenous immunomodulator as anti-inflammatory) and α-tocopherol (Vit. E as an antioxidant) by conjugating with squalene (endogenous lipid) NPs (SQ-Ad\Vit.E NPs) was presented. Importantly, SQ-Ad\Vit.E NPs exhibited an efficient inhibitory profile for pro-inflammatory cytokines and reactive nitrogen species (NOx) at a concentration of 10 µg/ml in LPS induced in vitro model of RAW 264.7 macrophage while in vivo endotoxemia mouse model showed a significant reduction of TNF-α, MLP-1, IL-6 with increased IL-10 along with the improved pharmacological response (bioavailability) compared to free drug. Accordingly, SQ-Ad\Vit.E NPs further demonstrated for its side effect study where they showed no significant reduction of BP with a protective effect of liver injury and hence improved survival rate in lethal LPS treated mouse model [147]. Recently, Hanafy et al., synthesized chitosan-coated bovine serum albumin NPs encapsulated with silymarin/curcumin (CH-Curc/Sily-BSA NPs) to determine its anti-inflammatory and anti-COVID19 activity. However, thus developed CH-Curc/Sily-BSA NPs were found to be optimally efficient for muco-inhalable delivery with higher dispersibility and improved lung deposition capability. Importantly, the developed CH-Curc/Sily-BSA NPs displayed a significant reduction of IL-6 at 64 ± 0.8 Pg/µL and superior inhibition of CRP at 6 ± 0.5 µg/µL in the oleic acid model. Further, dual curcumin and silymarin-loaded NPs at a concentration of 25 µg/ml showed potential antiviral activity against COVID 19 (44.4% viral inhibition) with improved histopathological evidence, contributing to its strong pharmacotherapeutic potentiality [148]. Besides that, Ding et al., formulated RBC incorporated methylprednisolone sodium succinate (MPSS) loaded chitosan NPs (RBC-MPSS-CH NPs) for effective delivery targeting the lung where COVID 19 primarily exists with a motive for reduced dosing frequency and hence toxicities. An ongoing study further identified the significant reduction of drug plasma concentration with improved circulation time through RBC Hitchhiking after in vivo pharmacokinetic study. Additionally, irrespective of MPSS-CH NPs and free MPSS, RBC- MPSS-CH NPs showed higher mean residence time (MRT) and area under curve (AUC) profile. On the other side, hepatocellular uptake of delivered drug through RBC- MPSS-CH NPs was superior with greater drug accumulation throughout the lungs as identified via in vivo fluorescent study in a mouse model. Equally important, RBC- MPSS-CH NPs exhibited a significant reduction of TNF-α and IL-6 in LPS induced lung injury rat model as compared to MPSS-CH NPs and free MPSS [149]. Another investigation conducted by Khater et al., introduced Fluoxetine HCL (FH) loaded lipid polymer hybrid NPs (FH-LPH NPs) to identify the potentiality and its efficacy against SARS-CoV-2 infection. The experiment initiated by determining the possible interaction between selective serotonin reuptake inhibitors (SSRIs) and SARS-CoV-2 main protease through molecular docking (MD) confirmed the highest possible interaction of FH through hydrogen bonding formation with a minimum required energy among other SSRIs. Most importantly, a cytotoxicity study performed for developed NPs in CCD-19 L4 cells (human lung fibroblast) exerted biocompatible to human cells. In addition, a cellular uptake study performed through Flow cytometry using Dil Labeled FH-LPH NPs of 50nm size in CCD-19 L4 cells showed satisfactory cellular internalization. Contrarily, in vitro release study of FH-LPH NPs carried out in phosphate buffer saline (PBS) and 10% v/v fetal bovine serum (FBS) showed 86 % and 92 % release profiles respectively after 24 hrs, indicating the controlled release behavior [150]. On the contrary, Idris et al., aimed to synthesize different types of siRNA-incorporated stealth lipid nanoparticles (SLNPs) to investigate the pharmacotherapeutic efficacy against SARS-CoV-2. As the study demonstrated that the suppressive potentiality of siRNAs-SLNPs against COVID 19 was satisfactory after in vivo study in an infected mice model. In addition to this, siUC7-SLNPs and siHeL2-SLNPs after intravenous administration in mice model greatly enhance the survival rate by causing lower weight loss as compared to viral infected and control-treated mice. Notably, an author identified that the siRNAs-SLNPs delivery system has the immense potential to block viral expression and replication in humans as well [151]. On other hand, Martins et al., further displayed the potential use of meso 2,3-dimercaptosuccinic acid (DMSA) fabricated iron oxide NPs (DMSA-Fe3O4 NPs) against COVID 19 disease. Thus developed DMSA-Fe3O4 NPs with a size range of 12 nm were physiochemically and magnetically characterized using different techniques owing to optimally synthesized for delivery. Moreover, a biocompatibility study of DMSA-Fe3O4 NPs further demonstrated the lack of toxicity to a normal cell, providing new insight for use of Nano-medicinal based delivery to mitigate the deadliest COVID 19 [152]. Besides the delivery of antiviral candidates site-specific through a nanoparticles-based approach, Hamouda et al., addressed the use of AgNPs in wearing a mask for its antiviral activity against COVID 19 and also measures the breathability performance of AgNPs treated masks. The cytotoxicity study and viral inhibitory study of AgNPs were performed in VERO-E6 cells which displayed cytotoxic concentration (CC50) 142.5 µl/ml indicating less toxic to host and inhibitory concentration (IC50) 25µl/ml for 21.49% viral inhibition and displayed potent antiviral activity. On other hand, air permeability and breathability of 100% cotton mask performed on volunteers showed no significant alteration in blood oxygen saturation (SpO2 %) and heart rate even at working and waking time compared to surgical mask. Noteworthy, the author concluded that AgNPs fabricated cloth masks could be a promising strategy to control COVID 19 [153]. Similarly, Archana et al., also examined the antimicrobial activity of flower extract (Hibiscus rosa Sinensis) incorporated copper iodide NPs (FE-CuI NPs) against COVID 19. The developed prismatic-shaped FE-CuI NPs with a size range of 552.45 nm were further fabricated into cotton fabric protective face masks to knock down the entry of COVID 19 into a human host. Additionally, in vitro cytotoxicity study of developed NPs was analyzed against cancer and spleen cell lines and was observed to be dose-dependent manner with IC50 value of 233.93µg/ml indicating biocompatible. Noteworthy, a molecular docking study confirmed the strong interaction between FE-CuI NPs and COVID 19 main protease that reinforces the approach for efficient and effective mitigation of existing COVID 19 [154]. In addition, Deng et al., explored the therapeutic efficacy of mRNA Antibody encapsulated lipid NPs (mRNA-LNPs) against SARS-CoV-2 infections. Importantly, in vivo study suggested that intravenous administration of mRNA-HB27- LNPs in mice model showed improved circulating half-life with longer protection for 1,7 and even 63 days against SARS-CoV-2, irrespective to free HB27 antibody. Additionally, study further revealed that the antiviral efficacy of mRNA-HB27- LNPs in the hamster model was observed to be dose-dependent manner after prophylactic administration [155]. In the next investigation, Sanna et al., demonstrated the antiviral activity of remdesivir-loaded targeted NPs (RDV-TNP-1) against SARS-CoV-2 infected Vero E6 cells. Afterward, the antiviral potentiality of RDV-TNP-1was observed to be significantly improved along with the highest ACE2 binding capability as compared to free RDV or non-targeted ones. Most interestingly, empty TNP-1E further exhibited antiviral efficacy which might be due to its direct competitive mechanism with viral particles for ACE2 receptors. Moreover, RDV-TNP1 further exposed its better biocompatible properties against normal cell lines irrespective to free RDV. Accordingly, an author concluded that TNP-based inhalable antiviral drug delivery for pulmonary targeting could be extremely favorable [156]. In addition to this, Zheng et al., highlighted the antiviral activity of Ivermectin (IVM) against SARS-CoV-2 via enhancing its pharmacokinetic attributes through the delivery of RBC- Hitchhiked IVM loaded PLGA NPs (RBC-IVM-PNPs) and chitosan-coated PLGA NPs (RBC-IVM-CSPNPs). Very importantly, both developed RBC-Hitchhiked NPs showed efficient pulmonary delivery with improved drug accumulation to the lung tissues which ultimately leads to potential suppression of inflammatory response and progression of acute lung injury. Although, cationic RBC-IVM-CSPNPs displayed superior anti-inflammatory activity due to their longer circulation times and minimal elimination rate irrespective of RBC-IVM-PNPs [157]. 5 Nanoparticles based vaccines delivery against COVID 19 The current situation facing mutational behaviours of corona virus frequently leads the global health and economy towards massive disruptions. Such multiple mutation and novel viral strain of COVID 19 primarily persists as a great wall barrier for the delivery of traditional vaccines like inactivated vaccines live attenuated vaccines, recombinant vaccines, DNA and viral vector-based vaccines that ultimately results in unexpected poor efficacy and effectiveness. Going towards, the development and deployment of future-proofing nanoparticles-based vaccine delivery against COVID 19 straightforward the emerging promise and become a selective track. Vaccine antigens either encapsulating at the core of nanoparticles (antigen encapsulated NPs) or trapped at the surface of nanoparticle (antigen-presenting NPs) offers feasible and tunable particulate structures that mimic the structural features of natural viruses [158], [159], [160]. In the present study, Walsh et al., developed lipidic nanoparticles for the delivery of two candidates (BNT162b1 and BNT162b2) based on nucleoside-modified RNA vaccines. They took trial on 195 candidates, in which 15 candidates are assembled in each of 13 batches. However, in adults, especially, BNT162b1 (secretes trimerized COVID-19 receptor binding sector) is accessorized with mild systemic reactions as compared to BNT162b2 (membrane-affixed COVID-19). Importantly, BNT162b2-based lipidic delivery showed superior safety and immunogenicity in younger and elder as compared to the BNT162b1 lipidic-based delivery system which further enforces to conduct of phase 2-3 clinical trials to investigate its safety and efficacy as well [161]. In another study, small interfering RNA (siRNA) incorporated three different lipidic nanoparticles (LNPs) were prepared by Idris et al., to target SARS-CoV-2. The study demonstrated that 90% of therapeutic effect was screened among all 3 developed LNPs when administered either singly or in combination form. In the meantime, siRNA-loaded two novel LNPs further possessed higher efficacy against SARS-CoV-2 infection with improved survivability rate after i.v administration into mice model. Noticeably, prepared LNPs can be scaled up to demonstrate in humans to treat the first arrival of viral symptoms [151]. Simultaneously, Elia et al., engineered mRNA vaccines based on LNPs conjugated SARS-CoV-2 human Fc-incorporated receptor binding domain (LNPs-RBD-hFc mRNA). Herein, in-vivo study demonstrated in BALB/c mice model showed improved Th-1 biased cellular response and robust humoral response, in addition to increased neutralizing antibodies [162]. Similarly, the next study performed by Rao et al., has further demonstrated the significance of genetically engineered derived decoy nanoparticles (nano-decoys) against SARS-CoV-2 infection. Most importantly, nano-decoys engineered by fusing cellular membrane nano-vesicles of ACE-2 receptors and human monocytes showed the potential to adsorb viral and inflammatory cytokines (interleukin-6 and granulocyte-macrophage colony-stimulating factor) in a mouse model, providing efficient protection against SARS-CoV-2 associated immune disorder and lung injury by competing with host cells [163]. In another investigation, Smith and co-workers et al., prepared a synthetic DNA vaccine to target the S protein of Covid-19. In in-vitro experiment, they found the robust nature of INO-4800. Furthermore, immunization with INO-4800 in mice and guinea model were tested for T-cell response which is antigen specified. Also, neutralized by functional antibodies and occlude the ACE-2 receptor from protein binding along with bio-distribution to the pulmonary region. Finally, we can conclude as the INO4800 vaccine stands as a possible applicant for translational analysis [164]. In addition, Gu et al., speculated the efficacy and safety profile of REVX-128 loaded trimeric spike protein impregnated NPs based vaccine delivery against SARS-CoV-2. Interestingly, the study exposed that even a single shot of NPs based vaccine immunization in mice model provides significantly higher serum antibody naturalizing potentiality as compared to NPs deficient vaccine immunization. Moreover, NPs based immunization in the Syrian golden hamster model further revealed superior antiviral activity and a greater safety profile irrespective of unprotected animals. Subsequently, NPs based REVX-128 vaccine also provides better thermostability up to 37°C for at least 4 weeks, enabling its propensity towards considerable progress on mitigation of COVID 19 [165]. In another experiment, Yang et al., formulated SW0123 incorporated SARS-CoV-2 spike protein packaged core-shell structured lipo-polyplex NPs (SW0123-LPP NPs) to fight against pandemic COVID 19. They reported that SW0123-LPP NPs-based vaccine immunization by intramuscular route to mice and non-human primates showed better uptake and prudent bio-distribution pattern with reduced liver targeting effect. In addition, inducement capability for Th-1 polarized T cells response and antibody neutralizing capacity of SW0123-LPP NPs based vaccine delivery associated with SARS-CoV-2 as well as D614G and N501Y variants was found to be significantly higher, depicting its potent immunogenicity [166]. In another similar study, Shinde et al., formulated an NVX-CoV2373 nanoparticle vaccine to determine its efficacy against the B.1.351 variant of COVID 19. The ongoing study demonstrated that NVX-CoV2373 nanoparticle vaccine exhibited superior efficacy of 60.1 % (95% CI,19.9 to 80.1) among HIV negative with seronegative at the base line while vaccine efficacy in seropositive at baseline was 52.2 % (95% CI, -24,8 to 81.7%). Notably, prototype sequenced NVX-CoV2373 nanoparticle-based vaccine immunization showed strong antibody neutralizing efficacy and improved antigen-specific poly-functional CD4 + T-cells response which may prevent cross-transmission of serological variants of SARS-CoV-2 [167]. Very hopefully, Zhang YN et al., investigated the mechanism of vaccine-induced immunity and antibody neutralizing potentiality of self-assembling protein NPs (SApNPs) against various variants of SARS-CoV-2. Afterward, the study revealed that S2GΔHR2 spikes (of ancestral Wuhan-Hu-1 strain) incorporated SApNPs exhibited a superior tendency to neutralize all B.1.1.7, B.1.351, P.1, and B.1.617 variants with comparable potency. Most importantly, 13-01v9-SApNPs possessed longer retentibility by 6-folds with 4-fold higher follicular dendritic cellular presentation and 5-folds greater germinal center reactibility in lymph node follicles of the mouse model as compared to soluble spikes [168]. Similarly, Thomas et al., designed BNT162b2-lipid NPs based mRNA vaccine to examine its efficacy and safety against COVID 19. In the study, BNT162b2-lipid NPs exhibited acceptable adverse event profiles with vaccine efficacy estimated as 96.7% (95% CI, 80.3 to 99.9) in severe disease cases while 91.3% (95% CI, 89.0 to 93.2) among participants without evidence of previous SARS-CoV-2 infections after 6 months follow up. Interestingly, almost 100% vaccine efficacy was observed against B.1.351 variants of SARS-CoV-2 in the south African region [169]. In one study, Kremsner et al., developed mRNA-based lipid nanoparticles-based viral S protein incorporated vaccine (CVnCoV-SLNPs) to determine its safety and efficacy profile in human volunteers. In the study, an experiment was performed with limited participants with different controlled groups at variable concentrations ranging from 2-12μg for 28 days, where safety, efficacy and immunogenicity were observed to be concentration dependent and hence regarded to be safe for human use [170]. In this regard, the same experimental group further investigated the antibody neutralizing efficacy of CVnCoV-SLNPs against SARS-CoV-2 in a large participant group. Herein, after immunization of the 2nd dose of 12μg CVnCoV-SLNPs, antibody neutralizing capacity was found to be 83% against seroconverted SARS-CoV-2 neutralizing titer (MN50) and 100% against SARS-CoV-2 S protein or RBD. After all, an author and group concluded that a concentration of CVnCoV-SLNPs with 12μg exhibiting strong efficacy and immunogenicity with acceptable side effects, could be an effective approach for COVID 19 prevention [171]. In addition, McKay et al., also formulated a self-amplifying RNA encoded viral S protein incorporated LNPs (saRNA-SLNPs) to find out its antibody neutralizing capacity and humoral response between injected murine and COVID 19 recovered patients. Here, saRNA-SLNPs injected mice model showed dose-dependent viral neutralization efficiency of IC50 of 5 × 103 to 105 and higher antibody titer compared to a naturally infectious human with IC50 of 103. Furthermore, saRNA-SLNPs injection into mice model also exhibited dose-dependent immunogenicity which was particularly higher than in COVID 19 recovered patients, enabling rapid translation for clinical use as well [172]. Sequentially, another research conducted by the same research group, prepared the same formulation of self-amplifying-RNA encoded lipidic nano-formulated vaccine (saRNA-SLNPs). In this investigation, saRNA-SLNPs exhibited robust neutralization against both pseudo virus and wild-type virus along with remarkable superior and dose-dependent SARS-CoV-2 specific antibody titers in mouse sera. Equally important, developed saRNA-SLNPs further showed enhanced cellular response with improved immunogenicity in the mice model [173]. 6 Nano-technological strategy in personal protection equipment (PPE) However, nanotechnology plays a crucial role in multi sectors such as in a diagnostic field, drug and vaccine delivery, against COVID-19 [174]. Besides, the applicability of nanotechnology in personal protection equipment (PPE) in the current scenario becomes equally fruitful for the prevention of SARS-CoV-2. Although, the world is currently consuming enormous PPE such as masks, face shields, respirators, etc. just to be protected from entry of the SARS-CoV-2 virus, but 100% effectiveness was not achieved through the use of such conventional PPE. Globally, consumers are limited to the use of conventional PPE right now, due to which they are still facing challenges with COVID-19. Therefore, the use of Nano-technological strategy in various PPE could easily surpass the theatrical situation of COVID-19 [175,176]. Recently, Zhong et al., synthesized a graphene-coated temperature-sensitive surgical mask by dual-mode laser fabrication method. Importantly, this reusable and recycled graphene-coated mask showed an efficient viral sterilization activity under solar illumination, due to high surface temperature (quickly increase to over 80 °C) and provides better protection from incoming viral droplets due to its super-hydrophobic surface. Additionally, pertaining to its outstanding salt-rejection performance, a photothermal graphene-coated mask could be directly used in solar-driven desalination for long-term use as well [177]. In addition to this, Huang et al., demonstrated that the bacterial inhibition rate of developed laser-induced graphene modified mask showed 80%, with additional 99.99% bacterial killing efficiency within 10 min, was observed after combination with graphene layer's photothermal effect. Thereby, the study further suggested the importance and applicability of nano-technological based PPE for effective prevention of COVID-19 [178]. In next study, Chen et al., further discussed the antiviral activity of graphene oxide (GO) fabricated with AgNPs (GO-AgNPs) against both enveloped and non-enveloped viruses, indicating an effective approach for the development of viral-resistant PPE [179]. Moreover, Nakamura et al also demonstrated the antibacterial and antiviral activity of devised AgNPs chitin nano fiber sheet (CNFS) against E. coli and H1N1 virus respectively. The study showed that AgNPs-CNFS possessed potential antiviral activity against the influenza virus, indicating that such NPs could be equally efficacious against SARS-CoV-2 when applied to protective cloths, masks and gloves [180]. Similarly, the study conducted by Borkow et al., suggested that impregnation of CuONPs into N95 respiratory mask exhibited improved biocidal and antiviral activity against influenza viruses without alteration in filtration efficiency of a mask [181]. On the other hand, Ungur et al., devised polyurethane fabricated CuO nano fiber to determine the air filtration efficiency. Such developed nano fibers showed improved air filtration efficiency and could be a better platform for its use in various surgical masks to be protected from virally contaminated air droplets [182]. 7 Nanotechnology in surface decontamination and sanitization Many studies suggested that SARS-CoV-2 viral contaminated micro-droplets released during sneezing and coughing may persist for 3 hrs to up to 9 days at 30 °C or even more, in an aerosolized form. These droplets can be transmitted from one person to another, if someone touches a contaminated surface and thereafter gets infected. To be free from such contamination, numerous conventional disinfectants and sanitizers such as alcohols, soap, sodium hypochlorite, hydrogen peroxide, etc. are available but still facing challenges to control COVID-19 [183,184]. Therefore, nano-technological based approach such as metallic NPs which are well known for their antibacterial, antifungal and antiviral activities, provides enormous strength and opens a new avenue for developing highly efficient and very effective disinfectants [185]. Additionally, activation of nano-technological devices upon electrothermal, photocatalytic and photothermal stimuli causes a release of active substance more efficiently, accompanying clear and upto 100 % disinfection potentiality [186]. In this regard, the antiviral activity of AgNPs has been studied by various researchers and group against the different virus. Elechiguerra et al., demonstrated the antiviral activity against HIV-1 virus [187]; Rogers et al. studied against monkeypox virus [188]; Orlowski et al., against herpes simplex virus [189]; Xiang et al., against H1N1 influenza virus [190], and so on, indicating potential option for disinfection of SARS-CoV-2 as well. Similarly, Gusseme et al., [191] and park et al.,[192] also demonstrate the antiviral activity of AgNPs by disinfecting viral contaminated water, signifying a better option for disinfection of SARS-CoV-2 from the contaminated surface as well. In addition to this, the antimicrobial activity of CuNPs has also been well established, which could be suitable technique to combat various viruses. In one study, Murray et al. demonstrate potential antiviral efficacy of Cu against poliovirus [193]. Moreover, Warnes et al., further examined viral inhibitory efficacy of CuNPs against HuCoV-229E corona virus. The study showed that CuNPs could efficiently inhibit viral growth, which further inhibits SARS-CoV-2, when used as disinfectant [194]. 8 Toxicological aspects of nanoparticles However, nanoparticles emerge as novel tools for bio-imaging, diagnosis and delivery of immense pharmacotherapeutic moieties against many deadliest diseases. In the meantime, potential investigation for its toxicological perspective seems to be equitability in respect to its use during delivery of multi cargos. Notably, various studies have been approached to confine the threatful adverse effects of nanoparticles on cardiovascular, neurological, pulmonary, cytological and other various epidemiological toxicities [195], [196], [197]. Numerous studies demonstrated that the toxic effect of such inorganic nanoparticles occurs through physicochemical (dispersity, size, shape and surface chemistry) and biochemical (cellular and molecular) mechanisms as shown in Figure 6 . More specifically, the cellular mechanism involves lysosomal impairment, mitochondrial dysfunctions, mitophagy, endoplasmic reticulum (ER) stress and endoplasmic reticulum autophagy, while molecular mechanism particularly associated with autophagy-related to signalling pathway, hypoxia-inducible factors and oxidative stress [198], [199], [200].Figure 6 Schematic representation of the toxicological effect of nanoparticles via the physicochemical and biochemical mechanism. Figure 6: Additionally, many researchers have investigated the potential pathway for toxicological aspects of inhaled or administered nanoparticles. Fortunately, most of the synthetic organic, supra-molecular and polymeric-based NPs showed bearable or no toxicity due to their biodegradable and biocompatible natures [201,202]. On other hand, the undesirable and unbearable toxicity of inorganic NPs is mainly due to apoptosis, necrosis, oxidative stress and autophagy [203,204]. Further, autophagy is the prime root for causing significant toxicities which mainly follows microautophagy, macroautophagy and chaperone-mediated autophagy. Microautophagy involves the direct engulfment of cytoplasmic components into lysosomes in non-selective manners whereas macroautophagy is associated with the formation of autophagosomes initially, followed by fusion of autophagosomes into lysosomes and hence its degradation, leading to cell death shown in Figure 7 . Similarly, chaperone-mediated autophagy is a degradative phenomenon of translocated cytosolic soluble proteins through chaperone-dependent selecting manners without the formation of additional vesicles [205], [206], [207]. In one investigation, Yu et al., elucidated that autophagic cell death was observed due to silica NPs in hepatoma HepG2 cell lines, which were then markedly inhibited by autophagy inhibitor [208]. Similarly, Liu et al., and park et al., further examined the autophagic cell death in A549 lung cell lines and BEAS-2B bronchial cell lines respectively, due to the toxicological effect of single-walled carbon nanotubes [209,210]. Moreover, Sun et al., highlighted the copper oxide NPs inducible autophagy in A549 cell lines, which was inhibited by autophagic inhibitors like wortmannin and 3-methyladenine [211]. In addition, Yu et al., and Johnson et al., further investigated the efficient toxicity of zinc oxide NPs leading to autophagic cell death [212,213]. Besides, cationic poly-amidoamine dendrimers also may cause intolerable adverse effects like liver injury, lung damage and neuronal dysfunctions associated with autophagic cell death [214], [215], [216]. In the next study, wang et al., confirmed the lysosomal dysfunctional associated autophagic cell death in hepatocytes due to silica NPs [217]. Importantly, Fe3O4 NPs exhibited potential lysosomal functional impairment, mitochondrial damage, ER and Golgi body stress, where after such destructive effect was mitigated by PLGA-coated Fe3O4 NPs [218].Figure 7 Nanoparticles induced macroautophagy pathway following three different steps: (1) formation of double membraned autophagosomes from cytoplasmic constituents (misfolded proteins or damaged organelles), (2) formation of autolysosomes by the fusion between autophagosomes and lysosomes, (3) degradation of autophagosomes leading to cell death. Figure 7: 9 Conclusions After the outbreak of SARS-CoV-2 throughout the world, millions of people were infected with the virus and lost their life. Notably, the Shortage of advanced technological diagnostic kits and lack of effective treatment alternatives further contribute to disastrous health and economic issues. Although, worldwide’ s health organizations, pharmaceutical industries, and many groups of scientists have collaborated to find out efficient approach for the prevention of COVID 19, but still facing multiple challenges to shout the pandemic. In contrast, nano-technological strategies for diagnosis via bio-sensing, treatment via medicinal delivery and prevention via vaccine delivery create an emerging hope for the entire world. The rapid, highly sensitive and more accurate potentiality of this diagnostic nano-technological equipment strongly provides arrays for early detection of COVID 19. Additionally, higher drug internalization, greater target ability, reduced dosing frequency and hence toxicities and cost-effective method of Nano-medicinal and vaccine delivery further helps in the treatment and prevention of corona virus. Subsequently, upcoming deadliest viral infections can also be easily diagnosed, treated and prevented through a nano-technological approach. However, limited potent toxicities associated with administered NPs, equally need to be focused on and comprehensively investigated but the beneficial outcomes of this novel nano-technological backbones greatly overcome the hurdles and barriers associated with drug delivery. Therefore, in the current situation, this Nano-medicinal based pathway may significantly draw the attention of entire global scientists. Author Agreement Statement We the undersigned declare that this manuscript is original, has not been published before and is not currently being considered for publication elsewhere. We confirm that the manuscript has been read and approved by all named authors and that there are no other persons who satisfied the criteria for authorship but are not listed. We further confirm that the order of authors listed in the manuscript has been approved by all of us. We understand that the Corresponding Author is the sole contact for the Editorial process. He/she is responsible for communicating with the other authors about progress, submissions of revisions and final approval of proofs Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper Acknowledgments The authors are thankful to ISF College of Pharmacy, Moga, Punjab for their continuous support and encouragement. ==== Refs References 1 Sun J He W-T Wang L Lai A Ji X Zhai X COVID-19: epidemiology, evolution, and cross-disciplinary perspectives Trends in molecular medicine 26 5 2020 483 495 32359479 2 Park M Cook AR Lim JT Sun Y Dickens BL. A systematic review of COVID-19 epidemiology based on current evidence Journal of clinical medicine 9 4 2020 967 32244365 3 Organization WH. Naming the coronavirus disease (COVID-19) and the virus that causes it Brazilian Journal Of Implantology And Health Sciences 2 3 2020 4 Shereen MA Khan S Kazmi A Bashir N Siddique R. COVID-19 infection: Emergence, transmission, and characteristics of human coronaviruses Journal of advanced research 24 2020 91 98 32257431 5 Yuen K-S Ye Z-W Fung S-Y Chan C-P Jin D-Y. SARS-CoV-2 and COVID-19: The most important research questions Cell & bioscience 10 1 2020 1 5 31911829 6 Chakraborty I Maity P. COVID-19 outbreak: Migration, effects on society, global environment and prevention Science of the Total Environment 728 2020 138882 7 Guo Y-R Cao Q-D Hong Z-S Tan Y-Y Chen S-D Jin H-J The origin, transmission and clinical therapies on coronavirus disease 2019 (COVID-19) outbreak–an update on the status Military medical research 7 1 2020 1 10 31928528 8 https://covid19.who.int/. 9 https://www.who.int/emergencies/diseases/novel-coronavirus-2019. 10 https://www.worldometers.info/coronavirus/. 11 Mallapaty S. Where did Omicron come from? Three key theories 2022 Nature Publishing Group 12 Wang Q, Iketani S, Li Z, Guo Y, Yeh AY, Liu M, et al. Antigenic characterization of the SARS-CoV-2 Omicron subvariant BA. 2.75. bioRxiv. 2022. 13 Tan C-W Lim B-L BE Young Yeoh AY-Y Yung C-F Yap W-C Comparative neutralisation profile of SARS-CoV-2 omicron subvariants BA. 2.75 and BA. 5 The Lancet Microbe 2022 14 Arora NK Mishra J. COVID-19 and importance of environmental sustainability 2020 Springer 117 119 15 Parthasarathy P Vivekanandan S. An extensive study on the COVID-19 pandemic, an emerging global crisis: Risks, transmission, impacts and mitigation Journal of Infection and Public Health 14 2 2021 249 259 33493922 16 Chaudhuri S Basu S Kabi P Unni VR Saha A. Modeling the role of respiratory droplets in Covid-19 type pandemics Physics of Fluids 32 6 2020 063309 17 Tang JW Marr LC Li Y Dancer SJ. Covid-19 has redefined airborne transmission British Medical Journal Publishing Group 2021 18 Gwenzi W. Leaving no stone unturned in light of the COVID-19 faecal-oral hypothesis? A water, sanitation and hygiene (WASH) perspective targeting low-income countries Science of the Total Environment 753 2021 141751 19 Mehraeen E Salehi MA Behnezhad F Moghaddam HR SeyedAlinaghi S. Transmission modes of COVID-19: a systematic review Infectious Disorders-Drug Targets (Formerly Current Drug Targets-Infectious Disorders) 21 6 2021 27 34 20 Zaki N Mohamed EA. The estimations of the COVID-19 incubation period: A scoping reviews of the literature Journal of infection and public health 14 5 2021 638 646 33848893 21 Quesada J López-Pineda A Gil-Guillén V Arriero-Marín J Gutiérrez F Carratala-Munuera C. Incubation period of COVID-19: A systematic review and meta-analysis Revista Clínica Española (English Edition) 221 2 2021 109 117 22 Mueller AL McNamara MS Sinclair DA. Why does COVID-19 disproportionately affect older people? Aging (albany NY) 12 10 2020 9959 32470948 23 Farshbafnadi M Zonouzi SK Sabahi M Dolatshahi M Aarabi MH. Aging & COVID-19 susceptibility, disease severity, and clinical outcomes: The role of entangled risk factors Experimental gerontology 154 2021 111507 24 Menni C Valdes AM Freidin MB Ganesh S Moustafa JSE-S Visconti A Loss of smell and taste in combination with other symptoms is a strong predictor of COVID-19 infection MedRxiv 2020 25 He X Cheng X Feng X Wan H Chen S Xiong M. Clinical symptom differences between mild and severe COVID-19 patients in China: a meta-analysis Frontiers in public health 8 2021 561264 26 Song Y Liu P Shi X Chu Y Zhang J Xia J SARS-CoV-2 induced diarrhoea as onset symptom in patient with COVID-19 Gut 69 6 2020 1143 1144 32139552 27 Zaim S Chong JH Sankaranarayanan V Harky A. COVID-19 and multiorgan response Current problems in cardiology 45 8 2020 100618 28 Devaux CA Rolain J-M Raoult D. ACE2 receptor polymorphism: Susceptibility to SARS-CoV-2, hypertension, multi-organ failure, and COVID-19 disease outcome Journal of Microbiology, Immunology and Infection. 53 3 2020 425 435 29 Li YC Bai WZ Hashikawa T. The neuroinvasive potential of SARS-CoV2 may play a role in the respiratory failure of COVID-19 patients Journal of medical virology 92 6 2020 552 555 32104915 30 Ye Q Wang B Mao J. The pathogenesis and treatment of theCytokine Storm'in COVID-19 Journal of infection 80 6 2020 607 613 32283152 31 Darif D Hammi I Kihel A Saik IEI Guessous F Akarid K. The pro-inflammatory cytokines in COVID-19 pathogenesis: What goes wrong? Microbial Pathogenesis 153 2021 104799 32 Ramasamy S Subbian S. Critical determinants of cytokine storm and type I interferon response in COVID-19 pathogenesis Clinical microbiology reviews 34 3 2021 e00299-20 33 Velavan TP Meyer CG. Mild versus severe COVID-19: laboratory markers International Journal of Infectious Diseases 95 2020 304 307 32344011 34 Zhu N Zhang D Wang W Li X Yang B Song J A novel coronavirus from patients with pneumonia in China, 2019 New England journal of medicine 2020 35 Gorbalenya AE Baker SC Baric RS de Groot RJ Drosten C Gulyaeva AA Severe acute respiratory syndrome-related coronavirus: The species and its viruses–a statement of the Coronavirus Study Group BioRxiv 2020 36 Yang P Wang X. COVID-19: a new challenge for human beings Cellular & Molecular Immunology 2020 37 Opriessnig T Huang YW. Coronavirus disease 2019 (COVID-19) outbreak: Could pigs be vectors for human infections? Xenotransplantation 27 2 2020 38 Vankadari N Wilce JA. Emerging COVID-19 coronavirus: glycan shield and structure prediction of spike glycoprotein and its interaction with human CD26 Emerging microbes & infections 9 1 2020 601 604 32178593 39 Chen J Wang R Wang M Wei G-W. Mutations strengthened SARS-CoV-2 infectivity Journal of molecular biology 432 19 2020 5212 5226 32710986 40 Bertoglio F Fühner V Ruschig M Heine PA Abasi L Klünemann T A SARS-CoV-2 neutralizing antibody selected from COVID-19 patients by phage display is binding to the ACE2-RBD interface and is tolerant to most known recently emerging RBD mutations BioRxiv 2021 2020.12. 03.409318 41 Sun Z Ren K Zhang X Chen J Jiang Z Jiang J Mass spectrometry analysis of newly emerging coronavirus HCoV-19 spike protein and human ACE2 reveals camouflaging glycans and unique post-translational modifications Engineering 7 10 2021 1441 1451 32904601 42 Chen X Li R Pan Z Qian C Yang Y You R Human monoclonal antibodies block the binding of SARS-CoV-2 spike protein to angiotensin converting enzyme 2 receptor Cellular & molecular immunology 17 6 2020 647 649 32313207 43 Silhol F Sarlon G Deharo J-C Vaïsse B. Downregulation of ACE2 induces overstimulation of the renin–angiotensin system in COVID-19: should we block the renin–angiotensin system? Hypertension Research 43 8 2020 854 856 32439915 44 Li G He X Zhang L Ran Q Wang J Xiong A Assessing ACE2 expression patterns in lung tissues in the pathogenesis of COVID-19 Journal of autoimmunity 112 2020 102463 45 Hu B Huang S Yin L. The cytokine storm and COVID-19 Journal of medical virology 93 1 2021 250 256 32592501 46 Yang L Xie X Tu Z Fu J Xu D Zhou Y. The signal pathways and treatment of cytokine storm in COVID-19 Signal transduction and targeted therapy 6 1 2021 1 20 33384407 47 Xu Z-S Shu T Kang L Wu D Zhou X Liao B-W Temporal profiling of plasma cytokines, chemokines and growth factors from mild, severe and fatal COVID-19 patients Signal transduction and targeted therapy 5 1 2020 1 3 32296011 48 Abers MS Delmonte OM Ricotta EE Fintzi J Fink DL de Jesus AAA An immune-based biomarker signature is associated with mortality in COVID-19 patients JCI insight 6 1 2021 49 Brzezinski RY Rabin N Lewis N Peled R Kerpel A Tsur AM Automated processing of thermal imaging to detect COVID-19 Scientific Reports 11 1 2021 1 10 33414495 50 Mohammed M Syamsudin H Al-Zubaidi S AKS RR Yusuf E. Novel COVID-19 detection and diagnosis system using IOT based smart helmet International Journal of Psychosocial Rehabilitation 24 7 2020 2296 2303 51 Rane KP. Design and development of low cost humanoid robot with thermal temperature scanner for COVID-19 virus preliminary identification International Journal 9 3 2020 52 Poljak M Korva M Knap Gašper N Fujs Komloš K Sagadin M Uršič T Clinical evaluation of the cobas SARS-CoV-2 test and a diagnostic platform switch during 48 hours in the midst of the COVID-19 pandemic Journal of Clinical Microbiology 58 6 2020 e00599-20 53 Kang T Lu J Yu T Long Y Liu G. Advances in nucleic acid amplification techniques (NAATs): COVID-19 point-of-care diagnostics as an example Biosensors and Bioelectronics 2022 114109 54 Lephart PR Bachman MA LeBar W McClellan S Barron K Schroeder L Comparative study of four SARS-CoV-2 Nucleic Acid Amplification Test (NAAT) platforms demonstrates that ID NOW performance is impaired substantially by patient and specimen type Diagnostic Microbiology and Infectious Disease 99 1 2021 115200 55 Butler-Laporte G Lawandi A Schiller I Yao M Dendukuri N McDonald EG Comparison of saliva and nasopharyngeal swab nucleic acid amplification testing for detection of SARS-CoV-2: a systematic review and meta-analysis JAMA internal medicine 181 3 2021 353 360 33449069 56 Hellou MM Górska A Mazzaferri F Cremonini E Gentilotti E De Nardo P Nucleic acid amplification tests on respiratory samples for the diagnosis of coronavirus infections: a systematic review and meta-analysis Clinical Microbiology and Infection 27 3 2021 341 351 33188933 57 Dai W-c Zhang H-w Yu J Xu H-j Chen H Luo S-p CT imaging and differential diagnosis of COVID-19 Canadian Association of Radiologists Journal 71 2 2020 195 200 32129670 58 Ahuja S Panigrahi BK Dey N Rajinikanth V Gandhi TK. Deep transfer learning-based automated detection of COVID-19 from lung CT scan slices Applied Intelligence 51 1 2021 571 585 34764547 59 Alshazly H Linse C Barth E Martinetz T. Explainable COVID-19 detection using chest CT scans and deep learning Sensors 21 2 2021 455 33440674 60 Harmon SA Sanford TH Xu S Turkbey EB Roth H Xu Z Artificial intelligence for the detection of COVID-19 pneumonia on chest CT using multinational datasets Nature communications 11 1 2020 1 7 61 Xiang J Yan M Li H Liu T Lin C Huang S Evaluation of enzyme-linked immunoassay and colloidal gold-immunochromatographic assay kit for detection of novel coronavirus (SARS-Cov-2) causing an outbreak of pneumonia (COVID-19) MedRxiv 2020 62 Ma H Zeng W He H Zhao D Jiang D Zhou P Serum IgA, IgM, and IgG responses in COVID-19 Cellular & molecular immunology 17 7 2020 773 775 32467617 63 Xiang F Wang X He X Peng Z Yang B Zhang J Antibody detection and dynamic characteristics in patients with coronavirus disease 2019 Clinical Infectious Diseases 71 8 2020 1930 1934 32306047 64 Eckbo EJ Locher K Caza M Li L Lavergne V Charles M. Evaluation of the BioFire® COVID-19 test and Respiratory Panel 2.1 for rapid identification of SARS-CoV-2 in nasopharyngeal swab samples Diagnostic microbiology and infectious disease 99 3 2021 115260 65 Creager HM Cabrera B Schnaubelt A Cox JL Cushman-Vokoun AM Shakir SM Clinical evaluation of the BioFire® Respiratory Panel 2.1 and detection of SARS-CoV-2 Journal of Clinical Virology 129 2020 104538 66 Consortium WST. Repurposed antiviral drugs for Covid-19—interim WHO solidarity trial results New England journal of medicine 384 6 2021 497 511 33264556 67 Yavuz S Ünal S. Antiviral treatment of COVID-19 Turkish journal of medical sciences 50 9 2020 611 619 32293834 68 Salmon DA Dudley MZ Brewer J Kan L Gerber JE Budigan H COVID-19 vaccination attitudes, values and intentions among United States adults prior to emergency use authorization Vaccine 39 19 2021 2698 2711 33781601 69 Moshkovits I Shepshelovich D. Emergency Use Authorizations of COVID-19–Related Medical Products JAMA Internal Medicine 182 2 2022 228 229 34928303 70 Calvo-Alvarez E Dolci M Perego F Signorini L Parapini S D'Alessandro S Antiparasitic Drugs against SARS-CoV-2: A Comprehensive Literature Survey Microorganisms. 10 7 2022 1284 35889004 71 Omrani AS Pathan SA Thomas SA Harris TR Coyle PV Thomas CE Randomized double-blinded placebo-controlled trial of hydroxychloroquine with or without azithromycin for virologic cure of non-severe Covid-19 EClinicalMedicine 29 2020 100645 72 https://www.who.int/news/item/04-07-2020-who-discontinues-hydroxychloroquine-and-lopinavir-ritonavir-treatment-arms-for-covid-19#:∼:text=WHO%20today%20accepted%20the%20recommendation. 73 MacNeil JR Su JR Broder KR Guh AY Gargano JW Wallace M Updated recommendations from the Advisory Committee on Immunization Practices for use of the Janssen (Johnson & Johnson) COVID-19 vaccine after reports of thrombosis with thrombocytopenia syndrome among vaccine recipients—United States, April 2021 Morbidity and Mortality Weekly Report 70 17 2021 651 33914723 74 Shay DK. Safety monitoring of the Janssen (Johnson & Johnson) COVID-19 vaccine—United States, March–April 2021 MMWR Morbidity and mortality weekly report 70 2021 75 Rosenblum HG Hadler SC Moulia D Shimabukuro TT Su JR Tepper NK Use of COVID-19 vaccines after reports of adverse events among adult recipients of Janssen (Johnson & Johnson) and mRNA COVID-19 vaccines (Pfizer-BioNTech and Moderna): update from the Advisory Committee on Immunization Practices—United States, July 2021 Morbidity and Mortality Weekly Report 70 32 2021 1094 34383735 76 Watkins K Griffin G Septaric K Simon EL. Myocarditis after BNT162b2 vaccination in a healthy male The American journal of emergency medicine 50 2021 815 e1-. e2 77 Vahedifard F Chakravarthy K. Nanomedicine for COVID-19: the role of nanotechnology in the treatment and diagnosis of COVID-19 Emergent materials 4 1 2021 75 99 33615140 78 Balkrishna A Arya V Rohela A Kumar A Verma R Kumar D Nanotechnology interventions in the management of COVID-19: prevention, diagnosis and virus-like particle vaccines Vaccines 9 10 2021 1129 34696237 79 Sheta SM El-Sheikh SM. Nanomaterials and metal-organic frameworks for biosensing applications of mutations of the emerging viruses Analytical Biochemistry 2022 114680 80 Martin J Tena N Asuero AG. Current state of diagnostic, screening and surveillance testing methods for COVID-19 from an analytical chemistry point of view Microchemical Journal 167 2021 106305 81 Shukla S Upadhyay V Maurya VK. Evaluating the efficiency of specimen (sample) pooling for real-time PCR based diagnosis of COVID-19 Indian Journal of Medical Microbiology 39 3 2021 339 342 33781658 82 Anselmo AC Mitragotri S. Nanoparticles in the clinic: An update post COVID-19 vaccines Bioengineering & translational medicine 6 3 2021 e10246 34514159 83 Abdelhamid HN Badr G. Nanobiotechnology as a platform for the diagnosis of COVID-19: a review Nanotechnology for Environmental Engineering 6 1 2021 1 26 84 Zhu X Wang X Han L Chen T Wang L Li H Reverse transcription loop-mediated isothermal amplification combined with nanoparticles-based biosensor for diagnosis of COVID-19 MedRxiv 2020 85 Medhi R Srinoi P Ngo N Tran H-V Lee TR. Nanoparticle-based strategies to combat COVID-19 ACS Applied Nano Materials 3 9 2020 8557 8580 86 Abid SA Muneer AA Al-Kadmy IM Sattar AA Beshbishy AM Batiha GE-S Biosensors as a future diagnostic approach for COVID-19 Life sciences 273 2021 119117 87 Li H Rothberg L. Colorimetric detection of DNA sequences based on electrostatic interactions with unmodified gold nanoparticles Proceedings of the National Academy of Sciences 101 39 2004 14036 14039 88 Moitra P Alafeef M Dighe K Frieman MB Pan D. Selective naked-eye detection of SARS-CoV-2 mediated by N gene targeted antisense oligonucleotide capped plasmonic nanoparticles ACS nano 14 6 2020 7617 7627 32437124 89 Aithal S Mishriki S Gupta R Sahu RP Botos G Tanvir S SARS-CoV-2 detection with aptamer-functionalized gold nanoparticles Talanta 236 2022 122841 90 Kim H Park M Hwang J Kim JH Chung D-R Lee K-s Development of label-free colorimetric assay for MERS-CoV using gold nanoparticles ACS sensors 4 5 2019 1306 1312 31062580 91 Fu Z Zeng W Cai S Li H Ding J Wang C Porous Au@ Pt nanoparticles with superior peroxidase-like activity for colorimetric detection of spike protein of SARS-CoV-2 Journal of colloid and interface science 604 2021 113 121 34265672 92 Karakuş E Erdemir E Demirbilek N Liv L. Colorimetric and electrochemical detection of SARS-CoV-2 spike antigen with a gold nanoparticle-based biosensor Analytica chimica acta 1182 2021 338939 93 T-a Yano T Kajisa Ono M Miyasaka Y Hasegawa Y Saito A Ultrasensitive detection of SARS-CoV-2 nucleocapsid protein using large gold nanoparticle-enhanced surface plasmon resonance Scientific reports 12 1 2022 1 8 34992227 94 Rasmi Y Saloua KS Nemati M Choi JR. Recent progress in nanotechnology for COVID-19 prevention, diagnostics and treatment Nanomaterials 11 7 2021 1788 34361174 95 Pishva P Yüce M. Nanomaterials to tackle the COVID-19 pandemic Emergent materials 4 1 2021 211 229 33615139 96 Lee AH Gessert SF Chen Y Sergeev NV Haghiri B. Preparation of iron oxide silica particles for Zika viral RNA extraction Heliyon 4 3 2018 e00572 97 Wang J Ali Z Si J Wang N He N Li Z. Simultaneous extraction of DNA and RNA from hepatocellular carcinoma (Hep G2) based on silica-coated magnetic nanoparticles Journal of nanoscience and nanotechnology 17 1 2017 802 806 29634167 98 Somvanshi S B. Kharat P B. Saraf T S. Somwanshi S B. Shejul S B. Jadhav K M. Multifunctional nano-magnetic particles assisted viral RNA-extraction protocol for potential detection of COVID-19 Materials Research Innovations 25 3 2021 169 174 99 Ibrahim Fouad G. A proposed insight into the anti-viral potential of metallic nanoparticles against novel coronavirus disease-19 (COVID-19) Bulletin of the National Research Centre 45 1 2021 1 22 100 Takemura K. Surface plasmon resonance (SPR)-and localized SPR (LSPR)-based virus sensing systems: Optical vibration of nano-and micro-metallic materials for the development of next-generation virus detection technology Biosensors 11 8 2021 250 34436053 101 Chaudhary V Royal A Chavali M Yadav S. Advancements in research and development to combat COVID-19 using nanotechnology Nanotechnology for Environmental Engineering 6 1 2021 1 15 102 Gowri A Kumar NA Anand BS. Recent advances in nanomaterials based biosensors for point of care (PoC) diagnosis of COVID-19–a minireview TrAC Trends in Analytical Chemistry 137 2021 116205 103 da Silva PB da Silva JR Rodrigues MC Vieira JA de Andrade IA Nagata T Detection of SARS-CoV-2 virus via dynamic light scattering using antibody-gold nanoparticle bioconjugates against viral spike protein Talanta 243 2022 123355 104 Díaz CR Lafuente-Gómez N Coutinho C Pardo D Alarcón-Iniesta H López-Valls M Development of colorimetric sensors based on gold nanoparticles for SARS-CoV-2 RdRp, E and S genes detection Talanta 243 2022 123393 105 Teengam P Siangproh W Tuantranont A Vilaivan T Chailapakul O Henry CS. Multiplex paper-based colorimetric DNA sensor using pyrrolidinyl peptide nucleic acid-induced AgNPs aggregation for detecting MERS-CoV, MTB, and HPV oligonucleotides Analytical chemistry 89 10 2017 5428 5435 28394582 106 Zhang Z Li D Wang X Wang Y Lin J Jiang S Rapid detection of viruses: based on silver nanoparticles modified with bromine ions and acetonitrile Chemical Engineering Journal 438 2022 135589 107 Gong P He X Wang K Tan W Xie W Wu P Combination of functionalized nanoparticles and polymerase chain reaction-based method for SARS-CoV gene detection Journal of nanoscience and nanotechnology 8 1 2008 293 300 18468073 108 Zhao Z Cui H Song W Ru X Zhou W Yu X. A simple magnetic nanoparticles-based viral RNA extraction method for efficient detection of SARS-CoV-2 BioRxiv 2020 109 Manivannan S Ponnuchamy K. Quantum dots as a promising agent to combat COVID-19 Applied organometallic chemistry 34 10 2020 e5887 32836625 110 Zhao Y Chen J Hu Z Chen Y Tao Y Wang L All-solid-state SARS-CoV-2 protein biosensor employing colloidal quantum dots-modified electrode Biosensors and Bioelectronics 202 2022 113974 111 Roh C Jo SK. Quantitative and sensitive detection of SARS coronavirus nucleocapsid protein using quantum dots-conjugated RNA aptamer on chip Journal of Chemical Technology & Biotechnology 86 12 2011 1475 1479 32336860 112 Mallakpour S Azadi E Hussain CM. Fight against COVID-19 pandemic with the help of carbon-based nanomaterials New Journal of Chemistry 45 20 2021 8832 8846 113 Serrano-Aroca Á Takayama K Tuñón-Molina A Seyran M Hassan SS Pal Choudhury P Carbon-based nanomaterials: promising antiviral agents to combat COVID-19 in the microbial-resistant era ACS nano 15 5 2021 8069 8086 33826850 114 Weiss C Carriere M Fusco L Capua I Regla-Nava JA Pasquali M Toward nanotechnology-enabled approaches against the COVID-19 pandemic ACS nano 14 6 2020 6383 6406 32519842 115 Seifi T Kamali AR. Antiviral performance of graphene-based materials with emphasis on COVID-19: a review Medicine in Drug Discovery 11 2021 100099 116 Srivastava A Dwivedi N Dhand C Khan R Sathish N Gupta M Potential of graphene-based materials to combat COVID-19: properties, perspectives, and prospects Materials Today Chemistry 18 2020 100385 117 Seo G Lee G Kim MJ Baek S-H Choi M Ku KB Rapid detection of COVID-19 causative virus (SARS-CoV-2) in human nasopharyngeal swab specimens using field-effect transistor-based biosensor ACS nano 14 4 2020 5135 5142 32293168 118 Li Y Peng Z Holl NJ Hassan MR Pappas JM Wei C MXene–graphene field-effect transistor sensing of influenza virus and SARS-CoV-2 ACS omega 6 10 2021 6643 6653 33748577 119 El-Said WA Al-Bogami AS Alshitari W Synthesis of gold nanoparticles@ reduced porous graphene-modified ITO electrode for spectroelectrochemical detection of SARS-CoV-2 spike protein Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 264 2022 120237 120 Shahdeo D Chauhan N Majumdar A Ghosh A Gandhi S. Graphene-Based Field-Effect Transistor for Ultrasensitive Immunosensing of SARS-CoV-2 Spike S1 Antigen ACS applied bio materials 5 7 2022 3563 3572 121 Payandehpeyman J Parvini N Moradi K Hashemian N. Detection of SARS-CoV-2 using antibody–antigen interactions with graphene-based nanomechanical resonator sensors ACS Applied Nano Materials 4 6 2021 6189 6200 122 Ang WL Lim RRX Ambrosi A Bonanni A Rapid electrochemical detection of COVID-19 genomic sequence with dual-function graphene nanocolloids based biosensor FlatChem 32 2022 100336 123 Li J Wu D Yu Y Li T Li K Xiao M-M Rapid and unamplified identification of COVID-19 with morpholino-modified graphene field-effect transistor nanosensor Biosensors and Bioelectronics 183 2021 113206 124 Zhao H Liu F Xie W Zhou T-C OuYang J Jin L Ultrasensitive supersandwich-type electrochemical sensor for SARS-CoV-2 from the infected COVID-19 patients using a smartphone Sensors and Actuators B: Chemical 327 2021 128899 125 Varghese R Salvi S Sood P Karsiya J Kumar D. Carbon nanotubes in COVID-19: A critical review and prospects Colloid and Interface Science Communications 46 2022 100544 126 Özmen EN Kartal E Turan MB Yazıcıoğlu A Niazi JH Qureshi A. Graphene and carbon nanotubes interfaced electrochemical nanobiosensors for the detection of SARS-CoV-2 (COVID-19) and other respiratory viral infections: A review Materials Science and Engineering: C 129 2021 112356 127 Thanihaichelvan M Surendran S Kumanan T Sutharsini U Ravirajan P Valluvan R Selective and electronic detection of COVID-19 (Coronavirus) using carbon nanotube field effect transistor-based biosensor: A proof-of-concept study Materials Today: Proceedings 49 2022 2546 2549 33996512 128 Zamzami MA Rabbani G Ahmad A Basalah AA Al-Sabban WH Ahn SN Carbon nanotube field-effect transistor (CNT-FET)-based biosensor for rapid detection of SARS-CoV-2 (COVID-19) surface spike protein S1 Bioelectrochemistry 143 2022 107982 129 Pinals RL Ledesma F Yang D Navarro N Jeong S Pak JE Rapid SARS-CoV-2 spike protein detection by carbon nanotube-based near-infrared nanosensors Nano Letters 21 5 2021 2272 2280 33635655 130 Kim SY Lee J-C Seo G Woo JH Lee M Nam J Computational Method-Based Optimization of Carbon Nanotube Thin-Film Immunosensor for Rapid Detection of SARS-CoV-2 Virus Small science 2 2 2022 2100111 131 Shao W Shurin MR Wheeler SE He X Star A. Rapid detection of SARS-CoV-2 antigens using high-purity semiconducting single-walled carbon nanotube-based field-effect transistors ACS applied materials & interfaces 13 8 2021 10321 10327 33596036 132 Jeong S González-Grandío E Navarro N Pinals RL Ledesma F Yang D Extraction of viral nucleic acids with carbon nanotubes increases SARS-CoV-2 quantitative reverse transcription polymerase chain reaction detection sensitivity ACS nano 15 6 2021 10309 10317 34105936 133 Rajil N Esmaeili S Neuman BW Nessler R Wu H-J Yi Z Quantum optical immunoassay: upconversion nanoparticle-based neutralizing assay for COVID-19 Scientific reports 12 1 2022 1 10 34992227 134 Rai M Bonde S Yadav A Bhowmik A Rathod S Ingle P Nanotechnology as a shield against COVID-19: current advancement and limitations Viruses 13 7 2021 1224 34202815 135 Martín Giménez VM Prado N Diez E Manucha W Reiter RJ New proposal involving nanoformulated melatonin targeted to the mitochondria as a potential COVID-19 treatment Future Medicine; 2020 2819 2821 136 Chauhan G Madou MJ Kalra S Chopra V Ghosh D Martinez-Chapa SO. Nanotechnology for COVID-19: therapeutics and vaccine research ACS nano 14 7 2020 7760 7782 32571007 137 Chaudhary KR Puri V Singh A Singh C. A review on recent advances in nanomedicines for the treatment of pulmonary tuberculosis Journal of Drug Delivery Science and Technology 2022 103069 138 Pooladanda V Thatikonda S Sunnapu O Tiwary S Vemula PK Talluri MK iRGD conjugated nimbolide liposomes protect against endotoxin induced acute respiratory distress syndrome Nanomedicine: Nanotechnology, Biology and Medicine 33 2021 102351 139 Lima TLC Feitosa RdC dos Santos-Silva E dos Santos-Silva AM Siqueira EMdS Machado PRL Improving encapsulation of hydrophilic chloroquine diphosphate into biodegradable nanoparticles: a promising approach against herpes virus simplex-1 infection Pharmaceutics 10 4 2018 255 30513856 140 AbouAitah K Allayh AK Wojnarowicz J Shaker YM Swiderska-Sroda A Lojkowski W. Nanoformulation Composed of Ellagic Acid and Functionalized Zinc Oxide Nanoparticles Inactivates DNA and RNA Viruses Pharmaceutics 13 12 2021 2174 34959455 141 Molinaro R Pasto A Taraballi F Giordano F Azzi JA Tasciotti E Biomimetic nanoparticles potentiate the anti-inflammatory properties of dexamethasone and reduce the cytokine storm syndrome: An additional weapon against COVID-19? Nanomaterials 10 11 2020 2301 33233748 142 He Q Lu J Liu N Lu W Li Y Shang C Antiviral Properties of Silver Nanoparticles against SARS-CoV-2: Effects of Surface Coating and Particle Size Nanomaterials 12 6 2022 990 35335803 143 Almanza-Reyes H Moreno S Plascencia-López I Alvarado-Vera M Patrón-Romero L Borrego B Evaluation of silver nanoparticles for the prevention of SARS-CoV-2 infection in health workers: In vitro and in vivo PLoS One 16 8 2021 e0256401 144 Jin H Zhao Z Lan Q Zhou H Mai Z Wang Y Nasal delivery of hesperidin/chitosan nanoparticles suppresses cytokine storm syndrome in a mouse model of acute lung injury Frontiers in Pharmacology 11 2021 592238 145 Alper Öztürk A Namlı İ Aygül A Cefaclor Monohydrate-Loaded Colon-Targeted Nanoparticles for Use in COVID-19 Dependent Coinfections and Intestinal Symptoms: Formulation, Characterization, Release Kinetics, and Antimicrobial Activity ASSAY and Drug Development Technologies 19 3 2021 156 175 33728979 146 Mehranfar A Izadyar M. Theoretical design of functionalized gold nanoparticles as antiviral agents against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) The journal of physical chemistry letters 11 24 2020 10284 10289 33226815 147 Dormont F Brusini R Cailleau C Reynaud F Peramo A Gendron A Squalene-based multidrug nanoparticles for improved mitigation of uncontrolled inflammation in rodents Science advances 6 23 2020 eaaz5466 32548259 148 Hanafy NA El-Kemary MA. Silymarin/curcumin loaded albumin nanoparticles coated by chitosan as muco-inhalable delivery system observing anti-inflammatory and anti COVID-19 characterizations in oleic acid triggered lung injury and in vitro COVID-19 experiment International Journal of Biological Macromolecules 198 2022 101 110 34968533 149 Ding Y Lv B Zheng J Lu C Liu J Lei Y RBC-hitchhiking chitosan nanoparticles loading methylprednisolone for lung-targeting delivery Journal of Controlled Release 341 2022 702 715 34933051 150 Khater SE El-Khouly A Abdel-Bar HM Al-Mahallawi AM Ghorab DM Fluoxetine hydrochloride loaded lipid polymer hybrid nanoparticles showed possible efficiency against SARS-CoV-2 infection International journal of pharmaceutics 607 2021 121023 151 Idris A Davis A Supramaniam A Acharya D Kelly G Tayyar Y A SARS-CoV-2 targeted siRNA-nanoparticle therapy for COVID-19 Molecular Therapy 29 7 2021 2219 2226 33992805 152 Martins ES Espindola A Britos TN Chagas C Barbosa E Castro CE Potential Use of DMSA-Containing Iron Oxide Nanoparticles as Magnetic Vehicles against the COVID-19 Disease ChemistrySelect 6 31 2021 7931 7935 34541297 153 Hamouda T Ibrahim HM Kafafy H Mashaly H Mohamed NH Aly NM. Preparation of cellulose-based wipes treated with antimicrobial and antiviral silver nanoparticles as novel effective high-performance coronavirus fighter International Journal of Biological Macromolecules 181 2021 990 1002 33864870 154 Archana K Rajagopal R Krishnaswamy VG Aishwarya S. Application of green synthesised copper iodide particles on cotton fabric-protective face mask material against COVID-19 pandemic Journal of Materials Research and Technology 15 2021 2102 2116 35864980 155 Deng Y-Q Zhang N-N Zhang Y-F Zhong X Xu S Qiu H-Y Lipid nanoparticle-encapsulated mRNA antibody provides long-term protection against SARS-CoV-2 in mice and hamsters Cell research 32 4 2022 375 382 35210606 156 Sanna V Satta S Hsiai T Sechi M. Development of targeted nanoparticles loaded with antiviral drugs for SARS-CoV-2 inhibition European journal of medicinal chemistry 231 2022 114121 157 Zheng J Lu C Ding Y Zhang J Tan F Liu J Red blood cell-hitchhiking mediated pulmonary delivery of ivermectin: Effects of nanoparticle properties International journal of pharmaceutics 619 2022 121719 158 Jafari A Danesh Pouya F Niknam Z Abdollahpour‑Alitappeh M Rezaei-Tavirani M Rasmi Y Current advances and challenges in COVID-19 vaccine development: from conventional vaccines to next-generation vaccine platforms Molecular biology reports 2022 1 15 159 Kisby T Yilmazer A Kostarelos K. Reasons for success and lessons learnt from nanoscale vaccines against COVID-19 Nature Nanotechnology 16 8 2021 843 850 160 Shin MD Shukla S Chung YH Beiss V Chan SK Ortega-Rivera OA COVID-19 vaccine development and a potential nanomaterial path forward Nature nanotechnology 15 8 2020 646 655 161 Walsh EE Frenck RW Jr Falsey AR Kitchin N Absalon J Gurtman A Safety and immunogenicity of two RNA-based Covid-19 vaccine candidates New England Journal of Medicine 383 25 2020 2439 2450 33053279 162 Elia U Ramishetti S Rosenfeld R Dammes N Bar-Haim E Naidu GS Design of SARS-CoV-2 hFc-conjugated receptor-binding domain mRNA vaccine delivered via lipid nanoparticles ACS nano 15 6 2021 9627 9637 33480671 163 Rao L Xia S Xu W Tian R Yu G Gu C Decoy nanoparticles protect against COVID-19 by concurrently adsorbing viruses and inflammatory cytokines Proceedings of the National Academy of Sciences 117 44 2020 27141 27147 164 Smith TR Patel A Ramos S Elwood D Zhu X Yan J Immunogenicity of a DNA vaccine candidate for COVID-19 Nature communications 11 1 2020 1 13 165 Gu M Torres JL Li Y Van Ry A Greenhouse J Wallace S One dose of COVID-19 nanoparticle vaccine REVC-128 protects against SARS-CoV-2 challenge at two weeks post-immunization Emerging Microbes & Infections 10 1 2021 2016 2029 34651563 166 Yang R Deng Y Huang B Huang L Lin A Li Y A core-shell structured COVID-19 mRNA vaccine with favorable biodistribution pattern and promising immunity Signal transduction and targeted therapy 6 1 2021 1 10 33384407 167 Shinde V Bhikha S Hoosain Z Archary M Bhorat Q Fairlie L Efficacy of NVX-CoV2373 Covid-19 vaccine against the B. 1.351 variant New England Journal of Medicine 384 20 2021 1899 1909 33951374 168 Zhang Y-N Paynter J Sou C Fourfouris T Wang Y Abraham C Mechanism of a COVID-19 nanoparticle vaccine candidate that elicits a broadly neutralizing antibody response to SARS-CoV-2 variants Science Advances 7 43 2021 eabj3107 34669468 169 Thomas SJ Moreira ED Jr Kitchin N Absalon J Gurtman A Lockhart S Safety and efficacy of the BNT162b2 mRNA Covid-19 vaccine through 6 months New England Journal of Medicine 385 19 2021 1761 1773 34525277 170 Kremsner P Mann P Bosch J Fendel R Gabor JJ Kreidenweiss A Phase 1 assessment of the safety and immunogenicity of an mRNA-lipid nanoparticle vaccine candidate against SARS-CoV-2 in human volunteers MedRxiv 2020 171 Kremsner PG Mann P Kroidl A Leroux-Roels I Schindler C Gabor JJ Safety and immunogenicity of an mRNA-lipid nanoparticle vaccine candidate against SARS-CoV-2 Wiener Klinische Wochenschrift 133 17 2021 931 941 34378087 172 McKay PF Hu K Blakney AK Samnuan K Bouton CR Rogers P Self-amplifying RNA SARS-CoV-2 lipid nanoparticle vaccine induces equivalent preclinical antibody titers and viral neutralization to recovered COVID-19 patients BioRXiv 2020 173 McKay PF Hu K Blakney AK Samnuan K Brown JC Penn R Self-amplifying RNA SARS-CoV-2 lipid nanoparticle vaccine candidate induces high neutralizing antibody titers in mice Nature communications 11 1 2020 1 7 174 Tang Z Zhang X Shu Y Guo M Zhang H Tao W. Insights from nanotechnology in COVID-19 treatment Nano today 36 2021 101019 175 Campos EV Pereira AE De Oliveira JL Carvalho LB Guilger-Casagrande M De Lima R How can nanotechnology help to combat COVID-19? Opportunities and urgent need Journal of Nanobiotechnology 18 1 2020 1 23 31898555 176 Rangayasami A Kannan K Murugesan S Radhika D Sadasivuni KK Reddy KR Influence of nanotechnology to combat against COVID-19 for global health emergency: A review Sensors International 2 2021 100079 177 Zhong H Zhu Z Lin J Cheung CF Lu VL Yan F Reusable and recyclable graphene masks with outstanding superhydrophobic and photothermal performances ACS nano 14 5 2020 6213 6221 32329600 178 Huang L Xu S Wang Z Xue K Su J Song Y Self-reporting and photothermally enhanced rapid bacterial killing on a laser-induced graphene mask ACS nano 14 9 2020 12045 12053 32790338 179 Chen Y-N Hsueh Y-H Hsieh C-T Tzou D-Y Chang P-L. Antiviral activity of graphene–silver nanocomposites against non-enveloped and enveloped viruses International journal of environmental research and public health 13 4 2016 430 27104546 180 Nakamura S Sato M Sato Y Ando N Takayama T Fujita M Synthesis and application of silver nanoparticles (Ag NPs) for the prevention of infection in healthcare workers International journal of molecular sciences 20 15 2019 3620 31344881 181 Borkow G Zhou SS Page T Gabbay J. A novel anti-influenza copper oxide containing respiratory face mask PloS one 5 6 2010 e11295 20592763 182 Ungur G Hrůza J. Modified polyurethane nanofibers as antibacterial filters for air and water purification RSC advances 7 78 2017 49177 49187 183 Kampf G. Potential role of inanimate surfaces for the spread of coronaviruses and their inactivation with disinfectant agents Infection Prevention in Practice 2 2 2020 100044 184 Dhama K Patel SK Kumar R Masand R Rana J Yatoo M The role of disinfectants and sanitizers during COVID-19 pandemic: advantages and deleterious effects on humans and the environment Environmental Science and Pollution Research 28 26 2021 34211 34228 33991301 185 Rai M Deshmukh SD Ingle AP Gupta IR Galdiero M Galdiero S. Metal nanoparticles: The protective nanoshield against virus infection Critical reviews in microbiology 42 1 2016 46 56 24754250 186 Dalawai SP Aly MAS Latthe SS Xing R Sutar RS Nagappan S Recent advances in durability of superhydrophobic self-cleaning technology: a critical review Progress in Organic Coatings 138 2020 105381 187 Elechiguerra JL Burt JL Morones JR Camacho-Bragado A Gao X Lara HH Interaction of silver nanoparticles with HIV-1 Journal of nanobiotechnology 3 1 2005 1 10 15661076 188 Rogers JV Parkinson CV Choi YW Speshock JL Hussain SM. A preliminary assessment of silver nanoparticle inhibition of monkeypox virus plaque formation Nanoscale Research Letters 3 4 2008 129 133 189 Orlowski P Tomaszewska E Gniadek M Baska P Nowakowska J Sokolowska J Tannic acid modified silver nanoparticles show antiviral activity in herpes simplex virus type 2 infection PloS one 9 8 2014 e104113 190 D-x Xiang Q Chen Pang L C-l Zheng Inhibitory effects of silver nanoparticles on H1N1 influenza A virus in vitro Journal of virological methods 178 1-2 2011 137 142 21945220 191 De Gusseme B Sintubin L Baert L Thibo E Hennebel T Vermeulen G Biogenic silver for disinfection of water contaminated with viruses Applied and Environmental Microbiology 76 4 2010 1082 1087 20038697 192 Park S Ko Y-S Jung H Lee C Woo K Ko G. Disinfection of waterborne viruses using silver nanoparticle-decorated silica hybrid composites in water environments Science of the Total Environment 625 2018 477 485 29291562 193 Murray JP Laband SJ. Degradation of poliovirus by adsorption on inorganic surfaces Applied and Environmental Microbiology 37 3 1979 480 486 222209 194 Warnes SL Little ZR Keevil CW. Human coronavirus 229E remains infectious on common touch surface materials MBio 6 6 2015 e01697-15 195 Cao Y Gong Y Liao W Luo Y Wu C Wang M A review of cardiovascular toxicity of TiO2, ZnO and Ag nanoparticles (NPs) Biometals 31 4 2018 457 476 29748744 196 Muheem A Jahangir MA Jaiswal CP Jafar M Ahmad MZ Ahmad J Recent patents, regulatory issues, and toxicity of nanoparticles in neuronal disorders Current Drug Metabolism 22 4 2021 263 279 33305703 197 Card JW Zeldin DC Bonner JC Nestmann ER. Pulmonary applications and toxicity of engineered nanoparticles American Journal of Physiology-Lung Cellular and Molecular Physiology 295 3 2008 L400 LL11 18641236 198 Cao Y Li S Chen J. Modeling better in vitro models for the prediction of nanoparticle toxicity: A review Toxicology Mechanisms and Methods 31 1 2021 1 17 32972312 199 Huang Y-W Cambre M Lee H-J. The toxicity of nanoparticles depends on multiple molecular and physicochemical mechanisms International journal of molecular sciences 18 12 2017 2702 29236059 200 Li Y Ju D. The role of autophagy in nanoparticles-induced toxicity and its related cellular and molecular mechanisms Cellular and Molecular Toxicology of Nanoparticles 2018 71 84 201 Tuncel D Demir HV. Conjugated polymer nanoparticles Nanoscale 2 4 2010 484 494 20644748 202 Sharifi-Rad J Quispe C Butnariu M Rotariu LS Sytar O Sestito S Chitosan nanoparticles as a promising tool in nanomedicine with particular emphasis on oncological treatment Cancer Cell International 21 1 2021 1 21 33397383 203 Fu PP Xia Q Hwang H-M Ray PC Yu H. Mechanisms of nanotoxicity: generation of reactive oxygen species Journal of food and drug analysis 22 1 2014 64 75 24673904 204 Manke A Wang L Rojanasakul Y. Mechanisms of nanoparticle-induced oxidative stress and toxicity BioMed research international 2013 2013 205 Stern ST Adiseshaiah PP Crist RM Autophagy and lysosomal dysfunction as emerging mechanisms of nanomaterial toxicity Particle and fibre toxicology 9 1 2012 1 17 22239852 206 Mao B-H Tsai J-C Chen C-W Yan S-J Wang Y-J Mechanisms of silver nanoparticle-induced toxicity and important role of autophagy Nanotoxicology 10 8 2016 1021 1040 27240148 207 Li JJ Hartono D Ong C-N Bay B-H Yung L-YL Autophagy and oxidative stress associated with gold nanoparticles Biomaterials 31 23 2010 5996 6003 20466420 208 Yu Y Duan J Yu Y Li Y Liu X Zhou X Silica nanoparticles induce autophagy and autophagic cell death in HepG2 cells triggered by reactive oxygen species Journal of hazardous materials 270 2014 176 186 24583672 209 Liu H Zhang Y Yang N Zhang Y Liu X Li C A functionalized single-walled carbon nanotube-induced autophagic cell death in human lung cells through Akt–TSC2-mTOR signaling Cell death & disease 2 5 2011 e159 -e 21593791 210 Park E-J Zahari NEM Lee E-W Song J Lee J-H Cho M-H SWCNTs induced autophagic cell death in human bronchial epithelial cells Toxicology in Vitro 28 3 2014 442 450 24389112 211 Sun T, Yan Y, Zhao Y, Guo F, Jiang C. Copper oxide nanoparticles induce autophagic cell death in A549 cells. 2012. 212 Yu K-N Yoon T-J Minai-Tehrani A Kim J-E Park SJ Jeong MS Zinc oxide nanoparticle induced autophagic cell death and mitochondrial damage via reactive oxygen species generation Toxicology in Vitro 27 4 2013 1187 1195 23458966 213 Johnson BM Fraietta JA Gracias DT Hope JL Stairiker CJ Patel PR Acute exposure to ZnO nanoparticles induces autophagic immune cell death Nanotoxicology 9 6 2015 737 748 25378273 214 Li C Liu H Sun Y Wang H Guo F Rao S PAMAM nanoparticles promote acute lung injury by inducing autophagic cell death through the Akt-TSC2-mTOR signaling pathway Journal of molecular cell biology 1 1 2009 37 45 19516051 215 Li Y Zeng X Wang S Sun Y Wang Z Fan J Inhibition of autophagy protects against PAMAM dendrimers-induced hepatotoxicity Nanotoxicology 9 3 2015 344 355 24983897 216 Wang S Li Y Fan J Wang Z Zeng X Sun Y The role of autophagy in the neurotoxicity of cationic PAMAM dendrimers Biomaterials 35 26 2014 7588 7597 24906346 217 Wang J Yu Y Lu K Yang M Li Y Zhou X Silica nanoparticles induce autophagy dysfunction via lysosomal impairment and inhibition of autophagosome degradation in hepatocytes International Journal of Nanomedicine 12 2017 809 28182147 218 Zhang X Zhang H Liang X Zhang J Tao W Zhu X Iron oxide nanoparticles induce autophagosome accumulation through multiple mechanisms: lysosome impairment, mitochondrial damage, and ER stress Molecular pharmaceutics 13 7 2016 2578 2587 27287467

0

PMC9749399

NO-CC CODE

2022-12-15 23:23:19

no

2022 Dec 14;:100118

utf-8

oa_other

==== Front J Colloid Interface Sci J Colloid Interface Sci Journal of Colloid and Interface Science 0021-9797 1095-7103 Published by Elsevier Inc. S0021-9797(20)31677-5 10.1016/j.jcis.2020.12.021 Regular Article Multidomain drug delivery systems of β-casein micelles for the local oral co-administration of antiretroviral combinations Singh Chauhan Prakram a Abutbul Ionita Inbal a Moshe Halamish Hen b Sosnik Alejandro b Danino Dganit ac⁎ a CryoEM Laboratory of Soft Matter, Faculty of Biotechnology and Food Engineering, Technion-Israel Institute of Technology, Haifa 3200003, Israel b Laboratory of Pharmaceutical Nanomaterials Science, Department of Materials Science and Engineering, Technion-Israel Institute of Technology, Haifa 3200003, Israel c Guangdong Technion – Israel Institute of Technology, Shantou, Guangdong Province 515063, China ⁎ Corresponding author at: CryoEM Laboratory of Soft Matter, Faculty of Biotechnology and Food Engineering, Technion-Israel Institute of Technology, Haifa 3200003, Israel. 1 1 2021 15 6 2021 1 1 2021 592 156166 31 8 2020 20 11 2020 8 12 2020 © 2020 Published by Elsevier Inc. 2020 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Graphical abstract The antiretroviral (ARV) cocktail revolved the treatment of the human immunodeficiency virus (HIV) infection. Drug combinations have been also tested to treat other infectious diseases, including the recent coronavirus disease 2019 (COVID-19) outbreak. To simplify administration fixed-dose combinations have been introduced, however, oral anti-HIV therapy still struggles with low oral bioavailability of many ARVs. This work investigated the co-encapsulation of two clinically relevant ARV combinations, tipranavir (TPV):efavirenz (EFV) and darunavir (DRV):efavirenz (EFV):ritonavir (RTV), within the core of β-casein (bCN) micelles. Encapsulation efficiency in both systems was ~100%. Cryo-transmission electron microscopy and dynamic light scattering of the ARV-loaded colloidal dispersions indicate full preservation of the spherical morphology, and x-ray diffraction confirm that the encapsulated drugs are amorphous. To prolong the physicochemical stability the formulations were freeze-dried without cryo/lyoprotectant, and successfully redispersed, with minor changes in morphology. Then, the ARV-loaded micelles were encapsulated within microparticles of Eudragit® L100, which prevented enzymatic degradation and minimized drug release under gastric-like pH conditions in vitro. At intestinal pH, the coating polymer dissolved and released the nanocarriers and content. Overall, our results confirm the promise of this flexible and modular technology platform for oral delivery of fixed dose combinations. Keywords Combination therapy Antiretrovirals Darunavir, efavirenz and ritonavir β-casein micelles (bCN) Colloidal dispersion cryo-TEM Nanoparticle-in-microparticle delivery system (NiMDS) ==== Body pmc1 Introduction The Human immunodeficiency virus (HIV-1) infection is one of the major health burdens worldwide with approximately 37.9 million people living with HIV in 2018 [1]. An estimated 1.7 million individuals worldwide became newly infected with HIV in 2018 – about 5,000 new infections per day, as well as 940,000 people that died of HIV that year [1]. Different antiretroviral (ARV) combinations have been clinically trialed. Based on clinical evidence, the chronic administration of a minimum of three ARVs of at least two different families (out of the existing six) is critical to maintain undetectable viral levels in plasma over time and to prevent the progress from the infection to the active phase of the disease, the Acquired Immunodeficiency Syndrome (AIDS) [2], [3]. The ARV cocktail known as the highly active antiretroviral therapy (HAART) revolutionized the treatment of the disease, making it manageable and chronic. The rationale behind HAART is that the inhibition of the HIV replication cycle at least at two different stages results in therapeutic synergy. More recently, the clinical use of ARV combinations has been implemented in pre-exposure prophylaxis and to prevent HIV infection in non-infected individuals at very high risk. ARV combinations have been also clinically assessed in the treatment of viral diseases such as the severe acute respiratory syndrome (SARS) and lately in the corona virus 2019 (COVID-19) outbreak [4], [5], [6]. In late March 2020, the World Health Organization (WHO) launched the global clinical trial called SOLIDARITY comprising several thousands of patients worldwide that aims to find a possible treatment for this new disease [7]. In this framework, the efficacy of the lopinavir/ritonavir (LPV/RTV) combination to prevent infection or improve clinical outcomes will be assessed. Pill burden and complex administration regimens jeopardize patient compliance and may lead to low therapeutic efficacy due to poor adherence. Fixed-dose combinations (FDCs) are pharmaceutical products containing two or more active pharmaceutical ingredients (APIs) in one single dosage form. FDCs have been designed to reduce pill burden, simplify administration regimens, improve patient compliance, and constrain the risk of monotherapy-associated drug resistance [8]. There are more than 10 2-in-1 and 3-in-1 ARV FDCs approved by the US-Food and Drug Administration (US-FDA) [9]. The non-nucleoside reverse transcriptase inhibitor efavirenz (EFV) and the protease inhibitors darunavir (DRV) and tipranavir (TPV) are among the ARVs approved by the FDA and the European Medicines Agency (EMA) (Fig. 1 ). EFV [10] is a first-line ARV listed in the WHO Model List of Essential Medicines [11]. DRV is also a first-line drug listed as essential by the WHO, [11] while TPV is second-line due to severe side-effects, though still used in patients that have shown resistance to other protease inhibitors [12]. Protease inhibitors are always associated with a boosting agent such as RTV, taken in low dose [13]. A more selective boosting agent, cobicistat, was introduced in FDCs in 2012 [14]. RTV and cobicistat show similar efficacy [15]  but different side effects [16] with limited clinical data for the latter in high-risk subpopulations [17]. Cobicistat-containing FDCs are more costly than those of RTV and less affordable in developing countries [18]. RTV (but not cobicistat) is catalogued by the WHO as an essential drug [11].Fig. 1 Chemical structure of (A) EFV, (B) DRV, (C) TPV and (D) RTV. EFV, DRV and TPV are classified into Class II of the Biopharmaceutics Classification System (BCS), displaying low water solubility, high permeability and low oral bioavailability [17]. Conversely, RTV belongs to Class IV (low water solubility and low permeability) (Table S1). EFV displays low water solubility in a broad pH range between 1 and 8 and limited oral bioavailability [19]. Protease inhibitors are highly soluble in acid aqueous solutions [20] and present a dramatic solubility decrease at neutral or basic pH. For example, the water solubility of RTV decreases from 1.2 mg mL−1 at pH 1 to 5 mg mL−1 at neutral pH [21]. Thus, even if administered in the form of water-soluble salts or solvates, protease inhibitors precipitate in the small intestine in the form of relatively large particles with slow dissolution rate, a phenomenon that result in low oral bioavailability [22], [23]. In this context, the investigation of advanced drug delivery systems that localize the delivery in the gut is called for, to improve the efficacy and the quality of life of patients and to reduce the medication costs in the therapy of HIV [24], [25], [26]. Beta-casein (bCN, 24 kDa, 209 amino acids) is an unstructured calcium-sensitive phosphoprotein, comprising about 36% of the cow milk [27], [28]. The most advantageous feature of bCN is its amphiphilic structure that leads to self-assembly in aqueous solution, thereby forming stable colloidal micellar structures [27], [29], [30]. The monomer radius of gyration (Rg) ranges between 7.3 and 13.5 nm [27], [28] depending on temperature, pH, and ionic strength. The critical micellar concentration (CMC) of bCN also depends on these parameters, and at pH 7.0 and the temperature interval relevant to this study it is between 0.5 and 2.0 mg mL−1 [28]. In our earlier studies, we showed that bCN micelles can efficiently encapsulate and orally deliver high amount of hydrophobic therapeutics such as celecoxib and except that becoming more round, and importantly no change in the morphology of the micelles occur even after freeze- drying and resuspension [29], [30], [31]. The oral route has been recognized as the most patient-compliant, mainly owing to the minimal invasiveness, painless self-administration and possible use of solid formulations with better physicochemical stability and long shelf life and the ability to sustain and localize the release in different portions of the gastrointestinal tract (GIT) [32]. Micelles are common and accepted colloidal particles in drug delivery, and bCN micelles emerge as a versatile, biocompatible platform for oral drug delivery. However, they undergo degradation in the harsh acidic conditions of the gastric fluids (pH 1–2.5) [33], a process that is catalyzed by gastric enzymes such as pepsin [34]. Aiming to make a sound contribution to the combined ARV therapy of HIV and pave the way for the use of bCN micelles in oral drug delivery, in this work we investigated and characterized a Nanoparticle-in-Microparticle Drug Delivery System (NiMDS) 2-in-1 and 3-in-1 FDCs to improve the oral bioavailability and reduce the administration frequency of ARVs. Our results demonstrate the promise of these modular and versatile delivery constructs for the local oral delivery of ARV FDCs in the therapy of HIV and possibly other viral infections. 2 Materials and methods 2.1 Production of drug-loaded bCN dispersions Bovine bCN ( >97% purity, Sigma-Aldrich) was dissolved in 50 mM phosphate buffered saline (PBS, pH 7.0, MP Biomedicals) containing 5.65 mM NaH2PO4  (99% purity, Merck Millipore), 3.05 mM Na2HPO4 (99% purity, Merck Millipore), 80 mM NaCl (98% purity, Loba Chemie) and 0.02% sodium azide (≥99.5% purity, Sigma-Aldrich). The bCN colloidal system was prepared at 10 mg mL−1 (0.41 mM), above its CMC (0.5–2.0 mg mL−1, 0.021–0.083 mM at pH 7.0, 25 °C), where physically stable bCN micelles exist [28], [29]. The system was stirred overnight at 4 °C, forming a transparent colloidal dispersion. For loading bCN micelles with TPV (≥98% purity, MW of 602.66 g mol−1, Boehringer Ingelheim) and EFV (≥98% purity, MW of 315.67 g mol−1, Gilead Sciences), the drugs were dissolved in absolute alcohol (Bio-Lab Ltd.). For the preparation of bCN micelles loaded with DRV (98% purity, MW of 547.66 g mol−1, Leap Chem Co.), EFV and RTV (99% purity, MW of 720.94 g mol−1, Leap Chem Co.), they were dissolved in dimethyl sulfoxide (≥99% purity, DMSO, Bio-Lab Ltd.). As controls, each drugs mixture - TPV:EFV in ethanol (99% purity, Parchem) or DRV:EFV:RTV in DMSO - was added dropwise to PBS. All control drug dispersions were cloudy, indicating poor dispersibility and poor solubility of the drugs in buffer. For the dispersion, a known amount of each drugs mixture was added dropwise to bCN micelles (10 mg mL−1) under stirring for 30 min at 25 °C, at predetermined protein:drugs mole ratios. Ethanol or DMSO concentration in the final dispersion was always <5% v v-1. The drug-loaded dispersion bCN:TPV:EFV is designated as bCTE (1:8:8 protein:drugs mole ratio) and bCN:DRV:EFV:RTV is designated bCDER (1:8:6:1 protein:drugs mole ratio), and compositions are given in Table 1 . All of these dispersions were transparent indicating good encapsulation of the drugs within the micelles.Table 1 Composition of different formulations used in this study. bCN:TPV:EFV Total amount (mg) bCN:DRV:EFV:RTV Total amount (mg) Component bCN TPV EFV bCN DRV EFV RTV Amount (mg) 10 1.90 0.99 12.9 10 1.7 0.74 0.28 12.7 Drug-loaded bCN micelles were encapsulated within Eudragit® L100 (Evonik) using direct Nano Spray-Drying (see below) to form gastro-resistant microparticles that are fully stable under gastric pH conditions and dissolve fast at pH = 6–7 (small intestine). To confirm the copolymer coating of the micelles, an acidic protease solution from Aspergillus saitoi (0.6 U mg−1; Sigma-Aldrich) was used to challenge their stability in vitro. 2.2 Freeze-drying The drug-free and drug-loaded bCN colloidal dispersions were frozen in liquid nitrogen and then freeze-dried in an Alpha 1–4 LSC basic lyophilizer (Martin Christ) for 24 h. Samples were stored at 4 °C, then resuspended in PBS, to reach the original concentration of 10 mg mL−1, or a higher concentration of 50 mg mL−1. Resuspension was performed by weighing the dry powder, adding a measured volume of PBS, and stirring for 30 min at room temperature. Transparent dispersions were obtained. 2.3 Turbidity Turbidity measurements were performed using an Ultrospec 2100 Pro spectrophotometer (Amersham Biosciences Corp.) at a wavelength of 600 nm with a light path of 1 cm. bCTE and bCDER dispersions at predetermined molar ratios were characterized. 2.4 Characterization of drug-loaded bCN dispersions The size distribution (scattering angle of 173o) and zeta-potential (Z-potential) were measured by dynamic light scattering (DLS) in a Zetasizer Nano ZSP (Malvern Panalytical Ltd.) at 25 °C. Z-potential of bCTE and bCDER was measured before freeze-drying, and after resuspension in PBS. An Olympus BX51 light microscope (LM, Olympus Corp.) was operated at Nomarski differential interference contrast (DIC) optics to examine the drugs in PBS and drug-loaded bCN dispersions. One drop (5 μL) was placed on a glass slide and covered with a cover slide. Images were recorded digitally at magnifications of 10- to 60-fold, with an Olympus DP71 camera connected to the LM. Image processing was done using the Cell A software (Olympus Corp.). 2.5 Direct-imaging cryogenic-transmission electron microscopy, Cryo-TEM Samples for cryo-TEM were prepared in a closed, home-made specimen chamber that was saturated with water and maintained at a controlled temperature (25 °C). A small drop (6 μL) of each colloidal suspension was placed on a 200 mesh carbon-coated copper grid (Ted Pella, Inc.) held by tweezers. Excess sample was removed by blotting with a filter paper to create a thin liquid film filling the holes of the grid of the dispersion of the dispersion. The blotted grids were vitrified upon plunging into the cryogen, liquid ethane maintained at its freezing temperature (−183 °C), transferred to liquid nitrogen (LN2) and then stored in liquid nitrogen (−196 °C) until analysis. Samples were examined in a Tecnai 12 G2 transmission electron microscopy (TEM, FEI) at 120 kV with a Gatan 626 cryo-holder (Pleasanton) maintained below − 170 °C to avoid the crystallization of the vitreous ice. To minimize beam exposure and radiation damage, images were recorded under low-dose conditions to minimize radiation damage on a cooled UltraScan 1000 2 k × 2 k high-resolution charge-coupled device (CCD) camera (Gatan), using the Digital Micrograph software package (Gatan), and using methodologies we have developed, as previously described [35]. 2.6 Wide-angle X-ray diffraction, XRD XRD experiments of freeze-dried bCN, unprocessed drugs, (EFV, TPV, DRV and RTV) and freeze-dried drug-loaded bCN samples (bCTE and bCDER) were performed using a Philips PW 3020 powder diffractometer equipped with a graphite crystal monochromator. The operating conditions were CuKα radiation (0.154 nm), 40 kV and 40 mA, in 2θ recording range from 0° to 90°, at room temperature. 2.7 In vitro pH-dependent dissolution assay To assess the dissolution and re-precipitation of drugs in PBS media that mimics the pH conditions of the gastrointestinal tract in vitro, 2.89 mg of TPV:EFV and 2.75 mg of DRV:EFV:RTV (total drugs amount used for encapsulation) were dispersed in 0.001 M HCl s (pH 2.0, 2.0 mL) at 37 °C and gently stirred at 350 RPM for 30 min. The dispersion was analysed by LM. Then, 5 mM KOH aqueous solution was added to rise the pH from 2.0 to 6.5, and the dispersion was examined again by LM. 2.8 Production and characterization of Nanoparticle-in-Microparticle delivery Systems, NiMDS NiMDS containing bCTE and bCDER as the nanoparticulate component were produced by the redispersion of the drug-loaded micelles in Eudragit® L100 ethanol solution and spray-drying [36] and designated EbCTE and EbCDER, respectively. Briefly, 50 mg of freeze-dried drug-loaded bCN micelles were dissolved in water and mixed slowly with the copolymer solution prepared in absolute ethanol (5.0 mg mL−1), set the pH solution (7.5) and then spray-dried to form microparticles using a BUCHI B-90 HP Nano Spray Dryer (Flawil) in a closed loop system with the following process setup: 4.0 μm mesh; 80 °C inlet temperature; 22–25 °C outlet temperature; 100 L min−1 gas flow rate; 20–25% pump speed; 80% spray power; and 80–90 kHz frequency. As a control, drug-free NiMDSs were produced by dispersing bCN micelles in water and then in Eudragit® L100 solution in absolute alcohol as described above. Microparticles were stored at room temperature until characterization. Microparticles size and size distribution were assessed after dispersion in 10 mM sodium chloride (25 μg mL−1) using a Zetasizer Nano ZSP, at 25 °C. The size and morphology were further analyzed by scanning electron microscopy (SEM) using a Zeiss Ultra-Plus microscope (Carl Zeiss NTS GmbH). Images were acquired using secondary electrons at 2 keV and at a working distance of 2.5–4.0 mm. 2.9 Fourier-transform infrared spectroscopy, FTIR To confirm the efficient coating of the micelles with Eudragit® L 100, that is critical to ensure their stability under gastric-like pH conditions, bCN, pristine Eudragit® L 100 and the drug-free bCN microparticles were analyzed by FTIR spectrometry using a Nicolet 6700 FTIR spectrophotometer (Thermo Fisher Scientific,) at scanning range of 4000–500 cm−1 and resolution of 2 cm−1. 2.10 Drug release in vitro The release of DRV, EFV and RTV from the bCDER designated as bCD, bCE, bCR and from EbCDER designated as EbCD, EbCE, EbCR were assessed in vitro in both gastric- and intestinal-like pH conditions. Drug-loaded samples (bCDER; 12.75 mg mL−1 and EbCDER; 125 mg mL−1) were placed in a dialysis membrane (12–14 kDa MWCO, Spectrum Laboratories Inc.) and immersed in pH 2.0 (lactic acid solution) or pH 6.8 (PBS), at 37 °C, and magnetically stirred at 50 RPM for 48 h. At predetermined time points, release medium aliquots were sampled and replaced by fresh pre-heated medium. Aliquots were freeze-dried and re-dissolved in DMSO. Then, the concentration of each drug was determined by UV–Vis spectrophotometry at 265, 247 and 238 nm respectively, using a calibration curve in DMSO with concentrations up to 1 mg. From this, the cumulative dissolution (expressed in percentage) was calculated. Experiments were conducted in triplicates and results expressed as Mean ± S.D. 2.11 Protection against gastric enzymes To evaluate the protective effect of Eudragit® L100 against gastrointestinal enzymes under gastric pH conditions, bCDER (12.75 mg mL−1 and EbCDER; 125 mg mL−1) were incubated with acidic protease enzyme (6 U) in simulated gastric conditions (pH 2.0, 37 °C) with continuous shaking for 60 min. Samples (500 µL) were taken at every 10 min and centrifuged at 8,000 RPM   for 5 min, and supernatant were used for protease activity calculation, knowing that one unit of enzyme hydrolyzes casein to produce color equivalent to 1.0 μmole of tyrosine per min under standard conditions. 2.12 Statistical analysis Statistical testing was performed by one-way ANOVA for group analysis using GraphPad Prism (GraphPad Software). The results from three independent experiments are presented as mean values ± S.D. 3 Results and discussion 3.1 Production of drug-loaded bCN micelles NiMDS are comprised of one nanoparticulate and one microparticulate component. Aiming to investigate the potential of bCN micelles to serve as the nanocarrier in ARV NiMDS FDCs for oral drug delivery, we encapsulated one 2-in-1 (TRP:EFV) and one 3-in-1 (DRV:EFV:RTV) drug combination. The former combination is a prototype used for the optimization of the production process, while the latter is a clinically relevant ARV combination. We first characterized the solubility of these combinations in PBS. As expected, due to their poor solubility in PBS, these drug combinations form large crystals that precipitate with time (Fig. 2 A1, A2, B1, B2).Fig. 2 LM and DIC images showing floating crystals of free drug combinations: (A) TPV:EFV, at 1:1 mol ratio, and (B) DRV:EFV:RTV, at 8:6:1 mol ratio, and complete solubilization after encapsulation in bCN micelle: (C) bCTE, at 1:8:8 protein:drugs mole ratio, and (D) bCDER, at 1:8:6:1 protein:drugs mole ratio. bCN concentration is 10 mg mL−1, and compositions according to Table 1. After encapsulation of these hydrophobic ARV combinations within bCN micelles, their aqueous solubility was dramatically increased, resulting in completely transparent dispersions and the absence of drug crystals (Fig. 2C1, C2, D1, D2). These results are consistent with the formation of strong interactions between the drug and the hydrophobic domains of the bCN micelles, that result in their efficient encapsulation and the creation of stable drug-loaded bCN dispersions, in agreement with our previous reports for Celecoxib [30], [31]. 3.2 Characterization of the drug-loaded bCN colloidal dispersions To evaluate the stability and shelf life of the drug-loaded dispersions, turbidity was measured at selected time points up to ~ 2 weeks (0.5, 1, 24, 48, 96, 120, 144 and 312 h). All dispersions showed good physical stability, and remained clear and transparent over the entire incubation time. The optical density (OD) was 0.08 ± 0.01 and 0.040 ± 0.01, for bCTE and bCDER, respectively. This indicates that the ARV combinations were efficiently retained inside the bCN micelles (Fig. 1S). Then, we produced dry powders that usually display much higher physicochemical stability and longer shelf life than aqueous liquid dispersions. Superior to many other nanoparticulate drug delivery systems [37], these dispersions could be successfully freeze-dried without the addition of cryo/lyoprotectants, and, in addition, the dry drug-loaded micelles were stable for at least 6 months in the dry form. As shown in Fig. 3 , the drug-free bCN colloidal dispersions showed a single narrow peak with an average diameter of 25 ± 2 nm by DLS. Each of bCN-encapsulated dispersions (bCTE and bCDER) showed a single peak as well, yet with larger average diameters of 35 ± 1 nm and 37 ± 3 nm, respectively, before lyophilization and after lyophilization and resuspension to the original concentration in PBS (Table S2). The size growth upon drug encapsulation was consistent with the enlargement of the hydrophobic domains of the micelle, upon drugs encapsulation, as was also found for celecoxib [29]. All the dispersions were transparent to the naked eye before (Fig. 3a1, b1, c1) and after freeze-drying and resuspension (Fig. 3a2, b2, c2). Measurements indicated a slight increase in turbidity for the drug-loaded bCN micelles with respect to the unloaded control, probably owing to the size growth (Fig. 3). Z-potential of all the tested samples remained unchanged (approximately −12 mV) (Fig. 3). These findings highlight the high stability and processability of ARV-loaded bCN micelles, in good agreement with our results with other hydrophobic drugs [30], [31]. Remarkably, in this work, double and triple drug combinations were successfully encapsulated with complete preservation of the micellar morphology.Fig. 3 Right: Size distribution expressed as hydrodynamic diameters, of drug-free and drug-loaded bCN micelles (solid lines), with a mean diameter 25 ± 2 nm growing to 35 ± 1 nm and 37 ± 2 nm after encapsulation. No influence of drying on the mean diameter was observed after resuspension the dry dispersions in PBS to the original compositions (doted lines). The corresponding solutions were transparent both before (a1, b1, c1) and after (a2, b2, c2) lyophilization. Left: Turbidity (upper panel) and Z-potential (lower panel) of drug-free and drug-loaded bCN micelles before and after lyophilization and resuspension shows no change in the overall electro-kinetic potential suggesting good stability of the colloidal dispersion. The results from three independent experiments are presented as mean values ± S.D. To confirm the efficient drug encapsulation and disclose the micellar morphology, bCTE and bCDER before and after freeze-drying and resuspension were analyzed by cryo-TEM. bCTE micrographs show the presence of a homogenous population of small round and swollen micelles, with a diameter of 21 ± 3 nm (Fig. 4 A1) for the fresh dispersion. The size remained almost unchanged after freeze-drying and resuspension (Fig. 4A2). Cryo-TEM images of bCDER showed round micelles of 25 ± 3 nm before (Fig. 4B1) as well as after freeze-drying (Fig. 4B2). The latter are also larger which is consistent with the increased drug loading.Fig. 4 (A) Cryo-TEM images of bCTE (1:8:8 protein:drugs mole ratio): (A1) fresh and (A2) after freeze-drying and resuspension to the original concentration. (B) Cryo-TEM images of bCDER (1:8:8 protein:drugs mole ratio): (B1) fresh and (B2) after freeze-drying and resuspension to the original concentration. We further resuspended the freeze-dried bCDER powder in a smaller aqueous medium volume to reach a higher concentration (50 mg mL−1) than the original one. This concentration change might be relevant for the bench-to-bedside translation, as it would enable oral administration of much higher doses in a similar volume. The suspensions remained transparent and the micelles size and morphology remained mostly unchanged (Fig. 5 A, B). Sizes measured by DLS were in good agreement with cryo-TEM data [38]. These results together with the absence of crystals at the micro scale (by LM) or the nano-scale (by cryo-TEM) provide solid evidence of successful drug encapsulation. These results support our previous work where we demonstrated encapsulation of a high concentration (~25% w.w-1) of the hydrophobic inflammatory drug celecoxib within bCN [29], [30]. The present data clearly supports the suitability of bCN micelles for the encapsulation and oral delivery of ARVs. Another advantage of our approach is the ability to resuspend dry powders in much smaller volumes, which increases the drugs concentration (i.e. from 2.75 to 11.00 mg mL−1), making the dispersion clinically feasible and relevant [31].Fig. 5 (A) Cryo-TEM shows uniform bCDER micelles (1:8:6:1, protein:drugs mole ratio) after lyophilization and resuspension to a 5-fold higher concentration (50 mg mL−1) than the original one. (B) The colloidal dispersion remains transparent. XRD was performed to EFV, TRP, DRV, RTV and freeze-dried samples of bCN, bCTE and bCDER. Fig. 6 shows that all the pristine drugs are crystalline in nature presenting typical diffraction peaks of small-molecule organic powders. Conversely, upon encapsulation within bCN micelles, no diffraction peaks were detected, indicating that all drugs are in amorphous form or, in other words, molecularly dispersed within the bCN matrix, which further increases their suitability for oral drug delivery [30]. This as well is consistent with our finding for celecoxib being molecularly dispersed in bCN micelles, as found by solution and solid-state nuclear magnetic resonance [29]. But here, importantly, the amorphous encapsulation is confirmed also for drug combinations.Fig. 6 Representative XRD spectra of free drugs (A to D) compared to bCN (E) and bCN with encapsulated drugs (F to G). (A) EFV, (B) TRP, (C) DRV, (D) RTV (E) bCN micelles, (F) bCTE, and (G) bCDER. Measurements clearly show no peaks of drugs in the presence of bCN micelles indicating encapsulated drugs in an amorphous form. In earlier works, different ARVs were encapsulated in various (mainly polymeric) nanocarriers especially integrase strand transfer inhibitors i.e. dolutegravir, cabotegravir [39], nucleoside/nucleotide reverse transcriptase inhibitors i.e. emtricitabine, tenofovir alafenamide [3], non-nucleoside reverse transcriptase inhibitors i.e EFV, rilpivirine and protease inhibitors i.e. DRV, lopinavir with boosting agent i.e. RTV [23]. Nowacek et al [40] manufactured nanoparticles of atazanavir, EFV, and RTV (termed nanoART) by using human monocyte-derived macrophages as a carrier but these were not tested for oral delivery as they were envisioned for the targeting the central nervous system. Our strategy is substantially distinct from others that in most of cases encapsulate one single drug, as we demonstrate in this work the co-encapsulation, with almost 100% efficiency, of double and triple ARV combinations. As stressed above, TRP:EFV was utilized to optimize the process conditions and the bCN:drugs ratio to maximize encapsulation, DRV:EFV:RTV is an ARV combination administered once daily in healthy volunteers in the amount of 900, 600 and 100 mg respectively [41]. 3.3 Production and characterization of gastro-resistant Nanoparticle-in-Microparticle Delivery Systems As previously shown, upon oral administration, protease inhibitors initially undergo dissolution in the stomach and reprecipitate in the small intestine [22], [23]. To assess this behavior for all the ARVs used in this study, free drug combinations of TRP:EFV and DRV:EFV:RTV (composition as in Table 1) were dissolved in 0.001 M HCl solution of pH 2.0. As seen in Fig. 7 A1, B1, in both experiment drugs dissolved and formed solutions that were transparent to both the naked eye and under the LM. Upon pH increase to pH 6.8 (this mimics in vitro the transit of the drugs from the stomach to the small intestine), we noticed the presence of large drugs particles of several microns in size (Fig. 7A2, B2). These results indicate that these pH-dependent drugs undergo dissolution at low pH and then, upon neutralization, they re-precipitate in the form of microparticles. This phenomenon was demonstrated in vivo where DRV/RTV pure nanoparticles administered to rats by the oral route showed very similar pharmacokinetics to unprocessed drugs [23]. To prevent this, protease inhibitor nanoparticles need to be encapsulated within gastro-resistant capsules that release the drug locally in the small intestine [22], [23], [42].Fig. 7 LM micrographs of free drug combinations (A1) TPV:EFV and (B1) DRV:EFV:RTV showing dissolution occur at acidic pH 2.0 and (A2) TPV:EFV and (B2) DRV:EFV:RTV showing reprecipitation at neutral pH 6.5. Cuvettes showing appearance of dispersion in both the conditions. (C) HR-SEM micrographs of (C1) EbCTE and (C2) EbCDER showing spherical round shape microparticles. Eudragit® L100, a polyanion random copolymer of methyl methacrylate and methacrylic acid (1:1 M ratio), is extensively used as a film-coating pharmaceutical excipient for improving the oral delivery of ARVs and other small-molecule drugs and proteins [22]. This copolymer is poorly soluble under the gastric pH conditions and freely soluble at the intestinal pH. Thus, microencapsulation within Eudragit® L100 is a simple, scalable and translatable approach to protect the bCN from degradation and minimize the release of the ARVs in the stomach [43]. Once in the small intestine, this copolymer is anticipated to undergo dissolution, releasing the encapsulated ARVs from the drug-loaded micelles [22]. To ensure the efficient encapsulation of the drug-loaded bCN micelles in the macroparticles and maximize the yield, we used the spray-drying technology. In this framework, bCTE and bCDER dispersions were dispersed in an ethanol solution of the copolymer and spray-dried under optimized conditions as described in the experimental section. Ethanol is an optimal solvent because it is approved for pharmaceutical use and it does not affect the structure and size of the bCN micelles in a substantial manner. In addition, it does not dissolve any of the ARVs used in this study within the timeframe of the production process, thus preserving the integrity of the drug-loaded micelles. We successfully prepared microparticles for both micellar dispersions, namely EbCTE and EbCDER, having mean hydrodynamic diameter of 590 ± 90 nm and 540 ± 70 nm, respectively, as determined by DLS (Fig. 2SA). Z-potential values were −35 ± 3 mV and –38 ± 4 mV, respectively, in accordance with the negative charge provided by the carboxylic acid groups in the side-chain of Eudragit® L100 (Fig. 2SB). Taken together, size and Z-potential values suggest that the produced NiMDS will be physically stable in suspension owing to electrostatic repulsion. To gain more insight into the size and the morphology of EbCTE and EbCDER, they were examined by HR-SEM. We identified mainly a population of round-shaped, smooth-surfaced microparticles with a diameter of approximately 435 ± 55 nm for EbCTE and 470 ± 70 nm for EbCDER (Fig. 7C1, C2), confirming the successful production of NiMDS. To confirm the integrity of the Eudragit® L100 coating around the bCN micelles, FTIR analysis of bCN, pure Eudragit® L 100 and the drug-free bCN microparticles was performed. bCN showed characteristic peaks of amide I and II stretching at 1646 and 1534 cm−1, respectively, and at 718 cm−1 due to the stretching of CH2– groups (Fig. 3S), in good agreement with the literature [44]. The spectrum of Eudragit® L100 showed the characteristic carbonyl vibrations of the ester group at 1728 cm−1 (Fig. 3S). In addition, peaks at 2996 and 2953 cm−1 due to stretching of CH2– groups could be found in both bCN and Eudragit® L100 spectra [45], [46]. Interestingly, spectra of drug-free bCN microparticles showed a peak at 1728 cm−1 and the decrease of bCN characteristic amide bands, which confirmed that the copolymer coated the nanoparticles. 3.4 Drug release in vitro Eudragit® L100 coating was anticipated to prevent drugs dissolution under gastric-like pH conditions. As a proof of concept, we chose EbCDER - a combination of three clinically relevant ARVs - and its respective uncoated version (bCDER) to assess the drug release in vitro first under gastric-like and then intestinal-like pH conditions. This protocol mimics the transit of the dispersions along the GIT after oral administration [22], [23], [47]. At pH 2.0, after 1 h, bCDER showed a relatively fast release of DRV, EFV and RTV of approximately 28.0 ± 2.0%, 11.0 ± 100% and 34.0 ± 4.0%, respectively, (Fig. 8 A and Table S2). Conversely, EbCDER showed a dramatic decrease in the release to 4.0 ± 0.3%, 4.0 ± 0.5% and 20.0 ± 1.0% (Fig. 8A and Table S2), respectively. These results clearly indicate that the copolymer microparticle successfully protected the micelles and limited drug release at low pH. The small amount of drug that was released might be by presence of some free drug-loaded bCN micelles/at the surface of the microparticles or in solution. At pH 6.8, the cumulative release of DRV, EFV and RTV from the uncoated (free) micelles after 4 h was 44.0 ± 3.0%, 24.0 ± 3.0% and 46.0 ± 2.0%, respectively (Fig. 8B and Table S2). With EbCDER, values were 37.0 ± 3.0%, 20.0 ± 2.0% and 39.0 ± 3.0%, for DRV, EFV and RTV, respectively (Fig. 8B and Table S2). The differences in drug release between bCDER and EbCDER under neutral conditions was not statistically significant, confirming the dissolution of the Eudragit® L100 microparticle and the release of the ARV-loaded micelles [47]. The high drug release under neutral pH is beneficial for oral drug delivery as most drugs undergo absorption in the small intestine owing to the large absorption surface area and the longer residence time with respect to the stomach [33]. Overall, our results demonstrate the distinct stability and cumulative release profile of all drugs from both dispersions under gastrointestinal-like conditions with a substantial decrease at low pH. Recent work explored strategies as ligand switchable and chain responsive delivery system [48], [49], imaging guided combination therapy [50] and single-wavelength near-infrared (NIR) light-triggered multifunctional micelles for combinational treatment of photothermal therapy (PTT) and photodynamic therapy (PDT) [51] to deliver hydrophobic drugs in targeted site safely, with improved therapeutic potential and lower adverse reactions.Fig. 8 DRV, EFV and RTV release profile from bCDER (solid line) and EbCDER (dotted line) under (A) gastric pH condition; pH 2.0 and (B) intestinal pH condition; pH 6.8. The results from two independent experiments are presented as mean values ± S.D. 3.5 Protection against gastric enzymes Eudragit® L100 plays a dual role by reducing drug release and, at the same time, protecting bCN micelles from enzymatic proteolysis. To evaluate the performance of both dispersions, bCDER and its corresponding EbCDER (10 mg) were incubated in the presence of acidic protease enzyme (6 U) in simulated gastric conditions and the degradation was monitored over time. Fig. 4S depicts the pattern of bCDER degradation as a function of protease enzyme activity. After 10 min, 18% protease activity was detected, and the maximum enzyme activity has been observed at 40 min, confirming the degradation of bCN. After this time point, a reduction in enzyme activity is detected, which could be due to absence of substrate [52]. As expected, no enzymatic degradation could be recorded for EbCDER due to the efficient isolation of bCN micelles from the degradation medium. These findings confirmed that Eudragit® L100 prevents the proteolysis of drug- loaded bCN micelles under simulated gastric conditions. 4 Conclusions Combination therapy is needed for difficult-to-treat infections because of their potency and reduced development of drug resistance. We previously introduced the use of β-casein micelles in dry form and as a colloidal dispersion as effective vehicles for oral delivery of poorly soluble drugs, using celecoxib as a model drug [29], [30], [31]. Compared to earlier research [3], [39] where mainly polymers were used as a carrier for ARV drug, in the present work we reportfor the first time on successful production of, an innovative oral drug delivery system composed of double and triple ARV combinations co-encapsulated within bCN micelles and further encapsulated within Eudragit® L100 microparticles. The drug-loaded micelles were produced without any additives (e.g., surfactants), have a spherical shape and uniform size, and can successfully encapsulate high concentrations of designed combinations of hydrophobic drugs. In addition, they could undergo freeze-drying to produce stable and redispersible powders without the addition of any cryo/lyoprotectant. Further, our engineered NiMDS prevents the release of the cargos and protect the protein micelles under the acid and degradative conditions of the stomach, and are expected to release to the small intestine, which could be beneficial to increase their oral absorption and bioavailability. Overall, our results demonstrate the promising performance of this drug delivery platform to reduce the dose and the frequency of administration of ARV drugs, a crucial step to overcome the current patient-incompliant therapy. Furthermore, the use of all FDA-approved ingredients and of scalable processes remarkably increases the chances of bench-to-bedside translation. Finally, the strategy is flexible and modular, enabling the encapsulation of different qualitative and quantitative hydrophobic drug combinations. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A Supplementary data The following are the Supplementary data to this article:Supplementary data 1 Acknowledgment This work was supported by PBC-postdoctoral fellowship from the Israel Council for Higher Education and the Technion-Guangdong Fellowship. AS thanks the support of the Phyllis and Joseph Gurwin Fund for Scientific Advancement. The protease was kindly donated by Prof. Yuval Shoham (Technion, Haifa, Israel). We thank Dr. Inbal Abutbul Ionita for professional analysis of the Cryo-TEM work. Author contribution PSC performed most of experiments; IAI performed the Cryo-TEM work; HMH contributed to data validation; PSC, AS, DD – designed the study, analysed the data, and wrote the manuscript. Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.jcis.2020.12.021. ==== Refs References 1 The global HIV/AIDS epidemic. Available from: https://www.hiv.gov/hiv-basics/overview/data-and-trends/global-statistics, 2019 (accessed on 3 May 2020). 2 Caplan M.R. Daar E.S. Corado K.C. Next generation fixed dose combination pharmacotherapies for treating HIV Expert Opinion Pharmacother. 19 6 2018 589 596 10.1080/14656566.2018.1450866 3 Hobson J.J. Al-khouja A. Curley P. Meyers D. Flexner C. Siccardi M. Owen A. Meyers C.F. Rannard S.P. Semi-solid prodrug nanoparticles for long-acting delivery of water-soluble antiretroviral drugs within combination HIV therapies Nat. Commun. 10 1 2019 10.1038/s41467-019-09354-z 4 Chu C. Cheng V. Hung I. Wong M. Chan K. Chan K. Kao R. Poon L. Wong C. Guan Y. Role of lopinavir/ritonavir in the treatment of SARS: initial virological and clinical findings Thorax. 59 2004 252 256 14985565 5 Cohen J. Can an anti-HIV combination or other existing drugs outwit the new coronavirus? Science 2020 10.1126/science:abb0659 6 H. C. Coronavirus puts drug repurposing on the fast track. 2020 Journal [2020 17 March].Available from: https://www.nature.com/articles/d41587-020-00003-1, 2020 (accessed on 17 March 2020). 7 “Solidarity” clinical trial for COVID-19 treatments. https://www.who.int/emergencies/diseases/novel-coronavirus-2019/global-research-on-novel-coronavirus-2019-ncov/solidarity-clinical-trial-for-covid-19-treatments. 8 Atta M.G. De Seigneux S. Lucas G.M. Clinical Pharmacology in HIV Therapy Clin. J. Am. Soc. Nephrol. 14 3 2019 435 444 10.2215/CJN.02240218 29844056 9 Zhang X. Anti-retroviral drugs: current state and development in the next decade Acta. Pharm. Sin. B. 8 2 2018 131 136 10.1016/j.apsb:2018.01.012 29719774 10 N. Paengsai, G. Jourdain, N. Salvadori, A. Tantraworasin, J.Y. Mary, T.R. Cressey, R. Chaiwarith, C. Bowonwatanuwong, S. Bhakeecheep, N. Kosachunhanun, Recommended first-line antiretroviral therapy regimens and risk of diabetes mellitus in HIV-infected adults in resource-limited settings, in Open Forum Infect. Dis. 2019. Oxford University Press US. 11 WHO model lists of essential medicines. https://www.who.int/medicines/publications/essentialmedicines/en/, 2019 (accessed on 12 May 2020). 12 Cooper D.A. Cordery D.V. Zajdenverg R. Ruxrungtham K. Arastéh K. Bergmann F. de Andrade Neto J.L. Scherer J. Chaves R.L. Robinson P. Tipranavir/ritonavir (500/200 mg and 500/100 mg) was virologically non-inferior to lopinavir/ritonavir (400/100 mg) at week 48 in treatment-naïve HIV-1-infected patients: a randomized, multinational, multicenter trial Plos One. 11 2016 e0144917 13 Tseng A. Hughes C.A. Wu J. Seet J. Phillips E.J. Cobicistat versus ritonavir: similar pharmacokinetic enhancers but some important differences Ann. Pharmacother. 51 2017 1008 1022 28627229 14 Deeks E.D. Cobicistat: a review of its use as a pharmacokinetic enhancer of atazanavir and darunavir in patients with HIV-1 infection Drugs. 74 2014 195 206 24343782 15 Marzolini C. Gibbons S. Khoo S. Back D. Cobicistat versus ritonavir boosting and differences in the drug–drug interaction profiles with co-medications J. Antimicrob. Chemother. 71 7 2016 1755 1758 10.1093/jac/dkw032 26945713 16 Navarro J. Curran A. Profile of once-daily darunavir/cobicistat fixed-dose combination for the treatment of HIV/AIDS HIV/AIDS (Auckl). 8 2016 175 27843352 17 von Hentig N. Clinical use of cobicistat as a pharmacoenhancer of human immunodeficiency virus therapy HIV/AIDS (Auckl). 8 2016 1 16 26730211 18 Putcharoen O. Do T. Avihingsanon A. Ruxrungtham K. Rationale and clinical utility of the darunavir–cobicistat combination in the treatment of HIV/AIDS Drug Des. Devel. Ther. 9 2015 5763 5769 19 Cristofoletti R. Nair A. Abrahamsson B. Groot D.W. Kopp S. Langguth P. Polli J.E. Shah V.P. Dressman J.B. Biowaiver Monographs for Immediate Release Solid Oral Dosage Forms: Efavirenz J. Pharm. Sci. 102 2 2013 318 329 10.1002/jps.23380 23175470 20 Rubbens J. Brouwers J. Tack J. Augustijns P. Gastrointestinal dissolution, supersaturation and precipitation of the weak base indinavir in healthy volunteers Eur. J. Pharm. Biopharm. 109 2016 122 129 10.1016/j.ejpb.2016.09.014 27693678 21 Law D. Schmitt E.A. Marsh K.C. Everitt E.A. Wang W. Fort J.J. Krill S.L. Qiu Y. Ritonavir–PEG 8000 amorphous solid dispersions: in vitro and in vivo evaluations J. Pharm. Sci. 93 2004 563 570 14762895 22 Imperiale J.C. Nejamkin P. del Sole M.J. E. Lanusse C. Sosnik A. Novel protease inhibitor-loaded Nanoparticle-in-Microparticle Delivery System leads to a dramatic improvement of the oral pharmacokinetics in dogs Biomaterials 37 2015 383 394 10.1016/j.biomaterials.2014.10.026 25453966 23 Augustine R. Ashkenazi D.L. Arzi R.S. Zlobin V. Shofti R. Sosnik A. Nanoparticle-in-microparticle oral drug delivery system of a clinically relevant darunavir/ritonavir antiretroviral combination Acta Biomater. 74 2018 344 359 10.1016/j.actbio.2018.04.045 29723705 24 Sosnik A. Chiappetta D.A. Carcaboso Á.M. Drug delivery systems in HIV pharmacotherapy: what has been done and the challenges standing ahead J. Control. Release. 138 2009 2 15 19445981 25 Neves J. Sarmento B. Sosnik A. Biomedical engineering approaches for HIV/AIDS Prophylaxis, diagnostics and therapy Adv. Drug Deliv. Rev. 103 2016 1 27378403 26 Sosnik A. Augustine R. Challenges in oral drug delivery of antiretrovirals and the innovative strategies to overcome them Adv. Drug Deliv. Rev. 103 2016 105 120 10.1016/j.addr.2015.12.022 26772138 27 Portnaya I. Cogan U. Livney Y.D. Ramon O. Shimoni K. Rosenberg M. Danino D. Micellization of Bovine β-Casein Studied by Isothermal Titration Microcalorimetry and Cryogenic Transmission Electron Microscopy J. Agric. Food Chem. 54 15 2006 5555 5561 10.1021/jf060119c 16848545 28 Moitzi C. Portnaya I. Glatter O. Ramon O. Danino D. Effect of Temperature on Self-Assembly of Bovine β-Casein above and below Isoelectric pH. Structural Analysis by Cryogenic-Transmission Electron Microscopy and Small-Angle X-ray Scattering Langmuir 24 7 2008 3020 3029 10.1021/la702802a 18327957 29 Bachar M. Mandelbaum A. Portnaya I. Perlstein H. Even-Chen S. Barenholz Y. Danino D. Development and characterization of a novel drug nanocarrier for oral delivery, based on self-assembled β-casein micelles J. Control. Release. 160 2 2012 164 171 10.1016/j.jconrel.2012.01.004 22266050 30 Turovsky T. Khalfin R. Kababya S. Schmidt A. Barenholz Y. Danino D. Celecoxib encapsulation in β-casein micelles: structure, interactions, and conformation Langmuir. 31 2015 7183 7192 26068530 31 Perlstein H. Turovsky T. Gimeson P. Cohen R. Rubinstein A. Danino D. Barenholz Y. Thermotropic behavior of celecoxib-loaded beta-casein micelles: relevance to the improved bioavailability Eur. J. Nanomed. 7 2015 303 312 32 Banerjee A. Qi J. Gogoi R. Wong J. Mitragotri S. Role of nanoparticle size, shape and surface chemistry in oral drug delivery J. Control. Release. 238 2016 176 185 10.1016/j.jconrel.2016.07.051 27480450 33 Homayun B. Lin X. Choi H.-J. Challenges and recent progress in oral drug delivery systems for biopharmaceuticals Pharmaceutics. 11 2019 129 30893852 34 Ruiz G.A. Opazo-Navarrete M. Meurs M. Minor M. Sala G. van Boekel M. Stieger M. Janssen A.E.M. Denaturation and in Vitro Gastric Digestion of Heat-Treated Quinoa Protein Isolates Obtained at Various Extraction pH Food Biophys. 11 2 2016 184 197 10.1007/s11483-016-9429-4 27212897 35 Danino D. Cryo-TEM of soft molecular assemblies Curr. Opin. Colloid Interface Sci. 17 6 2012 316 329 10.1016/j.cocis.2012.10.003 36 Sosnik A. Seremeta K.P. Advantages and challenges of the spray-drying technology for the production of pure drug particles and drug-loaded polymeric carriers Adv. Colloid Interface Sci. 223 2015 40 54 10.1016/j.cis.2015.05.003 26043877 37 Abdelwahed W. Degobert G. Stainmesse S. Fessi H. Freeze-drying of nanoparticles: formulation, process and storage considerations Adv. Drug Deliv. Rev. 58 2006 1688 1713 17118485 38 Dutta B. Barick K. Verma G. Aswal V. Freilich I. Danino D. Singh B. Priyadarsini K. Hassan P. PEG coated vesicles from mixtures of Pluronic P123 and l-α-phosphatidylcholine: structure, rheology and curcumin encapsulation Phys. Chem. Chem. Phys. 19 2017 26821 26832 28949348 39 Kirtane A.R. Abouzid O. Minahan D. Bensel T. Hill A.L. Selinger C. Bershteyn A. Craig M. Mo S.S. Mazdiyasni H. Cleveland C. Rogner J. Lee Y.-A. Booth L. Javid F. Wu S.J. Grant T. Bellinger A.M. Nikolic B. Hayward A. Wood L. Eckhoff P.A. Nowak M.A. Langer R. Traverso G. Development of an oral once-weekly drug delivery system for HIV antiretroviral therapy Nat. Commun. 9 1 2018 10.1038/s41467-017-02294-6 40 Nowacek A.S. McMillan JoEllyn Miller R. Anderson A. Rabinow B. Gendelman H.E. Nanoformulated Antiretroviral Drug Combinations Extend Drug Release and Antiretroviral Responses in HIV-1-Infected Macrophages: Implications for NeuroAIDS Therapeutics J. Neuroimmune Pharmacol. 5 4 2010 592 601 10.1007/s11481-010-9198-7 20237859 41 Soon G.H. Shen P. Yong E.-L. Pham P. Flexner C. Lee L. Pharmacokinetics of Darunavir at 900 Milligrams and Ritonavir at 100 Milligrams Once Daily when Coadministered with Efavirenz at 600 Milligrams Once Daily in Healthy Volunteers Antimicrob. Agents Chemother. 54 7 2010 2775 2780 10.1128/AAC.01564-09 20385850 42 Xu B. Zhang W. Chen Y. Xu Y. Wang B.o. Zong L.i. Eudragit® L100-coated mannosylated chitosan nanoparticles for oral protein vaccine delivery Int. J. Biol. Macromol. 113 2018 534 542 10.1016/j.ijbiomac.2018.02.016 29408613 43 Kim J.S. Park J.H. Jeong S.C. Kim D.S. Yousaf A.M. Din F.U. Kim J.O. Yong C.S. Youn Y.S. Oh K.T. Jin S.G. Choi H.-G. Novel revaprazan-loaded gelatin microsphere with enhanced drug solubility and oral bioavailability J. Microencapsul. 35 5 2018 421 427 10.1080/02652048.2018.1515997 30136606 44 Farrell H. Jr Wickham E. Unruh J. Qi P. Hoagland P. Secondary structural studies of bovine caseins: temperature dependence of β-casein structure as analyzed by circular dichroism and FTIR spectroscopy and correlation with micellization Food Hydrocoll. 15 2001 341 354 45 Ptiček Siročić A. Kratofil Krehula L. Katančić Z. Hrnjak-Murgić Z. Characterization of casein fractions–Comparison of commercial casein and casein extracted from cow’s milk Chem. Biochem. Eng. 30 2016 501 509 46 Sharma M. Sharma V. Panda A.K. Majumdar D.K. Development of enteric submicron particle formulation of papain for oral delivery Int. J. Nanomedicine. 6 2011 2097 22114474 47 Thakral S. Thakral N.K. Majumdar D.K. Eudragit®: a technology evaluation Expert Opin. Drug. Deliv. 10 1 2013 131 149 10.1517/17425247.2013.736962 23102011 48 Ding Y. Liu J. Zhang Y. Li X. Ou H. Cheng T. Ma L. An Y. Liu J. Huang F. Liu Y. Shi L. Nanoscale Horiz. 4 2019 658 666 49 Ding X. Hong C. Zhang G. Liu J. Ouyang H. Wang M. Dong L. Zhang W. Xin H. Wang X. Nanoscale Horiz. 4 2019 1277 1285 50 Yang Z. Dai Y. Shan L. Shen Z. Wang Z. Yung B.C. Jacobson O. Liu Y. Tang W. Wang S. Lin L. Niu G. Huang P. Chen X. Nanoscale Horiz. 4 2019 426 433 31565239 51 Hu X. Tian H. Jiang W. Song A. Li Z. Luan Y. Small 14 2018 e1802994 52 George N. Chauhan P.S. Kumar V. Puri N. Gupta N. Approach to ecofriendly leather: characterization and application of an alkaline protease for chemical free dehairing of skins and hides at pilot scale J. Cleaner Prod. 79 2014 249 257

33652169

PMC9749415

NO-CC CODE

2022-12-15 23:23:19

no

J Colloid Interface Sci. 2021 Jun 15; 592:156-166

utf-8

J Colloid Interface Sci

10.1016/j.jcis.2020.12.021

oa_other

==== Front J Psychiatr Res J Psychiatr Res Journal of Psychiatric Research 0022-3956 1879-1379 Elsevier Ltd. S0022-3956(21)00007-8 10.1016/j.jpsychires.2021.01.006 Correspondence Increasing risks of domestic violence in India during COVID-19 pandemic Mondal Dinabandhu ∗ Paul Pintu Centre for the Study of Regional Development, School of Social Sciences, Jawaharlal Nehru University, New Delhi, India Karmakar Suranjana Department of Women's Studies, The University of Burdwan, West Bengal, India ∗ Corresponding author. 14 1 2021 3 2021 14 1 2021 135 9495 23 12 2020 4 1 2021 © 2021 Elsevier Ltd. All rights reserved. 2021 Elsevier Ltd Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Keywords Domestic violence COVID-19 Pandemic India ==== Body pmcDomestic violence against women is a violation of human rights and a serious public health concern worldwide (WHO, 2005). Women and girls are subject to different forms of violence even before their birth and until their death. Domestic violence has various dimensions and is often manifested in a complex forms of physical aggression, sexual coercion, emotional and psychological abuse, and controlling behaviours (Garcia-Moreno et al., 2006). Women of all ages, irrespective of socio-cultural and religious identities, geographic boundaries, and economic status are vulnerable to this inhuman act. As per estimates of World Health Organization, one in every third woman (35%) worldwide experiences physical and/or sexual violence in their lifetime and in most cases, the perpetrator is none other than their intimate partner, as about 30% women report experiences of violence committed by the partner (WHO, 2005). The incidence of violence against women remains unacceptably high in Indian society. The recent National Family Health Survey (NFHS-4) survey reveals that 31% ever-married women in this country experienced physical, sexual abuse, or emotional violence in 2015–16. Around one-fourth (24.5%) women are exposed to various violence-related injuries including cut, bruises, burns, deep wounds or broken bones, etc. Furthermore, nearly half (49%) of women reported that their behaviour is controlled by their partners (IIPS & ICF, 2017). As per National Crime Record Bureau (NCRB), a total of 89,097 cases were registered related to crimes against women across the country in 2018 with a 3.6% increase from the previous year. Domestic violence tops in the category of different crimes against women (Outlook, 2020). Violence against women in India is deeply rooted in the patriarchal family structure and rigid socio-cultural norms of society. In a patriarchal society, men are positioned in a higher order of societal and family structure, and they control over women in several ways. In fact, violence is an extreme form of discrimination linked to a continuum of beliefs that men think gives them the right to control women's behaviours (Heise, 1993; Jewkes et al., 2002). When several countries are under strict lockdown to restrain the spread of infectious COVID-19 virus, there has been a surge in cases of domestic violence all over the world. 10.13039/100004420 United Nations Population Fund (UNFPA), in collaboration with 10.13039/100007880 Johns Hopkins University (10.13039/100011408 USA ) and 10.13039/501100001784 Victoria University (Australia) estimated that 15 million additional cases of gender-based violence are likely to occur in every three months of lockdown. India is not an exception. Since the lockdown was imposed from March 25, the number of domestic violence cases has been increasing at an alarming pace across the country. Home, which is considered as the safest place, appears not safe for all. The National Commission for Women (NCW) received 587 complaints from women in just 24 days of lockdown in the country between March 23 and April 16. Out of these, 239 cases are related to domestic violence with a drastic rise from the previous month's 123 cases (February 27 to March 22) (The Economic Times, 2020). However, the actual number of cases is expected to be much more than the reported cases. During lockdown, women cannot go out for registering complaints due to restriction of movement, and some women cannot even communicate with their parents and friends regarding it as everyone is staying at home. The options, therefore, for lodging complaints are limited now under COVID lockdown. The pandemic has increased the risk of violence against women. This crisis has locked the perpetrators and the victim together. Amid lockdown, with everyone at home and unavailability of the domestic maid, the load of domestic work which is traditionally demarcated as ‘women's work’ has increased and so do the chances of violence, if they fail to fulfill these works. Another reason for the steep rise of domestic violence could be joblessness and economic distress in the family due to lockdown. Similar experiences were observed in United States during the Great Depression of the 1930s and the Great Recession between 2007 and 2009. The study of “Intimate Partner Violence in the Great Recession” found that men feel increasingly anxious about losing jobs and financial security, which tend to increase controlling behaviour and sometimes abuse their partners. It's a psychological dynamic that “loss of control in one domain, like the economy, leads men to assert greater control in another domain, in this case their intimate relationships” which stems from the patriarchal mindset (Schneider et al., 2016). However, it is more intensified in COVID-19 crisis as the lockdown restricts the options of escape routes for women. Earlier, they could go to the parent's home or friend's place to take shelter when violence escalates. Despite legal safeguard under Protection of Women from Domestic Violence Act (2005), domestic violence is rampant in Indian society, especially in this current emergency. Jammu and Kashmir High Court took Suo Motu cognizance of an increasing number of domestic violence cases and directed the government to make redress mechanism of violence as a key part of the national response plans for COVID-19. Similarly, Delhi High Court, hearing a PIL by an NGO, directed the State and Central Governments, the national and state commissions of women to take protective measures for women (Hindustan Times, 2020). During this lockdown, the situation is worst for women who are facing domestic violence as only few redressal systems are functional now. Keeping this in mind, the National Commission for Women launched a WhatsApp number to receive complaints from women in this lockdown (The Economic Times, 2020). Women should be informed regarding various modes of complaint registration to prevent domestic violence in a similar way, social distancing and use of masks are promoted to fight against COVID-19. Further, awareness should be spread about the risks of domestic violence at the community level. It has to be remembered that domestic violence is not only a violation of human rights but also it has far-reaching consequences on the psychological and physical health of women as well as on their children. All women deserve dignity, respect and freedom; only then, society will progress. Funding The authors did not receive any specific grant from any funding agency for this research. Author contributions All authors contributed equally to conceiving, writing, revising and proof reading this manuscript. Declaration of competing interest The authors have no conflicts of interest to declare. Acknowledgments None. ==== Refs References Garcia-Moreno C. Jansen H.A. Ellsberg M. Heise L. Watts C.H. Prevalence of intimate partner violence: findings from the WHO multi-country study on women's health and domestic violence The lancet 368 9543 2006 1260 1269 Heise L. Violence against women: the hidden health burden World Health Stat. Q. 46 1 1993 78 85 8237054 Hindustan Times Retrieved from https://www.hindustantimes.com/india-news/domestic-violence-during-covid-19-lockdown-emerges-as-serious-concern/story-mMRq3NnnFvOehgLOOPpe8J.html 2020 International Institute for Population Sciences (IIPS) & ICF National Family Health Survey (NFHS-4), 2015-16: India 2017 Retrieved from http://rchiips.org/nfhs/NFHS-4Reports/India.pdf Jewkes R. Levin J. Penn-Kekana L. Risk factors for domestic violence: findings from a South African cross-sectional study Soc. Sci. Med. 55 9 2002 1603 1617 12297246 Outlook Retrieved from https://www.outlookindia.com/newsscroll/domestic-violence-tops-crime-against-women-in-2018-ncrb/1704114 2020 09 January 2020 Schneider D. Harknett K. McLanahan S. Intimate partner violence in the Great recession Demography 53 2 2016 471 505 27003136 The Economic Times Retrieved from https://economictimes.indiatimes.com/news/politics-and-nation/india-witnesses-steep-rise-in-crime-against-women-amid-lockdown-587-complaints-received-ncw/articleshow/75201412.cms 2020 World Health Organization WHO Multi-Country Study on Women's Health and Domestic Violence against Women: Summary Report of Initial Results on Prevalence, Health Outcomes and Women's Responses 2005 Retrieved from https://www.who.int/gender/violence/who_multicountry_study/summary_report/summary_report_English2.pdf

33454588

PMC9749426

NO-CC CODE

2022-12-15 23:23:20

no

J Psychiatr Res. 2021 Mar 14; 135:94-95

utf-8

J Psychiatr Res

10.1016/j.jpsychires.2021.01.006

oa_other

==== Front J Comput Phys J Comput Phys Journal of Computational Physics 0021-9991 0021-9991 The Author(s). Published by Elsevier Inc. S0021-9991(21)00486-1 10.1016/j.jcp.2021.110591 110591 Article Computational modeling of protein conformational changes - Application to the opening SARS-CoV-2 spike Kucherova Anna a Strango Selma a Sukenik Shahar b Theillard Maxime a⁎ a Department of Applied Mathematics, University of California, Merced, CA 95343, USA b Department of Chemistry and Chemical Biology, University of California, Merced, CA 95343, USA ⁎ Corresponding author. 26 7 2021 1 11 2021 26 7 2021 444 110591110591 © 2021 The Author(s) 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. We present a new approach to compute and analyze the dynamical electro-geometric properties of proteins undergoing conformational changes. The molecular trajectory is obtained from Markov state models, and the electrostatic potential is calculated using the continuum Poisson-Boltzmann equation. The numerical electric potential is constructed using a parallel sharp numerical solver implemented on adaptive Octree grids. We introduce novel a posteriori error estimates to quantify the solution's accuracy on the molecular surface. To illustrate the approach, we consider the opening of the SARS-CoV-2 spike protein using the recent molecular trajectory simulated through the Folding@home initiative. We analyze our results, focusing on the characteristics of the receptor-binding domain and its vicinity. This work lays the foundation for a new class of hybrid computational approaches, producing high-fidelity dynamical computational measurements serving as a basis for protein bio-mechanism investigations. Keywords SARS-CoV-2 Covid-19 Spike protein Molecular trajectory Poisson-Boltzmann Multiscale modeling ==== Body pmc1 Introduction Proteins are polymers composed of amino acid chains that are folded into well-defined three-dimensional shapes. The structure and of this shape is crucial for proper protein function. However, in many proteins the three-dimensional structure is not enough, and function must be facilitated by specific intra-protein motions. [15], [18], [37], [39]. The SARS-CoV-2 spike (S) protein is one such protein [15], [18], [37], [39]. Naturally found as a homotrimer, the S-protein contains a buried receptor binding domain (RBD). The RBD has a high affinity to the human angiotensin-converting enzyme 2 (ACE2), and binding mediates viral entry to the cell. However, to facilitate ACE2 binding, the RBD must be exposed through a series of complex motions that occur in the unbound S protein trimer. [2], [18], [20], [33]. Recently, Zimmerman et al. [43] produced the first Markov state models simulation of the SARS-CoV-2 spike opening. This computational tour de force involved millions of citizen scientists collaborating through the Folding@home initiative [1], and produced an overall 0.1s of molecular trajectories. For a full characterization of the protein interaction, the molecular trajectory may not be enough. One aspect that may provide some insight into the interactions and function of a protein is the electrostatic potential it generates. For the SARS-CoV-2 spike, according to a previous study, the affinity constant (derived from the electrostatic potential) for the RBD of SARS-CoV-2 to the ACE2 is 10 to 15 times greater than that of SARS-CoV, potentially contributing to its transmission efficiency [37]. The reason for the higher binding affinity was attributed to several mutations, most notably from the residue Val404, found in SARS-CoV, to the positively charged Lys417 in SARS-CoV-2. This mutation resulted in an intensified electrostatic potential complementarity between the negatively charged ACE2 binding site and the now more positively charged RBD of SARS-CoV-2 [15], [37]. To understand the contribution of protein charge repositioning and also to help inform drug design strategies that leverage this distribution, it is imperative to know how the protein's electrostatic potential changes as it deforms. The Poisson-Boltzmann equation has long been recognized as the representation of choice to model the electric potential generated by proteins in solvents. It has drawn significant interest from the computational community since the pioneering calculations of Warwicker and Watson in the early 1980s [38], which has lead to the production of a broad variety of numerical solvers [4], [3], [6], [7], [13], [21], [17], [22] and open source software [12], [16], [32], built over the traditional spectrum of numerical methods. Such tools are, for example, employed in the context of drug development and discovery to calculate solvation free energies [10], [14], [28]. Using massively parallel architectures [11], these calculations can be carried out on considerably large proteins, such as the entire HIV-1 capsid (4,884,312 atoms, Protein database entry: 3J3Q). In this work, we combine Markov state model simulations with partial differential equation modeling to obtain dynamical electric potential maps of deforming proteins. To illustrate this novel approach, we leverage the S protein opening trajectory created by the Folding@home initiative to examine and characterize the potential of the SARS-CoV-2 S protein during trimer opening. At each frame of the simulated trajectory, we reconstruct the protein surface and calculate the generated electric potential with the approach developed by Mirzadeh et al. [26], [27]. Using adaptive non-graded Octree grids and sharp discretizations, we efficiently produce high-fidelity solutions to the non-linear Poisson-Boltzmann equation. In the continuum model, all temporal variations are neglected, and as a consequence, all potential maps are independent, making their computation embarrassingly parallelizable. To verify the accuracy of our method, we construct practical a posteriori error estimates for the surface representation and electrostatic properties. Our study of the electrostatic dynamics of the S protein opening reveals dramatic rearrangements of the electrostatic field during this process. These rearrangements act to localize a negatively charged field towards the interior of the S protein and expose a positive surface of its residue binding domain. This is in line with the negative charge of the target binding region on the ACE2 receptor and may aid the S protein in binding to its target with high affinity. The paper is arranged as follows: in section 2, we present the trajectory used for this study. Next, section 3 entails a thorough mathematical design for the calculation of the potential field that is done on each frame in the trajectory. Section 4 encapsulates the numerical method used along with a convergence study for computational validation. The dynamical electrostatic-geometric properties of the spike protein example are characterized in section 5, and conclusions are drawn in section 6. 2 Opening of the SARS-CoV-2 spike protein The S protein exists primarily in the closed configuration, hiding the three identical RBDs in its core [43]. The Folding@home trajectory focuses on the opening of the unglycosylated, uncleaved spike configuration through the detachment of a single monomer from the spike's core, revealing its associated binding site. A similar (but not identical) monomeric-opening state has recently shown to be populated in ≈16% of the unbound spike population in a recent cryoEM study [5]. 2.1 Simulated trajectory The trajectory, from Zimmerman et al. [43] contains a 71-frame sequence that captures the most populated pathway of the spike protein's transition from its closed to open state. This path was calculated utilizing a goal-oriented adaptive sampling algorithm (FAST, [42]) to favorably sample spike opening. Each frame consists of a list of N=51,671 atoms, represented by their positions xi=1..N, radius ri=1..N and fixed partial charge zi=1..N. Throughout these frames, the protein undergoes continuous deformations, which shift the receptor-binding domain of one of the monomers from being hidden while the spike is in its closed conformation to being fully exposed once the spike opens. Fig. 1 a and 1 b represent the protein's molecular structure in the open and closed configurations. Each one of its chains is depicted in a different color to illustrate the protein's trimeric structure. The revealed receptor-binding domain is located at the opening extremity of the red chain (see Fig. 1 a, 1 b).Fig. 1 SARS-CoV-2 Spike protein visualized in its closed (a) and open (b) states. The four pairs of atoms, along with the separating distances, used to quantify the gap opening are represented with straight solid colored lines. (c) The relative distance (Å) between arbitrarily chosen atom pairs is represented as a function of time. One atom is selected on the receptor-binding domain (RBD), and the other opposite the RBD across the opening of the spike protein. This distance is relative to the average distance between each respective atom pair (1, 2, 3, and 4 in the graph), hence the term' relative distance'. (d) The root mean squared deviation (RMSD) is calculated as a function of time with frame 1 (closed state) as the reference frame. (For interpretation of the colors in the figure(s), the reader is referred to the web version of this article.) Fig. 1 2.2 Relative opening measurement To provide preliminary quantification of the spike opening, dozens of atom pairs on opposing sides of the spike were chosen, and the distance between them measured as the spike transitions from one conformation to another (i.e. as a function of frame) (Fig. 1 a and b). One of the atoms in the pair was chosen from the binding interface, the monomer depicting in red in Fig. 1(a and b), while the second atom in the pair was chosen across the top of the spike opening. Although all pairs presented the same general trajectory, four of these atom pairs were arbitrarily selected as a subset for illustration. The resulting relative distance between the atom pairs is measured in Angstroms (Å), as shown in Fig. 1 c. The lines labeled 1, 2, 3, and 4 refer to the behavior of the four-pair subset, with their average behavior depicted in black. All four measurements generally follow the same trend. This extension happens continuously outside of frames 50 to 55, which show abrupt variations. 2.3 Magnitude of the conformation change To measure the total structural deformation, we compute the root-mean-square deviation (RMSD). It measures the average distance between atoms in the current position and some reference configuration, defined here as the initial closed configuration. Specifically(1) RMSD=1N∑i=1N|xin−xi1|2, where the superscripts n and 1 are for the current and initial states. Despite large variations at the initial and final stages, the RMSD evolution (depicted in Fig. 1 d), shares similar features with the evolution of the relative opening. These similarities suggest that the significant transformations that occur during the opening are recapitulated by the four distances in Fig. 1 c. 3 Continuum modeling This section describes the reconstruction of the potential map at each iteration from the current protein structure using the non-linear Poisson-Boltzmann (PB) equation. Because of the way the trajectory has been obtained, all frames are independent. Thus, the potential maps are decoupled and can be computed separately. A more comprehensive model, such as the Poisson–Boltzmann–Nernst–Planck model [41], [25], would consider the diffusion of the ions and introduce time derivatives in the partial differential equations. Because at the atomic scale (L=1 Å), for typical diffusivities (D≈10−9m2s−1), the ionic diffusion time scale (τD≈L2D=10−11s) is orders of magnitude smaller than the estimated opening duration (τO>>10−9s), these effects can be neglected and the static PB equation is a pertinent model. Readers not interested in the numerical specifics can skip the following sections (3 and 4). 3.1 Molecular surfaces A common way of portraying a molecular surface is by use of the van der Waals Surface depiction (see Fig. 2 b). Each atom in a molecule is depicted by a sphere with location xi=1...N and radius ri=1...N that is defined by an isosurface on their electron density. The union of these spheres, depicted in gray in Fig. 2, forms the van der Waals Surfaces (vdWS) of that molecule. The nature of the vdWS means that some regions might be identified as being exposed to the solvent, while in fact, their geometry makes them inaccessible to the solvent particles. As a consequence, we define the Solvent Accessible Surface (SAS [19]) of the protein, depicted as the blue outline in Fig. 2 b; it is formed through the addition of the solvent's particle radius, rP, to each ri=1..N resulting in a buffer around the vdWS. While the puffed-up SAS depiction may be useful for some areas of study, the Solvent Excluded Surface (SES) is preferable when discussing surface details of a molecule [31], as shown in Fig. 2b. As the name suggests, the inner molecule region defined by this surface includes all locations the solvent cannot occupy, including the vdWS and the tiny crevasses on its exterior, shown as concave black triangles at the meeting of atoms A and B in Fig. 2b.Fig. 2 (a) Visualization of the top view of the Spike protein molecule (colored by monomer). Color denotes the three monomers (b) A cutaway diagram of a general molecular surface depicting the Van der Waals Surface (vdWS), solvent-excluded surface (SES), and solvent-accessible surface (SAS). (c) SES representation of the Spike protein oriented the same way as Fig. (a) (top view). Fig. 2 3.2 Mathematical representation and numerical construction To represent the biomolecule, we employ the level set method [30] and capture the SES location as the zero level set of an auxiliary field ϕ(x) defined over the domain of interest Ω. The solvent Ω−, the Solvent Excluded Surface Γ, and the inside of the molecule Ω+ are defined as(2) Ω−={x∈R3|ϕ(x)<0} (3) Γ={x∈R3|ϕ(x)=0} (4) Ω+={x∈R3|ϕ(x)>0} The normal n to the interface Γ is defined as pointing toward Ω+. It is calculated as the normalized level set gradient(5) n=∇ϕ|∇ϕ|. The level set function is constructed following the approach proposed by Mirzadeh et al. in [27]. We start by constructing the level set function ϕSAS(x) representing the SAS as(6) ϕSAS(x)=maxi=1..N⁡(ri+rp−|x−x|). The SAS level set function is then reinitialized to be a signed-distance function (i.e. |∇ϕSAS|=1) by solving the reinitialization equation in fictitious time τ until the steady state is reached(7) ∂ϕSAS∂τ+sign(ϕSAS)|∇ϕSAS|−1)=0,∀x∈Ω. From the reinitialized function R(ϕSAS), the SES level set function is then obtained as(8) ϕ(x)=R(ϕSAS)(x)−rp,∀x∈Ω. As it was pointed out in [27], this procedure can create non-physical inner cavities. They are identified by finding physical points disconnected from the contour of the computational domain ∂Ω where the level set function ϕ(x) is positive. In practice, such points are isolated by solving the following Laplace problem(9) −△c=0,∀x∈Ω−, (10) c=0,∀x∈Γ, (11) c=1,∀x∈∂Ω, and detecting where ϕ(x)<0 and c=0. The cavities are removed by switching the sign of the level set function at these problematic positions. Finally, for computational purposes the SES level set is systematically reinitialized. 3.3 Poisson-Boltzmann equation The electrostatic potential, Ψ, around a biomolecule immersed in a binary z:z electrolyte solution can be described by the following non-linear Poisson-Boltzmann (PB) equation,(12) −∇⋅(ϵϵ0∇Ψ)+2cb(x)NAezsinh⁡(eΨkBT)=∑i=1Nqiδ(x−xi),∀x∈Ω∖Γ, where ϵ is the relative permittivity, being equal to ϵ+ in the molecule (Ω+) and ϵ− in the electrolyte (Ω−). ϵ0 is the permittivity of a vacuum, NA is the Avogadro number, cb is the bulk salt concentration, e is the elementary charge, kB is the Boltzmann constant, T is the temperature, z is the valence of the background electrolyte, qi and xi are the partial charge and position of the ith atom respectively, N is the total number of atoms in the molecule. cb(x)=0 inside the molecule. In non-dimensional form, the Poisson-Boltzmann equation takes the following form(13) −∇⋅(ϵ∇ψ)+κD2(x)sinh⁡(ψ)=∑i=1NλBziδ(x−xi),∀x∈Ω∖Γ where the characteristic length L is chosen to be 1Å, the potential is scaled by the thermal voltage (kBTe), zi is the non-dimensional partial charge on the ith atom, and κD=2cbNAe2zϵ0kBTL2 is the non-dimensional inverse of the Debye length. Inside the molecule, κD is null. The constant λB=e2kBTϵ0L is the non-dimensional Bjerrum length. The above non-dimensional PDE is completed with the following jump conditions on non-dimensional potential(14) [ψ]Γ=0,[ϵ∇ψ⋅n]Γ=0,∀x∈Γ, where the jump operator is defined for any quantity ζ defined in both domain as [ζ]Γ=ζ+−ζ−. All parameters and non-dimensional numbers, along with their values for this study, are summarized in Table 1 .Table 1 Problem parameters and non-dimensional numbers. Table 1Parameter Symbol Definition Value Characteristic length L - 1 Å Characteristic potential (thermal voltage) ψ0 kBTe 2.570 × 10−2 V Vacuum permittivity ϵ0 - 8.854 × 10−12 F m−2 Spike relative permittivity ϵ+ - 2 Solvent relative permittivity ϵ− - 7.830 × 10 Boltzmann constant kB - 1.381 × 10−23 J K−1 Avogadro number NA - 6.022 × 1023 M−1 Elementary charge e - 1.602 × 10−19 C Temperature T - 298.15 K Salt concentration cb - 1 × 10.0−3 Mm−3 Probe radius rp - 1.4Å Valence of the background electrolyte z - 1 Non-dimensional inverse of Debbye length κD 2cbNAe2zϵ0kBTL2 9.207 × 10−3 Non-dimensional Bjerrum length λB e2kBTϵ0L 7.039 × 103 3.4 Solution decomposition Following [27], we treat the singularities arising in the solution due to the singular charges inside the molecules by using the decomposition proposed by Chern et al. [8]. Doing so we split the non-dimensional potential ψ, into regular and singular parts: ψˆ and ψ¯, respectively(15) ψ=ψˆ+ψ¯. The singular part is itself split into two parts ψ⁎ and ψ0 (16) ψ¯(x)={ψ⁎(x)+ψ0(x)if x∈Ω+,0if x∈Ω− where ψ⁎ is the Coulombic potential due to singular charges,(17) ψ⁎(x)=λB4πϵ+∑i=1Nzi|x−xi|, and ψ0 satisfies the following Poisson's problem(18) ∇ψ0=0,∀x∈Ω+, (19) ψ0=−ψ⁎,∀x∈Γ. Utilizing the decomposition shown above, the regular (non-singular) part of the solution is given by solving(20) −∇⋅(ϵ∇ψˆ)+κ2(x)sinh⁡(ψˆ)=0,∀x∈Ω∖Γ, subject to the following jump conditions:(21) [ψˆ]Γ=0,∀x∈Γ (22) [ϵ∇ψˆ⋅n]Γ=−ϵ+∇(ψ⁎+ψ0)⋅n|Γ,∀x∈Γ Since ψ⁎ is known analytically, the gradient, ∇ψ⁎, appearing in the right hand side of (22) can be computed exactly(23) ∇ψ⁎=−λB4πϵ+∑i=1Nzix−xi|x−xi|3/2, while ∇ψ0 must be numerically approximated. 4 Numerical method The numerical approach for the resolution of the above Poisson-Boltzmann problem is presented in this section, along with novel practical a posteriori error estimates. We then verify the method for the entire trajectory by conducting a systematic convergence study for these estimators. 4.1 Implementation The numerical method, implemented on non-graded adaptive Octree grids, follows the general description given in [27], with each snapshot of the protein treated independently. The surface and grid generation is done using the level set framework developed by Min and Gibou [23]. In particular, as Fig. 3 illustrates, the mesh is systematically adapted to the SES location. All quantities are stored at the nodes of the mesh for improved accuracy and facilitated manipulation. The numerical solutions of the Poisson systems (9)-(10)-(11) and (18)-(19) (for the cavities detection and the construction of the regular part of the solution ψ0 respectively) are obtained using the second-order approach presented in [35], itself based on [24]. The solution, ψˆ, to the problem defined by Eqs. (20), (21) and (22) is constructed using a nodal version of the jump solver presented in [34]. The non-linearity of Eq. (20) is addressed using Newton's method [8], [26], [27], with a relative error tolerance of 10−6, chosen to be orders of magnitude smaller than the desired overall numerical error. The whole method is parallelized in a shared memory fashion using OpenMP [9], [29].Fig. 3 Visualizations of the entire computational domain, SARS-CoV-2 spike protein (frame 1), and finest mesh used for this study (level 11 with 34,362,796 nodes). Computational cells are colored by their corresponding tree level (i.e. the number of successive mesh subdivisions required for their construction). For visual purposes only half of the mesh is depicted. Fig. 3 The entire computational domain is defined as the cube of side length 400Å (about twice the size of the whole protein) and center xc=1N∑i=1Nxi. Calculations were performed on Octrees of maximum level ranging from 7 to 11. On the finest grids, the minimal spatial resolution is 0.18Å. 4.2 Error estimates To monitor the convergence of the overall method, we construct a posteriori numerical error estimators. They only rely on the decomposition presented in section 3.4, and can therefore be employed independently of the numerical approach. For the current implementation, detailed in section 4, we refer the interested reader to [26], [27] for formal convergence studies, using analytic solutions and order estimations. The error on the interface representation, gradient of the solution (i.e. electric field) and solution itself can be estimated using the following metrics eΓ,eE,eψ (24) eΓ=1|Γ||∫Γ∇ψ⁎⋅n−∑iNλzi4πϵ+|, (25) eE=1|Γ||∫Γ∇ψ0⋅n|, (26) eψ=1|Γ|∫Γ|ψ⁎+ψ0|, where |Γ| denotes the surface area of Γ. In virtue of Gauss's Theorem the first two integrals are null. Because of the boundary condition (19), the third quantity should also be null. When computing eΓ, because the gradient of the Coulombic and the total charge are known exactly, numerical errors can only arise from calculating the local normal n or approximating the surface integral. Therefore, this metric focuses on the geometry and its manipulation only. The numerical errors in the gradient of the component ψ0 are the primary source of errors in the second metric eE. Thus, it can be interpreted as a lower bound estimate for the average total normal electric field on the interface. The metric eψ, the average error in ψ0 on the interface, can similarly be used to estimate the numerical error in the total potential on the interface. 4.3 Convergence study Fig. 4a depicts the spike protein's electrostatic potential at the initial frame (closed configuration) as the mesh is refined. Positive electrostatic potential is colored in blue, neutral (no charge) in white, and the negative electrostatic potential is shown in red. The coarsest simulations (maxlevel=7,8) are only able to reproduce the general structure of the protein but fail at creating an accurate electrostatic map. As the maximal resolution reaches the characteristic atomic radius (rmin=1Å), the finest geometrical features are correctly reproduced, leading to appreciably more accurate results (maxlevel=9, minimal resolution 0.79Å). Further increasing the spatial resolution refines these molecular structures and the small scale potential variations even more (maxlevel=10,11).Fig. 4 (a) Evolution of the spike's Solvent Excluded Surface and electrostatic potential as the maximal resolution increases. (top down view in the closed configuration (frame 1)). (b − d) Convergence of the interface representation (eΓ), surface normal electric potential (eE), and surface potential (eψ). Fig. 4 The time evolution of our three error estimates for all examined resolutions is presented in Fig. 4. As expected, all three metrics converge with increasing resolution. The impact of using subatomic resolution (i.e. maxlevel≥9) is well illustrated with the convergence of the electric field and potential error estimates: it is unclear for super-atomic resolutions (maxlevel=7,8), and evident for subatomic ones (maxlevel=9,10,11). The error estimation for the total potential (eψ) is significantly larger than the variations between consecutive maximum levels observed in Fig. 4 a, which is a proxy for the error on the regular part of the solution ψˆ. Since eψ involves the singular part of the solution, which exhibits large spatial variations over small length scales, it is prone to higher numerical errors and expected to be larger than the actual error on the regular part of the solution. Closer inspection of our measurements reveals that these two metrics may differ by at least one order of magnitude. The error estimation for ψ, using maxlevel=10 is approximately equal to the maximum absolute surface potential value (≈10) observed on Fig. 4 a. This misleadingly suggests the relative error on the total solution is as large as 100%. The comparison between the potential maps for maxlevel=10 and maxlevel=11 indicates that in practice the relative error on the surface potential probably lies between 1% and 10%. From all these remarks, we are confident that the method is correctly implemented and that the most refined simulations accurately capture the S protein's electrostatic potential. For the finest resolution, the estimated average error on the total potential is close to 2, which in light of the above discussion, suggests that the average practical relative error on the surface potential is under 2%. 5 Results The S protein remains predominantly in the closed conformation to mask its receptor-binding domains (RBDs), thereby impeding their binding. To bind with ACE2, the S protein transforms into its open conformation, revealing its binding interface. In describing our results, we refer to the part of the spike containing the three RBDs as the top part and the predominantly negatively charged portion binding to the virus membrane as the bottom part. [43]. Our simulations (see Fig. 5 a) illustrate that the top part of the spike protein is predominantly positively charged, in accordance with the negative charge of the ACE2 binding site (see Fig. 6 ). Surprisingly, the top of the spike also reveals an underlying core with a surface area that has a dense negative charge. As the spike opens, this area could be exposed to the solvent, generating a negatively charged electrostatic cloud in the upper part of the protein, which may repulse the negatively charged ACE2 receptor. Therefore, we characterize the dynamics of the geometrical and electric properties of this negatively charged core χ (Fig. 5 b), the resulting repulsive cloud C (Fig. 5 c), the binding site B (Fig. 6 c), and the far electric field of the spike protein as they shift from the dynamics that occur during spike opening.Fig. 5 (a) Potential map of the SARS-Cov-2 spike open (frame 1), intermediate (frame 35), and close configuration (frame 71). (b) Evolution of the negatively charged core as the spike opens. (c) Repulsive electric cloud generated by the negatively charged core. (d − e) Characterization of the negatively charged core χ and repulsive electric cloud C: relative size (d) and potential (e) variations over the entire molecular trajectory. Fig. 5 Fig. 6 (a) Portion of the ACE2 receptor displaying surface charge (left). ACE2 and RBD of the spike protein connected (right). (b) Binding site of ACE2 highlighted in green (left). The binding site of ACE2 is highlighted within the ACE2 (orange) and RBD (yellow) connection (right). (c) SARS-Cov-2 spike in open conformation with RBD (yellow) and binding site, B (green) highlighted. (d − e) Characteristics of the binding site, B, of the spike protein. (d) Relative variance of the area and absolute curvature along with average potential. (e) average positive and negative potential. Fig. 6 5.1 Negatively charged core We define the negatively charged core χ, illustrated in Fig. 5 b, as the area of the SES in the top of the protein where the electric potential is more negative than a threshold value cχ=−5. Fig. 5 c and d depict the evolution of its surface and average charge. As the spike opens, the area shrinks by about 50%, while the variation in the average potential remains small (≈10%). In both cases, the most significant variations happen during the first ten iterations. In comparison, the molecular structure analysis (see Fig. 1) displays continuous variation over the entirety of the trajectory, indicating that the structural and electro-geometric transformations are non-trivially coupled. The diminution of both quantities could be explained by the fact that the spike opening removes a barrier, one of the monomers, between the core and solvent, causing the dilution of the surface charges in the solvent, and therefore a contraction of the core and its charge. 5.2 Repulsive electric cloud We define the negatively charged electric cloud C, generated by the core χ, as the region in the upper part of the solvent where the potential is below the threshold value cC=−2 (see Fig. 5 c). The measurements presented in Fig. 5 d and e indicate C expands by ≈42% as the spike opens, while its average potential decreases in magnitude by 30%. Again we interpret this phenomenon as an effective dissolution of the negative charges induced by removing the protecting monomer. For both geometries χ and C, we observe large variations in the first ten frames. The abrupt change (around frame 50) in the protein structure, discussed in section 2, is only perceptible in the size variations of the electric cloud C. 5.3 Binding site characterization The interface between the SARS-CoV-2 receptor-binding domain (RBD) and ACE2 are of particular interest as the binding of the two facilitates virus entry into cells [2], [18], [20], [33]. Lan et al. [18] undertook the task of determining the residues of the RBD and ACE2 interface which form a connection, finding the location of the binding site on the spike, B, to consist of 17 residues between Lys417-Tyr505 and the ACE2 site to be formed of 20 residues between Gln24-Tyr83 and Asn330-Arg393. For this analysis, we define the binding site, B, for the spike and the ACE2 receptor as the portion of the proteins SES generated by the residues sequences [417−505] and [24−42]∪[79−83]∪[330]∪[354−357]∪[393] respectively. The spike RBD is generated by the sequence [333−527]. When the spike is in the closed position B is located near its center, but as the spike shifts to its open conformation, B is moved outward and exposed to the solvent, as seen in Fig. 7 . As the spike opens, we monitor the evolution of the binding site area, average absolute curvature, and potential. The average is computed over the binding site surface. We interpret the average absolute curvatures as measures of the global convexity of the binding site. For the potential evolution, we distinguish between the average potential, average positive potential, and average negative potential.Fig. 7 Streamlines and direction of the electric field in the close (top) and open (bottom) configurations. As the spike opens, the streamlines emanating from the upper part of the molecule are observed to open as well. This effect is particularly noticeable for the lines emerging from the RBD domain (opening extremity on the right). Fig. 7 Fig. 6 provides a depiction of the resulting information. A relative decrease is observed for the area and average absolute curvature: as the spike opens, the binding site shrinks and flattens, in each case by 10−15%, suggesting minor conformational changes. The average negative potential also experiences a decrease, becoming less negative as the spike opens. Meanwhile, the average positive potential remains relatively constant, resulting in the total average potential of B following very closely with the changes in average ψ− and ultimately nearing 0 in the final frame. This increase in average potential is consistent with the knowledge that the ACE2 receptor is designed to bind to negative charges, and may be necessary in directing the ACE2 binding to the correct region. 5.4 Electric field structure The spike electric far-field streamlines are depicted in Fig. 7. In the closed configuration, most streamlines emanating from the bottom part of the spike are pointing away from the spike. Only a few of them, caused by the rare presence of positive charges, return rapidly to the protein. In the upper part, the streamlines are predominantly closed, suggesting that in this configuration, the spike protein may not be able to attract charges of any polarity at long range, and therefore has limited attracting potential. This may cause a lower affinity. As the spike opens, the streamlines in the bottom remain unchanged. However, above the top of the protein, the streamlines are now predominantly open. Fig. 8(a)-(b) depicts the entire electric field structure along with the region in the solvent where the electric field point away from the protein, or equivalently where negative charges would be drawn to the spike. In the closed configuration, this region is predominantly located above the spike, while in the open configuration, it is split into three sub-regions, each centered around one of the RBDs. In the latter, the sub-region centered around the revealed RBD represents 45% of the entire region and carries 34% of its total electric potential energy.Fig. 8 Streamlines of the electric field (E=−∇ψˆ) in the close (a) and open (b) configurations over the entire computational domain. The portion of the streamlines where the electric field is pointing away from the protein (i.e. ∇ψ ⋅ x = −E ⋅ x < 0), and therefore negative charges would be drawn to it, is depicted in blue. (c − e) Electric potential multipole structure: charge density decomposition over its first four moments C,D,Q and O. (c) and (e) represent the principal directions of the dipole D (in black) and the quadrupole Q, in the initial and final configurations, respectively. For the quadrupole, the purple directions (eigenvectors) are associated with negative eigenvalues. The gray one has a positive eigenvalue. All segment lengths are proportional to the strength of the corresponding pole (i.e. the norm of the dipole or of the corresponding eigenvalue). The protein binding site is depicted as opaqued. (d) Depicts the evolution of the norm of all four first moments. We use the standard L2 entry-wise norm. Fig. 8 To characterize the restructuring of the electric field of the S protein, we define the first four moments of the non-dimensional charge distribution q(x)=sinh(−ψ(x)) in the solvent(27) Total ChargeC=∫Ω−q(x)d(3)x, (28) DipoleDi=∫Ω−q(x)xid(3)x, (29) QuadrupoleQij=∫Ω−q(x)(3xixj−x⋅xδij)d(3)x, (30) OctupoleOijk=∫Ω−q(x)(15xixjxk−x⋅x(δijxk+δikxj+δjkxi))d(3)x. Note that the last two moments are defined in their symmetric traceless form, as they would be for a standard electrostatic multipole expansion. Because here our continuum model is more complex, it should be reminded that these four moments are not guaranteed to be the ones appearing in the multipole expansion. Nonetheless, they remain a pertinent tool to characterize the structure of the electrostatic solution. Their evolution throughout the opening, illustrated in Fig. 8 c, show that the strength of the dipole (D) is decreasing, while all other moments norms are increasing. The two most informative moments for the structure of the electric field, the dipole and the quadrupole, are diminished by 9% and increased by 67% respectively. The principal direction of Q associated to the positive eigenvector appears to remain quasi-parallel to the dipole direction. In fact, all four directions (the dipole direction and the three principal directions of Q) appear to undergo the same rotation. The opening also reinforces the anisotropy in the tensor Q, by amplifying the difference between the magnitude of its eigenvalues, compressing one of its directions and effectively turning it into a two-dimensional quadrupole. 6 Conclusions We examined the SARS-CoV-2 spike protein's transition from its open to closed conformation, observing the protein's electrostatic potential dynamics with a particular focus on its receptor-binding domains. In the Folding@home trajectory we analyzed, one of the three monomers detaches from the core of the complex and becomes visible to the surrounding environment, consistent with recent cryo-EM derived structures [5]. Our results show that despite the dramatic molecular displacement engendered by this strategic repositioning, the geometric and electric properties of the RBD itself remain largely unaltered. Instead, as we describe below, changes emanating from the exposure of the core of the S protein cause a change in the electric fields surrounding the RBD. Our continuum analysis, both of the surface potential and the volumetric potential in the vicinity of the binding region, points to the existence of an inner negatively charged core on the surface of the spike, which is revealed to the solvent as the spike opens. This negatively charged surface shrinks as the spike opens, inducing a negatively charged region between the exposed RBD and the two hidden ones. The emergence of this electric cloud is a priori puzzling: the binding site of the ACE2 receptor being almost entirely negatively charged, we would expect this cloud to repulse the receptor. However, this cloud does not envelop the spike's binding site, B, which becomes more positively charged. This is in line with what we expect is a strong binder of the ACE2 receptor and indicates that targeting the open state of the S protein may be a more viable drug design strategy than the closed configuration. Indeed, recent cryoEM studies showed that ≈16% of a recombinantly expressed S protein population is in an open state, with a single monomer “erected” from the core [5]. The electric field, which we observe uncoiling above the protein, exhibits a dramatic rearrangement. This is correlated to the emergence of the negatively charged cloud and manifests in the multipole moments decomposition. In particular, we observe the quadrupole moment growing in magnitude, rotating, and compressing one of its principal directions. This transformation can be interpreted as a strategic transition from an undirected configuration where the entire top part of the spike attracts negatively charged structures, such as the ACE2 binding site, to one where the attraction is directed toward a specific RBD. The progression of time will uncover a multitude of breakthroughs regarding the behavior of the SARS-CoV-2 S protein. In the short time since the conception of our investigation, numerous discoveries by other researchers have been made public already. For example, the trajectory presented in [43] contains glycans, chemical compounds known to coat the exterior of many viruses, which may have an impact on the results presented here. Recall that the trajectory we explored here is simply one possible path the spike may follow. So there is a possibility that a different trajectory would be a more accurate representation of the true function of the spike protein. Recent studies displayed cryo_EM structures for the spike, presenting an open configuration that differs from the open configuration represented here [40], [36]. A similar analysis to the one presented here may be required on different structures, as more probable trajectories are predicted and novel molecular structures uncovered, to capture a more relevant image of the electric potential on and around the spike. Utilizing the same scientific pipeline, we are confident high-fidelity quantitative insights — into the electric potential of molecules unrestricted to the one investigated here — could be obtained efficiently, supporting the global scientific effort. At the computational level, we have developed a novel framework, combining Markov states models simulations with a continuum physics numerical solver to produce high-resolution multiscale dynamical potential maps of deforming proteins and their surrounding environment. Our framework includes practical mathematical tools to quantify the numerical error and characterize protein electro-geometric properties. The molecular trajectories alone are insufficient for understanding protein electrostatic interactions and, thus, protein core bio-mechanisms. On the other hand, it is unrealistic to envision an approach where trajectories of such a large protein could be obtained solely through continuum modeling. We believe hybrid approaches, such as the one presented here, can provide the scientific community with invaluable information, which we hope can be used to elucidate how changes to physical properties such as electrostatics translate into function. CRediT authorship contribution statement M. Theillard and S. Sukenik conceived the presented project. M. Theillard developed and implemented the numerical method. S. Strango, A. Kucherova, and M. Theillard carried the computations and analyzed the Results. All authors contributed to the redaction of the manuscript and approved its final form. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements The authors would like to thank M. Zimmerman, G. Bowman, and the Folding@Home project for creating and providing us with the S protein opening trajectory. We also thank D. Strubbe and J. Grasis for valuable discussions. This research was supported by a COVID-19 seed grant from the 10.13039/100008652 Center for Information Technology Research in the Interest of Society (CITRIS) at UC Merced awarded to M. Theillard and S. Sukenik. The authors acknowledge computing time on the Multi-Environment Computer for Exploration and Discovery (MERCED) cluster at the University of California, Merced, which was funded by 10.13039/100000001 National Science Foundation Grant No. ACI-1429783. M. Theillard and S. Sukenik are members of the NSF-CREST Center for Cellular and Biomolecular Machines at the University of California, Merced (NSF-HRD-1547848). ==== Refs References 1 FoldingImage 1home http://foldingathome.org 2 Amin M. Sorour M.K. Kasry A. Comparing the binding interactions in the receptor binding domains of sars-cov-2 and sars-cov J. Phys. Chem. Lett. 11 12 2020 4897 4900 PMID: 32478523 32478523 3 Baker N.A. Improving implicit solvent simulations: a Poisson-centric view Curr. Opin. Struct. Biol. 15 2 April 2005 137 143 15837170 4 Baker N.A. Bashford D. Case D.A. Implicit solvent electrostatics in biomolecular simulation New Algorithms for Macromolecular Simulation 2006 263 295 5 Benton D. Wrobel A. Xu P. Roustan C. Martin S. Rosenthal P. Skehel J. Gamblin S. Receptor binding and priming of the spike protein of SARS-CoV-2 for membrane fusion Nature September 2020 6 Boschitsch A.H. Fenley M.O. A fast and robust Poisson–Boltzmann solver based on adaptive Cartesian grids J. Chem. Theory Comput. 7 5 2011 1524 1540 21984876 7 Chen J. Brooks C.L. Khandogin J. Recent advances in implicit solvent-based methods for biomolecular simulations Curr. Opin. Struct. Biol. 18 2 April 2008 140 148 18304802 8 Chern I.-L. Liu J.-G. Wang W.C. Accurate evaluation of electrostatics for macromolecules in solution Methods Appl. Anal. 10 2003 309 328 9 Dagum L. Menon Openmp R. An industry-standard api for shared-memory programming IEEE Comput. Sci. Eng. 5 1 January 1998 46 55 10 Decherchi S. Masetti M. Vyalov I. Rocchia W. Implicit solvent methods for free energy estimation Eur. J. Med. Chem. 91 Feb 2015 27 42 PMID: 25193298 25193298 11 Egan R. Gibou F. Fast and scalable algorithms for constructing solvent-excluded surfaces of large biomolecules J. Comput. Phys. 374 2018 91 120 12 Felberg L.E. Brookes D.H. Yap E.-H. Jurrus E. Baker N.A. Pb-am T. Head-Gordon An open-source, fully analytical linear Poisson-Boltzmann solver J. Comput. Chem. 38 15 June 2017 1275 1282 27804145 13 Geng W. Yu S. Wei G.W. Treatment of charge singularities in implicit solvent models J. Chem. Phys. 127 11 2007 114106 14 Harris R.C. Mackoy T. Fenley M.O. Problems of robustness in Poisson-Boltzmann binding free energies J. Chem. Theory Comput. 11 2 Feb 2015 705 712 PMID: 26528091 26528091 15 Hassanzadeh K. Perez Pena H. Dragotto J. Buccarello L. Iorio F. Pieraccini S. Sancini G. Feligioni M. Considerations around the sars-cov-2 spike protein with particular attention to covid-19 brain infection and neurological symptoms ACS Chem. Neurosci. 11 15 Aug 2020 2361 2369 PMID: 32627524 32627524 16 Jurrus E. Engel D. Star K. Monson K. Brandi J. Felberg L.E. Brookes D.H. Wilson L. Chen J. Liles K. Chun M. Li P. Gohara D.W. Dolinsky T. Konecny R. Koes D.R. Nielsen J.E. Head-Gordon T. Geng W. Krasny R. Wei G. Holst M.J. McCammon J.A. Baker N.A. Improvements to the apbs biomolecular solvation software suite Protein Sci. 27 1 2018 112 128 28836357 17 Koehl P. Electrostatics calculations: latest methodological advances Curr. Opin. Struct. Biol. 2006 142 151 16540310 18 Lan J. Ge J. Yu J. Shan S. Zhou H. Fan S. Zhang Q. Shi X. Wang Q. Zhang L. Wang X. Structure of the sars-cov-2 spike receptor-binding domain bound to the ace2 receptor Nature 581 7807 May 2020 215 220 32225176 19 Lee B. Richards F.M. The interpretation of protein structures: estimation of static accessibility J. Mol. Biol. 55 3 1971 379 400 5551392 20 Letko M. Marzi A. Munster V. Functional assessment of cell entry and receptor usage for sars-cov-2 and other lineage b betacoronaviruses Nat. Microbiol. 5 4 2020 562 569 32094589 21 Li C. Li L. Petukh M. Alexov E. Progress in developing Poisson-Boltzmann equation solvers Molecular Based Mathematical Biology Comput. Math. Biophys. 1 Mar 2013 10.2478/mlbmb-2013-0002 PMID: 24199185 22 Lu B.Z. Zhou Y.C. Holst M.J. McCammon J.A. Recent progress in numerical methods for the Poisson Boltzmann equation in biophysical applications Commun. Comput. Phys. 3 5 2008 973 1009 23 Min C. Gibou F. A second order accurate level set method on non-graded adaptive Cartesian grids J. Comput. Phys. 225 2007 300 321 24 Min C. Gibou F. Ceniceros H. A supra-convergent finite difference scheme for the variable coefficient Poisson equation on non-graded grids J. Comput. Phys. 218 2006 123 140 25 Mirzadeh M. Gibou F. Squires T.M. Enhanced charging kinetics of porous electrodes: surface conduction as a short-circuit mechanism Phys. Rev. Lett. 113 Aug 2014 097701 26 Mirzadeh M. Theillard M. Gibou F. A second-order discretization of the nonlinear Poisson-Boltzmann equation over irregular geometries using non-graded adaptive Cartesian grids J. Comput. Phys. 230 2010 2125 2140 27 Mirzadeh M. Theillard M. Helgadöttir A. Boy D. Gibou F. An adaptive, finite difference solver for the nonlinear Poisson-Boltzmann equation with applications to biomolecular computations Commun. Comput. Phys. 13 2013 150 173 28 Nguyen D.D. Wang B. Wei G.-W. Accurate, robust, and reliable calculations of Poisson-Boltzmann binding energies J. Comput. Chem. 38 13 May 2017 941 948 PMID: 28211071 28211071 29 OpenMP Architecture Review Board OpenMP Application Program Interface version 5.0 May 2018 30 Osher S. Sethian J. Fronts propagating with curvature-dependent speed: algorithms based on Hamilton-Jacobi formulations J. Comput. Phys. 79 1988 12 49 31 Pan Q. Tai X.-C. Model the solvent-excluded surface of 3D protein molecular structures using geometric pde-based level-set method Commun. Comput. Phys. 6 2009 777 792 32 Rocchia W. Alexov E. Honig B. Extending the applicability of the nonlinear Poisson-Boltzmann equation: multiple dielectric constants and multivalent ions J. Phys. Chem. B 105 28 Jul 2001 6507 6514 33 Tai W. He L. Zhang X. Pu J. Voronin D. Jiang S. Zhou Y. Du L. Characterization of the receptor-binding domain (rbd) of 2019 novel coronavirus: implication for development of rbd protein as a viral attachment inhibitor and vaccine Cell. Mol. Immunol. 17 6 Jun 2020 613 620 32203189 34 Theillard M. Gibou F. Saintillan D. Sharp numerical simulation of incompressible two-phase flows J. Comput. Phys. 391 2019 91 118 35 Theillard M. Rycroft C. Gibou F. A multigrid method on non-graded adaptive Octree and Quadtree Cartesian grids J. Sci. Comput. 55 2013 1 15 36 Walls A.C. Park Y.-J. Tortorici M.A. Wall A. McGuire A.T. Veesler D. Structure, function, and antigenicity of the SARS-CoV-2 spike glycoprotein Cell 181 2 April 2020 281 292 32155444 37 Wang Y. Liu M. Gao J. Enhanced receptor binding of sars-cov-2 through networks of hydrogen-bonding and hydrophobic interactions Proc. Natl. Acad. Sci. 117 25 2020 13967 13974 32503918 38 Warwicker J. Watson H.C. Calculation of the electric potential in the active site cleft due to alpha-helix dipoles J. Mol. Biol. 157 4 1982 671 679 6288964 39 Wrapp D. Wang N. Corbett K.S. Goldsmith J.A. Hsieh C.-L. Abiona O. Graham B.S. McLellan J.S. Cryo-em structure of the 2019-ncov spike in the prefusion conformation Science 367 6483 2020 1260 1263 32075877 40 Wrobel A.G. Benton D.J. Xu P. Roustan C. Martin S.R. Rosenthal P.B. Skehel J.J. Gamblin S.J. Sars-cov-2 and bat ratg13 spike glycoprotein structures inform on virus evolution and furin-cleavage effects Nat. Struct. Mol. Biol. 27 8 Aug 2020 763 767 32647346 41 Zheng Q. Wei. Poisson-boltzmann-nernst-planck model G.-W. J. Chem. Phys. 134 19 May 2011 194101 PMID: 21599038 42 Zimmerman M.I. Bowman G.R. Fast conformational searches by balancing exploration/exploitation trade-offs J. Chem. Theory Comput. 11 12 2015 5747 5757 26588361 43 Zimmerman M.I. Porter J.R. Ward M.D. Singh S. Vithani N. Meller A. Mallimadugula U.L. Kuhn C.E. Borowsky J.H. Wiewiora R.P. Hurley M.F.D. Harbison A.M. Fogarty C.A. Coffland J.E. Fadda E. Voelz V.A. Chodera J.D. Bowman G.R. Citizen scientists create an exascale computer to combat covid-19 2020 bioRxiv

0

PMC9749448

NO-CC CODE

2022-12-15 23:23:20

no

J Comput Phys. 2021 Nov 1; 444:110591

utf-8

J Comput Phys

10.1016/j.jcp.2021.110591

oa_other

==== Front Environ Res Environ Res Environmental Research 0013-9351 1096-0953 Elsevier Inc. S0013-9351(20)31507-3 10.1016/j.envres.2020.110610 110610 Article Re: An ecological analysis of long-term exposure to PM2.5 and incidence of COVID-19 in Canadian health regions Goldberg Mark S. 1∗ Department of Medicine, McGill University, McGill University Health Centre-Research Institute, Centre for Outcomes Research and Evaluation, Research Institute, Montreal General Hospital, R2-105, 1650 Cedar Ave, Montreal, QC, H3G 1A4, Canada Villeneuve Paul J. School of Mathematics and Statistics and Department of Neurosciences, Faculty of Science Carleton University, Herzberg Building, Room 54131125 Colonel By Drive, Ottawa, ON, K1S 5B6, Canada ∗ Corresponding author. 1 Web pagehttp://www.med.mcgill.ca/epidemiology/goldberg/Environmental epidemiology grouphttp://www.mcgill.ca/environ-epi. 15 12 2020 3 2021 15 12 2020 194 110610110610 11 11 2020 7 12 2020 © 2020 Elsevier Inc. All rights reserved. 2020 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcDear Editor, We read with interest this paper on air pollution in Canada (Stieb et al., 2020) and its possible contribution to an increased incidence of COVID-19. In this ecological study, the authors used as the unit of observation very broad geographical regions defined administratively (referred to as “health regions”), which vary dramatically in size and constitution within and between Canadian provinces. They then made use of data on incident cases of COVID-19 in these 111 regions (73,390 cases until the end of May 2020). To these grouped counts, they juxtaposed, in a negative binomial statistical model, past satellite observations of fine particulate matter (PM2.5) over a 17-year period (2000–2016) that had a spatial resolution of 1 × 1 km. These were then averaged to produce a summary exposure measure for each region. A number of ecological covariates were included in the model. They found that a 1 μg/m3 increase in PM2.5 increased the incidence of COVID-19 by 7% that they indicated “was several fold larger per unit PM2.5 than hazard ratios typically observed in cohort studies of mortality”. The authors indicated that due to the study design “the findings should be interpreted with caution”, and they also discussed some limitations due to the “coarseness” of these regions as well as cross-level bias (“ecological fallacy”). Nevertheless, it is our view that the limitations of their data are so severe that they do not advance public health as their risk estimates are neither credible nor interpretable. The basis for this conclusion is our recent detailed methodological review of mortality studies of both SARS and COVID-19 that we published in October 2020 in Environmental Health Perspectives, where we concluded that all studies of associations between these infectious diseases and environmental factors are seriously biased (Villeneuve and Goldberg, 2020). Our concerns expressed in our review paper (Villeneuve and Goldberg, 2020) about the validity of the risk estimates apply to this study as well. To briefly summarize some of our arguments, studies that make use of the ecological design suffer from cross-level bias that arise by using groups rather than individuals as the unit of analysis. Heterogeneity of populations within these large areas is a key issue. Differences between and within jurisdictions as to the timing on the pandemic curve, behaviour of populations, the measures taken to reduce infections, and the R0 value will affect incidence rates. Although the authors attempted to account for some of these essential components of the pandemic, by including a term for days since peak daily incidence of new cases, deaths at the health region level, and date of declaration of emergency, these factors cannot possibly account for infection dynamics within and between these vast regions and through time. Not being able to account for clustering of disease, spatiotemporal variations in the strains of COVID-19 that may affect sequelae differently, and, of course, heterogeneity in spatial-temporal assignment of air pollution within regions, that correlates with socio-economic status, leads to bias. Use of very large areas in a grouped analysis is a major cause of concern. Because cases were aggregated and assigned to each region, it does not matter where the person lived in the region, where they may have travelled, where they may have worked, what social activities they engaged in, and what levels of air pollution they may have been exposed to. This means that any differences within the region are ignored by design. To be concrete, consider a specific example of population dynamics from Montreal, in which there are a number of “hot spots” that have varied since the beginning of the pandemic. In the spring of 2020, a religious group living in an affluent area of the city, with low levels of air pollution, had the highest counts in the city because many members were in the United States where they became infected. The virus rampaged through the community until it was controlled. Summer came, cases went down, restrictions were reduced, and the members started congregating again. Now in October incidence rates are skyrocketing again. All in a low air pollution part of the city. A study based on individuals would not ignore these or other types of circumstances, but an ecological design where grouped data across these large regions is contrasted is subject to serious error given the vast differences in case identification, screening and implementation of public health practices its correlation with air pollution, and other factors. Other deficiencies in the design, over and above the ecological fallacy, include serious misclassification and under-reporting of the incidence of COVID-19. In the United States, the CDC and others have recently estimated that deaths in the United States are underestimated by about 20% (Rossen et al., 2020; Woolf et al., 2020). There is clear evidence of undercounting of cases throughout the world and in Canada, and population-based testing and contact tracing is deficient. In a recent paper (Russell et al., 2020), about 20% of cases have not been identified in Canada. The Institute for Health Metrics and Evaluation modelled the under ascertainment of incident COVID-19 cases in Canada and estimated that during the first peak (march–June 2020) less than one in 5 cases of COVID-19 in the population were identified (Institute for Health Metrics and Evaluation, 2020). Similar estimates of under-ascertainment were produced by a research group at Imperial College (Imperial College, 2020). Canadian data suggests that per capita, a greater number of cases are being identified in urban areas which have higher concentrations of air pollution (Government of Alberta, 2020; BC Centre for Disease Control, 2020). The increased numbers of cases in urban areas may be driven by a number of factors other than air pollution, including increased person to person interaction, greater capacity for screening, higher probabilities, and hence greater awareness of being infected, or other potential sources of exposure such as travel, or attending large gatherings. The pandemic is about between-person spread, which implies clustering in different settings. In Canada, many clusters of COVID-19 have been driven by its spread in vulnerable communities, (e.g., long term care homes, occupational settings (e.g., meat packing plants), and large gatherings (e.g., funerals, sports tournaments, and weddings), and now in schools. And the sources of infection have varied across Canada; for example, in the Maritime region, virtually all recent cases have been attributable to travel. In Quebec in the spring, the beginning of the pandemic coincided with the spring break, with families and students bringing the virus back from southern climates. In contrast, the vast majority of cases in Ontario have been due to spread from close contact (Public Health Ontario, 2020). These patterns of COVID-19 incidence raise two critical points. First, attempts to characterize associations between ambient air pollution and COVID-19 should at a minimum be able to account for sources of infection. Second, occurrences of COVID-19 are not independent, and they are more likely to cluster in areas of high population density, which importantly, tend to be areas with higher concentrations of air pollution. Therefore, statistical methods need to account for clustering of cases. The data in the study by Stieb et al. could not be used to take into account the clustered nature of the data, nor did it allow for sources of infection. Our critical review focussed on mortality from COVID-19 in which there is plausible mechanism by which underlying conditions may affect the clinical course after infection. It is clear that air pollution does not cause COVID-19 and whether it should be considered as a modifier is possible, but these and other data do not shed light on this because of the inherent biases. In terms of public health, context is essential. Studies of air pollution and COVID-19 may detract from the needed implementation of public health measures needed to control the spread of COVID-19. Stay in place orders, the use of face masks, hand washing, and physical distancing are the most important policies in reducing the spread in COVID-19. The possible contributions of air pollution on increasing incidence of COVID-19, put in this context, is at best a drop in a very large bucket. South Korea, with ambient levels of air pollution far greater than in Canada, or the United States, has been able to minimize the impact of COVID-19 relative to other countries. Is the success of mitigating COVID-19 in New Zealand due to its lower levels of air pollution? Public health experts would unanimously say otherwise (Baker et al., 2020). For these reasons, and particularly when jurisdictions are lacking the political courage or ability to implement these basic measures, it is critically important that authors of these types of papers make some effort to put their findings into context. Other authors have also expressed concern over these studies Heederik et al. (2020). As we stated in our review: “In fact, we feel that the public is not served well by these studies, many of which have undergone the scrutiny of peer review, especially because the press are on the lookout for sensational stories. All observational studies are not created equal, and the rush to use a flawed design to investigate the association between air pollution and mortality from COVID-19 jeopardizes the clear and compelling evidence of chronic exposure to air pollution as a threat to human health and deflects from the increased rates of infection and health consequences caused by problems of social and income disparities, overcrowding, and other societal issues.” About the authors As epidemiologists working for over 20 years in air pollution, we have played a role on the narrative about the health impacts from exposure to air pollution. We have analyzed data from the Harvard Six Cities and American Cancer Society's Cancer Prevention cohorts. These cohorts were among the first to establish links between long-term exposure to air pollution and death from cardiovascular and respiratory disease. In Canada, we have participated in multiple national and local air pollution studies, including those that have used national census data. Our experiences provide the foundation for us to critically review the research, and outline what we consider to be the key weaknesses of this study. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ==== Refs References Baker M.G. Wilson N. Anglemyer A. Successful elimination of covid-19 transmission in New Zealand N. Engl. J. Med. 383 2020 e56 32767891 BC Centre for Disease Control COVID-19 Data 2020 http://www.bccdc.ca/health-info/diseases-conditions/covid-19/data Government of Alberta COVID-19 Alberta Statistics 2020 https://www.alberta.ca/stats/covid-19-alberta-statistics.htm Heederik D.J.J. Smit L.A.M. Vermeulen R.C.H. Go slow to go fast: a plea for sustained scientific rigour in air pollution research during the covid-19 pandemic Eur. Respir. J. 56 2020 Imperial College Age-structured SEIR Model Focused on Low- and Middle-Income Countries 2020 https://mrc-ide.github.io/global-lmic-reports/ Institute for Health Metrics and Evaluation Hybrid SEIR Model 2020 https://covid19.healthdata.org/2020 Public Health Ontario Daily Epidemiologic Summary. COVID-19 in Ontario: January 15, 2020 to November 9 2020 https://files.ontario.ca/moh-covid-19-report-en-2020-11-10.pdf Rossen L.M. Branum A.M. Ahmad F.B. Sutton P. Anderson R.N. Excess deaths associated with COVID-19, by age and race and ethnicity — United States, january 26–october 3, 2020 MMWR Morb. Mortal. Wkly. Rep. 69 2020 1522 1527 10.15585/mmwr.mm6942e2 33090978 Russell T.W. Golding N. Hellewell J. Reconstructing the early global dynamics of under-ascertained COVID-19 cases and infections BMC Med. 18 2020 332 10.1186/s12916-020-01790-9 33087179 Stieb D.M. Evans G.J. To T.M. Brook J.R. Burnett R.T. An ecological analysis of long-term exposure to PM2.5 and incidence of COVID-19 in Canadian health regions Environ. Res. 191 2020 110052 10.1016/j.envres.2020.110052 32860780 Villeneuve P.J. Goldberg M.S. Methodological considerations for epidemiological studies of air pollution and the SARS and COVID-19 coronavirus outbreaks Environ. Health Perspect. 128 2020 95001 32902328 Woolf S.H. Chapman D.A. Sabo R.T. Weinberger D.M. Hill L. Taylor D.D.H. Excess deaths from COVID-19 and other causes, march-july 2020 J. Am. Med. Assoc. 324 2020 Number 15

33333035

PMC9749452

NO-CC CODE

2022-12-15 23:23:20

no

Environ Res. 2021 Mar 15; 194:110610

utf-8

Environ Res

10.1016/j.envres.2020.110610

oa_other

==== Front Eur J Pharmacol Eur J Pharmacol European Journal of Pharmacology 0014-2999 1879-0712 Elsevier B.V. S0014-2999(21)00826-8 10.1016/j.ejphar.2021.174670 174670 Article In vitro ion channel profile and ex vivo cardiac electrophysiology properties of the R(-) and S(+) enantiomers of hydroxychloroquine Ballet Véronique a Bohme G. Andrees b∗ Brohan Eric c Boukaiba Rachid b Chambard Jean-Marie b Angouillant-Boniface Odile c Carriot Thierry a Chantoiseau Céline b Fouconnier Sophie b Houtmann Sylvie b Prévost Céline c Schombert Brigitte b Schio Laurent d Partiseti Michel b a Preclinical Safety Investigative Toxicology, Sanofi-Aventis R&D, Chilly-Mazarin, France b High Content Biology, Integrated Drug Discovery, Sanofi-Aventis R&D, Vitry-sur-Seine, France c Early Development, Advanced Preparative Chromatography, Sanofi-Aventis R&D, Vitry-sur-Seine, France d Integrated Drug Discovery, Sanofi-Aventis R&D, Vitry-sur-Seine, France ∗ Corresponding author. Sanofi-Aventis R&D, 13 quai Jules Guesde, F-94403, Vitry-sur-Seine, France. 2 12 2021 15 1 2022 2 12 2021 915 174670174670 7 9 2021 1 12 2021 1 12 2021 © 2021 Elsevier B.V. All rights reserved. 2021 Elsevier B.V. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Hydroxychloroquine (HCQ) is a derivative of the antimalaria drug chloroquine primarily prescribed for autoimmune diseases. Recent attempts to repurpose HCQ in the treatment of corona virus disease 2019 has raised concerns because of its propensity to prolong the QT-segment on the electrocardiogram, an effect associated with increased pro-arrhythmic risk. Since chirality can affect drug pharmacological properties, we have evaluated the functional effects of the R(-) and S(+) enantiomers of HCQ on six ion channels contributing to the cardiac action potential and on electrophysiological parameters of isolated Purkinje fibers. We found that R(-)HCQ and S(+)HCQ block human Kir2.1 and hERG potassium channels in the 1 μM–100 μM range with a 2–4 fold enantiomeric separation. NaV1.5 sodium currents and CaV1.2 calcium currents, as well as KV4.3 and KV7.1 potassium currents remained unaffected at up to 90 μM. In rabbit Purkinje fibers, R(-)HCQ prominently depolarized the membrane resting potential, inducing autogenic activity at 10 μM and 30 μM, while S(+)HCQ primarily increased the action potential duration, inducing occasional early afterdepolarization at these concentrations. These data suggest that both enantiomers of HCQ can alter cardiac tissue electrophysiology at concentrations above their plasmatic levels at therapeutic doses, and that chirality does not substantially influence their arrhythmogenic potential in vitro. Keywords CiPA initiative Torsade de pointes Patch-clamp Purkinje fibers Covid-19 Chiral switch ==== Body pmc1 Introduction Hydroxychloroquine (HCQ, Plaquenil®) is a well-known antimalarial drug currently used in immuno-inflammatory conditions such as cutaneous and systemic lupus erythematosus or rheumatoid arthritis (Schrezenmeier and Dorner, 2020). Recent attempts to repurpose HCQ for the treatment of corona virus disease 2019 (Covid-19) have raised concerns because this drug can prolong the QT-segment on the electrocardiogram (Mazzanti et al., 2020; Oscanoa et al., 2020). Drug-induced QT interval prolongation is associated with an increased risk of a distinctive type of ventricular arrythmias known as Torsade de Pointes (TdP) which can lead to cardiac arrest (Roden, 2004). The mechanism underlying TdP results from a delay of the cardiac action potential repolarization towards the end of its plateau phase, which prematurely reactivates the slowly inactivating L-type CaV1.2 calcium channels. Action potential repolarization is driven by outward potassium currents flowing through channels encoded by the human ether-a-go-go related gene (i.e. hERG channels). Inhibition of hERG function was the first molecular mechanism associated with TdP Rampe and Brown (2013); (Sanguinetti et al., 1995). However, drug effects at the voltage gated KV7.1 or the inwardly rectifying Kir2.1 channels can also influence repolarization and induce arrhythmic events. Moreover, blockade of inward Na+ and Ca2+ currents could in principle counterbalance the QT elongating effects of outward K+ currents inhibition, a hypothesis that underlies the Comprehensive in vitro Pro-Arrhythmia (CiPA) initiative aimed at refining the electrocardiographic risk assessment of new drugs (Gintant et al., 2016). HCQ is a chiral 4-aminoquinoline which has been developed and marketed as a racemate. Yet, it has been documented for several racemic drugs that the pharmacology of each enantiomer can give rise to a eutomer and a distomer where the desired properties can be ascribed primarily to the Rectus- or the Sinister-enantiomer (Tucker, 2000). Chirality can also influence pharmacodynamic behavior, a feature that has been demonstrated for R(-)HCQ and S(+)HCQ in the mid-90's (Ducharme et al., 1994; Tett et al., 1994). Several methods have been described to obtain the enantiomers of HCQ separately, including direct synthesis using optically resolved reactants (Blaney et al., 1994) and chromatographic separation on a preparative scale (Wilson et al., 2020). Here we have evaluated the effects of R(-)HCQ and S(+)HCQ in in vitro electrophysiological models pertaining to cardiovascular safety pharmacology. First, we evaluated the effects of each enantiomer on six cardiac ion channels using automated patch-clamp and cloned human channels expressed in mammalian cells. Secondly, we tested the effects of R(-)HCQ and S(+)HCQ on transmembrane currents recorded from ex vivo isolated rabbit Purkinje fibers, a native tissue that recapitulates the essential features of human ventricular conductive tissue (Baczko et al., 2016). We found that the enantiomers of HCQ selectively inhibit Kir2.1 channels and, less potently, hERG channels, and significantly alter several cardinal AP metrics at concentrations about one order of magnitude above the free plasma concentrations circulating at therapeutically effective doses. 2 Material and methods 2.1 Chiral preparative HPLC Separation of the enantiomers of HCQ was performed using a carbamate-derivatized amylose chiral stationary phase (Chiralpak® AD, Daicel chemical industries). About 1 g hydroxychloroquine sulfate was dissolved in 200 mL of 90:10 acetonitrile/methanol containing 5% triethylamine and loaded in two halves on a 76.5 mm × 350 mm column packed with 1.2 kg chiral phase at 20 μm particle size. Elution was performed with the same solvent mix containing 0.1% triethylamine at a flowrate of 400 mL/min. Efficiency was improved by recycling the feed after the desorption of the first enantiomer. This second passage on the same column allowed complete resolution of both enantiomers. The fractions monitored at 265 nm UV were collected and evaporated to obtain free-base oils. Absolute configurations were assigned as S(+)HCQ (optical rotation OR = +76.4 [c] = 320 g/mL in DMSO at 589 nm; enantiomeric excess ee = 98.5%) and R(-)HCQ (OR = −68.1 same conditions; ee = 99%) based on published data (Ibrahim and Fell, 1990). 2.2 Patch-clamp on human cardiac ion channels Studies were performed as described previously (Le Marois et al., 2020). Briefly, CHO cells expressing hERG or KV7.1/minK were obtained from B'Sys (Switzerland), and CaV1.2/β2/α2δ1, Kir2.1 or KV4.3/KChIP2.2 from Charles River (USA), respectively. HEK cells expressing NaV1.5 were constructed in house. All expressions were constitutive except Kir2.1, CaV1.2 and NaV1.5 which were induced overnight by exposure to 1 μg/mL doxycycline (in the presence of 3 μM verapamil for CaV1.2). Whole-cell recordings were performed at room temperature on a Qpatch® 48X planar patch-clamp workstation (Sophion, Denmark). The pulse-protocols used and cycle-times applied to elicit each current are schematized as insets in their respective figures. Leak currents were handled either on-line by P/n subtraction or offline by subtracting currents remaining in the presence of a specific reference inhibitor added at the end of each recording session at a maximally active (i.e. “full-block”) concentration. Extracellular buffers for potassium channels contained (in mM) NaCl, 150; KCl, 4; CaCl2, 2; MgCl2, 1 and HEPES, 10. For the hERG channels, glucose (10) was added. The pH was adjusted to 7.4 with NaOH. The intracellular buffers were composed of (in mM): KF, 120; KCl, 20; EGTA, 10; MgCl2, 1; HEPES, 10. For KV7.1 channels, 10 mM EDTA was added as a second chelator, EGTA was decreased to 5 mM and no MgCl2 was added. The pH was adjusted to 7.2 with KOH. The extracellular buffer for CaV1.2 channels contained (in mM): NaCl, 145; KCl, 4; BaCl2, 10 and HEPES, 10, and pH was adjusted to 7.4 with NaOH. The intracellular buffer contained: CsF, 27; CsCl, 112; EGTA, 8.2; NaCl, 2; HEPES, 10 and Mg-ATP, 4. The pH was adjusted to 7.2 with CsOH. For NaV1.5 channels, the extracellular buffer contained: NaCl, 137; KCl, 4; CaCl2, 2; MgCl2, 1 and HEPES, 10. The pH was adjusted to 7.4 with NaOH. The intracellular buffer contained: CsF, 150; EGTA/CsOH, 1/5; NaCl, 10; MgCl2, 1; CaCl2, 1; HEPES, 10. The pH was adjusted to 7.2 with CsOH. All channels were exposed to six concentrations of R(-)HCQ or S(+)HCQ applied cumulatively in an ascending order up to 90 μM. Specific inhibitors applied at full-block concentrations to establish assay sensitivity and isolate leak currents were E−4031 (10 μM) for the hERG channels, lidocaine (3 mM) for the NaV1.5 channels, CdCl2 (0.2 mM) for CaV1.2, HMR-1556 (30 μM) for KV7.1, SKF-96365 (30 μM) for KV4.3 and BaCl2 (3 mM) for the Kir2.1 channels. Bepridil was used as a positive control and tested at concentrations up to 30 μM in parallel to the HCQ enantiomers for all channels which were insensitive to them. Treatments were done in buffer containing 0.3% or 1% DMSO, and 1% Pluronic F-68. All inhibitions were quantified as change in normalized peak current amplitude except for KV4.3 currents which were quantified as change in normalized integral charge transferred calculated as the area under the current trace versus time. 2.3 Current clamp on excised rabbit Purkinje fibers Methods were reviewed and endorsed by Sanofi's internal animal ethics committee and conducted in accordance with European Directives 86/609/EEC and 2010/63/EU under specific approval by the French Ministry of Research (APAFiS authorization N° 2018080110555740). As previously described (Le Marois et al., 2020), Purkinje fibers were dissected out from excised hearts of male New Zealand rabbits weighing 1.7–2.1 kg and fixed in a small ex vivo tissue bath continuously perfused with a buffer containing (in mM): NaCl 120; KCl 4; MgCl2 1; NaH2PO4 1.8; NaHCO3 25; glucose 11; CaCl2 1.8 maintained at 36 ± 1 °C and delivered at a flow-rate of 10 mL/min. Oxygen supply and physiological pH 7.4 were obtained by bubbling 95% CO2/5% O2 in the buffer reservoirs. The tissue action potential (AP) firing was driven by electrical pulses delivered through stainless-steel electrodes. The membrane resting potential, AP amplitude, depolarization speed (Vmax), and AP duration at 50% and 90% repolarization (ADP50 and APD90, respectively) were recorded using 3 M KCl-filled fine glass micro-electrodes and standard signal amplification and computerized acquisition and monitoring devices. Concentrated (12 mM) stock solutions of S(+)HCQ or R(-)HCQ were prepared in 100% DMSO before being formulated in study buffer to the final concentrations tested. Each fiber (n = 6 per drug) was exposed to four increasing concentrations of R(-)HCQ or S(+)HCQ according to the protocol depicted in Fig. 7. During the recordings, 1 mL samples of the formulated buffers were collected before (in the reservoirs at room temperature) and after the fibers (at the outlet of the tissue chamber 20 min following onset of each concentration perfusion) to determine their actual drug exposure. 2.4 Data analysis Patch-clamp current-trace captures and current-size measurements were exported from Sophion's proprietary Assay Software for Qpatch to a professional graphing package (Prism 8.3.0, GraphPad Software, San Diego, CA, USA) which served also for half-maximal inhibitory concentration (IC50) estimations by fitting percent inhibition data (normalized to pre-drug levels) to a sigmoidal curve with minimum and maximum constrained to 0% and 100%, respectively. Calculated IC50 values are reported along with their 95% confidence interval (95% CI). Statistical analyses of the Purkinje fiber data were performed by one-way repeated factor analysis of variance (ANOVA) followed by Dunnett's test versus baseline for each drug at each pacing rate using SAS 9.4 software (SAS Institute Inc., Cary, NC, USA). A rank-transformation was applied in case of normality hypothesis violation or lack of variance homogeneity. Due to a high number of missing values, the 30 μM data set for R(-)HCQ was excluded from statistical analysis at all pacing rates and for all metrics except for the resting potential data. Likewise, the 10 μM data set for R(-)HCQ at the lowest pacing rate was analyzed only for the resting potential due to missing data for the other parameters. Significance levels were set to 5%. Descriptive statistics and graphs were generated with Prism 8.3.0. 3 Results 3.1 Ion channel studies Cloned human cardiac ion channels expressed in CHO or HEK cells were recorded on a QPatch® 48X automated patch-clamp workstation in population patch mode. Fig. 1 A illustrates the effects of R(-)HCQ and S(+)HCQ on hERG current traces elicited by a classical two-step long-duration (5 s per step) voltage-protocol allowing ample time for channel blockade at room temperature and triggering large outward tail currents upon its + 20 mV to −50 mV repolarizing step. Both R(-)HCQ and S(+)HCQ concentration-dependently inhibited these currents. The time-course of the effect of each of the six concentrations applied cumulatively indicated that the active concentrations could reach steady-state inhibition at the end of each exposure period (Fig. 1B). The final application of the selective and potent hERG blocker E−4031 (1 μM) eliminated all currents, thereby establishing the sensitivity of the assay. The half-maximal inhibitory concentrations (IC50) estimated from the sigmoidal curve fitting of the percent block data indicate a 3.7-fold greater potency of R(-)HCQ over S(+)HCQ (Table 1 ).Fig. 1 R(-)HCQ and S(+)HCQ inhibit hERG potassium currents. Panel A: representative outward currents generated by the voltage-command protocol schematized at reduced scale above the traces in the absence (vehicle, green current trace) or presence of 30 μM of either enantiomer (blue and red traces for R(-)HCQ and S(+)HCQ, respectively). Panel B: time-course of hERG tail current amplitude in response to cumulative application of six increasing concentrations of R(-)HCQ or S(+)HCQ, preceded by three applications of vehicle (Veh.) to test current stability, and followed by a final application of a reference hERG inhibitor (E−4031) at full-block concentration (10 μM). Current traces and symbol colors are matched in panels A and B. Fig. 1 Table 1 Inhibitory potencies of R(-)HCQ and S(+)HCQ on six cardiac ion channels of the Comprehensive in vitro Pro-arrhythmia Assay (CiPA) panel. Table 1Channel R(-)HCQ N = S(+)HCQ N = hERG 20 μM [19–22] 7 74 μM [65–86] 5 NaV1.5 >90 μM 6 >90 μM 7 CaV1.2/β2/α2δ1 >90 μM 6 >90 μM 5 KV4.3/KChIP2.2 >90 μM 7 >90 μM 7 KV7.1/minK >90 μM 4 >90 μM 5 Kir2.1 3.3 μM [2.9–3.7] 10 6.5 μM [5.7–7.6] 6 Subunit compositions of the human recombinant channels tested are detailed in first column. IC50 values are shown along with their 95% confidence intervals. Fig. 2 summarizes the effects of the enantiomers on human NaV1.5 channels expressed in HEK cells. These were held at −80 mV outside of the depolarizing voltage-steps to populate both the closed and inactivated states of the channels as described (Donovan et al., 2011). The fast inactivating inward currents were generated by short depolarizing steps to −40 mV. Interestingly, neither R(-)HCQ nor S(+)HCQ affected NaV1.5 to a meaningful extent (i.e. below about 10% decrease which can be attributed to current rundown) up to the highest tested concentration of 90 μM. In contrast, the local anesthetic and NaV1.5 blocker lidocaine at 3 mM systematically abolished all currents in each recording, confirming the sensitivity of the experimental conditions employed. Furthermore, we evaluated in parallel the multi-channel inhibitor bepridil, once used as an antianginal agent, and found an IC50 value of 2.5 μM along with a 95% confidence interval ranging from 2.3 μM to 2.7 μM, which is fully in accordance with published data (Crumb et al., 2016).Fig. 2 R(-)HCQ and S(+)HCQ do not alter NaV1.5 sodium currents. Panel A: representative inward currents elicited by the voltage-command protocol schematized at reduced scale above the traces in the presence of the indicated concentrations of enantiomers (blue and red traces for R(-)HCQ and S(+)HCQ, respectively) or of 10 μM of the NaV1.5 inhibitor bepridil (black trace). Panel B: time-course of peak NaV1.5 current amplitude in response to cumulative application of six increasing concentrations of R(-)HCQ or S(+)HCQ, preceded by three applications of vehicle (Veh.) to test current stability, and followed by a final application of a reference NaV1.5 inhibitor (lidocaine) at full-block concentration (3 mM). Current traces and symbol colors are matched in panels A and B. Note that a half-log left-shifted concentration-range was tested for the positive control bepridil compared to the concentration-range tested for the enantiomers (see arrow labels in graph). Fig. 2 We then went on to examine the effects of HCQ's enantiomers on CaV1.2 channels expressed in CHO cells along with their β2 and α2δ1 regulatory proteins. During the patch-clamp recordings, the cells were maintained at a depolarized holding of - 50 mV to enrich the population of channels in the inactivated state as described (Kuryshev et al., 2014). The slowly inactivating inward currents were obtained by stepping the membrane to 0 mV with 200 ms square pulses. To minimize current rundown over time, barium was used as a charge carrier. Fig. 3 A depicts the current traces obtained in the presence of 90 μM R(-)HCQ or 90 μM S(+)HCQ. Neither of the enantiomers was able to block CaV1.2 currents, while CdCl2 (0.2 mM) applied at the end of the recording sessions consistently inhibited all the currents (Fig. 3B). Moreover, the positive control bepridil tested in parallel with the enantiomers was active, exhibiting an IC50 value of 5.2 μM [95% CI: 4.4 μM–6.2 μM] in agreement with published values (Crumb et al., 2016).Fig. 3 R(-)HCQ and S(+)HCQ do not alter CaV1.2 calcium currents. Panel A: representative inward currents elicited by the voltage-command protocol schematized at reduced scale above the traces in the presence of the indicated concentrations of enantiomers (blue and red traces for R(-)HCQ and S(+)HCQ, respectively) or of the CaV1.2 inhibitor bepridil (black trace). Panel B: time-course of maximal CaV1.2 current amplitude in response to cumulative application of six increasing concentrations of R(-)HCQ or S(+)HCQ, preceded by three applications of vehicle (Veh.), and followed by a final application of CdCl2 at full-block concentration (0.2 mM). Current traces and symbol colors are matched in panels A and B. Bepridil was applied at a left-shifted concentration-range compared to the enantiomers (see arrow labels in graph). Fig. 3 We then profiled R(-)HCQ and S(+)HCQ on KV4.3 channels. CHO cells expressing these channels along with KChIP2.2 accessory subunits were held at −80 mV and stepped to +20 mV using 500 ms long square pulses. The latter generated large, rapidly inactivating outward currents as expected (Fig. 4 A). Unlike bepridil, neither 90 μM R(-)HCQ nor 90 μM S(+)HCQ were able to affect the current peak amplitude nor inactivation. Fig. 4B depicts the change of the total charge transferred (i.e. the integral of the area-under- the-current versus time curve) as a function of applications of each drug concentration and ultimately of 30 μM SKF-96365, a non-selective cationic channel blocker with micromolar potencies at several cardiac potassium channels including human KV4.3 (Liu et al., 2016). Bepridil, again tested in parallel, exhibited a concentration-dependent inhibitory effect on KV4.3 currents, yielding an IC50 value of 11 μM [95% CI: 9.7 μM–12 μM] (Fig. 4B).Fig. 4 R(-)HCQ and S(+)HCQ do not alter KV4.3 potassium currents. Panel A: representative outward currents elicited by the voltage-command protocol schematized at reduced scale above the traces in the presence of the indicated concentrations of enantiomers (blue and red traces for R(-)HCQ and S(+)HCQ, respectively) or of the KV4.3 inhibitor bepridil (black trace). Panel B: time-course of change in area under the current trace (i.e. total charge transferred) in response to cumulative application of six increasing concentrations of R(-)HCQ or S(+)HCQ, preceded by three applications of vehicle (Veh.) to stabilize current, and followed by a final application of SKF-96365 (SKF) at full-block concentration (30 μM). Current traces and symbol colors are matched in panels A and B. Bepridil was applied at a left-shifted concentration-range compared to the enantiomers (see arrow labels in graph). Fig. 4 To further explore the activities of R(-)HCQ and S(+)HCQ on repolarizing currents of the cardiac AP, we patched CHO cells expressing KV7.1 subunits along with the ancillary protein minK which is needed to reproduce the native I Ks current (Fig. 5 A). The cells were stepped to +40 mV from a holding potential of −80 mV using prolonged 2 s square-pulses. The latter triggered a slowly activating current that resisted the application of either 90 μM R(-)HCQ or the same concentration of S(+)HCQ, while the positive control bepridil was active as expected, giving an IC50 value of 7.2 μM [95% CI: 6.6 μM–7.8 μM] (Fig. 5B).Fig. 5 R(-)HCQ and S(+)HCQ do not alter KV7.1 potassium currents. Panel A: representative outward currents elicited by the voltage-command protocol schematized at reduced scale above the traces in the presence of the indicated concentrations of enantiomers (blue and red traces for R(-)HCQ and S(+)HCQ, respectively) or of the KV7.1 inhibitor bepridil. Panel B: time-course of change in normalized outward current amplitude measured at the end of a 2s-long depolarizing pulse in response to cumulative application of six increasing concentrations of R(-)HCQ or S(+)HCQ preceded by three applications of vehicle (Veh.) to stabilize current, and followed by a final application of HMR-1556 (HMR) at full-block concentration (30 μM). Current traces and symbol colors are matched in panels A and B with blue traces and circles for R(-)HCQ, red traces and squares for S(+)HCQ, and black traces and triangles for bepridil which was applied at a left-shifted concentration-range compared to the enantiomers (see arrows in graph). Fig. 5 Finally, we examined the effects of the enantiomers on Kir2.1 expressed in CHO cells (Fig. 6 ). The cells were maintained at a depolarized holding of −20 mV, and the outward component of the current relevant to AP repolarization was elicited by a step-ramp protocol from +40 mV down to −120 mV at −0.46 V/s followed by a short plateau. As previously noted for hERG, both R(-)HCQ and S(+)HCQ inhibited the Kir2.1-mediated repolarizing currents (Fig. 6A), although with approximately one order of magnitude stronger potency (Fig. 6B), albeit with a similarly modest enantiomeric separation (Table 1).Fig. 6 R(-)HCQ and S(+)HCQ inhibit Kir2.1 potassium currents. Panel A: representative outward currents elicited by the hyperpolarizing ramp of the voltage-command protocol schematized at reduced scale above the traces in the absence (vehicle, green current trace) or presence of the indicated concentration of enantiomer (blue and red traces for R(-)HCQ and S(+)HCQ, respectively). Panel B: time-course of Kir2.1 peak outward current amplitude in response to cumulative application of six increasing concentrations of R(-)HCQ or S(+)HCQ preceded by three applications of vehicle (Veh.) to test current stability and followed by a final application of BaCl2 at full-block concentration (3 mM). Current traces and symbol colors are matched in panels A and B. Fig. 6 Fig. 7 Experimental protocol for the Purkinje fiber studies. The perfused tissue samples were first exposed to normal buffer under pacing at the indicated frequency and for the indicated durations, before being exposed to drug-containing buffer. R(-)HCQ or S(+)HCQ were tested at 0.3 μM, 3 μM, 10 μM and 30 μM, each applied for 30 min in ascending order. After the highest concentration, a 30 min washout with normal buffer was added to evaluate reversibility of drug effects. Five action potential parameters were measured at the end of each frequency period. The slowest pacing rate (i.e. 0.25 Hz or 15 beats/min) was used to favor the occurrence of abnormal repolarization events, if drug should induce any. The fastest pacing rate (i.e. 3 Hz or 180 bpm) was used to evaluate use-dependency of changes observed. Fig. 7 3.2 Purkinje fiber studies Action potentials (AP) were recorded from rabbit Purkinje fibers with conventional KCl-buffer-filled fine glass microelectrodes manually inserted through the conductive tissue membrane. Fig. 7 schematizes the drug application sequence employed and pacing changes made to assess drug effects. Fig. 8 and Fig. 9 depict the effects of R(-)HCQ and S(+)HCQ, respectively, on five parameters analyzed at three different pacing rates, along with their statistical significance. Fig. 10 compares the percent change versus baseline observed for each of these metrics.Fig. 8 Effects of R(-)HCQ on rabbit Purkinje fiber membrane potential and action potential (AP) parameters. Upper panels A, B and C illustrate typical AP-waveforms paced at 3 Hz (180 beats/min, bpm), 1 Hz (60 bpm) and 0.25 Hz (15 bpm), respectively, in the presence of the indicated concentrations of enantiomer. The graphs in panels D through R show median (with interquartile range) and individual values for each of five metrics (arranged vertically) at the different pacing rates (arranged horizontally) during cumulative perfusion of vehicle (black symbols), the four tested concentrations of R(-)HCQ (blue, dark yellow, red and pink symbols) and after 30 min washout with vehicle (green symbols). Symbol colors in graphs D through R match with waveform colors in panels A, B and C. Arrows in panel A, B and C highlight the strong depolarization of membrane resting potential. Typefaces on graphs denote significant differences from baseline (‡ = p < 0.05; ¶ = p < 0.01; § = p < 0.001). Fig. 8 Fig. 9 Effects of S(+)HCQ on rabbit Purkinje fiber membrane potential and action potential (AP) parameters. Upper panels A, B and C illustrate typical AP-waveforms paced at 3 Hz (180 beats/min, bpm), 1 Hz (60 bpm) and 0.25 Hz (15 bpm), respectively, in the presence of the indicated concentrations of enantiomer. The graphs in panels D through R show median (with interquartile range) and individual values for each of five metrics (arranged horizontally) at the different pacing rates (arranged vertically) during cumulative perfusion of vehicle (black symbols), the four tested concentrations of S(+)HCQ (blue, dark yellow, red and pink symbols) and after 30 min washout with vehicle (green symbols). Symbol colors in graphs D through R match with waveform colors in panels A, B and C. Arrows in panel C show occasional early after depolarizations (EADs) due to the strong action potential elongation triggering calcium channel reactivations. Typefaces on graphs denote significant differences from baseline (‡ = p < 0.05; ¶ = p < 0.01; § = p < 0.001). Fig. 9 Fig. 10 Effect-size of R(-)HCQ- and S(+)HCQ-induced changes in Purkinje fiber membrane and AP parameters. The graphs in panels A through O show percent change of median from baseline for each of the five metrics (arranged horizontally) measured at the different pacing rates (arranged vertically) in the presence of the four tested concentrations and after washout from R(-)HCQ (blue bars grouped on the left side of each graph) or S(+)HCQ (red bars grouped on the right side of each graph). Note the same range in vertical axis of all graphs in this figure to allow global effects comparison. Fig. 10 Overall, at the lowest tested concentration of 0.3 μM, neither R(-)HCQ nor S(+)HCQ had a significant effect on any of the parameters whatever the beating rate imposed (compare first and second scatter dot plots in each graph labelled D through R in Fig. 8, Fig. 9). In contrast, both enantiomers started to develop effects that occasionally reached statistical significance at 10-fold higher concentrations. R(-)HCQ consistently decreased the resting potential of the conductive tissue membranes. This effect was concentration-dependent over the 3 μM–30 μM range, and of similar magnitude whatever the pacing rate driven by the electrical stimulations (Fig. 8, graphs D, E and F and Fig. 10, graphs A, B and C). Specifically, and for example, at the highest tested concentration of 30 μM, the median membrane potential was decreased by 23% from – 91 mV to – 70 mV at 180 beats/min (bpm), by 20% from – 90 mV to – 72 mV at 60 bpm and by 21% from – 89 mV to – 70 mV at 15 bpm. These effects were almost completely reversible after washout with vehicle buffer for most metrics (Fig. 8, compare baseline and washout scatter dot plots in graphs D, E and F). R(-)HCQ also produced a significant decrease in the AP amplitude and the depolarization velocity at 10 μM for the 180 bpm and 60 bpm pacing rates (Fig. 8, graphs G through L). However, the rather large membrane depolarization induced by R(-)HCQ resulted in the fibers escaping electrical pacing at the intermediate 60 bpm rate, and even more so at the very slow 15 bpm rate. Hence, AP amplitude, AP depolarization velocity, and APD50 and APD90 measurements could not be collected at 10 μM and 30 μM at the slowest pacing rates, and at 10 μM at the intermediate pacing rate. A nearly complete data set could only be collected from fibers stimulated at the fastest rate (180 bpm) which overrode the autogenic activity, although several data-points were still missing at 30 μM. At this rapid 180 bpm pacing rate, R(-)HCQ significantly decreased the AP amplitude and its upstroke depolarization velocity at 10 μM. The latter was halved at the highest tested concentration (– 51% at 30 μM, with n = 2 only, Fig. 10G). R(-)HCQ also elongated the AP duration by a small, but significant extent at near complete repolarization (Fig. 8, graphs P and Q), an effect culminating at 25% increase during exposure to 30 μM of the drug for the n = 2 readings that could be captured (Fig. 10, graph M). Similar effects could also be inferred at the intermediate (60 bpm) and slow (15 bpm) pacing rates despite the high amount of lacking data, the effects on AP amplitude and depolarization velocity displaying some degree of use-dependency as suggested by the data available at 3 μM (Fig. 10, graphs D through O) while the AP elongating effects globally persisted upon washout. Of note, notwithstanding the few complete data sets, no torsadogenic early after depolarizations (EADs) events were observed in response to R(-)HCQ at any concentration and pacing rate. Compared to R(-)HCQ, its companion enantiomer S(+)HCQ had qualitatively similar effects on the AP metrics, although quantitatively it affected the AP duration more (notably at the very slow pacing rate) and the resting membrane potential less (see in particular Fig. 10, graphs B and O). The prolongation of the action potential at near complete repolarization was concentration-dependent, reverse use-dependent and only partially reversible upon washout (Fig. 9, graphs P through R, and Fig. 10, graphs M through O). As a quantitative illustration, the median APD90 duration at the concentration of 10 μM was increased by 12% from 220 ms to 246 ms at 180 bpm, by 34% from 358 ms to 481 ms at 60 bpm, and by 56% from 531 ms to 827 ms at 15 bpm. Autogenic activity escaping electrical pacing was seen in two out of six fibers exposed to 30 μM at all rates. Of note, at the extreme slow pacing rate of 15 bpm, abnormal AP exhibiting EADs were seen in one of the six fibers exposed to 10 μM, and in one of the four fibers that could follow pacing in the presence of 30 μM S(+)HCQ, suggesting some degree of torsadogenic potential for this enantiomer beyond 10 μM (see AP waveform in Fig. 9C). Regarding the other AP metrics, S(+)HCQ also substantially decreased the speed of the action potential upstroke depolarization (Fig. 9, graphs J, K and L), an effect that was partially reversible and use-dependent (Fig. 10, graphs G, H and I). At the most rapid pacing rate (180 bpm), a depolarization of the resting potential was observed which culminated at 11% decrease from – 91 mV to – 81 mV in the presence of 30 μM S(+)HCQ (Fig. 9D), a significant and concentration-dependent change, although clearly less pronounced than the effect produced by R(-)HCQ on this same parameter (Fig. 10 A, B and C). Table 2 compares concentration estimates made by standard LC-MS analytic methods on the R(-)HCQ and S(+)HCQ content in drug-formulated buffer samples collected in the reservoir and at the outlet of the tissue chamber after 20 min perfusion of the Purkinje fibers. The data indicate that none of the tested compounds was lost by retention in the conductive tissue pieces or adherence to components of the experimental setup such as tissue chamber walls or the buffer circulation tubing and pumps.Table 2 R(-) and S(+)HCQ content in the formulated buffer samples collected before and after exposure to the Purkinje fibers. Table 2Drug Target (μM) Actual (μM) Loss R(-)HCQ 0.3 0.29 - 3% 3 2.9 - 3% 10 9.8 - 2% 30 29.6 - 1% S(+)HCQ 0.3 0.31 +3% 3 3.1 +3% 10 10.1 +1% 30 30.1 <1% 4 Discussion The present experiments indicate that, among the cardiac channels comprising the CiPA panel, the R(-)- and S(+)-enantiomers of HCQ selectively inhibit the human Kir2.1- and hERG-potassium channels, leaving all four other major cardiac currents unaffected up to the highest tested concentrations. We also found that this inhibitory profile can translate into significant alterations of rabbit Purkinje fibers action potential (AP) parameters recorded in an ex vivo situation at concentrations above the therapeutically effective free circulating levels of HCQ. The automated patch-clamp studies show that the Kir2.1 channels are the most sensitive to inhibition, with IC50 values occurring within a low 1 μM–10 μM concentration range for both enantiomers, while the blockade of hERG channels by R(-)HCQ and S(+)HCQ developed within a tenfold higher concentration range. In contrast, neither R(-)HCQ nor S(+)HCQ affected NaV1.5, CaV1.2, KV4.3 nor KV7.1 channels by a meaningful amount up to 90 μM. This cardiac ion channel inhibitory profile agrees with the profile of chloroquine (CQ), the antiparasitic predecessor of HCQ. Indeed, using a manual patch-clamp approach (Crumb et al., 2016), observed less than 20% inhibition of human recombinant NaV1.5, CaV1.2, KV4.3 and KV7.1 channels exposed to 30 μM CQ, while hERG and Kir2.1 channels where inhibited with IC50 values of 7 μM and 11 μM, respectively. Similarly, in thorough mechanistic study aimed at identifying the biophysical and molecular basis of Kir2.1 blockade by CQ, the IC50 value on the outward component was found to be 9 μM (Rodriguez-Menchaca et al., 2008). Hence, hydroxylation of CQ's side chain, which reduces its toxicity, does not affect its selectivity nor greatly change its potency at those of the CiPA cardiac ion channels that are sensitive to 4-aminoquinolines. More specifically, our data are also fully in line with those recently obtained on a new generation QPatch II platform, with IC50 values for racemic HCQ on hERG and Kir2.1 reaching 17 μM and 30 μM, respectively (Okada et al., 2021). While it is unsure to which extent the differences in IC50 values are significant, the reasons why the channel sensitivity is inverted in both studies is unclear. Mechanistically, it is tempting to ascribe the non-use-depend depolarizing effects of R(-)HCQ on the Purkinje fibers to its prominent potency at blocking non-voltage-gated Kir2.1 channels. These inwardly rectifying pores are chiefly involved in setting the diastolic resting membrane potential close to the potassium equilibrium potential between consecutive AP firings. Kir2.1 channels also contribute to the repolarization phase of the AP in tandem with hERG and KV7.1 channels (Hibino et al., 2010), providing a way by which R(-)HCQ prolongs the APD90. This prolongation of the AP duration could indeed be observed over the full R(-)HCQ concentration-range at the fastest pacing rate, and up to 10 μM and 3 μM at the intermediate and slow pacing rates, respectively. In the absence of direct inhibition of NaV1.5 channels, the slowing of the AP depolarization velocity and the decrease of the AP amplitude induced by both R(-)HCQ and S(+)HCQ can also be regarded as consequences of the Kir2.1 channel blockade, given that the rising membrane depolarization will engage fast inactivation of a substantially growing fraction of NaV1.5 channels, thereby removing them from the population available in the resting-state ready for activation. This implies that the effects of S(+)HCQ, which is slightly less potent at blocking Kir2.1 channels, on AP amplitude and depolarization velocity would be somewhat less pronounced. This is however difficult to establish given the incomplete sets of measurements we could acquire across pacing rates due to the autogenic activity induced by R(-)HCQ beyond 3 μM. Overall, we found only a modest degree of enantiomeric separation between R(-)HCQ and S(+)HCQ. Their effects on the cardinal AP metrics were largely similar (see Fig. 10) and in the patch-clamp experiments, their block-potency ratio against Kir2.1 was only two-fold, and hardly reached four-fold against hERG, although in both cases the levorotatory enantiomer was the most potent. The pharmacodynamics of HCQ's enantiomers had previously been compared in other in vitro models without showing potency differences. For example, R(-)HCQ and S(+)HCQ were found equipotent at inhibiting human immune-deficiency virus serotype 1 replication in human T-cell lines and macrophage hybridoma (Chiang et al., 1996). Likewise, the proliferation of Plasmodium falciparum isolates grown in human erythrocytes was inhibited with identical IC50 values by R(-)HCQ and S(+)HCQ (Warhurst et al., 2003). Our present in vitro data also do not suggest that racemic HCQ harbors a clear-cut eutomer and distomer combination of drugs from the arrhythmogenic cardiac safety standpoint. Chirality affects somewhat more meaningfully the pharmacokinetics of HCQ. Indeed, besides the fact that its fate in the body is characterized by very large volumes of distributions and long half-life, the circulating levels at therapeutic doses display enantioselectivity. Specifically, in rheumatoid arthritis patients stably treated with racemic HCQ at the recommended dose, the R(-)/S(+) ratio in blood levels was found to be greater than one in all individuals despite a large two-to three-fold variability between patients (Tett et al., 1994). Moreover, the pharmacokinetics of HCQ were confirmed to be stereoselective in rats (Wei et al., 1995) and rabbits (Ducharme et al., 1994) with no hints for in vivo racemization and a preferential accumulation of the R(-)HCQ enantiomer in ocular tissue (Wainer et al., 1994). The recent consideration of HCQ as a potential treatment for Covid-19 has generated concerns because of its 4-aminoquinoline structure common to CQ which is known to prolong the QT-segment of the ECG. In a meta-analysis of 28 clinical studies in which patients took HCQ with or without other QT-prolonging drugs, the occurrence of QT-prolongations associated with HCQ monotherapy was below 2%, although patients over 60 appeared significantly more sensitive (Oscanoa et al., 2020). Our in vitro cardiosafety data should be discussed with respect to the circulating drug levels at therapeutic doses. In a population pharmacokinetics model developed from rheumatoid arthritis patients receiving daily doses of 400 mg HCQ, average concentrations in the whole blood reached 2.8 μM (Carmichael et al., 2003). However, recent data comparing whole blood, serum and plasma levels demonstrated that the plasma concentrations of HCQ are two-fold lower than those in blood (Carlsson et al., 2020). Furthermore, in vitro data indicate that the enantiomer of HCQ bind plasma proteins, with the free fraction of R(-)HCQ being about twice as high as that of S(+)HCQ at 63% and 36%, respectively (McLachlan et al., 1993). Given that only the free fraction exercises pharmacodynamic effects such as those described here, a conservative estimate of the therapeutic free plasma concentrations of HCQ enantiomers for comparison with our electrophysiology findings are clearly below 1 μM. Specifically, such a sub-micromolar estimate is comforted by data calculated in a population PK model derived from lupus patients in which mean plasma HCQ ranges between 0.5 μM–0.9 μM for Cmax and 0.3 μM–0.4 μM for Ctrough (Morita et al., 2016). These conservative estimations may however gain relevance in specific circumstances where the cardiac AP repolarization reserve is altered, such as for example in carriers of congenital long-QT syndromes (the prevalence of which could be estimated to be as high as 1:2500 according to recent studies (Schwartz et al., 2009)), or when medications known to prolong the QT interval are administered concomitantly (Roden, 2004). The present study has limitations, mostly inherent to the in vitro and ex vivo nature of our experimental approaches. First, the small sample size of the Purkinje fibers studies, that was driven by animal ethics considerations and the fastidiousness of manual Purkinje fiber recordings, may reduce the general relevance of our experimental observations. This applies notably to the effects of R(-)HCQ since several parameters could not be properly assessed over all pacing rates and concentrations because of the autogenic activity induced by this enantiomer. It would be relevant to compare the effects of both enantiomers in a more integrated model such as, for example, the rabbit whole-heart preparation perfused according to the Langendorff method, which might offer a better translation to the clinical situation given its higher degree of physiologic integrity (Ellermann et al., 2021). Second, there are uncertainties over the drug effective concentrations when attempting to translate to the clinical situation. Indeed, the defined buffers used in the in vitro patch-clamp and the ex vivo voltage-clamp experiments do not exhaustively mimic the plasma milieu, even if the enantiomers of HCQ bind plasmatic proteins relatively loosely (McLachlan et al., 1993) and are readily soluble, and not particularly suspected of stickiness due to lipophilicity, which can be a source of compound loss in automated patch-clamp experiments (Kramer et al., 2020). Third, the Purkinje fiber studies were performed in standard conditions. However, in real life, HCQ may need to be given to patients with associated pro-arrhythmic risk factors, such as for example hypokalemia, which is well known to promote cardiac arrhythmias (Weiss et al., 2017). It could have been useful to evaluate the effects of R(-)HCQ and S(+)HCQ in the presence of reduced potassium concentrations, as has been recently been done when racemic HCQ was tested in human adult cardiomyocytes. Indeed, the proarrhythmic risk rose significantly when the potassium concentration in the experimental buffer was decreased from 4 mM to 2.5 mM (Jordaan et al., 2021). Finally, the translatability to humans of our Purkinje fibers data needs to consider the fact that I to currents are primarily conducted through KV1.4 channels in rabbits rather than KV4.3 as in humans, a feature that may result in differences in phase 1 repolarization and propensity to engage arrhythmia (Patel and Campbell, 2005). In conclusion, ion channel inhibitions and cardiac cell membrane alterations were seen with both R(-)HCQ and S(+)HCQ at concentrations in the single to double-digit micromolar range. Given the sub-micromolar free plasma concentrations that can reasonably be estimated for HCQ enantiomers at therapeutic doses, the present in vitro data agree with the low incidence of arrhythmogenic events such as TdP reported over the long-standing clinical practice of HCQ monotherapies in its approved indications. However, caution should be exercised when this drug is given in conjunction with other QT-prolonging drugs or in patients with additional risk factors such as electrolyte imbalance or congenital long QT syndromes. CRediT authorship contribution statement Véronique Ballet: designed and directed the studies, interpreted the results and wrote the manuscript.. G. Andrees Bohme: interpreted the results and wrote the manuscript.. Eric Brohan: designed and directed the studies. Rachid Boukaiba: designed and directed the studies. Jean-Marie Chambard: designed and directed the studies. Odile Angouillant-Boniface: generated and curated the data. Thierry Carriot: generated and curated the data. Céline Chantoiseau: generated and curated the data. Sophie Fouconnier: generated and curated the data. Sylvie Houtmann: generated and curated the data. Céline Prévost: generated and curated the data. Brigitte Schombert: generated and curated the data. Laurent Schio: initiated and supervised the project. Michel Partiseti: initiated and supervised the project, interpreted the results and wrote the manuscript, All authors approved the final version and concur with the conclusions. Declaration of competing interest All authors are current or former Sanofi employees and may hold shares and/or stock options in the company. Acknowledgements The authors are grateful to Dorothée Tamarelle for her expert counseling in the statistical analysis of the data and critical reading of the manuscript. The contributions of Christelle Musso for HPLC quantification of R(-)HCQ and S(+)HCQ containing buffers and of Fiona Ducrey for proofreading of the final manuscript are also gratefully acknowledged. ==== Refs References Baczko I. Jost N. Virag L. Bosze Z. Varro A. Rabbit models as tools for preclinical cardiac electrophysiological safety testing: importance of repolarization reserve Prog. Biophys. Mol. Biol. 121 2016 157 168 27208697 Blaney P.M.B.S.J. Carr G. Ellames G.J. Herbert J.M. Peace J.E. Smith D.I. A practical synthesis of the enantiomers of hydroxychloroquine Tetrahedron: Asymmetry 5 1994 1815 1822 Carlsson H. Hjorton K. Abujrais S. Ronnblom L. Akerfeldt T. Kultima K. Measurement of hydroxychloroquine in blood from SLE patients using LC-HRMS-evaluation of whole blood, plasma, and serum as sample matrices Arthritis Res. Ther. 22 2020 125 32475347 Carmichael S.J. Charles B. Tett S.E. Population pharmacokinetics of hydroxychloroquine in patients with rheumatoid arthritis Ther. Drug Monit. 25 2003 671 681 14639053 Chiang G. Sassaroli M. Louie M. Chen H. Stecher V.J. Sperber K. Inhibition of HIV-1 replication by hydroxychloroquine: mechanism of action and comparison with zidovudine Clin. Therapeut. 18 1996 1080 1092 Crumb W.J. Jr. Vicente J. Johannesen L. Strauss D.G. An evaluation of 30 clinical drugs against the comprehensive in vitro proarrhythmia assay (CiPA) proposed ion channel panel J. Pharmacol. Toxicol. Methods 81 2016 251 262 27060526 Donovan B.T. Bakshi T. Galbraith S.E. Nixon C.J. Payne L.A. Martens S.F. Utility of frozen cell lines in medium throughput electrophysiology screening of hERG and NaV1.5 blockade J. Pharmacol. Toxicol. Methods 64 2011 269 276 21996251 Ducharme J. Wainer I.W. Parenteau H.I. Rodman J.H. Stereoselective distribution of hydroxychloroquine in the rabbit following single and multiple oral doses of the racemate and the separate enantiomers Chirality 6 1994 337 346 8068491 Ellermann C. Wolfes J. Eckardt L. Frommeyer G. Role of the rabbit whole-heart model for electrophysiologic safety pharmacology of non-cardiovascular drugs Europace 23 2021 828 836 33200170 Gintant G. Sager P.T. Stockbridge N. Evolution of strategies to improve preclinical cardiac safety testing Nat. Rev. Drug Discov. 15 2016 457 471 26893184 Hibino H. Inanobe A. Furutani K. Murakami S. Findlay I. Kurachi Y. Inwardly rectifying potassium channels: their structure, function, and physiological roles Physiol. Rev. 90 2010 291 366 20086079 Kramer J. Himmel H.M. Lindqvist A. Stoelzle-Feix S. Chaudhary K.W. Li D. Bohme G.A. Bridgland-Taylor M. Hebeisen S. Fan J. Renganathan M. Imredy J. Humphries E.S.A. Brinkwirth N. Strassmaier T. Ohtsuki A. Danker T. Vanoye C. Polonchuk L. Fermini B. Pierson J.B. Gintant G. Publisher Correction: cross-site and cross-platform variability of automated patch clamp assessments of drug effects on human cardiac currents in recombinant cells Sci. Rep. 10 2020 11884 Kuryshev Y.A. Brown A.M. Duzic E. Kirsch G.E. Evaluating state dependence and subtype selectivity of calcium channel modulators in automated electrophysiology assays Assay Drug Dev. Technol. 12 2014 110 119 24579774 Le Marois M. Ballet V. Sanson C. Maizieres M.A. Carriot T. Chantoiseau C. Partiseti M. Bohme G.A. Cannabidiol inhibits multiple cardiac ion channels and shortens ventricular action potential duration in vitro Eur. J. Pharmacol. 886 2020 173542 Liu H. Yang L. Chen K.H. Sun H.Y. Jin M.W. Xiao G.S. Wang Y. Li G.R. SKF-96365 blocks human ether-a-go-go-related gene potassium channels stably expressed in HEK 293 cells Pharmacol. Res. 104 2016 61 69 26689773 Mazzanti A. Briani M. Kukavica D. Bulian F. Marelli S. Trancuccio A. Monteforte N. Manciulli T. Morini M. Carlucci A. Viggiani G. Cannata F. Negri S. Bloise R. Memmi M. Gambelli P. Carbone A. Molteni M. Bianchini R. Salgarello R. Sozzi S. De Cata P. Fanfulla F. Ceriana P. Locatelli C. Napolitano C. Chiovato L. Tomasi L. Stefanini G.G. Condorelli G. Priori S.G. Association of hydroxychloroquine with QTc interval in patients with COVID-19 Circulation 142 2020 513 515 32501756 McLachlan A.J. Cutler D.J. Tett S.E. Plasma protein binding of the enantiomers of hydroxychloroquine and metabolites Eur. J. Clin. Pharmacol. 44 1993 481 484 8359187 Morita S. Takahashi T. Yoshida Y. Yokota N. Population pharmacokinetics of hydroxychloroquine in Japanese patients with cutaneous or systemic lupus erythematosus Ther. Drug Monit. 38 2016 259 267 26587870 Okada J.I. Yoshinaga T. Washio T. Sawada K. Sugiura S. Hisada T. Chloroquine and hydroxychloroquine provoke arrhythmias at concentrations higher than those clinically used to treat COVID-19: a simulation study Clin. Transl. Sci. 14 2021 1092 1100 33404133 Oscanoa T.J. Vidal X. Kanters J.K. Romero-Ortuno R. Frequency of long QT in patients with SARS-CoV-2 infection treated with hydroxychloroquine: a meta-analysis Int. J. Antimicrob. Agents 2020 106212 Patel S.P. Campbell D.L. Transient outward potassium current, 'Ito', phenotypes in the mammalian left ventricle: underlying molecular, cellular and biophysical mechanisms J. Physiol. 569 2005 7 39 15831535 Jordaan P. Dumotier B. Traebert M. Miller P.E. Ghetti A. Urban L. Abi-Gerges N. Cardiotoxic potential of hydroxychloroquine, chloroquine and azithromycin in adult human primary cardiomyocytes Toxicol. Sci. 180 2021 356 368 33483756 Rampe D. Brown A.M. A history of the role of the hERG channel in cardiac risk assessment J. Pharmacol. Toxicol. Methods 68 2013 13 22 23538024 Roden D.M. Drug-induced prolongation of the QT interval N. Engl. J. Med. 350 2004 1013 1022 14999113 Rodriguez-Menchaca A.A. Navarro-Polanco R.A. Ferrer-Villada T. Rupp J. Sachse F.B. Tristani-Firouzi M. Sanchez-Chapula J.A. The molecular basis of chloroquine block of the inward rectifier Kir2.1 channel Proc. Natl. Acad. Sci. U. S. A. 105 2008 1364 1368 18216262 Sanguinetti M.C. Jiang C. Curran M.E. Keating M.T. A mechanistic link between an inherited and an acquired cardiac arrhythmia: HERG encodes the IKr potassium channel Cell 81 1995 299 307 7736582 Schrezenmeier E. Dorner T. Mechanisms of action of hydroxychloroquine and chloroquine: implications for rheumatology Nat. Rev. Rheumatol. 16 2020 155 166 32034323 Schwartz P.J. Stramba-Badiale M. Crotti L. Pedrazzini M. Besana A. Bosi G. Gabbarini F. Goulene K. Insolia R. Mannarino S. Mosca F. Nespoli L. Rimini A. Rosati E. Salice P. Spazzolini C. Prevalence of the congenital long-QT syndrome Circulation 120 2009 1761 1767 19841298 Tett S.E. McLachlan A.J. Cutler D.J. Day R.O. Pharmacokinetics and pharmacodynamics of hydroxychloroquine enantiomers in patients with rheumatoid arthritis receiving multiple doses of racemate Chirality 6 1994 355 359 8068493 Tucker G.T. Chiral switches Lancet 355 2000 1085 1087 10744105 Wainer I.W. Chen J.C. Parenteau H. Abdullah S. Ducharme J. Fieger H. Iredale J. Distribution of the enantiomers of hydroxychloroquine and its metabolites in ocular tissues of the rabbit after oral administration of racemic-hydroxychloroquine Chirality 6 1994 347 354 8068492 Warhurst D.C. Steele J.C. Adagu I.S. Craig J.C. Cullander C. Hydroxychloroquine is much less active than chloroquine against chloroquine-resistant Plasmodium falciparum, in agreement with its physicochemical properties J. Antimicrob. Chemother. 52 2003 188 193 12837731 Wei Y. Nygard G.A. Ellertson S.L. Khalil S.K. Stereoselective disposition of hydroxychloroquine and its metabolite in rats Chirality 7 1995 598 604 8593253 Weiss J.N. Qu Z. Shivkumar K. Electrophysiology of hypokalemia and hyperkalemia Circ Arrhythm Electrophysiol 10 2017 Wilson L.J. Mi C. Kraml C.M. A preparative chiral separation of hydroxychloroquine using supercritical fluid chromatography J. Chromatogr. A 1634 2020 461661

34863995

PMC9749463

NO-CC CODE

2022-12-15 23:23:20

no

Eur J Pharmacol. 2022 Jan 15; 915:174670

utf-8

Eur J Pharmacol

10.1016/j.ejphar.2021.174670

oa_other

==== Front Environ Health Perspect Environ Health Perspect EHP Environmental Health Perspectives 0091-6765 1552-9924 Environmental Health Perspectives EHP11652 10.1289/EHP11652 Research Letter Federal PFAS Testing and Tribal Public Water Systems Mok Kira 1 Salvatore Derrick 2 Powers Martha 1 3 Brown Phil 1 3 Poehlein Maddy 4 Conroy-Ben Otakuye 5 https://orcid.org/0000-0001-5223-2848 Cordner Alissa 6 1 Department of Sociology and Anthropology, Northeastern University, Boston, Massachusetts, USA 2 Department of Marine and Environmental Sciences, Northeastern University, Boston, Massachusetts, USA 3 Department of Health Sciences, Northeastern University, Boston, Massachusetts, USA 4 PFAS Project Lab, Northeastern University, Boston, Massachusetts, USA 5 School of Sustainable Engineering and the Built Environment, Arizona State University, Tempe, Arizona, USA 6 Department of Sociology, Whitman College, Walla Walla, Washington, USA Address correspondence to Alissa Cordner, Whitman College, 345 Boyer Ave., Walla Walla, WA 99362 USA. Telephone: (509) 527-5124. Email: cordneaa@whitman.edu 14 12 2022 12 2022 130 12 12770131 5 2022 26 8 2022 17 11 2022 https://ehp.niehs.nih.gov/about-ehp/license EHP is an open-access journal published with support from the National Institute of Environmental Health Sciences, National Institutes of Health. All content is public domain unless otherwise noted. The authors have no conflicts of interest or competing interest to disclose. Note to readers with disabilities: EHP strives to ensure that all journal content is accessible to all readers. However, some figures and Supplemental Material published in EHP articles may not conform to 508 standards due to the complexity of the information being presented. If you need assistance accessing journal content, please contact ehpsubmissions@niehs.nih.gov. Our staff will work with you to assess and meet your accessibility needs within 3 working days. ==== Body pmcIntroduction Systemic environmental health disparities exist for residents of Tribal Nations in the United States, who are disproportionately burdened by diseases and experience lower life expectancy compared to non-Native individuals.1 Research on Tribal drinking water is limited but includes documentation of high rates of unsafe levels of inorganic contaminants, nitrates, and foul odor and taste.2 Per- and polyfluoroalkyl substances (PFAS), a large class of persistent, toxic, and water-soluble chemicals, are a leading concern for safe drinking water.3 Exposure to PFAS has been associated with decreased antibody response, decreased fetal and infant growth, and increased risk of kidney cancer, and the evidence also suggests a relationship between PFAS exposure and the risk of breast cancer, testicular cancer, and thyroid disease.3 An estimated 200 million U.S. residents receive PFAS-contaminated public drinking water,4 but no federal regulatory drinking water standards currently exist.5 Large gaps exist in knowledge about PFAS contamination on Tribal lands. To explore these gaps, we conducted a comparative analysis of past and future drinking water testing for Tribal and non-Tribal public water systems (PWS). Methods From 2013 to 2015, the U.S. Environmental Protection Agency (U.S. EPA) conducted drinking water sampling through Unregulated Contaminant Monitoring Rule 3 (UCMR3) for 21 contaminants, including six PFAS, in community water systems and nontransient noncommunity PWS serving more than 10,000 people (large PWS), as well as 800 PWS serving <10,000 people (small PWS).6 To analyze PWS tested for PFAS in UCMR3 and the populations they served, we obtained data on PWS that submitted data to the U.S. EPA’s Safe Drinking Water Information System (SDWIS) and were listed as active in quarter 1 of 2013. We identified Tribal PWS as those with a Native American owner type in SDWIS in 2013. The U.S. EPA’s planned UCMR5 (2023–2025) will sample PWS serving >3,300 people and a random sample of 800 PWS serving ≤3,300 people.7 To calculate the projected inclusion of Tribal PWS in UCMR5, we analyzed PWS that submitted data to SDWIS and were listed as active in quarter 2 of 2022, which was the most up-to-date PWS data available at time of submission. We assumed that all PWS serving >3,300 people will be sampled. We projected the random sampling of 800 small PWS based on the proportion and average populations served by Tribal and non-Tribal PWS serving ≤3,300 people. Analysis was conducted in RStudio (version 2021.09.3; RStudio, PBC). To determine the extent of additional PFAS testing on Tribal lands, we communicated with U.S. EPA representatives to identify sampling plans, engagement with state programs, and funding sources. Results Table 1 shows that 3.2% (n=27) of Tribal PWS were tested for PFAS in UCMR3, in comparison with 7.2% (n=4,892) of non-Tribal systems. A total of 27.8% (n=352,790) of the population served by Tribal PWS were included in UCMR3, in comparison with 79.1% (n=242,265,582) of the population served by non-Tribal PWS. No data were provided for 16.7% (n=3) of large Tribal PWS and 4.3% (n=175) of large non-Tribal PWS due to missing data or lack of sampling in UCMR3. Additionally, of PWS sampled in UCMR3, no PFAS results were provided for 18.2% (n=6) of Tribal PWS and 11.5% (n=637) of non-Tribal PWS due to missing data or lack of sampling for PFAS. The population served by Tribal PWS are disproportionately served by small systems, with 68.5% (n=869,892) of the population served by Tribal PWS receiving water from PWS serving ≤10,000 people, in comparison with just 18.8% (n=57,726,562) of the population served by non-Tribal PWS. Table 1 Analysis of completed (UCMR3) and planned (UCMR5) sampling inclusion of PWS serving Tribal and non-Tribal populations. Tribal Non-Tribal Systems [n (%)] Population [n (%)] Systems [n (%)] Population [n (%)] Total PWS, 2013 SDWIS 847 1,269,153 67,864 306,347,928  Serving >10,000 people 18 (2.1%) 399,261 (31.5%) 4,258 (6.3%) 248,621,366 (81.2%)  Serving ≤10,000 people 829 (97.9%) 869,892 (68.5%) 63,606 (93.7%) 57,726,562 (18.8%) PWS sampled for PFAS in UCMR3 (2013–2015) 27 (3.2%) 352,790 (27.8%) 4,892 (7.2%) 242,265,582 (79.1%)  PWS serving >10,000 people reporting data for PFAS (% of same-size PWS) 15 (83.3%) 305,466 (76.5%) 4,077 (95.7%) 239,356,389 (96.3%)  PWS serving ≤10,000 people reporting data for PFAS (% of same-size PWS) 12 (1.4%) 47,324 (5.4%) 815 (1.3%) 2,909,193 (5.0%) Total PWS, 2022 SDWIS 855 1,400,197 65,904 322,312,628  Serving >3,300 people 98 (11.6%) 896,474 (63.8%) 9,553 (14.5%) 294,503,029 (91.3%)  Serving ≤3,300 people 757 (88.5%) 503,723 (36.0%) 56,351 (85.5%) 27,802,557 (8.6%) PWS projected to be sampled for PFAS in UCMR5 (2023–2025) 109 (12.9%) 903,530 (64.3%) 10,342 (15.7%) 294,510,071 (91.5%)  PWS serving >3,300 people to be sampled for PFAS (% of same-size PWS) 98 (100%) 896,474 (100%) 9,553 (100%) 294,510,071 (100%)  PWS serving ≤3,300 people to be sampled for PFAS (% of same-size PWS) 11 (1.5%) 7,056 (1.4%) 789 (1.4%) 389,473 (1.4%) Note: Sources include U.S. EPA.6–7 PFAS, per-and polyfluoroalkyl Substances; PWS, public water systems; SDWIS, Safe Drinking Water Information System; UCMR, Unregulated Contaminant Monitoring Rule; U.S. EPA, U.S. Environmental Protection Agency. We projected that 12.7% (n=109) of Tribal PWS and 15.7% (n=10,342) of non-Tribal PWS will be sampled for PFAS in 2023–2025 (Table 1). Just 64.5% (n=903,503) of the population served by Tribal PWS will be included in UCMR5; in comparison, 91.5% (n=294,899,544) of the population served by non-Tribal PWS will be included in UCMR5. Over one-third (36.0%, n=503,723) of the population served by Tribal PWS receives water from PWS serving ≤3,300 people, in comparison with just 8.6% (n=27,802,557) of the population served by non-Tribal PWS. Each U.S. EPA region has a Public Water System Supervision (PWSS) State and Tribal Support Program Grant that provides regulatory support and funding related to PWS and emerging contaminants on Tribal lands.8 Per conversations with representatives, 6 of 10 U.S. EPA regions plan to conduct “limited, voluntary” sampling in Tribal PWS for PFAS in 2021–22 (Table 2).9 Table 2 Tribal drinking water PFAS testing under Public Water System Supervision State and Tribal Support Program grants for emerging contaminants. U.S. EPA Region PFAS sampling planned Priority contaminants by region Status of PFAS sampling 1 No NA No planned PFAS sampling 2 Yes PFAS Sampling will be conducted for two Tribes 3 No NA No Tribal PWS in region 4 No NA Sampling may be conducted by U.S. EPA contractor 5 Yes PFAS Sampling projected to begin early 2022 6 No Manganese No planned PFAS sampling 7 Yes PFAS Sampling completed in 2021, results not yet available 8 Yes Manganese, PFAS Sampling projected to begin early 2022 9 Yes PFAS Started sampling late 2021, projected to continue through 2022 10 Yes PFAS Sampling projected to begin early 2022 Note: Testing results may have been released since this paper was finalized. Results available at (reference 9). Source: U.S. EPA Tribal Drinking Water Headquarters and Regions, personal communications, 2021–2022. NA, not available; PFAS, per-and polyfluoroalkyl substances; U.S. EPA, Environmental Protection Agency. U.S. EPA representatives identified policy, funding, and staffing as limiting factors related to the implementation of such PFAS testing. Multiple regions anticipated challenges should PFAS be detected in Tribal PWS, citing the absence of current regulations for PFAS and insufficient remediation funding. Representatives also pointed to a lack of U.S. EPA-certified labs and the need to divide scarce resources between multiple priority contaminants (U.S. EPA Tribal Drinking Water Headquarters and Regions, personal communications, 2021–2022). Discussion Our study has several limitations. Missing or incomplete data from UCMR3 add uncertainty to our analysis of historical testing. Additionally, Tribal PWS are identified by owner type and not by the demographics of the population served, because demographic data are not available at the PWS level. Comprehensive PFAS drinking water testing for Tribal communities is needed. Future research should examine other potential sources of PFAS exposure for Tribal communities. Assessing and managing environmental health risks must incorporate culturally significant practices and traditional ecological knowledge, as well as Tribally defined boundaries and traditional hunting and fishing areas.10,11 Our analysis shows that even systematic research may fail to equitably include certain populations. Therefore, we suggest that UCMR5 be amended to provide resources and support for the inclusion of more Tribal PWS and that the U.S. EPA should support testing of additional Tribal water sources, such as private wells. Other measures, such as education and remediation, should be pursued in locations where contamination is detected, especially in Tribal communities that have historically been excluded from PFAS action. Small PWS may need targeted resources given the substantial remediation costs associated with PFAS contamination. State agencies could offer greater support for focused PFAS monitoring and remediation in Tribal Nations. Although developing data on environmental inequalities for Tribal communities is not a sufficient condition for addressing environmental injustice, it is a necessary step. Acknowledgments This research was supported by the National Science Foundation (SES-1827817 and SES-2120510) and the National Institute of Environmental Health Sciences (1R01ES028311-01A1, 1T32ES023769-01, and R25ES025496). The authors thank P. Hingst, M. Junker, and members of the PFAS Project Lab for their useful suggestions and comments. The authors are also grateful to the U.S. EPA representatives who generously shared their time to describe their programmatic work. ==== Refs References 1. Indian Health Service. 2019. Disparities. https://www.ihs.gov/newsroom/factsheets/disparities/ [accessed 27 April 2022]. 2. Teodoro MP, Haider M, Switzer DU. 2018. U.S. Environmental policy implementation on tribal lands: trust, neglect, and justice. Policy Stud J 46 (1 ):37–59, 10.1111/psj.12187. 3. National Academies of Sciences, Engineering, and Medicine. 2022. Guidance on PFAS Exposure, Testing, and Clinical Follow-Up. Washington, DC: National Academies Press, PMID: , 10.17226/26156. 4. Andrews DQ, Naidenko OV. 2020. Population-wide exposure to per- and polyfluoroalkyl substances from drinking water in the United States. Environ Sci Technol Lett 7 (12 ):931–936, 10.1021/acs.estlett.0c00713. 5. U.S. EPA (U.S. Environmental Protection Agency). 2022. Per- and Polyfluoroalkyl Substances (PFAS). https://www.epa.gov/pfas [accessed 26 May 2022]. 6. U.S. EPA. Third Unregulated Contaminant Monitoring Rule. 2021. https://www.epa.gov/dwucmr/third-unregulated-contaminant-monitoring-rule [accessed 11 November 2021]. 7. U.S. EPA. 2022. The Fifth Unregulated Contaminant Monitoring Rule (UCMR 5). https://www.epa.gov/dwucmr/fifth-unregulated-contaminant-monitoring-rule [accessed 4 April 2022]. 8. U.S. EPA. 2021. Tribal Public Water System Supervision Program. https://www.epa.gov/tribaldrinkingwater/tribal-public-water-system-supervision-program [accessed 4 April 2022]. 9. U.S. EPA. 2022. Safe Drinking Water on Tribal Lands: Tribal PFAS Monitoring Results. https://sdwis.epa.gov/ords/sfdw_pub/f?p=SDWIS_FED_REPORTS_PUBLIC:TRIBAL_PFAS [accessed 29 November 2022]. 10. Cummins C, Doyle J, Kindness L, Lefthand MJ, Bear Dont Walk UJ, Bends AL, et al. 2010. Community-based participatory research in Indian country: improving health through water quality research and awareness. Fam Community Health 33 (3 ):166–174, PMID: , 10.1097/FCH.0b013e3181e4bcd8.20531097 11. Finn S, Herne M, Castille D. 2017. The value of traditional ecological knowledge for the environmental health sciences and biomedical research. Environ Health Perspect 125 (8 ):085006, PMID: , 10.1289/EHP858.28858824

36515533

PMC9749477

NO-CC CODE

2022-12-16 23:24:03

no

Environ Health Perspect. 2022 Dec 14; 130(12):127701

utf-8

Environ Health Perspect

10.1289/EHP11652

oa_other

==== Front Environ Health Perspect Environ Health Perspect EHP Environmental Health Perspectives 0091-6765 1552-9924 Environmental Health Perspectives EHP12187 10.1289/EHP12187 Invited Perspective Invited Perspective: Tribal Water Issues Exemplified by the Navajo Nation Jones Lindsey 1 Ingram Jani C. 2 1 Water Infrastructure Finance Authority, Phoenix, Arizona, USA 2 Department of Chemistry and Biochemistry, Northern Arizona University, Flagstaff, Arizona, USA Address correspondence to Jani C. Ingram. Email: Jani.Ingram@nau.edu 14 12 2022 12 2022 130 12 12130121 9 2022 09 11 2022 18 11 2022 https://ehp.niehs.nih.gov/about-ehp/license EHP is an open-access journal published with support from the National Institute of Environmental Health Sciences, National Institutes of Health. All content is public domain unless otherwise noted. The authors declare they have no conflicts of interest to disclose. Note to readers with disabilities: EHP strives to ensure that all journal content is accessible to all readers. However, some figures and Supplemental Material published in EHP articles may not conform to 508 standards due to the complexity of the information being presented. If you need assistance accessing journal content, please contact ehpsubmissions@niehs.nih.gov. Our staff will work with you to assess and meet your accessibility needs within 3 working days. Refers to https://doi.org/10.1289/EHP11652 ==== Body pmcMore than 40% of the U.S. public water supply comes from groundwater.1 In the semiarid southwestern United States, where surface water can be scarce and drought conditions make these sources unreliable, groundwater is an important source of drinking water for many people. For example, in northeastern Arizona, groundwater delivers up to 60% of the public water supply.2 In sparsely populated areas where the cost-to-benefit ratio makes it challenging to develop water infrastructure,3 unregulated wells and springs are important water sources.4 However, these sources carry a heavy risk of contamination. Tribal lands are especially affected by the risk of drinking contaminated water, and in many cases testing is absent altogether, as Mok et al. write in this issue of Environmental Health Perspectives.5 The Navajo Nation is a case in point of the water-related challenges faced by Tribes. Nearly half of the 300,000 members of the Navajo Tribe live on the Reservation,6 which spreads over parts of Arizona, New Mexico, and Utah. The Navajo Nation has one of the world’s largest uranium reserves on its lands.7 More than 500 abandoned uranium mine claims have created human health and environmental concerns on Navajo lands,8 and our team reported that many unregulated water sources on the Navajo Nation have elevated levels of arsenic, uranium, manganese, and other elements from former mining operations.9,10 Although mining-related contamination is one of the better-known problems in Navajo water, other drinking water contaminants, such as per- and polyfluoroalkyl substances (PFAS), are understudied on Tribal lands, as Mok et al. point out.5 Municipal wells on the Navajo Nation are regulated by the Navajo Tribal Utility Authority (www.ntua.com). However, the reliance on unregulated water sources on Navajo lands for household use, including drinking water, makes PFAS and other emerging contaminants increasingly important as a potential health risk to the Navajo people, especially because the increase in wildfires, trash burning, and other potential sources of these compounds makes their presence in water sources likely.11,12 The dangers of unregulated water sources are especially relevant to the estimated 70,000 people on the Navajo Nation who lack running water in their homes.13 Those without access to public water systems must haul water from unregulated sources.14 Unregulated wells are prone to contamination issues because they may not be as deep or as well-constructed as municipal wells.15 Also, unregulated wells are not regularly tested for contaminants and often lack water treatment systems.16 Sovereignty is especially important to Native Americans when it comes to Tribal management of drinking water and other resources. Sovereignty provided Tribes with water rights at the time their reservations were established. These rights are based on the 1908 Supreme Court decision in Winters v. United States (the Winters Doctrine), which guaranteed that land established as a reservation would be permanently owned by the Tribe. Further, it guaranteed rights to adequate amounts of water to meet the needs of the reservation. The Winters Doctrine has been essential in negotiations between Tribes and other water users. For example, Tribes have successfully leased their water rights to nontribal entities, which has provided an important source of revenue.17 However, this practice is controversial among stakeholders, including Tribal communities, especially in western states where non-Native water users have historically used Tribal water for free.17 Navajo sovereignty also came into play in 2005 when the Tribe banned uranium mining with the Diné Natural Resources Protection Act (DNRPA) of 2005.18 The DNRPA was enacted to ensure that no further harm will be caused from uranium mining. The act highlights the importance of protecting the natural environment for cultural and spiritual reasons. It further states that a healthy physical and mental environment is the right of all Navajo people. In this way, it aligns with the Fundamental Laws of the Diné, Navajo traditional beliefs that provide the foundation for addressing the complex relationship between nature and humans.20 These Laws, which have played a role in Tribal life for many generations, were codified in 2002 and directed by the Navajo Nation Council (2002) to be incorporated into Tribal self-governance.19 The connections that the DNRPA provides among laws, culture, and environmental protection establishes a framework to create other socially appropriate legislation that not only safeguards resources such as drinking water but also considers Navajo principles and sovereignty.20 Often, environmental policy is developed for issues facing Tribes without input from those Tribes. However, Tribes should have the opportunity to address environmental issues using traditional ecological knowledge in accordance with the Navajo Fundamental Laws. It is essential that these Laws are used today and into the future to provide the basis for Navajo decision making. Acknowledgments The authors would like to acknowledge support from our Navajo partners, financial support from the National Institute of Environmental Health Sciences (Grant P50 ES026089), and editing support from J. Ingram. ==== Refs References 1. Alley WM, Reilly TE, Franke OL. 1999. Sustainability of Ground-Water Resources. U.S. Geological Survey Circular 1186. https://pubs.usgs.gov/circ/circ1186/pdf/circ1186.pdf [accessed 18 November 2022]. 2. Arizona Department of Water Resources. 2009. Arizona Water Atlas: Volume 2–Eastern Plateau Planning Area. Phoenix, AZ: Arizona Department of Water Resources. https://infoshare.azwater.gov/docushare/dsweb/Get/Document-10427/Volume_2_final_web.pdf [accessed 18 November 2022]. 3. Navajo Nation Department of Water Resources. 2011. Draft Water Resource Development Strategy for the Navajo Nation. http://www.frontiernet.net/∼nndwr_wmb/PDF/Reports/DWRReports/DWR2011%20Water%20Resource%20Development%20Strategy%20for%20the%20Navajo%20Nation.pdf [accessed 18 November 2022]. 4. Hoover J, Gonzales M, Shuey C, Barney Y, Lewis J. 2017. Elevated arsenic and uranium concentrations in unregulated water sources on the Navajo Nation, USA. Expo Health 9 (2 ):113–124, PMID: , 10.1007/s12403-016-0226-6.28553666 5. Mok K, Salvatore D, Powers M, Brown P, Poehlein M, Conroy-Ben O, et al. 2022. Federal PFAS testing and tribal public water systems. Environ Health Perspect 130 (12 ):127701, 10.1289/EHP11652. 6. National Congress of American Indians Policy Research Center. 2021. Research Policy Update: 2020 Census Results–NCAI Navajo Region Tribal Land Data. https://www.ncai.org/policy-research-center/research-data/prc-publications/2020_Census_NCAI_Region_Navajo_Summary_9_14_2021_FINAL.pdf [accessed 18 November 2022]. 7. Delemos J, Rock T, Brugge D, Slagowski N, Manning T, Lewis J. 2007. Lessons from the Navajo: assistance with environmental data collection ensures cultural humility and data relevance. Prog Community Health Partnersh 1 (4 ):321–326, PMID: , 10.1353/cpr.2007.0039.19655034 8. Ram NM, Moore C, McTiernan L. 2016. Cleanup options for Navajo abandoned uranium mines. Remediation 26 (3 ):131–148, 10.1002/rem.21473. 9. Credo J, Torkelson J, Rock T, Ingram JC. 2019. Quantification of elemental contaminants in unregulated water across western Navajo Nation. Int J Environ Res Public Health 16 (15 ):2727–4210, PMID: , 10.3390/ijerph16152727.31370179 10. Jones L, Credo J, Parnell R, Ingram JC. 2020. Dissolved uranium and arsenic in unregulated groundwater sources–Western Navajo Nation. J Contemp Water Res Educ 169 (1 ):27–43, PMID: , 10.1111/j.1936-704x.2020.03330.x.34790284 11. Walling M. 2022. Pipeline Fire on the Navajo Nation: A Sacred Mountain on Fire and Smoke in Their Lungs. AZcentral.com, Wildfires section, 15 June 2022. https://www.azcentral.com/story/news/local/arizona-wildfires/2022/06/15/navajo-nation-members-dread-loss-sacred-mountain-pipeline-fire/7641963001/ [accessed 18 November 2022]. 12. Navajo Nation Climate Change Program. 2022. Climate Action for Dinétah! https://www.navajoclimatechange.org/post/air-quality-pollution-adaptation-plan-goals-and-adaptation-strategies [accessed 18 November 2022]. 13. Nania J, Cozzetto K, Eitner M. 2014. Considerations for Climate Change and Variability Adaptation on the Navajo Nation. Boulder, CO: University of Colorado Boulder. http://scholarship.law.colorado.edu/books_reports_studies/3 [accessed 18 November 2022]. 14. Navajo Public Water. 2019. Guidelines for Hauling and Transporting Regulated Water for Human Consumption (Navajo Nation Environmental Protection Agency Public Water Systems Supervision Program). http://www.navajopublicwater.org/Guidelines_Hauling-TransportingRegulatedWater_HumanConsumption_WordDoc.pdf. 15. Johnson TD, Belitz K. 2017. Domestic well locations and populations served in the contiguous U.S.: 1990. Sci Total Environ 607–608 :658–668, PMID: , 10.1016/j.scitotenv.2017.07.018.28709100 16. Malecki KMC, Schultz AA, Severtson DJ, Anderson HA, VanDerslice JA. 2017. Private-well stewardship among a general population based sample of private well-owners. Sci Total Environ 601–602 :1533–1543, PMID: , 10.1016/j.scitotenv.2017.05.284.28605871 17. Williams SM. 1997. Overview of Indian water rights. J Contemp Water Res Educ 107 (1 ):6–8. 18. Navajo Nation Council. 2005. Resolution of the Navajo Nation Council. An Act Relating to Resources, and Diné Fundamental Law; Enacting the Diné Natural Resources Protection Act of 2005; Amending Title 18 of the Navajo Nation Code. https://www.grandcanyontrust.org/sites/default/files/gc_uranium_navajoCouncilResolution.pdf [accessed 18 November 2022]. 19. Navajo Nation Council. 2002. Resolution of the Navajo Nation Council CN-69-02: Amending Title 1 of the Navajo Nation Code to Recognize the Fundamental Laws of the Diné. https://courts.navajo-nsn.gov/Resolutions/CN-69-02Dine.pdf [accessed 18 November 2022]. 20. Rock T, Ingram JC. 2020. Traditional ecological knowledge policy considerations for abandoned uranium mines on Navajo Nation. Hum Biol 92 (1 ):19–26, PMID: , 10.13110/humanbiology.92.1.01.33231023

36515532

PMC9749478

NO-CC CODE

2022-12-16 23:24:04

no

Environ Health Perspect. 2022 Dec 14; 130(12):121301

utf-8

Environ Health Perspect

10.1289/EHP12187

oa_other

==== Front Environ Health Perspect Environ Health Perspect EHP Environmental Health Perspectives 0091-6765 1552-9924 Environmental Health Perspectives EHP10947 10.1289/EHP10947 Research Pollinator Deficits, Food Consumption, and Consequences for Human Health: A Modeling Study https://orcid.org/0000-0001-5207-2370 Smith Matthew R. 1 Mueller Nathaniel D. 2 https://orcid.org/0000-0001-6028-5712 Springmann Marco 3 4 https://orcid.org/0000-0001-7128-5283 Sulser Timothy B. 5 https://orcid.org/0000-0003-0725-4049 Garibaldi Lucas A. 6 7 https://orcid.org/0000-0002-6890-0481 Gerber James 8 https://orcid.org/0000-0001-6035-620X Wiebe Keith 5 https://orcid.org/0000-0002-5808-2454 Myers Samuel S. 1 9 1 Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA 2 Department of Ecosystem Science and Department of Soil and Crop Sciences, Colorado State University, Fort Collins, Colorado, USA 3 Centre on Climate Change and Planetary Health, London School of Hygiene and Tropical Medicine, London, UK 4 Environmental Change Institute and Oxford Martin Programme on the Future of Food, University of Oxford, Oxford, UK 5 Environment and Production Technology Division, International Food Policy Research Institute, Washington, District of Columbia, USA 6 Instituto de Investigaciones en Recursos Naturales, Agroecología y Desarrollo Rural, Universidad Nacional de Río Negro, Miter 630, San Carlos de Bariloche, Río Negro, Argentina 7 Consejo Nacional de Investigaciones Científicas y Técnicas. Instituto de Investigaciones en Recursos Naturales, Agroecología y Desarrollo Rural. Mitre 630, CP 8400, San Carlos de Bariloche, Río Negro, Argentina 8 Institute on the Environment, University of Minnesota, Saint Paul, Minnesota, USA 9 Harvard University Center for the Environment, Cambridge, Massachusetts, USA Address correspondence to Matthew R. Smith, 665 Huntington Ave., Bldg. 1, Room 1301, Boston, MA 02115 USA. Email: msmith@hsph.harvard.edu 14 12 2022 12 2022 130 12 12700314 1 2022 10 11 2022 15 11 2022 https://ehp.niehs.nih.gov/about-ehp/license EHP is an open-access journal published with support from the National Institute of Environmental Health Sciences, National Institutes of Health. All content is public domain unless otherwise noted. Background: Animal pollination supports agricultural production for many healthy foods, such as fruits, vegetables, nuts, and legumes, that provide key nutrients and protect against noncommunicable disease. Today, most crops receive suboptimal pollination because of limited abundance and diversity of pollinating insects. Animal pollinators are currently suffering owing to a host of direct and indirect anthropogenic pressures: land-use change, intensive farming techniques, harmful pesticides, nutritional stress, and climate change, among others. Objectives: We aimed to model the impacts on current global human health from insufficient pollination via diet. Methods: We used a climate zonation approach to estimate current yield gaps for animal-pollinated foods and estimated the proportion of the gap attributable to insufficient pollinators based on existing research. We then simulated closing the “pollinator yield gaps” by eliminating the portion of total yield gaps attributable to insufficient pollination. Next, we used an agriculture–economic model to estimate the impacts of closing the pollinator yield gap on food production, interregional trade, and consumption. Finally, we used a comparative risk assessment to estimate the related changes in dietary risks and mortality by country and globally. In addition, we estimated the lost economic value of crop production for three diverse case-study countries: Honduras, Nepal, and Nigeria. Results: Globally, we calculated that 3%–5% of fruit, vegetable, and nut production is lost due to inadequate pollination, leading to an estimated 427,000 (95% uncertainty interval: 86,000, 691,000) excess deaths annually from lost healthy food consumption and associated diseases. Modeled impacts were unevenly distributed: Lost food production was concentrated in lower-income countries, whereas impacts on food consumption and mortality attributable to insufficient pollination were greater in middle- and high-income countries with higher rates of noncommunicable disease. Furthermore, in our three case-study countries, we calculated the economic value of crop production to be 12%–31% lower than if pollinators were abundant (due to crop production losses of 3%–19%), mainly due to lost fruit and vegetable production. Discussion: According to our analysis, insufficient populations of pollinators were responsible for large present-day burdens of disease through lost healthy food consumption. In addition, we calculated that low-income countries lost significant income and crop yields from pollinator deficits. These results underscore the urgent need to promote pollinator-friendly practices for both human health and agricultural livelihoods. https://doi.org/10.1289/EHP10947 Supplemental Material is available online (https://doi.org/10.1289/EHP10947). The authors declare they have nothing to disclose. Note to readers with disabilities: EHP strives to ensure that all journal content is accessible to all readers. However, some figures and Supplemental Material published in EHP articles may not conform to 508 standards due to the complexity of the information being presented. If you need assistance accessing journal content, please contact ehpsubmissions@niehs.nih.gov. Our staff will work with you to assess and meet your accessibility needs within 3 working days. ==== Body pmcIntroduction Despite large increases in global food production over the past half-century, providing adequate nutrition on a global scale has remained elusive for many populations. Approximately 768 million people are undernourished worldwide, and that number has been growing steadily since 2015, following a decade of decline.1 In addition to those suffering from hunger, 2 billion people globally have been estimated to experience micronutrient deficiencies, although global monitoring data is infrequently collected. The most commonly reported deficiencies are in iron2,3 as well as widespread inadequate zinc,4–6 vitamin A,7,8 and protein for particular population groups.9,10 Meanwhile, populations in many countries are also facing a pandemic of obesity and metabolic diseases from excess caloric intake, with >2 billion adults worldwide being overweight and obese.11,12 Inadequate intake of healthy foods, such as fruits, vegetables, and nuts, is also driving large burdens of disease.13 Considering these persistent challenges, strategies for global food and nutritional security have begun to shift from strictly producing adequate calories to providing more nutritious diets.14,15 Coincident with recognition of the need for more nutritious diets has been a growing awareness that we need to reduce the environmental toll of global food production. Agriculture is the single largest driver of biodiversity loss, land-use change, growing scarcity of freshwater, and land degradation globally.16–19 It is also a significant contributor to climate change, responsible for one-fourth to one-third of global greenhouse gas emissions.20 As such, growing more nutritious foods with lower environmental impact has become one of the great challenges of the 21st century.16,18,19 Pollinators Are Key for Healthy Foods Ensuring an abundance and diversity of pollinators is one effective approach to address the nutritional and environmental challenges facing global food systems. Animal pollination increases the production of three-fourths of agricultural crop varieties21 for several reasons. Pollinators are more efficient at delivering pollen than wind or self-pollination, which increases successful fertilization and improves seed and fruit set (transition of ovule/ovary to fruit/seed), resulting in greater yields. In addition, animal pollinators improve cross-pollination among different plants, thereby increasing genetic diversity by limiting inbreeding. Plants that rely on animal pollination include cash crops (coffee, cocoa, spices) and many food groups important for global health (fruits,22 vegetables,22 nuts,23,24 legumes23) that, when eaten in greater amounts, have been shown in human epidemiological studies to be protective against a range of chronic noncommunicable diseases (NCDs), including heart disease, stroke, many cancers, and diabetes. Moreover, because wild pollinators increase yields without requiring regular external inputs, they can generate significant income for farmers, thereby improving farmers’ livelihoods, with potential downstream implications for their health.25 These benefits are realized without any associated negative environmental impacts. Multiple studies have estimated the contribution of animal pollination to the annual value of global agricultural output at USD $224–577 billion (in 2015 USD).26,27 Animal Pollinators are Under Pressure from Environmental Degradation Yet global pollinators are increasingly in peril, mainly from anthropogenic alteration of their environment, nutrition, and biological networks.28 Wild pollinators, in particular, are under growing threat. Pervasive land-use changes are fracturing, shrinking, and degrading suitable habitats for pollinators worldwide, not only reducing available areas for nesting but limiting pollinators’ ability to migrate as an adaptation strategy in an increasingly disjointed landscape. Furthermore, reductions in wild lands and the dominance of farms growing large monocultures have shrunk the diversity of flowering plants and thereby the duration of flowering, causing nutritional stress. Intensive farming techniques, such as frequent tilling, disturb and destroy nesting sites and disrupt wild plant communities on farms. The ongoing use of pesticides, such as neonicotinoids, have inflicted lethal and sublethal harm to bees both on treated farms and in nearby areas.28 In addition, the overarching impact of climate change is causing a host of deleterious effects: driving pollinators out of their historical range to find suitable new environmental conditions; causing novel predators, competitors, and pathogens to invade newly habitable environments; and increasing the asynchrony between pollinators and their coevolved plant species who may be motivated by different environmental cues.29 Although scarce monitoring data currently limits our ability to definitively link individual drivers to pollinator declines, wherever measured, pollinator communities are decreasing in abundance, range, or diversity.30–32 Managed honeybees, facing sometimes catastrophic hive collapse caused by pest and nutrition pressures, have not been able to compensate for wild pollinator losses nor keep pace with the growth in pollinator-dependent crops that rely on them,33 which makes the use of managed bees an increasingly risky solution to compensate for wild pollinator losses. Furthermore, managed pollinators are not fully interchangeable for wild pollinators,34 and cropping systems with large managed honeybee industries (e.g., blueberries, cherries, apples) could still see additional yield benefit from even greater animal pollination.35 This lack of pollinators is already reducing food production. A 2016 study by Garibaldi et al.36 used a global sample of 344 fields in 33 different pollinator-dependent farming systems across Africa, Asia, Latin America, and Europe to identify the yield penalty currently attributable to insufficient pollination (i.e., the pollinator deficit). To do this, they collected a range of data on farming practices, proximity to natural habitats, and crop yields, as well as pollinator visitation and richness in each location to isolate the role of pollination in supporting yields. They found that, of the yield gap between the low- and high-producing fields across all crop systems, roughly a quarter of the difference could be explained by insufficiently abundant and diverse pollinator populations. However, this previous work36 has not yet been extended to quantify the current burden of lost pollination for food intake, nutrition, incomes, and global health. In this study, we aimed to make an advance over earlier work by applying these empirically derived estimates of lost yields from inadequate pollination with the following motivating research questions: How much additional food would have been produced if global pollination were adequate (hereafter called the pollinator deficit)? Who would have consumed that food, and what health benefits would they have experienced? How many diet-related diseases and deaths could have been averted? For lower-income countries especially, what are the economic costs of insufficient pollination? In this study, we explored the first two questions at a global scale and by country by applying and connecting well-established and vetted analytical tools using the following steps: a) comparing existing global agricultural yields with climate-specific theoretically attainable yields for the 63 most important pollinator-dependent crops to identify the total yield gap for each crop and country, b) using empirical relationships of the percentage of these yield gaps attributable to insufficient pollination36 to quantify the pollinator deficit for each crop and country, c) employing an international economic trade model to identify who would be most likely to have consumed this additional food, and d) using relative risks (RRs) linking dietary risk factors to health outcomes to quantify the implications for global, regional, and country-specific mortality of closing the pollinator yield gap. To evaluate the economic penalty of insufficient pollination, we analyzed, as case studies, three developing countries of different size, geography, and crop specialty—Honduras, Nepal, and Nigeria—to quantify examples of the economic value lost to unrealized agricultural productivity for an individual country. Methods The model framework underpinning our analysis comprised several interconnected modules (Figure 1) that are discussed individually below. “Example: Poland” in the “Results” section contains a representative example of how the model works for a subset of crops. Figure 1. Schematic depicting the chain of modules that constituted our overall model. Arrows show where outputs of one module serve as inputs to another. Module C received inputs from both modules A and B, which connected to module D, then to module E. Note: mod, module. Figure 1 is a schematic flowchart with three steps, namely, Inputs, Modules, and Outputs. Step 1: Inputs: Globally gridded precipitation and temperature characteristics and gridded yields for 63 crops, Modules: A. Total yield gap model. Outputs: Fiftieth and ninetieth percentile yields for each country across 63 animal-pollinated crops and difference between fiftieth and ninetieth percentile is total yield gap. Step 2: Input: Characteristics of 344 fields (for example: Size, management, rainfall), crop characteristics (type, yield, pollinator dependence), and measured pollinator visitation and diversity. Modules: B. Pollinator dependent portion of total yield gap. Outputs: Percentage of yield gap that could be closed if pollination was as on high-yielding farms (pollinator deficit), independent of crop. Step 3: Input: Current total yield gaps: all countries, 63 crops (module A), percentage of total yield gap from pollinator deficit (Module B), and harvested area. Modules: C. From total yield gap model and pollinator dependent portion of total yield gap, there is a missing production from closing pollinator yield gaps. Outputs: Amount of agricultural production lost caused by pollinator deficits. Step 4: Input: Price elasticities of supply and demand for 62 commodity groups; global per capita incomes; starting year data on global prices, production, consumption, and trade; global technological trends; and lost production from pollinator deficits (module C). Modules: D. Economic and trade model to redistribute food. Outputs: Trends in global prices, production, consumption, and trade and per capita amount of available animal-pollinated foods lost due to pollinator deficits by country. Step 5: Inputs: Lost consumption of pollinator-dependent foods accounting for food waste (Module D), current chronic diseases prevalence and mortality rates per country, and relative risks of chronic disease mortality caused by dietary factors. Modules: E. Estimated lost benefits for health. Outputs: Mortality from lost animal-pollinated healthy foods. Yield Gap Analysis Crop yield gaps for 63 pollinator-dependent crops (Excel Table S1) were calculated by subtracting circa 2000 yields37 from circa 2000 climatically attainable yields, which were estimated using a climate zonation approach (Figure 1, module A).38–40 The particular method used here was developed by Mueller et al.40 and is described in the supplemental material therein. Globally, we established 100 zones of equal harvested area and similar climatic properties (i.e., annual precipitation and temperature characteristics derived from WorldClim41) for each crop. Within each zone, we determined an attainable yield defined as the area-weighted 90th percentile yield (i.e., the yield value, which is exceeded by only 10% of area within that zone, with maps of yield and area from Monfreda et al.42) The resulting map of yield ceilings, which has 10-km resolution, has 100 values for each crop, allocated according to the positions of the zones, which are in turn based on the distributions of the maps of crop yield and area from Monfreda et al.42 These attainable yields were then summarized using area-weighted averages for each crop and country. The resulting attainable yield layers represent a ceiling of regionally averaged actual yields, in contrast to agronomic potential yields that may be economically unrealistic. Yield gaps were estimated for all country–crop combinations where those crops were grown. For countries whose yields were currently greater than the 90th percentile attainable yield, the gap was set to zero. Yield gap modeling was performed in MATLAB (version 2018b; MathWorks). These methods are equivalent to the attainable yield modeling carried out by Mueller et al.40 with the exception that the earlier analysis used a 95th percentile cutoff. Closing the Pollinator Yield Gap Next, we used data on the relationship between pollination and total yield gaps36 to calculate the size of the yield gap for each crop attributable to insufficient pollination (Figure 1, module B). The model and data set used to quantify the proportion of the yield gap (i.e., difference between the 50th and 90th percentile yield of each crop and country) attributable to insufficient pollination were taken from an earlier study by Garibaldi et al. from 2016.36 In their study, they collected data from 344 fields from 33 pollinator-dependent crop systems from both small and large farms within a 5-y window from 2010 to 2014. Regions were chosen to focus primarily on developing countries, although the inclusion of many large farms with industrialized practices common in developed countries allows for their results to be applied more broadly across income levels. Crops grown included a diversity of pollinator-dependent food groups (fruits, vegetables, nuts, oilseeds, spices), as well as inedible cash crops (coffee, cotton). Crop systems were chosen specifically to span the range of management styles (conventional agriculture, organic agriculture, traditional practices), field sizes, biotic/abiotic variables, and landscape settings. Data collection variables and methods were uniform at all sites to build a model that identified whether the availability of pollinating insects contributed significantly to the crop yield. The following variables were collected from all sample sites: flower–visitor density, flower–visitor richness, crop yield, index of agricultural intensification (ranging from −3 to 5, with five variables of conventional intensification, each adding 1 to the index—presence of monoculture, synthetic fertilizers, herbicides, pesticides, and fungicides—and three agroecological variables, each subtracting 1 from the index—presence of polyculture, organic certification, and organic fertilizers), latitude, longitude, field size, and isolation from natural or semi-natural habitats. Additional calculated or external variables were also included: percentage pollinator dependence of each crop from Klein et al.,21 baseline level of flower–visitor density (10th percentile: n per 100 flowers), yield gap within each crop system (10th/90th percentile), flower–visitor gap within each crop system (10th/90th percentile), and interaction terms between these variables. All variables were then included as inputs into a general linear mixed-effects model to predict the crop yield. Their model found that several of these variables were significantly predictive of crop yields (sign in parentheses indicates direction of relationship): flower–visitor density (+), flower–visitor richness (+), field size (+), intensification index (+), isolation (in kilometers) from natural habitats (–), and flower–visitor gap percentage (+), as well as several interaction terms: flower–visitor density × field size (–), flower–visitor density × flower–density richness × field size (+), flower–visitor density × isolation (–). Many other model variables were not found to be significant: percentage pollinator dependence of each crop, latitude, longitude, baseline flower–visitor density (10th percentile), yield gap percentage, and several interaction terms. The lack of significance of the crop’s percentage pollinator dependence was particularly surprising given that certain crops are believed to derive much more of their yield from animal pollination, although neither the percent dependence nor its interaction term with crop–visitor density were found to be significant. Two additional findings from Garibaldi et al.36 that are relevant to our present study were that the relationship between flower–visitor density and crop yield was significant for small farms regardless of their flower–visitor richness (number of species per field in the sampling window), whereas density was only significant for larger farms when flower–visitor richness was high (>3 species). This difference is presumed to be caused by the dominance of generalist honeybee species in large fields, given that they are less efficient pollinators than a diverse community of wild insects despite having high foraging ranges.34 Based on these findings, some additional steps were required before applying this previous work to our present study. First, we began with an assumption that our counterfactual high-pollinator scenario would need to include both a high density and richness of pollinating insects. This is because simply having more pollinators was only beneficial for yields on small fields (<2 ha), whereas having both higher density and diversity of pollinators also increased yields in large fields.36 Therefore, we needed to use a subset of the original data to include only crop systems with species richness ≥3 from all field sizes. This subset is found in Excel Table S2. Notably, this subset preserves the diversity of management intensity, pollinator dependence, field size, and crop diversity as in the full original data set, lending confidence to our ability to apply these findings to developed countries where conventional intensely managed agriculture in large fields is more common. We then fit a Gaussian distribution to the pollinator-attributable yield gap percentages from Excel Table S2 and identified the mean along with its 95% confidence interval (CI) (25.5%; 95% CI: 5.5%, 45.4%) for use in our model. Of note are two data points from Excel Table S2 that were anomalously high or low: −37% for agraz (native blueberry) in Colombia and 121% for sunflower in South Africa. All Excel Table S2 values were generated using the best mixed-effects model from Garibaldi et al.,36 although these particular data points had very high or low values for some inputs, causing their irregularity: Colombian agraz was not at all isolated from natural areas (0km) and South African sunflower had very high species richness given its large field size—eight species per 30-min sampling window. We preserved these extreme values to maintain congruency with the previous study although removing them would have only slightly altered our yield gap percentage for later steps (22.9%; 95% CI: 10.5%, 35.3%). Percentage changes in production caused by closing the pollinator deficits were then multiplied by current production values to estimate the additional food that would be produced under scenarios of enhanced pollinator density and richness (Figure 1, module C). The percentage increase in production was capped at 100% to preclude unrealistically high modeled yield gains from greater pollination; country–crop combinations with yield increases >100% constituted a small percentage of all data points (∼1%). Modeling of percentage production changes under high pollination scenarios was performed in MATLAB (version 2018a; MathWorks). One caveat to our model is that we did not account for potential losses to agricultural productivity that might be necessary to achieve the adequate pollinator levels in our high-pollinator scenario. Many practices can boost on-farm pollinator populations without harming productivity, such as the installation of bordering hedgerows or the use of adjacent marginal lands for habitat. Nevertheless, where nearby pollinator habitat is not available, setting aside farmland to serve as undisturbed habitat may be necessary. Doing so would reduce the productive land for crops and could lessen our projected yield gains. Trade and Economic Model Changes in a country’s crop production do not directly translate to equivalent changes in domestic consumption. That is because the place where a food will be eaten, regardless of where it is grown, is dictated by its price (itself governed by supply and demand for that food) and consumers’ ability and willingness to pay for it. Furthermore, farmers change their own behavior—how much area to devote to which crop, the amount of effort to expend—based on crop prices in a given season, which feeds back to influence a crop’s supply, price, and demand. To capture the complexity of these market forces that intercede between production and consumption, we relied on the International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT), version 3 (Figure 1, module D).43 Fuller descriptions of the IMPACT model and its core elements are available,43–45 but a summary is given here, and a schematic of IMPACT, its components, and how they intersect, in Figure S1. IMPACT is a global, partial equilibrium multimarket model of the food system that links pixel-scale modeling of climate, hydrology, and crops to national level supply and demand and further to global-scale international markets through trade. IMPACT simulates national and global agricultural market behavior annually.43 Starting with FAOSTAT data37 that have been balanced globally using a Bayesian maximum entropy algorithm to reconcile imbalanced FAO-reported production and consumption data, IMPACT computes annual solutions that balance global agricultural supply and demand with prices that clear international trade markets such that global net trade equals zero. Food, feed, and industrial demand are determined through a combination of endogenous and exogenous drivers (prices and population/income growth, respectively) with market behavior determined by income and price elasticities derived from the literature and expert judgment. Therefore, any increases in crop production are distributed among consumers (either domestically or in other countries via trade) using a combination of model-generated food price changes, as well as per capita incomes and price elasticities, which govern a consumer’s willingness to purchase a certain quantity of each food. The supply side is, likewise, simultaneously determined through a combination of endogenous and exogenous parameters. Producers react to price changes to make within-year adjustments to production systems that are defined by assumptions on technological potential and trends determined through analysis of historical data and expert judgment about likely structural change within the agricultural sector.46,47 Productivity impacts of levels of theoretically replete pollinator populations were translated into yield increases according to the food-group mapping found in Excel Table S1. In IMPACT, two types of scenarios were run: a) baseline scenarios, where production, trade, and consumption patterns proceeded as usual; and b) an alternative set, where production was increased assuming higher levels of pollination (i.e., a reduction in the pollinator deficit). Three levels of high-pollinator scenarios were run: a median case where production was increased assuming a median closure of the pollinator yield gap, as well as low and high 95% confidence-bound scenarios. Levels of yield increases for pollinated crops at aggregate world regions are detailed in Table S1. In our higher-pollination scenarios, pollinated-yield boosts were introduced in the year 2010 and allowed to come into equilibrium until the comparison year 2020. The economic equilibrium modeling process includes simultaneous farmer and consumer responses to the changes in production and the ensuing international commodity prices, encompassing changes in farming intensity, crop choice, and farmed area to account for farmers’ varying profit incentives; changing consumer diets after accounting for food prices and their elasticities of demand; and shifting trade markets to rebalance the flow of food from more or less productive regions. IMPACT is anchored to empirically derived data in year 2005 (an average of 2004–2006) and calibrated to available data up to 2012 using smoothed 3-y averages, after which it is then based on assumptions that change in 5-y increments. This multiyear smoothing limits the model’s ability to account for volatile annual changes to crop production or price (e.g., the 2007–2008 global food price spike and crisis) but, instead, is designed to capture the behavior of the agricultural and food system in response to long-run trends, such as population and income growth, or the influence of sustained global temperature increases. After 2012, simulations were run to 2020 using five global climate models [National Oceanic and Atmospheric Administration’s Geophysical Fluid Dynamic Laboratory model (GFDL), Hadley Centre’s Global Environment Model (HGEM), Institut Pierre Simon Laplace’s Earth System Model (IPSL), Model for Interdisciplinary Research on Climate (MIRO), Norwegian Earth System Model (NORE)] under a moderate radiative forcing model [Representative Concentration Pathway (RCP) 4.5]48 and a moderate shared socioeconomic pathway scenario (SSP2).49 The various SSPs do not diverge significantly until after 2020, and SSP2 was used solely to bridge 2010 to 2020 by accounting for changes in population and gross domestic product (GDP), as well as elasticities calibrated to a business-as-usual future. This allowed for the model to resolve changes in global food production, consumption, and trade that flow from these demographic and economic drivers. The average and standard deviation from among the five global climate models were used for each of the four scenarios—namely, the baseline and three high-pollinator scenarios—in subsequent health modeling. Nut consumption, unlike other food categories, required additional steps before being used as an input to health models. Groundnuts were explicitly reported in IMPACT, whereas tree nuts were included in a broader IMPACT category of “other crops” that included other pollinator-benefitting crops (cloves and “spices, other”), and nonpollinated or inedible crops (e.g., jute, tobacco, rubber). To estimate the food availability of tree nuts, we used per capita estimates from the FAO food balance sheet estimates37 as our baseline values. We then calculated the difference between current and high-pollinator production and consumption values for the “other crops” category in each country from IMPACT. Separately, using FAO data, we estimated the percentage that tree nuts constituted in the production and consumption of pollination-benefitting “other crops” in each country. For example, if Country X consumes 9g/d of tree nuts, 1g/d of cloves, and 5g/d of “spices, other,” we would estimate that tree nuts account for 60% of the consumption of pollination-benefitting “other crops”; a similar calculation was also made for production in each country. We then multiplied these percentages by the total absolute change in production and consumption of “other crops” under high-pollinator scenarios from IMPACT to approximate the pollinator-attributable change in tree nut production and consumption. Tree nut production and consumption values were added to groundnut values and used as inputs in estimating the health implications. Health Outcomes We used a global risk–disease model focused on dietary and weight-related risk factors to quantify what impacts the changes in pollination could have on mortality in each country (Figure 1, module E). The model is based on a comparative risk assessment framework with eight risk factors and five disease endpoints.50 In comparative risk assessments, one compares different levels of risk exposure—for example, a situation where lead levels in municipal water are high compared with a counterfactual case where lead in the water is zero—and calculates the consequences of that relative level of risk in terms of health outcomes, using an empirically established relationship between risk and outcome. Risks can be binary (e.g., lead exposure exceeding a certain threshold) or continuous (e.g., functions equating higher lead levels to increasingly severe cognitive or developmental consequences). Based on several variables—an exposure variable (e.g., lead concentrations), the RR of a health outcome given a certain risk exposure (e.g., empirical relationship between lead levels and developmental impairment), and the current occurrence of health outcomes under consideration (e.g., prevalence of developmental impairment in the community)—one can then compute what amount of existing health burden is attributable to the current risk exposure. In our study, the risk factors included high consumption of red meat and low consumption of fruits, vegetables, nuts, and legumes, as well as being underweight [body mass index (BMI)<18.5], overweight (25<BMI<30), and obese (BMI>30). The disease endpoints included coronary heart disease, stroke, type-2 diabetes mellitus, cancer (in aggregate and as site-specific ones, such as colon and rectum cancers), and an aggregate endpoint of all-cause mortality associated with changes in weight.50 Further information on RRs associated with each of these endpoints is described below. A table showing the groupings of individual foods into health-related food groups is found in Excel Table S1 (“other crops,” “palm crop,” and “other grains,” also included in Excel Table S1, were not included in the health modeling). For specifying the exposure levels of the dietary and weight-related risk factors, we used and adjusted the food availability estimates from IMPACT. For the dietary risk assessment, we used regional data on food wastage at the consumption level (Excel Table S3), combined with conversion factors into edible matter51 to convert food availability estimates into proxies for food consumption. For reference, consumption values of major food groups after removing waste are listed in Excel Table S4. For the weight-related risk assessment, we used current weight levels and the historical relationship between food availability and BMI to estimate changes in weight levels in the different scenarios. The mortality and disease burden attributable to dietary and weight-related risk factors were then estimated by calculating population impact fractions (PIFs) and applying those to age and country-specific mortality rates.52–54 PIFs represent the proportions of disease cases that would be avoided when the risk exposure was changed from a baseline situation (the current diet) to a counterfactual situation (the pollination scenarios). For specifying the exposure levels of the dietary and weight-related risk factors, we used and adjusted the food availability estimates from IMPACT. RR estimates that relate the risk factors to the disease endpoints were adopted from meta-analyses of prospective cohort studies for dietary risks22–24,55–59 and pooled cohort studies for weight-related risks.52,60 RRs used in our calculations are provided in Excel Table S5. In line with the meta-analyses, we included nonlinear dose–response relationships for fruits and vegetables22 and nuts24 and assumed linear dose–response relationships for the remaining risk factors.23,57,58 Because our analysis was primarily focused on mortality from chronic diseases, we focused on adults ≥20 years of age, and we adjusted the RR estimates for attenuation with age based on a pooled analysis of cohort studies focused on metabolic risk factors53 in line with other assessments.54,61 In addition to changes in total mortality, we also calculated years of life lost. Health modeling was performed in GAMS (version 38; GAMS). Sources and Treatment of Uncertainty in the Global Model Three main sources of uncertainty were propagated through the model to capture the full range of possible outcomes. The first and largest quantifiable source was introduced by the relationship between insufficient pollination and crop yields. For this, we used the empirically derived distribution of possible pollinator yield gaps (Excel Table S2) and, described above, as inputs into a Monte Carlo simulation (N=1,000) to quantify the possible percentage loss of each pollinated food caused by insufficient pollination. A randomly chosen pollinator-attributable yield gap percentage was chosen for each iteration, the total yield gap for that food and country was closed by that amount, and a percentage change in food production was calculated. We then identified the median, 2.5th percentile, and 97.5th percentile values from among the 1,000 Monte Carlo runs for each food and country. Additional uncertainty was also introduced from the global trade and health modeling. As mentioned previously, IMPACT used multiple climate and socioeconomic models to bridge the gap between the base year in IMPACT (2010) and our analysis year (2020). Because of the complexity and time required to run IMPACT, it was not feasible to use all Monte Carlo model output from module C (Figure 1) to generate 1,000 IMPACT model (module D) outputs. Instead, we used three high-pollinator scenario outputs from among the 1,000 outputs from module C: median, 2.5th percentile (“low pollinator”) and 97.5th percentile (“high pollinator”). Therefore, IMPACT was run four times (benchmark plus three high-pollinator scenarios) and the average and standard deviation of the output (driven by the various general circulation models) was calculated. Finally, the outputs of these four scenarios (plus standard deviations) were passed on to be used as inputs in health modeling, module E. Here, additional uncertainty was introduced from the RR parameters of each food–outcome pair. Using averages and 95% CIs from previously reported RRs, we ran three comparisons between each high-pollinator scenario and the benchmark to estimate the change in mortality that would be caused by changes in diet. We used 95% RR CIs associated with RR estimates using error propagation methods. Final median values in Tables 1 and 2 reflect the average of the “average” scenario, whereas the 95% uncertainty interval (UI) reflects the upper bound of the high-pollinator scenario and the lower bound of the low-pollinator scenario. Table 1 Modeled lost production and consumption (95% UIs) of healthy food groups due to insufficient pollination. Country groups Lost food due to insufficient pollination (percentage change relative to high-pollinator scenarios) Production Consumption Fruit Vegetables Nuts Fruit Vegetables Nuts Income group  Low −4.7 (−8.2, −0.5) −26.0 (−32.5, −16.5) −8.4 (−14.0, −1.9) −1.5 (−2.3, −0.9) −3.1 (−4.0, −2.1) −6.5 (−8.9, −3.5)  Lower middle −10.3 (−12.5, −5.3) −13.0 (−18.2, −5.4) −1.8 (−3.8, 0.6) −4.0 (−5.4, −2.0) −3.3 (−4.9, −1.3) −5.4 (−7.0, −3.2)  Upper middle −4.5 (−6.8, −2.0) 1.6 (−3.4, 4.9) −5.3 (−8.8, 2.1) −4.9 (−6.7, −2.1) −3.0 (−4.7, −0.9) −3.7 (−5.3, −1.5)  High 5.3 (1.9, 8.6) 1.2 (−2.1, 4.9) 7.1 (−4.6, 16.4) −4.8 (−6.3, −2.9) −3.1 (−4.2, −1.8) −11.9 (−15.7, −6.7) Region  East Asia and Pacific −4.2 (−7.4, −1.3) 4.6 (−0.9, 7.7) −5.4 (−8.7, 1.4) −5.7 (−8.0, −2.1) −3.0 (−4.8, −0.6) −3.6 (−5.2, −1.3)  Europe and Central Asia −5.3 (−8.7, −0.5) −11.1 (−16.7, −4.7) −0.8 (−18.2, 15.4) −4.8 (−6.1, −3.3) −3.0 (−3.9, −2.1) −11.2 (−14.2, −7.8)  Latin America −2.3 (−4.3, −0.2) −5.3 (−10.7, 0.0) 0.0 (−9.6, 11.0) −2.4 (−3.7, −1.1) −3.1 (−4.7, −1.4) −10.6 (−15.6, −4.6)  Middle East/N. Africa −6.7 (−9.6, −2.9) −8.9 (−15.4, −1.7) −5.6 (−20.7, 17.1) −5.2 (−7.4, −2.7) −2.1 (−3.0, −1.3) −4.4 (−5.9, −2.6)  North America 11.8 (1.7, 19.2) 7.5 (1.1, 12.5) 12.4 (0.6, 19.2) −5.7 (−8.6, −1.1) −4.4 (−7.0, −0.7) −11.3 (−16.9, −2.4)  South Asia −13.4 (−16.1, −3.8) −1.9 (−8.2, 5.0) 0.1 (−2.5, 3.1) −4.5 (−6.5, −1.1) −3.9 (−6.1, −0.7) −3.9 (−5.7, −0.6)  Sub-Saharan Africa −1.7 (−3.9, 0.4) −36.1 (−41.6, −19.8) −6.9 (−10.4, −3.0) −1.3 (−1.9, −0.6) −3.4 (−4.5, −2.1) −7.0 (−9.2, −4.6)  World −4.7 (−7.1, −0.8) −3.2 (−5.3, −0.4) −4.7 (−6.9, −0.5) −4.6 (−7.0, −0.8) −3.2 (−5.3, −0.4) −6.1 (−9.5, −0.8) Note: Positive values in production indicate where production is currently higher than it would be if pollination were sufficient given that they would likely import these foods rather than grow them domestically. Definitions for each food category are found in Excel Table S1. Supporting data on the population and current production tonnage by food group for each income group and region are found in Table S2. Income and region classifications are from the World Bank.62 Table 2 Excess mortality (95% UIs) attributable to insufficient pollination by risk factor. Country groups Current excess annual deaths attributable to insufficient pollination, total and by dietary risk factor (thousands) Current deaths from insufficient pollination as percentage of total mortality Fruit Vegetables Nuts Other risk factors Total Income group  Low 2 (0, 4) 2 (0, 3) 2 (0, 3) 2 (0, 4) 9 (2, 15) 0.3 (0.1, 0.5)  Lower middle 44 (9, 71) 43 (8, 72) 18 (4, 26) 6 (1, 9) 110 (22, 179) 0.6 (0.1, 1.0)  Upper middle 111 (21, 180) 78 (16, 129) 32 (7, 50) −12 (−20, −2) 208 (41, 338) 1.0 (0.2, 1.6)  High 31 (6, 50) 28 (6, 46) 47 (10, 72) −5 (−8, −1) 101 (21, 159) 1.0 (0.2, 1.5) Region  East Asia and Pacific 98 (19, 160) 59 (12, 97) 21 (4, 32) −6 (−9, −1) 171 (34, 278) 1.0 (0.2, 1.7)  Europe and Central Asia 37 (7, 58) 35 (7, 59) 42 (9, 65) −7 (−12, −1) 107 (22, 169) 1.2 (0.2, 1.9)  Latin America and Caribbean 8 (1, 13) 5 (1, 9) 4 (1, 6) −2 (−4, −0) 14 (3, 25) 0.4 (0.1, 0.7)  Middle East and North Africa 10 (2, 16) 5 (1, 9) 4 (1, 5) −2 (−4, −0) 16 (3, 27) 0.8 (0.2, 1.3)  North America 9 (2, 15) 12 (2, 19) 16 (3, 26) −1 (−2, −0) 36 (7, 57) 1.1 (0.2, 1.8)  South Asia 23 (5, 37) 30 (6, 52) 8 (2, 10) 6 (1, 10) 67 (13, 109) 0.6 (0.1, 1.0)  Sub-Saharan Africa 4 (1, 6) 4 (1, 6) 4 (1, 7) 4 (1, 6) 16 (3, 26) 0.3 (0.1, 0.5)  World 189 (37, 305) 151 (31, 251) 99 (21, 151) −9 (−14, −2) 427 (86, 691) 0.8 (0.2, 1.3) Note: “Other risk factors” include changes in red meat and legume consumption, as well as changes in overweight, underweight, and obese populations. Negative values indicate current risk-attributable deaths that may increase under higher-pollination scenarios, such as those caused by a greater prevalence of overweight and obese populations and increases in red meat consumption. It is worth mentioning that not all model components had associated uncertainties, and our final estimates do not reflect a true encapsulation of all potential outcomes. Specifically, most IMPACT model (module D) inputs—such as price elasticities, intrinsic productivity growth rates, and farmer planting/effort choices—did not have quantifiable uncertainties attached. Likewise, our global yield gap measurements (module A) were not capable of reporting uncertainties. Therefore, our reported uncertainties are likely too narrow. Economic Analysis for Case-Study Countries To quantify the economic penalty of insufficient pollination, we chose three case-study countries—Honduras, Nepal, and Nigeria—to identify the implications for the lost economic value of agricultural production due to inadequate pollination in an individual country. Case-study countries were chosen based on the following criteria: low or lower-middle income; diverse population size, economy, agricultural system, and geography; being economically reliant on pollination-dependent crops (including cash crops); and a stated interest or strategy to protect pollinators. An exception to these criteria was that Honduras did not have an established plan nor official interest to protect pollinators. However, very few Latin American countries have officially prioritized protecting pollinators, and Honduras’s reliance on coffee production for their national economy, coupled with increasingly narrow profit margins driven by escalating climate change, suggested that identifying opportunities to increase yields that did not further degrade the environment may be salient. Finally, they were chosen to be illustrative of more typical response behavior for different categories of lower-income regions globally, instead of outliers to represent the upper bound of potential losses. For these countries, we performed the counterfactual high-pollinator scenario only in each country of interest individually, and all other countries were parameterized as in the baseline scenario. As in the production and consumption modules, global markets were allowed to come into equilibrium between the production boost in 2010 and the comparison year 2020. To give an overall indication of the economic impact associated with the high-pollinator scenario, the economic value of production was calculated as the product of the estimated international price of each commodity multiplied by its total production. This metric gives a conservative indication of the benefits of increased pollinator populations with a focus on the supply side while consumer-side benefits, although likely positive but more complicated to estimate in this limited modeling exercise, are left aside. Results Globally, we estimated that the world is currently losing 4.7% (0.8%, 7.1%) of total production of fruit, 3.2% (0.4%, 5.3%) of vegetables, and 4.7% (0.5%, 6.9%) of nuts due to insufficient pollination (Table 1). All parenthetical ranges in Results indicate 95% UIs. Had these foods been produced, distributed through the global food trade system, and consumed (assuming current percentage rates of food loss and waste), we estimated that 427,000 (86,000, 691,000) excess global annual deaths, mostly from chronic NCDs, would have been averted (Table 2). Our model showed that lost food production was greatest in lower-income countries, primarily because these countries had the largest yield gaps based on our estimates and would experience greater absolute yield increases from adequate pollination than countries with smaller overall yield gaps (Table 1). Importantly, this result was underpinned by the earlier finding that the contribution of insufficient pollinators to a farm’s overall yield gap was independent of its geography, degree of agricultural intensification, and several other agronomic and landscape characteristics.36 In some areas, pollinator deficits in our models were found to be substantial; an estimated 26% (17%, 33%) of vegetable production and 8% (2%, 14%) of nut production in low-income countries was estimated lost due to inadequate pollination, as well as 10% (5%, 13%) of fruit production and 13% (5%, 18%) of vegetable production in lower–middle-income countries. These modeled production losses, when mediated by the global trade system, led to decreases in fruit and vegetable consumption that ranged from 2% to 5% compared with high-pollinator scenarios, and from 4% to 12% less nut consumption. On average, our model estimated that trade would have transferred production from lower-income countries to higher-income countries. This was especially evident when looking across regions (Table 1), where, for example, we saw that North America consistently experienced the greatest reductions in consumption (in percentage terms) across all food categories. Meanwhile, sub-Saharan Africa saw much more modest impacts on estimated consumption of fruits and vegetables. Calculated Mortality Burden Resulting from Lost Pollination Our model-generated results of reduced production and consumption of pollinator-dependent crops drove large mortality burdens (Table 2). Pollinator deficits were estimated to be responsible for 1% of total annual mortality in both upper–middle- and high-income countries. Globally, decreased fruit and vegetable intakes accounted for the highest amount of increased mortality, 189,000 (37,000, 305,000) and 151,000 (31,000, 251,000) deaths respectively, primarily due to stroke, coronary heart disease, and cancer (Figure 2). Low nut intake also contributed an estimated 99,000 deaths annually (21,000–151,000) Other minor beneficial factors under high-pollinator scenarios (higher legume consumption, fewer underweight) as well as detrimental factors (higher obese and overweight, higher red meat consumption), reduced our final mortality estimates relative to a high-pollinator scenario by 9,000 (2,000, 14,000) annual deaths, or ∼2%. For total avoidable deaths, the largest number was found in upper–middle-income countries and the lowest number in low-income countries. This was due in part to the very large populations in upper–middle-income countries (3 billion people; including China, Indonesia, and Brazil) compared with low-income countries (675 million people), coupled with the higher baseline rates of chronic diseases in upper–middle-income countries that could have been ameliorated by eating more healthy foods. Figure 2. Current annual mortality estimated to be attributable to inadequate pollination and its dietary effects, by cause of death and country income. Inadequate pollination is defined as a combination of too-little flower visitation and scant pollinator diversity to achieve optimal yields. More information of “other risk factors” and causes of negative values may be found in the caption for Table 2. Source data may be found in Excel Table S6. Figure 2 is a stacked bar graph titled Current mortality attributable to insufficient pollination by cause of death, plotting thousands of deaths, ranging from negative 50 to 250 in increments of 50 (y-axis) across Income level, ranging as low, low middle, upper middle, and high (x-axis) for other, cancer, stroke, coronary heart disease, and type 2 diabetes. Our modeled health effects of inadequate pollination were not evenly distributed. Figure 3 and Table 2 show the regional distribution of the health burden from the pollinator deficit. Areas with particularly high health burdens included China, India, Central Asia, Eastern Europe, and Russia and parts of Southeast Asia and North Africa. These regions shared high baseline prevalence rates of underlying dietary-affected NCDs, as well a greater loss of the protective effect from consumption of pollinator-dependent foods. Much of Southern and Eastern Africa, Latin America, Western Europe, and Australia would have seen relatively little difference in mortality under higher-pollination scenarios compared with the present day. Figure 3. Life-years lost per capita estimated to be attributable to insufficient pollination. Insufficient-pollination–related health conditions include dietary and weight factors. Values represent median of model runs. Source data may be found in Excel Table S7. Map outline sourced from https://thematicmapping.org/downloads/world_borders.php. Figure 3 depicts a world map of insufficient-pollination–related health conditions, including dietary and weight factors. A color scale that depicts the current life-years lost attributable to insufficient pollination (number per million people) ranges from 100 to 2,500 in increments of 300, 2,500 to 4,000 in increments of 1,500, and 4,000 to 6,000 in increments of 2,000. Calculated Economic Losses Resulting from Lost Pollination When comparing current and modeled yields assuming replete pollination, the annual lost economic value across all agricultural commodities—represented here as a commodity’s annual production quantity multiplied by its international price—for our three case-study countries amounted to −12% (−3%, −19%) in Honduras, −17% (−5%, −22%) in Nigeria, and −31% (−13%, −32%) in Nepal (Table S3). These economic losses were attributable to crop production losses of −3% (−1%, −5%) in Honduras, −15% (−5%, −18%) in Nigeria, and −19% (−7%, −20%) in Nepal. The greater percentage economic loss compared with production loss (by weight) suggests that pollinated crops constituted high-value commodities for these countries. Dividing modeled economic losses (Table S3) by the 2020 population related to the agricultural sector62,63 amounted to an annual lost value per person in Honduras of USD $209 ($42, $363), USD $250 ($83, $264) for Nepal, and USD $325 ($81, $442) for Nigeria (all in 2005 USD). Although lost agricultural value is only a rough proxy for lost income, for reference the total 2019 Nepalese agricultural GDP divided by the population employed in agriculture was USD $326 (all in constant 2010 USD). Honduras’s 2019 agricultural per capita GDP is USD $799 and Nigeria’s is USD $1,486, also suggesting a considerable potential loss to incomes. Because such a large share of the population is employed in agriculture in these countries—30% in Honduras, 35% in Nigeria, and 64% in Nepal62—this effect could be substantial for these countries and for other agriculture-dependent nations globally. In all countries, the lost value of production was dominated by fruits and vegetables: 84% (84%, 85%) of total lost production value in Honduras, 84% (63%, 86%) in Nigeria, and 100% (99%, 100%) in Nepal (Table S3). The importance of these food categories in economic terms reflected their current dominance in each country: In Honduras and Nepal, fruits and vegetables are currently the highest-value agricultural commodity category, and they are the second-most valuable commodity in Nigeria behind roots and tubers. Following fruits and vegetables, several additional crops were estimated to contribute to significant economic losses: Honduras lost most value from pulses (10% of total loss; 11%, 11%) and “other crops” (7% of total loss; 6%, 7%), primarily coffee. Nigeria lost significant value from underperforming vegetable oil (8% of total loss; 7%, 26%) and oil crop (7%; 5%, 10%) categories, led by palm fruit and oil, as well as smaller losses among “other crops” (2%; 2%, 3%), mainly cocoa. Nepal was estimated to lose nearly all its value from fruits and vegetables, but these estimates have large accompanying error bars due to the poor specificity of its reporting of specific fruit and vegetable production. Example: Poland Here we expand our analysis in a single country in detail to help demonstrate how the models operate individually and together. Unless otherwise noted, all data is derived from sources listed in the “Methods” section above. Furthermore, because values are presented to inform the reader of how our individual models interconnect rather than to highlight Poland’s results per se, all numbers represent median values of our modeled uncertainties for simplicity. Three-quarters of Poland’s animal-pollinated crop production, excluding oilseeds, comes from three crops: apples (53%), cucumbers (11%), and tomatoes (10%).37 To understand Poland’s missing potential due to insufficient pollination, we first compared Poland’s average crop yields with its potential yield based on a selection of yields globally grown on cropland that has a similar climate to Poland’s. Poland’s average reported yield1 for apple is 12.1 metric tons (t)/ha compared with an attainable yield of 15.8 t/ha; cucumbers, 11.5 t/ha on average vs. 18.9 t/ha attainable; and tomatoes, 15.5 t/ha on average vs. 37.8 t/ha attainable. Of the gap between the average and estimated climatically attainable yields, we relied on robust field-based empirical work to assume that roughly a quarter of the difference is caused by insufficient pollination after controlling for other potential variables (e.g., industrialized farming techniques, input use, irrigation, soil properties, proximity to natural areas) (for details, see “Closing the pollinator yield gap” the “Methods” section above). From this, we estimated that if pollinators were abundant and diverse, Poland could produce 8% more apples, 12% more cucumbers, and 28% more tomatoes. Widening the lens to both animal- and nonpollinated crops, we estimate that, under greater pollination, Poland could produce 13% more fruit and 3% more vegetables than at present. The larger value for fruit reflects that animal-pollinated vegetables make up a smaller proportion of total production based on our calculations. If Poland and the rest of the world were producing greater volumes of fruit and vegetables (among other crops) in a more pollinated world, economic principles that underpin our IMPACT global economic model would predict that farmers would follow new price incentives by changing what they plant or the effort they expend. Global consumers would also buy and eat differently based on food prices, accommodated by shifts in global trade flows. These local and global forces combine in our modeled results for Poland by driving up exports of surplus fruit to meet new demand elsewhere, increasing imports of now less expensive vegetables to meet domestic demand, and increasing domestic consumption of fruits by 6% and vegetables by 4%. Interestingly, nut intake is estimated to increase from 4.2 to 5.0g/d, driven entirely by increased imports caused by higher supply and lower prices (very few nuts are grown domestically). Our model suggests that these relatively modest diet changes nevertheless would have the benefit of reducing avoidable mortality from chronic disease. Higher fruit intake could be estimated to help avoid 1,400 deaths annually in Poland due to decreased risk of stroke (900 deaths avoided), cancer (300 deaths), and coronary heart disease (200 deaths). Furthermore, higher vegetable intake could lead to reduced mortality from coronary heart disease (1,000 deaths), cancer (500 deaths), and stroke (200 deaths). Higher nut intake could help avoid 1,700 deaths, all from coronary heart disease. All together, these beneficial changes to the diet could be estimated to avoid 4,700 deaths per year. Discussion Our results suggest that suboptimal pollination appears to be already driving significant excess mortality globally and loss of economic value in producing regions. Furthermore, they suggest that it is also likely widening inequality in diets and health outcomes given that a reduced supply of pollinated foods would raise prices and narrow access within and across countries. Today’s estimated health impacts of insufficient pollination would be comparable to other major global risk factors: those attributable to substance use disorders, interpersonal violence, or prostate cancer.64 We found that in percentage terms, this health burden was estimated to be borne disproportionately by upper–middle- and high-income countries, and much of the absolute burden was estimated to be suffered in middle-income countries with large populations, namely, China, India, Indonesia, and Russia. In addition, our analysis showed that the lower-income countries we examined could also be losing considerable agricultural income from depressed yields, potentially on the order of 10%–30% of total agricultural production value. It is worth noting that our estimates of the health impacts of global pollinator deficits are likely to be conservative. In this analysis, we focused on a single pathway: the impact of lost pollinator-dependent crop production and consumption on deaths from NCDs. However, the loss of pollinator-dependent crops is likely to impact health in other important ways not addressed by our analysis. One way is increased prevalence of micronutrient deficiency, particularly for vitamin A and folate. Although falling globally, there are still substantial global burdens of disease from these deficiencies,8 and pollinator-dependent crops are responsible for a large share of these nutrients in the global diet.65 Another pathway is the indirect effect of reduced income among farmers in low and lower–middle-income countries. Presumably, the higher incomes associated with higher per capita crop yields in producing regions would translate into health benefits, particularly for lower-income countries. Finally, other health opportunities, such as reduced access to health-benefitting bee products (e.g., honey, propolis, royal jelly) and pollinated medicinal plants that are important in both industrialized and traditional medicines, may be lost. Analysis of these effects, however, was beyond the scope of this paper. This study represents a unique, cross-disciplinary combination of data and modeling to quantify the health implications of inadequate pollination on a global scale. A previous analysis66 investigated more extreme theoretical future scenarios of 50%, 75%, and 100% removal of global pollinators and their implications on diets and health, finding a predictably more severe impact on agricultural production and health. With 100% removal of pollinators, they found that supplies of fruit, vegetables, nuts, and seeds could fall by 16%–23%, leading to 1.4 million additional annual deaths globally due to the ensuing dietary and nutritional changes. However, unlike this prior study, which examined the implications for extreme hypothetical scenarios of severe or complete pollinator loss, the present analysis aimed to quantify the present-day penalty being paid by inadequate global pollination compared with our achievable potential. As such, it serves to inform and target strategies aimed at boosting pollinating insect populations by quantifying the potential health and economic benefits of adopting such policies. We provide four accompanying notes to explain and justify our results. First, our data sets for both the total yield gap of each crop and country, as well as the pollinator-attributable yield gap,36 are derived from recorded or empirical measurements, making them congruent and applicable to our research question. Our second note is a justification for the use of a single percentage—25.5% (95% UI: 5.5%, 45.4%)—to characterize the pollinator-attributable yield gap for all pollinator-dependent crops globally. This percentage was identified from an empirically derived regression model after controlling for many other confounding variables: location (latitude and longitude), management intensity, isolation from semi-natural or natural habitats, the estimated percentage pollinator dependence of each crop from Klein et al.,21 baseline floral density, and the size of the total yield gap. Some of these variables were also found to be significantly correlated with yields—such as management intensity, isolation from natural habitats—whereas most were not, most importantly, the crop’s estimated pollinator dependence percentage. This last finding can also make intuitive sense given that farmers of highly pollinator-dependent crops may recognize the necessity of pollinators and therefore manage them, whereas farmers of low-to-intermediate pollinator-dependent crops may not cultivate pollinators so aggressively. In this case, a pumpkin farmer (high pollinator-dependent crop) near a peanut farmer (low pollinator-dependent crop) may have similar pollinator deficits despite very different crop pollinator dependence. Because the pollinator yield relationship was found to be independently significant after controlling for a large slate of predictor variables, we believe it is a robust empirical finding that we could confidently apply broadly. Third, we address how our model results may underestimate the true effect. Given that many agricultural inputs will continue to be optimized globally over time (e.g., fertilizers, other agrochemicals, improved seed, increased mechanization), whereas pollinator populations are expected to continue their decline, these countervailing trends are likely to increase the pollinator-attributable portion of the yield gap globally. Therefore, it is possible that our results may ultimately underestimate the true effect of insufficient pollination on global health and diets. Fourth and finally, we note that our estimates of global food production and consumption are underpinned by FAO data drawn from nationally reported accounts, which can be unreliable in some countries given a low prioritization of agricultural data collection or poor quality data. It is difficult to estimate the magnitude or direction of these errors because of a slim literature systematically assessing data quality. Most available studies comparing FAO data with more reliable survey data have found that FAO tends to produce higher estimates of fruit and vegetable consumption. This includes non-starchy vegetable consumption in sub-Saharan Africa,67 although fruit and total fruit/vegetable intakes did not show a consistent and significant bias when measured by FAO compared with different methods. However, we have attempted to correct for this in our estimates by removing retail and domestic food waste, which is included in FAO food availability estimates but not others. Our results underscore the importance of pollinators for human health and increase the urgency of implementing pollinator-friendly policies to halt and reverse the trends of pollinator declines. Diverse research investigating the optimal policies to benefit pollination have shown remarkable consensus around a short list of highly effective strategies: increase flower abundance and diversity on farms, reduce pesticide use, and preserve or restore nearby natural habitat.28,68–71 This encouraging scientific agreement has already spurred action worldwide, with many countries creating and implementing their own national pollinator protection strategies. Despite this promising momentum, immense challenges remain for the restoration of pollinator populations globally. In this analysis, we have demonstrated that the protection of animal pollinators is not solely an ecological or environmental issue but also has significant implications for human health and economic well-being. Supplementary Material Click here for additional data file. Click here for additional data file. Click here for additional data file. Acknowledgments M.R.S., L.A.G., and S.S.M. conceived of the study. N.D.M. and J.G. conducted the attainable crop yield calculations. M.R.S. and L.A.G. conducted the calculations of the pollinator-attributable yield gap. T.B.S. and K.W. performed the International Model for Policy Analysis of Agricultural Commodities and Trade modeling of agricultural trade and economic value, as well as food production and consumption. M.S. performed the health modeling of ensuing dietary changes. M.R.S. collated all data outputs and produced data tables and figures. M.R.S. wrote the manuscript, with contributions from N.D.M., T.B.S., M.S., and S.S.M. All authors edited the manuscript. M.R.S. and S.S.M. are funded in part by grants from the Gordon and Betty Moore Foundation (to S.S.M.), Weston Foods Inc., and Fifth Generation Inc. T.B.S and K.W. were funded by the CGIAR Research Program on Policies, Institutions, and Markets although this research builds on previous work supported by the CGIAR Research Program on Climate Change, Agriculture and Food Security; the United States Agency for International Development (ID0EKYBG33202 to K.W.); the Bill & Melinda Gates Foundation (to K.W.); and the Wellcome Trust’s Our Planet Our Health program (205212/Z/16/Z to M.S., T.S., and K.W.) through the Livestock, Environment and People project (https://leap.web.ox.ac.uk/). ==== Refs References 1. FAO (Food and Agriculture Organization of the United Nations), IFAD (International Fund for Agricultural Development), UNICEF (United Nations Children’s Fund), WFP (World Food Program), WHO (World Health Organization). 2022. The State of Food Security and Nutrition in the World 2022. Rome, Italy: FAO. 2. Stoltzfus RJ. 2003. Iron deficiency: global prevalence and consequences. Food Nutr Bull 24 (suppl 4 ):S99–S103, PMID: , 10.1177/15648265030244S206.17016951 3. Zimmermann MB, Hurrell RF. 2007. Nutritional iron deficiency. Lancet 370 (9586 ):511–520, PMID: , 10.1016/S0140-6736(07)61235-5.17693180 4. Brown KH, Wuehler SE, Peerson JM. 2001. The importance of zinc in human nutrition and estimation of the global prevalence of zinc deficiency. Food Nutr Bull 22 (2 ):113–125, 10.1177/156482650102200201. 5. Caulfield LE, Black RE. 2004. Zinc deficiency. In: Comparative Quantification of Health Risks: Global and Regional Burden of Disease Attributable to Selected Major Risk Factors, vol 1 . Ezzati M, Lopez AD, Rodgers A, Murray CJL, eds. 257–280. 6. Wessells KR, Brown KH. 2012. Estimating the global prevalence of zinc deficiency: results based on zinc availability in national food supplies and the prevalence of stunting. PLoS One 7 (11 ):e50568, PMID: , 10.1371/journal.pone.0050568.23209782 7. World Health Organization. 2009. Global Prevalence of Vitamin A Deficiency in Populations at Risk 1995–2005: WHO Global Database on Vitamin A Deficiency. https://apps.who.int/iris/bitstream/handle/10665/44110/9789241598019_eng.pdf [accessed 30 September 2020]. 8. Stevens GA, Bennett JE, Hennocq Q, Lu Y, De-Regil LM, Rogers L, et al. 2015. Trends and mortality effects of vitamin A deficiency in children in 138 low-income and middle-income countries between 1991 and 2013: a pooled analysis of population-based surveys. Lancet Glob Health 3 (9 ):e528–e536, PMID: , 10.1016/S2214-109X(15)00039-X.26275329 9. de Onís M, Monteiro C, Akré J, Glugston G. 1993. The worldwide magnitude of protein–energy malnutrition: an overview from the WHO Global Database on Child Growth. Bull World Health Organ 71 (6 ):703–712, PMID: .8313488 10. Medek DE, Schwartz J, Myers SS. 2017. Estimated effects of future atmospheric CO2 concentrations on protein intake and the risk of protein deficiency by country and region. Environ Health Perspect 125 (8 ):087002, PMID: , 10.1289/EHP41.28885977 11. NCD-RisC (NCD Risk Factor Collaboration). 2016. A century of trends in adult human height. eLife 5 :e13410, PMID: , 10.7554/eLife.13410.27458798 12. Development Initiatives Poverty Research Ltd. 2020. 2020 Global Nutrition Report: Action on Equity to End Malnutrition. https://globalnutritionreport.org/reports/2020-global-nutrition-report/ [accessed 30 September 2020]. 13. GBD 2017 Risk Factor Collaborators. 2018. Global, regional, and national comparative risk assessment of 84 behavioural, environmental and occupational, and metabolic risks or clusters of risks for 195 countries and territories, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 392 (10159 ):1923–1994, PMID: , 10.1016/S0140-6736(18)32225-6.30496105 14. FAO, IFAD, UNICEF, WFP, WHO. 2021. The State of Food Security and Nutrition in the World 2021. Rome, Italy: FAO. 15. GLOPAN (Global Panel on Agriculture and Food Systems for Nutrition). 2020. Future Food Systems: For People, Our Planet, and Prosperity. https://www.glopan.org/wp-content/uploads/2020/09/Foresight-2.0_Future-Food-Systems_For-people-our-planet-and-prosperity.pdf [accessed 5 October 2020]. 16. Godfray HCJ, Garnett T. 2014. Food security and sustainable intensification. Philos Trans R Soc Lond B Biol Sci 369 (1639 ):20120273, PMID: , 10.1098/rstb.2012.0273.24535385 17. Willett W, Rockström J, Loken B, Springmann M, Lang T, Vermeulen S, et al. 2019. Food in the anthropocene: the EAT-Lancet Commission on healthy diets from sustainable food systems. Lancet 393 (10170 ):447–492, PMID: , 10.1016/S0140-6736(18)31788-4.30660336 18. Myers SS. Food and nutrition on a rapidly changing planet. 2020. In: Planetary Health: Protecting Nature to Protect Ourselves. Myers SS, Frumkin H, eds. Washington, DC: Island Press, 536. 19. Rockström J, Edenhofer O, Gaertner J, DeClerck F. 2020. Planet-proofing the global food system. Nat Food 1 (1 ):3–5, 10.1038/s43016-019-0010-4. 20. Mbow C, Rosenzweig C, Barioni LG, Benton TG, Herrero M, Krishnapillai M, et al. 2019. Food security. In: Climate Change and Land: An IPCC Special Report on Climate Change, Desertification, Land Degradation, Sustainable Land Management, Food Security, and Greenhouse Gas Fluxes in Terrestrial Ecosystems. Shukla PR, Skea J, Calvo Buendia E, Masson-Delmotte V, Pörtner HO, Roberts DC, et al. , eds. https://www.ipcc.ch/srccl/chapter/chapter-5/ [accessed 8 December 2020]. 21. Klein AM, Vaissière BE, Cane JH, Steffan-Dewenter I, Cunningham SA, Kremen C, et al. 2007. Importance of pollinators in changing landscapes for world crops. Proc Biol Sci 274 (1608 ):303–313, PMID: , 10.1098/rspb.2006.3721.17164193 22. Aune D, Giovannucci E, Boffetta P, Fadnes LT, Keum N, Norat T, et al. 2017. Fruit and vegetable intake and the risk of cardiovascular disease, total cancer and all-cause mortality—a systematic review and dose-response meta-analysis of prospective studies. Int J Epidemiol 46 (3 ):1029–1056, PMID: , 10.1093/ije/dyw319.28338764 23. Afshin A, Micha R, Khatibzadeh S, Mozaffarian D. 2014. Consumption of nuts and legumes and risk of incident ischemic heart disease, stroke, and diabetes: a systematic review and meta-analysis. Am J Clin Nutr 100 (1 ):278–288, PMID: , 10.3945/ajcn.113.076901.24898241 24. Aune D, Keum N, Giovannucci E, Fadnes LT, Boffetta P, Greenwood DC, et al. 2016. Nut consumption and risk of cardiovascular disease, total cancer, all-cause and cause-specific mortality: a systematic review and dose-response meta-analysis of prospective studies. BMC Med 14 (1 ):207, PMID: , 10.1186/s12916-016-0730-3.27916000 25. Garibaldi LA, Gemmill-Herren B, D’Annolfo R, Graeub BE, Cunningham SA, Breeze TD. 2017. Farming approaches for greater biodiversity, livelihoods, and food security. Trends Ecol Evol 32 (1 ):68–80, PMID: , 10.1016/j.tree.2016.10.001.27793463 26. Lautenbach S, Seppelt R, Liebscher J, Dormann CF. 2012. Spatial and temporal trends of global pollination benefit. PLoS One 7 (4 ):e35954, PMID: , 10.1371/journal.pone.0035954.22563427 27. Gallai N, Salles JM, Settele J, Vaissière BE. 2009. Economic valuation of the vulnerability of world agriculture confronted with pollinator decline. Ecol Econ 68 (3 ):810–821, 10.1016/j.ecolecon.2008.06.014. 28. IPBES (Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services). 2016. The Assessment Report on Pollinators, Pollination and Food Production: Summary for Policymakers. http://digitallibrary.un.org/record/1664349 [accessed 30 September 2020]. 29. Vasiliev D, Greenwood S. 2021. The role of climate change in pollinator decline across the Northern Hemisphere is underestimated. Sci Total Environ 775 :145788, PMID: , 10.1016/j.scitotenv.2021.145788.33618305 30. Potts SG, Imperatriz-Fonseca V, Ngo HT, Aizen MA, Biesmeijer JC, Breeze TD, et al. 2016. Safeguarding pollinators and their values to human well-being. Nature 540 (7632 ):220–229, PMID: , 10.1038/nature20588.27894123 31. Wagner DL, Grames EM, Forister ML, Berenbaum MR, Stopak D. 2021. Insect decline in the anthropocene: death by a thousand cuts. Proc Natl Acad Sci USA 118 (2 ):e2023989118, PMID: , 10.1073/pnas.2023989118.33431573 32. Zattara EE, Aizen MA. 2021. Worldwide occurrence records suggest a global decline in bee species richness. One Earth 4 (1 ):P114–P123, 10.1016/j.oneear.2020.12.005. 33. Aizen MA, Harder LD. 2009. The global stock of domesticated honey bees is growing slower than agricultural demand for pollination. Curr Biol 19 (11 ):915–918, PMID: , 10.1016/j.cub.2009.03.071.19427214 34. Garibaldi LA, Steffan-Dewenter I, Winfree R, Aizen MA, Bommarco R, Cunningham SA, et al. 2013. Wild pollinators enhance fruit set of crops regardless of honey bee abundance. Science 339 (6127 ):1608–1611, PMID: , 10.1126/science.1230200.23449997 35. Reilly JR, Artz DR, Biddinger D, Bobiwash K, Boyle NK, Brittain C, et al. 2020. Crop production in the USA is frequently limited by a lack of pollinators. Proc Biol Sci 287 (1931 ):20200922, PMID: , 10.1098/rspb.2020.0922.33043867 36. Garibaldi LA, Carvalheiro LG, Vaissière BE, Gemmill-Herren B, Hipólito J, Freitas BM, et al. 2016. Mutually beneficial pollinator diversity and crop yield outcomes in small and large farms. Science 351 (6271 ):388–391, PMID: , 10.1126/science.aac7287.26798016 37. FAOSTAT. 2020. Food balance sheets. http://www.fao.org/faostat/en/#data/FBSH/report [accessed 19 November 2020]. 38. Licker R, Johnston M, Foley JA, Barford C, Kucharik CJ, Monfreda C, et al. 2010. Mind the gap: how do climate and agricultural management explain the ‘yield gap’ of croplands around the world? Glob Ecol Biogeogr 19 (6 ):769–782, 10.1111/j.1466-8238.2010.00563.x. 39. Johnston M, Licker R, Foley J, Holloway T, Mueller ND, Barford C, et al. 2011. Closing the gap: global potential for increasing biofuel production through agricultural intensification. Environ Res Lett 6 (3 ):034028, 10.1088/1748-9326/6/3/034028. 40. Mueller ND, Gerber JS, Johnston M, Ray DK, Ramankutty N, Foley JA. 2012. Closing yield gaps through nutrient and water management. Nature 490 (7419 ):254–257, PMID: , 10.1038/nature11420.22932270 41. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A. 2005. Very high resolution interpolated climate surfaces for global land areas. Int J Climatol 25 (15 ):1965–1978, 10.1002/joc.1276. 42. Monfreda C, Ramankutty N, Foley JA. 2008. Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types, and net primary production in the year 2000. Global Biogeochem Cycles 22 (1 ):GB1022, 10.1029/2007GB002947. 43. Robinson S, Mason-D’Croz D, Islam S, Sulser TB, Robertson RD, Zhu T, et al. 2015. The International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT): Model Description for Version 3. International Food Policy Research Institute Discussion Paper 1483. Washington, DC: IFPRI. 44. Mason-D’Croz D, Bogard JR, Herrero M, Robinson S, Sulser TB, Wiebe K, et al. 2020. Modelling the global economic consequences of a major African swine fever outbreak in China. Nat Food 1 (4 ):221–228, PMID: , 10.1038/s43016-020-0057-2.33634268 45. Springmann M, Mason-D’Croz D, Robinson S, Garnett T, Godfray HCJ, Gollin D, et al. 2016. Global and regional health effects of future food production under climate change: a modelling study. Lancet 387 (10031 ):1937–1946, PMID: , 10.1016/S0140-6736(15)01156-3.26947322 46. Mason-D’Croz D, Sulser TB, Wiebe K, Rosegrant MW, Lowder SK, Nin-Pratt A, et al. 2019. Agricultural investments and hunger in Africa modeling potential contributions to SDG2 – Zero Hunger. World Dev 116 :38–53, PMID: , 10.1016/j.worlddev.2018.12.006.30944503 47. Rosegrant MW, Koo J, Cenacchi N, Ringler C, Robertson R, Fisher M, et al. 2014. Food Security in a World of Natural Resource Scarcity: The Role of Agricultural Technologies. Washington, DC: International Food Policy Research Institute. 48. van Vuuren DP, Edmonds J, Kainuma M, Riahi K, Thomson A, Hibbard K, et al. 2011. The representative concentration pathways: an overview. Clim Change 109 (1–2 ):5–31, 10.1007/s10584-011-0148-z. 49. Riahi K, van Vuuren DP, Kriegler E, Edmonds J, O’Neill BC, Fujimori S, et al. 2017. The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: an overview. Glob Environ Change 42 :153–168, 10.1016/j.gloenvcha.2016.05.009. 50. Springmann M, Wiebe K, Mason-D’Croz D, Sulser TB, Rayner M, Scarborough P. 2018. Health and nutritional aspects of sustainable diet strategies and their association with environmental impacts: a global modelling analysis with country-level detail. Lancet Planet Health 2 (10 ):e451–e461, PMID: , 10.1016/S2542-5196(18)30206-7.30318102 51. Gustavsson J, Cederberg C, Sonesson U, van Otterdijk R, Meybeck A. 2011. Global Food Losses and Food Waste: Extent, Causes and Prevention; Study Conducted for the International Congress Save Food! At Interpack2011 Düsseldorf, Germany. Düsseldorf, Germany. Food and Agriculture Organization of the United Nations. 52. Berrington de Gonzalez A, Hartge P, Cerhan JR, Flint AJ, Hannan L, MacInnis RJ, et al. 2010. Body-mass index and mortality among 1.46 million white adults. N Engl J Med 363 (23 ):2211–2219, PMID: , 10.1056/NEJMoa1000367.21121834 53. Singh GM, Danaei G, Farzadfar F, Stevens GA, Woodward M, Wormser D, et al. 2013. The age-specific quantitative effects of metabolic risk factors on cardiovascular diseases and diabetes: a pooled analysis. PLoS One 8 (7 ):e65174, PMID: , 10.1371/journal.pone.0065174.23935815 54. Micha R, Shulkin ML, Peñalvo JL, Khatibzadeh S, Singh GM, Rao M, et al. 2017. Etiologic effects and optimal intakes of foods and nutrients for risk of cardiovascular diseases and diabetes: systematic reviews and meta-analyses from the Nutrition and Chronic Diseases Expert Group (NutriCoDE). PLoS One 12 (4 ):e0175149, PMID: , 10.1371/journal.pone.0175149.28448503 55. Chan DSM, Lau R, Aune D, Vieira R, Greenwood DC, Kampman E, et al. 2011. Red and processed meat and colorectal cancer incidence: meta-analysis of prospective studies. PLoS One 6 (6 ):e20456, PMID: , 10.1371/journal.pone.0020456.21674008 56. Chen GC, Lv DB, Pang Z, Liu QF. 2013. Red and processed meat consumption and risk of stroke: a meta-analysis of prospective cohort studies. Eur J Clin Nutr 67 (1 ):91–95, PMID: , 10.1038/ejcn.2012.180.23169473 57. Feskens EJM, Sluik D, van Woudenbergh GJ. 2013. Meat consumption, diabetes, and its complications. Curr Diab Rep 13 (2 ):298–306, PMID: , 10.1007/s11892-013-0365-0.23354681 58. Micha R, Michas G, Mozaffarian D. 2012. Unprocessed red and processed meats and risk of coronary artery disease and type 2 diabetes—an updated review of the evidence. Curr Atheroscler Rep 14 (6 ):515–524, PMID: , 10.1007/s11883-012-0282-8.23001745 59. Zheng J, Huang T, Yu Y, Hu X, Yang B, Li D. 2012. Fish consumption and CHD mortality: an updated meta-analysis of seventeen cohort studies. Public Health Nutr 15 (4 ):725–737, PMID: , 10.1017/S1368980011002254.21914258 60. Prospective Studies Collaboration, Whitlock G, Lewington S, Sherliker P, Clarke R, Emberson J, et al. 2009. Body-mass index and cause-specific mortality in 900 000 adults: collaborative analyses of 57 prospective studies. Lancet 373 (9669 ):1083–1096, PMID: , 10.1016/S0140-6736(09)60318-4.19299006 61. GBD 2013 Risk Factors Collaborators, Forouzanfar MH, Alexander L, Anderson HR, Bachman VF, Biryukov S, et al. 2015. Global, regional, and national comparative risk assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters of risks in 188 countries, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 386 (10010 ):2287–2323, PMID: , 10.1016/S0140-6736(15)00128-2.26364544 62. World Bank Country and Lending Groups. https://datahelpdesk.worldbank.org/knowledgebase/articles/906519 [accessed 5 November 2020]. 63. United Nations. 2022. Population Division: World Population Prospects 2022. https://population.un.org/wpp/ [accessed 8 November 2022]. 64. IHME (Institute for Health Metrics and Evaluation). 2019. GBD Results Tool. http://ghdx.healthdata.org/gbd-results-tool [accessed 8 November 2022]. 65. Eilers EJ, Kremen C, Smith Greenleaf S, Garber AK, Klein AM. 2011. Contribution of pollinator-mediated crops to nutrients in the human food supply. PLoS One 6 (6 ):e21363, PMID: , 10.1371/journal.pone.0021363.21731717 66. Smith MR, Singh GM, Mozaffarian D, Myers SS. 2015. Effects of decreases of animal pollinators on human nutrition and global health: a modelling analysis. Lancet 386 (10007 ):1964–1972, PMID: , 10.1016/S0140-6736(15)61085-6.26188748 67. Mensah DO, Nunes AR, Bockarie T, Lillywhite R, Oyebode O. 2021. Meat, fruit, and vegetable consumption in sub-Saharan Africa: a systematic review and meta-regression analysis. Nutr Rev 79 (6 ):651–692, PMID: , 10.1093/nutrit/nuaa032.32556305 68. Kennedy CM, Lonsdorf E, Neel MC, Williams NM, Ricketts TH, Winfree R, et al. 2013. Global quantitative synthesis of local and landscape effects on wild bee pollinators in agroecosystems. Ecol Lett 16 (5 ):584–599, PMID: , 10.1111/ele.12082.23489285 69. Garibaldi LA, Carvalheiro LG, Leonhardt SD, Aizen MA, Blaauw BR, Isaacs R, et al. 2014. From research to action: enhancing crop yield through wild pollinators. Front Ecol Environ 12 (8 ):439–447, 10.1890/130330. 70. Dicks LV, Viana B, Bommarco R, Brosi B, Arizmendi MDC, Cunningham SA, et al. 2016. Ten policies for pollinators. Science 354 (6315 ):975–976, PMID: , 10.1126/science.aai9226.27884996 71. Sardiñas H, Code A, Cruz JK, et al. 2021. Bee Better Certified: Background to the Production Standards. Portland, OR: Xerces Society for Invertebrate Conservation.

36515549

PMC9749483

NO-CC CODE

2022-12-16 23:24:04

no

Environ Health Perspect. 2022 Dec 14; 130(12):127003

utf-8

Environ Health Perspect

10.1289/EHP10947

oa_other

==== Front J Surg Res J Surg Res The Journal of Surgical Research 0022-4804 1095-8673 Academic Press S0022-4804(22)00302-X 10.1016/j.jss.2022.04.075 Education and Career Development Virtual Interactions and the 2020-2021 Residency Application Cycle in General Surgery: A Look Ahead DeLay Thomas K. BA a Singh Nikhi P. BS a Duong Teressa A. BS a Rais-Bahrami Soroush MD abcd King Timothy W. MD, PhD ef Chen Herbert MD ad Corey Britney L. MD ag∗ a Heersink School of Medicine, University of Alabama at Birmingham, Birmingham, Alabama b Department of Urology, University of Alabama at Birmingham, Birmingham, Alabama c Department of Radiology, University of Alabama at Birmingham, Birmingham, Alabama d O'Neal Comprehensive Cancer Center, University of Alabama at Birmingham, Birmingham, Alabama e Division of Plastic and Reconstructive Surgery, Department of Surgery, Loyola University Stritch School of Medicine, Maywood, Illinois f Plastic Surgery Section, Hines VA Medical Center, Hines, Illinois g Birmingham Veteran's Affairs Medical Center, Birmingham, Alabama ∗ Corresponding author. The University of Alabama at Birmingham School of Medicine, Birmingham, AL. Tel.: +1 205 975 3000; fax: +1 205 975 0286. 31 5 2022 10 2022 31 5 2022 278 331336 3 1 2022 22 3 2022 27 4 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Introduction General surgery residency training programs adapted to the COVID-19 pandemic by going online instead of in-person, through virtual interviews, social media engagement, and virtual open houses. The impact of these virtual interactions is unknown. We sought to understand their effectiveness as per residency program directors and assistant program directors. Materials and methods An institutional review board approval was obtained to conduct this anonymous survey. A Qualtrics XM survey containing multiple-choice and short-answer questions was distributed to 590 residency program and assistant program directors through the Association of Program Directors in Surgery (APDS) listserv on July 6, July 13, and July 20. Results We observed a response rate of approximately 11% across the 590 surgeons contacted. Nearly all (90%) respondents offered virtual preinterview interactions, primarily virtual open houses, virtual facility tours, and virtual question and answer (Q&A) sessions with residents and faculty; 48% of respondents were unsure of the utility of virtual interactions and the majority (54%) felt that virtual interaction limits a program's ability to evaluate applicants. Virtual Q&As were ranked to be the most effective interaction (7.6/10); 80% of respondents felt that visiting rotations were “somewhat important” to “very important,” the two highest options available. In addition, 74% felt that applicants missed out on fully experiencing the program by forgoing these rotations. Most respondents (78%) noted that evaluation of applicants’ preinterview did not change as a result of virtual interactions. Nearly half (48%) of the respondents offered more interview days due to the virtual format. A fifth (21%) of respondents stated that virtual interactions resulted in a change in the rank position of an applicant. Respondents ranked Twitter and Instagram higher in applicant engagement than Facebook. Factors that impacted interview or rank order list the most were late/absent step two CK scores (33%) and a lack of away rotations (31%), both being limitations largely due to the pandemic. With respect to future application cycles, most (71%) raised concerns regarding disparities between applicants applying in-person and virtually if both or either are offered. Conclusions Our study suggests that program directors and associate program directors have reservations about the use of virtual interactions with applicants. Interestingly, these data suggest that visiting subinternships are useful for programs in evaluating applicants. This may encourage students to pursue rotations at other institutions at the expense of already-limited resources. It remains unclear whether virtual interactions will be used in the future, but respondents largely agreed that the virtual means of interacting with and disseminating information to the applicants of the 2020-2021 general surgery Match were a success. Keywords General surgery Interview Match Residency Subinternship Virtual reality Virtual ==== Body pmcIntroduction In an unprecedented effort to limit transmission in the context of the COVID-19 pandemic, United States medical students were largely barred from partaking in many of their usual hospital-based clinical duties for a substantial portion of 2020 and 2021.1 These measures were necessary for the mutual safety of students and patients alike, although their impact on the 2021 Association of American Medical Colleges residency applicant pool and programs respective responses have been equally unprecedented. Efforts were taken to bridge the gap for students who were unable to participate in standard in-person clinical clerkships, namely simulated shelf-like questions and virtual group learning activities.2 , 3 However, the limited availability of extracurricular academic interaction, namely in-person interviews and rotations at other institutions, were largely unaccounted for. Visiting subinternships were brought to a halt, limiting students’ ability to assess and be assessed by programs outside of their home institutions.4 Rotations at outside institutions are of particular significance among more competitive institutions and historically competitive specialties.5 , 6 Students reported significantly higher levels of stress surrounding the virus and fear of transmission and these changes in the academic environment are likely to have compounded that stress.7 In response to the COVID-19 pandemic, residency programs appeared to enhance their online presences through the creation of virtual opportunities and social media accounts.8, 9, 10 Some specialties saw the introduction of a virtual subinternship, although these virtual subinternship opportunities represent a small fraction of applicants compared to prior years.11 , 12 A recent study estimated the prevalence of away rotators to be 59% of all fourth-year medical students.13 The same study reported 36% of applicants matching at an institution where they had rotated, either their home institution or one at which a visiting rotation was performed. These unique experiences offer students a useful insight into programs that they would likely not otherwise have access to. Given the absence of in-person interviews and the near-elimination of subinternships, residency applicants in 2020 were left without two primary mainstays of interacting with programs other than their home institution. This constraint hindered students’ ability to assess institutions where they may be interested in matching and assess other programs in relation to programs known to them. Program directors are similarly limited and their programs face an equally difficult task of adequately evaluating individual applicants and managing ever-growing applicant pools.14 The effectiveness of virtual interactions remains unclear. Significant opportunities exist to investigate the quality of virtual interactions between residency programs and prospective applicants.15 This article seeks to enumerate both how programs accommodated COVID-19 restrictions and their retrospective perception of the successes and failures of those accommodations. These perspectives may offer a unique perspective on general surgery residency program directors’ impressions of the 2020 Match and a possible insight into what elements of virtual interaction may remain in the application years to come, independent of the progress and hopeful resolution of the ongoing global pandemic. Methods Institutional board exemption (IRB #300006990) was obtained to conduct this anonymous survey. A Qualtrics XM survey containing multiple-choice and short-answer questions was distributed to 590 residency program and assistant program directors through the Association of Program Directors in Surgery (APDS) over the course of three consecutive weeks. No identifying data were collected. The survey was closed 1 week following the last distribution email. A full list containing the survey's questions is provided within the supplemental document (Survey Questions). Results Approximately 11% of the 590 APDS members contacted responded to the survey with generally even distribution across regions of the United States (Table 1 ). Among respondents, almost all (90%) offered virtual preinterview interactions, primarily open house and question and answers (Q&As), with residents and faculty. Of preinterview offerings, the most highly rated were Q&A sessions with residents and faculty, 7.6/10 and 7.1/10, respectively (Fig. ). Of our respondents, 43% thought these preinterview virtual interactions were a success while a slightly greater portion (48%) was unsure of their utility. Of those who thought preinterview virtual events were not a success (9%), the major reason was difficulty assessing applicants over a virtual platform. Most general surgery residency program directors and assistant program directors (78%) concluded that their evaluations of students did not change because of preinterview virtual interactions. However, those whose evaluations of students did change due to preinterview virtual interactions reported that virtual opportunities allowed the opportunity to gauge students’ interest and ability to interact with others.Table 1 Respondent demographics. Respondent demographics % Count Region  Northeastern (Connecticut, Delaware, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont, Washington, D.C.) 38.33% 23  Southern (Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, Oklahoma, Puerto Rico, South Carolina, Tennessee, Texas, Virginia, West Virginia) 26.67% 16  Central (Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, Wisconsin) 20.00% 12  Western (Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, Wyoming) 15.00% 9  Total 100% 60 Role  Program director 83.33% 50  Associate program director 16.67% 10  Total 100% 60 Response rate  Completed at least some of the survey 10.5% 62  Received but did not complete survey 89.5% 528  Total 100% 590 Fig. Histogram with standard deviation of responses to Q25: “Please rate the effectiveness of each preinterview virtual opportunity you offered. (0, Not influential at all; 10, The most beneficial metric available to you this application cycle).” Of respondents, 69% said their virtual events were conducted to provide information about their programs to applicants, whereas 45% of responses reported that the events were used to advertise their programs to applicants (Table 2 ), and 43% said they were held to facilitate resident-student interactions. A minority, 26%, of programs answered that events served as a way to get to know applicants and determine their fit within the program. Although most programs developed social media accounts for their residency programs after March 1, 2020, their interactions with applicants via those accounts largely did not affect how applicants were viewed (mean 3/10). The respondents who said they evaluated candidates differently in this cycle were more likely to have offered virtual internships and online didactics and less likely to have offered other less structured events like Q&As and meet and greets (Table 3 ).Table 2 Responses to Q20: “Why did you offer these preinterview virtual opportunities? Select all that apply.” # Answer % Count 1 To provide information about our program to applicants 30.08% 40 2 To get to know applicants 11.28% 15 3 To facilitate resident and applicant interaction 18.80% 25 4 To evaluate applicant interest in our program 11.28% 15 5 To determine applicant fit in our program 8.27% 11 6 To advertise our program 19.55% 26 7 Other 0.75% 1 Total 100% 133 Table 3 Responses to Q16: “In response to the COVID-19 pandemic, did you change your preinterview selection criteria for candidates during the 2020-2021 application cycle? How so?” Number of responses Question Yes (%), we evaluated candidates differently in this application cycle than in prior cycles No (%), we evaluated candidates in a similar manner as in prior cycles Total 1 Virtual open houses 24.14% 7 75.86% 22 29 2 Virtual subinternships 100.00% 1 0.00% 0 1 3 Didactic lectures 100.00% 3 0.00% 0 3 8 Virtual grand rounds 50.00% 3 50.00% 3 6 4 Q&A session with program director and faculty 18.52% 5 81.48% 22 27 5 Q&A session with residents 19.44% 7 80.56% 29 36 6 Virtual facility tour 25.00% 7 75.00% 21 28 7 Other 25.00% 1 75.00% 3 4 9 We did not offer any preinterview virtual opportunities 0.00% 0 100.00% 2 2 Respondents reported factors that impacted interview decisions or applicants' ranking most significantly were late/absent Step two CK scores (33%) and a lack of visiting subinternships (31%). Of respondents, 75% said they did not change preinterview selection criteria, and 74% of respondents felt applicants missed out on fully experiencing their program by not doing an away rotation there. Of the responding program directors and assistant program directors, 54% believed virtual interviews limit a program's ability to gauge applicants, although residency programs remain divided on whether absence of an away rotation opportunity would limit their ability to assess applicants: 37% answered “yes;” and 46% answered “no.” However, a large majority of respondents (80%) believed visiting rotations were at least somewhat important and 35% believed they were very important. Among those who answered, 50% said the expense of in-person interviews will impact how they interview students in the future, whereas 30% said it will not, and 20% remained unsure. Respondents remain equally split regarding how cost will affect future structures; 40% said expense will impact the decision to offer virtual versus in-person interviews, whereas 40% said it will not, and 20% remained unsure. Discussion In this study, we sought to characterize the changes made for the 2021 Residency Match process and residency program leadership impressions of their degrees of success and viability for future cycles. The restrictions surrounding travel and in-person interaction in the era of COVID-19 have affected physicians in nearly every corner of their practice. Residency applicants were faced with changes to the traditional aspects of application and their application cycle. All United States medical students have experienced some degree of limitation in their clinical exposure in the era of COVID-19, even within their own institutions. These limitations may vary from institution to institution, but Association of American Medical Colleges restrictions on rotations at visiting institutions and in-person interviews stand out as a broad and uniformly limiting change to a once-crucial aspect of applying to the Match, especially in more competitive fields like surgery. The most highly rated virtual preinterview interactions were Q&A sessions with residents and faculty. Faculty viewed the interactions primarily as a means of familiarizing applicants with their programs, rather than a method of primary evaluation of applicants for preinterview vetting. For the minority who said that preinterview interactions altered their perceptions of applicants, a majority said their ranks of these students were positively affected, causing them to be ranked higher. Of identifiable areas affected by COVID-19 restrictions, program directors and assistant program directors cited absent or late Step two CK scores and a lack of visiting subinternships as areas that negatively affected the strength of students' applications. Although Step two CK testing was not directly affected in the way visiting subinternships were, many students experienced unexpected test date cancellations and had difficulty in rescheduling tests at nearby proctored locations.16 Visiting subinternships were virtually eliminated altogether for most students. It would be difficult to estimate all the impacts COVID-19 had on applicants, but the lack of visiting subinternships and affected Step two CK testing stand out among respondents and may merit a continued examination. Although this study evaluated the perspective of residency program leadership on the virtual offerings only for the 2020-2021 application cycle, the results remain relevant because they pose implications for current and future application cycles given the unexpected length of the COVID-19 pandemic. Cycles in the future stand to benefit from this information and will be well-served to consider the efficacy of various virtual interactions as they plan for an uncertain future. With these changes, we hope to see a greater degree of transparency in program directors’ plans for applicant interactions so that students can also prepare appropriately. Concerns facing students are academic, logistical, and more now than ever, financial. Outside of academic concerns, the changing landscape of cross-institutional interactions before the residency Match raises the issue of equity in access to visiting subinternships and interviews. Of the responding general surgery residency program directors and assistant program directors, 71% worry about disparities between applicants applying in-person and virtually. Visiting subinternships provide an excellent opportunity for exposure to specialties that may not be present at a student's home institution and an opportunity to potentially audition for a spot at an outside institution. Like many aspects of medical education, however, these subinternships require students to incur a significant personal expense. A 2016 study estimated the cost of a single visiting rotation to be $958.13 With many students, particularly those applying to surgery and surgical subspecialties applying for multiple visiting subinternships, these expenses can easily reach several thousands of dollars.17 Total expenditures among specialties ranged from $1312 to $3465.13 Respondents to our survey are largely divided on how to address student expenses, with only half saying student expenses will affect how they interview in the future, including the decision to offer virtual or in-person interviews. In one sense, the restriction of visiting subinternships controls for financial privilege among applicants. On the other hand, it disadvantages students coming from smaller institutions without a large availability of subspecialty rotations. In-person interviews present another disparity in access by requiring students to incur costs of travel and lodging. Rather than assuaging this disparity, virtual interviews allow wealthier applicants to apply to as many programs as they choose without concerns surrounding the expense of extra applications and without having to consider the logistic implications of in-person interviews. The COVID-19 restrictions on travel offer a unique kind of case study on virtual interactions and their impact on access and outcomes and thereby warrant further investigation. Continuous studies evaluating these online opportunities can better provide applicants and programs with understanding of the impact of these interactions and how to improve in the future. Limitations to this study include a small sample size, potential unknown bias in response, possible subjectivity in the survey itself, and the uncertainty of the accuracy of retrospect among respondents. Our response rate was particularly low at 11.5% (Table 1). We are unaware of any other concomitant surveys at the time of issuance, but time to complete survey and a lack of incentive or compensation likely played some role in the poor turnout. Further studies may benefit from more concise surveys, possibly multiple. This small sample size reduces the statistical power of our study. However, as a purely descriptive study, our results still provide an insight into the experiences of those who chose to participate. These data are based on a collection period that may not have allowed all programs to fully develop their response for the next cycle of the residency Match. Future directions following this study could include postmatch interviews with applicants on their impressions of social media interactions, online open houses, and virtual away electives in comparison with a similar survey given prior to any of these virtual interactions. In addition, as the opportunity cost of advanced medical training and higher education continues to grow out of proportion to students ‘and applicants’ financial means, future studies would be well-served to further characterize the impact of financial burden on United States medical trainees before and after the global COVID-19 pandemic and possible accommodations that may be adopted beyond the era of COVID-19 in the future. There remains much to be seen regarding what will be done in future application cycles, not to mention what faculty and applicants believe ought to be done. Choosing between fully virtual and fully physical interviews makes an implicit argument about the significance of the interview itself. In-person interviews indirectly assert the importance of first-hand observation of social interaction not otherwise seen on a visiting rotation. Virtual interviews, however, may suggest that applicants have been vetted socially by their letters of recommendation. Allowing applicants to choose presents the problem of stratifying applicants by their presumed interest based on whether they chose to invest the time and resources needed for a physical interview over a virtual one. Programs may benefit from offering two rounds of interviews, the first being virtual and the latter physical. This would allow applicants greater freedom to apply to more programs without a fear of being limited by travel resources. The secondary interview would be in-person, acting to gage an applicant's compatibility with the program and serve as a demonstration of committed interest in a program. In-person interviews could also possibly be waived if a student had already completed a visiting rotation in the department to which they are applying. Disclosure None declared. Funding None. Availability of Data Not applicable. Supplementary Materials Survey COVID-19 Gensurg Application process Supplementary data to this article can be found online at https://doi.org/10.1016/j.jss.2022.04.075. ==== Refs References 1 Sell N.M. Qadan M. Delman K.A. Implications of COVID-19 on the general surgery match Ann Surg 272 2020 e155 e156 32675523 2 Negrete Manriquez J.A. Bazargan-Hejazi S. Nahm S.J. de Virgilio C. Exploring a novel approach to surgery clerkship didactics during the COVID-19 pandemic: a qualitative study Am J Surg 223 2021 662 669 34284882 3 Kempthorne D. Williams L. Brazelle M. Navigating the virtual medical school experience during COVID - what comes next? Am J Surg 223 2021 1013 1014 34861988 4 Boyd C.J. Inglesby D.C. Corey B. Impact of COVID-19 on away rotations in surgical field J Surg Res 255 2020 96 98 32543384 5 Higgins E. Newman L. Halligan K. Miller M. Schwab S. Kosowicz L. Do audition electives impact match success? Med Educ Online 21 2016 31325 27301380 6 Huffman E.M. Athanasiadis D.I. Anton N.E. How resilient is your team? Exploring healthcare providers' well-being during the COVID-19 pandemic Am J Surg 221 2021 277 284 32994041 7 Singh N.P. Fang H.A. Lopez R. DeAtkine A. Burns Z. Boyd C.J. Thoughts on 2021- 2022 visiting rotation recommendations from current and recent applicants in surgery Am J Surg 222 2021 903 904 33810835 8 DeAtkine A.B. Grayson J.W. Singh N.P. Nocera A.P. Rais-Bahrami S. Greene B.J. #ENT: otolaryngology residency programs create social media platforms to connect with applicants during COVID-19 pandemic [e-pub ahead of print] Ear Nose Throat J 2020 10.1177/0145561320983205 145561320983205 9 Pruett C. Deneen K. Kozar T. Social media changes in pediatric residency programs during COVID-19 pandemic Acad Pediatr 21 2021 1104 1107 34126258 10 Fang H.A. Boudreau H. Khan S. An evaluation of social media utilization by general surgery programs in the COVID-19 era Am J Surg 222 2021 937 943 33906728 11 Cutshall H. Hattaway R. Singh N.P. Rais-Bahrami S. McCleskey B. The #Path2Path virtual landscape during the COVID-19 pandemic: preparing for the 2020 pathology residency recruitment season Acad Pathol 8 2021 23742895211002783 12 Pasala M.S. Anabtawi N.M. Farris R.L. Family medicine residency virtual adaptations for applicants during COVID-19 pandemic Fam Med 53 2021 684 688 34587263 13 Winterton M. Ahn J. Bernstein J. The prevalence and cost of medical student visiting rotations BMC Med Educ 16 2016 291 27842590 14 Boyd C.J. Ananthasekar S. Greene B.J. Reply to: virtual interviews for the 2020-2021 national residency matching program during the COVID-19 pandemic: a curse or blessing? [e-pub ahead of print] Am Surg 2021 10.1177/0003134821995068 3134821995068 15 DeAtkine A.B. Chisolm P.F. Singh N.P. Interviewing otolaryngology applicants in a virtual setting: a perspective after 2020 to 2021 match [e-pub ahead of print] Ear Nose Throat J 2021 10.1177/01455613211040377 1455613211040377 16 Price S. Testing boundaries COVID-19 made the USMLE, clerkships a moving target for med students Tex Med 116 2020 34 17 Robin J.X. Boyd C.J. Rais-Bahrami S. Ponce B.A. Addressing the financial burdens of away rotations in surgical fields [e-pub ahead of print] Am Surg 2021 10.1177/00031348211023400 31348211023400

35659708

PMC9749614

NO-CC CODE

2022-12-15 23:23:20

no

J Surg Res. 2022 Oct 31; 278:331-336

utf-8

J Surg Res

10.1016/j.jss.2022.04.075

oa_other

==== Front Geochim Cosmochim Acta Geochim Cosmochim Acta Geochimica et Cosmochimica Acta 0016-7037 0016-7037 Pergamon Press S0016-7037(21)00059-4 10.1016/j.gca.2021.01.033 Awards Ceremony Speech Citation for the 2020 F.W. Clarke Medal to Daniel Stolper Bender Michael L. Princeton University, Dept. of Geosciences, Washington Rd, Princeton, NJ 08540, USA 28 1 2021 1 4 2021 28 1 2021 298 244245 20 1 2021 20 1 2021 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc Daniel Stolper was born and raised in Pasadena, California. He earned an A. B. degree from Harvard, was Fulbright Scholar for a year at the University of Southern Denmark, and did his Ph. D. at Caltech. After a postdoc at Princeton, Daniel moved to the University of California, where he is Assistant Professor at the time of writing. Daniel is a geochemist whose research has 3 broad themes. The first is the isotope geochemistry of methane. Daniel did pioneering work using a reimagined isotope ratio mass spectrometer conceived and developed by his Ph. D. advisor, John Eiler. This instrument enabled the user to determine the abundance of methane isotopologues with 2 heavy atoms, 13CH3D or 12CH2D2. The property of merit is the clumped isotope anomaly, the departure in the abundance of an isotopologue with 2 heavy atoms from its stochastic abundance. Daniel and his collaborators used experiments and theory to determine the temperature dependence of the clumped isotope anomalies in methane. They then showed that many methane sources had a clumped isotope composition at isotopic equilibrium, given their temperature of formation or alteration. They also showed that kinetics induced disequilibrium in biogenic methane sources, especially when methane production was rapid. They developed a concept for classifying methane according to its C and H isotope composition, its clumped isotope composition, and the abundance of methane relative to that of other light hydrocarbons. They also explored the application of clumped isotopes to studies of methane geochemistry in particular deposits. The second theme of Daniel’s research is an investigation of clumped isotope diagenesis in carbonates. With colleagues, Daniel made models of diagenetic changes to clumped isotope abundances, and challenged the models’ simulations with data. Daniel had the deep insight that apparently large changes observed when calcites were first heated could be attributed to isotope exchanges with neighboring calcite molecules. He and his colleagues worked out the roles of diagenesis in the presence and absence of water. They also showed that one could reconstruct environmental information with diagenetic models of calcite recrystallization in deep-sea sediments. The third theme of Daniel’s research is centered around O2, in both geological and biogeochemical contexts. Daniel reasoned that he could use the oxidation state of ocean crust, accessed via ophiolites and island arc volcanics, to determine the oxidation state of the deep ocean. The connection comes from the fact that ocean crust is altered by reaction with seawater. Seawater-basalt exchange then imprints the ocean crust, and eventually arc volcanics, with the oxidation state of the deep ocean. This approach then gave a time of about 400–500 Ma for elevating the deep ocean O2 concentration. Since gases in the deep ocean mix with air over a timescale of ∼ 1 kyr, the history of deep ocean oxygenation may be similar for the surface ocean and atmosphere. Daniel has also worked on the more recent history of atmospheric O2. He repurposed a large database of O2 concentration in ice core trapped gases to determine changes over the past 800 kyr. This work showed that the O2 concentration is decreasing at a rate of about 1200 ppm/Myr (out of 210,000 ppm O2 in air). The imbalance is very small, pointing to the role of strong feedbacks in the carbon cycle that remain to be identified. On the biogeochemical side, Daniel characterized oxygen isotope fractionation associated with respiration with exceptional depth. He showed that the fractionation patterns he observed implied a 2-step process for O2 consumption at cytochrome oxidase: a reversible step in which O2 is bound, followed by a kinetic step in which O2 is reduced. He also showed that O2 isotope fractionation at cold temperatures was much smaller than heretofore estimated. This observation accounts for weak O2 isotope fractionation associated with O2 consumption in the deep ocean, solving a mystery going back nearly 50 years. Daniel is a brilliant scholar who takes great joy in the doing of science. He is a wonderful colleague: interactive, stimulating, and deeply knowledgeable about a very broad range of topics. He is hardworking and efficient. He is generous intellectually and personally. Daniel will contribute to the community of earth scientists, and provide leadership in the research, for many years to come.

0

PMC9749615

NO-CC CODE

2022-12-15 23:23:20

no

Geochim Cosmochim Acta. 2021 Apr 1; 298:244-245

utf-8

Geochim Cosmochim Acta

10.1016/j.gca.2021.01.033

oa_other

==== Front Geochim Cosmochim Acta Geochim Cosmochim Acta Geochimica et Cosmochimica Acta 0016-7037 0016-7037 Pergamon Press S0016-7037(21)00047-8 10.1016/j.gca.2021.01.025 Awards Ceremony Speech Citation for the 2020 V.M. Goldschmidt Award to Richard W. Carlson Hofmann Albrecht W. Max Planck Institute for Chemistry, 55128 Mainz, Germany 4 2 2021 1 4 2021 4 2 2021 298 258259 20 1 2021 20 1 2021 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmc Rick Carlson was brought into the world of Nd isotope geochemistry by Günter Lugmair at Scripps, himself a Goldschmidt medalist. In many ways, neodymium isotopes have revolutionized planetary and especially terrestrial geochemistry because of the robustness and geological-memory-preserving properties of the Sm-Nd decay system, and Rick has been a leading neodymium isotopicist from the day one. As a Ph.D. student, he did some of the first lunar and terrestrial Nd isotope analyses back in 1978. This work established the age of the lunar crust at about 4.4 billion years, a value that has pretty much stood the test of time, though some people now claim that the lunar crust may be as old as 4.5 Ga. Also, quite early on, he did definitive work on the American type locality of continental flood basalts, the Columbia River basalt. He showed that crustal assimilation played a decisive role in the chemical evolution of these flood basalts, and they cannot simply be derived from a primitive mantle reservoir. He came to the Department of Terrestrial Magnetism of the Carnegie Institution Washington in 1980, just after I left to go Germany, and he proceeded to build a world-leading isotope geochemistry group there that is now going as strong as ever. I will jump to the year 2005, when Carlson with his postdoc Maud Boyet dropped a veritable bombshell on the community: They showed that the atomic abundance of 142Nd, daughter of the “short-lived” parent nuclide 146Sm (T1/2 ≤ 100 Myr), in all terrestrial rocks differs from that found in chondritic meteorites. This completely unexpected discovery demonstrated that, if Earth has a chondritic Sm/Nd ratio and initial 142Nd atomic abundance, then Earth must have been differentiated permanently into an early “enriched” (i.e. low-Sm/Nd) and an early “depleted” (high-Sm/Nd) reservoir, and this differentiation event must have occurred more than 4.53 Gyr ago. The past fifteen years have seen an enormous research effort directed at trying to understand this dilemma and to sort out the first 500 Myr of Earth-Moon history. Carlson has remained firmly at the forefront of this work. For example, Boyet et al. (2015) found that the 142Nd systematics of the lunar crust is consistent either with a chondritic 142Nd/144Nd ratio combined with an increased Sm/Nd ratio, or alternatively, with the Moon having a chondritic Sm/Nd, but a lower-than-terrestrial 142Nd/144Nd ratio. We may still not know the ultimate solution to these puzzles, but whatever it is, it will have a profound impact on our understanding of the early history of our planet. For the past several years, Rick and his postdoc Jonathan O’Neil have been leading an effort to unravel Earth’s Hadean history. Hades is the god of the Underworld; its entrance, the 4 billion year age barrier, is guarded by a three-headed dog called Cerberus. It’s hard to get in (unless you are dead!), and it’s even harder to get anything back out of it. You may know what happened to Orpheus! Anyway, rather than trying to decipher ancient zircons from younger rocks, Rick’s group looked at the Nd isotopic composition of the Nuvvuagittuq greenstone belt in eastern Canada, and found that these rocks yield an apparent 142Nd/144Nd vs. Sm/Nd isochron with an age of 4.27 Gyr (O’Neil et al., 2008). Although it is still being debated whether this age dates the actual emplacement of these rocks or the age of differentiation of their mantle sources, this is the first time that the 4 Gyr age barrier has been breached by anything other than the detrital or xenocrystic zircon grains preserved in much younger rocks. Subsequently, O’Neil and Carlson (2017) showed that a large block of Archean continental crust in northeastern Canada has inherited some of this much more ancient 142Nd variability, and this led them to the conclusion that much of this younger Archean crust was generated by remelting of a >4.2 billion year old (“Hadean”) basaltic protocrust having the same isotopic and chemical characteristics as the Nuvvuagittuq greenstone belt they had previously analyzed. Thus, largely as a result of the research of Carlson and his collaborators, the Hadean (>4 billion year) evolution of our planet is gradually being unraveled. Rick Carlson has, in my opinion, one remarkable weakness: His fondness of fancy cars and car racing. Before I knew him better, I initially thought this was completely out of character. Why would Rick go out racing some vintage Corvette??? Now that he invited me to take a ride in his Tesla, I think I am beginning to understand. Particularly the part when you call your Tesla from your restaurant table and tell it: “Come and get me!” Ladies and Gentlemen, it is my honor to present Rick Carlson, a scientist nearly at the pinnacle of his career, recipient of GSA’s Day Medal, AGU’s Bowen Award, member of the National Academy of Sciences, Fellow of the American Academy of Arts and Sciences, director of Carnegie’s Earth and Planets Laboratory, and Tesla driver, to receive the Victor Moritz Goldschmidt Award of the Geochemical Society.

0

PMC9749616

NO-CC CODE

2022-12-15 23:23:20

no

Geochim Cosmochim Acta. 2021 Apr 1; 298:258-259

utf-8

Geochim Cosmochim Acta

10.1016/j.gca.2021.01.025

oa_other

==== Front J Thromb Thrombolysis J Thromb Thrombolysis Journal of Thrombosis and Thrombolysis 0929-5305 1573-742X Springer US New York 2735 10.1007/s11239-022-02735-0 Article Association between thromboembolic events and COVID-19 infection within 30 days: a case–control study among a large sample of adult hospitalized patients in the United States, March 2020–June 2021 http://orcid.org/0000-0003-3137-8594 Huang Ya-Lin A. yhuang@cdc.gov Yusuf Hussain Adamski Alys Hsu Joy Baggs James Auf Rehab Adjei Stacey Stoney Rhett Hooper W. Craig Llata Eloisa Koumans Emilia H. Ko Jean Y. Romano Sebastian Boehmer Tegan K. Harris Aaron M. grid.416738.f 0000 0001 2163 0069 COVID-19 Response Team, Centers for Disease Control and Prevention, 1600 Clifton Road NE Mailstop US8-4, Atlanta, GA 30329 USA 14 12 2022 16 17 11 2022 © This is a U.S. Government work and not under copyright protection in the US; foreign copyright protection may apply 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The association between thromboembolic events (TE) and COVID-19 infection is not completely understood at the population level in the United States. We examined their association using a large US healthcare database. We analyzed data from the Premier Healthcare Database Special COVID-19 Release and conducted a case–control study. The study population consisted of men and non-pregnant women aged ≥ 18 years with (cases) or without (controls) an inpatient ICD-10-CM diagnosis of TE between 3/1/2020 and 6/30/2021. Using multivariable logistic regression, we assessed the association between TE occurrence and COVID-19 diagnosis, adjusting for demographic factors and comorbidities. Among 227,343 cases, 15.2% had a concurrent or prior COVID-19 diagnosis within 30 days of their index TE. Multivariable regression analysis showed a statistically significant association between a COVID-19 diagnosis and TE among cases when compared to controls (adjusted odds ratio [aOR] 1.75, 95% CI 1.72–1.78). The association was more substantial if a COVID-19 diagnosis occurred 1–30 days prior to index hospitalization (aOR 3.00, 95% CI 2.88–3.13) compared to the same encounter as the index hospitalization. Our findings suggest an increased risk of TE among persons within 30 days of being diagnosed COVID-19, highlighting the need for careful consideration of the thrombotic risk among COVID-19 patients, particularly during the first month following diagnosis. Supplementary Information The online version contains supplementary material available at 10.1007/s11239-022-02735-0. Keywords COVID-19 Thromboembolic events Case–control study Healthcare database ==== Body pmcHighlights A strong association was found between thromboembolic events and COVID-19 diagnoses in a large U.S. healthcare database. The association was particularly pronounced for a COVID-19 diagnosis identified in a previous hospitalization Our findings suggest opportunity for better screening and medical interventions to prevent thromboembolic events and subsequent admission among COVID-19 patients Background Numerous reports have described the occurrence of coagulation events, including venous thromboembolism, among people with SARS-CoV-2 infection [1–5]. However, the association between thromboembolic events and COVID-19 infection is not completely understood at the population level in the United States. Using a large U.S. healthcare database, we examined the association between thromboembolic events and a COVID-19 diagnosis. Methods A case–control study was conducted using March 2020–June 2021 data from the Premier Healthcare Database Special COVID-19 Release (PHD-SR). The PHD-SR is an extensive hospital-based administrative all-payer database, which includes longitudinal patient data from > 900 facilities in the United States, representing approximately 20% of U.S. hospital admissions. Thromboembolic events (hereafter referred to as TE) were defined as the occurrence of any of the following: venous thromboembolism (deep vein thrombosis and/or pulmonary embolism), cerebral venous thrombosis, and portal vein thrombosis. The study population consisted of men and non-pregnant women ≥ 18 years of age. Further, cases were defined as persons with an inpatient ICD-10-CM diagnosis of TE between 3/1/2020 and 6/30/2021. Controls were inpatients without TE and matched 2:1 to the cases by sex, age (± 2 years), admission month, and facility; if there were more than two potential controls matching to a case, two were randomly selected (Supplementary Table 1 and 2). The index healthcare related encounter (hereafter referred to as encounter) date was defined as the date of the initial TE healthcare encounter for cases or a patient’s first matched encounter during the matching month (if a matched control had more than one encounter during that month) for controls. We compared the proportion of both cases and controls who had concurrent or prior COVID-19 diagnosis during either inpatient or non-hospital (including emergency department and other type of non-inpatient services, such as clinics, or rehabilitation/skilled nursing/hospice settings) encounters within 30 days of their index TE/non-TE encounter date; we also observed the distributions of timing and encounter type of the COVID-19 diagnosis. Using multivariable logistic regression, we assessed the association between TE occurrence and COVID-19 diagnosis, adjusting for sex, age groups, race/ethnicity, insurance type, urban vs rural location of the facility, U.S. Census region of the facility, and comorbidities associated with TE, which occurred within 6 months prior to the index hospitalization [6]. We also performed separate models to assess the effect of the timing of the COVID-19 diagnosis (concurrent with vs prior to TE); and the combination effect of the timing and the encounter type of the COVID-19 diagnosis (inpatient vs non-hospital encounter type). All ICD-10 and CPT codes used in this study are listed in Supplementary Table 3. All analyses were performed using SAS 9.4 (SAS Institute, Cary NC). Because of the use of secondary data, patient consent was not required for this study. This study was among those using PHD data determined by a CDC review to be research not involving human subjects. Results Among 227,343 cases and 454,686 controls, 34,500 (15.2%) cases and 47,021 (10.3%) controls had a concurrent or prior COVID-19 diagnosis within 30 days of their index TE or non-TE inpatient encounter (Table 1). Among 34,500 cases with a COVID-19 diagnosis, 83.1% were diagnosed with COVID-19 during the same encounter as their TE diagnosis, while 16.9% were diagnosed during a separate encounter up to 30 days prior. Among cases with a concurrent or prior COVID-19 diagnosis, 91.8% of the COVID-19 diagnoses occurred during inpatient encounters and 8.2% occurred during non-hospital encounters.Table 1 Number and percentage of COVID-19 diagnosis among cases and controls*, by timing and encounter type, Premier Healthcare Database Special COVID-19 Release, March 2020–June 2021 TE cases (n = 227,343) Non-TE controls (n = 454,686) n (%) n (%) Had concurrent or prior COVID-19 diagnosis within 30 days of TE or non-TE encounter  Yes 34,500 (15.2) 47,021 (10.3)  No 192,156 (84.5) 407,665 (89.7) Among those with COVID-19 diagnosis within 30 days of TE or non-TE encounter  Timing of the COVID-19 diagnosis   Same encounter as TE or non-TE 28,684 (83.1) 42,293 (89.9)   Encounter 1–30 days prior to TE or non-TE encounter 5,816 (16.9) 4728 (10.1)  Encounter type of the COVID-19 diagnosis   Inpatient or observation encounters 31,664 (91.8) 42,597 (90.6)   Non-hospital* encounters 2,836 (8.2) 4424 (9.4)  Encounter type and timing of the COVID-19 diagnosis   Inpatient/observation encounters of COVID-19 diagnosis same encounter as TE or non-TE 28,684 (83.1) 42,293 (89.9)   Inpatient/observation encounters of COVID-19 diagnosis 1–30 days prior to TE or non-TE encounter 2,980 (8.6) 304 (0.7)   Non-hospital* encounters of COVID-19 diagnosis 1–30 days prior to TE or non-TE encounter 2,836 (8.2) 4424 (9.4) TE thromboembolic events defined as the occurrence of any of the following: venous thromboembolism (deep vein thrombosis and/or pulmonary embolism), cerebral venous thrombosis, and portal vein thrombosis *Non-hospital encounters included emergency department and other type of non-inpatient services, such as clinics, or rehabilitation/skilled nursing/hospice settings Multivariable regression analysis showed a statistically significant association between a COVID-19 diagnosis and TE among cases compared to controls (adjusted odds ratio [aOR] 1.75, 95% CI 1.72–1.78; Table 2, Model 1). The association was more substantial if a COVID-19 diagnosis occurred 1–30 days prior to index hospitalization (aOR 3.00, 95% CI 2.88–3.13; Table 2, Model 2) compared to the same encounter as the index hospitalization. When considering both timing and encounter type, a COVID-19 diagnosis from a prior inpatient encounter showed a remarkedly strong association of TE occurrence compared to controls (aOR 21.77, 95% CI 19.31–24.55; Table 2, Model 3). Several covariates showed significant positive associations with TE, including female sex, age ≥ 85 years, non-Hispanic black race/ethnicity, and most comorbidities.Table 2 Multivariable analysis of demographic and clinical characteristics associated with occurrence of thromboembolic events, premier healthcare database special COVID-19 release, March 2020–June 2021 (n = 609,441) Model 1 Model 2 Model 3 aOR (95% CI) p–value aOR (95% CI) p–value aOR (95% CI) p–value Had concurrent or prior COVID-19 diagnosis within 30 days of TE or non-TE encounter  Yes 1.75 (1.72–1.78)  < .0001  No Reference Timing of the COVID-19 diagnosis  No COVID-19 diagnosis Reference  COVID-19 diagnosed during same encounter as TE or non-TE 1.61 (1.59–1.64)  < .0001  COVID-19 diagnosed1-30 days prior to TE or non-TE encounter 3.00 (2.88–3.13)  < .0001 Encounter type and timing of the COVID-19 diagnosis  No COVID-19 diagnosis Reference  Inpatient/observation encounters of COVID-19 diagnosis same encounter as TE or non-TE 1.61 (1.59–1.64)  < .0001  Inpatient/observation encounters of COVID-19 diagnosis 1–30 days prior to TE or non-TE encounter 21.77 (19.31–24.55)  < .0001  Non-hospital* encounters of COVID-19 diagnosis 1–30 days prior to TE or non-TE encounter 1.64 (1.56–1.73)  < .0001 Sex  Male 0.96 (0.95–0.97)  < .0001 0.96 (0.95–0.97)  < .0001 0.96 (0.95–0.97)  < .0001  Female Reference Reference Reference Age group (years)  18–49 Reference Reference Reference  50–64 0.92 (0.90–0.93)  < .0001 0.92 (0.91–0.94)  < .0001 0.92 (0.90–0.93)  < .0001  65–74 0.97 (0.95–0.99) 0.0068 0.97 (0.95–0.99) 0.0124 0.97 (0.95–0.99) < .0001  75–84 1.00 (0.98–1.02) 0.9826 1.00 (0.98–1.03) 0.8399 1.00 (0.98–1.02) 0.8593  ≥ 85 1.08 (1.06–1.11) < .0001 1.09 (1.06–1.21) < .0001 1.08 (1.05–1.11) < .0001 Race/ethnicity  Asian, non-Hispanic 0.79 (0.76–0.83) < .0001 0.79 (0.76–0.83) < .0001 0.79 (0.76–0.83) < .0001  Black, non-Hispanic 1.21 (1.20–1.23)  < .0001 1.21 (1.20–1.23)  < .0001 1.21 (1.20–1.23)  < .0001  Hispanic 0.94 (0.92–0.96)  < .0001 0.94 (0.92–0.95)  < .0001 0.94 (0.92–0.96)  < .0001  White, non-Hispanic Reference Reference Reference  Other 1.01 (0.98–1.04) 0.4265 1.01 (0.99–1.04) 0.3412 1.02 (0.99–1.04) 0.2747  Unknown 1.28 (1.24–1.33)  < .0001 1.28 (1.24–1.34)  < .0001 1.29 (1.24–1.34)  < .0001 Insurance type  Medicare 0.86 (0.83–0.88)  < .0001 0.86 (0.84–0.88)  < .0001 0.86 (0.83–0.88)  < .0001  Medicaid 0.93 (0.91–0.96)  < .0001 0.94 (0.91–0.96)  < .0001 0.93 (0.91–0.96)  < .0001  Managed care 0.99 (0.97–1.02) 0.6446 0.99 (0.97–1.02) 0.6495 1.00 (0.97–1.02) 0.6760  Commercial Reference Reference Reference  Self-pay 1.00 (0.96–1.04) 0.9818 1.00 (0.97–1.04) 0.8828 1.00 (0.96–1.04) 0.9791  Other 0.91 (0.88–0.94)  < .0001 0.91 (0.88–0.94)  < .0001 0.91 (0.87–0.94)  < .0001 Facility urbanicity  Urban 0.94 (0.92–0.95)  < .0001 0.94 (0.92–0.96)  < .0001 0.94 (0.92–0.95)  < .0001  Rural Reference Reference Reference Facility region  Midwest Reference Reference Reference  Northeast 1.03 (1.01–1.05) 0.0006 1.04 (1.02–1.05) 0.0001 1.03 (1.02–1.05) 0.0002  South 1.01 (0.99–1.02) 0.2682 1.01 (1.00–1.02) 0.2181 1.01 (1.00–1.02) 0.1950  West 1.05 (1.03–1.07) < .0001 1.05 (1.03–1.07) < .0001 1.05 (1.03–1.07) < .0001 Comorbidities associated with TE**  Thrombocytopenia 1.60 (1.57–1.63) < .0001 1.60 (1.57–1.63) < .0001 1.59 (1.57–1.62) < .0001  Hemorrhagic stroke 2.30 (2.21–2.40) < .0001 2.30 (2.21–2.40) < .0001 2.31 (2.21–2.40) < .0001  Ischemic stroke 0.99 (0.97–1.01) 0.4075 0.99 (0.97–1.01) 0.4240 0.99 (0.97–1.01) 0.3703  Myocardial infarction 1.25 (1.22–1.28)  < .0001 1.25 (1.23–1.28)  < .0001 1.25 (1.22–1.28)  < .0001  Other arterial thrombosis 2.79 (2.66–2.94)  < .0001 2.79 (2.66–2.94)  < .0001 2.79 (2.66–2.94)  < .0001  Meningitis, encephalitis, or other CNS 2.22 (2.07–2.37)  < .0001 2.22 (2.08–2.37)  < .0001 2.22 (2.07–2.37)  < .0001  Head or neck infection 1.41 (1.36–1.46)  < .0001 1.40 (1.35–1.45)  < .0001 1.39 (1.35–1.45)  < .0001  Prior venous thromboembolism 3.55 (3.50–3.61)  < .0001 3.55 (3.49–3.60)  < .0001 3.55 (3.49–3.60)  < .0001  Thrombophilia 3.48 (3.36–3.60)  < .0001 3.48 (3.37–3.60)  < .0001 3.47 (3.36–3.60)  < .0001  Malignancy 1.77 (1.74–1.79)  < .0001 1.77 (1.74–1.79)  < .0001 1.77 (1.74–1.79)  < .0001  Head injury 1.11 (1.07–1.15)  < .0001 1.11 (1.08–1.15)  < .0001 1.11 (1.07–1.15)  < .0001  Thyroid disorder 0.98 (0.96–0.99) 0.0006 0.97 (0.96–0.99) 0.0005 0.97 (0.96–0.99) 0.0003  Cardiovascular disease 1.26 (1.24–1.27)  < .0001 1.26 (1.24–1.27)  < .0001 1.26 (1.24–1.27)  < .0001  Hypertension 0.83 (0.82–0.84)  < .0001 0.83 (0.82–0.84)  < .0001 0.83 (0.82–0.84)  < .0001  Obesity 1.30 (1.29–1.32)  < .0001 1.30 (1.29–1.32)  < .0001 1.30 (1.29–1.32)  < .0001  Type 2 diabetes 0.86 (0.85–0.87)  < .0001 0.86 (0.85–0.87)  < .0001 0.86 (0.85–0.87)  < .0001  Hemorrhagic disorder 1.68 (1.63–1.73)  < .0001 1.68 (1.63–1.74)  < .0001 1.68 (1.63–1.73)  < .0001  Systemic lupus 1.11 (1.07–1.16)  < .0001 1.11 (1.06–1.15)  < .0001 1.11 (1.06–1.15)  < .0001  Renal disease 1.32 (1.31–1.34)  < .0001 1.33 (1.31–1.34)  < .0001 1.32 (1.31–1.34)  < .0001  Liver disease 1.48 (1.46–1.51)  < .0001 1.48 (1.46–1.51)  < .0001 1.48 (1.46–1.51)  < .0001 aOR adjusted odds ratio, CI confidence interval, CNS central nervous system Model 1 included cases and controls assessing the association of concurrent or prior COVID-19 diagnosis (exposure) and occurrence of thromboembolic events (outcome) adjusting for covariates. Model 2 included cases and controls assessing the association of timing of COVID-19 diagnosis to index encounter and occurrence of thromboembolic events adjusting for covariates. Model 3 included cases and controls assessing the association of the encounter type and timing of COVID-19 diagnosis and occurrence of thromboembolic events adjusting for covariates. Covariates adjusted for in all three models included sex, age group, race/ethnicity, insurance type, urban vs rural location of the facility, U.S. Census region of the facility, thrombocytopenia, hemorrhagic stroke, ischemic stroke, myocardial infarction, other arterial thrombosis, meningitis, encephalitis, other CNS conditions, head or neck infection, prior venous thromboembolism, thrombophilia, malignancy, head injury, thyroid disorder, cardiovascular disease, hypertension, obesity, type 2 diabetes, hemorrhagic disorder, systemic lupus erythematosus (SLE) or other connective tissue disorder, renal disease, and liver disease *Non-hospital encounters included emergency department and other type of services not considered as inpatient services **Comorbidities were identified within 6 months prior to the admission of interest Discussion This case–control study using a large U.S. hospital-based data source demonstrated a strong association between TE and prior (within 30 days) or concurrent COVID-19 diagnoses, consistent with data from other studies [3, 7–9]. The association was particularly pronounced for a COVID-19 diagnosis identified in a previous hospitalization. Although in the majority of cases in our study a COVID-19 diagnosis and TE occurred during the same encounter, about 9% of the TE hospitalizations were among patients who were previously admitted with a COVID-19 diagnosis, representing an opportunity for better screening and medical intervention to prevent TE and subsequent readmission. As another consideration, the potential benefits of prolonged TE prophylaxis for selected patients hospitalized with COVID-19 could be assessed. Additionally, patients seen in non-hospital settings with a COVID-19 diagnosis and subsequently admitted to hospital may need additional clinical considerations related to TE prevention based on the potential additional risk that the COVID-19 infection may have conferred. Future studies could assess the relationship between the use and timing of anticoagulation therapy and risk for subsequent TE. We also observed that the association between COVID-19 and TE was higher among non-Hispanic black people than other racial/ethnical groups, a finding also seen in other studies [8, 9]. Racial disparities in thrombotic risk were found in the pre-COVID-19 era, and COVID-19 infection may have aggravated the gap [10]. Racial differences in the prevalence and severity of COVID-19 have been documented in the United States, possibly due to a higher rate of comorbidities and social disparities in access to health care in black populations [11, 12]. It is important to note that higher ORs among non-Hispanic black persons were found even after controlling for several comorbidities associated with TE. While comorbidities associated with TE were controlled for in our analysis, further confirmatory investigation to better understand other conditions not controlled for is warranted. The adjusted ORs for TE were slightly lower among Medicare and Medicaid patients in this study. We are not sure why this may be; however, the finding merits further investigation. This study has limitations. First, COVID-19 and TE were defined by ICD-10 diagnosis codes, and some may have been misclassified. Second, our generalizability is limited as our study population was primarily patients diagnosed with COVID-19 during hospitalization prior to or concurrent with the TE admission; the relationship between potentially less severe COVID-19 disease (e.g., patients not requiring hospitalization) and TE risk could be assessed further. Third, although we stratified the associations between TE and COVID-19 by hospital vs non-hospital encounter type, we didn’t quantify the risk based on COVID-19 severity of disease or inflammatory markers. Future cohort studies are needed to understand what factors predispose hospitalized COVID-19 patients to develop TE and thus could benefit from preventive measures. Finally, causality between COVID-19 diagnosis and TE cannot be inferred from our findings due to the case–control design. The strengths of this study are that it is one of the largest U.S.-based studies on this topic that included > 225,000 patients who experienced a thromboembolic event and twice as many controls who were matched based on several characteristics. This study also controlled for several comorbidities known to increase the risk of TE. In conclusion, our study suggests an increased risk of TE among persons within 30 days of being diagnosed COVID-19. These findings highlight the need for careful consideration of the thrombotic risk among COVID-19 patients, particularly during the first month following diagnosis, as coagulation events can result in worse outcomes. It is essential to implement evidence-based public health prevention strategies, including COVID-19 vaccination, to reduce the risk of COVID-19 and associated morbidity and mortality, as well as measures to prevent TE among patients hospitalized with COVID-19. In this context, it is also important to note that in January of 2022, the Advisory Committee for Immunization Practices (ACIP) recommended that primary and booster doses of COVID-19 vaccinations among persons ≥ 18 years of age be administered preferentially using mRNA vaccines. As more patients have been vaccinated, future studies could assess the effect of vaccination on reducing the TE from COVID-19 infection. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 80 kb) Funding No funding external to the Centers for Disease Control and Prevention was provided for this study. Declarations Conflict of interest No conflicts of interest were reported by any authors. Disclaimer The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Acharya Y Alameer A Calpin G Alkhattab M Sultan S A comprehensive review of vascular complications in COVID-19 J Thromb Thrombolysis 2021 53 586 593 10.1007/s11239-021-02593-2 34724155 2. Tan BK Mainbourg S Friggeri A Arterial and venous thromboembolism in COVID-19: a study-level meta-analysis Thorax 2021 76 10 970 979 10.1136/thoraxjnl-2020-215383 33622981 3. Helms J Tacquard C Severac F High risk of thrombosis in patients with severe SARS-CoV-2 infection: a multicenter prospective cohort study Intensive Care Med 2020 46 6 1089 1098 10.1007/s00134-020-06062-x 32367170 4. Centers for Disease Control and Prevention. Science brief: evidence used to update the list of underlying medical conditions associated with higher risk for severe COVID-19. https://www.cdc.gov/coronavirus/2019-ncov/science/science-briefs/underlying-evidence-table.html. Accessed June. 5. MacNeil JR Su JR Broder KR Updated recommendations from the advisory committee on immunization practices for use of the Janssen (Johnson & Johnson) COVID-19 vaccine after reports of thrombosis with thrombocytopenia syndrome among vaccine recipients: United States, April 2021 MMWR Morb Mortal Wkly Rep 2021 70 17 651 656 10.15585/mmwr.mm7017e4 33914723 6. Otite FO Patel S Sharma R Trends in incidence and epidemiologic characteristics of cerebral venous thrombosis in the United States Neurology 2020 95 16 e2200 e2213 10.1212/WNL.0000000000010598 32847952 7. Klok FA Kruip M van der Meer NJM Incidence of thrombotic complications in critically ill ICU patients with COVID-19 Thromb Res 2020 191 145 147 10.1016/j.thromres.2020.04.013 32291094 8. Go AS Reynolds K Tabada GH COVID-19 and risk of VTE in ethnically diverse populations Chest 2021 160 4 1459 1470 10.1016/j.chest.2021.07.025 34293316 9. Metra B Summer R Brooks SE George G Sundaram B Racial disparities in COVID-19 associated pulmonary embolism: a multicenter cohort study Thromb Res 2021 205 84 91 10.1016/j.thromres.2021.06.022 34274560 10. Chaudhary R Bliden KP Kreutz RP Race-Related disparities in COVID-19 thrombotic outcomes: beyond social and economic explanations EClinicalMedicine 2020 29 100647 10.1016/j.eclinm.2020.100647 33251501 11. Romano SD Blackstock AJ Taylor EV Trends in racial and ethnic disparities in COVID-19 hospitalizations, by region: United States, March–December 2020 MMWR Morb Mortal Wkly Rep 2021 70 15 560 565 10.15585/mmwr.mm7015e2 33857068 12. Price-Haywood EG Burton J Fort D Seoane L Hospitalization and mortality among black patients and white patients with Covid-19 N Engl J Med 2020 382 26 2534 2543 10.1056/NEJMsa2011686 32459916

36515793

PMC9749619

NO-CC CODE

2022-12-15 23:23:20

no

J Thromb Thrombolysis. 2022 Dec 14;:1-6

utf-8

J Thromb Thrombolysis

10.1007/s11239-022-02735-0

oa_other

==== Front Russ J Org Chem Russian Journal of Organic Chemistry 1070-4280 1608-3393 Pleiades Publishing Moscow 3763 10.1134/S1070428022100086 Article Synthesis of Schiff Bases and Isoindolyl- and Thiazolyl-Substituted Quinolines from 6-Amino-2-methylquinolin-4-ol https://orcid.org/0000-0002-4039-2323 Aleqsanyan I. L. ialeksanyan@ysu.am https://orcid.org/0000-0003-1210-0052 Hambardzumyan L. P. grid.21072.36 0000 0004 0640 687X Yerevan State University, 375025 Yerevan, Armenia 14 12 2022 2022 58 10 14341437 20 1 2022 20 4 2022 26 4 2022 © Pleiades Publishing, Ltd. 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Reactions of 6-amino-2-methylquinolin-4-ol with salicylaldehyde, phthalic anhydride, phenyl isothiocyanate, and ammonium thiocyanate afforded 6-[(2-hydroxybenzylidene)amino]-2-methylquinolin-4-ol, 2-(4-hydroxy-2-methylquinolin-6-yl)-1H-isoindole-1,3(2H)-dione, N-(4-hydroxy-2-methylquinolin-6-yl)-N′-phenylthiourea, and 1-(4-hydroxy-2-methylquinolin-6-yl)thiourea, respectively. Heterocyclizations of the latter with ethyl bromoacetate and bromacetophenone led to the formation of 2-[(4-hydroxy-2-methylquinolin-6-yl)­imino]-1,3-thiazolidin-4-one and 2-methyl-6-[(4-phenyl-1,3-thiazol-2(3H)-ylidene)amino]quinolin-4-ol, respectively. Keywords: quinoline thiourea phenylthiourea aminoquinoline phenyl isothiocyanate phthalic anhydride bromoacetophenone ethyl bromoacetate isoindole thiazolidine issue-copyright-statement© Pleiades Publishing, Ltd. 2022 ==== Body pmcINTRODUCTION Quinoline and its derivatives constitute an important class heterocyclic compounds that are promising for the design of new drugs, including those for the treatment of COVID-19 [1–6]. The quinoline scaffold could give rise to a broad spectrum of biological activity [7–9], such as antimicrobial, antiviral, anti­protozoal, antimalarial, antitumor, cardiovascular, psycho­tropic, antioxidant, anticonvulsant, analgesic, anti-inflammatory, anthelminthic, etc. [8]. Numerous methods have been developed for the synthesis of quinoline and its derivatives. The goal of the present work was to synthesize new quinoline derivatives containing five-membered heterocyclic fragments starting from 6-amino-2-methylquinolin-4-ol. RESULTS AND DISCUSSION In continuation of our studies on the synthesis of biologically active compounds, herein we report the synthesis of new Schiff bases and isoindole-1,3-dione and thiazole derivatives containing a quinoline ring using 6-amino-2-methylquinolin-4-ol (1) as starting material. The reaction of aminoquinoline 1 with salicylaldehyde in boiling ethanol gave 6-[(2-hydroxy­benzylidene)amino]-2-methylquinolin-4-ol (2). Com­pound 1 reacted with phthalic anhydride in dioxane–acetic acid (5:1) under reflux conditions to produce 2-(4-hydroxy-2-methylquinolin-6-yl)-1H-isoindole-1,3(2H)-dione (3). N-(4-Hydroxy-2-methylquinolin-6-yl)-N′-phenylthiourea (4) was synthesized in a good yield by the reaction of aminoquinoline 1 with an equimolar amount of phenyl isothiocyanate in boiling ethanol. The reaction of 1 with ammonium thiocyanate in aqueous medium in the presence of concentrated aqueous HCl at 150°C for 5–6 h afforded N-(4-hydroxy-2-methylquinolin-6-yl)thiourea (5) (Scheme 1). Scheme 1. Taking into account functional potential of thiourea 5, it was reacted with ethyl bromoacetate and bromo­acetophenone in the presence of sodium acetate in anhydrous ethanol. These reactions led to the formation of 2-[(4-hydroxy-2-methylquinolin-6-yl)imino]-1,3-thiazolidin-4-one (6) and 2-methyl-6-{[4-phenyl-1,3-thiazol-2(3H)-ylidene]amino}quinolin-4-one (7), respectively (Scheme 2). Scheme 2. EXPERIMENTAL The 1H and 13C NMR spectra were recorded on a Varian Mercury-300 spectrometer (Germany) in DMSO-d6–CCl4 (1:3). The progress of reactions and the purity of the isolated compounds were monitored by TLC on Alugram® XtraSIL G UV254 plates (Germany) using iodine vapor for visualization. All solvents were distilled just before use, and com­mercially available reagents were purchased from Merck (Darmshtadt, Germany) and/or its branches. 6-[(2-Hydroxybenzylidene)amino]-2-methyl­quinolin-4-ol (2). A mixture of 0.174 g (1 mmol) of compound 1 [10], 10 mL of methanol, 0.122 g (1 mmol) of salicylaldehyde, and one drop of concen­trated aqueous HCl was refluxed with stirring for 7 h. The solvent was distilled off, the residue was dissolved in a dilute alkali solution, the solution was filtered, and the filtrate was acidified to pH 5.0–5.5. The precipitate was filtered off and washed with water. Yield 0.20 g (72%), mp 326–327°C, Rf 0.52 (EtOH–xylene, 1:1). 1H NMR spectrum, δ, ppm: 2.63 s (3H, CH3), 6.73 s (1H, Harom), 7.11–7.16 m (2H, Harom), 7.34 d (1H, Harom, J = 2.4 Hz), 7.43 d.d (1H, Harom, J = 9.0, 2.5 Hz), 7.51–7.59 m (2H, Harom), 7.88 d (1H, Harom, J = 9.0 Hz), 8.01 d (1H, Harom, J = 8.3 Hz), 9.82 s (1H, NH), 10.83 br.s (1H, OH). Found, %: C 73.20; H 5.14.; N 10.23. C17H14N2O2. Calculated, %: C 73.38; H 5.04; N 10.07. 2-(4-Hydroxy-2-methylquinolin-6-yl)-1H-isoin­dole-1,3(2H)-dione (3) was synthesized according to the procedure described in [11]. A mixture of 0.174 g (1 mmol) of compound 1 [10], 10 mL of dioxane, 2 mL of acetic acid, and 0.18 g (1.2 mmol) of phthalic anhydride was refluxed with stirring for 3 h. After cooling, the precipitate was filtered off and washed with dioxane. Yield 0.27 g (89%), mp 350°C (decomp.), Rf 0.57 (EtOH–xylene, 1:1.5). 1H NMR spectrum, δ, ppm: 2.29 s (3H, CH3), 5.92 s (1H, Harom), 7.36 d (1H, Harom, J = 8.9 Hz), 7.43 d.d (2H, Harom, J = 11.1, 3.9 Hz), 7.74–7.87 m (2H, Harom), 7.97 s (1H, Harom), 8.41 d (1H, Harom, J = 2.3 Hz), 11.89 br.s (1H, OH). Found, %: C 71.18; H 3.79; N 9.37. C18H12N2O3. Calculated, %: C 71.05; H 3.95; N 9.21. N-(4-Hydroxy-2-methylquinolin-6-yl)-N′-phenylthiourea (4). A mixture of 0.87 g (5 mmol) of compound 1 [10], 10 mL of ethanol, and 0.675 g (0.6 mL, 5 mmol) of phenyl isothiocyanate was refluxed with stirring for 6 h. After cooling, the precipitate was filtered off and washed with ethanol. Yield 1.30 g (85%), mp 325°C (decomp.), Rf 0.60 (EtOH–xylene, 1:2.5). 1H NMR spectrum, δ, ppm: 2.35 d (3H, CH3, J = 0.7 Hz), 5.81 s (1H, Harom), 7.06–7.12 m (1H, Harom), 7.26–7.34 m (2H, Harom), 7.43 d (1H, Harom, J = 8.8 Hz), 7.52–7.57 m (2H, Harom), 7.88 d.d (2H, Harom, J = 8.8, 2.2 Hz), 7.97 d (1H, Harom, J = 2.5 Hz), 9.87 br.s (2H, NH), 11.39 br.s (1H, OH). 13C NMR spectrum, δC, ppm: 19.23, 39.49, 107.67, 117.16, 118.42, 123.13, 123.69, 124.50, 127.81, 128.26, 134.39, 137.07, 139.30, 148.25, 176.09, 179.43. Found, %: C 66.18; H 4.71; N 13.43; S 10.20. C17H15N3OS. Calculated, %: C 66.02; H 4.85; N 13.59; S 10.36. N-(4-Hydroxy-2-methylquinolin-6-yl)thiourea (5). A mixture of 1.74 g (10 mmol) of compound 1 [10], 30 mL of water, 2.5 mL of aqueous HCl (pH ~ 2.0), and 2.28 g (30 mmol) of ammonium thio­cyanate was heated with stirring at ~150°C for 5–6 h. After cooling, the precipitate was filtered off and washed with water. Yield 1.51 g (65%), mp 242–243°C, Rf 0.52 (EtOH–xylene, 1:2). 1H NMR spectrum, δ, ppm: 2.74 s (3H, CH3), 6.95 s (1H, Harom), 7.62 br.s (2H, NH2), 7.99 d (1H, Harom, J = 9.1 Hz), 8.16 d.d (1H, Harom, J = 9.1, 2.5 Hz), 8.52 d (1H, Harom, J = 2.5 Hz), 10.50 s (1H, NH), 14.72 br.s (1H, OH). 13C NMR spectrum, δC, ppm: 19.59, 39.39, 39.78, 40.06, 40.33, 95.45, 105.54, 113.44, 118.95, 119.72, 129.42, 135.85, 138.25. Found, %: C 56.78; H 4.69; N 18.12; S 13.87. C11H11N3OS: Calculated, %: C 56.65; H 4.72; N 18.03; S 13.73. 2-[(4-Hydroxy-2-methylquinolin-6-yl)imino]-1,3-thiazolidin-4-one (6). A mixture of 0.233 g (1 mmol) of compound 5, 10 mL of anhydrous ethanol, 0.246 g (3 mmol) of anhydrous sodium acetate, and 0.22 g (0.15 mL, 1.3 mmol) of ethyl bromoacetate was re­fluxed with stirring for 5–6 h. After cooling, the precipitate was filtered off, washed with ethanol, and dried. Yield 0.23 g (85%), mp 375°C (decomp.), Rf 0.50 (EtOH–xylene, 1:3). 1H NMR spectrum, δ, ppm: 2.31 s (3H, CH3), 3.96 t (2H, CH2, J = 19.1 Hz), 5.89 s (1H, Harom), 7.20–7.70 m (2H, Harom), 7.93 d (1H, Harom, J = 8.4 Hz), 11.26 br.s (1H, NH), 11.60 br.s (1H, OH). Found, %: C 57.26; H 4.89; N 15.23; S 11.59. C13H11N3O2S. Calculated, %: C 57.14; H 4.72; N 15.38; S 11.72. 2-Methyl-6-{[4-phenyl-1,3-thiazol-2(3H)ylidene]amino}quinolin-4-ol (7). A mixture of 0.233 g (1 mmol) of compound 5, 10 mL of anhydrous ethanol, 0.246 g (3 mmol) of anhydrous sodium acetate, and 0.199 g (1 mmol) of bromoacetophenone was refluxed with stirring for 6–7 h. After cooling, the precipitate was filtered off, washed with ethanol, and dried. Yield 0.31 g (93%), mp 305–306°C, Rf 0.67 (EtOH–PhMe, 1:1). 1H NMR spectrum, δ, ppm: 2.30 s (3H, CH3), 5.82 s (1H, Harom), 7.25–7.34 m (2H, Harom), 7.40 d.d (2H, Harom, J = 10.3, 4.7 Hz), 7.49 d (1H, Harom, J = 8.9 Hz), 7.89 d.d (1H, Harom, J = 8.9, 2.7 Hz), 7.92–7.97 m (2H, Harom), 8.46 d (1H, Harom, J = 2.6 Hz), 10.41 s (1H, NH), 11.58 br.s (1H, OH). 13C NMR spec­trum, δC, ppm: 19.26, 39.23, 40.06, 40.33, 102.85, 107.45, 110.46, 118.54, 122.29, 125.23, 125.66, 127.50, 128.53, 134.50, 135.07, 136.72, 148.52, 149.99, 162.94, 176.21. Found, %: C 68.32; H 4.68; N 12.49; S 9.72. C19H15N3OS. Calculated, %: C 68.48; H 4.50; N 12.61; S 9.61. CONCLUSIONS The reactions of 6-amino-2-methylquinolin-4-ol with salicylaldehyde, phthalic anhydride, phenyl iso­thiocyanate, and ammonium thiocyanate have been found to produce 6-[(2-hydroxybenzylidene)amino]-2-methylquinolin-4-ol, 2-(4-hydroxy-2-methylquinolin-6-yl)-1H-isoindole-1,3(2H)-dione, N-(4-hydroxy-2-methylquinolin-6-yl)-N′-phenylthiourea, and N-(4-hy­droxy-2-methylquinolin-6-yl)thiourea, respectively. Methods have been developed for the synthesis of 2-[(4-hydroxy-2-methylquinolin-6-yl)imino]-1,3-thia­zolidin-4-one and 2-methyl-6-{[4-phenyl-1,3-thiazol-2(3H)-ylidene]amino}quinolin-4-ol via heterocycliza­tions of N-(4-hydroxy-2-methylquinolin-6-yl)thiourea with ethyl bromoacetate and bromoacetophenone, respectively. CONFLICT OF INTEREST The authors declare the absence of conflict of interest. ==== Refs REFERENCES 1 Man R.-J. Jeelani N. Zhou Ch. Yang Y.-Sh. Anti-Cancer Agents Med. Chem. 2021 21 825 10.2174/1871520620666200516150345 2 Zhou W. Wang H. Yang Y. Chen Z.S. Zou C. Zhang J. Drug Discovery Today 2020 25 2012 10.1016/j.drudis.2020.09.010 32947043 3 De Barros, C.M., Almeida, C.A.F., Pereira, B., Costa, K.C.M., Pinheiro, F.A., Maia, L.D.B., Trindade, C.M., Garcia, R.C.T., Torres, L.H., and Diwan, S., Pain Physician, 2020, vol. 23, no. 4S, p. S351. 4 Shah S. Das S. Jain A. Misra D.P. Negi V.S. Int. J. Rheum. Dis. 2020 23 613 10.1111/1756-185X.13842 32281213 5 Adeel A.A. Sudan J. Paediatr. 2020 20 4 10.24911/SJP.106-1587122398 32528194 6 Halcrow, P.W., Geiger, J.D., and Chen, X., Front. Cell. Dev. Biol., 2021, vol. 9, article ID 627639. 10.3389/fcell.2021.627639 7 Matada, B.S., Pattanashettar, R., and Yernale, N.G., Bioorg. Med. Chem., 2021, vol. 32, article ID 115973. 10.1016/j.bmc.2020.115973 8 Radini I.A.M. Khidre R.E. El-Telbani E.M. Lett. Drug Des. Discovery 2016 13 921 10.2174/1570180813666160712234454 9 Suresh N. Nagesh H.N. Sekhar K.V. Kumar A. Shirazi A.N. Parang K. Bioorg. Med. Chem. Lett. 2013 23 6292 10.1016/j.bmcl.2013.09.077 24138941 10 Rubtsov M.B. Bunina V.I. Zh. Obshch. Khim. 1944 14 1129 11 Aleksanyan I.L. Hambardzumyan L.P. Russ. J. Org. Chem. 2020 56 2114 10.1134/S1070428020120118

0

PMC9749625

NO-CC CODE

2022-12-15 23:23:20

no

Russ J Org Chem. 2022 Dec 14; 58(10):1434-1437

utf-8

oa_other

==== Front Russ J Org Chem Russian Journal of Organic Chemistry 1070-4280 1608-3393 Pleiades Publishing Moscow 3776 10.1134/S1070428022100219 Article A Convenient One-Pot Synthesis of Bis(indolyl)methane Derivatives and Evaluation of Their Nematicidal Activity against the Root Knot Nematode Meloidogyne incognita Rani M. 1 Utreja D. utrejadivya@yahoo.com 1 Dhillon N. K. 2 Kaur K. 1 1 grid.412577.2 0000 0001 2176 2352 Department of Chemistry, Punjab Agricultural University, 141027 Ludhiana, Punjab India 2 grid.412577.2 0000 0001 2176 2352 Department of Plant Pathology, Punjab Agricultural University, 141027 Ludhiana, Punjab India 14 12 2022 2022 58 10 15271533 11 2 2022 18 4 2022 22 4 2022 © Pleiades Publishing, Ltd. 2022 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Seven arylbis(indolyl)methanes were synthesized by electrophilic substitution reaction of aromatic aldehydes on indole using glacial acetic acid as catalyst in aqueous media under ultrasonic irradiation. The synthesized bis-indole derivatives were characterized using elemental analysis, 1H and 13C NMR, FT-IR spectroscopy, and mass spectrometry and were screened for their nematicidal activity against the root knot nematode Meloidogyne incognita. The efficiency of the synthesized compounds was evaluated in vitro by egg hatching and mortality tests. All tested compounds showed significant nematicidal potential, and the nitro substituted derivative, 3,3′-[(4-nitrophenyl)methylene]di(1H-indole), exhibited the highest activity. Keywords: indole glacial acetic acid bis(indolyl)methanes aromatic aldehydes nematicidal activity Meloidogyne incognita issue-copyright-statement© Pleiades Publishing, Ltd. 2022 ==== Body pmcINTRODUCTION Heterocycles constitute a wide class of organic compounds which contribute significantly in every facet of pure and applied chemistry [1]. Heterocyclic compounds have played an important role in bio­chemical industries as they are present in large propor­tion in biomolecules like vitamins, enzymes, biologi­cally active compounds, and natural products [2]. Indole derivatives have attracted considerable attention [3] due to the broad range of their biological properties such as antiviral [4], antibacterial [5], anticancer [6], antidepressant [7], and antifungal activities [8]. Derivatives of indole have a variety of medicinal uses as antihypertensive and antiparasitic agents, antidepressants, as well as in cardiology, neurology, and endocrinology [9]. One of the indole alkaloids is delavirdine which prevents human immunodeficiency virus (HIV) from proliferating in human body [10]. Arbidol is an antiviral agent which is used for the treatment of influenza viruses, and it also acts as an ef­ficient inhibitor of SARS-CoV-2 virus [11]. Rhopaladin is a marine indole alkaloid which showed repres­sive activity against c-ErbB-2 kinase and cyclin-depen­dent kinase 4 [12] (Fig. 1). Indole derivatives such as tadalafil, chaetoindolone A, eudistomin C, strepto­chlorin analogs [13], etc., have also been evaluated for the effective control of plant pathogens. Plant patho­gens cause damage to flora, fauna, and microorganisms [14]. Nowadays, the major threat to agriculture is the root-knot nematode of the Meloidogyne genus [15] out of which Meloidogyne incognita species majorly cause yield loss in different crops like tomato, brinjal, turmeric, melon, etc. [16] which damage crops worth $125 billion with a yield loss of 14% all over the world [17]. In India, annual loss in 30 crops caused by nematodes is in crores each year. Therefore, efforts have been made for the prevention of infectious nematodes. Fig. 1. Structures of some indole-containing drugs. Various reagents have been used for the synthesis of indole dimers. The reagents commonly used for the substitution reaction of indole with aromatic aldehydes are mineral acids like HCl, H2SO4, and HF, but these are listed as hazardous catalysts [18]. Herein, we report a convenient one-step reaction for the synthesis of bis-indolyl methanes using glacial acetic acid under ultra­sonication conditions as it has emerged as a new lead in green organic synthesis [19]. Also, the use of glacial acetic acid offers many advantages such as cost effec­tive­ness, easy availability, lower toxicity, air stability, and easy separation of products by simple filtration, thereby eradicating the necessity of purification protocols such as chromatography and liquid–liquid extractions which are very time-consuming [20]. As far as nematicidal activity [21] is concerned, only limited information is available in this regard. To bridge this gap, bis(indolyl)methanes have been synthesized from indole and aromatic aldehydes and examined for their nematicidal activity against the root knot nematode M. incognita [22]. RESULTS AND DISCUSSION Bis(indolyl)methane derivatives 3a–3g were syn­the­sized as shown in Scheme 1 by the reaction of indole (1) with aromatic aldehydes 2a–2g in water in the presence of acetic acid as catalyst at 40°C under ultrasonic irradiation. The crude products were recrys­tallized from ethanol to afford pure compounds 3a–3g in good yields. The structure of the synthesized com­pounds was confirmed by using various spectroscopic techniques, including 1H and 13C NMR, FT-IR spec­troscopy, and mass spectrometry. Compounds 3a–3g are colored crystalline solids that are stable in air and readily soluble in dimethyl sulfoxide, methylene chloride, chloroform, and other polar solvents. Scheme 1. The proposed mechanism of indole dimerization with aromatic aldehydes involves electrophilic substi­tu­tion at the 3-position of indole by the aldehyde carbonyl carbon atom in acidic environment to give intermediate A which is converted to another interme­diate B via elimination of water molecule. The addition of the second indole molecule to B produces protonated bis(indolyl)methane C whose deprotonation affords final product 3 (Scheme 2). Scheme 2. Indole (1) and its derivatives 3a–3g were evaluated for their nematicidal activity against Meloidogyne incognita at different concentrations (1500, 1000, 750, 500, 250, and 100 ppm) after exposure for 24, 48, 72, 96, and 120 h. The efficiency of compounds 1 and 3a–3g on the egg hatch inhibition of M. incognita at different concentrations and durations are shown in Figs. 2 and 3 (see also Supplementary Materials). The highest percent egg hatch inhibition was exhibited by the compounds at the maximum concentration (1500 ppm), and their efficiency declined as the con­cen­tration decreased (Fig. 2). Similarly, the maximum egg hatch inhibition was observed after 120 h of dura­tion, followed by 96, 72, 48, and 24 h (Fig. 3). Fig. 2. Effect of indole (1) and its derivatives 3a–3g on percent egg hatch inhibition of M. incognita at different concentrations. Fig. 3. Effect of indole (1) and its derivatives 3a–3g on percent egg hatch inhibition of M. incognita at different durations. Figures 4 and 5 illustrate the effect of indole (1) and its derivatives 3a–3g on the percent mortality of second stage juveniles (J2) of M. incognita at different con­centrations and durations. The observed pattern was similar to that of percent egg hatch inhibition, i.e., the percent mortality decreased with decrease in the con­centration and exposure time (Supplementary Table 2). Fig. 4. Effect of indole (1) and its derivatives 3a–3g on percent mortality of second stage juveniles of M. incognita at different concentrations. Fig. 5. Effect of indole (1) and its derivatives 3a–3g on percent mortality of second stage juveniles of M. incognita at different durations. The maximum percent mortality against second stage juvenile of M. incognita was seen after 120 h duration, followed by 96, 72, 48, and 24 h of exposure (Fig. 5). Thus, all the compounds showed both concen­tration and duration dependent manner for both percent egg hatch inhibition and percent mortality of root knot nematodes. 3,3′-[(4-Nitrophenyl)methyl]di(1H-indole) (3a) exhibited the highest percent egg hatch inhibition potential (96.83%) and maximum percent mortality potential (100.00%). For percent egg hatch inhibition, the order of efficiency was as follows: 3a > 3b > 3c > 3d > 3e > 3f > 1 > 3g which showed that 3a was the most effective and that 3g was the least effective. The same order was observed for percent mortality. Thus, the compounds having electron-withdrawing groups showed better nematicidal activity [23, 24]. EXPERIMENTAL The melting points were measured in open-end capil­laries with a Nutronics digital melting point apparatus. The reactions were carried out using a Helix Biosciences Ultra Sonicator (220 V, 700 W) from the Central Instrumentation Laboratory (Punjab Agricul­tural University, Ludhiana). The 1H and 13C NMR spectra were recorded at 25°C on a Bruker Avance II spectrometer (500 and 125 MHz, respectively) using CDCl3 or DMSO-d6 as solvent and tetramethylsilane (TMS) as internal standard. The IR spectra (400– 4000 cm–1) were recorded on a Perkin Elmer Spectrum Two FT-IR spectrometer from samples prepared as KBr pellets. Elemental analyses were obtained on a Thermo Scientific instrument (Department of Chemistry, Guru Nanak Dev University, Amritsar). General procedure for the synthesis of 3,3′-(aryl­methylene)di(1H-indoles) 3a–3g. A solution of indole (1, 0.138 mol, 0.162 g) in acetonitrile (5.0 mL) was added dropwise at room temperature to a solution of aromatic aldehyde 2a–2e (0.039 mol) in acetonitrile (5.0 mL). Water (20.0 mL) and a catalytic amount of acetic acid (0.5 mol %) were then added, and the mixture was irradiated at 40°C in an ultrasonicator. After completion of the reaction (2–8 h; TLC), the mixture was poured onto crushed ice, and the solid product was filtered off and recrystallized from hot methylene chloride. 3,3′-[(4-Nitrophenyl)methylene]di(1H-indole) (3a). Yield 78%, yellow crystalline solid, mp 220–222°C (from CH2Cl2), Rf 0.52 (EtOAc–hexane, 20:80). IR spectrum, ν, cm–1: 1219 (C–N), 1344 (NO2), 1515 (NO2), 1554 (C=Carom), 2858 (C–H), 3023 (C–Harom), 3397 (N–H). 1H NMR spectrum (CDCl3), δ, ppm: 5.92 s (1H, CH), 6.62 d (2H, J = 1.7 Hz, Harom), 6.94–6.98 m (2H, Harom), 7.11–7.14 m (2H, Harom), 7.28 d (2H, J = 7.95 Hz, Harom), 7.31 d (2H, J = 8.15 Hz, Harom), 7.36–7.39 q (1H, J = 7.9 Hz, Harom), 7.62 d (1H, J = 7.65 Hz, Harom), 7.94 s (2H, NH, D2O exchange­able), 8.01 d.d (1H, J = 1.3, 1.45 Hz, Harom), 8.14 t (1H, J = 1.7 Hz, Harom). 13C NMR spectrum (CDCl3), δC, ppm: 111.26, 118.23, 119.55, 121.51, 122.34, 123.63, 123.67, 126.65, 129.13, 134.87, 136.75, 146.38, 148.51. Mass spectrum: m/z 368.09 [M + 1]+. Found, %: C 75.19; H 4.66; N 11.44; O 8.71. C23H17N3O2. Calculated, %: C 74.09; H 4.26; N 11.34; O 8.21. M 367.13. 3,3′-[(3-Nitrophenyl)methylene]di(1H-indole) (3b). Yield 82%, yellow crystalline solid, mp 218–220°C (from CH2Cl2), Rf 0.40 (EtOAc–hexane, 20:80). IR spectrum, ν, cm–1: 1232 (C–N), 1338 (NO2), 1522 (NO2), 1551 (C=Carom), 2856 (C–H), 3019 (C–Harom), 3356 (N–H). 1H NMR spectrum (DMSO-d6), δ, ppm: 6.03 s (1H, Harom), 6.87–6.90 m (4H, Harom), 7.04– 7.07 m (2H, Harom), 7.29 d (2H, J = 7.95 Hz, Harom), 7.37 d (2H, J = 8.1 Hz, Harom), 7.61 d (2H, J = 8.7 Hz, Harom), 8.15 d.d (2H, J = 1.85, 1.32 Hz, Harom), 10.92 d (2H, J = 1.4 Hz, NH, D2O exchangeable). 13C NMR spec­trum (DMSO-d6), δC, ppm: 111.46, 116.55, 118.29, 118.79, 120.97, 123.30, 123.73, 126.24, 129.33, 136.47, 145.64, 153.02. Mass spectrum: m/z 368.23 [M + 1]+. Found, %: C 75.19; H 4.66; N 11.44; O 8.71. C23H17N3O2. Calculated, %: C 75.02; H 4.45; N 11.21; O 8.11. M 367.13. 3,3′-[(4-Fluorophenyl)methylene]di(1H-indole) (3c). Yield 75%, light brown crystalline solid, mp 78–80°C (from CH2Cl2), Rf 0.34 (EtOAc–hexane, 20:80). IR spectrum, ν, cm–1: 1224 (C–N), 1332 (C–F), 1521 (C=Carom), 2860 (C–H), 3013 (C–Harom), 3448 (N–H). 1H NMR spectrum (CDCl3), δ, ppm: 5.93 s (1H, CH), 6.61 d (2H, J = 2.2 Hz, Harom), 6.93–6.98 m (2H, Harom), 7.11–7.15 m (2H, Harom), 7.27 d (2H, J = 5.90 Hz, Harom), 7.32 d (2H, J = 12.3 Hz, Harom), 7.37 q (1H, J = 12.9 Hz, Harom), 7.63 d (1H, J = 7.60 Hz, Harom), 7.94 s (2H, NH, D2O exchangeable), 8.02 d.d (1H, J = 3.8, 0.54 Hz, Harom), 8.15 t (1H, J = 3.15 Hz, Harom). 13C NMR spectrum (CDCl3), δC, ppm: 112.26, 118.35, 119.75, 119.91, 121.13, 122.34, 123.75, 124.93, 126.15, 129.56, 134.67, 136.65, 146.68, 148.15. Mass spectrum: m/z 341.23 [M + 1]+. Found, %: C 81.16; H 5.03; F 5.58; N 8.23. C23H17FN2. Calculated, %: C 80.86; H 4.91; F 5.28; N 8.13. M 340.39. 3,3′-[(3,4-Dimethoxyphenyl)methylene]di(1H-indole) (3d). Yield 90%, light pink crystalline solid, mp 198–200°C (from CH2Cl2), Rf 0.53 (EtOAc–hexane, 20:80). IR spectrum, ν, cm–1: 1084 (C–O), 1198 (C–N) 1539 (C=Carom), 2831 (C–H), 3008 (C–Harom), 3240 (N–H). 1H NMR spectrum (CDCl3), δ, ppm: 3.74 s (3H, OMe), 3.83 s (3H, OMe), 5.82 s (1H, CH), 6.63 t (2H, J = 1.4 Hz, Harom), 6.75 d (1H, J = 8.25 Hz, Harom), 6.82 d.d (1H, J = 3.40 Hz, Harom), 6.92 d (1H, J = 1.95 Hz, Harom), 6.98–7.01 m (2H, Harom), 7.14–7.17 m (2H, Harom), 7.33 d (2H, J = 8.2 Hz, Harom), 7.39 d (2H, J = 7.95 Hz, Harom), 7.88 s (2H, NH, D2O exchangeable). 13C NMR spectrum (CDCl3), δC, ppm: 55.82, 55.85, 110.94, 111.04, 112.27, 119.23, 119.93, 119.97, 120.61, 121.93, 123.55, 127.10, 136.73, 136.75, 147.33, 148.71. Mass spectrum: m/z 383.29 [M + 1]+. Found, %: C 78.51; H 5.80; N 7.32; O 8.37. C25H22N2O2. Calculated, %: C 78.13; H 5.49; N 7.22; O 8.24. calculated: M 382.45. 3,3′-[(4-Methoxyphenyl)methylene]di(1H-indole) (3e). Yield 94%, light red crystalline solid, mp 180–182°C (from CH2Cl2), Rf 0.50 (EtOAc–hexane, 20:80). IR spectrum, ν, cm–1: 1029 (C–O), 1147 (C–N), 1488 (C=Carom), 2689 (C–H), 3051 (C–Harom), 3197 (N–H). 1H NMR spectrum (CDCl3), δ, ppm: 3.83 s (3H, OMe), 5.82 s (1H, CH), 6.62 t (2H, J = 2.85 Hz, Harom), 6.75 d (1H, J = 8.30 Hz, Harom), 6.82 d.d (1H, J = 1.95, 0.76 Hz, Harom), 6.93 d (1H, J = 7.05 Hz, Harom), 6.93–7.04 m (2H, Harom), 7.14–7.17 m (2H, Harom), 7.32 d (2H, J = 9.95 Hz, Harom), 7.38 d (2H, J = 4.20 Hz, Harom), 7.88 s (2H, NH, D2O exchangeable). 13C NMR spectrum (CDCl3), δC, ppm: 55.84, 110.47, 111.49, 113.67, 119.22, 119.39, 119.91, 121.61, 122.13, 124.75, 127.90, 137.63, 137.65, 147.13, 148.53. Mass spec­trum: m/z 353.41 [M + 1]+. Found, %: C 81.79; H 5.72; N 7.95; O 4.54. C24H20N2O. Calculated, %: C 81.87; H 5.65; N 7.45; O 4.41. M 352.43. 4-[Di(1H-indol-3-yl)methyl]-2-methoxyphenol (3f). Yield 88%, grey crystalline solid, mp 111–113°C (from CH2Cl2), Rf 0.43 (EtOAc–hexane, 20:80). IR spectrum, ν, cm–1: 1134 (C–O), 1214 (C–N), 1545 (C=Carom), 2899 (C–H), 3112 (C–Harom), 3342 (N–H), 3365 (O–H). 1H NMR spectrum (CDCl3), δ, ppm: 3.86 s (3H, OMe), 5.52 s (1H, CH), 5.79 s (1H, OH, D2O exchangeable), 6.68 d (2H, J = 1.50 Hz, Harom), 6.76 d (1H, J = 8.25 Hz, Harom), 6.84 d.d (1H, J = 2.05, 0.71 Hz, Harom), 6.92 d (1H, J = 2.05 Hz, Harom), 6.70 t (2H, J = 7.20 Hz, Harom), 7.14–7.17 m (2H, Harom), 7.34 d (2H, J = 8.15 Hz, Harom), 7.40 d (2H, J = 8.15 Hz, Harom), 7.90 s (2H, NH, D2O exchangeable). 13C NMR spectrum (CDCl3), δC, ppm: 55.93, 110.37, 110.96, 115.07, 119.18, 119.91, 119.96, 120.11, 121.87, 123.46, 127.10, 136.71, 144.93, 145.30. Mass spec­trum: m/z 369.19 [M + 1]+. Found, %: C 78.24; H 5.47; N 7.60; O 8.69. C24H20N2O2. Calculated, %: C 78.14; H 5.24; N 7.57; O 8.45. M 368.43. 4-[Di(1H-indol-3-yl)methyl]phenol (3f). Yield 78%, pink crystalline solid, mp 118–120°C (from CH2Cl2), Rf 0.47 (EtOAc–hexane, 20:80). IR spectrum, ν, cm–1: 1221 (C–N), 1322 (C–O), 1524 (C=Carom), 2864 (C–H), 3054 (C–Harom), 3247 (N–H), 3373 (O–H). 1H NMR spectrum (DMSO-d6), δ, ppm: 5.71 s (1H, CH), 6.66 d.d (2H, J = 8.55, 5.9 Hz, Harom), 6.78 d (2H, J = 1.80 Hz, Harom), 6.84–6.87 m (2H, Harom), 7.01–7.04 m (2H, Harom), 7.14 d (2H, J = 8.45 Hz, Harom), 7.26 d (2H, J = 8.11 Hz, Harom), 7.33 d (2H, J = 8.10 Hz, Harom), 9.11 s (1H, OH, D2O exchangeable), 10.75 d (2H, J = 1.65 Hz, NH, D2O exchangeable). 13C NMR spectrum (DMSO-d6), δC, ppm: 111.25, 114.62, 117.92, 118.54, 119.04, 120.63, 123.24, 126.54, 128.99, 135.07, 136.46, 155.13. Mass spec­trum: m/z 339.29 [M + 1]+. Found, %: C 81.63; H 5.36; N 8.28; O 4.73. C23H18N2O. Calculated, %: C 81.49; H 5.26; N 8.18; O 4.65. M 338.40. Nematicidal activity. a. Egg hatch inhibition assay. A pure culture of root knot nematodes was raised on the crop of brinjal. Five egg masses were taken and placed in a solution of a tested compound in water (5 mL) with a concentration of 1500, 1000, 750, 500, 250, and 100 ppm. A small quantity of acetone was added to dissolve the compounds in distilled water for preparing stock solutions which were then diluted to a required concentration. Distilled water with the same amount of acetone was used as control in each treat­ment, and three replications of each treatment were made. Egg hatching was observed after 24, 48, 72, 96, and 120 h, maintaining the temperature at 27±2°C. Statistical analysis was performed, and critical differ­ences were calculated. The percent hatch inhibition was calculated as (C – T)/C×100, where C is the number of nematodes in the control sample, and T is the number of nematodes after treatment. b. Second stage juvenile mortality assay. Freshly hatched second stage juveniles (J2) were taken for the mortality test. An average of 20 stage two juveniles were placed in 1 mL of distilled water, and 5 mL of a solution of each test compound prepared as in the hatching test was added with 4 replications each together with control. For each concentration, the results were recorded after 24, 48, 72, 96, and 120 h of exposure, and the percent ratio of the number of dead nematodes to the total number of nematodes was determined. CONCLUSIONS Arylbis(indolyl)methanes were synthesized by a one-step procedure from aromatic aldehydes and indole under ultrasonic irradiation. The proposed proce­dure is relatively easy, less time consuming, and greener than those reported previously. 3,3′-[(4-Nitro­phenyl)methylene]di(1H-indole) (3a) showed the best nematicidal activity against the root knot nematode M. incognita according to the egg hatch inhibition and second stage juvenile mortality assays. Arylbis­(indolyl)­methanes having electron-withdrawing groups were more effective. Supplementary information 11178_2022_3776_MOESM1_ESM.pdf Supplementary information The online version contains supplementary material available at 10.1134/S1070428022100219. ACKNOWLEDGMENTS The spectral data were obtained at the Sophisticated Analytical Instrumentation Facility (SAIF), Panjab University, Chandigarh. CONFLICT OF INTEREST The authors declare no conflict of interest. ==== Refs REFERENCES 1 Kalaria P.N. Karad S.C. Raval D.K. Eur. J. Med. Chem. 2018 158 917 10.1016/j.ejmech.2018.08.040 30261467 2 Thakral S. Singh V. Curr. Bioact. Compd. 2019 15 312 10.2174/1573407214666180614121140 3 Utreja D. Sharma S. Goyal A. Kaur K. Kaushal S. Curr. Org. Chem. 2020 23 2271 10.2174/1385272823666191023122704 4 Wei C. Zhang J. Shi J. Gan X. Hu D. Song B. J. Agric. Food Chem. 2019 67 13882 10.1021/acs.jafc.9b05357 31721582 5 Utreja D. Kaur J. Kaur K. Jain P. Mini-Rev. Org. Chem. 2020 17 991 10.2174/1570193X17666200129094032 6 Martins P. Jesus J. Santos S. Raposo L.R. Roma-Rodrigues C. Baptista P.V. Fernandes A.R. Molecules 2015 20 16851 10.3390/molecules200916852 7 Hamid H.A. Ramli A.N.M. Yusoff M.M. Front. Pharmacol. 2017 8 1 10.3389/fphar.2017.00096 28149278 8 Mi Y. Zhang J. Han X. Tan W. Miao Q. Cui J. Guo Z. Int. J. Biol. Macromol. 2021 181 581 10.1016/j.ijbiomac.2021.03.109 9 Zhang M.Z. Chen Q. Yang G.F. Eur. J. Med Chem. 2015 89 421 10.1016/j.ejmech.2014.10.065 25462257 10 Scott L.J. Perry C.M. Drugs 2000 60 1411 10.2165/00003495-200060060-00013 11152019 11 Blaising J. Polyak S.J. Pécheur E.I. Antiviral Res. 2014 107 84 10.1016/j.antiviral.2014.04.006 24769245 12 Yang C.G. Huang H. Jiang B. Curr. Org. Chem. 2004 8 1691 10.2174/1385272043369656 13 Abdellatif K.R. Lamie P.F. Omar H.A. J. Enzyme Inhib. Med. Chem. 2016 31 318 10.3109/14756366.2015.1022174 25798690 14 Zhang M.Z. Jia C.Y. Gu Y.C. Mulholland N. Turner S. Beattie D. Clough J. Eur. J. Med. Chem. 2017 126 669 10.1016/j.ejmech.2016.12.001 27936445 15 Ji X. Wang Z. Dong J. Liu Y. Lu A. Wang Q. J. Agric. Food Chem. 2016 64 9143 10.1021/acs.jafc.6b04020 27933985 16 Yan W. Zhao S.S. Ye Y.H. Zhang Y.Y. Zhang Y. Xu J.Y. Tan R.X. J. Nat. Prod. 2019 82 2132 10.1021/acs.jnatprod.8b01101 31329433 17 MesaValle, C.M., Garrido Cardenas, J.A., Carmona, C.J., Talavera, M., and Agugliaro, M.F., Agronomy, 2020, vol. 10, article nol. 1148. 10.3390/agronomy10081148 18 Mo L.P. Ma Z.C. Synth. Commun. 2005 35 1997 10.1081/SCC-200066653 19 Bedi P. Malkania L. Sharma S. Pramanik G. Pramanik T. Drug Invent. Today. 2019 11 1282 20 Van Eijk H.M. Bloemen J.G. Dejong C.H. J. Chromatogr. B 2009 877 719 10.1016/j.jchromb.2009.01.039 21 Kaur J. Utreja D. Dhillon N.K. Sharma S. Lett. Org. Chem. 2018 15 870 10.2174/1570178615666180330155049 22 Truong N.M. Nguyen C.N. Abad P. Quentin M. Favery B. Adv. Bot. Res. 2015 73 293 10.1016/j.arabjc.2017.10.001 23 Kaur J. Utreja D. Dhillon N.K. Sharma S. Lett. Org. Chem. 2019 16 759 10.2174/1570178616666190219131042 24 Kaur, K., Utreja, D., Dhillon, N.K., Pathak, R.K., and Singh, K., Pestic. Biochem. Physiol., 2021, vol. 171, article ID 104736. 10.1016/j.pestbp.2020.104736

0

PMC9749629

NO-CC CODE

2022-12-15 23:23:20

no

Russ J Org Chem. 2022 Dec 14; 58(10):1527-1533

utf-8

oa_other

==== Front Dysphagia Dysphagia Dysphagia 0179-051X 1432-0460 Springer US New York 10544 10.1007/s00455-022-10544-z Original Article Telehealth Management of Dysphagia in Adults: A Survey of Speech Language Pathologists’ Experiences and Perceptions Sevitz Jordanna S. 1 Bryan Jennine L. 2 Mitchell Samantha S. 2 Craig Bruce A. 3 Huber Jessica E. 2 Troche Michelle S. 1 http://orcid.org/0000-0002-7599-7728 Malandraki Georgia A. malandraki@purdue.edu 245 1 grid.21729.3f 0000000419368729 Laboratory for the Study of Upper Airway Dysfunction, Department of Biobehavioral Sciences, Teachers College, Columbia University, New York, NY USA 2 grid.169077.e 0000 0004 1937 2197 Department of Speech, Language, and Hearing Sciences, Purdue University, West Lafayette, IN USA 3 grid.169077.e 0000 0004 1937 2197 Department of Statistics, Purdue University, West Lafayette, IN USA 4 grid.169077.e 0000 0004 1937 2197 Weldon School of Biomedical Engineering, Purdue University, West Lafayette, IN USA 5 grid.169077.e 0000 0004 1937 2197 Purdue University, 715 Clinic Drive/Lyles-Porter Hall Rm.3152, West Lafayette, IN 47907 USA 14 12 2022 116 13 4 2022 16 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The goal of this study was to explore telehealth use for dysphagia management in response to COVID-19 to understand variables associated with clinician confidence and perceived effectiveness of this service delivery model and determine clinician-perceived benefits and challenges of managing dysphagia via telehealth. Speech-language pathologists (SLPs, n = 235) completed a web-based survey, providing information on demographics, telehealth use during the pandemic, and perspectives on current and future tele-management of dysphagia. Analyses included descriptive statistics to examine usage patterns; logistic regression to determine which variables were associated with telehealth use, clinician confidence, and perceived-effectiveness; and conventional content analysis to analyze responses to open-ended questions. Results revealed a sharp increase in the tele-management of dysphagia during the pandemic. Years of experience with dysphagia management (p = .031) and pre-pandemic use of telehealth (p < .001) were significantly associated with current use patterns. Working in the outpatient setting was associated with greater clinician confidence (p = .003) and perceived effectiveness (p = .007), and use of guidelines (p = .042) was also associated with greater clinician confidence. Key challenges identified included inadequate technological infrastructure, inadequate patient digital literacy, and reimbursement restrictions. Key benefits were treatment continuity, improving access to care, and time savings. The majority (67%) of respondents reported that they would use telehealth in the future. These findings demonstrate SLPs’ abilities and desire to expand their practice patterns to include telehealth for dysphagia management. Therefore, clinician training and more research on best practices for assessment and treatment of dysphagia via telehealth is warranted to refine models of care for dysphagia tele-management. Supplementary Information The online version contains supplementary material available at 10.1007/s00455-022-10544-z. Keywords Telehealth Dysphagia Management Survey COVID-19 Purdue University, College of Health and Human SciencesCOVID-19 Rapid Response grant Malandraki Georgia A. ==== Body pmcIntroduction At the onset of the SARS-CoV-2 (COVID-19) pandemic in March 2020, the healthcare landscape changed abruptly, and it became a global healthcare priority to mitigate viral transmission [1]. Given the close-contact and aerosol-generating nature of most dysphagia procedures, there was an urgent need to minimize in-person service delivery, and one way to do so was to adopt telehealth [2, 3]. This was further facilitated by the lifting of many federal, state, and international reimbursement and licensure restrictions at the onset of the pandemic [4, 5] that allowed for continued provision of dysphagia services without putting patients or providers at risk of contracting COVID-19. While telehealth quickly became widely used as a platform for dysphagia service delivery during this period, prior to the pandemic, this modality was not commonly used in dysphagia practice [6]. However, an amassing body of literature had already begun to support its use to manage dysphagia across adult populations. Indeed, completing a dysphagia-specific case-history via telehealth can be effective and efficient [7] and telehealth clinical swallowing evaluations with on-site facilitators have repeatedly been shown to be feasible and reliable in patients with a variety of diagnoses including head and neck cancer [8], stroke [9], and neurodegenerative disease [10], including patients with varying cognitive abilities [10]. Research has also supported that tele-clinical swallowing evaluations reduce wait times and costs [8], and are associated with excellent patient satisfaction ratings [8, 11]. Instrumental swallowing evaluations have also been explored using telehealth. Tele-videofluoroscopic swallow studies (tele-VFSS) directed in real-time by a remote clinician have demonstrated feasibility and reliability [12, 13] and in cases where a local dysphagia expert is not available, asynchronous teleconsultation based on VFSS assessments may play an important role in improving patient care [14]. There is also emerging evidence to support dysphagia treatment via telehealth [15]. In the head and neck cancer population, home-based exercise programs with telehealth treatment sessions have been associated with high levels of satisfaction and reduced costs of attendance for patients [16]. Applications and websites to facilitate home treatment and enhance patient adherence have also started to emerge with overall positive outcomes in regard to patient satisfaction, adherence, and functional outcomes [17–24]. For adults with neurogenic dysphagia, there is preliminary evidence to support the implementation of compensatory strategies [25] and exercise-based treatments [26] via telehealth, with reported improvements in physiologic function [27], swallowing performance [26], and treatment adherence [24, 26]. Many of the studies discussed above have utilized specialized equipment (e.g., cameras and microphones) and specific technology platforms which were not widely available to clinicians and patients during the COVID-19 pandemic. Providing additional equipment to patients was generally not feasible due to uncertainties regarding the way in which the novel coronavirus could spread and due to financial limitations. This was a possible barrier in translating the existing research to clinical practice during the pandemic. Despite the significant body of literature supporting the tele-management of dysphagia, its integration in clinical practice prior to the pandemic had been limited and many clinicians reported feeling unprepared to adopt this model when the pandemic started [6]. Clinician experience with and acceptance of telehealth is a crucial determinant of the implementation, expansion, success, and sustainability of telehealth services [28–30]. To date, clinician satisfaction with dysphagia tele-services has been assessed in few controlled studies [16, 31, 32]. However, to our knowledge, no extensive work has been done to help us broadly examine clinician experiences and perceptions with tele-management of dysphagia across settings, procedures, and patient-populations, and to explore variables associated with current or future telehealth usage for dysphagia care. This work is both critical and timely as telehealth starts to become part of standard healthcare [2, 3] – and has the potential to significantly improve access to care for patients with dysphagia [16, 33, 34]. Therefore, we aimed to complete an international survey of dysphagia clinicians to better understand telehealth usage patterns and clinicians’ experiences and perceptions using telehealth for dysphagia management during this time. Specifically, we aimed to answer 5 primary questions:To what extent were in-person dysphagia services disrupted during the COVID-19 pandemic? What were the primary usage patterns of telehealth (i.e., types of procedures, facilitator use, trainings) during this time? What clinician variables were significantly associated with telehealth use during the pandemic, and for those that used telehealth—what were the variables that were associated with clinician confidence and perceived effectiveness? What were the clinician-perceived benefits and challenges of using telehealth to manage dysphagia during and after the pandemic? What variables were significantly associated with clinicians reporting that they would or would not use telehealth in the future (i.e., after the pandemic)? Methods Participants Study participants completed a web-based anonymous survey via the Qualtrics online software [35] between January 15th and April 1st 2021. Participants were Speech Language Pathologists (SLPs) recruited through social media including Facebook, Twitter, and Instagram; emails to colleagues around the U.S. and internationally; and the American Speech-Language Hearing Association listserv for Special Interest Group 13, Swallowing and Swallowing Disorders. Survey enrollment was voluntary and participants were offered to enroll in a raffle to win one of two $50 vouchers by entering their email at the end of the survey. Inclusion criteria were (1) being a licensed SLP (in their country/area) or clinical fellow in Speech Language Pathology and (2) currently treating adult patients (> 18 years of age) for dysphagia. Development and Pilot Testing of Survey A pilot survey was first distributed to 12 SLPs who provided feedback regarding the survey quality and format to examine face validity. All field testers agreed that all questions were relevant and useful; however, minor wording suggestions were made to improve clarity. Survey Structure The survey (Online Appendix 1) consisted of 47 questions in three main categories with several subsections in each category. The first section titled “Demographics and Clinical Data” included 13 questions related to participant demographics and experience (n = 8), clinical data such as work setting and patient populations (n = 3), and prior experience with telehealth (n = 2). The second section titled “Dysphagia and Telehealth during COVID-19” included 27 questions pertaining to dysphagia management during the most and least restrictive periods of COVID-19. Most and least restrictive periods were not pre-specified time periods, but rather were defined according to local restrictions and interpreted individually for each survey respondent. The “most restrictive (MR) period” was described in the survey as “during the period that local or state government restrictions on in-person contact and movement were the strictest (e.g., May 2020 in NYC).” The “least restrictive (LR) period” was defined as “when most local and government restrictions were lifted, and clinicians were able to start seeing patients in-person regularly.” Questions in this section related to cancellation patterns of in-person procedures during the COVID-19 pandemic (n = 4), patterns of telehealth use during the pandemic (n = 9), use of guidelines and trainings (n = 4), clinician confidence and perceived-effectiveness of telehealth services (n = 8), and challenges of telehealth dysphagia management (n = 2). The final (third) section titled “Dysphagia and Telehealth after COVID-19” included seven questions regarding the likelihood of using telehealth in the future and clinician-perceived challenges and benefits of telehealth for dysphagia management. Statistical Analysis Descriptive statistics were used to answer research questions related to cancellations and telehealth usage patterns. All variables were categorical or ordinal and were summarized using frequencies and percentages. Across models, predictors included: years of experience managing dysphagia, prior experience with telehealth across any area of SLP, work setting, use of a facilitator, use of guidelines, completion of trainings, and use of telehealth during the pandemic. We did not consider the predictor “years of experience as an SLP” because it was highly correlated with “years of experience managing dysphagia” (Spearman ρ = 0.93), nor did we consider “prior experience with telehealth specifically for dysphagia” because few participants (n = 29) had dysphagia-specific telehealth experience prior to COVID-19. Binary logistic regression was used to determine which variables were significantly associated with the use of telehealth during the pandemic and projected future telehealth use. Ordinal logistic regression was used to determine which variables were significantly associated with clinician confidence and perceived effectiveness of telehealth to manage dysphagia during the pandemic. Statistical analyses were performed in R Version 4.0.1 [36]. A ranking system was used to analyze clinician-perceived challenges of telehealth. Nine possible challenges were presented in the survey, and clinicians ranked them in terms of their relative importance. A mean rank was calculated for each obstacle and the top three challenges are reported. Conventional content analysis [37] was used to answer research questions related to clinician-perceived benefits of telehealth and other free-response data, in which an inductive approach was used to develop themes from open-ended survey responses. Because this was an inductive process, a reliability analysis may dilute theme complexity [38] and therefore, this analysis was completed in pairs of raters. This allowed raters to develop consensus between the pairs and increase validity of theme identification. The raters familiarized themselves with the data and coded the data for key themes. This was done by tagging each response with a meaningful label or code (1–2 words) that represented it. The first author subsequently reviewed all codes to remove duplicates (if any), combine similar codes, and ensure the codes best represented the underlying constructs. The frequency (i.e., number of occurrences) of each code was calculated. The first author then categorized the codes into descriptive themes, with a focus on the quantification of trends and patterns reported by clinicians [39]. Results A total of 278 participants submitted the survey; however, 24 did not meet at least one of the inclusion criteria and 19 survey responses were removed due to responses deemed to be incomplete/suspicious. Specifically, 14 of 19 survey responses had identical IP addresses and responded “I don’t know” to most questions, and the remaining five had identical IP addresses with limited responses and total response time of less than three minutes. Thus, 235 surveys were included in the final analyses (Fig. 1). Questions were not mandatory; therefore, the number of survey respondents who answered each question varied and is reported in the results section.Fig. 1 Flowchart of data cleaning procedure Demographics of Survey Participants Survey respondents were Speech Language Pathologists (SLPs) (n = 224) or SLP clinical fellows (n = 10), 94.4% were females, with 61% of participants from the U.S. and the remaining 39% from 16 other countries. See Table 1 for detailed demographic and clinical background information.Table 1 Survey respondent demographics Case (percent) Age  21–30 years old 52 (22.1%)  31–40 years old 80 (34.0%)  41–50 years old 54 (23.0%)  51–60 years old 31 (13.2%)  > 60 years old 18 (7.6%) Gender  Female 222 (94.4%)  Male 10 (4.3%)  Other 0 (0%)  Prefer not to answer 3 (1.3%) Race  American Indian or Native Alaskan 1 (0.43%)  Asian or pacific Islander 12 (5.1%)  Black, not Hispanic 7 (3.0%)  White, not Hispanic 188 (80.3%)  Hispanic 11 (4.7%)  Prefer not to answer 15 (6.4%) Education  Bachelors 28 (11.9%)  Masters 166 (70.3%)  Doctorate 23 (9.7%)  Post-doctorate 9 (3.8%)  SLPD 5 (2.1%)  Other 2 (0.9%) Region  North America 162 (68.9%)  South America 24 (10.2%)  Europe 36 (15.3%)  Asia 1 (0.4%)  Australasia 5 (2.1%)  Middle east 3 (1.3%)  Africa 3 (1.3%) Years of experience as an SLP  < 1 year (CF) 10 (4.2%)  1–5 years 41 (17.4%)  6–10 years 51 (21.6%)  11–15 years 43 (18.2%)  > 15 years 88 (37.3%) Years of experience with dysphagia care  < 1 year 16 (6.8%)  1–5 years 44 (18.7%)  6–10 years 60 (25.5%)  11–15 years 48 (20.4%)  > 15 years 67 (28.5%) Work settinga  Acute care 124  Sub-acute rehab hospital 40  Skilled nursing facility 41  Outpatient 83  Private practice 42  University clinic 29  Home health 26  Other 4 Patient age rangea  Younger adults (18–39) 119  Middle-age adults (40–60) 188  Older adults (> 60) 222 Patient populationsa  Stroke 187  TBI 107  Neurodegenerative disease 192  Head and neck Cancer 133  Other 55 aSurvey respondents may belong to more than one grouping in this category (therefore percentages are not provided for this category) Research Question 1: Disruption to In-Person Services—Cancellation of in-Person Procedures During the COVID-19 Pandemic Cancellation patterns during the most and least restrictive periods of the COVID-19 pandemic are shown in Supplemental Table 1. Specifically, during the most restrictive (MR) period of the pandemic, 48.2% of respondents to this question (n = 112/232) reported cancelling clinical evaluations, 71.5% (n = 163/228) canceled instrumental evaluations, and 56% (n = 129/230) reported cancelling treatment sessions, at least half of the time (i.e., those that responded “about half the time,” “often,” or “always”). During the least restrictive (LR) period of the pandemic, 12.99% (n = 30/231) of participants canceled clinical evaluations, 18.7% (n = 43/230) canceled instrumental evaluations, and 13.36% (n = 31/232) canceled treatment sessions, at least half of the time (i.e., those that responded “about half the time,” “often,” or “always”). Conventional content analysis revealed seven key reasons for these cancellations including: (1) COVID restrictions (e.g., government regulations, facility closures, COVID testing requirements; MR n = 69; LR n = 20); (2) COVID risk (e.g., testing positive for COVID-19, high COVID-19 case counts, surge in infections; MR n = 39, LR n = 22); (3) COVID uncertainty (e.g., lack of clarity regarding how to safely carry out procedures, awaiting policy development and clearance; MR n = 9, LR n = 0); (4) insufficient resources (e.g., lack of personal protective equipment (PPE), lack of access to instrumental assessments; MR n = 25, LR n = 13); (5) patient concern (i.e., patient perceived risk of contracting COVID-19; MR n = 22, LR n = 21); (6) dysphagia services not being prioritized (e.g., dysphagia services were only completed in cases of emergency; MR n = 10, LR n = 4); and (7) barriers to telehealth (e.g., lack of equipment, clinician inexperience with this modality; MR n = 6, LR n = 1). Research Question 2: Telehealth Usage Patterns Before COVID-19 All survey respondents (n = 235) answered questions regarding telehealth use prior to COVID-19 (Fig. 2). When considering experience with telehealth across any area of SLP practice prior to the pandemic, 20.4% (n = 48) reported having some experience, 33.2% (n = 78) reported familiarity with research but no personal experience, and 46.4% (n = 109) reported no familiarity or experience at all. When considering using telehealth to manage dysphagia specifically prior to COVID-19, 12.3% (n = 29) of participants reported having some experience, 26.0% (n = 61) reported familiarity with research but no personal experience, and 61.7% (n = 145) reported no familiarity with research or experience.Fig. 2 Telehealth use prior to COVID-19 During COVID-19 A total of 50.2% (n = 118) of survey respondents reported using telehealth to manage dysphagia during the COVID-19 pandemic. Of those, 34.7% (n = 41) used telehealth in the inpatient setting, 89.0% (n = 105) in the outpatient setting, and 23.7% (n = 28) used telehealth in both the in- and out-patient settings. Details regarding the frequency with which clinicians used telehealth for different procedures during the most and least restrictive periods of the pandemic are reported in Fig. 3. Across settings, during the most restrictive period of the pandemic, telehealth was most frequently used for therapy sessions (n = 97), followed by consultations (n = 79), and then clinical (bedside) assessments (n = 62). Clinicians also reported using telehealth for other procedures (n = 27) such as family training, education, or counseling, consultations with other professionals, multidisciplinary clinics, case meetings, supervision meetings, telephone consultations, research visits, pre/post-operative counseling and troubleshooting, and interpreter services. Telehealth was least commonly used for instrumental assessments (n = 7). During the least restrictive period of the pandemic, a similar pattern was observed. Telehealth was most frequently used for therapy sessions (n = 88), followed by consultations (n = 69), clinical (bedside) assessments (n = 53), and other procedures as described above (n = 19). It was least commonly used for instrumental assessments (n = 9) during this time as well.Fig. 3 Frequency of telehealth dysphagia services provided during the pandemic Facilitator Use We defined a “facilitator” as a caregiver/nurse/aid who facilitated patients in the use of technology, environmental requirements (e.g., position of the patient and the cameras, lighting), feeding the patient (if needed), and was present during the session for safety reasons. Of the 131 participants who responded to this question, 16.8% (n = 22) reported always using a facilitator, 19.8% (n = 26) used a facilitator most of the time, 6.9% (n = 9) used a facilitator about half of the time, 34.4% (n = 45) used a facilitator sometimes, and 22.1% (n = 29) never used a facilitator. Facilitators most commonly were a caregiver relative (n = 81) and less commonly a caregiver aid (n = 25), a nurse (n = 22), or a physician (n = 1). Conventional content analysis of the related open text question revealed that the main components of the remote sessions with which facilitators helped were technology set-up and use (n = 52), food preparation (n = 20) and administering bolus trials (n = 29), clarification of instructions and cueing (n = 25), providing case history information (n = 19), facilitating equipment (e.g., holding cameras, providing patient with necessary devices/equipment, holding treatment devices, n = 13), assisting with implementation of strategies, home practice, and carry-over (n = 9), participating in patient/family education (n = 4), providing translator services (n = 3), and providing verification to the clinician regarding patient performance (n = 2). Guidelines and Trainings Of those who utilized telehealth to manage dysphagia (118/235), published guidelines were utilized by 53.4% (n = 63), and 55.1% (n = 65) reported completing specific trainings. The most commonly used guidelines were those provided on the American Speech Language Hearing Association (ASHA) Telepractice Portal and ASHA recommendations (n = 25), the Purdue I-EAT laboratory guidelines (n = 14), and facility-specific guidelines and recommendations (n = 17). Many other resources were also utilized, including the University of Queensland guidelines (n = 6) and country-specific resources, such as guidance by Speech Pathology Australia (SPA), the Brazilian Federal Council of Speech Therapy, the Irish Association of Speech and Language Therapists (IASLT), the Canadian Personal Health Information Act (PHIA), the College of Speech Language Pathologists and Audiologists of Ontario (CASLPO), the Alberta College of Speech Language Pathologists and Audiologists (ACSLPA), and Speech Pathology and Audiology Canada (SAC). Other respondents included resources or guidelines by other professional associations such as the Dysphagia Research Society (DRS), the Texas Speech and Hearing Association (TSHA), the Royal College of Speech and Language Therapists (RCSLT), the American Telemedicine Association (ATA), and peer-reviewed publications. Trainings most commonly included country-specific webinars or courses, such as those from ASHA (n = 19), the IASLT (n = 4), or other country or facility-specific trainings. Webinars from experts in the field (e.g., Dr. Malandraki) (n = 7), online resources/trainings from Purdue University (n = 4), the University of Queensland (n = 3), Australian-based webinars (n = 2), and other unspecified webinars were also reported. Clinician Confidence Participants who utilized telehealth to manage dysphagia were asked to rate their level of confidence in providing dysphagia services via telehealth at the start of the pandemic (n = 104 participants responded) and at the time of completing the survey (n = 109 participants responded). Confidence was rated on a scale from 1 to 5, with 1 being “not confident at all” and 5 being “very confident.” At the start of the pandemic, 43 respondents (41.3%) rated themselves on the low end of the confidence scale (i.e., as a “1” or a “2”), and 18 respondents (17.3%) rated themselves on the high end of the confidence scale (i.e., as a “4” or a “5”). At the time of completing the survey, these numbers were almost reversed, with 12 respondents (11.0%) rating themselves on the low end of the confidence scale (i.e., as a “1” or a “2”) and 72 respondents (66.1%) rating themselves on the high end of the confidence scale (i.e., as a “4” or a “5”) (Fig. 4a).Fig. 4 Clinician confidence and perceived effectiveness of telehealth services Clinician Perceived Effectiveness Clinicians also rated their level of perceived effectiveness of dysphagia services via telehealth at the start of the pandemic (n = 101 participants responded) and at the time of completing the survey (n = 104 participants responded). Effectiveness was also rated on a scale from 1 to 5, with 1 being “not effective at all” and 5 being “very effective.” At the start of the pandemic, 34 respondents (33.7%) rated themselves as effective or very effective (i.e., a “4” or a “5”) in providing tele-services for dysphagia. At the time of this survey, this number had doubled, and 70 respondents (67.3%) rated their services as effective or very effective (i.e., a “4” or a “5”) (Fig. 4b). Self-reported Reasons for Ratings of Clinician Confidence and Effectiveness (Qualitative Analysis) Survey respondents were then asked to justify their chosen levels of confidence and effectiveness, respectively. Conventional content analysis revealed similar reasons provided to justify levels of both parameters. Five key domains that influenced clinician confidence and perceived effectiveness were identified. These were: (1) prior experience, (2) treatment factors, (3) patient factors, (4) technology-related factors, and (5) perceived efficacy or success. See Table 2 for examples offered by the participants on the ways in which these domains enhanced and lowered clinician confidence and perceived effectiveness.Table 2 Factors clinicians reported to have influenced clinician confidence and perceived effectiveness of telehealth dysphagia management Domains Enhanced confidence/perceived effectiveness Lowered confidence/perceived effectiveness Prior experience • Dysphagia management experience • Telehealth experience • Use of trainings and guidelines • General experience with technology platforms • Inexperience with telehealth • Lack of training Treatment factors • Facilitator use • Access to instrumental assessment • Omitting bolus trials • Engaging in treatment only • Lack of physical contact with patient • Lack of instrumental assessment or alternative objective measures • Inability to control the environment (e.g., distractions in patients’ homes) • Difficulty working with facilitators • Safety concerns Patient factors • Higher cognition • Higher motivation • Lower cognition • Lower motivation Technology-related factors • Higher confidence, experience, and skill • Lower confidence, experience, and skill Clinician-perceived efficacy • Feeling a treatment was ‘successful’ • Feeling uncertainty about treatment ‘success’ Research Question 3: Variables Significantly Associated with the Usage of Telehealth and Self-reported Confidence and Effectiveness Levels Variables Significantly Associated with the Use of Telehealth Binary logistic regression revealed that experience managing dysphagia (p = .003) and experience with telehealth prior to the pandemic (p < .001) were significantly associated with the use of telehealth to manage dysphagia during COVID-19 (Table 3). Specifically, 6–10 years of experience managing dysphagia (OR 2.06; CI 0.49–2.13) and more than 15 years of experience managing dysphagia (OR 4.83; CI 0.92–27.6) were associated with higher odds of using telehealth to manage dysphagia during the pandemic. Compared to those who had no experience or familiarity with telehealth, those who were familiar with research on telehealth had 2.63 times the odds of using telehealth to manage dysphagia during the pandemic (OR 2.63, CI 1.40–5.00); those who used telehealth before, although not regularly, had 4.22 times the odds (OR 4.22, CI 1.84–10.2), and those who used telehealth regularly has 6.83 times the odds (OR 6.83, CI 0.85–144).Table 3 Variables associated with the use of telehealth to manage dysphagia during COVID-19 Predictors ORa 95% CIa p-value Age .221  21–30 years old – –  31–40 years old 1.09 0.37, 3.10  41–50 years old 0.45 0.11, 1.69  51–60 years old 0.29 0.06, 1.29  > 60 years old 0.60 0.12, 2.85 Geographic region .460  USA – –  Canada 2.35 0.81, 7.34  Europe 1.15 0.50, 2.63  Africa, Asia, & the Middle East 1.23 0.22, 7.44  South America 1.36 0.53, 3.51  Australia/New Zealand 4,617,769 0.00, NA Experience managing dysphagia .031  < 1 year – –  1–5 years 0.78 0.21, 2.90  6–10 years 2.06 0.49, 9.13  11–15 years 1.23 0.26, 6.16  > 15 years 4.83 0.92, 27.6 Prior experience with telehealth  < .001  None – –  Familiar with research 2.63 1.40, 5.00  Used telehealth but not regularly 4.22 1.84, 10.2  Used telehealth regularly 6.83 0.85, 144  odds ratio, CI confidence interval Variables Significantly Associated with Clinician Confidence Using Telehealth for Dysphagia Ordinal logistic regression revealed that work setting (p = .003) and the use of guidelines (p = .042) were significantly associated with higher clinician confidence in telehealth services at the time of the survey (Table 4). Specifically, working in the outpatient setting was associated with 94% increased odds of reporting higher confidence (OR 0.06; CI 0.01–0.29), when compared to clinicians working in the inpatient setting. Clinicians who used guidelines had over two times the odds of reporting higher confidence in their telehealth dysphagia care (p = .044; OR 2.39; CI 1.03–5.67).Table 4 Variables associated with SLP confidence of telehealth dysphagia management Predictors ORa 95% CIa p-value Experience managing dysphagia .401  < 1 year – –  1–5 years 4.41 0.44, 43.9  6–10 years 5.00 0.55, 44.6  11–15 years 9.80 0.91, 107  > 15 years 6.25 0.73, 53.2 Prior experience with telehealth .220  None – –  Familiar with research 1.97 0.72, 5.48  Used telehealth but not regularly 3.44 1.07, 11.5  Used telehealth regularly 1.49 0.27, 8.74 Work setting .003  Outpatient only – –  Inpatient only 0.06 0.01, 0.29  Both 0.75 0.29, 1.95 Trainings completed .314  No – –  Yes 1.56 0.66, 3.72 Use of guidelines .042  No – –  Yes 2.39 1.03, 5.67 Presence of a facilitator .052  Never – –  Sometimes 1.86 0.45, 7.88  Half of the time 2.80 0.42, 19.7  Most of the time 7.39 1.66, 35.3  Always 2.93 0.64, 13.9 aOR odds ratio, CI confidence interval Variables Significantly Associated with Clinician-Perceived Effectiveness Using Telehealth for Dysphagia Similarly, work setting was significantly associated with higher clinician perceived effectiveness in telehealth services at the time of the survey (p = .007) (Table 5). Again, working in the outpatient setting only was associated with higher odds of clinician-perceived effectiveness (OR 0.07; CI 0.01–0.37) when compared to those who worked in the inpatient setting.Table 5 Variables associated with clinician-perceived effectiveness of telehealth dysphagia management Predictors ORa 95% CIa p-value Experience managing dysphagia .155  < 1 year – –  1–5 years 0.63 0.06, 6.34  6–10 years 0.72 0.08, 6.03  11–15 years 3.39 0.33, 35.9  > 15 years 1.03 0.13, 8.28 Prior experience with telehealth .674  None – –  Familiar with research 1.29 0.47, 3.53  Used telehealth but not regularly 2.07 0.64, 6.81  Used telehealth regularly 1.55 0.25, 9.93 Work setting .007  Outpatient only – –  Inpatient only 0.07 0.01, 0.37  Both 0.92 0.34, 2.50 Trainings completed .486  No – –  Yes 1.37 0.57, 3.31 Use of guidelines .177  No – –  Yes 1.80 0.77, 4.31 Presence of a facilitator .256  Never – –  Sometimes 2.14 0.46, 10.3  Half of the time 0.72 0.10, 5.28  Most of the time 3.81 0.76, 20.0  Always 1.94 0.38, 10.0 aOR odds ratio, CI confidence interval Research Question 4: Understanding the Benefits and Challenges of Telehealth to Manage Dysphagia Benefits of Telehealth (Qualitative Analysis) The most frequently reported benefits of telehealth during COVID-19 were safety (n = 69), access to care (n = 58), and allowing for treatment continuity (n = 50). There were many factors that clinicians reported to have facilitated telehealth use during COVID-19. These most prominently included factors pertaining to the pandemic and the need to mitigate viral transmission (n = 27; e.g., government and facility restrictions on in-person procedures, the need to avoid public transportation, the safety of staying at home). External support factors such as administrative support (n = 10), technical support (n = 12) and technology trainings (n = 6), and the use of a facilitator (n = 13) were also reported to facilitate telehealth use. Telehealth-specific support and resources that abounded at this time, such as evidence to support telehealth dysphagia management (n = 8), telehealth protocols from organizations or facilities (n = 16), telehealth trainings (n = 3), and telehealth education (n = 4), also facilitated telehealth use. Finally, reimbursement of telehealth services (n = 16), improving patient accessibility to dysphagia services (n = 15), facilities providing equipment to patients who needed it (e.g., iPads, laptops) (n = 9), and a positive experience with telehealth (n = 7) were all also reported as factors facilitating telehealth usage during the pandemic. Clinician-reported benefits of telehealth beyond the pandemic generally fell into two categories: (1) enhancing patient accessibility to care (n = 110) and (2) benefits inherent to the telehealth modality (n = 86; e.g., incorporating family in treatment; naturalistic environment; enabling treatment continuity after an initial intense burst of treatment; and time, cost, and travel savings). Challenges of Telehealth (Rank-Order Analysis) Key challenges of telehealth dysphagia management from clinicians’ perspectives were lack of telehealth infrastructure (e.g., cameras, equipment) (n = 110), reimbursement limitations (n = 81), and lack of telehealth training (n = 106). Key challenges that clinicians identified on the part of patients generally surrounded technology—such as patients not being “tech-savvy” (n = 134), not having a computer/device (n = 106), or not having adequate internet connectivity (n = 90) (Table 6).Table 6 Clinician reported challenges of telehealth dysphagia management Rank Clinician perspective Patient perspective During the pandemic (mean rank) For the future (mean rank) During the pandemic (mean rank) For the future (mean rank) 1 Lack of infrastructure (2.17) Reimbursement/insurance coverage/payment issues (1.89) Reduced knowledge about technology (‘not tech savvy’) (2.26) Reduced knowledge about technology (‘not tech savvy’) (2.40) 2 Reimbursement/insurance coverage/payment issues (2.41) Lack of infrastructure (2.36) Lack of computer/device (2.68) Reimbursement/insurance coverage/payment issues (2.63) 3 Lack of knowledge/training in telehealth (2.55) Licensure restrictions (2.77) Internet connectivity issues (3.11) Lack of computer/device (2.94) *Mean ranking is the mean rank out of 9 (1 = highest, 9 = lowest) Research Question 5: Variables Significantly Associated with Projected Use of Telehealth to Manage Dysphagia in the Future Of all survey respondents, 36% said they would continue to use telehealth after COVID-19 and an additional 31% said they would continue to use telehealth after COVID-19, if reimbursement of services is possible. When comparing clinicians who said they would use telehealth in the future (i.e., those who said “yes” and those who said “yes, if reimbursement allows”) to those who said they would not use telehealth in the future, the only variable significantly associated with future telehealth use was the use of telehealth during COVID-19 (p < .001) (Table 7). Specifically, those who used telehealth in the outpatient setting (only) had 5.71 times the odds (p < .001, OR 5.71, CI 2.49–14.5) and those who used telehealth in the in-patient and out-patient settings had 6.02 times the odds (p = .006, OR 6.02, CI 1.88–27.0) of reporting that they would use telehealth in the future as compared to those who did not use telehealth during the pandemic. A sub-analysis consisting of just those who used telehealth during COVID-19 revealed no additional significant predictors of future telehealth adoption (Supplemental Table 2).Table 7 Variables associated with the likelihood of using telehealth in the future: “yes” vs “no” Predictors ORa 95% CIa p-value Experience managing dysphagia .077  < 1 year – –  1–5 years 2.05 0.56, 8.14  6–10 years 5.06 1.38, 20.2  11–15 years 3.23 0.88, 12.9  > 15 years 4.34 1.16, 17.6 Prior experience with telehealth .200  None – –  Familiar with research 1.27 0.61, 2.66  Used telehealth but not regularly 2.16 0.82, 6.24  Used telehealth regularly 3,394,358 0.00, NA Use of telehelath during COVID-19  < .001  No – –  Yes—Outpatient only 5.71 2.49, 14.5  Yes—Inpatient only 0.89 0.25, 3.14  Yes—Inpatient and outpatient 6.02 1.88, 27.0 aOR odds ratio, CI confidence interval Self-reported Reasons for Clinician Likelihood of Using Telehealth in the Future (Qualitative Analysis) Reasons why clinicians would be likely to use telehealth in the future related to increasing accessibility (n = 48) and improving efficiency (n = 12) of dysphagia care. Reasons why clinicians stated they would be less likely to use telehealth in the future were related to a lack of ability to perform instrumental evaluations and/or physically manipulate patients, and a resulting concern regarding diagnostic decision making via telehealth (n = 42). For example, one respondent stated, “I would not use telehealth if I suspected a patient may be silently aspirating” and others said they would prefer to use telehealth for follow-up only. Few clinicians also reported barriers in access to technology and connectivity as limiting factors (n = 7). Discussion This study surveyed speech language pathologists (61% within US, and 39% in other countries) and identified patterns of telehealth use prior to and during the COVID-19 pandemic, as well as clinician-perceived benefits and challenges of using this service delivery model to manage dysphagia now and in the future. The data demonstrate limited use of telehealth for dysphagia management prior to the pandemic, a large increase during the pandemic, and a further projected increase in the future. Years of experience with dysphagia management and pre-pandemic use of telehealth were significantly associated with current use patterns, and use of telehealth during the pandemic was significantly associated with projected future use. Working in the outpatient setting was associated with greater clinician confidence with and perceived effectiveness of telehealth, and use of guidelines was also associated with greater clinician confidence of tele-dysphagia management. Several key challenges and benefits were also identified and are discussed. Telehealth Usage Patterns Specifically, when evaluating telehealth usage patterns, we identified that prior to the pandemic only 20.4% of survey respondents had used telehealth in their SLP practice, and a mere 12.3% had utilized telehealth to manage dysphagia specifically, with only 2.98% of respondents reporting regular use of tele-management of dysphagia prior to the pandemic. The number of clinicians who used telehealth to manage dysphagia grew to an impressive 50% during the pandemic. This sharp increase is not surprising and reflects what was seen across other healthcare fields servicing populations that may experience dysphagia (e.g., head and neck cancer [40]; Parkinson’s disease [41]) and across other SLP services [42–45]. This is likely due, at least in part, to the cancellation of in-person services during this time. We found that during the most restrictive (MR) period of the pandemic dysphagia services were frequently cancelled. Over 50% of clinicians reported cancelling treatment sessions at least half the time and 72% cancelling instrumental evaluations at least half the time. This is in accordance with the cancellation rates of elective surgeries, non-urgent office visits, and SLP voice and swallowing services in many facilities at the onset of the pandemic [46, 47]. During the least restrictive (LR) period of the pandemic, fewer procedures were cancelled; however, cancellations were still observed, explaining the rapid uptake in telehealth, to safely provide dysphagia management to patients in need. We further identified that more years of experience with dysphagia management and prior experience using telehealth (in any area of SLP) were significantly associated with the probability of using telehealth to manage dysphagia during the pandemic. Experience has been associated with greater perceived usefulness of information technology systems [48]. It stands to reason that clinicians who were already using remote services would continue using this modality, as they would have the necessary tools, knowledge, and self-efficacy to do so [48]. Those familiar with research on the topic, but with no experience, were more likely to use telehealth, suggesting that being equipped with the necessary knowledge, even in the absence of experience, facilitated the adoption of telehealth. This suggests that educating students and clinicians on the current research on best telehealth dysphagia practices should be considered as an important component of clinical education. In fact, a recent study highlighted the importance of extensive and specific training for successful implementation of telehealth clinical swallowing evaluations [49]. Further, it is likely that more experienced clinicians possess greater skill and clinical decision-making abilities, which may have enabled clinicians with more years of dysphagia experience to adopt telehealth more easily. Looking more specifically at the types of dysphagia services provided via telehealth during both the MR and LR periods of the pandemic, treatment sessions were conducted most frequently, followed by clinical evaluations, with instrumental evaluations conducted infrequently. There are few validated tele-VFSS systems and those that do exist are not widely available in mainstream practice [12–14]. Therefore, most clinicians did not have the resources to conduct tele-VFSS, and further, given the restrictions on in-person activity, availability of instrumental evaluations was limited early in the pandemic. The greater frequency of telehealth treatment visits, as compared to assessments, corroborates qualitative findings from the present study which revealed greater clinician uncertainty regarding tele-assessments and some degree of hesitation in making diagnostic decisions via telehealth. This finding has been reported across SLP services [45, 50–52] and has been previously linked to dysphagia services as well [2, 53]. Uncertainty regarding telehealth assessments was also reported by clinicians as a factor that lowered their confidence and perceived effectiveness, with lack of instrumental assessment specifically reported by some as a diagnostic barrier. However, lack of instrumental assessment is not unique to telehealth and while it limits the conclusions that can be drawn from any clinical swallowing evaluation, clinical and instrumental assessments serve unique purposes in the continuum of dysphagia care [54, 55]. Findings from telehealth clinical swallowing evaluations have shown to reliably match findings from in-person clinical evaluations [9, 10] address patient needs, overcome barriers such as geographic distance, and enable more timely intervention [49]. Additionally, clinical swallowing tele-evaluations may be bolstered by objective measures, such as the Timed Water Swallow Test [56] and the Test of Masticating and Swallowing Solids [57], which have shown adequate reliability via telehealth [58]. Continued efforts to develop smart teledynamic systems that can collect more objective data via telehealth are ongoing across healthcare fields and should be examined in future research. Clinician Confidence and Perceived Effectiveness Clinician reported confidence and perceived effectiveness improved substantially from the start of the pandemic to the time of the survey. This was likely influenced by a number of factors, including increased experience gained with telehealth [48] and factors that clinicians reported to have enhanced their confidence, such as increased availability of guidelines and trainings and increased availability of instrumental assessments. This highlights the importance of addressing multiple domains including experience, training, guidelines, and access to instrumental assessments to increase the confidence of clinicians in tele-management. The improvement in clinician confidence and perceived effectiveness further demonstrates the potential for telehealth to be incorporated into clinical care post-pandemic for parts of the evaluation or treatment process when needed but not as a complete replacement for in-person services. An integrated model of care that incorporates in-person and telehealth services has recently been proposed as the way forward to enhance dysphagia care [53]. Fritz et al. (2020) propose a framework for using telehealth as an initial starting point for dysphagia management, during which one can obtain medical and case history information, obtain patient-reported outcomes, and complete certain components of the clinical swallowing evaluation [3, 59]. This information can then be used to determine next steps for a patient, which may include an instrumental evaluation and/or in-person follow-up. Depending on the circumstance, this model could easily be reversed—beginning with an in-person evaluation, which may include an instrumental assessment, and following up with treatment via telehealth [6]. Further, facilitators and use of trainings and guidelines all appeared to play an important role in telehealth dysphagia management. Across settings and procedures, among those who used telehealth during the pandemic, over 75% reported using a facilitator at least sometimes. The high proportion of survey respondents who utilized facilitators reflects prior research [8, 10, 60], guidelines highly recommending the use of a facilitator [2, 6], and more recent data suggesting that the presence of a family member may optimize remote dysphagia evaluations [3]. Among clinicians who used telehealth, 50% used guidelines and 55% completed trainings. While guidelines and trainings from various parts of the world were reportedly utilized in this sample, given that the majority of the sample was based in the U.S., our data may not capture the extent of training opportunities available internationally. Clinical guidelines provide specific recommendations about best practices, support care-providers with readily available information, and serve to improve patient care [61]. They may be specifically helpful in situations with uncertainty around best practices [61], such as during the COVID-19 pandemic. Indeed, clinicians reported the use of guidelines and completion of trainings to enhance both their confidence and perceived effectiveness of telehealth dysphagia services. Further, we identified that the use of guidelines was statistically significantly associated with confidence in telehealth dysphagia management. These findings are supported by a recent implementation trial which highlighted the essential role of training and expert guidance for successful implementation of telehealth dysphagia services [49]. Thus, the expansion and wider dissemination of evidence-based trainings and guidelines is an important area for future work. Statistical analysis also revealed that working in the outpatient (as compared to inpatient) setting was significantly associated with greater clinician-perceived confidence and effectiveness. Given illness severity and acuity, and variable patient alertness in inpatient settings, it is understandable that clinicians reported lower confidence and perceived-effectiveness of their telehealth services in these settings. Defining the ways in which telehealth can best be utilized for inpatient dysphagia management may be an important direction for future study. For example, exploring ways to facilitate valid and reliable remote clinical swallowing evaluations in under-resourced hospitals may be an area of study with potential to improve access to critical dysphagia services. However, safety safeguards and unique patient-factors would need to be carefully considered. Benefits and Challenges of Telehealth Understanding benefits and challenges of telehealth from clinicians’ perspectives has important implications for how telehealth models of care may be optimized in both the in- and outpatient settings. Clinician-described benefits of telehealth during the pandemic generally surrounded the topics of safety, access to care, and treatment continuity. Telehealth enabled patients to continue (or begin) receiving dysphagia services without putting themselves or providers at risk for virus transmission [62–64]. Clinicians also reported numerous benefits of telehealth that extend well beyond the COVID-19 pandemic. These included improving patient access to care and numerous other benefits, such as time, cost, and travel savings, incorporating family into the sessions, utilizing the patient’s naturalistic environment, and allowing for treatment continuity after an initial intense burst of treatment, among others. Indeed, it has been established that telehealth may improve access to care [16, 33, 34, 49, 65], reduce patient time, cost, and travel burden associated with receiving therapeutic services [8, 18], enable greater intensity and frequency of rehabilitation while allowing patients to remain in a comfortable and familiar environment [66], incorporate family and carers [53], and result in excellent patient and provider satisfaction [11, 17, 49, 67–69]. Of the respondents who used telehealth, most reported that they will continue to do so in the future and using telehealth during the pandemic was significantly associated with projected future use. Moreover, while 50% of participants in this study utilized telehealth to manage dysphagia during the pandemic, 67% reported that they plan to use telehealth in the future. This underscores clinician willingness and desire to adopt telehealth as an integrated delivery model for dysphagia care. However, to optimize the tele-management of dysphagia, a number of reported challenges need to be addressed. Clinicians reported that from their perspective, the primary obstacle to providing telehealth dysphagia services during the pandemic was a lack of infrastructure (e.g., technology available, Internet connectivity). In fact, one in four Americans does not have access to a smartphone device and/or sufficient broadband Internet to engage in synchronous videoconferencing [70, 71], and in many developing countries, lack of technological and infrastructure availability is even more significant and has been identified as a barrier to the adoption of this service delivery model [52]. Addressing these barriers is critical because quality of technology and stability of internet connection are important determinants of successful telehealth implementation [48]. Even for those with adequate internet and technology, facility constraints on resources and staff training may still be limiting factors that need to be addressed [49]. Clinicians also reported that they believed the primary obstacle to telehealth from the patients’ perspective was inadequate “tech savviness,” or digital literacy. It has been reported that, among patients with neurologic disease, computer self-efficacy and computer anxiety are significant predictors of participation in tele-rehabilitation [65]. Additionally, a patient’s lack of experience with technology has been identified as a factor that may negatively impact patient-perceived usefulness of tele-treatment [65] and influence adherence [48]. Clinicians may be able to play a role in orienting patients to basic technology platforms, and facilitators may play an assisting role in this regard. However, clinicians likely need more training as well [53]. In a survey of SLPs in Hong Kong, 50% of respondents who did not use telehealth reported that “technology” was a key barrier, and that technology should be the focus of further training [51]. Moreover, a lot of the pre-pandemic research on tele-dysphagia management has utilized custom-built platforms, additional web-cameras, and trained facilitators—all of which were not available during the pandemic and may not be financially or logistically feasible for routine clinical use in the future. Simple, easy-to-use tele-dynamic systems may help to circumvent some of the barriers surrounding experience and training [72] as well as access to equipment. Thus, another critical need of the field is to develop smarter, easier, and more accessible telehealth platforms. Finally, licensure and reimbursement restrictions were reported by clinicians as a key challenge of utilizing telehealth for dysphagia management. While telehealth has been reimbursed due to the pandemic “public health emergency” in the United States, policies have yet to be formally amended for the future. Advocacy at both the state and federal levels will be essential for working toward reimbursement schemes that include telehealth. Federal, state, and international reimbursement schemes will need to recognize and reimburse for telehealth services such that it can become a standard and enduring model of service delivery. Limitations This survey study had a relatively small sample size (n = 235), the majority (61%) of clinicians were from the United States, all were SLPs, and only ~ 50% of survey respondents (n = 118) utilized telehealth to manage dysphagia during the pandemic. However, the survey remained open for three months and extensive recruitment efforts via social media and personal contacts were conducted. Additionally, because it was not mandatory to answer each question, the number of respondents per question varied. Due to the heterogeneity in geographic region, work-setting, and patient populations that survey respondents worked with, there may be different standards of care, resources available, and reimbursement structures relating to telehealth practice that influenced telehealth use. Further, because many survey respondents worked in multiple settings and worked with patients of varying diagnoses, the data collected in this survey cannot differentiate the impact of specific work environments and patient populations on telehealth use. Future work should include other professionals who treat dysphagia, especially given the variety of disciplines involved in dysphagia care across the world. Given the nature of this study, we acknowledge the possibility of a sampling bias, and it remains possible that the respondents who chose to participate in the survey may somehow differ from clinicians who did not complete the survey. Of note, the survey was limited to respondents who were able to read and understand English. Additionally, survey questions are subject to individual interpretation, which may differ among respondents and all responses are subjective. Clinicians’ experiences and perceptions reflected in this survey represent a moment in time for each respondent. Especially given the ongoing and constantly evolving nature of the COVID-19 pandemic and the shifting work landscape during this time, clinicians’ experiences with telehealth are likely also changing and evolving. Longitudinal and/or follow-up studies will be needed to better understand the evolution of telehealth in the field of dysphagia. Lastly, in the present survey, the challenges of telehealth that patients may experience were reported by clinicians. Future research should seek to explore patient-perceived benefits and challenges of receiving dysphagia care via telehealth. Conclusions and Future Directions This survey of speech-language pathologists highlighted the limited use of telehealth to manage dysphagia prior to COVID-19, and the sharp rise in such use driven by the pandemic. However, this study also highlighted that, currently and even during the height of the pandemic, telehealth was and is under-utilized for dysphagia management. It is well-established that telehealth can improve access to care and expand provision of critical dysphagia services. Findings from this study revealed numerous clinician-reported benefits of dysphagia tele-management and clinician willingness to utilize this service delivery model. Several challenges were also identified; however, the majority of clinicians reported that they will use telehealth in the future. Technological developments, such as wearable devices, portable treatment and evaluative tools, smartphone applications, and adherence tracking mechanisms using standard consumer-grade equipment (e.g., iPhones and iPads) have already started playing a role in the tele-management of dysphagia and need to be further developed and evaluated in future research. Importantly, future research should also examine the experiences of multiple stakeholders—including patients, caregivers, and clinicians—to understand the barriers and facilitators of this mode of service delivery to refine and optimize dysphagia tele-management and improve access to care for our patients. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 13 KB) Supplementary file2 (DOCX 15 KB) Supplementary file3 (DOCX 26 KB) Acknowledgements This work was partially supported by a Purdue University College of Health and Human Sciences COVID-19 Rapid Response Grant (PI: last author). Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Pasnick S Carlos WG Dela Cruz CS Gross JE Garrison G Jamil S SARS-CoV-2 transmission and the risk of aerosol-generating procedures Am J Respir Crit Care Med 2020 202 4 P13 P14 10.1164/rccm.2024P13 32795140 2. Miles A Connor NP Desai RV Jadcherla S Allen J Brodsky M Garand KL Malandraki GA McCulloch TM Moss M Murray J Pulia M Riquelme LF Langmore SE Dysphagia care across the continuum: a multidisciplinary dysphagia research society taskforce report of service-delivery during the COVID-19 global pandemic Dysphagia 2021 36 170 182 10.1007/s00455-020-10153-8 32654059 3. Soldatova L Williams C Postma GN Falk GW Mirza N Virtual dysphagia evaluation: practical guidelines for dysphagia management in the context of the COVID-19 pandemic Otolaryngol Head Neck Surg 2020 163 455 458 10.1177/0194599820931791 32450732 4. List of Telehealth Services for Calendar Year 2021. Centers for Medicare and Medicaid services. https://www.cms.gov/Medicare/Medicare-General-Information/Telehealth/Telehealth-Codes (2021). accessed 30 Apr 2021. 5. Providing Telehealth Services Under Medicare During the COVID-19 Pandemic, American Speech-Language-Hearing Association. https://www.asha.org/practice/reimbursement/medicare/providing-telehealth-services-under-medicare-during-the-covid-19-pandemic/ (n.d.). accessed 30 Apr 2021. 6. Malandraki GA Arkenberg RH Mitchell SS Malandraki JB Telehealth for dysphagia across the life span: using contemporary evidence and expertise to guide clinical practice during and after COVID-19 Am J Speech Lang Pathol 2021 30 532 550 10.1044/2020_AJSLP-20-00252 33555933 7. Kantarcigil C Malandraki GA First step in telehealth assessment: a randomized controlled trial to investigate the effectiveness of an electronic case history form for dysphagia Dysphagia 2017 32 548 558 10.1007/s00455-017-9798-y 28424897 8. Burns CL Ward EC Gray A Baker L Cowie B Winter N Rusch R Saxon R Barnes S Turvey J Implementation of speech pathology telepractice services for clinical swallowing assessment: an evaluation of service outcomes, costs and consumer satisfaction J Telemed Telecare 2019 25 545 551 10.1177/1357633X19873248 31631757 9. Morrell K Hyers M Stuchiner T Lucas L Schwartz K Mako J Spinelli KJ Yanase L Telehealth stroke dysphagia evaluation is safe and effective Cerebrovasc Dis 2017 44 225 231 10.1159/000478107 28848110 10. Ward EC Sharma S Burns C Theodoros D Russell T Validity of conducting clinical dysphagia assessments for patients with normal to mild cognitive impairment via telerehabilitation Dysphagia 2012 27 460 472 10.1007/s00455-011-9390-9 22271284 11. Sharma S Ward EC Burns C Theodoros D Russell T Assessing dysphagia via telerehabilitation: patient perceptions and satisfaction Int J Speech-Lang Pathol 2013 15 176 183 10.3109/17549507.2012.689333 22663016 12. Malandraki GA McCullough G He X McWeeny E Perlman AL Teledynamic evaluation of oropharyngeal swallowing J Speech Lang Hear Res 2011 54 1497 1505 10.1044/1092-4388(2011/10-0284) 22052284 13. Burns CL Ward EC Hill AJ Phillips N Porter L Conducting real-time videofluoroscopic swallow study via telepractice: a preliminary feasibility and reliability study Dysphagia 2016 31 473 483 10.1007/s00455-016-9701-2 26979971 14. Malandraki GA Markaki V Georgopoulos VC Bauer JL Kalogeropoulos I Nanas S An international pilot study of asynchronous teleconsultation for oropharyngeal dysphagia J Telemed Telecare 2013 19 75 79 10.1177/1357633x12474963 23470448 15. Nordio S Innocenti T Agostini M Meneghello F Battel I The efficacy of telerehabilitation in dysphagic patients: a systematic review Acta Otorhinolaryngol Ital 2018 38 79 85 10.14639/0392-100X-1816 29967554 16. Collins A Burns CL Ward EC Comans T Blake C Kenny L Greenup P Best D Home-based telehealth service for swallowing and nutrition management following head and neck cancer treatment J Telemed Telecare 2017 23 866 872 10.1177/1357633X17733020 29081270 17. Wall LR Ward EC Cartmill B Hill AJ Porceddu SV Examining user perceptions of SwallowIT : a pilot study of a new telepractice application for delivering intensive swallowing therapy to head and neck cancer patients J Telemed Telecare 2017 23 53 59 10.1177/1357633X15617887 26670210 18. Wall LR Kularatna S Ward EC Cartmill B Hill AJ Isenring E Byrnes J Porceddu SV Economic analysis of a three-arm RCT exploring the delivery of intensive, prophylactic swallowing therapy to patients with head and neck cancer during (chemo)radiotherapy Dysphagia 2019 34 627 639 10.1007/s00455-018-9960-1 30515560 19. Wall LR Ward EC Cartmill B Hill AJ Isenring E Byrnes J Porceddu SV Prophylactic swallowing therapy for patients with head and neck cancer: a three-arm randomized parallel-group trial Head Neck 2020 42 873 885 10.1002/hed.26060 31903689 20. Shinn EH Jensen K McLaughlin J Garden AS Fellman BM Liang L Peterson SK Interactive website for head and neck cancer patients: adherence and coping program to prevent dysphagia after radiation Internet Interv 2019 18 100289 10.1016/j.invent.2019.100289 31890636 21. Starmer HM Abrams R Webster K Kizner J Beadle B Holsinger FC Quon H Richmon J Feasibility of a mobile application to enhance swallowing therapy for patients undergoing radiation-based treatment for head and neck cancer Dysphagia 2018 33 227 233 10.1007/s00455-017-9850-y 28965209 22. Constantinescu G Rieger J Seikaly H Eurich D Adherence to home-based swallowing therapy using a mobile system in head and neck cancer survivors Am J Speech Lang Pathol 2021 10.1044/2021_AJSLP-21-00026 23. Constantinescu G Loewen I King B Brodt C Hodgetts W Rieger J Designing a mobile health app for patients with dysphagia following head and neck cancer: a qualitative study JMIR Rehabil Assist Technol 2017 4 e3 10.2196/rehab.6319 28582245 24. Srp M Korteová R Kliment R Jech R Růžička E Hoskovcová M Expiratory muscle strength training in patients with Parkinson’s disease: a pilot study of mobile monitoring application, movement disorders clinical Practice 2021 8 1148 1149 10.1002/mdc3.13313 25. Cassel SG Case reports: Trial dysphagia interventions conducted via telehealth Int J Telerehab 2016 8 71 76 10.5195/IJT.2016.6193 26. Bascuñana-Ambrós H Renom-Guiteras M Nadal-Castells MJ Beranuy-Rodríguez M Perrot-González JC Ramirez-Mirabal E Trejo-Omeñaca A Monguet-Fierro JM Swallowing muscle training for oropharyngeal dysphagia: a non-inferiority study of online versus face-to-face therapy J Telemed Telecare 2021 10.1177/1357633X211035033 27. Sevitz JS Borders JC Dakin AE Kiefer BR Alcalay RN Kuo S-H Troche MS Rehabilitation of airway protection in individuals with movement disorders: a telehealth feasibility study Dysphagia 2022 31 6 2741 2758 28. Brewster L Mountain G Wessels B Kelly C Hawley M Factors affecting front line staff acceptance of telehealth technologies: a mixed-method systematic review J Adv Nurs 2014 70 21 33 10.1111/jan.12196 23786584 29. Brooks E Turvey C Augusterfer EF Provider barriers to telemental health: obstacles overcome Obstac Remain Telemed E-Health 2013 19 433 437 10.1089/tmj.2013.0068 30. Wade VA Eliott JA Hiller JE Clinician acceptance is the key factor for sustainable telehealth services Qual Health Res 2014 24 682 694 10.1177/1049732314528809 24685708 31. Ward EC Burns CL Theodoros DG Russell TG Impact of dysphagia severity on clinical decision making via telerehabilitation Telemed E-Health 2014 20 296 303 10.1089/tmj.2013.0198 32. Burns CL Ward EC Hill AJ Malcolme K Bassett L Kenny LM Greenup P A pilot trial of a speech pathology telehealth service for head and neck cancer patients J Telemed Telecare 2012 18 443 446 10.1258/jtt.2012.gth104 23209274 33. Haulman A Geronimo A Chahwala A Simmons Z The use of telehealth to enhance care in ALS and other neuromuscular disorders Muscle Nerve 2020 61 682 691 10.1002/mus.26838 32297678 34. Peacock D Baumeister P Monaghan A Siever J Yoneda J Wile D Perception of healthcare access and utility of telehealth among Parkinson’s disease patients Can J Neurol Sci 2020 47 700 704 10.1017/cjn.2020.99 32450924 35. Qualtrics and all other Qualtrics product or service names are registered trademarks or trademarks of Qualtrics, Provo, UT, USA. https://www.qualtrics.com, (n.d.). 36. R Core Team, A language and environment for statistical computing, (2018). https://www.R-project.org/. 37. Hsieh H-F Shannon SE Three approaches to qualitative content analysis Qual Health Res 2005 15 1277 1288 10.1177/1049732305276687 16204405 38. Morse JM Critical analysis of strategies for determining rigor in qualitative inquiry Qual Health Res 2015 25 1212 1222 10.1177/1049732315588501 26184336 39. Vaismoradi M Turunen H Bondas T Content analysis and thematic analysis: implications for conducting a qualitative descriptive study Nurs Health Sci 2013 15 398 405 10.1111/nhs.12048 23480423 40. Chua MLK Ma DJ Anderson CM Karam SD Margalit DN Kimple RJ Follow-up and management of patients with head and neck cancer during the 2019 novel coronavirus (SARS-CoV-2) disease pandemic Adv Radiat Oncol 2019 5 2020 631 636 10.1016/j.adro.2020.04.031 41. Feeney MP Xu Y Surface M Shah H Vanegas-Arroyave N Chan AK Delaney E Przedborski S Beck JC Alcalay RN The impact of COVID-19 and social distancing on people with Parkinson’s disease: a survey study Npj Parkinsons Dis 2021 7 10 10.1038/s41531-020-00153-8 33479241 42. Dimer NA do Canto-Soares N dos Santos-Teixeira L de Goulart BNG The COVID-19 pandemic and the implementation of telehealth in speech-language and hearing therapy for patients at home: an experience report CoDAS 2020 10.1590/2317-1782/20192020144 43. Aggarwal K Patel R Ravi R Uptake of telepractice among speech-language therapists following COVID-19 pandemic in India Speech Lang Hear 2021 24 228 234 10.1080/2050571X.2020.1812034 44. Carter JM Killan CF Ridgwell JJ Telehealth rehabilitation for adults with cochlear implants in response to the Covid-19 pandemic: platform selection and case studies Cochlear Implants Int 2021 10.1080/14670100.2021.1949524 45. Gibbs V Cai RY Aldridge F Wong M Autism assessment via telehealth during the Covid 19 pandemic: experiences and perspectives of autistic adults, parents/carers and clinicians Res Autism Spectrum Disorders 2021 88 101859 10.1016/j.rasd.2021.101859 46. Stein J Visco CJ Barbuto S Rehabilitation medicine response to the COVID-19 pandemic Am J Phys Med Rehabil 2020 10.1097/PHM.0000000000001470 47. Searl J Kearney A Genoa K Doyle PC Clinical experiences of people with a laryngectomy during the SARS COVID-19 pandemic Am J Speech Lang Pathol 2021 30 2430 2445 10.1044/2021_AJSLP-21-00117 34665653 48. Venkatesh V Bala H Technology acceptance model 3 and a research agenda on interventions Decis Sci 2008 39 273 315 10.1111/j.1540-5915.2008.00192.x 49. Ward EC Burns CL Gray A Baker L Cowie B Winter N Rusch R Saxon R Barnes S Turvey J Establishing clinical swallowing assessment services via telepractice: a multisite implementation evaluation Am J Speech Lang Pathol 2021 30 2456 2464 10.1044/2021_AJSLP-21-00109 34432993 50. Campbell DR Goldstein H Evolution of telehealth technology, evaluations, and therapy: effects of the COVID-19 pandemic on pediatric speech-language pathology services Am J Speech Lang Pathol 2021 10.1044/2021_AJSLP-21-00069 51. Fong R Tsai CF Yiu OY The implementation of telepractice in speech language pathology in Hong Kong during the COVID-19 pandemic Telemed E-Health 2021 27 30 38 10.1089/tmj.2020.0223 52. Leochico CFD Espiritu AI Ignacio SD Mojica JAP Challenges to the emergence of telerehabilitation in a developing country: a systematic review Front Neurol 2020 11 1007 10.3389/fneur.2020.01007 33013666 53. Ward EC Raatz M Marshall J Wishart LR Burns CL Telepractice and dysphagia management: the era of COVID-19 and beyond Dysphagia 2022 10.1007/s00455-022-10444-2 54. Speyer R Cordier R Farneti D Nascimento W Pilz W Verin E Walshe M Woisard V White paper by the European Society for Swallowing Disorders: screening and non-instrumental assessment for dysphagia in adults Dysphagia 2021 10.1007/s00455-021-10283-7 55. Mccullough G Martino R Shaker R Easterling C Belafsky PC Postma GN Clinical evaluation of patients with dysphagia: importance of history taking and physical exam Manual of diagnostic and therapeutic techniques for disorders of deglutition 2013 New York Springer 11 30 56. Hughes TAT Wiles CM Clinical measurement of swallowing in health and in neurogenic dysphagia QJM 1996 89 109 116 10.1093/qjmed/89.2.109 8729551 57. Huckabee M McIntosh T Fuller L Curry M Thomas P Walshe M McCague E Battel I Nogueira D Frank U den Engel-Hoek L Sella-Weiss O The Test of Masticating and Swallowing Solids (TOMASS): reliability, validity and international normative data Int J Lang Commun Disord 2018 53 144 156 10.1111/1460-6984.12332 28677236 58. Borders JC Sevitz JS Malandraki JB Malandraki GA Troche MS Objective and subjective clinical swallowing outcomes via telehealth: reliability in outpatient clinical practice Am J Speech Lang Pathol 2021 30 598 608 10.1044/2020_AJSLP-20-00234 33555954 59. Fritz MA Howell RJ Brodsky MB Suiter DM Dhar SI Rameau A Richard T Skelley M Ashford JR O’Rourke AK Kuhn MA Moving forward with dysphagia care: implementing strategies during the COVID-19 pandemic and beyond Dysphagia 2021 36 161 169 10.1007/s00455-020-10144-9 32519150 60. Ward EC Burns CL Theodoros DG Russell TG Evaluation of a clinical service model for dysphagia assessment via telerehabilitation Int J Telemed Appl 2013 2013 1 7 10.1155/2013/918526 61. Wollersheim H Burgers J Grol R Clinical guidelines to improve patient care Neth J Med 2005 63 6 62. Fisk M Livingstone A Pit SW Telehealth in the context of COVID-19: changing perspectives in Australia, the United Kingdom, and the United States J Med Internet Res 2020 22 e19264 10.2196/19264 32463377 63. Smith AC Thomas E Snoswell CL Haydon H Mehrotra A Clemensen J Caffery LJ Telehealth for global emergencies: implications for coronavirus disease 2019 (COVID-19) J Telemed Telecare 2019 26 2020 309 313 10.1177/1357633X20916567 64. Leite H Gruber T Hodgkinson IR Flattening the infection curve—understanding the role of telehealth in managing COVID-19 LHS 2020 33 221 226 10.1108/LHS-05-2020-084 65. Klaic M Galea MP Using the technology acceptance model to identify factors that predict likelihood to adopt tele-neurorehabilitation Front Neurol 2020 11 1637 10.3389/fneur.2020.580832 66. Maresca G Maggio MG De Luca R Manuli A Tonin P Pignolo L Calabrò RS Tele-neuro-rehabilitation in Italy: state of the art and future perspectives Front Neurol 2020 11 1144 10.3389/fneur.2020.563375 67. Kruse CS Krowski N Rodriguez B Tran L Vela J Brooks M Telehealth and patient satisfaction: a systematic review and narrative analysis BMJ Open 2017 7 e016242 10.1136/bmjopen-2017-016242 68. Tenforde AS Borgstrom H Polich G Steere H Davis IS Cotton K O’Donnell M Silver JK Outpatient physical, occupational, and speech therapy synchronous telemedicine: a survey study of patient satisfaction with virtual visits during the COVID-19 pandemic Am J Phys Med Rehabil 2020 99 977 981 10.1097/PHM.0000000000001571 32804713 69. Wilkinson JR Spindler M Wood SM Marcus SC Weintraub D Morley JF Stineman MG Duda JE High patient satisfaction with telehealth in Parkinson disease: a randomized controlled study Neurol Clin Pract 2016 6 241 251 10.1212/CPJ.0000000000000252 27347441 70. Demographics of Internet and home broadband usage in the United States, Pew Research Center: Internet, Science & Tech. (2021). https://www.pewresearch.org/internet/fact-sheet/internet-broadband/ accessed 22 Apr 2021. 71. Benda NC Veinot TC Sieck CJ Ancker JS Broadband Internet access is a social determinant of health! Am J Public Health 2020 110 1123 1125 10.2105/AJPH.2020.305784 32639914 72. Dodakian L McKenzie AL Le V See J Pearson-Fuhrhop K Burke Quinlan E Zhou RJ Augsberger R Tran XA Friedman N Reinkensmeyer DJ Cramer SC A home-based telerehabilitation program for patients with stroke Neurorehabil Neural Repair 2017 31 923 933 10.1177/1545968317733818 29072556

36515730

PMC9749630

NO-CC CODE

2022-12-15 23:23:20

no

Dysphagia. 2022 Dec 14;:1-16

utf-8

Dysphagia

10.1007/s00455-022-10544-z

oa_other

==== Front Biol Bull Rev Biology Bulletin Reviews 2079-0864 2079-0872 Pleiades Publishing Moscow 1218 10.1134/S207908642206007X Article Phosphorilated Polyprenols as Universal Agents of Viral Reproduction Suppression Sanin A. V. saninalex@inbox.ru 1 Pronin A. V. 1 Narovlyanskiy A. N. 1 Ozherelkov S. V. 2 Sedov A. M. 1 1 grid.415738.c 0000 0000 9216 2496 Gamaleya National Research Center for Epidemiology and Microbiology of the Ministry of Health of the Russian Federation, Moscow, Russia 2 grid.4886.2 0000 0001 2192 9124 Chumakov Federal Scientific Center for Research and Development of Immunobiological Preparations, Russian Academy of Sciences, Moscow, Russia 14 12 2022 2022 12 6 609624 9 3 2022 13 3 2022 25 3 2022 © Pleiades Publishing, Ltd. 2022, ISSN 2079-0864, Biology Bulletin Reviews, 2022, Vol. 12, No. 6, pp. 609–624. © Pleiades Publishing, Ltd., 2022.Russian Text © The Author(s), 2022, published in Uspekhi Sovremennoi Biologii, 2022, Vol. 142, No. 3, pp. 268–285. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This review is focused on polyprenyl phosphatis (PPFs) (phosphorilated polyprenols), a unique class of natural compounds with a wide spectrum of biological activity. The main emphasis is put on their antiviral properties that were studied in numerous in vitro and in vivo experiments. The results formed the basis for the development of new drugs containing PPFs as the active substance. These drugs, due to their high efficiency and harmlessness, are widely used in veterinary practice in the treatment of viral diseases of both small domestic and agricultural animals. Also, the world’s first PPF-based drug is registered in Russia, and it can be used as part of a comprehensive therapy of chronic recurrent genital infection caused by the herpes simplex virus. Keywords: polyprenyl phosphatis viruses Phosprenyl Gamapren Fortepren Polyprenyl Immunostimulant antiviral activity cell cultures veterinary medicine therapy herpes virus genital herpes issue-copyright-statement© Pleiades Publishing, Ltd. 2022 ==== Body pmcINTRODUCTION Polyprenols, or polyisoprenoids are long-chain isoprenoid alcohols consisting of 5 or more (up to 40) isoprene links connected by the principle of “head to tail.” They are present in a variety of structural elements of the cell and play a crucial role in the functioning of the body, being the predecessors of dolichols. While polyprenols (α-impatient isoprenoid alcohols) are mainly contained in the leaves of many plants, then dolichols (α-saturated isoprenoid alcohols) are an integral component of animal cells and tissues (Sagami et al., 2018). Dolichols are formed in the liver and play a key role in the dolichylphosphate cycle: one of the key processes taking place in the body, resulting in glycosylation of proteins and lipids with the formation of cellular receptors, enzymes, immunoglobulins, some growth factors, and hormones. After entering the animal’s body, polyprenols are converted into dolichylphosphate, playing the role of the modulator of the dolichylphosphate cycle and glycosylation processes (Antipina et al., 2021). It should be noted that free polyprenols are metabolically inactive, while the intermediate acceptors of sugars during glycosylation of proteins and lipids are polyprenylphosphates (PPFs) (or phosphorilated polyprenols). Polyprenol’s predecessors, farnezyl and geranylgeranyl-pyrophosphate, are also necessary for prenylation of the most important intracellular proteins, including most representatives of the super family of small GTPases, as well as the heterotrimer G‑proteins and nuclear lamines, catalyzed by prenyl-transferases (Jeong et al., 2018). The lipophilic prenyl group facilitates the consolidation of proteins in cell membranes, mediating protein-white interactions and signal transmission. This explains the extraordinary variety of physiological properties of polyprenols: they have immunomodulating (Pronin et al., 2002), adjuvant (Pronin et al., 2011), antioxidant (Sanin et al., 2017a), anti-inflammatory (Ganshina et al., 2011), and other important properties (Narovlyansky et al., 2018), which makes them attractive objects for the development of therapeutic remedies (Narovlyansky et al., 2007). The purpose of this article is to analyze the wealth of information about the antiviral activity of these unique compounds. To date, only four officially registered treatment agents, the current principle of which is the PPF, are known: Phosprenyl® (FP) and Fortepren® ( in which the active substance is PPF from Siberian fir), as well as Gamapren® (GP) and Polyprenyl Immunostimulant™ (in which the active substance is the moraprenyl phosphates from mulberry leaves). Phosprenyl®—international unpatented name: disodium salt of polyprenol phosphate. This antiviral drug with immunomodulating properties was registered in 1995. Manufacturer: CJSC Micro-Plus, Russia. Dosage form: a solution for injection containing 0.4% disodium salt of PPF. Fortepren®—registered by the Ministry of Health of the Russian Federation December 15, 2021, the number of the registration certificate of the drug LP-00770l-151221, manufacturer: the Federal State Budgetary Institution NIREM named after N.F. Gamaleya of the Ministry of Health of Russia (branch “Medgamal”), Moscow. Dosage form: a solution for intramuscular administration with the concentration of the active substance, sodium polyprenylphosphate, 4 mg/mL. Gamapren®—international unpatented name: disodium salt of polyprenol phosphate. It has been registered since 2008. It refers to immunomodulating drugs with antiviral properties. Manufacturer: LLC “GamaVetFarm,” Russia. Dosage form: a sterile solution containing 0.5% disodium salt of PPF from mulberry leaves as an active substance. Polyprenyl Immunostimulant™. Manufacturer: Sass & Sass, Inc., USA. Dosage form: PPF solution from mulberry leaves with a concentration of 2 mg/mL. ANTIVIRAL EFFECTIVENESS OF PPF-BASED DRUGS IN VITRO Studies included infections caused by viruses belonging to various families and childbirths, which included: Measles virus (fam. Paramyxoviridae, genus Morbillivirus)1; Canine distemper virus (fam. Paramyxoviridae, genus Morbillivirus); Mumps virus (fam. Paramyxoviridae, genus Ruba-lavirus); Influenza virus A (fam. Orthomyxoviridae, genus Influenzavirus A); Feline rhinotracheitis virus (fam. Herpesviridae, genus Alphaherpesvirus); Bovine rhinotracheitis virus (fam. Herpesviridae, genus Varicellovirus); Aujeszky’s disease virus (fam. Herpesviridae, genus Varicellovirus); Vesicular stomatitis virus (fam. Rhabdoviridae, genus Vesiculovirus); Human immunodeficiency virus (fam. Retroviridae, genus Lentivirus); Hepatitis A virus (fam. Picornaviridae, genus Hepatovirus); Theiler’s murine encephalomyelitis virus (fam. Picornaviridae, genus Cardiovirus); Hepatitis C virus (fam. Flaviviridae, genus Hepacivirus); Yellow fever virus (fam. Flaviviridae, genus Flavivirus); Bovine diarrhea virus (fam. Flaviviridae, genus Pestivirus); Tick-borne encephalitis virus (fam. Flaviviridae, genus Flavivirus); Bovine leucosis virus (fam. Retroviridae, genus Oncovirus of the type C); Infectious canine hepatitis virus (fam. Adenoviridae, genus Mastadenovirus). An experimental study on the action of PPF on the measles virus of the Leningrad-16 (L-16) strain in an infected Vero cell culture showed that the introduction of FP at a dose of 200 μg/mL led to a decrease in the viral CPE (the cytopathic effect of viruses), which was expressed in the formation of typical multinuclear symplasts (Deeva et al., 1998). Experiments studying the influence of FP on the reproduction of the canine distemper virus (Rockborn strain) were carried out on the 4647 cell culture. The virus was introduced into the culture after the formation of the monolayer, and an agar coating with the addition of FP in various concentrations was applied after 2 hours of incubation. As a result, it was shown that FP in the concentration of 200 μg/mL has a pronounced antivirus effect manifested in a decrease in the infectious capacity of the virus by 2.3 log PFU/mL. The effect of PPF on the reproduction of epidemic mumps, the Leningrad-3 strain (L-3), was carried out on the Vero cell culture. The titer of virus was determined by the degree of inhibition of hemadsorption to infected cells of chicken red blood cells and expressed in the HADU50 (50% of heme adsorbing units). It was found that FP caused a significant decrease in the intensity of the hemadsorption process. The antiviral efficiency of FP in relation to the influenza virus A was studied on the cell culture of embryonic chicken fibroblasts (CF). The allantois version of the type A influenza virus, a WSN strain of the H1N1 serotype with various multiplicity of infection was used. The original titer of the virus was 4 × 105 PFU/mL. The introduction of 200 μg/mL of FP into the culture after 1 hour of contact of cells with the virus led to 80% protection of the cell layer when using 1 (5 × 102 PFU/cell) and 10 infectious doses of the virus. Pre-treatment of the virus with FP for 1 h increased the protective effect of the drug. It was 100% at 1 and 10 doses and 80% at 100 doses of the virus. 100% protection was also achieved during preliminary processing of cells with FP within 1 hour before the administration of the virus (Pronin et al., 2005). Similar results were obtained in the study of antiviral activity of PPF preparations in relation to the highly pathogenic strains of the bird influenza virus A (H5N1), allocated from a deceased domestic chicken in the Novosibirsk region during the epizootic in July 2005, the strain of the influenza virus was highly pathogenic for the cultures of the kidney cells of the pig embryo (SPEV cell), which was accompanied by the development of intensive CPE. It was found that FP and GP suppressed the infectious activity of the virus in the SPEV cells infected with the virus at doses of 10.0 and 100.0 TCD50/mL (tissue cytopathogenic dose of the virus) within 48 hours of cell infection (observation period). The maximum antiviral activity of FP and GP preparations was manifested during the processing of SPEV cell culture 1 hour before infection and at the time of infection of the cells with the bird influenza virus at a dose of 10 TCD50/mL (Pronin et al., 2006; Sanin et al., 2017b). When studying the effects of PPF preparations on the reproduction of the feline rhinotracheitis virus grand strain, the sensitive culture of CRFK cells (Crandell Feline Kidney cell line) was infected with the virus with a multiplicity of infection 5–10 PFU/cell. Using gel electrophoresis, it was shown that FP and GP at a dose of 200 μg/mL were supplied with the accumulation of viral proteins in this culture of cells (Sanin et al., 2015). With the titration of the bovine rhinotracheitis virus (type 1 herpes virus), strain 4016, on the Taurus-1 cell culture, the titer of virus under the influence of FP (200 μg/mL) decreased 100 times (by 2 log CPE50/mL) compared with the titer of the virus in control. A similar result was obtained when GP at a dose of 100 μg/mL was added to the lung culture of cow embryos (LCE). At the same time, they also observed the inhibition of the development of the CPE of the virus (Ozherelkov et al., 2001). FP also suppressed the reproduction of the bovine rhinotracheitis virus for 3–4 log TCD50/mL (Glotov et al., 2004). Experiments on the study of in vitro antiviral activity of PPF in relation to the Aujeszky’s disease virus (SSCI strain) were carried out on a sensitive culture of CF cells in the presence of GP (200 μg/mL). At the same time, the suppression of the reproduction of the virus at 2.25 log CPE50/mL was observed (Sanin et al., 2018a). When studying the influence of PPF on the reproduction of the vesicular stomatitis virus (fam. Rhabdoviridae, genus Vesiculovirus) in the culture of human fibroblasts (M-21, M-27 lines), it was found that FP in the concentration of 200 μg/mL significantly inhibits the infectious titers of the virus. The MT4 CD4-cells permissive for the reproduction of HIV-1 were used in experiments on the study of the influence of PPF on the human immunodeficiency virus. PPFs with Azidothymidine (AZT) PPF-AZT or AZT (as positive control) at concentrations of 10, 25, 50, 100, and 200 μg/mL were added to the infected HIV-1 cells. The results were evaluated by the number of antigen-positive cells and the degree of inhibitory of HIV-1 reproduction. As a result, it was found that in vitro FP has a dose-dependent inhibitory effect in relation to HIV-1. A decrease in the number of antigen -positive cells to 5% (at 50% in control) was found at the concentration of PPF of 100 and 200 μg/mL. At the same time, the degree of inhibition of HIV-1 reproduction was 90% (with zero value in control). The use of the PPF-AZT conjugate with a molar ratio of PPF : AZT 9 : 1 showed, just like in the positive control, a 100% inhibition of HIV-1 virus in relation to the permissive for reproduction of the HIV-1 line of human CD4-cells MT4 in all used PPF concentrations (Danilov et al., 1997). The FRhK-4 monolayer culture, which was infected with a suspension of the hepatitis A virus (HAV) and incubated for 24 days at 37°C with a weekly change in the maintenance environment, was used for studying the influence of PPF on the reproduction of the hepatitis A virus (fam. Picornaviridae, genus Hepatovirus). The content of HAV antigen was determined in cell lysates on the 7th and 24th day after infection with the solid-phase immunoenzyme assay, the suppression of the products of HAV antigens was established when introducing FP an hour before infection of cell cultures and simultaneously with infection after 7 and 24 days, respectively (Danilov et al., 1997). The experiments on the influence of PPF on the reproduction of the Theiler’s murine encephalomyelitis virus (fam. Picornaviridae, genus Cardiovirus) in the VNK-21 and P388 cells showed that GP in a dose of 50 and 100 μg/mL suppressed the reproduction of the virus, which was evaluated to suppress the ripening of the structural protein VP3 (2-3-fold) in both cell cultures (Kozhevnikova et al., 2007). The antiviral effect of PPF was also studied with an experimental model of infection caused by a hepatitis C virus (fam. Flaviviridae, genus Hepacivirus) in the SPEV cell culture. The cytopathogenic version of the virus isolated from blood serum of a chronically infected patient was used. It was found that FP has antiviral properties when it is added to the cell culture immediately after infection with the HCV. Under the influence of PPF, the HCV titers were reduced by 3.0 log (60 μg of PPF) and by 1.9 log (30 μg of PPF). A positive effect was also obtained in with preventive use of PPF (Narovlyansky et al., 2012). When evaluating the action of PPF on the reproduction of another representative of fam. Flaviviridae, the virus of yellow fever in the PS cell culture (pig fetal kidney cells) it was found that both FP and GP at doses of 100 μg/mL reduced the infectious titer of the virus by 1.9–2.3 log PFU50/mL (Ozherelkov et al., 2017a). Similar data were obtained under the influence of PPF preparations in relation to two more flaviviruses: bovine diarrhea virus, BDV (fam. Flaviviridae, genus Pestivirus) and tick-borne encephalitis virus (fam. Flaviviridae, genus Flavivirus). FP caused a statistically reliable decrease in the BDV model in the MDBK and the KST cell cultures, and its antiviral effectiveness significantly exceeded the virulicidale activity of Ribavirin, Anandin, Polyprenol, Erakond, and Gumiton preparations (Glotova et al., 2005; Glotov et al., 2004). It is important that the cultivation of the BDV in the presence of non-phosphorilated polyprenols led to the accumulation of it in culture fluid in the same amount as in control, which indicates that the phosphorylation of polyprenols enhances their antiviral efficiency by several orders (Sergeyev et al., 2004). FP and GP at a dose of 200 μg/mL suppressed the ability of BDV to multiply in KST cell culture by 30–80 times compared to control. At the same time, a slowdown in the manifestation of BDV for 24 and 48 hours was observed compared to control (Ozherelkov et al., 2017a). The influence of GP on the dynamics of the accumulation of viral proteins was evaluated in the precipitation reaction. It was established that GP in the early stages significantly suppresses the accumulation of proteins in the BDV in the KST cell culture (Sanin et al., 2018b). When evaluating the in vitro antiviral action of PPF in relation to the tick-borne encephalitis virus (TBEV) (Soph’in strain) a suspension of the brain of diseased mice-lugs infected intracerebrally was used. Experiments were carried out using J-96 monocytes on SPEV cell culture and human monocytic culture. As a result, it was found that FP and GP caused a decrease in the TBEV titer from 8.5 to 5.8–6.1 log PFU/mL. We also observed a decrease in the viral CPE by 2.3–2.5 times and a decrease in the release of mature virions from infected cells from с 6.4 × 108 to 1.2 × 107 PFU. A decrease in protein E from 0.69 to 0.07 μg/mL was found (Godunov, 2006; Ozherelkov et al., 2000, 2017a). The CC81 indicator system (feline kidney cells, transformed by the sarcoma mouse virus) was used for studying the action of PPF on the reproduction of cattle leukemia virus (fam. Retroviridae, the genus Oncovirus type C). Assessment of the reproduction of the virus was carried out according to the number of syncytia formed in cells under the influence of the virus. It was found that the introduction of FP (200–500 μg/mL) leads to a decrease in the infectious capacity of the virus by 56.3% (Danilov et al., 1997). The virus of canine infectious hepatitis (fam. Adenoviridae, genus Mastadenovirus) of the 1st type was cultivated in the cells of the MDCK (Madin-Darby canine kidney) line. Under the influence of FP (200 μg/mL), the suppression of the reproduction of the virus was found (Danilov et al., 1997). Thus, the experimental models of in vitro viral infections showed (Table 1) that PPF-based preparations have antiviral activity in relation to DNA-containing (Aujeszky’s disease virus, feline rhinotracheitis virus, cattle infectious rhinotracheitis virus) and RNA-containing enveloped viruses (canine distemper virus, measles virus, epidemic mumps virus, bovine diarrhea virus, bovine leukemia virus, vesicular stomatitis virus, yellow fever virus, tick-borne encephalitis virus), and also suppress the replication of non-enveloped type 1 adenovirus and picornaviruses (hepatitis A, Theiler’s murine encephalomyelitis virus). Table 1. Antiviral efficacy of PPF-based drugs in vitro Virus NA Capsid PPF drug, dose Observed effect Measles virus (fam. Paramyxoviridae, genus Morbillivirus) RNA + FP, 200 µg/mL Reducing the CPE of the virus on cells Canine distemper virus (fam. Paramyxoviridae, genus Morbillivirus) strain Rockborn RNA + FP, 200 µg/mL Infectious titer decreased in 4647 cell culture by 2.3 log Mumps virus (fam. Paramyxoviridae, genus Rubalavirus) RNA + FP, 200 µg/mL Reducing the degree of damage to virus-infected cells and reducing the infectious titer of the virus Influenza A virus (fam. Orthomyxoviridae, genus Influenzavirus А) RNA + FP, 200 µg/mL Suppression of the viral CPE, protection of the cell monolayer Avian influenza A virus serotype H5N1 (fam. Orthomyxoviridae. genus Influenzavirus А) RNA + FP, GP, 200 µg/mL Complete suppression of the viral CPE when administered 1 hour before infection of SPEV cells with the virus at a dose of 10–100 TCD50 Feline rhinotracheitis virus (fam. Herpesviridae, genus Alphaherpesvirus) Grand strain DNA + FP, 200 µg/mL GP, 200 µg/mL Suppression of viral protein accumulation in sensitive CFRK cell culture Infectious bovine rhinotracheitis virus (fam. Herpesviridae, genus Varicellovirus), strain 4016 DNA + GP, 100–200 µg/mL FP, 200 µg/mL Suppression of virus reproduction in LCE cell culture, decrease in infectious titer by 2.0 log Suppression of maturation of viral proteins in Taurus cell culture Aujeszky’s disease virus (fam. Herpesviridae, genus Varicellovirus) SSCI strain DNA + GP, 200 µg/mL Suppression of virus reproduction by 2.25 log CPD50/mL Vesicular stomatitis virus (fam. Rhabdoviridae, genus Vesiculovirus), Indiana strain RNA + FP, 200 µg/mL Decreased CPD of the virus per cell and decreased infectious titer Human immunodeficiency virus (fam. Retroviridae, genus Lentivirus) RNA + FP, 200 µg/mL Reduction in the number of antigen-positive cells to 5% (at 50% in control). Suppression of HIV-1 reproduction in MT4 cells by 90% Hepatitis A virus (fam. Picornaviridae, genus Hepatovirus) RNA – FP, 100–200 µg/mL Inhibition of the production of HAV antigens upon the introduction of FP an hour before infection of cell cultures and simultaneously with infection Theiler murine encephalomyelitis virus (fam. Picornaviridae, genus Cardiovirus), GDVII strain RNA – FP, 50–100 µg/mL Suppression of virus reproduction and VP3 protein accumulation in BHK-21 and P388D1 cell cultures Hepatitis C virus (fam. Flaviviridae, genus Hepacivirus) RNA + FP, 30–60 µg/mL Decrease in HCV titers by 3.0–3.5 log TCD50 when PPF is administered immediately after cell infection or 24 hours before cell infection Yellow fever virus (fam. Flaviviridae, genus Flavivirus) RNA + FP, GP, 100 µg/mL Suppression of YFV reproduction in PS cell culture (decrease by 1.9–2.3 log PFU50/mL) Bovine diarrhea virus (fam. Flaviviridae, genus Pestivirus) RNA + GP, 200 µg/mL FP, 200 µg/mL Suppression of virus reproduction in CST cell culture by 1.8 log. Slowing down the manifestation of viral CPE by 24–48 hours Tick-borne encephalitis virus (TBEV) (fam. Flaviviridae, genus Flavivirus) RNA + FP, 400 µg/mL GP, 400 µg/mL TBEV titer decreased from 8.5 to 5.8 log PFU/mL. 2.3–2.5-fold decrease in CPE of TBEV per cell. Reducing the yield of mature virions from 6.4 × 108 to 1.2 × 107 PFU. Protein E content decreased from 0.69 to 0.07 µg/mL Bovine leukemia virus (fam. Retroviridae, genus oncovirus type C) RNA + FP, 200–500 µg/mL Infectious titer of virus decreased by 0.8–1.75 log in CC81 cell culture Canine infectious hepatitis virus (fam. Adenoviridae, genus Mastadenovirus) DNA – FP, 200 µg/mL Decreased infectious titer of virus NA—nucleic acid; CPE—cytopathogenic effect (the ratio of the number of destroyed cells to the number of survived cells); PFU— plaque-forming unit (the smallest amount of virus that can cause the formation of one negative colony (plaque); TCD50—tissue cytopathogenic dose of the virus that causes the death of 50% of the cells of the monolayer. ANTIVIRAL EFFICACY OF PPF IN IN VIVO EXPERIMENTS The antiviral activity of PPF-based drugs was studied with infections caused by the following viruses: Murine hepatitis virus (fam. Coronaviridae, genus Betacoronavirus); Western equine encephalitis virus (alpha virus infection) (fam. Togaviridae, genus Alphavirus); Herpes simplex virus type 1 (fam. Herpesviridae, genus Simplexvirus); Herpes simplex virus type 2 (fam. Herpesviridae, genus Simplexvirus); Aujeszky’s disease virus (fam. Herpesviridae, genus Varicellovirus); Feline rhinotracheitis virus (fam. Herpesviridae, genus Alphaherpesvirus); Cytomegalovirus (fam. Herpesviridae, subfam. Betaherpesvirinae); Feline panleukopenia virus (fam. Parvoviridae, genus Parvovirus); Murine ectromelia virus (fam. Poxviridae, genus Orthopoxvirus); Feline calicivirus (fam. Caliciviridae, genus Vesivirus); Rabies virus (fam. Rhabdoviridae, genus Lissavirus); Tick-borne encephalitis virus (fam. Flaviviridae, genus Flavivirus); Influenza A virus (fam. Orthomyxoviridae, genus Influenzavirus А). The antiviral efficiency of PPF was evaluated on the experimental model of coronavirus infection caused by a Murine hepatitis virus (Meshcherin’s strain). Mice were infected intra-abdominally (i/a) at a dose of 10 LD50, or orally (p/o) at a dose of 100 LD50. FP at a dose of 200 μg was introduced intramuscularly (i/m) or p/o daily for two weeks. As a result, the increase in the survival of mice was 40–60% (Narovlyansky et al., 2018). Alphaviral infection of mice caused by the Western equine encephalitis virus (California strain) at a dose of 10 LD50/0.1 mL showed that the introduction of FP subcutaneously (s/c) or i/a in doses of 10–100 μg/mouse according to treatment-and-prophylactic or treatment schemes increased the protection of mice by 40–50% and extended the ALE (average life expectancy) by 4–5 days (Narovlyansky et al., 2018). The administration of GP according to the treatment and preventive scheme (400 μg s/c) to the experimental model of lethal herpetic meningoencephalitis in mice weighing 7–8 g, caused by an intracerebral introduction of 0.03 mL of material containing 10 LD50 of the herpes simplex virus type 1 (HSV-1 strain L2) provided 47% protection in mice and extended the ALE to 7.8 days (5.9 days in control). In the therapeutic scheme of administration, the mortality rate decreased from 84 to 61%, and the ALE increased to 6–8 days (4.5 days in control) (Narovlyansky et al., 2014; Sanin et al., 2018f). Fortepren (FTP) administered i/m at a dose of 4 mg/kg of weight, had a pronounced statistically reliable (p < 0.01) antiviral effect in the model of the genital herpes of males of guinea pigs infected with the type 2 herpes virus. The maximum therapeutic effect of FTP (39.4%) was noted on the fifth day after infection. The average duration of the disease was reduced by 4.3 days compared to control (Narovlyansky et al., 2015b). The study of the influence of GP on the course of infection caused by the rhinotracheitis virus was carried out on non-pedigree male and female kittens aged 1.5 to 3.5 months. Infection was carried out at a dose of 0.5–1.0 mL (virus activity 6.0 TCD50/mL). At the onset of the appearance of clinical signs of the disease, GP was administered p/o at a dose of 0.3–0.7mL. Clinical recovery happened on the 3rd–8th day (disappearance of ulcers, a decrease in dehydration, an increase in activity, and improvement of the state of the wool), while the recovery in the control group was recorded on days 22–30 (Narovlyansky et al., 2005). The administration of GP to rabbits infected with the Aujeszky’s disease virus (ADV) reduced the mortality of ADV by 33%, while the ALE increased from 5.3 days in the control to 11.8 days. Infection of the rabbits with ADV diluted in GP, followed by the treatment with GP led to an increase in animal survival to 90%, which may indicate the presence of a virucidal action in relation to ADV (Narovlyansky et al., 2005). The administration of FP at a dose of 0.25 mL to the experimental model of Macaques, in which antibodies to cytomegalovirus were detected, led to an increase in the level of products of interferon-α (IFN-α), an important indicator of the state of antiviral immunity (Karal-ogly et al., 2007). The antiviral activity of PPF in relation to flavivirus infection was studied on an experimental model of Syrian hamsters of 4–5 weeks of age, which were infected subcutaneously with the tick-borne encephalitis virus (TBEV) B-383 strain at a dose of 103 LD50/1.0 mL. On the 6–7th day after infection, against the background of a developed clinical picture of the disease, hamsters were administered with 0.5–1.0 mg of FP i/m. 100% of animals in the control group of hamsters infected with TBEV died after 7–8 days. In the group of hamsters who received FP, the death of animals came only on the 18th day after infection, which corresponds to the normal course of acute infection. Daily injections of FP to animals in an extremely serious condition made it possible to keep them alive throughout the experiment (Narovlyansky et al., 2018). The inoculation of FP or GP to experimental mice infected by TBEV (Absettarov and Soph’in strains) led to a significant decrease in the mortality of mice and an increase in the ALE of mice by 2–2.5 times (Vasiliev et al., 2008; Ozherelkov et al., 2017a). At the same time, stimulation of the early products of the cytokines of IFN-γ, TNFα, IL-6 and IL-12 was noted (Kozhevnikova et al., 2008; Sanin et al., 2018e). The antiviral efficiency of FP was also studied in a lethal infection in mice caused by a type A influenza virus H1N1, the WSN strain. Two schemes were used: preventive, with a single introduction of FP at a dose of 5 μg/mouse at the time of infection, and therapeutic, with the introduction of FP 1 day after infection. The use of a prophylactic scheme has led to a decrease in mortality by 61.5% and an increase in ALE by 4.5 days with i/m administration. In the therapeutic scheme, a decrease in mortality by 50% and an increase in ALE by 3.7 days was shown. Similar results were obtained on mice with intranasal introduction of the influenza type A virus, AICHI 2/68 strain (Grigoryeva, 2004; Pronin et al., 2005). THERAPEUTIC EFFECTIVENESS OF PPF AGAINST VIRAL INFECTIONS IN VETERINARY PRACTICE All drugs, of which the active substances are PPFs, have confirmed antiviral efficiency (Deeva et al., 1998; Ozherelkov et al., 2001; Narovlyansky et al., 2012), which has been used in veterinary practice in the treatment of viral diseases of both small domestic and agricultural animals (Sanin et al., 2011b). In the Russian Federation, two drugs based on PPF: FP and GP, have been used successfully for a long time. FP has been used for the prevention and treatment of canine (canine distemper, infectious hepatitis, parvovirus enteritis, adenovirosis, papillomatosis, Aujeszky’s disease, infectious tracheobronchitis, etc.), as well as feline (panleukopenia, herpetic rhinotracheitis, calicivirosis, coronaviral infections, etc.) and other small pets viral infections (Deeva et al., 1998; Narovlyansky et al., 2005; Sanin, 2005; Sanin et al., 2008, 2015; Savoyskaya et al., 2008; Pereslegina et al., 2013; Rudneva et al., 2016; Savoyskaya and Kozhevnikova, 2019) for more than 25 years. The drug is also widely used in the animal husbandry practice (Deeva et al., 1998; Sanin et al., 2011b). GP is used mainly for the treatment of feline viral diseases (Sanin et al., 2009, 2018c). PPF-BASED DRUGS USAGE IN PREVENTION AND TREATMENT OF THE VIRAL INFECTIONS OF SMALL DOMESTIC ANIMALS In recent years, the incidence of domestic dogs has sharply decreased due to the improvement of the quality of vaccines and the ubiquitous vaccination of dogs against the most significant viral infections (rabies, canine distemper, infectious hepatitis, and parvoviral enteritis) (Shulyak, 2004; Polozyuk and Sergeyev, 2021; Schultz, 2006). Nevertheless, FP is still successfully used as a means of etiotropic therapy and as part of a comprehensive treatment regimen for certain cases of viral infections, for example, canine distemper (Glazunov and Kornev, 2016), parvoviral enteritis (Petrakova, 2018; Koshlyak and Kankalov, 2022), “aviary cough” syndrome (Usikova and Rachikhina, 2019), and outbreaks of diseases in nurseries (Dmytryshyn, 2012; Savoyskaya and Kozhevnikova, 2019). In addition, FP is successfully used as an adjuvant for vaccines (Kozhevnikova et al., 2006), contributing to both an increase in the titer of specific antibodies and increasing the immunity strength (Pronin et al., 2011; Ozherelkov et al., 2012). Under experimental conditions, the adjuvant properties of FP were detected with combined use with tick-borne encephalitis, parvoviral enteritis, rabies (Ozherelkov et al., 2017b; Ozherelkov and Kozhevnikova, 2020), and hepatitis C vaccines (Onishchuk et al., 2017; Masalova et al., 2018). The adjuvant activity of FP was also demonstrated after the immunization of dogs against rabies (Ozherelkov, 2018). Vaccinating dogs against rabies with various vaccines with an anti-rabid component (Multican-8, Nobivak-DHPPIL+R, and Eurican-DHPPI2-LR), combined with FP revealed an increase in the level of protective antibodies against rabies virus after 21 and 51 days after vaccination and the improvement of cellular and humoral factors of non-specific resistance (Gulyukin et al., 2013). In recent years, an increase in the incidence of domestic dogs with viral papillomatosis of the oral cavity has been noted. Although papillomas are regarded as benign neoplasms, the search for effective therapeutic agents is still relevant. This is explained by the fact that this infection can occur in a latent form, in which the carrier of viruses poses a threat to other dogs. In addition, there are frequent cases in which papillomas undergo a malignant transformation, degenerating into scaly carcinomas (Calvert, 1990). FP showed high therapeutic effectiveness in this disease, both at the initial stage of the formation of papilloma (Gordeeva et al., 2008) and in cases of chronic papillomatosis (Rudneva et al., 2016; Okolelova and Bobina, 2020). In cases of adenovirus infections and parainfluenza in dogs, the use of GP reduced the severity and duration of the disease (Savoyskaya et al., 2008). PPF is successfully used for the treatment of viral diseases: panleukopenia, calicivirus infection, rhinotracheitis, and coronavirus infections in domestic cats (Sanin et al., 2009). Panleukopenia is an acute viral infection with high mortality, reaching 90% in kittens and 60% in young and adult cats (Kruse et al., 2010). The causative agent is a non-enveloped DNA-containing parvovirus, which affects rapidly dividing cells, including hematopoietic stem cells (HSCs) of the bone marrow (Uzal et al., 2016). It results in pronounced immunodeficiency and panleukopenia, which determine both the severity and outcome of the disease. In a severe form of the disease at the peak of leukopenia, almost all proliferating elements in the bone marrow can be exhausted. The number of HSCs is minimal, and the remaining cells lose the ability to compensate for the lack of mature elements (Uzal et al., 2016). That is why the proven ability of PPF-based drugs to stimulate the mobilization of the HSCs is important in the treatment of panleukopenia, contributing to their emission from bone marrow partially devastated by parvoviruses into the bloodstream (Sanin et al., 2008). The circulating HSCs then purposefully migrate into the affected tissues and target organs, contributing to the reproduction of the cellular composition and restoring lymphohemopoiesis. Apparently, this property of PPF-based drugs is responsible for their increased therapeutic efficiency for panleukopenia compared to other antiviral agents. The inclusion of both FP (Pereslegina et al., 2017) and GP (Sanin et al., 2018с) in the treatment protocol of cats with panleukopenia in controlled studies significantly reduces the terms of clinical recovery and helps to restore structural and functional blood indicators. Similarly, the use of GP in complex therapy of cats with rhinotracheitis (Glotova et al., 2008), as well as calicivirus infection (Sanin et al., 2018d) weakens the inflammatory reaction, reduces treatment periods, and prevents the activation of secondary microflora. The use of PPF-based preparations is effective in the treatment of coronavirus infections in cats. Thus, the inclusion of FP in complex therapy of coronavirus gastroenteritis (the causative agent is alphacoronavirus AlphaCov 1) reduces the terms of clinical recovery (Zhavnis et al., 2019) and prevents the transition of the chronic form of infection into acute form, almost always resulting in death (Savoyskaya et al., 2021). The FECV coronavirus receptor, which causes gastroenteritis in cats, is membrane metalloprotease-alanyl aminopeptidase, or aminopeptidase N (APN)/CD13, the expression of which in all types of cells is regulated by Th1-cytokines and increases under the influence of IFN-γ and IL-4 (Tani et al., 2000). This is why PPFs possession of immunomodulating (Pronin et al., 2000), anti-inflammatory (Ganshina et al., 2011) and antioxidant (Sanin et al., 2017a) properties are so important in the therapeutic efficiency of these drugs against viral infections. The clinical efficiency of FP is also confirmed in the treatment of an effusive form of the infectious peritonitis caused by the FIP coronavirus (Pereslegina et al., 2013; Pereslegina and Zhavnis, 2019). In turn, GP is also effective (allows prolonged remission and improvement of the quality of life) in the treatment of the dry, or non-effusive form of infectious peritonitis of cats (Furman et al., 2010). Another PPF-based drug, Polyprenyl Immunostimulant, developed in the United States, also prolonged remission in cats with the dry FIP (Legendre et al., 2017). This disease is characterized by antibody-dependent enhancement of the viral infection (Takano et al., 2008). A similar phenomenon is also typical for a number of other infections, in particular, flaviviruses (Ozherelkov et al., 2008), in which the therapeutic efficiency of FP and GP is shown (Kozhevnikova et al., 2008; Ozherelkov et al., 2017a). EFFECTIVENESS OF PPF-BASED DRUGS IN THE ANIMAL HUSBANDRY AND POULTRY FP is used for the treatment and prevention of viral infections in productive animals and poultry (Sanin et al., 2011b). In poultry farming, FP is used at all stages of ontogenesis: for eggs in the incubator, for chicks, and for adult chicken (Sanin et al., 2011c). The use of FP aerosol on chickens led to a decrease in the incidence and mortality from the “respiratory syndrome” (Dementieva et al., 2007). In young chickens, FP reduces the incidence and increases natural resistance to illness (Goloveshchenko et al., 2002; Tyurina et al., 2006). In adult chickens, FP is used for vaccination against Newcastle disease, infectious bronchitis (pathogen is avian gammacoronavirus—ACoV), and infectious bursal disease (Goloveshchenko et al., 2002; Kushniruk et al., 2005). There are also good experimental grounds to recommend FP for the prevention of bird flu (Pronin et al., 2005, 2006). In pig farming, FP was used for prevention and therapy of transmissive gastroenteritis (Deeva et al., 2004). The use of FP as an adjuvant to sow and piglets, together with vaccines against European plague of the pigs and Aujeszky’s disease, led to a 2–4-fold increase in antibodies titers and a decrease in the incidence by 42% (Pronin, 2005). The use of FP in calves with a mixed infection caused by viruses of infectious rhinotracheitis, adenovirus diarrhea and parainfluenza-3 led to reduction of mortality by 15.6%, and reduction of the treatment length by 3 days (Krasota et al., 2011). The use of FP in calves with diarrhea, concomitantly infected with rotavirus, coronavirus, and bovine diarrhea virus, contributed to a decrease in mortality by 13.3% (Deeva et al., 2005). The use of FP as a part of complex therapy of calves with rotavirus infection also reduced mortality and treatment length, and prevented the loss of body weight (Avakyants et al., 2002). FP is also recommended for use against viral papillomatosis of cattle, the pathogen of which is BPV, bovine papillomavirus (online resource: https://fermer.ru/forum/veterinariya-krs/164503). The use of FP during the vaccination of cows against parainfluenza, infectious rhinotracheitis, and viral diarrhea as an adjuvant led to an increase in antibodies 2–4-fold compared to control (Pronin, 2005). In lambs, the use of FP resulted in a reduction in the treatment of animals from respiratory viral infections by 2–3 days and increased safety (Murzaliev and Zaitsev, 2017). In horses, FP is effective for the treatment of viral arteritis and influenza: the use of the drug relieved the course of diseases and reduced treatment periods by more than 2-fold. An increase in antibodies and a decrease in the frequency of post-vaccination complications from 20 to 3% was observed when the horses were vaccinated against leptospirosis and influenza using FP as an adjuvant (Zaitseva et al., 2006). It is also reported that FP is effective for the treatment of infectious stomatitis of rabbits caused by a filtering virus (Asadullina, 2008). The use of FP in minks infected with the Aleutian disease virus (fam. Parvoviridae, genus Amdoparvovirus) contributed to the increase in natural resistance and the preservation of young animals (Bespalova et al., 2007; Rostrosa et al., 2019). FP was used in the beekeeping for treating infections to honey bees caused by acute bee paralysis virus (RNA virus causing the mass death of bees) and a threaded virus (the only DNA-containing bee virus from the fam. Dicistroviridae, genus Cripavirus), which causes the death of larvae, pupae and adult bees. Bees received sugar syrup containing FP in an amount of 0.1 mg/mL as therapy. As a result, a significant (46.8%) increase in the life expectancy of bees was achieved (Batuev et al.,2003). PPF USAGE IN THE TREATMENT OF VIRAL INFECTIONS IN MEDICINE To date, only one drug based on PPF, Fortepren® (FTP), which belongs to the pharmacotherapy group of antiviral agents, has been registered in human medicine (Narovlyansky et al., 2008). The drug Ropren® is also known, but, unlike FTP, non-phosphorilated polyprenols serve as its active substance. The main indication for the use of FTP is the therapy of chronic recurrent genital infection (HRGVI) caused by the herpes simplex virus, in order to increase the duration of remission and reduce the severity of symptoms of relapse in the inter-relapse period in adults (Ershov et al., 2017). Clinical studies of FTP were conducted on 80 patients with a confirmed diagnosis of HRGVI of genital localization caused by HSV corresponding to the criteria of inclusion and selected in the screening process (Sedov et al., 2015). As a result, it was found that the inter-relapse in a group of patients who received FTP for the entire study statistically significantly increased from 29.36 ± 2.16 to 42.98 ± 3.29 days. Meanwhile, this indicator did not change in the control group. Also, patients who received FTP had a statistically reliable reduction in the frequency of relapse of the HRGVI from 3.03 ± 0.02 before the start of treatment to 1.94 ± 0.19 during treatment in the absence of a decrease in the frequency of relapse in the control (Narovlyansky et al., 2015a). In addition, the level of leukocyte virus-induced interferon had increased significantly in 64% of patients who received FTP at the end of the clinical study, while growth of interferon titers were not observed in the control group (Pronin et al., 2016). FTP can be used as part of a comprehensive therapy of chronic recurrent genital infection caused by the herpes simplex virus to increase the duration of remission, reduce the frequency of exacerbations and the reduce severity of the symptoms of relapse (Ershov et al., 2020). POSSIBLE MECHANISMS OF THE ANTIVIRAL ACTIVITY OF PPF-BASED DRUGS The ability of PPF to suppress the reproduction of a number of DNA and RNA-containing viruses (enveloped as well as non-enveloped ones) that play an important role in human and animal pathology was shown in both in vitro and in vivo experiments and clinical studies (Table 2). Since viral infections are almost always accompanied by immunosuppression, the search for therapeutic agents that have not only antiviral activity, but are also able to have an immunomodulating effect and show anti-inflammatory activity, is relevant. Studies of many scientists have shown that PPF-based drugs satisfy all these requirements. All PPFs stimulate the antiviral immune response (Pronin et al., 2021). After their administration into the body of a patient with a viral infection, they induce the early formation of IFN-α, IFN-γ, TNFα, IL-6, IL-12, and other key cytokines (Pronin et al., 2002; Kozhevnikova et al., 2008; Konyushko et al., 2020; Ozherelkov et al., 2002). Thus, PPFs can restore the balance of the Th1/Th2 immune response development necessary for the formation of effective antiviral immunity, that is violated by viruses (Pronin et al., 2000; Ozherelkov et al., 2012; Sanin et al., 2018e). The weakening of the inflammatory reaction under the effect of PPFs is already noted in the early stages of therapy. PPFs suppress the activity of 5- and 15-lipoxygenases (Ganshina et al., 2011), and counter-regulate the action of an important pro-inflammatory cytokine, the macrophage migration inhibitory factor, or MIF (Sanin et al., 2011a). Table 2.   Antiviral efficacy of PPF-based drugs in vivo Virus NA Capsid PPF drug, dose Observed effect Mouse hepatitis virus (fam. Coronaviridae, genus Betacoronavirus), Meshcherin strain RNA + FP, 200 µg/mouse Therapeutic effect (increase in the survival of mice by 40–60%) with repeated daily p/o or i/p administration for 2 weeks Feline coronavirus (fam. Coronaviridae, genus Alphacoronavirus) RNA + FP, GP, 2.5–3 mL PI, 0.25–0.5 mL/kg Reduced duration and severity of disease, improved quality of life Western equine encephalitis virus (fam. Togaviridae, genus Alphavirus), California strain RNA + FP, 10–100 μg/mouse A pronounced therapeutic and prophylactic effect with s/c and i/p administration. Increase in ALE up to 9.5 days compared to 4.3 days in control Herpes simplex virus type 1 (fam. Herpesviridae, genus Simplexvirus), L-2 strain DNA + FP, GP, 200 µg/mouse Protective effect in the treatment-and-prophylactic scheme of administration (47% protection with s/c injection). Lethality of mice decreased from 84% to 61%. Increase in ALE up to 6.8 days compared to 4.5 days in control Herpes simplex virus type 2 (fam. Herpesviridae, genus Simplexvirus), ЕС strain DNA + FTP, 4 mg, 4-fold Reducing the average duration of the disease by 4.3 days compared to control Aujeszky’s disease virus, (fam. Herpesviridae, genus Varicellovirus), SSCI strain DNA + GP, 2.5–5 mL Decrease in mortality by 33%, increase in ALE from 5.3 days (in control) up to 11.8 days Feline rhinotracheitis virus (fam. Herpesviridae, genus Alphaherpesvirus) Grand strain DNA + GP, 0.5–1.0 mL GP, 1.0 mL Reducing the symptoms of the disease in kittens and significant acceleration of recovery. Reducing the recovery time of cats and suppressing the reproduction of the virus on the nasal mucosa Feline herpes virus type 1 (fam. Herpesviridae, genus Alphaherpesvirus), SGE strain DNA + PI, 0.25–0.5 mL/kg Reducing the severity of the disease and reducing the recovery time Cytomegalovirus (fam. Herpesviridae, subfam. Betaherpesvirinae) DNA + FP, 0.25 mL Increased production of IFN-α in rhesus monkeys with antibodies to cytomegalovirus Feline panleukopenia virus (fam. Parvoviridae genus Parvovirus) DNA, single stranded – FP, 0.5 mL GP, 0.5 mL Reduction of terms of clinical recovery, restoration of structural and functional parameters of blood Mouse ectromelia virus (fam. Poxviridae, genus Orthopoxvirus) DNA + FP, 5–25 µg/mouse Reducing the severity of the disease, clinical recovery Feline calicivirus (fam. Caliciviridae, genus Vesivirus) RNA – GP, 0.5–1.0 mL Rapid attenuation of the inflammatory response, reducing the severity of the disease in kittens, reducing the recovery time Rabies virus (fam. Rhabdoviridae, genus Lissavirus), CVS strain RNA + FP, 5–25 µg/mouse An increase in the survival rate of mice up to 65–68% (20% in the control) with a single and 3-fold i/p administration after infection Tick-borne encephalitis virus (fam. Flaviviridae, genus Flavivirus) B-383 strain RNA + FP, 0.5–1.0 mg Increased life expectancy of infected Syrian hamsters up to 18 days (7–8 days in control) Tick-borne encephalitis virus (fam. Flaviviridae, genus Flavivirus) Absettarov’s strain RNA + FP, 5–20 µg/mouse GP, 100 µg/mouse Reducing the lethality in mice to 70% (with 100% in the control) and increasing the life expectancy of mice by 2–2.5 times. Stimulation of early production of IFN-γ and IL-12 Influenza A virus (fam. Orthomyxoviridae. genus Influenzavirus А) WSN A (H1N1) strain RNA + FP, 5 µg/mouse Decreased lethality in mice by 61.5% and ALE increased by 4.5 days Papillomavirus (fam. Papillomaviridae, genus Papillomavirus) DNA – FP, 0.3 mL/kg Reducing symptoms, shortening clinical recovery p/o—oral administration: i/p—intraperitoneal administration: s/c—subcutaneous administration; i/m—intramuscular injection; i/n— intranasal administration; ALE—average life expectancy; FTP—Fortepren; PI—Polyprenyl Immunostimulant, IFN—interferon. The direct antiviral effect of PPFs is manifested in violation of one or more of the main stages of the virus life cycle: the adsorption of the virus on the surface of the cell, penetration into the cell, assembly and/or prenylation and glycosylation of viral proteins. The mechanisms by which PPF suppresses the reproduction of viruses include non-specific binding with virions outside the cell. Thus, FP has a pronounced hemagglutinating activity against goose erythrocytes (EG) and does not have such activity in relation to canine or feline red blood cells. TBEV also agglutinates the EG, so the reaction of hemagglutination inhibition in the presence of antiviral antibodies is used to determine antibodies titers to TBEV. FP is shown to block the hemagglutinating activity of TBEV, and vice versa: TBEV slows down the hemagglutination of the EG by FP, which indicates the ability of FP to interact with the virions outside the cell, forming stable FP-virus complexes, thereby preventing cell infection (Ozherelkov, 2003). PPFs can also prevent the adsorption of the virus on cells, violating the process of reception of viruses on the membrane. Increasing the fluidity and permeability of membranes, PPFs can violate the fusion process of the lipid membrane of a number of capsid viruses with a membrane of the target cell. So, it was shown that FP added into a culture 1 hour before infection suppresses the cytopathogenic activity of the H5N1 bird influenza virus. After inoculation of FP simultaneously with the virus all cells survive, while in the control all of them die. FP prevents the desialization of the virus glycoproteins and the host cell viral neuraminidase, contributing to the formation of conglomerates of viral particles, in which the virions life cycle breaks off. Sialic acids are N- or O-acyl derivatives of neuraminic acid, the attachment of which occurs only in the presence of PPF. In addition, the key event in the stage of adsorption of capsid viruses is the merging of their lipid membrane with a plasma membrane. One can affect the adsorption process by changing the lipid composition of the membranes. FP can violate the fusion process of the lipid capsid of capsid viruses with a membrane of the target cell by increasing the fluidity and permeability of membranes (Pronin et al.,2005). The suppression of the synthesis of viral proteins is another mechanism of PPF antiviral activity. Thus, radioimmunoprecipitation assay showed that the introduction of FP into the cell culture 8 hours after their infection with TBEV suppresses the accumulation of capsid protein E (Ozherelkov et al., 2000; Godunov, 2006). FP also reduced the synthesis of the capsid protein VP3 of the Theiler’s murine encephalomyelitis virus 2–3-fold, which was detected using Western blot analysis (Kozhevnikova et al., 2007). In addition, FP suppressed the ripening of structural proteins of viruses of infectious rhinotracheitis, hepatitis C, and adenovirus in cell culture (Narovlyansky et al., 2018). Since PPF serves as an intermediate acceptor of sugars during protein glycosylation, it affects almost all the stages of the interaction of the virus with the cell. According to some parameters, PPF behaves as lectin, specific to the mannose, galactose, and N-acetylglucosamine glycoproteins, which can lead to suppression of the binding of the virus with receptors. The key mechanisms of antiviral activity of PPF include the suppression of prenylation of viral proteins. Prenylation is the process of post-translation modification of proteins, in which the lipophilic isoprenyl group joins the viral protein synthesized de novo. The first stages of viral prenylation occur in the cytoplasm of the target cell, while the final one occurs in the endoplasmic reticulum. Prenylated proteins participate in almost all stages of the virus life cycle: binding to the cell, penetration into the cell and the nucleus, and replication of the viral genome (Jeong et al., 2018). Inhibition of prenylation violates the assembly and production of viral particles. Therefore, farnesyl-transferase (enzyme that transmit farnesyl to the C-terminal cysteine of target protein) inhibitors suppress the replication of many viruses (Glenn et al., 1998; Asselah et al., 2020). PPF presumably inhibits viral prenyltransferases, which, in combination with the suppression of glycosylation, leads to a violation of the assembly of the virions and the formation of defective particles. This process is described for FP in relation to TBEV and is confirmed by electronic micrographs (Fig. 1). Fig. 1. Defective TBEV virions after treatment with FP (top). Normal virions are at the bottom. The administration of PPF in vivo leads to the production of interleukin-1, the induction of IFN, and the launch of the IFN-mediated mechanisms for the suppression of the synthesis of isoprenoid metabolites in the mevalonate pathway. It is also shown that the synthesis of IFN-α and IFN-γ is induced after the interaction of PPF with TLR2/TLR4 and the calcium signal supply. It is assumed that the IFN inhibits the expression of the SREBP-2 transcription factor, and as a result blocks the path of mevalonic acid and the formation of early precursors of polyprenols, which are necessary for prenylation of viral proteins and formation of mature virulent viral particles (Pronin et al., 2021). While some mechanisms of antiviral activity of phosphorilated polyprenols are clarified in relation to the enveloped viruses (obstruction of their penetration into a target cell, violation of synthesis, prenylation and glycosylation of viral proteins, formation of defective virions, etc.,) there remains a lot unclear in the suppression of the reproduction of non-enveloped viruses. It has been shown that GP suppressed the reproduction of the Theiler’s murine encephalomyelitis virus by inhibiting the synthesis of viral protein VP3 (Kozhevnikova et al., 2007). At the same time, it is known that many non-enveloped viruses (parvoviruses, adenoviruses, picornaviruses, and caliciviruses) use clathrin-mediated endocytosis (Stuart and Brown, 2006). Other non-enveloped viruses, for example Simian virus 40 (SV40), use caveolae to enter the cell, thus avoiding degradation in lysosomes (Simons and Ehehalt, 2002), while the human echovirus 11 and the Coxsackie B4 virus penetrate into the cell through a cholesterol-dependent mechanism and lipid rafts: nanostructural complexes of cholesterol and sphingolipids, whose integrity is maintained by cholesterol, act as a kind of mobile spacer between the molecules of sphingolipid. Their removal leads to a loss of functional activity of raft (Stuart and Brown, 2006). Caveolae participate in membrane transport and the formation of the reaction to an external signal. Transmembrane signal transmission, which is apparently carried out in rafts and, possibly, the caveolae, is associated with receptors of many growth factors, as well as some GTP-binding proteins and protein kinases. In addition, the caveolae are involved in the regulation of calcium signaling pathways. All this suggests the existence of mechanisms by which PPF-based drugs can suppress the reproduction of non-enveloped viruses. Thus, PPFs can violate the penetration of the mentioned agents into target cells, preventing cholesterol synthesis (Pronin et al., 2014) and competitively displacing it from lipid rafts. It is curious that plant flavonoids (Seo et al., 2015) have a pronounced antivirus effect in relation to caliciviruses, including feline calicivirus, which are characterized by the same changes in antioxidant and prooxidant properties as phosphorylated polyprenols (Sanin et al., 2017a). CONCLUSIONS Thus, taking into account the variety of mechanisms of suppressing the reproduction of viruses belonging to a large variety of taxonomic categories containing both RNA and DNA (single-stranded and double-stranded), enveloped and non-enveloped ones, we may conclude that phosphorilated polyprenols act as a universal master key that breaks the evolutionary security codes of viruses. COMPLIANCE WITH ETHICAL STANDARDS Conflict of Interests The authors declare the existence of a conflict of interest. A.V. Pronin, A.N. Narovlyansky, A.V. Sanin, and S.V. Ozherelkov participated in the theoretical development and practical use of the drugs Phosprenyl and Gamapren, and A.V. Pronin, A.N. Narovlyansky, and A.V. Sanin are also the developers of the medical drug Fortepren®, which is registered by the Ministry of Health of the Russian Federation. All applicable international, national and/or institutional guidelines for the care and use of animals have been observed. 1 Taxonomic affiliation of viruses was determined by: Virus Taxonomy…, 2012. Translated by M. Shulskaya ==== Refs REFERENCES 1 Antipina A.A. Popov V.S. ad Balaban’yan, V.Yu., Polyprenols as an original class of natural compounds with a wide spectrum of pharmacological activity Farmatsiya 2021 70 15 21 10.29296/25419218-2021-05-02 2 Asadullina I.I. The effect of the antiviral drug fosprenil on the growth and meat productivity of rabbits with infectious stomatitis Vestn. Bashkirskogo Gos. Agrar. Univ. 2008 11 8 9 3 Asselah T. Loureiro D. Tout I. Future treatments for hepatitis delta virus infection Liver International 2020 40 54 60 10.1111/liv.14356 32077603 4 Avakyants, B.M., Belousova, R.V., Rodin, A.N., et al., Experience in the treatment of rotavirus infection, Veterinarnyi Konsul’tant, 2002, no. 14, p. 8. 5 Batuev, Yu.M., Berezina, L.K., Narovlyansky, A.N., et al., Fosprenil activity on a honey bee model, Veterinariya, 2003, no. 5, pp. 26–28. 6 Bespalova, T.A., Sidorov, G.N., and Khitrova, E.A., Correction of the immune status of healthy minks and minks infected with the Aleutian disease virus, Veterinarnaya Patologiya, 2007, no. 3 (22), pp. 250–253. 7 Calvert C.A. Infectious Diseases of the Dog and Cat 1990 Philadelphia WB Saunders 8 Danilov L.L. Maltsev S.D. Deyeva A.V. Phosprenyl: a novel drug with antiviral and immunomodulatory activity Arch. Immunol. Ther. Exp. 1997 44 395 400 9 Deeva, A.V., Ozherelkov, S.V., Zhukova, S.L., et al., Fosprenil is a broad-spectrum antiviral drug, Veterinarnaya Praktika, 1998, no. 1 (4), pp . 12–22. 10 Deeva, A. V., Mishchenko, N.K., Lobova, T.P., et al., The use of Fosprenil for the prevention and treatment of transmissible gastroenteritis in pigs, Veterinariya, 2004, no. 2, pp. 12–15. 11 Deeva, A.V., Rakova, T.N., Lobova, T.P., et al., Efficiency of Fosprenil for increasing nonspecific resistance of organism and treatment of acute viral infections in young cattle, Veterinarnaya Patologiya, 2005, no. 1, pp. 96–98. 12 Dement’eva, V.A., Mekhdikhanov, G.G., Deeva, A.V., et al., Nonspecific prevention of respiratory diseases in birds with aerosol application of Fosprenil, Veterinariya, 2007, no. 12, pp. 16–17. 13 Dmytryshyn, O.P., The use of immunomodulators in viral infections of dogs, Naukovii Visnik L’vivs’kogo Natsional’nogo Universitetu Veterinarnoï Meditsini ta Biotekhnologii imeni S.Z. Gzhits’kogo, 2012, vol. 14, nos. 2–1 (52), pp. 102–108. 14 Ershov F.I. Pronin A.V. Sanin A.V. Narovlyansky A.N. Combination of traditional therapy of genital herpes with immunotherapy: experience of using domestic immunotherapeutic drugs Usp. Sovrem. Biol. 2020 140 263 277 15 Ershov F.I. Narovlyansky A.N. Pronin A.V. Sanin A.V. Advances in Medicine and Biology. 2017 New York Nova Science Publishers, Inc 16 Furman, I.M., Vasil’ev, I.K., Narovlyansky, A.N., et al., Phosphorilated plant polyprenyl compounds are probably able to control the dry form of FIP in cats, Rossiskii Veterinarnyi Zhurnal. MDZh, 2010, no. 3, pp. 42–43. 17 Ganshina I.V. Sud’ina G.F. Sanina V.Yu. Phosphorylated polyprenols—a novel class of compounds with anti-inflammatory and bronchial spasmolytic activity Infektsii i Immunitet 2011 1 355 360 18 Glazunov Yu.V. Korneva V.S. The effectiveness of “Fosprenil” in the treatment of the intestinal form of canine distemper Molodoi Uchenyi 2016 26 225 227 19 Glenn J.S. Marsters J.C. Greenberg H.B. Use of prenylation inhibitor as a novel antiviral agent J. Virol. 1998 72 9303 9306 10.1128/JVI.72.11.9303-9306.1998 9765479 20 Glotov, A.G., Glotova, T. I., Sergeev, A.A., et al., Antiviral activity of different drugs in vitro against viruses of bovine infectious rhinotracheitis and bovine diarrhea, Vopr. Virusol., 2004, no. 5, pp. 43–45. 21 Glotova, T.I., Glotov, A.G., Belkina, T.V., et al., Activity of drugs against bovine VDV, Veterinariya, 2005, no. 8, pp. 26–28. 22 Glotova, T.I., Glotov, A.G., Russkikh, V.V., and Tugu-nova, T.B., Antiviral activity of the new agent Gamapren against feline rhinotracheitis virus, Sib. Vestn. S-kh. Nauki, 2008, no. 10, pp. 63–68. 23 Godunov, R.S., Immunomodulatory and antiviral activity of polyprenols of natural origin, Extended Abstract of Cand. Sci. (Med.) Dissertation, Moscow: Research Institute of Epidemiology and Microbiology named after N.F. Gamalei RAMS, 2006. 24 Goloveshchenko, A.A., Deeva, A.V., Goloveshchenko, K.A., et al., The use of Fosprenil in the fattening of broiler chickens, Veterinariya, 2002, no. 12, pp. 14–16. 25 Gordeeva, E.V., Vasil’ev, I.K., Narovlyansky, A.N., et al., Papillomatosis of the oral cavity of dogs—a new approach to treatment, Rossiskii Veterinarnyi Zhurnal. MDZh,, 2008, no. 2, pp. 15–17. 26 Grigor’eva, E.A., Immunomodulatory effect of the antiviral drug Fosprenil in normal conditions and in experimental influenza infection, Extended Abstract of Cand. Sci. (Biol.) Dissertation, Moscow: Research Institute of Epidemiology and Microbiology named after N.F. Gamalei RAMS, 2004. 27 Gulyukin, A.M., Khismatullina, N.A., Gafarova, A.Z., et al., Assessment of the immune status of dogs vaccinated with various rabies vaccines in combination with immunostimulants, Veterinarnaya Meditsina, 2013, no. 97, pp. 266–268. 28 Jeong A. Suazo K.F. Wood W.G. Isoprenoids and protein prenylation: implications in the pathogenesis and therapeutic intervention of Alzheimer’s disease Crit. Rev. Biochem. Mol. Biol. 2018 53 279 310 10.1080/10409238.2018.1458070 29718780 29 Karal-ogly, D.D., Mezentseva, M.V., Agrba, V.Z., et al., The effectiveness of the use of new domestic drugs Fosprenil and Gamavit in monkeys infected with cytomegalovirus, in Interferonu – 50 let. Yubileinyi sbornik, posvyashchennyi otkrytiyu interferonov, 100-letiyu akademika AMN SSSR V.D. Solov’eva i 45-letiyu otdela interferonov GU NIIEM im. N.F. Gamalei RAMN (Interferon Is 50 Years Old. Anniversary Collection Dedicated to the Discovery of Interferons, the 100th Anniversary of Academician of the Academy of Medical Sciences of the USSR V.D. Solovyov and the 45th Anniversary of the Department of Interferons of the Research Institute of Epidemiology and Microbiology named after N.F. Gamaleya RAMS), Moscow, 2007, pp. 204–213. 30 Konyushko O.I. Ozherelkov S.V. Vorovich M.F. Lines of diploid human fibroblasts—a test system for studying the immunomodulatory properties of modern immunobiological preparations, Byulleten Eksperimental’noy Biologii i Meditsiny 2020 170 188 191 31 Koshlyak, V.V. and Kankalova, A.V., Breed predisposition of dogs to parvovirus enteritis, therapeutic and cost-effectiveness of treatment methods, Mezhdunar. Nauch.–Issled. Zh., 2022, nos. 1–2 (115), pp. 42–46. 32 Kozhevnikova, T.N., Vorovich, M.F., Kozlov, V.G., et al., The use of Fosprenil as an adjuvant for vaccines and a stimulator of the production of specific antibodies in the manufacture of hyperimmune sera, Rossiskii Veterinarnyi Zhurnal. MDZh, 2006, no. 2, pp. 8–10. 33 Kozhevnikova, T.N., Viktorova, E.G., Kozlov, V.G., et al., Moraprenyl phosphates inhibit replication of Theiler encephalomyelitis virus and accumulation of VP3 viral protein in sensitive BHK-21 and P388D1 cell cultures, Zh. Mikrobiol., Epidemiol. Immunobiol., 2007, no. 3, pp. 26–30. 34 Kozhevnikova, T.N., Ozherelkov, S.V., Izmest’eva, A.V., et al., Influence of drugs Gamapren and Fosprenil, created on the basis of plant polyprenols, on the production of some regulatory cytokines in normal and experimental tick-borne encephalitis in mice, Rossiiskii Immunologicheskii Zhurnal, 2008, vol. 2, no. 11, nos. 2–3, pp. 250. 35 Krasota, A.Yu., Alekseeva, N.Yu., Kurbanbekova, Z.D., et al., New approaches to the treatment of respiratory and intestinal viral infections in cattle using the drug “Fosprenil,” Veterinariya Sel’skokhozyaistvennykh Zhivotnykh, 2011, no. 5, pp. 24–25. 36 Kruse B.D. Unterer S. Horlacher K. Prognostic factors in cats with feline panleukopenia J. Vet. Intern. Med. 2010 24 1271 1276 10.1111/j.1939-1676.2010.0604.x 21039863 37 Kushniruk, T.N., Segal, I.N., and Yakovleva, E.G., Effect of fosprenil and echinacea on the intensity of immunity to Newcastle disease in broiler chickens, Byulleten’ Nauchnogo Rabotnika, Belgorod: Belgorod. Gos. Sel’skokhoz. Akad., 2005, no. 4, pp. 55–58. 38 Legendre A.M. Kuritz T. Galyon G. Polyprenyl immunostimulant treatment of cats with presumptive non-effusive feline infectious peritonitis in a field study Frontiers in Veterinary Science 2017 4 7 20 10.3389/fvets.2017.00007 28261584 39 Masalova O.V. Lesnova E.I. Onishcuk A.A. Polyprenyl phosphates induce a high humoral and cellular response to immunization with recombinant proteins of the replicative complex of the hepatitis C virus Doklady Biochem. Biophys. 2018 482 261 263 10.1134/S1607672918050083 40 Murzaliev I. Dzh. and Zaitseva, O.O., Evaluation of the immunological activity of antiviral drugs fosprenil and forvet Uch. Zap. UO VGAVM 2017 53 104 106 41 Narovlyansky, A.N., Berezina, L.K., Vetkova, L.G., et al., Treatment of herpesvirus infections with polyprenols, Veterin. Klin., 2005, no. 2, pp. 11–15. 42 Narovlyansky, A.N., Vasil’ev, A.N., Savoiskaya, S.L., et al., Isoprenoid system: role in antiviral immunity, Vedomosti Nauchn. Tsentra Ekspert. Sredstv Med. Primen., 2007, no. 3, pp. 66–78. 43 Narovlyansky, A.N., Parfenova, T.M., Vasil’ev, A.N., et al., Fortepren is a new promising antiherpes drug based on polyprenyl phosphates, Rossiiskii Immunologicheskii Zhurnal, 2008, vol. 2, no. 11, nos. 2–3, p. 256. 44 Narovlyansky, A.N., Deryabin, P.G., Sedov, A.M., et al., Antiviral activity of polyprenyl phosphates in experimental infection caused by hepatitis C virus in vitro, Zh. Mikrobiol. Epidemiol. Immunol., 2012, no. 5, pp. 80–84. 45 Narovlyansky, A.N., Ozherelkov, S.V., Sanin, A.V., et al., Antiviral activity and possible mechanisms of action of moraprenyl phosphates during experimental infection caused by herpes simplex type 1 virus, Zh. Mikrobiol. Epidemiol. Immunol., 2014, no. 5, pp. 54–60. 46 Narovlyansky A.N. Ivanova A.M. Shevlyagina N.V. Efficacy of polyprenyl phosphates in the experimental genital herpes model Vopr. Virusol. 2015 60 9 13 47 Narovlyansky, A.N., Sedov, A.M., Pronin, A.V., et al., Treatment of patients with chronic recurrent herpes virus infection of genital localization: a clinical study of fortepren preparation, Zh. Mikrobiol. Epidemiol. Immunol., 2015a, no. 4, pp. 112–118. 48 Narovlyansky A.N. Pronin A.V. Sanin A.V. Isoprenoids: Polyprenols and Polyprenyl Phosphates as Physiologically Important Metabolic Regulators 2018 New York Nova Sci. Publishers. Inc. 49 Okolelova, A.I. and Bobina, E.A., Identification of the most effective regimen for the treatment of viral papillomatosis in dogs, Colloquium–Journal, 2020, nos. 2–2 (54), pp. 34–35. 50 Onishchuk, A.A., Masalova, O.V., Lesnova, E.I., et al., Comparative analysis of the humoral and cellular immune response of mice to the nucleotide and amino acid sequences of the hepatitis C virus introduced with a new adjuvant based on polyprenyl phosphates, Aktual’naya Biotekhnol., 2017, no. 2 (21), pp. 235–237. 51 Ozherelkov, S.V., The role of natural immunomodulatory factors in the pathogenesis of experimental viral infections, Extended Abstract of Doct. Sci. (Biol.) Dissertation, Institute of Poliomyelitis and Viral Encephalitis im. M.P. Chumakova RAMS, 2003. 52 Ozherelkov S.V. Phosprenyl as an adjuvant for rabies vaccine inocula, Tr. Vses. Nauchno-Issled. Inst. Eksp. Vet. Ya.R. Kovalenko 2018 80 277 282 53 Ozherelkov, S.V. and Kozhevnikova, T.N., Study of adjuvant effect of fosprenil on immunogenicity of vaccines against parvovirus enteritis and rabies in experiment, Veterinariya Kubani, 2020, no. 1, pp. 33–35. 54 Ozherelkov S.V. Timofeev A.V. Novikova G.P. Protective effect of a new antiviral preparation of phosprenyl in experimental tick-borne encephalitis Vopr. Virusol. 2000 45 33 37 55 Ozherelkov S.V. Belousova R.V. Danilov L.L. The preparation Phosprenyl suppresses diarrhea and cattle infectious rhinotracheitis virus multiplication in sensitive cell cultures Vopr. Virusol. 2001 5 43 45 56 Ozherelkov, S.V., Sanin, A.V., Deeva, A.V., et al., Phosprenyl stimulates in vivo production of IL-4, IL-5, and IL-12 in BALB/c mice, J. Interferon Cytokine Res., 2002, vol. 22, no. S, p. 183. 57 Ozherelkov, S.V., Kalinina, E.S., and Kozhevnikova, T.N., Experimental study of the phenomenon of antibody dependent tick-borne encephalitis virus infectivity enhancement in vitro, Zh. Mikrobiol., Epidemiol. Immunobiol., 2008, no. 6, pp. 39–43. 58 Ozherelkov S.V. Deeva A.V. Sanin A.V. Polyprenyl phosphates as adjuvants, polarizing the immune response to Th1 Infektsiya i Immunitet 2012 2 645 650 59 Ozherelkov, S.V., Kozhevnikova, T.N., Narovlyansky, A.N., et al., Antiviral effect of Fosprenil and Gamapren against flaviviruses, Veterinariya i Kormlenie, 2017a, no. 3, pp. 78–80. 60 Ozherelkov, S.V., Kozhevnikova, T.N., Narovlyansky, A.N., et al., Fosprenil stimulates the immunogenicity of vaccines against rabies and tick-borne encephalitis, Veterinariya i Kormlenie, 2017b, no. 3, pp. 80–81. 61 Pereslegina, I.O. and Zhavnis, S.E., Achieving remission in the complex therapy of feline infectious peritonitis: a clinical case, Veterinariya, 2019, no. 6, pp. 48–52. 62 Pereslegina, I.O., Videnina, A.A., Narovlyansky, A.N., Novel approach to FIP therapy, Rossiskii Veterinarnyi Zhurnal. MDZh, 2013, no. 1, pp. 6–10. 63 Pereslegina, I.O., Dubrovina, T.S., Klintsova, T.Yu., et al., Comparison of two protocols of feline panleukopenia treatment, Rossiskii Veterinarnyi Zhurnal. MDZh, 2017, no. 5, pp. 24–28. 64 Petrakova, A.O., Review of the main principles of etiopathogenetic therapy of canine parvovirus enteritis, Alleya Nauki, 2018, vol. 2, no. 1 (17), pp. 388–393. 65 Polozyuk, O.N., and Sergeev, A.A., Canine plane and method for its treatment, in Agrarnaya nauka v usloviyakh stanovleniya tsifrovoi ekonomiki i proizvodstva ekologicheski chistoi produktsii v Rossiiskoi Federatsii. Mat. Mezhdunar. Nauch.-Prakt. Konf. Persianovskii (Agricultural Science in the Context of the Formation of the Digital Economy and the Production of Environmentally Friendly Products in the Russian Federation. Proc. Intl. Sci. Conf. Persianovsky Settlement), 2021, pp. 77–80. 66 Pronin, A.V., Immunomodulation and the vaccinoprophilaxis: experience of application phosprenyl, Rossiiskii Veterinarnyi Zhurnal. SKhZh, 2005, no. 1, pp. 42–44. 67 Pronin A.V. Ozherelkov S.V. Narovlyansky A.N. Role of cytokines in immunomodulatory effects of polyprenyl phosphate: new generation of antiviral drugs Rossiiskii Immunologicheskii Zhurnal 2000 5 155 164 68 Pronin A.V. Grigor’eva E.A. Sanin A.V. Polyprenols as possible factors that determine an instructive role of the innate immunity in the acquired immune response Rossiiskii Immunologicheskii Zhurnal 2002 7 135 142 69 Pronin, A.V., Narovlyansky, A.N., Deryabin, P.G., et al., Antiviral effect of Fosprenil in influenza infection, Veterinariya i Kormlenie, 2005, no. 6, p. 25. 70 Pronin, A.V., Narovlyansky, A.N., Deryabin, P.G., et al., Fosprenil and prevention of bird flu, Veterinariya Kubani, 2006, no. 2, pp. 27–28. 71 Pronin A.V. Ozherelkov S.V. Deeva A.V. Polyprenyl phosphates as adjuvants, polarizing the immune response to Th1 Med. Immunol. 2011 13 529 72 Pronin, A.V., Danilov, L.L., Narovlyansky, A.N., et al., Plant polyisoprenoids and control of cholesterol level, Arch. Immunol. Ther. Exp., vol. 2014, no. 62 (1), pp. 31–39. 73 Pronin, A.V., Narovlyansky, A.N., Shulzhenko, A.E., et al., New polyprenyl phosphate based preparation fortepren as promising cytokine regulating antiviral remedy, Cytokine Growth Factor Rev., 2016, pp. 119–126. 74 Pronin A.V. Narovlyansky A.N. Sanin A.V. New approaches to the prevention and treatment of viral diseases Arch. Immunol. Ther. Exp. (Warsz) 2021 69 10 10.1007/s00005-021-00613-w 33811524 75 Rostrosa, P.A., Sanin, A.V., Narovlyansky, A.N., et al., Increasing the natural resistance and survival of minks in the case of Aleutian disease, Rossiskii Veterinarnyi Zhurnal, 2019, no. 6, pp. 14–19. 76 Rudneva, S.Yu., Narovlyansky, A.N., Pronin, A.V., et al., Treatment of papillomatosis of the oral cavity in a dog using Fosprenil, Rossiskii Veterinarnyi Zhurnal. MDZh, 2016, no. 3, pp. 9–11. 77 Sagami H. Swiezewska E. Shidoji Y. The history and recent advances in research of polyprenol and its derivatives Biosci., Biotechnol., Biochem. 2018 82 947 955 10.1080/09168451.2017.1411775 29297247 78 Sanin, A.V., Veselovskii, V.V., Danilov, L.L., et al., Increased mobilization of hematopoietic stem cells by phosphorylated polyisoprenoids, Rossiiskii Immunologicheskii Zhurnal, 2008, vol. 2 (11), nos. 2–3, p. 113. 79 Sanin, A.V., Savoiskaya, S.L., Vasil’ev, I.K., et al., Gamapren efficacy against respiratory viral infections in cats, Veterinariya Kubani, 2009, no. 6, pp. 29–30. 80 Sanin, A.V., Suslov, A.P., Tret’yakov, O.Yu., et al., Multidirectional effect of MIF and sodium polyprenyl phosphate on the course of experimental flavivirus infection in mice, Zh. Mikrobiol. Epidemiol. Immunol., 2011a, no. 5, pp. 56–61. 81 Sanin, A.V., Narovlyansky, A.N., and Pronin, A.V., Immunomodulators in agriculture—fad or necessity, Rossiiskii Veterinarnyi Zhurnal. SKhZh., 2011b, no. 1, pp. 37–40. 82 Sanin, A.V., Videnina, A.A., Narovlyansky, A.N., and Pronin A.V., On the use of immunomodulators in poultry farming, Ptitsa i Ptitseprodukty, 2011c, no. 6, pp. 34–36. 83 Sanin, A.V., Pronin, A.V., and Narovlyansky, A.N., et al., Experimental study of the antiviral efficacy of phosphorylated polyprenols against feline rhinotracheitis virus in vitro and in vivo, Veterinariya, 2015, no. 11, pp. 17–21. 84 Sanin, A.V., Narovlyansky, A.N., Pronin, A.V., et al., Study of phosprenyl antioxidant activity in various biological test systems, Rossiskii Veterinarnyi Zhurnal. MDZh, 2017a, no. 10, pp. 28–31. 85 Sanin A.V. Deryabin P.G. Narovlyansky A.N. Antiviral activity of Phosprenyl and Gamapren against infection caused by influenza a (H5N1) virus in cell culture Vopr. Virusol. 2017 62 168 173 10.18821/0507-4088-2017-62-4-168-173 29733166 86 Sanin, A.V., Narovlyansky, A.N., Pronin, A.V., et al., Experimental study of Gamapren antiviral activity against herpes simplex virus in vitro, Rossiskii Veterinarnyi Zhurnal. MDZh, 2018a, no. 3, pp. 28–33. 87 Sanin, A.V., Ozherelkov, S.V., and Narovlyansky, A.N., et al., Antiviral activity of Gamapren against the bovine diarrhea virus in the in vitro experiments, Rossiiskii Veterinarnyi Zhurnal. SKhZh, 2018b, no. 2, pp. 12–16. 88 Sanin, A.V., Annikov, V.V., Annikova L.V., et al., Clinical efficacy of Gamapren® in feline panleukopenia: a controlled study, Veterinariya i Kormlenie, 2018c, no. 5, pp. 45–48. 89 Sanin, A.V., Narovlyansky, A.N., Pronin A.V., et al., Clinical efficacy of Gamapren® in feline calicivirus infection, Veterinariya, 2018d, no. 5, pp. 25–31. 90 Sanin, A.V., Narovlyansky, A.N., Pronin, A.V., et al., Effect of Gamapren on the production of regulatory cytokines in mice following experimental flavivirus infection, Rossiskii Veterinarnyi Zhurnal. MDZh, 2018e, no. 4, pp. 18–24. 91 Sanin, A.V., Narovlyansky, A.N., Pronin, A.V., et al., Therapeutic and prophylactic efficacy of Gamapren in experimental infection of mice caused by herpes simplex virus type 1, Veterinariya Kubani, 2018f, no. 3, pp. 15–19. 92 Savoiskaya, S.L., Klitsunova, N.V., Narovlyansky, A.N., et al., Efficacy of the drug Gamapren in obtaining adenovirus and parainfluenza of dogs, Veterinarnyi Doctor, 2008, no. 11, pp. 15–16. 93 Savoiskaya, S.L. and Kozhevnikova, T.N., Phosprenyl efficacy in the treatment of canine parvoviral enteritis, Veterinariya i Kormlenie, 2019, no. 7, pp. 22–24. 94 Savoiskaya, S.L., Ogorodnikova, I.V., and Kozhevnikova, T.N., Control of chronic coronavirus infection in cats with the use of Phosprenyl and Gamavit, Veterinariya i Kormlenie, 2021, no. 3, pp. 46–48. 95 Schultz, R., Duration of immunity for canine and feline vaccines: a review, Vet. Microbiol., 2006, vol. 117, no. 1, pp. 75–79. 96 Sedov, A.M., Narovlyansky, A.N., Pronin, A.V., et al., Mechanism of antiviral action and evaluation of the efficacy of the new preparation Fortepren® in the complex therapy of chronic recurrent herpesvirus infection of genital localization, Zh. Mikrobiol., Epidemiol. Immunobiol., 2018, no. 5, pp. 81–87. 97 Seo, D.J., Jeon, S.B., Oh, H.L., et al., Comparison of the antiviral activity of flavonoids against murine norovirus and feline calicivirus, Food Control, 2016, vol. 60, no. 6, pp. 25–30. 98 Sergeev, A.N., Belkina, T.V., Glotov, A.G., et al., Antiviral activity of different drugs in vitro against viruses of bovine infectious rhinotracheitis and bovine diarrhea, Vopr. Virusol., 2004, no. 5, pp. 43–46. 99 Shulyak, B.F., Virusnye infektsii sobak (Canine Viral Infections), Moscow: Olita, 2004. https://fermer.ru/forum/veterinariya-krs/164503. 100 Simons K. Ehehalt R. Cholesterol, lipid rafts, and disease J. Clin. Invest. 2002 110 597 603 10.1172/JCI0216390 12208858 101 Stuart A.D. Brown T.K. Entry of feline calicivirus is dependent on clathrin-mediated endocytosis and acidification in endosomes J. Virol. 2006 80 7500 7509 10.1128/JVI.02452-05 16840330 102 Takano T. Katada Y. Moritoh S. Analysis of the mechanism of antibody-dependent enhancement of feline infectious peritonitis virus infection: aminopeptidase N is not important and a process of acidification of the endosome is necessary J. Gen. Virol. 2008 89 1025 1029 10.1099/vir.0.83558-0 18343845 103 Tani K. Ogushi F. Huang L. CD13/aminopeptidase N, a novel chemoattractant for T lymphocytes in pulmonary sarcoidosis Am. J. Respir. Crit. Care Med. 2000 161 1636 1642 10.1164/ajrccm.161.5.9902008 10806168 104 Tyurina, O. L., Deeva, A.V., Mekhdikhanov, G.G., et al., Increasing the safety and productivity of broilers using Phosprenil, Veterinariya, 2006, no. 12, pp. 13–14. 105 Usikova, T.I. and Rachikhina, S.S., Kennel cough syndrome in dogs, in Ekologiya Yuzhnoi Sibiri i sopredel’nykh territorii. Mat. XXIII Mezhdunar. nauch. shkoly-konf. studentov i molodykh uchenykh (Ecology of Southern Siberia and Adjacent Territories. Proc. XXIII Int. Sci. School-Conference of Students and Young Scientists), Anyushin, V.V., Ed., 2019, pp. 39–41. 106 Uzal F.A. Plattner B.L. Hostetter J.M. Jubb 2016 107 Vasil’ev A.N. Ozherelkov S.V. Kozlov V.V. Antiviral and immunomodulating activity of polyprenyl phosphates in viral infections Antibiot. Khimioter. 2008 53 3 8 108 King A.M.Q. Adams M.J. Carstens E.B. and Lefkowitz E.J. Virus Taxonomy: Classification and Nomenclature of Viruses. Ninth Report of the International Committee on Taxonomy of Viruses 2012 Amsterdam Elsevier, Academic Press 109 Zaitseva M.L. Deeva A.V. Andreeva M.V. Experience in the use of Fosprenil in the treatment of infectious diseases of horses, Mat. konf. po nauch.-prakt. rabote MGAVMiB im. 2006 110 Zhavnis, S.E., Pereslegina, I.O., and Sanina, A.A., Comprehensive treatment of coronavirus gastroenteritis in kittens: a clinical case, Izv. Orenburgskogo Gos. Agrar. Univ., 2019, no. 5 (79), pp. 205–207.

0

PMC9749632

NO-CC CODE

2022-12-15 23:23:20

no

Biol Bull Rev. 2022 Dec 14; 12(6):609-624

utf-8

oa_other

==== Front Biol Bull Rev Biology Bulletin Reviews 2079-0864 2079-0872 Pleiades Publishing Moscow 1224 10.1134/S2079086422060044 Article Neurotropism as a Mechanism of the Damaging Action of Coronavirus Gomazkov O. A. oleg-gomazkov@yandex.ru grid.418846.7 0000 0000 8607 342X Orekhovich Scientific Research Institute of Biomedical Chemistry, Moscow, Russia 14 12 2022 2022 12 6 667678 15 4 2022 18 4 2022 18 4 2022 © Pleiades Publishing, Ltd. 2022, ISSN 2079-0864, Biology Bulletin Reviews, 2022, Vol. 12, No. 6, pp. 667–678. © Pleiades Publishing, Ltd., 2022.Russian Text © The Author(s), 2022, published in Uspekhi Sovremennoi Biologii, 2022, Vol. 142, No. 4, pp. 404–416. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Abstract— Clinical evidence suggests that COVID-19 is accompanied by many symptoms of damage to the central and peripheral nervous system. This article outlines new aspects of pathogenesis that consider the principle of neurotropism as the leading cause of SARS-CoV-2 infection and central nervous system dysfunction. New data demonstrate additional mechanisms for coronavirus transfection. The description of some transmembrane proteins (neuropilin, etc.) serve as an additional argument for SARS-CoV-2 neurotropism, these molecules act as cofactors for virus transfection in the tissues of the lungs, brain, and other organs. The study of the damaging effect of SARS-CoV-2 at the level of an individual neuron is formulated as a task of neurotropism investigation. The use of the organoid methodology as a new approach in biomedical analysis for modeling the relationship between the host and the pathogen is described. Numerous data on the pathogenesis of COVID-19 indicate that astrocytes and microglia are targets of SARS-CoV-2. Neuroinflammation is considered as an inverse manifestation of neurotropism and a consequence of the neural and mental complications of pathogenesis. Keywords: COVID-19 neurotropism SARS-CoV-2 neurological complications organelles neuroglia issue-copyright-statement© Pleiades Publishing, Ltd. 2022 ==== Body pmcINTRODUCTION The outbreak of the COVID-19 pandemic led to large-scale studies of the pathogenesis of this disease, which is a tricky complex of concomitant negative processes and consequences. An analysis of the clinical experience shows that pathogenesis of the respiratory distress syndrome caused by the SARS-CoV-2 virus exhibits a huge range of manifestations. These include clinical disorders of whole systems, individual organs, tissues, and biochemical reactions. COVID-19 represents a disturbance of cellular and molecular processes that gives reasons to identify pathogenic links. Diffuse alveolar lung injury with pronounced microangiopathy in the form of bilateral pneumonia is a typical clinical manifestation of COVID-19. Systemic infection is accompanied by a rapid increase in circulating chemokines and interleukins in the blood, which cross the blood-brain barrier (BBB) to enter the brain. Clinical materials indicate a variety of symptoms related to immediate or long-term neurodegenerative and mental disorders. Data on the neuroinvasive potential of SARS-CoV-2 confirm damage to the structures of the brain and peripheral nervous system. A detailed understanding of the pathogenesis, and identification of cellular and biochemical targets of SARS-CoV-2 are important in order to elaborate a therapeutic anti-COVID strategy. This paper takes into account aspects of COVID-19 pathogenesis that allow us to analyze the cellular and biochemical mechanisms of viral invasion leading to various forms of neurodegenerative and mental complications. Neurotropism is considered a leading mechanism involved in the neurodestructive effect of SARS-CoV-2. Experimental data are a basis for interpretation of the mechanisms associated with cellular tropism of coronaviruses. In addition to the traditional consideration of ACE2 (angiotensin-converting enzyme 2) as the main transporter in coronavirus entry, we assess the involvement of other molecules (neuropilins and other proteins), which facilitate transfection and contribute to SARS-CoV-2 neurotropism. The virus entry into the brain tissue is associated with a processes wherein disturbance of the immune defense plays a leading role. Neuroinflammation with an altered phenotype of microglial cells and astrocytes results in damage to brain systems. Clinical studies indicate that astrocytes and microglia are targets of neurotropic viruses, including SARS-CoV-2. The neurotropism of SARS-CoV-2 frequently extends to the postcovid period, leaving a trace in the form of neurodegenerative manifestations that can even develop into mental disorders. COVID AND NEURAL DAMAGE Clinical experience has generated significant material indicating the negative effects of coronaviruses on brain function, manifestations of neurotropism and neuroinvasion, which cause neurodegenerative and mental disorders in COVID-19. The initial processes of pathogenesis, i.e., expression of cyto- and chemokines, endothelial dysfunction, neuroinflammation, hypercoagulation, and immunothrombosis, determine the severity of multiple organ failure (Correia et al., 2020; Tsivgoulis et al., 2020; Morgado et al., 2021). An analysis of clinical histories made it possible to identify brain cells as the second most important pathogenetic target in COVID-19, which needs early protection from neuroinvasion (Li, Z. et al., 2020). Comorbidities, primarily age-related, neurodegenerative, and mental conditions, are associated with higher COVID-19 susceptibility. Postcovid complications have become a separate area of study (Costas-Carrera et al., 2022). Neurological symptoms in the acute period of COVID-19 include disorders of cerebral circulation and systemic cerebral disorders. The development of ischemia affects small perforating vessels and disrupts blood supply to limbic areas of the brain (Sokolova and Fedin, 2020). Such disorders of cerebral circulation are usually characterized as complications of ischemic disease, arterial hypertension, or thrombogenic changes in diabetes (Gusev et al., 2020). Epidemiological data and postmortem brain examination suggest that viral infections, including SARS-CoV-2, may contribute to the exacerbation of Alzheimer’s disease. The results of the previous research showed that viral fragments of CoV strains were detected in brain samples together with pathogenic beta-amyloid deposits. The areas of the brain affected by the virus include the limbic system of the cortex and subcortical structures that are associated with memory and cognitive processes (Arbour et al., 2000). A growing number of cases provide evidence of mental manifestations associated with COVID-19. Clinical protocols document a variety of characteristic symptoms: post-traumatic stress disorder, depression, anxiety, obsessive-compulsive phenomena, first-episode psychosis, neurocognitive syndrome, and others (Fedin, 2021). Viral entry into brain tissue may lead to cerebral dysfunction with cognitive impairment, primarily in compromised or elderly patients. These disorders can be caused by impaired endothelial function and neuroinflammation (Steardo et al., 2020). Therefore, a large number of clinical reports and reviews describe a wide range of neurological symptoms that occur in COVID-19 patients. The key question remains: which clinical signs are due to the direct influence of the virus, and which are consequence of the associated pathological processes caused by the infection? This formulation of the issue makes it possible to separate neurotropism for a better understanding of the cellular and molecular components involved in COVID pathogenesis and for better identification of new pharmacologic targets. NEURTROPISM. BIOCHEMISTRY AND CELLULAR BIOLOGY OF THE SELECTIVE AGGRESSIVENESS OF SARS-CoV-2 In modern medical literature neurotropism is defined as an upregulation of biochemical mechanisms to facilitate entry, replication, and propagation of certain viral strains in nervous tissue. The presence of complementary chemical structures of the host cell and virion ligands is a determinant of virus host tropism. Comparison with other coronavirus strains indicates that neurotropism is a common feature of infection, which is expressed to the greatest extent in SARS-CoV-2 (Desforges et al., 2014; Hu et al., 2020). There are a number of reasons as to why SARS-CoV-2 neurotropism should be isolated among the mechanisms of COVID-19 clinical symptoms. Although new facts about the mechanisms of neuroinvasion have become clearer, it remains necessary to clarify whether SARS-CoV-2 is a truly neurotropic agent or if neurodestruction is a consequence of systemic cellular and biochemical processes elicited by COVID-19 (ElBini Dhouib, 2021). It is emphasized that the brain, being a super-complex and dynamic cellular-tissue system, may represent an attractive environment for SARS-CoV-2 replication. At the same time, cytokine storm and cellular inflammation lead to stochastic activation of components within the immune system that trigger vascular disorders in the endothelium, immunothrombosis, and injury to the parenchyma and neurons of the brain. Neurodegenerative mechanisms, i.e., phosphorylation of tau protein, synuclein aggregation, and accumulation of toxic proteins, were reported for many infected brain structures (Nath and Smith, 2021). The idea of neurotropism is reinforced by the information about the presence of ACE2 in some areas of the brain. The results of genetic analysis showed that the enzyme has a high concentration in the substantia nigra, the ventricles of the brain, and the middle temporal gyrus. According to cell-type distribution analysis, ACE2 expression was detected in excitatory and inhibitory neurons of the temporal gyrus and the cerebral cortex (Chen et al., 2021). It is assumed that the regional presence of ACE2 in the brain as an indispensable entry receptor for coronavirus is the primary argument in favor of a neuroinvasive mechanism. The Main Pathways of SARS-CoV-2 Entry into the Brain The primary task of researchers is to determine how the virus implements its neurotropic action after its entry into the brain structures? Current clinical experience indicates that SARS-CoV-2, interacting with the ACE2 protein, takes advantage of complex hematogenous transfection and/or direct neurogenic invasion into the brain. The so-called hematogenous route involves injury to the vascular endothelium and disruption of the protective function of the BBB. Model experiments revealed how the infected cells allow the infection to enter into brain areas (Buzhdygan et al., 2020). Previous studies with various strains of SARS-CoV showed that neurons located in the centers of the medulla oblongata may be affected (Netland et al., 2008). When applying this information to the current COVID-19 situation, it is assumed (Li, Y.C. et al., 2020) that negative outcomes are often associated with neuroinvasive dysfunction of the cardiorespiratory center in the brain. The death of endothelial cells disrupts microenvironment of the brain parenchyma and allows the virus to reach other brain regions (Alquisiras-Burgos et al., 2021). Pathoanatomical investigation revealed SARS-CoV-2 virus particles in microvascular endothelial cells in the frontal lobe of the brain (Paniz-Mondolf et al., 2020). The presence of ACE2 in the endothelium is associated with multiple organ failure due to disturbance of the vasculature (Baig et al., 2020). Other arguments support a trans-neuronal hypothesis: SARS-CoV-2 enters the brain via the olfactory, the neural taste, and trigeminal pathways, especially at the early stages of infection (Liu et al., 2021). Direct connectivity of the olfactory bulb to the amygdala, the entorhinal cortex, and the hippocampus represents a pathway of viral infection from olfactory mucosa to reach the neural network and cause injury (Aghagoli et al., 2021). Some authors associate loss of sense of smell, a typical symptom of COVID-19, with infection of the olfactory system by SARS-CoV-2. Introduction of infection may occur by means of axonal transport of the virus via the olfactory nerve and its entry to the olfactory cortical region (Brann et al., 2020). МRI studies revealed the presence of the virus in the epithelium of the nasal cavities and ciliated cells in patients at the early phase of the disease (Politi et al., 2020). SARS-CoV-2 RNA is detected in autopsy studies in the brain regions involved in recognizing environmental signals. Fragments of the viral RNA were found in the olfactory bulb, amygdala, entorhinal cortex, temporal and frontal neocortex. These areas of the brain are responsible for emotional and spatial memory responses and cognitive functions (Serrano et al., 2021). SARS-CoV-2 in Cerebrospinal Fluid The cerebrospinal fluid regulates trophic and metabolic processes between the blood and the brain due to cerebrospinal fluid circulation and turnover. SARS-CoV-2, similarly to other coronaviruses, can use cerebrospinal fluid for its invasion. SARS-CoV-2 antibodies were detected in the cerebrospinal fluid of patients, and this may serve as an indicator of infection. Viral fragments were also revealed in the cerebrospinal fluid in meningoencephalitis associated with COVID-19 (Moriguchi et al., 2020). Electron microscopy reveals traces of coronavirus in neurons and endothelial cells in brain autopsies (Baig et al., 2020). The detected products of SARS-CoV-2 include proteins S1 and S2, fragments of the viral envelope, and nucleoproteins (Benameur et al., 2020). Therefore, comparison of clinical and biochemical studies confirms the signs of SARS-CoV-2 neuroinvasion. (1) When the virus was detected in the cerebrospinal fluid, the patients hardly exhibited any respiratory symptoms, but manifested signs of encephalitis or demyelinating pathology, which apparently occurred due to direct entry of the virus into the cerebrospinal fluid. (2) Comparative analysis shows that the quantity of SARS-CoV-2 in cerebrospinal fluid is associated with the severity of neurological symptoms: the highest content was related to encephalitis and a lower quantity was observed in disorders of cerebral circulation, encephalopathy, and Guillain-Barre syndrome (Li, Y.C. et al., 2021). (3) The presence of SARS-CoV-2 antibodies in the cerebrospinal fluid in patients with an intact BBB may indicate direct invasion of the virus into the brain. Upon entry into the cerebrospinal fluid, coronavirus invades into the main areas, including the brain stem that contains the nuclei that control cardiorespiratory functions. The invasion of SARS-CoV-2 into the brain stem may be a cause of acute respiratory failure in COVID-19 (Dey et al., 2021). Neuropilin and Other Receptors for Viral Entry The data indicate that in addition to ACE2, COVID-19 pathogenesis involves additional routes that facilitate viral entry into the host cells. The participation of transmembrane proteins as independent docking sites with virus fragments served to explain the tissue tropism of SARS-CoV-2: selective ligands of the host cell molecules act as infection cofactors, including membrane-bound serine protease TMPRSS2, furin, cathepsin L, basigin, and neuropilin-1 (NRP-1) (Coutard et al., 2020; Daly et al., 2020). One of the new players in the pathogenesis of COVID-19 is the neuropilin protein (Sarabipour and Mac Gabhann, 2021). Neuropilins are a group of glycoproteins that have long been beyond the interest of biologists and pathophysiologists. Neuropilin was discovered in the developing brain and was identified as a neuropil, an axon-guidance receptor (Kawakami et al., 1996). The studies performed during the pre-COVID period provide a general picture of the processes involving the NRP-1 and NRP-2 neuropilins. Due to the variety of connections, neuropilins are involved in migration and invasion of various cells, membrane disorders, angiogenesis, etc. Neuropilins are connecting components involved in the control of many physiological processes (Plein et al., 2014; Kofler and Simons, 2016). NRP-1 is both the host factor of cellular entry of SARS-CoV-2 and a component of increased contagiousness (Perez-Miller et al., 2021). In the traditional interpretation, the binding domain of the coronavirus interacts exclusively with ACE2. According to new data, the S1 and the S2 domains cleaved by furin protease mediate the fusion of virus membrane with host neuropilins. Inhibition of this interaction reduced SARS-CoV-2 infectivity in cell culture (Daly et al., 2020; Kielian, 2020). Autopsy analysis of olfactory epithelial cells in COVID-19 patients showed that NRP-1 facilitates viral entry and enhances the pathogenic effects of coronavirus (Cantuti-Castelvetri et al., 2020). NRP-1 was investigated in individual structures of the human brain using a sequencing method. RNA expression was highest in the hippocampus when compared to the olfactory region, basal ganglia, thalamus, hypothalamus, and midbrain. This leads to the conclusion about a wide variety of possible neurological symptoms in COVID-19. An analysis of viral invasion routes in COVID-19 reveals a complex mechanism of neurological complications. The discovery of a new factor, NRP-1, provides new details related to the neurotropic mechanisms of SARS-CoV-2 (Davies et al., 2020). Therefore, while until recently SARS-CoV-2 entry into host cells was mainly attributed to ACE2 as a unique cofactor (Hopkins et al., 2021), new data present other mechanisms that enhance coronavirus transfection in COVID-19. The involvement of some transmembrane proteins is new evidence in favor of the tissue tropism of SARS-CoV-2, in which these endogenous compounds act as cofactors of virus transfection into the tissues of the lungs, brain, and other organs (Sarabipour and Mac Gabhann, 2021). However, it should be noted that, using the example of neuropilins, their presence in many tissues and association with pathogenetic processes of immunothrombosis and organ injury requires control and a selective strategy of targeted therapy (Gomazkov, 2022). ORGANOIDS—A NEW METHOD TO STUDY NEUROTROPISM Organoids as Subcellular Model Systems The study of the mechanisms involved in the damaging effect of SARS-CoV-2 at the neuron level is a useful task in the investigation of neurotropism. The evident difficulties of the in-life study of cellular and molecular processes encourage the use of new methodologies. A new approach using organoids has gained significant popularity in the study of cellular tropism and mechanisms of SARS-CoV-2-induced injury (Katsura et al., 2020; Ramani et al., 2020; Yang et al., 2020). Brain organoids are derived from embryonic stem cells or induced pluripotent stem cells and find various applications in regenerative medicine. These systems are three-dimensional (3D) cellular structures, since they reflect tissue cytoarchitecture similar to the developing elements of the brain (Renner et al., 2017). The initial experiments with a three-dimensional (tissue) system of organoids were devoted to the phenotypic analysis of the brain. This method is used in the study of the human cerebral cortex and some forms of neuronal disorders (Lancaster et al., 2013). These studies showed the significant value of organoids as experimental platforms, including for studies related to COVID-19 pathogenesis (Lv et al., 2021). The use of organoids provides an opportunity to identify tissue-specific forms of SARS-CoV-2 interaction with host cells (Tiwari et al., 2021). Modeling of SARS-CoV-2 Effects at the Level of an Individual Neuron The neurotropism of SARS-CoV-2, i.e., its selective invasiveness to various brain structures, which was revealed experimentally, is considered the cause of COVID-19 neuronal dysfunction. Experiments were conducted using a hPSC-based platform of subcellular microstructures of the cerebral cortex, hippocampus, hypothalamus and midbrain exposed to convalescent serum of a patient. Immune labeling of MAP2, PU.1, and GFAP proteins confirmed that the cells obtained in the model system were identical (Ramani et al., 2021). A range of studies revealed that the virus can cause a productive infection of cortical neurons and neural progenitors in three-dimensional systems (Mao and Jin, 2020; Zhang et al., 2020). Comparative studies of coronavirus strains demonstrated a specific ability of SARS-CoV-2 to replicate in various types of nerve cells (Chu et al., 2020). Incubation with SARS-CoV-2 led to the accumulation of viral particles, which was accompanied by an increase in the level of viral RNA (Bullen et al., 2020). Autopsies confirmed the presence of SARS-CoV-2 fragments in cortical neurons in COVID-19 patients (Song et al., 2021). The results of subsequent experiments revealed that choroid plexus organoids were destroyed under the action of SARS-CoV-2 (Jacob et al., 2020). Infection of choroid cells causes increased viral RNA replication, and triggers the immune response, neuroinflammation, and massive apoptosis (Pellegrini et al., 2020). Models of Organoids and Identification of Cofactors in Viral Transfection After identification of the key role of ACE2 in SARS-CoV-2 entry into the lung tissue, this enzyme received major attention. Indeed, SARS-CoV-2 uses ACE2 as a Trojan horse to invade target cells, and ACE2 acts as a high-affinity receptor for the virus. New cellular technologies offer more subtle arguments for the essential role of ACE2 in the tissue transfection by the virus. Experiments using human endothelial cell cultures derived from pulmonary, cardiac, brain, and kidney tissues showed that ACE2 knockout blocked the infecting capacity of SARS-CoV-2. In contrast, cultured endothelial cells transduced with recombinant ACE2 receptors were infected by SARS-CoV-2 and revealed high viral titers and endothelial cell lysis. SARS-CoV-2 infection of ACE2-expressing endothelial cells elicits the procoagulative and inflammatory responses observed in COVID-19 patients (Conde et al., 2020). Studies using 3D-models of human neurons have shown that although the neurons of human brain organoids express low levels of ACE2, SARS-CoV-2 preferably targets the human neurons. SARS-CoV-2 exposure to the organelles of the cerebral cortex is associated with tau protein hyperphosphorylation and neuronal death (Ramani et al., 2020). In addition, experiments with human brain organoids treated with antibodies to ACE2 showed only slightly decreased viral transfection, which indirectly indicates that additional factors may contribute to SARS-CoV-2 replication (Yang et al., 2020). These new facts suggest the participation of other cofactors of viral entry and may explain why SARS-CoV-2 has such a large range of targets in various human organs. The use of model platforms of lung and brain organoids confirmed that neuropilin-1 (NRP-1), cathepsin L1 (CTSL1), furin (PACE), and basigin (BSG = CD147) are involved (Tiwari et al., 2021) in SARS-CoV-2 infection (Tiwari et al., 2021). In summary, some conclusions illustrating the concepts of SARS-CoV-2 neurotropism are as follows. (1) The use of organoids as a new approach for biomedical analysis allows one to model the host and pathogen interaction. The range of human tissues susceptible to virus infection includes a significant variety of organs and cellular structures. (2) Experiments using the brain organoid platform confirmed the permissiveness of neurons and other cell types to SARS-CoV-2 as evidence of neurotropic capacity and neuronal replication (Fig. 1). These materials confirmed the results of autopsy and analytical in silico studies of SARS-CoV-2 in brain structures in COVID-19. Fig. 1. Modeling of SARS-CoV-2 infection using in vitro organoid technology (according to Trevisan et al., 2021, adapted). Using stem cell and organoid technology, the mechanisms involved in the action of SARS-CoV-2 coronavirus on the epithelium of the human lungs, central nervous system, gastrointestinal tract and cardiovascular system were investigated. The architecture and physiology of various tissues and organs can be reproduced in two-dimensional and three-dimensional cultures of differentiated human cells and organoids. These in vitro cellular systems are derived from induced pluripotent stem cells, primary cells, cell lines and ex vivo tissue biopsies. Combined cultivation with immune cells allows us to represent possible responses of the anti-inflammatory immune systems of host cells to the viral agent. The results of the studies demonstrated the potential of in vitro models for analyzing viral tropism and host cell response. (3) The destructive activity in organoids confirms capacity of the virus to directly attack brain cells. Organoid methods illustrate a limited tropic capacity of SARS-CoV-2 toward neurons and astrocytes of several brain regions, but the virus has a significantly higher preference to injure the choroid vascular plexus of the ventricles. (4) The results illustrate the concept of neurotropism as a mechanism of SARS-CoV-2 pathogenesis. As discussed earlier, neurotoxic effects are not caused only by the virus, but they are also caused by the induced cytoimmune toxicity, vascular inflammation, and thrombosis. Infection of organoids with SARS-CoV-2 reveals disrupted transcription regulation, impaired cellular function, and increased cell death. The use of organoid combinations offers a new level in the pathophysiological investigation of SARS-CoV-2 neuropathology. The results provide evidence of pronounced SARS-CoV-2 neurotropism and specify the brain regions and cells susceptible to viral aggression (Song et al., 2021). Brain Cell Organoids. Limitations and Prospects Despite the documented evidence, the universality of organoid platforms may be limited. Authors describing the results of evidence-based research propose a number of ideas that influence interpretation (Jacob et al., 2020; Ramani et al., 2021; Trevisan et al., 2021). Since artificial models are subject to limited contact with the cytosolic medium, one may doubt the structural and functional relevance of platforms derived from transformed stem cells. The direct effect of SARS-CoV-2 in culture only reflects the general scheme of contagiousness on the organoid material. The clinical picture attributed to the influence of a huge number of biochemical components encountered in the cytosolic medium, as well as the intensity and length of exposure to the virus are apparently far from the results observed in the models. Nevertheless, according to (Ramani et al., 2021), organoid platforms represent almost ideal model systems for studying the damaging potential of viruses that target neural cells. The use of organoid systems has shown that SARS-CoV-2 exhibits neurotropic capacity with molecular and pathochemical traces of infection. However, in COVID-19 modeling these data remain tip of the iceberg. In general, this methodology offers significant opportunities for experimental variations and demonstrates viral neurotropism and the key points in regulatory control. This approach serves as a laboratory supplement to a huge array of clinical work. The model cell screening system can also be used in the search for new targeted therapeutic agents. NEUROGLIA AS A FACTOR OF CORONAVIRUS NEUROTROPISM Microglia are the resident mononuclear phagocytes of the central nervous system with a characteristic cellular organization and specific gene expression. Microglial cells, as the main neuroimmune sentinels of the brain, constantly monitor changes in their environment, and recognize pathogens, toxins or cellular debris. In experiments using direct RNA sequencing to determine transcriptomes (Hickman et al., 2013), it was found that microglia synthesizes special sensory recognition proteins. At the same time, they elicit neuroprotective or neurotoxic responses as a form of reaction to foreign agents. Cytokine storm is a phenomenon characterized by amplified release of pro-inflammatory cytokines, which is the main factor in the development of acute respiratory distress syndrome and multiple organ failure in COVID-19. However, clinical data show that cytokine storm and subsequent cellular inflammation frequently lead to stochastic activation of components of the immune system, provoking damage to the parenchyma and neurons of the brain. As a result, cytokines may act as direct (due to selective neurotropism) or indirect (via disruption of BBB) factors of neurotoxicity, damage, and cellular death (Vargas et al., 2020; Aghagoli et al., 2021). This information encourages one to further consider SARS-CoV-2 as the cause of neuroinvasive processes in COVID-19, since the infection triggers reactivity of resident immune and glial cells that are intended to provide protection to neurons. Microglia as a Sensor of Virus Invasion It is generally believed that microglia act as the main sensor of viral infections in the central nervous system (Chen et al., 2019). Equipped with a variety of molecular sensors that activate intracellular signaling cascades, microglia promote the expression of antiviral cytokines (Dantzer, 2018). Entry of the virus into the nervous system and its replication lead to direct infection of resident immune cells with SARS-CoV-2, interfere with the innate defense mechanisms, impair regulation of cyto- and chemokines, dysregulate autoimmunity, and cause neuronal dysfunction (Awogbindin et al., 2021). A range of publications confirm that microglia use molecular recognition patterns, DAMPs/PAMPs (damage-associated molecular patterns/pathogen-associated molecular patterns) with upregulation of intracellular signaling pathways to trigger transcription and expression of protective cytokines (Furr and Marriott, 2012). Viral epitopes fuse with with the plasma membrane receptors in endosomes and in the cytoplasm of immune cells (Pichlmair and Reis e Sousa, 2007; Pedraza et al., 2010). Specialized membrane structures of monocytes and macrophages identified as Toll-like receptors recognize fragments of pathogens (Ronald and Beutler, 2010). Up-regulation of Toll-like receptors by the PAMPs and DAMPs patterns triggers a signaling cascade that culminates in the synthesis and release of proinflammatory cytokines (O’Neill et al., 2013). Clinical analysis of COVID-19 indicates that encephalopathy with multiple organ failure is associated with an elevated level of systemic inflammation markers (Cummings et al., 2020; Helms et al., 2020; Koralnik and Tyler, 2020). Neuroinflammation. Immune Cells as Mediators of Viral Aggression An important link in pathogenesis is the infiltration of the brain tissue with infected immune cells. Viruses can enter the brain with cells that act as transporters. Monocytes, neutrophils, and T cells enter the brain through the vasculature of the meninges and the vascular plexus, which represent a path for viral aggression (Bergmann et al., 2006; Engelhardt et al., 2017). The features of the immune response make it possible to formulate (Tavčar et al., 2021) a general viewpoint of the intricate picture of pathophysiological processes. Systemic inflammation, which develops after infection with SARS-CoV-2, is caused by hyperactivation of the innate immune system and the release of pro-inflammatory mediators. These related reactions serve as a clinical sign of COVID-19. Neuroinflammation includes extensive activation of glial cells, release of pro-inflammatory cytokines, antioxidants, free radicals, and neurotrophic factors. In accordance with the evidence supporting that glial cells have a duality in their phenotype, neurotoxic or neuroprotective properties, these processes depend on age, infectious stimuli, and pathophysiological condition, which is especially relevant in COVID-19 (Matias et al., 2019). Clinical materials indicate that the reactive phenotype induced by the SARS-CoV-2 virus plays a role in maintaining neuroinflammation as a leading cause of neurodegenerative and mental disorders (Merad and Martin, 2020; Murta et al., 2020; Valenza et al., 2021). Activation of microglia in the affected areas is observed in most of COVID-19 cases. The frequency of detecting coronavirus fragments is much higher in brain regions with microgliosis or lymphocytic infiltration, compared to regions exposed to hypoxic or vascular damage (Li, Y.C. et al., 2021). In summary, the sequence of neurotropic processes in viral invasion may be as follows. (1) Injury to endothelial cells and disruption of the protective function of the BBB triggers pro-inflammatory signals from the periphery into the brain parenchyma with upregulation of the inflammatory response in microglia and astrocytes. (2) Glial cells, astrocytes, and microglia play a key role in the pathogenesis of inflammatory and degenerative disorders and can be regarded (Vargas et al., 2020) as neurotropic targets of SARS-CoV-2 viruses (Fig. 2). Fig. 2. The relationships between the neurodestructive effect of SARS-CoV-2 and antiviral protection with the participation of immune mechanisms in microglia (according to Vargas et al., 2020, adapted). Microglial cells act as a mechanism of the immune response necessary to prevent viral influence and activate the systemic antiviral response. This event includes recruitment of peripheral monocytes/macrophages, an increased innate immunity response, elevated cytokine release, and activation of T cells, together, a complex of mechanisms controlling spread of the virus. In severe cases of COVID-19, excessive activation of microglial cells can contribute to negative effects by reactivating astrocytes or T-lymphocyte-mediated neurotoxicity, i.e., phenomena that contribute to synapse loss and neuron degeneration. (3) Astrocytes are the main agents in the neuroinflammation system: fusion of the virus with ACE2 upregulates microglia and triggers the release of proinflammatory cytokines. (4) These reactions cause transformation of the astrocyte phenotype into a reactive form and stimulate neuroinflammation along with neurodestructive processes in COVID-19. CONCLUSIONS The concept of neurotropism in terms of COVID-19 pathogenesis can be seen as a summary of the evidence in favor of a complex process. An analysis of recent data obtained in experiments and clinical data allowed us to find new therapeutic solutions. In conclusion, we present the main statements discussed in the article in order to advance arguments in favor of this viewpoint. (1) A range of experimental studies and clinical data on the neuroinvasive potential of SARS-CoV-2 confirm the facts of damage to the brain and peripheral nervous system. A better understanding of pathogenesis, identification of cellular and biochemical targets are important for a therapeutic strategy. This paper outlines the basic principles that allow us to assess new aspects of neurodegenerative and mental complications in the pathogenesis of COVID-19. Herein, the idea of neurotropism as the leading cause of SARS-CoV-2 infection, as well as subsequent dysfunction and damage to the patient’s central nervous system, deserves separate consideration. (2) The brain, being an extremely complex and dynamic cellular-tissue system, can form an attractive environment for SARS-CoV-2 replication. The results presented illustrate neurotropism as a complex mechanism of SARS-CoV-2 pathogenesis. Although new facts are constantly added constantly added to the description of neuroinvasion, it remains necessary to clarify whether SARS-CoV-2 is a direct neurotropic agent or if neurodestructive processes are a consequence of systemic cellular and biochemical changes. (3) Analysis of viral invasion pathways in COVID-19 reveals a staged mechanism of neurological complications. Until recently, the idea of virus entry into host cells was mainly associated with the role of ACE2 as a unique mediator; new data demonstrate additional mechanisms of SARS-CoV-2 transfection. The description of some transmembrane proteins (neuropilin, etc.) served as a new argument in the concept of neurotropism, when such molecules act as cofactors of viral entry into the tissues of the lungs, brain, and other organs. (4) The complexities of in-life study of cellular and molecular processes encourage the use of new methods. The use of organoid models as a new approach to biomedical analysis has provided opportunities for modeling host-pathogen interactions. The range of human tissues permissive to infection includes a variety of organs and cellular structures. The study of the mechanisms involved in the damaging effect of SARS-CoV-2 at the level of an individual neuron is formulated as an important task in the study of neurotropism. Organoid platforms represent a new level of research into COVID-19 neuropathology. (5) The entry of the infecting agent into the patient’s brain and the subsequent disturbance of the immune defense are manifestations of neurotropism. Clinical materials on COVID-19 indicate that astrocytes and microglia are targets of SARS-CoV-2 viruses. Neurodestructive processes in COVID-19 may be associated with manifestations of vascular inflammation, immunothrombosis, and cytoimmune toxicity. Therefore, microglia and neuroinflammation are considered factors of neurotropism, and the neural and mental complications of COVID illness. ACKNOWLEDGMENTS The authors thank Professor V.V. Poroikov, Corresponding Member of the Russian Academy of Sciences for long-term cooperation and assistance in the work. FUNDING This study was supported by the Program of Basic Research in the Russian Federation for the long-term period (2021–2030, no. 122030100170-5). COMPLIANCE WITH ETHICAL STANDARDS The authors declare that they have no conflict of interest. This article does not contain any studies involving animals or human participants performed by any of the authors. Translated by M. Novikova ==== Refs REFERENCES 1 Aghagoli G. Neurological involvement in COVID-19 and potential mechanisms Neurocrit. Care 2021 34 1062 1071 10.1007/s12028-020-01049-4 32661794 2 Alquisiras-Burgos I. Peralta-Arrieta I. Alonso-Palomares L.A. Neurological complications associated with the blood-brain barrier damage induced by the inflammatory response during SARS-CoV-2 infection Mol. Neurobiol. 2021 58 520 535 10.1007/s12035-020-02134-7 32978729 3 Arbour N. Day R. Newcombe J. Talbot P.J. Neuroinvasion by human respiratory coronaviruses J. Virol. 2000 74 8913 8921 10.1128/JVI.74.19.8913-8921.2000 10982334 4 Awogbindin I.O. Ben-Azu B. Olusola B.A. Microglial implications in SARS-CoV-2 infection and COVID-19: lessons from viral RNA neurotropism and possible relevance to Parkinson’s disease Front. Cell. Neurosci. 2021 15 670298 10.3389/fncel.2021.670298 34211370 5 Baig A.M. Khaleeq A. Ali U. Syeda H. Evidence of the COVID-19 virus targeting the CNS: tissue distribution, host–virus interaction, and proposed neurotropic mechanisms ACS Chem. Neurosci. 2020 11 995 998 10.1021/acschemneuro.0c00122 32167747 6 Benameur K. Agarwal A. Auld S.C. Encephalopathy and encephalitis associated with cerebrospinal fluid cytokine alterations and coronavirus disease, Atlanta, Georgia, USA, 2020 Emerging Infect. Dis. 2020 26 2016 2021 10.3201/eid2609.202122 7 Bergmann C.C. Lane T.E. Stohlman S.A. Coronavirus infection of the central nervous system: host–virus stand-off Nat. Rev. Microbiol. 2006 4 121 132 10.1038/nrmicro1343 16415928 8 Brann, D.H., Tsukahara, T., Weinreb, C., et al., Non-neuronal expression of SARS-CoV-2 entry genes in the olfactory system suggests mechanisms underlying COVID-19-associated anosmia, Sci. Adv., 2020, vol. 6, no. 31, p. eabc5801. 9 Bullen C.K. Hogberg H.T. Bahadirli-Talbott A. Infectability of human BrainSphere neurons suggests neurotropism of SARS-CoV-2 ALTEX 2020 37 665 671 32591839 10 Buzhdygan T.P. DeOre B.J. Baldwin-Leclair A. The SARS-CoV-2 spike protein alters barrier function in 2D static and 3D microfluidic in vitro models of the human blood–brain barrier Neurobiol. Dis. 2020 146 105131 10.1016/j.nbd.2020.105131 33053430 11 Cantuti-Castelvetri L. Ojha R. Pedro L.D. Neuropilin-1 facilitates SARS-CoV-2 cell entry and infectivity Science 2020 370 856 860 10.1126/science.abd2985 33082293 12 Chen Z. Zhong D. Li G. The role of microglia in viral encephalitis: a review J. Neuroinflammation 2019 16 76 10.1186/s12974-019-1443-2 30967139 13 Chen R. Wang K. Yu J. The spatial and cell-type distribution of SARS-CoV-2 receptor ACE2 in human and mouse brain Front. Neurol. 2021 11 573095 10.3389/fneur.2020.573095 33551947 14 Chu H. Chan J.F. Yuen T.-T. Comparative tropism, replication kinetics, and cell damage profiling of SARS-CoV-2 and SARS-CoV with implications for clinical manifestations, transmissibility, and laboratory studies of COVID-19: an observational study Lancet Microbe 2020 1 e14 e23 10.1016/S2666-5247(20)30004-5 32835326 15 Conde J.N. Schutt W.S. Gorbunova E.E. Mackow E.R. Recombinant ACE2 expression is required for SARS-CoV-2 to infect primary human endothelial cells and induce inflammatory and procoagulative responses mBio 2020 11 e03185 20 33310781 16 Correia A.O. Feitosa P.W.G. Moreira J.L.S. Neurological manifestations of COVID-19 and other coronaviruses Neurology, Psychiatry and Brain Research 2020 37 27 32 10.1016/j.npbr.2020.05.008 32834527 17 Costas-Carrera A. Sánchez-Rodríguez M.M. Cañizares S. Neuropsychological functioning in post-ICU patients after severe COVID-19 infection: the role of cognitive reserve 2022 18 Coutard B. Valle C. De Lamballerie X. The spike glycoprotein of the new coronavirus 2019-nCoV contains a furin-like cleavage site absent in CoV of the same clade Antiviral Res. 2020 176 104742 10.1016/j.antiviral.2020.104742 32057769 19 Cummings M.J. Baldwin M.R. Abrams D. Epidemiology, clinical course, and outcomes of critically ill adults with COVID-19 in New York City: a prospective cohort study Lancet 2020 395 1763 1770 10.1016/S0140-6736(20)31189-2 32442528 20 Daly J.L. Simonetti B. Klein K. Neuropilin-1 is a host factor for SARS-CoV-2 infection Science 2020 370 861 865 10.1126/science.abd3072 33082294 21 Dantzer R. Neuroimmune interactions: from the brain to the immune system and vice versa Physiol. Rev. 2018 98 477 504 10.1152/physrev.00039.2016 29351513 22 Davies J. Randeva H.S. Chatha K. Neuropilin-1 as a new potential SARS-CoV-2 infection mediator implicated in the neurologic features and central nervous system involvement of COVID-19 Mol. Med. Rep. 2020 22 4221 4226 33000221 23 Desforges M. Le Coupanec A. Stodola J.K. Human coronaviruses: viral and cellular factors involved in neuroinvasiveness and neuropathogenesis Virus Res. 2014 194 145 158 10.1016/j.virusres.2014.09.011 25281913 24 Dey J. Alam M.T. Chandra S. Neuroinvasion of SARS-CoV-2 may play a role in the breakdown of the respiratory center of the brain J. Med. Virol. 2021 93 1296 1303 10.1002/jmv.26521 32964419 25 ElBini Dhouib I. I., Does coronaviruses induce neurodegenerative diseases? A systematic review on the neurotropism and neuroinvasion of SARS-CoV-2 Drug Discoveries Ther. 2021 14 262 272 10.5582/ddt.2020.03106 26 Engelhardt B. Vajkoczy P. Weller R.O. The movers and shapers in immune privilege of the CNS Nat. Immunol. 2017 18 123 131 10.1038/ni.3666 28092374 27 Fedin, A.I., Neurological complications of COVID-19, Novosti Nevrologii, 2021, no. 11 (85), pp. 1–10. 28 Furr S.R. Marriott I. Viral CNS infections: role of glial pattern recognition receptors in neuroinflammation Front. Microbiol. 2012 3 201 10.3389/fmicb.2012.00201 22723794 29 Gomazkov O.A. Neuropilin is a new player in the pathogenesis of COVID-19 Neurochem. J. 2022 16 130 135 10.1134/S1819712422020064 30 Gusev E.I. Martynov M.Yu. Boiko A.N. Novel coronavirus infection (COVID-19) and damage to the nervous system: mechanisms of neurological disorders, clinical manifestations, organization of neurological care Zh. Nevrol. Psikhiatr. im. S.S. Korsakova 2020 120 7 16 31 Helms J. Kremer S. Merdji H. Neurologic features in severe SARS-CoV-2 infection The New England Journal of Medicine 2020 382 2268 2270 10.1056/NEJMc2008597 32294339 32 Hickman S.E. Kingery N.D. Ohsumi T.K. The microglial sensome revealed by direct RNA sequencing Nat. Neurosci. 2013 16 1896 1905 10.1038/nn.3554 24162652 33 Hopkins C. Lechien J.R. Saussez S. More that ACE2? NRP1 may play a central role in the underlying pathophysiological mechanism of olfactory dysfunction in COVID-19 and its association with enhanced survival Med. Hypotheses 2021 146 110406 10.1016/j.mehy.2020.110406 33246692 34 Hu J. Jolkkonen J. Zhao C. Neurotropism of SARS-CoV-2 and its neuropathological alterations: Similarities with other coronaviruses Neurosci. Biobehav. Rev. 2020 119 184 193 10.1016/j.neubiorev.2020.10.012 33091416 35 Jacob F. Pather S.R. Huang W.K. Human pluripotent stem cell-derived neural cells and brain organoids reveal SARS-CoV-2 neurotropism predominates in choroid plexus epithelium Cell Stem Cell 2020 27 937 950. e9 10.1016/j.stem.2020.09.016 33010822 36 Katsura, H., Sontake, V., Tata, A., et al., Human lung stem cell–based alveolospheres provide insights into SARS-CoV-2-mediated interferon responses and pneumocyte dysfunction, Cell Stem Cell, 2020, vol. 27, pp. 890–904. e8. 37 Kawakami A. Kitsukawa T. Takagi S. Fujisawa H. Developmentally regulated expression of a cell surface protein, neuropilin, in the mouse nervous system J. Neurobiol. 1996 29 1 17 10.1002/(SICI)1097-4695(199601)29:1<1::AID-NEU1>3.0.CO;2-F 8748368 38 Kielian M. Enhancing host cell infection by SARS-CoV-2 Science 2020 370 765 766 10.1126/science.abf0732 33184193 39 Kofler N. Simons M. The expanding role of neuropilin: regulation of transforming growth factor-β and platelet-derived growth factor signaling in the vasculature Curr. Opin. Hematol. 2016 23 260 267 10.1097/MOH.0000000000000233 26849476 40 Koralnik I.J. Tyler K.L. COVID-19: a global threat to the nervous system Ann. Neurol. 2020 88 1 11 10.1002/ana.25807 32506549 41 Lancaster M.A. Renner M. Martinet C.-A. Cerebral organoids model human brain development and microcephaly Nature 2013 501 373 379 10.1038/nature12517 23995685 42 Li Y.C. Bai W.Z. Hashikawa T. The neuroinvasive potential of SARS-CoV2 may play a role in the respiratory failure of COVID-19 patients J. Med. Virol. 2020 92 552 555 10.1002/jmv.25728 32104915 43 Li Y.C. Zhang Y. Tan B.-H. What can cerebrospinal fluid testing and brain autopsies tell us about viral neuroinvasion of SARS-CoV-2 J. Med. Virol. 2021 93 4247 4257 10.1002/jmv.26943 33724490 44 Li Z. Huang Y. Guo X. The brain, another potential target organ, needs early protection from SARS-CoV-2 neuroinvasion Sci. China: Life Sci. 2020 63 771 773 10.1007/s11427-020-1690-y 32246403 45 Liu J.-M. Tan B.-H. Wu S. Evidence of central nervous system infection and neuroinvasive routes, as well as neurological involvement, in the lethality of SARS-CoV-2 infection J. Med. Virol. 2021 93 1304 1313 10.1002/jmv.26570 33002209 46 Lv T. Meng F. Yu M. Defense of COVID-19 by human organoids Phenomics 2021 1 113 128 10.1007/s43657-021-00015-0 35233559 47 Mao X.-Y. Jin W.-L. iPSCs-derived platform: a feasible tool for probing the neurotropism of SARS-CoV-2 ACS Chem. Neurosci. 2020 11 2489 2491 10.1021/acschemneuro.0c00512 32840109 48 Matias I. Morgado J. Gomes F.C.A. Astrocyte heterogeneity: impact to brain aging and disease Front. Aging Neurosci. 2019 11 59 10.3389/fnagi.2019.00059 30941031 49 Merad M. Martin J.C. Pathological inflammation in patients with COVID-19: a key role for monocytes and macrophages Nat. Rev. Immunol. 2020 20 355 362 10.1038/s41577-020-0331-4 32376901 50 Morgado A.M. Bezerra A.L.M.S. Felix E.B.G. COVID-19’s clinical-pathological evidence in relation to its repercussion on the central and peripheral nervous system Adv. Exp. Med. Biol. 2021 1353 197 215 10.1007/978-3-030-85113-2_11 35137375 51 Moriguchi T. Harii N. Goto J. A first case of meningitis/encephalitis associated with SARS-coronavirus-2 Int. J. Infect. Dis. 2020 94 55 58 10.1016/j.ijid.2020.03.062 32251791 52 Murta V. Villarreal A. Ramos A.J. Severe acute respiratory syndrome coronavirus 2 impact on the central nervous system: are astrocytes and microglia main players or merely bystanders? ASN Neuro 2020 12 1759091420954960 10.1177/1759091420954960 32878468 53 Nath A. Smith B. Neurological issues during COVID-19: an overview Neurosci. Lett. 2021 742 135533 10.1016/j.neulet.2020.135533 33248163 54 O’Neill L.A. Golenbock D. Bowie A.G. The history of Toll-like receptors—redefining innate immunity Nat. Rev. Immunol. 2013 13 453 460 10.1038/nri3446 23681101 55 Netland J. Meyerholz D.K. Moore S. Severe acute respiratory syndrome coronavirus infection causes neuronal death in the absence of encephalitis in mice transgenic for human ACE2 J. Virol. 2008 82 7264 7275 10.1128/JVI.00737-08 18495771 56 Paniz-Mondolfi A. Bryce C. Grimes Z. Central nervous system involvement by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) J. Med. Virol. 2020 92 699 702 10.1002/jmv.25915 32314810 57 Pedraza S.T. Guillermo B. Urcuqui-Inchima S. Viral recognition by the innate immune system: the role of pattern recognition receptors Colomb. Med. 2010 41 377 387 10.25100/cm.v41i4.731 58 Pellegrini L. Albecka A. Mallery D.L. SARS-CoV-2 infects the brain choroid plexus and disrupts the blood-CSF barrier in human brain organoids Cell Stem Cell 2020 27 951 961. e5 10.1016/j.stem.2020.10.001 33113348 59 Perez-Miller S. Patek M. Moutal A. Novel compounds targeting neuropilin receptor 1 with potential to interfere with SARS-CoV-2 virus entry ACS Chem. Neurosci. 2021 12 1299 1312 10.1021/acschemneuro.0c00619 33787218 60 Pichlmair A. Reis e Sousa C. Innate recognition of viruses Immunity 2007 27 370 383 10.1016/j.immuni.2007.08.012 17892846 61 Plein A. Fantin A. Ruhrberg C. Neuropilin regulation of angiogenesis, arteriogenesis, and vascular permeability Microcirculation 2014 21 315 323 10.1111/micc.12124 24521511 62 Politi L.S. Salsano E. Grimaldi M. Magnetic resonance imaging alteration of the brain in a patient with coronavirus disease 2019 (COVID-19) and anosmia JAMA Neurology 2020 77 1028 1029 10.1001/jamaneurol.2020.2125 32469400 63 Ramani A. Müller L. Ostermann P.N. SARS-COV-2 targets neurons of 3D human brain organoids EMBO J. 2020 39 e106230 10.15252/embj.2020106230 32876341 64 Ramani A. Pranty A.-I. Gopalakrishnan J. Neurotropic effects of SARS-CoV-2 modeled by the human brain organoids Stem Cell Rep. 2021 16 373 384 10.1016/j.stemcr.2021.02.007 65 Renner M. Lancaster M.A. Bian S. Self-organized developmental patterning and differentiation in cerebral organoids EMBO J. 2017 36 1316 1329 10.15252/embj.201694700 28283582 66 Ronald P.C. Beutler B. Plant and animal sensors of conserved microbial signatures Science 2010 330 1061 1064 10.1126/science.1189468 21097929 67 Sarabipour S. Mac Gabhann F. Targeting neuropilins as a viable SARS-CoV-2 treatment FEBS J. 2021 288 5122 5129 10.1111/febs.16096 34185437 68 Serrano, G.E., Walker, J.E., and Arce, R., Mapping of SARS-CoV-2 brain invasion and histopathology in COVID-19, medRxiv, 2021. 10.1101/2021.02.15.21251511 69 Sokolova, L.P. and Fedin, A.I., Cognitive and autonomic disorders in a new coronavirus disease, Novosti Nevrologii, 2020, no. 11 (73), pp. 3–12. 70 Song E. Zhang C. Israelow B. Neuroinvasion of SARS-CoV-2 in human and mouse brain J. Exp. Med. 2021 218 e20202135 10.1084/jem.20202135 33433624 71 Steardo L. Steardo L. Zorec R. Verkhratsky A. Neuroinfection may contribute to pathophysiology and clinical manifestations of COVID-19 Acta Physiol. (Oxf.) 2020 229 e13473 10.1111/apha.13473 32223077 72 Tavčar P. Potokar M. Kolenc M. Neurotropic viruses, astrocytes, and COVID-19 Front. Cell. Neurosci. 2021 15 662578 10.3389/fncel.2021.662578 33897376 73 Tiwari S.K. Wang S. Smith D. Revealing tissue-specific SARS-CoV-2 infection and host responses using human stem cell-derived lung and cerebral organoids Stem Cell Rep. 2021 16 437 445 10.1016/j.stemcr.2021.02.005 74 Trevisan M. Riccetti S. Sinigaglia A. Barzon L. SARS-CoV-2 infection and disease modelling using stem cell technology and organoids Int. J. Mol. Sci. 2021 22 2356 10.3390/ijms22052356 33652988 75 Tsivgoulis G. Palaiodimou L. Katsanos A.H. Neurological manifestations and implications of COVID-19 pandemic Ther. Adv. Neurol. Disord. 2020 13 1756286420932036 10.1177/1756286420932036 32565914 76 Valenza M. Steardo L. Steardo L. Systemic inflammation and astrocyte reactivity in the neuropsychiatric sequelae of COVID-19 Front. Cell. Neurosci. 2021 15 748136 10.3389/fncel.2021.748136 34912192 77 Vargas G. Geraldo L.H.M. Salomão N.G. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and glial cells: insights and perspectives Brain Behav. Immun.: Health 2020 7 100127 32838339 78 Yang L. Han Y. Nilsson-Payant B.E. A human pluripotent stem cell-based platform to study SARS-CoV-2 tropism and model virus infection in human cells and organoids Cell Stem Cell 2020 27 125 136. e7 10.1016/j.stem.2020.06.015 32579880 79 Zhang B.Z. Chu H. Han S. SARS-CoV-2 infects human neural progenitor cells and brain organoids Cell Res. 2020 30 928 931 10.1038/s41422-020-0390-x 32753756

0

PMC9749633

NO-CC CODE

2022-12-15 23:23:20

no

Biol Bull Rev. 2022 Dec 14; 12(6):667-678

utf-8

oa_other

==== Front Biol Bull Rev Biology Bulletin Reviews 2079-0864 2079-0872 Pleiades Publishing Moscow 1224 10.1134/S2079086422060044 Article Neurotropism as a Mechanism of the Damaging Action of Coronavirus Gomazkov O. A. oleg-gomazkov@yandex.ru grid.418846.7 0000 0000 8607 342X Orekhovich Scientific Research Institute of Biomedical Chemistry, Moscow, Russia 14 12 2022 2022 12 6 667678 15 4 2022 18 4 2022 18 4 2022 © Pleiades Publishing, Ltd. 2022, ISSN 2079-0864, Biology Bulletin Reviews, 2022, Vol. 12, No. 6, pp. 667–678. © Pleiades Publishing, Ltd., 2022.Russian Text © The Author(s), 2022, published in Uspekhi Sovremennoi Biologii, 2022, Vol. 142, No. 4, pp. 404–416. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Abstract— Clinical evidence suggests that COVID-19 is accompanied by many symptoms of damage to the central and peripheral nervous system. This article outlines new aspects of pathogenesis that consider the principle of neurotropism as the leading cause of SARS-CoV-2 infection and central nervous system dysfunction. New data demonstrate additional mechanisms for coronavirus transfection. The description of some transmembrane proteins (neuropilin, etc.) serve as an additional argument for SARS-CoV-2 neurotropism, these molecules act as cofactors for virus transfection in the tissues of the lungs, brain, and other organs. The study of the damaging effect of SARS-CoV-2 at the level of an individual neuron is formulated as a task of neurotropism investigation. The use of the organoid methodology as a new approach in biomedical analysis for modeling the relationship between the host and the pathogen is described. Numerous data on the pathogenesis of COVID-19 indicate that astrocytes and microglia are targets of SARS-CoV-2. Neuroinflammation is considered as an inverse manifestation of neurotropism and a consequence of the neural and mental complications of pathogenesis. Keywords: COVID-19 neurotropism SARS-CoV-2 neurological complications organelles neuroglia issue-copyright-statement© Pleiades Publishing, Ltd. 2022 ==== Body pmcINTRODUCTION The outbreak of the COVID-19 pandemic led to large-scale studies of the pathogenesis of this disease, which is a tricky complex of concomitant negative processes and consequences. An analysis of the clinical experience shows that pathogenesis of the respiratory distress syndrome caused by the SARS-CoV-2 virus exhibits a huge range of manifestations. These include clinical disorders of whole systems, individual organs, tissues, and biochemical reactions. COVID-19 represents a disturbance of cellular and molecular processes that gives reasons to identify pathogenic links. Diffuse alveolar lung injury with pronounced microangiopathy in the form of bilateral pneumonia is a typical clinical manifestation of COVID-19. Systemic infection is accompanied by a rapid increase in circulating chemokines and interleukins in the blood, which cross the blood-brain barrier (BBB) to enter the brain. Clinical materials indicate a variety of symptoms related to immediate or long-term neurodegenerative and mental disorders. Data on the neuroinvasive potential of SARS-CoV-2 confirm damage to the structures of the brain and peripheral nervous system. A detailed understanding of the pathogenesis, and identification of cellular and biochemical targets of SARS-CoV-2 are important in order to elaborate a therapeutic anti-COVID strategy. This paper takes into account aspects of COVID-19 pathogenesis that allow us to analyze the cellular and biochemical mechanisms of viral invasion leading to various forms of neurodegenerative and mental complications. Neurotropism is considered a leading mechanism involved in the neurodestructive effect of SARS-CoV-2. Experimental data are a basis for interpretation of the mechanisms associated with cellular tropism of coronaviruses. In addition to the traditional consideration of ACE2 (angiotensin-converting enzyme 2) as the main transporter in coronavirus entry, we assess the involvement of other molecules (neuropilins and other proteins), which facilitate transfection and contribute to SARS-CoV-2 neurotropism. The virus entry into the brain tissue is associated with a processes wherein disturbance of the immune defense plays a leading role. Neuroinflammation with an altered phenotype of microglial cells and astrocytes results in damage to brain systems. Clinical studies indicate that astrocytes and microglia are targets of neurotropic viruses, including SARS-CoV-2. The neurotropism of SARS-CoV-2 frequently extends to the postcovid period, leaving a trace in the form of neurodegenerative manifestations that can even develop into mental disorders. COVID AND NEURAL DAMAGE Clinical experience has generated significant material indicating the negative effects of coronaviruses on brain function, manifestations of neurotropism and neuroinvasion, which cause neurodegenerative and mental disorders in COVID-19. The initial processes of pathogenesis, i.e., expression of cyto- and chemokines, endothelial dysfunction, neuroinflammation, hypercoagulation, and immunothrombosis, determine the severity of multiple organ failure (Correia et al., 2020; Tsivgoulis et al., 2020; Morgado et al., 2021). An analysis of clinical histories made it possible to identify brain cells as the second most important pathogenetic target in COVID-19, which needs early protection from neuroinvasion (Li, Z. et al., 2020). Comorbidities, primarily age-related, neurodegenerative, and mental conditions, are associated with higher COVID-19 susceptibility. Postcovid complications have become a separate area of study (Costas-Carrera et al., 2022). Neurological symptoms in the acute period of COVID-19 include disorders of cerebral circulation and systemic cerebral disorders. The development of ischemia affects small perforating vessels and disrupts blood supply to limbic areas of the brain (Sokolova and Fedin, 2020). Such disorders of cerebral circulation are usually characterized as complications of ischemic disease, arterial hypertension, or thrombogenic changes in diabetes (Gusev et al., 2020). Epidemiological data and postmortem brain examination suggest that viral infections, including SARS-CoV-2, may contribute to the exacerbation of Alzheimer’s disease. The results of the previous research showed that viral fragments of CoV strains were detected in brain samples together with pathogenic beta-amyloid deposits. The areas of the brain affected by the virus include the limbic system of the cortex and subcortical structures that are associated with memory and cognitive processes (Arbour et al., 2000). A growing number of cases provide evidence of mental manifestations associated with COVID-19. Clinical protocols document a variety of characteristic symptoms: post-traumatic stress disorder, depression, anxiety, obsessive-compulsive phenomena, first-episode psychosis, neurocognitive syndrome, and others (Fedin, 2021). Viral entry into brain tissue may lead to cerebral dysfunction with cognitive impairment, primarily in compromised or elderly patients. These disorders can be caused by impaired endothelial function and neuroinflammation (Steardo et al., 2020). Therefore, a large number of clinical reports and reviews describe a wide range of neurological symptoms that occur in COVID-19 patients. The key question remains: which clinical signs are due to the direct influence of the virus, and which are consequence of the associated pathological processes caused by the infection? This formulation of the issue makes it possible to separate neurotropism for a better understanding of the cellular and molecular components involved in COVID pathogenesis and for better identification of new pharmacologic targets. NEURTROPISM. BIOCHEMISTRY AND CELLULAR BIOLOGY OF THE SELECTIVE AGGRESSIVENESS OF SARS-CoV-2 In modern medical literature neurotropism is defined as an upregulation of biochemical mechanisms to facilitate entry, replication, and propagation of certain viral strains in nervous tissue. The presence of complementary chemical structures of the host cell and virion ligands is a determinant of virus host tropism. Comparison with other coronavirus strains indicates that neurotropism is a common feature of infection, which is expressed to the greatest extent in SARS-CoV-2 (Desforges et al., 2014; Hu et al., 2020). There are a number of reasons as to why SARS-CoV-2 neurotropism should be isolated among the mechanisms of COVID-19 clinical symptoms. Although new facts about the mechanisms of neuroinvasion have become clearer, it remains necessary to clarify whether SARS-CoV-2 is a truly neurotropic agent or if neurodestruction is a consequence of systemic cellular and biochemical processes elicited by COVID-19 (ElBini Dhouib, 2021). It is emphasized that the brain, being a super-complex and dynamic cellular-tissue system, may represent an attractive environment for SARS-CoV-2 replication. At the same time, cytokine storm and cellular inflammation lead to stochastic activation of components within the immune system that trigger vascular disorders in the endothelium, immunothrombosis, and injury to the parenchyma and neurons of the brain. Neurodegenerative mechanisms, i.e., phosphorylation of tau protein, synuclein aggregation, and accumulation of toxic proteins, were reported for many infected brain structures (Nath and Smith, 2021). The idea of neurotropism is reinforced by the information about the presence of ACE2 in some areas of the brain. The results of genetic analysis showed that the enzyme has a high concentration in the substantia nigra, the ventricles of the brain, and the middle temporal gyrus. According to cell-type distribution analysis, ACE2 expression was detected in excitatory and inhibitory neurons of the temporal gyrus and the cerebral cortex (Chen et al., 2021). It is assumed that the regional presence of ACE2 in the brain as an indispensable entry receptor for coronavirus is the primary argument in favor of a neuroinvasive mechanism. The Main Pathways of SARS-CoV-2 Entry into the Brain The primary task of researchers is to determine how the virus implements its neurotropic action after its entry into the brain structures? Current clinical experience indicates that SARS-CoV-2, interacting with the ACE2 protein, takes advantage of complex hematogenous transfection and/or direct neurogenic invasion into the brain. The so-called hematogenous route involves injury to the vascular endothelium and disruption of the protective function of the BBB. Model experiments revealed how the infected cells allow the infection to enter into brain areas (Buzhdygan et al., 2020). Previous studies with various strains of SARS-CoV showed that neurons located in the centers of the medulla oblongata may be affected (Netland et al., 2008). When applying this information to the current COVID-19 situation, it is assumed (Li, Y.C. et al., 2020) that negative outcomes are often associated with neuroinvasive dysfunction of the cardiorespiratory center in the brain. The death of endothelial cells disrupts microenvironment of the brain parenchyma and allows the virus to reach other brain regions (Alquisiras-Burgos et al., 2021). Pathoanatomical investigation revealed SARS-CoV-2 virus particles in microvascular endothelial cells in the frontal lobe of the brain (Paniz-Mondolf et al., 2020). The presence of ACE2 in the endothelium is associated with multiple organ failure due to disturbance of the vasculature (Baig et al., 2020). Other arguments support a trans-neuronal hypothesis: SARS-CoV-2 enters the brain via the olfactory, the neural taste, and trigeminal pathways, especially at the early stages of infection (Liu et al., 2021). Direct connectivity of the olfactory bulb to the amygdala, the entorhinal cortex, and the hippocampus represents a pathway of viral infection from olfactory mucosa to reach the neural network and cause injury (Aghagoli et al., 2021). Some authors associate loss of sense of smell, a typical symptom of COVID-19, with infection of the olfactory system by SARS-CoV-2. Introduction of infection may occur by means of axonal transport of the virus via the olfactory nerve and its entry to the olfactory cortical region (Brann et al., 2020). МRI studies revealed the presence of the virus in the epithelium of the nasal cavities and ciliated cells in patients at the early phase of the disease (Politi et al., 2020). SARS-CoV-2 RNA is detected in autopsy studies in the brain regions involved in recognizing environmental signals. Fragments of the viral RNA were found in the olfactory bulb, amygdala, entorhinal cortex, temporal and frontal neocortex. These areas of the brain are responsible for emotional and spatial memory responses and cognitive functions (Serrano et al., 2021). SARS-CoV-2 in Cerebrospinal Fluid The cerebrospinal fluid regulates trophic and metabolic processes between the blood and the brain due to cerebrospinal fluid circulation and turnover. SARS-CoV-2, similarly to other coronaviruses, can use cerebrospinal fluid for its invasion. SARS-CoV-2 antibodies were detected in the cerebrospinal fluid of patients, and this may serve as an indicator of infection. Viral fragments were also revealed in the cerebrospinal fluid in meningoencephalitis associated with COVID-19 (Moriguchi et al., 2020). Electron microscopy reveals traces of coronavirus in neurons and endothelial cells in brain autopsies (Baig et al., 2020). The detected products of SARS-CoV-2 include proteins S1 and S2, fragments of the viral envelope, and nucleoproteins (Benameur et al., 2020). Therefore, comparison of clinical and biochemical studies confirms the signs of SARS-CoV-2 neuroinvasion. (1) When the virus was detected in the cerebrospinal fluid, the patients hardly exhibited any respiratory symptoms, but manifested signs of encephalitis or demyelinating pathology, which apparently occurred due to direct entry of the virus into the cerebrospinal fluid. (2) Comparative analysis shows that the quantity of SARS-CoV-2 in cerebrospinal fluid is associated with the severity of neurological symptoms: the highest content was related to encephalitis and a lower quantity was observed in disorders of cerebral circulation, encephalopathy, and Guillain-Barre syndrome (Li, Y.C. et al., 2021). (3) The presence of SARS-CoV-2 antibodies in the cerebrospinal fluid in patients with an intact BBB may indicate direct invasion of the virus into the brain. Upon entry into the cerebrospinal fluid, coronavirus invades into the main areas, including the brain stem that contains the nuclei that control cardiorespiratory functions. The invasion of SARS-CoV-2 into the brain stem may be a cause of acute respiratory failure in COVID-19 (Dey et al., 2021). Neuropilin and Other Receptors for Viral Entry The data indicate that in addition to ACE2, COVID-19 pathogenesis involves additional routes that facilitate viral entry into the host cells. The participation of transmembrane proteins as independent docking sites with virus fragments served to explain the tissue tropism of SARS-CoV-2: selective ligands of the host cell molecules act as infection cofactors, including membrane-bound serine protease TMPRSS2, furin, cathepsin L, basigin, and neuropilin-1 (NRP-1) (Coutard et al., 2020; Daly et al., 2020). One of the new players in the pathogenesis of COVID-19 is the neuropilin protein (Sarabipour and Mac Gabhann, 2021). Neuropilins are a group of glycoproteins that have long been beyond the interest of biologists and pathophysiologists. Neuropilin was discovered in the developing brain and was identified as a neuropil, an axon-guidance receptor (Kawakami et al., 1996). The studies performed during the pre-COVID period provide a general picture of the processes involving the NRP-1 and NRP-2 neuropilins. Due to the variety of connections, neuropilins are involved in migration and invasion of various cells, membrane disorders, angiogenesis, etc. Neuropilins are connecting components involved in the control of many physiological processes (Plein et al., 2014; Kofler and Simons, 2016). NRP-1 is both the host factor of cellular entry of SARS-CoV-2 and a component of increased contagiousness (Perez-Miller et al., 2021). In the traditional interpretation, the binding domain of the coronavirus interacts exclusively with ACE2. According to new data, the S1 and the S2 domains cleaved by furin protease mediate the fusion of virus membrane with host neuropilins. Inhibition of this interaction reduced SARS-CoV-2 infectivity in cell culture (Daly et al., 2020; Kielian, 2020). Autopsy analysis of olfactory epithelial cells in COVID-19 patients showed that NRP-1 facilitates viral entry and enhances the pathogenic effects of coronavirus (Cantuti-Castelvetri et al., 2020). NRP-1 was investigated in individual structures of the human brain using a sequencing method. RNA expression was highest in the hippocampus when compared to the olfactory region, basal ganglia, thalamus, hypothalamus, and midbrain. This leads to the conclusion about a wide variety of possible neurological symptoms in COVID-19. An analysis of viral invasion routes in COVID-19 reveals a complex mechanism of neurological complications. The discovery of a new factor, NRP-1, provides new details related to the neurotropic mechanisms of SARS-CoV-2 (Davies et al., 2020). Therefore, while until recently SARS-CoV-2 entry into host cells was mainly attributed to ACE2 as a unique cofactor (Hopkins et al., 2021), new data present other mechanisms that enhance coronavirus transfection in COVID-19. The involvement of some transmembrane proteins is new evidence in favor of the tissue tropism of SARS-CoV-2, in which these endogenous compounds act as cofactors of virus transfection into the tissues of the lungs, brain, and other organs (Sarabipour and Mac Gabhann, 2021). However, it should be noted that, using the example of neuropilins, their presence in many tissues and association with pathogenetic processes of immunothrombosis and organ injury requires control and a selective strategy of targeted therapy (Gomazkov, 2022). ORGANOIDS—A NEW METHOD TO STUDY NEUROTROPISM Organoids as Subcellular Model Systems The study of the mechanisms involved in the damaging effect of SARS-CoV-2 at the neuron level is a useful task in the investigation of neurotropism. The evident difficulties of the in-life study of cellular and molecular processes encourage the use of new methodologies. A new approach using organoids has gained significant popularity in the study of cellular tropism and mechanisms of SARS-CoV-2-induced injury (Katsura et al., 2020; Ramani et al., 2020; Yang et al., 2020). Brain organoids are derived from embryonic stem cells or induced pluripotent stem cells and find various applications in regenerative medicine. These systems are three-dimensional (3D) cellular structures, since they reflect tissue cytoarchitecture similar to the developing elements of the brain (Renner et al., 2017). The initial experiments with a three-dimensional (tissue) system of organoids were devoted to the phenotypic analysis of the brain. This method is used in the study of the human cerebral cortex and some forms of neuronal disorders (Lancaster et al., 2013). These studies showed the significant value of organoids as experimental platforms, including for studies related to COVID-19 pathogenesis (Lv et al., 2021). The use of organoids provides an opportunity to identify tissue-specific forms of SARS-CoV-2 interaction with host cells (Tiwari et al., 2021). Modeling of SARS-CoV-2 Effects at the Level of an Individual Neuron The neurotropism of SARS-CoV-2, i.e., its selective invasiveness to various brain structures, which was revealed experimentally, is considered the cause of COVID-19 neuronal dysfunction. Experiments were conducted using a hPSC-based platform of subcellular microstructures of the cerebral cortex, hippocampus, hypothalamus and midbrain exposed to convalescent serum of a patient. Immune labeling of MAP2, PU.1, and GFAP proteins confirmed that the cells obtained in the model system were identical (Ramani et al., 2021). A range of studies revealed that the virus can cause a productive infection of cortical neurons and neural progenitors in three-dimensional systems (Mao and Jin, 2020; Zhang et al., 2020). Comparative studies of coronavirus strains demonstrated a specific ability of SARS-CoV-2 to replicate in various types of nerve cells (Chu et al., 2020). Incubation with SARS-CoV-2 led to the accumulation of viral particles, which was accompanied by an increase in the level of viral RNA (Bullen et al., 2020). Autopsies confirmed the presence of SARS-CoV-2 fragments in cortical neurons in COVID-19 patients (Song et al., 2021). The results of subsequent experiments revealed that choroid plexus organoids were destroyed under the action of SARS-CoV-2 (Jacob et al., 2020). Infection of choroid cells causes increased viral RNA replication, and triggers the immune response, neuroinflammation, and massive apoptosis (Pellegrini et al., 2020). Models of Organoids and Identification of Cofactors in Viral Transfection After identification of the key role of ACE2 in SARS-CoV-2 entry into the lung tissue, this enzyme received major attention. Indeed, SARS-CoV-2 uses ACE2 as a Trojan horse to invade target cells, and ACE2 acts as a high-affinity receptor for the virus. New cellular technologies offer more subtle arguments for the essential role of ACE2 in the tissue transfection by the virus. Experiments using human endothelial cell cultures derived from pulmonary, cardiac, brain, and kidney tissues showed that ACE2 knockout blocked the infecting capacity of SARS-CoV-2. In contrast, cultured endothelial cells transduced with recombinant ACE2 receptors were infected by SARS-CoV-2 and revealed high viral titers and endothelial cell lysis. SARS-CoV-2 infection of ACE2-expressing endothelial cells elicits the procoagulative and inflammatory responses observed in COVID-19 patients (Conde et al., 2020). Studies using 3D-models of human neurons have shown that although the neurons of human brain organoids express low levels of ACE2, SARS-CoV-2 preferably targets the human neurons. SARS-CoV-2 exposure to the organelles of the cerebral cortex is associated with tau protein hyperphosphorylation and neuronal death (Ramani et al., 2020). In addition, experiments with human brain organoids treated with antibodies to ACE2 showed only slightly decreased viral transfection, which indirectly indicates that additional factors may contribute to SARS-CoV-2 replication (Yang et al., 2020). These new facts suggest the participation of other cofactors of viral entry and may explain why SARS-CoV-2 has such a large range of targets in various human organs. The use of model platforms of lung and brain organoids confirmed that neuropilin-1 (NRP-1), cathepsin L1 (CTSL1), furin (PACE), and basigin (BSG = CD147) are involved (Tiwari et al., 2021) in SARS-CoV-2 infection (Tiwari et al., 2021). In summary, some conclusions illustrating the concepts of SARS-CoV-2 neurotropism are as follows. (1) The use of organoids as a new approach for biomedical analysis allows one to model the host and pathogen interaction. The range of human tissues susceptible to virus infection includes a significant variety of organs and cellular structures. (2) Experiments using the brain organoid platform confirmed the permissiveness of neurons and other cell types to SARS-CoV-2 as evidence of neurotropic capacity and neuronal replication (Fig. 1). These materials confirmed the results of autopsy and analytical in silico studies of SARS-CoV-2 in brain structures in COVID-19. Fig. 1. Modeling of SARS-CoV-2 infection using in vitro organoid technology (according to Trevisan et al., 2021, adapted). Using stem cell and organoid technology, the mechanisms involved in the action of SARS-CoV-2 coronavirus on the epithelium of the human lungs, central nervous system, gastrointestinal tract and cardiovascular system were investigated. The architecture and physiology of various tissues and organs can be reproduced in two-dimensional and three-dimensional cultures of differentiated human cells and organoids. These in vitro cellular systems are derived from induced pluripotent stem cells, primary cells, cell lines and ex vivo tissue biopsies. Combined cultivation with immune cells allows us to represent possible responses of the anti-inflammatory immune systems of host cells to the viral agent. The results of the studies demonstrated the potential of in vitro models for analyzing viral tropism and host cell response. (3) The destructive activity in organoids confirms capacity of the virus to directly attack brain cells. Organoid methods illustrate a limited tropic capacity of SARS-CoV-2 toward neurons and astrocytes of several brain regions, but the virus has a significantly higher preference to injure the choroid vascular plexus of the ventricles. (4) The results illustrate the concept of neurotropism as a mechanism of SARS-CoV-2 pathogenesis. As discussed earlier, neurotoxic effects are not caused only by the virus, but they are also caused by the induced cytoimmune toxicity, vascular inflammation, and thrombosis. Infection of organoids with SARS-CoV-2 reveals disrupted transcription regulation, impaired cellular function, and increased cell death. The use of organoid combinations offers a new level in the pathophysiological investigation of SARS-CoV-2 neuropathology. The results provide evidence of pronounced SARS-CoV-2 neurotropism and specify the brain regions and cells susceptible to viral aggression (Song et al., 2021). Brain Cell Organoids. Limitations and Prospects Despite the documented evidence, the universality of organoid platforms may be limited. Authors describing the results of evidence-based research propose a number of ideas that influence interpretation (Jacob et al., 2020; Ramani et al., 2021; Trevisan et al., 2021). Since artificial models are subject to limited contact with the cytosolic medium, one may doubt the structural and functional relevance of platforms derived from transformed stem cells. The direct effect of SARS-CoV-2 in culture only reflects the general scheme of contagiousness on the organoid material. The clinical picture attributed to the influence of a huge number of biochemical components encountered in the cytosolic medium, as well as the intensity and length of exposure to the virus are apparently far from the results observed in the models. Nevertheless, according to (Ramani et al., 2021), organoid platforms represent almost ideal model systems for studying the damaging potential of viruses that target neural cells. The use of organoid systems has shown that SARS-CoV-2 exhibits neurotropic capacity with molecular and pathochemical traces of infection. However, in COVID-19 modeling these data remain tip of the iceberg. In general, this methodology offers significant opportunities for experimental variations and demonstrates viral neurotropism and the key points in regulatory control. This approach serves as a laboratory supplement to a huge array of clinical work. The model cell screening system can also be used in the search for new targeted therapeutic agents. NEUROGLIA AS A FACTOR OF CORONAVIRUS NEUROTROPISM Microglia are the resident mononuclear phagocytes of the central nervous system with a characteristic cellular organization and specific gene expression. Microglial cells, as the main neuroimmune sentinels of the brain, constantly monitor changes in their environment, and recognize pathogens, toxins or cellular debris. In experiments using direct RNA sequencing to determine transcriptomes (Hickman et al., 2013), it was found that microglia synthesizes special sensory recognition proteins. At the same time, they elicit neuroprotective or neurotoxic responses as a form of reaction to foreign agents. Cytokine storm is a phenomenon characterized by amplified release of pro-inflammatory cytokines, which is the main factor in the development of acute respiratory distress syndrome and multiple organ failure in COVID-19. However, clinical data show that cytokine storm and subsequent cellular inflammation frequently lead to stochastic activation of components of the immune system, provoking damage to the parenchyma and neurons of the brain. As a result, cytokines may act as direct (due to selective neurotropism) or indirect (via disruption of BBB) factors of neurotoxicity, damage, and cellular death (Vargas et al., 2020; Aghagoli et al., 2021). This information encourages one to further consider SARS-CoV-2 as the cause of neuroinvasive processes in COVID-19, since the infection triggers reactivity of resident immune and glial cells that are intended to provide protection to neurons. Microglia as a Sensor of Virus Invasion It is generally believed that microglia act as the main sensor of viral infections in the central nervous system (Chen et al., 2019). Equipped with a variety of molecular sensors that activate intracellular signaling cascades, microglia promote the expression of antiviral cytokines (Dantzer, 2018). Entry of the virus into the nervous system and its replication lead to direct infection of resident immune cells with SARS-CoV-2, interfere with the innate defense mechanisms, impair regulation of cyto- and chemokines, dysregulate autoimmunity, and cause neuronal dysfunction (Awogbindin et al., 2021). A range of publications confirm that microglia use molecular recognition patterns, DAMPs/PAMPs (damage-associated molecular patterns/pathogen-associated molecular patterns) with upregulation of intracellular signaling pathways to trigger transcription and expression of protective cytokines (Furr and Marriott, 2012). Viral epitopes fuse with with the plasma membrane receptors in endosomes and in the cytoplasm of immune cells (Pichlmair and Reis e Sousa, 2007; Pedraza et al., 2010). Specialized membrane structures of monocytes and macrophages identified as Toll-like receptors recognize fragments of pathogens (Ronald and Beutler, 2010). Up-regulation of Toll-like receptors by the PAMPs and DAMPs patterns triggers a signaling cascade that culminates in the synthesis and release of proinflammatory cytokines (O’Neill et al., 2013). Clinical analysis of COVID-19 indicates that encephalopathy with multiple organ failure is associated with an elevated level of systemic inflammation markers (Cummings et al., 2020; Helms et al., 2020; Koralnik and Tyler, 2020). Neuroinflammation. Immune Cells as Mediators of Viral Aggression An important link in pathogenesis is the infiltration of the brain tissue with infected immune cells. Viruses can enter the brain with cells that act as transporters. Monocytes, neutrophils, and T cells enter the brain through the vasculature of the meninges and the vascular plexus, which represent a path for viral aggression (Bergmann et al., 2006; Engelhardt et al., 2017). The features of the immune response make it possible to formulate (Tavčar et al., 2021) a general viewpoint of the intricate picture of pathophysiological processes. Systemic inflammation, which develops after infection with SARS-CoV-2, is caused by hyperactivation of the innate immune system and the release of pro-inflammatory mediators. These related reactions serve as a clinical sign of COVID-19. Neuroinflammation includes extensive activation of glial cells, release of pro-inflammatory cytokines, antioxidants, free radicals, and neurotrophic factors. In accordance with the evidence supporting that glial cells have a duality in their phenotype, neurotoxic or neuroprotective properties, these processes depend on age, infectious stimuli, and pathophysiological condition, which is especially relevant in COVID-19 (Matias et al., 2019). Clinical materials indicate that the reactive phenotype induced by the SARS-CoV-2 virus plays a role in maintaining neuroinflammation as a leading cause of neurodegenerative and mental disorders (Merad and Martin, 2020; Murta et al., 2020; Valenza et al., 2021). Activation of microglia in the affected areas is observed in most of COVID-19 cases. The frequency of detecting coronavirus fragments is much higher in brain regions with microgliosis or lymphocytic infiltration, compared to regions exposed to hypoxic or vascular damage (Li, Y.C. et al., 2021). In summary, the sequence of neurotropic processes in viral invasion may be as follows. (1) Injury to endothelial cells and disruption of the protective function of the BBB triggers pro-inflammatory signals from the periphery into the brain parenchyma with upregulation of the inflammatory response in microglia and astrocytes. (2) Glial cells, astrocytes, and microglia play a key role in the pathogenesis of inflammatory and degenerative disorders and can be regarded (Vargas et al., 2020) as neurotropic targets of SARS-CoV-2 viruses (Fig. 2). Fig. 2. The relationships between the neurodestructive effect of SARS-CoV-2 and antiviral protection with the participation of immune mechanisms in microglia (according to Vargas et al., 2020, adapted). Microglial cells act as a mechanism of the immune response necessary to prevent viral influence and activate the systemic antiviral response. This event includes recruitment of peripheral monocytes/macrophages, an increased innate immunity response, elevated cytokine release, and activation of T cells, together, a complex of mechanisms controlling spread of the virus. In severe cases of COVID-19, excessive activation of microglial cells can contribute to negative effects by reactivating astrocytes or T-lymphocyte-mediated neurotoxicity, i.e., phenomena that contribute to synapse loss and neuron degeneration. (3) Astrocytes are the main agents in the neuroinflammation system: fusion of the virus with ACE2 upregulates microglia and triggers the release of proinflammatory cytokines. (4) These reactions cause transformation of the astrocyte phenotype into a reactive form and stimulate neuroinflammation along with neurodestructive processes in COVID-19. CONCLUSIONS The concept of neurotropism in terms of COVID-19 pathogenesis can be seen as a summary of the evidence in favor of a complex process. An analysis of recent data obtained in experiments and clinical data allowed us to find new therapeutic solutions. In conclusion, we present the main statements discussed in the article in order to advance arguments in favor of this viewpoint. (1) A range of experimental studies and clinical data on the neuroinvasive potential of SARS-CoV-2 confirm the facts of damage to the brain and peripheral nervous system. A better understanding of pathogenesis, identification of cellular and biochemical targets are important for a therapeutic strategy. This paper outlines the basic principles that allow us to assess new aspects of neurodegenerative and mental complications in the pathogenesis of COVID-19. Herein, the idea of neurotropism as the leading cause of SARS-CoV-2 infection, as well as subsequent dysfunction and damage to the patient’s central nervous system, deserves separate consideration. (2) The brain, being an extremely complex and dynamic cellular-tissue system, can form an attractive environment for SARS-CoV-2 replication. The results presented illustrate neurotropism as a complex mechanism of SARS-CoV-2 pathogenesis. Although new facts are constantly added constantly added to the description of neuroinvasion, it remains necessary to clarify whether SARS-CoV-2 is a direct neurotropic agent or if neurodestructive processes are a consequence of systemic cellular and biochemical changes. (3) Analysis of viral invasion pathways in COVID-19 reveals a staged mechanism of neurological complications. Until recently, the idea of virus entry into host cells was mainly associated with the role of ACE2 as a unique mediator; new data demonstrate additional mechanisms of SARS-CoV-2 transfection. The description of some transmembrane proteins (neuropilin, etc.) served as a new argument in the concept of neurotropism, when such molecules act as cofactors of viral entry into the tissues of the lungs, brain, and other organs. (4) The complexities of in-life study of cellular and molecular processes encourage the use of new methods. The use of organoid models as a new approach to biomedical analysis has provided opportunities for modeling host-pathogen interactions. The range of human tissues permissive to infection includes a variety of organs and cellular structures. The study of the mechanisms involved in the damaging effect of SARS-CoV-2 at the level of an individual neuron is formulated as an important task in the study of neurotropism. Organoid platforms represent a new level of research into COVID-19 neuropathology. (5) The entry of the infecting agent into the patient’s brain and the subsequent disturbance of the immune defense are manifestations of neurotropism. Clinical materials on COVID-19 indicate that astrocytes and microglia are targets of SARS-CoV-2 viruses. Neurodestructive processes in COVID-19 may be associated with manifestations of vascular inflammation, immunothrombosis, and cytoimmune toxicity. Therefore, microglia and neuroinflammation are considered factors of neurotropism, and the neural and mental complications of COVID illness. ACKNOWLEDGMENTS The authors thank Professor V.V. Poroikov, Corresponding Member of the Russian Academy of Sciences for long-term cooperation and assistance in the work. FUNDING This study was supported by the Program of Basic Research in the Russian Federation for the long-term period (2021–2030, no. 122030100170-5). COMPLIANCE WITH ETHICAL STANDARDS The authors declare that they have no conflict of interest. This article does not contain any studies involving animals or human participants performed by any of the authors. Translated by M. Novikova ==== Refs REFERENCES 1 Aghagoli G. Neurological involvement in COVID-19 and potential mechanisms Neurocrit. Care 2021 34 1062 1071 10.1007/s12028-020-01049-4 32661794 2 Alquisiras-Burgos I. Peralta-Arrieta I. Alonso-Palomares L.A. Neurological complications associated with the blood-brain barrier damage induced by the inflammatory response during SARS-CoV-2 infection Mol. Neurobiol. 2021 58 520 535 10.1007/s12035-020-02134-7 32978729 3 Arbour N. Day R. Newcombe J. Talbot P.J. Neuroinvasion by human respiratory coronaviruses J. Virol. 2000 74 8913 8921 10.1128/JVI.74.19.8913-8921.2000 10982334 4 Awogbindin I.O. Ben-Azu B. Olusola B.A. Microglial implications in SARS-CoV-2 infection and COVID-19: lessons from viral RNA neurotropism and possible relevance to Parkinson’s disease Front. Cell. Neurosci. 2021 15 670298 10.3389/fncel.2021.670298 34211370 5 Baig A.M. Khaleeq A. Ali U. Syeda H. Evidence of the COVID-19 virus targeting the CNS: tissue distribution, host–virus interaction, and proposed neurotropic mechanisms ACS Chem. Neurosci. 2020 11 995 998 10.1021/acschemneuro.0c00122 32167747 6 Benameur K. Agarwal A. Auld S.C. Encephalopathy and encephalitis associated with cerebrospinal fluid cytokine alterations and coronavirus disease, Atlanta, Georgia, USA, 2020 Emerging Infect. Dis. 2020 26 2016 2021 10.3201/eid2609.202122 7 Bergmann C.C. Lane T.E. Stohlman S.A. Coronavirus infection of the central nervous system: host–virus stand-off Nat. Rev. Microbiol. 2006 4 121 132 10.1038/nrmicro1343 16415928 8 Brann, D.H., Tsukahara, T., Weinreb, C., et al., Non-neuronal expression of SARS-CoV-2 entry genes in the olfactory system suggests mechanisms underlying COVID-19-associated anosmia, Sci. Adv., 2020, vol. 6, no. 31, p. eabc5801. 9 Bullen C.K. Hogberg H.T. Bahadirli-Talbott A. Infectability of human BrainSphere neurons suggests neurotropism of SARS-CoV-2 ALTEX 2020 37 665 671 32591839 10 Buzhdygan T.P. DeOre B.J. Baldwin-Leclair A. The SARS-CoV-2 spike protein alters barrier function in 2D static and 3D microfluidic in vitro models of the human blood–brain barrier Neurobiol. Dis. 2020 146 105131 10.1016/j.nbd.2020.105131 33053430 11 Cantuti-Castelvetri L. Ojha R. Pedro L.D. Neuropilin-1 facilitates SARS-CoV-2 cell entry and infectivity Science 2020 370 856 860 10.1126/science.abd2985 33082293 12 Chen Z. Zhong D. Li G. The role of microglia in viral encephalitis: a review J. Neuroinflammation 2019 16 76 10.1186/s12974-019-1443-2 30967139 13 Chen R. Wang K. Yu J. The spatial and cell-type distribution of SARS-CoV-2 receptor ACE2 in human and mouse brain Front. Neurol. 2021 11 573095 10.3389/fneur.2020.573095 33551947 14 Chu H. Chan J.F. Yuen T.-T. Comparative tropism, replication kinetics, and cell damage profiling of SARS-CoV-2 and SARS-CoV with implications for clinical manifestations, transmissibility, and laboratory studies of COVID-19: an observational study Lancet Microbe 2020 1 e14 e23 10.1016/S2666-5247(20)30004-5 32835326 15 Conde J.N. Schutt W.S. Gorbunova E.E. Mackow E.R. Recombinant ACE2 expression is required for SARS-CoV-2 to infect primary human endothelial cells and induce inflammatory and procoagulative responses mBio 2020 11 e03185 20 33310781 16 Correia A.O. Feitosa P.W.G. Moreira J.L.S. Neurological manifestations of COVID-19 and other coronaviruses Neurology, Psychiatry and Brain Research 2020 37 27 32 10.1016/j.npbr.2020.05.008 32834527 17 Costas-Carrera A. Sánchez-Rodríguez M.M. Cañizares S. Neuropsychological functioning in post-ICU patients after severe COVID-19 infection: the role of cognitive reserve 2022 18 Coutard B. Valle C. De Lamballerie X. The spike glycoprotein of the new coronavirus 2019-nCoV contains a furin-like cleavage site absent in CoV of the same clade Antiviral Res. 2020 176 104742 10.1016/j.antiviral.2020.104742 32057769 19 Cummings M.J. Baldwin M.R. Abrams D. Epidemiology, clinical course, and outcomes of critically ill adults with COVID-19 in New York City: a prospective cohort study Lancet 2020 395 1763 1770 10.1016/S0140-6736(20)31189-2 32442528 20 Daly J.L. Simonetti B. Klein K. Neuropilin-1 is a host factor for SARS-CoV-2 infection Science 2020 370 861 865 10.1126/science.abd3072 33082294 21 Dantzer R. Neuroimmune interactions: from the brain to the immune system and vice versa Physiol. Rev. 2018 98 477 504 10.1152/physrev.00039.2016 29351513 22 Davies J. Randeva H.S. Chatha K. Neuropilin-1 as a new potential SARS-CoV-2 infection mediator implicated in the neurologic features and central nervous system involvement of COVID-19 Mol. Med. Rep. 2020 22 4221 4226 33000221 23 Desforges M. Le Coupanec A. Stodola J.K. Human coronaviruses: viral and cellular factors involved in neuroinvasiveness and neuropathogenesis Virus Res. 2014 194 145 158 10.1016/j.virusres.2014.09.011 25281913 24 Dey J. Alam M.T. Chandra S. Neuroinvasion of SARS-CoV-2 may play a role in the breakdown of the respiratory center of the brain J. Med. Virol. 2021 93 1296 1303 10.1002/jmv.26521 32964419 25 ElBini Dhouib I. I., Does coronaviruses induce neurodegenerative diseases? A systematic review on the neurotropism and neuroinvasion of SARS-CoV-2 Drug Discoveries Ther. 2021 14 262 272 10.5582/ddt.2020.03106 26 Engelhardt B. Vajkoczy P. Weller R.O. The movers and shapers in immune privilege of the CNS Nat. Immunol. 2017 18 123 131 10.1038/ni.3666 28092374 27 Fedin, A.I., Neurological complications of COVID-19, Novosti Nevrologii, 2021, no. 11 (85), pp. 1–10. 28 Furr S.R. Marriott I. Viral CNS infections: role of glial pattern recognition receptors in neuroinflammation Front. Microbiol. 2012 3 201 10.3389/fmicb.2012.00201 22723794 29 Gomazkov O.A. Neuropilin is a new player in the pathogenesis of COVID-19 Neurochem. J. 2022 16 130 135 10.1134/S1819712422020064 30 Gusev E.I. Martynov M.Yu. Boiko A.N. Novel coronavirus infection (COVID-19) and damage to the nervous system: mechanisms of neurological disorders, clinical manifestations, organization of neurological care Zh. Nevrol. Psikhiatr. im. S.S. Korsakova 2020 120 7 16 31 Helms J. Kremer S. Merdji H. Neurologic features in severe SARS-CoV-2 infection The New England Journal of Medicine 2020 382 2268 2270 10.1056/NEJMc2008597 32294339 32 Hickman S.E. Kingery N.D. Ohsumi T.K. The microglial sensome revealed by direct RNA sequencing Nat. Neurosci. 2013 16 1896 1905 10.1038/nn.3554 24162652 33 Hopkins C. Lechien J.R. Saussez S. More that ACE2? NRP1 may play a central role in the underlying pathophysiological mechanism of olfactory dysfunction in COVID-19 and its association with enhanced survival Med. Hypotheses 2021 146 110406 10.1016/j.mehy.2020.110406 33246692 34 Hu J. Jolkkonen J. Zhao C. Neurotropism of SARS-CoV-2 and its neuropathological alterations: Similarities with other coronaviruses Neurosci. Biobehav. Rev. 2020 119 184 193 10.1016/j.neubiorev.2020.10.012 33091416 35 Jacob F. Pather S.R. Huang W.K. Human pluripotent stem cell-derived neural cells and brain organoids reveal SARS-CoV-2 neurotropism predominates in choroid plexus epithelium Cell Stem Cell 2020 27 937 950. e9 10.1016/j.stem.2020.09.016 33010822 36 Katsura, H., Sontake, V., Tata, A., et al., Human lung stem cell–based alveolospheres provide insights into SARS-CoV-2-mediated interferon responses and pneumocyte dysfunction, Cell Stem Cell, 2020, vol. 27, pp. 890–904. e8. 37 Kawakami A. Kitsukawa T. Takagi S. Fujisawa H. Developmentally regulated expression of a cell surface protein, neuropilin, in the mouse nervous system J. Neurobiol. 1996 29 1 17 10.1002/(SICI)1097-4695(199601)29:1<1::AID-NEU1>3.0.CO;2-F 8748368 38 Kielian M. Enhancing host cell infection by SARS-CoV-2 Science 2020 370 765 766 10.1126/science.abf0732 33184193 39 Kofler N. Simons M. The expanding role of neuropilin: regulation of transforming growth factor-β and platelet-derived growth factor signaling in the vasculature Curr. Opin. Hematol. 2016 23 260 267 10.1097/MOH.0000000000000233 26849476 40 Koralnik I.J. Tyler K.L. COVID-19: a global threat to the nervous system Ann. Neurol. 2020 88 1 11 10.1002/ana.25807 32506549 41 Lancaster M.A. Renner M. Martinet C.-A. Cerebral organoids model human brain development and microcephaly Nature 2013 501 373 379 10.1038/nature12517 23995685 42 Li Y.C. Bai W.Z. Hashikawa T. The neuroinvasive potential of SARS-CoV2 may play a role in the respiratory failure of COVID-19 patients J. Med. Virol. 2020 92 552 555 10.1002/jmv.25728 32104915 43 Li Y.C. Zhang Y. Tan B.-H. What can cerebrospinal fluid testing and brain autopsies tell us about viral neuroinvasion of SARS-CoV-2 J. Med. Virol. 2021 93 4247 4257 10.1002/jmv.26943 33724490 44 Li Z. Huang Y. Guo X. The brain, another potential target organ, needs early protection from SARS-CoV-2 neuroinvasion Sci. China: Life Sci. 2020 63 771 773 10.1007/s11427-020-1690-y 32246403 45 Liu J.-M. Tan B.-H. Wu S. Evidence of central nervous system infection and neuroinvasive routes, as well as neurological involvement, in the lethality of SARS-CoV-2 infection J. Med. Virol. 2021 93 1304 1313 10.1002/jmv.26570 33002209 46 Lv T. Meng F. Yu M. Defense of COVID-19 by human organoids Phenomics 2021 1 113 128 10.1007/s43657-021-00015-0 35233559 47 Mao X.-Y. Jin W.-L. iPSCs-derived platform: a feasible tool for probing the neurotropism of SARS-CoV-2 ACS Chem. Neurosci. 2020 11 2489 2491 10.1021/acschemneuro.0c00512 32840109 48 Matias I. Morgado J. Gomes F.C.A. Astrocyte heterogeneity: impact to brain aging and disease Front. Aging Neurosci. 2019 11 59 10.3389/fnagi.2019.00059 30941031 49 Merad M. Martin J.C. Pathological inflammation in patients with COVID-19: a key role for monocytes and macrophages Nat. Rev. Immunol. 2020 20 355 362 10.1038/s41577-020-0331-4 32376901 50 Morgado A.M. Bezerra A.L.M.S. Felix E.B.G. COVID-19’s clinical-pathological evidence in relation to its repercussion on the central and peripheral nervous system Adv. Exp. Med. Biol. 2021 1353 197 215 10.1007/978-3-030-85113-2_11 35137375 51 Moriguchi T. Harii N. Goto J. A first case of meningitis/encephalitis associated with SARS-coronavirus-2 Int. J. Infect. Dis. 2020 94 55 58 10.1016/j.ijid.2020.03.062 32251791 52 Murta V. Villarreal A. Ramos A.J. Severe acute respiratory syndrome coronavirus 2 impact on the central nervous system: are astrocytes and microglia main players or merely bystanders? ASN Neuro 2020 12 1759091420954960 10.1177/1759091420954960 32878468 53 Nath A. Smith B. Neurological issues during COVID-19: an overview Neurosci. Lett. 2021 742 135533 10.1016/j.neulet.2020.135533 33248163 54 O’Neill L.A. Golenbock D. Bowie A.G. The history of Toll-like receptors—redefining innate immunity Nat. Rev. Immunol. 2013 13 453 460 10.1038/nri3446 23681101 55 Netland J. Meyerholz D.K. Moore S. Severe acute respiratory syndrome coronavirus infection causes neuronal death in the absence of encephalitis in mice transgenic for human ACE2 J. Virol. 2008 82 7264 7275 10.1128/JVI.00737-08 18495771 56 Paniz-Mondolfi A. Bryce C. Grimes Z. Central nervous system involvement by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) J. Med. Virol. 2020 92 699 702 10.1002/jmv.25915 32314810 57 Pedraza S.T. Guillermo B. Urcuqui-Inchima S. Viral recognition by the innate immune system: the role of pattern recognition receptors Colomb. Med. 2010 41 377 387 10.25100/cm.v41i4.731 58 Pellegrini L. Albecka A. Mallery D.L. SARS-CoV-2 infects the brain choroid plexus and disrupts the blood-CSF barrier in human brain organoids Cell Stem Cell 2020 27 951 961. e5 10.1016/j.stem.2020.10.001 33113348 59 Perez-Miller S. Patek M. Moutal A. Novel compounds targeting neuropilin receptor 1 with potential to interfere with SARS-CoV-2 virus entry ACS Chem. Neurosci. 2021 12 1299 1312 10.1021/acschemneuro.0c00619 33787218 60 Pichlmair A. Reis e Sousa C. Innate recognition of viruses Immunity 2007 27 370 383 10.1016/j.immuni.2007.08.012 17892846 61 Plein A. Fantin A. Ruhrberg C. Neuropilin regulation of angiogenesis, arteriogenesis, and vascular permeability Microcirculation 2014 21 315 323 10.1111/micc.12124 24521511 62 Politi L.S. Salsano E. Grimaldi M. Magnetic resonance imaging alteration of the brain in a patient with coronavirus disease 2019 (COVID-19) and anosmia JAMA Neurology 2020 77 1028 1029 10.1001/jamaneurol.2020.2125 32469400 63 Ramani A. Müller L. Ostermann P.N. SARS-COV-2 targets neurons of 3D human brain organoids EMBO J. 2020 39 e106230 10.15252/embj.2020106230 32876341 64 Ramani A. Pranty A.-I. Gopalakrishnan J. Neurotropic effects of SARS-CoV-2 modeled by the human brain organoids Stem Cell Rep. 2021 16 373 384 10.1016/j.stemcr.2021.02.007 65 Renner M. Lancaster M.A. Bian S. Self-organized developmental patterning and differentiation in cerebral organoids EMBO J. 2017 36 1316 1329 10.15252/embj.201694700 28283582 66 Ronald P.C. Beutler B. Plant and animal sensors of conserved microbial signatures Science 2010 330 1061 1064 10.1126/science.1189468 21097929 67 Sarabipour S. Mac Gabhann F. Targeting neuropilins as a viable SARS-CoV-2 treatment FEBS J. 2021 288 5122 5129 10.1111/febs.16096 34185437 68 Serrano, G.E., Walker, J.E., and Arce, R., Mapping of SARS-CoV-2 brain invasion and histopathology in COVID-19, medRxiv, 2021. 10.1101/2021.02.15.21251511 69 Sokolova, L.P. and Fedin, A.I., Cognitive and autonomic disorders in a new coronavirus disease, Novosti Nevrologii, 2020, no. 11 (73), pp. 3–12. 70 Song E. Zhang C. Israelow B. Neuroinvasion of SARS-CoV-2 in human and mouse brain J. Exp. Med. 2021 218 e20202135 10.1084/jem.20202135 33433624 71 Steardo L. Steardo L. Zorec R. Verkhratsky A. Neuroinfection may contribute to pathophysiology and clinical manifestations of COVID-19 Acta Physiol. (Oxf.) 2020 229 e13473 10.1111/apha.13473 32223077 72 Tavčar P. Potokar M. Kolenc M. Neurotropic viruses, astrocytes, and COVID-19 Front. Cell. Neurosci. 2021 15 662578 10.3389/fncel.2021.662578 33897376 73 Tiwari S.K. Wang S. Smith D. Revealing tissue-specific SARS-CoV-2 infection and host responses using human stem cell-derived lung and cerebral organoids Stem Cell Rep. 2021 16 437 445 10.1016/j.stemcr.2021.02.005 74 Trevisan M. Riccetti S. Sinigaglia A. Barzon L. SARS-CoV-2 infection and disease modelling using stem cell technology and organoids Int. J. Mol. Sci. 2021 22 2356 10.3390/ijms22052356 33652988 75 Tsivgoulis G. Palaiodimou L. Katsanos A.H. Neurological manifestations and implications of COVID-19 pandemic Ther. Adv. Neurol. Disord. 2020 13 1756286420932036 10.1177/1756286420932036 32565914 76 Valenza M. Steardo L. Steardo L. Systemic inflammation and astrocyte reactivity in the neuropsychiatric sequelae of COVID-19 Front. Cell. Neurosci. 2021 15 748136 10.3389/fncel.2021.748136 34912192 77 Vargas G. Geraldo L.H.M. Salomão N.G. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and glial cells: insights and perspectives Brain Behav. Immun.: Health 2020 7 100127 32838339 78 Yang L. Han Y. Nilsson-Payant B.E. A human pluripotent stem cell-based platform to study SARS-CoV-2 tropism and model virus infection in human cells and organoids Cell Stem Cell 2020 27 125 136. e7 10.1016/j.stem.2020.06.015 32579880 79 Zhang B.Z. Chu H. Han S. SARS-CoV-2 infects human neural progenitor cells and brain organoids Cell Res. 2020 30 928 931 10.1038/s41422-020-0390-x 32753756

36515854

PMC9749636

NO-CC CODE

2022-12-15 23:23:20

no

Acta Neurol Belg. 2022 Dec 14;:1

latin-1

Acta Neurol Belg

10.1007/s13760-022-02149-y

oa_other

==== Front J Community Health J Community Health Journal of Community Health 0094-5145 1573-3610 Springer US New York 1160 10.1007/s10900-022-01160-x Original Paper Parental and Other Caregiver Loss Due to COVID-19 in the United States: Prevalence by Race, State, Relationship, and Child Age http://orcid.org/0000-0003-3612-9669 Treglia Dan dtreglia@upenn.edu 1 http://orcid.org/0000-0002-5185-2866 Cutuli J. J. 2 Arasteh Kamyar 2 Bridgeland John 3 1 grid.25879.31 0000 0004 1936 8972 University of Pennsylvania, 3814 Walnut St, Philadelphia, PA 19104 USA 2 Nemours Children’s Health, Wilmington, DE USA 3 COVID Collaborative, Washington, DC USA 14 12 2022 18 19 10 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The more than one million COVID-19 deaths in the United States include parents, grandparents, and other caregivers for children. These losses can disrupt the social, emotional, and economic well-being of children, their families, and their communities, and understanding the number and characteristics of affected children is a critical step in responding. We estimate the number of children who lost a parent or other co-residing caregiver to COVID-19 in the U.S. and identify racial, ethnic, and geographic disparities by aligning COVID-19 death counts through mid-May 2022 with household information from a representative sample of individuals. We estimate that 216,617 children lost a co-residing caregiver to COVID-19; 77,283 lost a parent and more than 17,000 children lost the only caregiver with whom they lived. Non-White children were more than twice as likely as White children to experience caregiver loss, and children under 14 years old experienced 70% of caregiver loss. These losses are a salient threat to the functioning of families and the communities in which COVID-19 deaths are concentrated, compounding additional challenges to physical and mental health and economic stability disproportionately imposed by the pandemic on historically disadvantaged populations. Policymakers and systems should take steps to ensure access to appropriate supports. Keywords COVID-19 Parental loss Caregiver loss Socioeconomic disparities Child well-being http://dx.doi.org/10.13039/100010536 Walton Family Foundation http://dx.doi.org/10.13039/100007028 Leona M. and Harry B. Helmsley Charitable Trust New York Life Foundationhttp://dx.doi.org/10.13039/100013592 Allstate Foundation Skoll Foundation ==== Body pmcIntroduction The deaths of more than one million Americans to COVID-19 through mid-May 2022 threaten the functioning of family and social networks [4]. Among these deaths are parents, custodial grandparents, or others on whom children and youth relied [14], [19], and their sudden loss has ramifications for the children, the entire household, and their community as a whole. The loss of a parent or other caregiver is likely to be a transformative moment in a child’s life [16]. Though most who experience parental loss manifest resilience, 5 to 10 percent demonstrate problems that include higher rates of mental health diagnoses like depression and anxiety [1], [18]; [30], higher rates of alcohol and substance abuse disorders[12], worse peer relationships [2], higher rates of suicide, and lower levels of educational attainment, adult employment, and other indicators of human capital [2], [10]; [20]; [22]. Though no research has documented the impacts of caregiver loss during the COVID-19 pandemic, the overall decline in youth mental health and disruptions in systems that would normally be available to bereaved youth likely increase the risk of negative outcomes among youth losing a parent since March 2020. The risk for poor outcomes for children who lost a parent or caregiver to COVID-19 is compounded by the concentration of COVID-19 deaths in disadvantaged communities. Non-White populations have higher COVID-19 death rates than their White counterparts, with disparities concentrated among adults under age 65 [4]. Low-income individuals are at highest risk of death from COVID-19 and low-income counties faced disproportionately large shares of COVID-19 deaths, even after controlling for other indicators of well-being in two studies focused on individual and community-level associations with COVID-19 mortality rates of all persons [11, 28]. In addition, Black and Hispanic workers in low-wage, often consumer-facing, positions, have notably high rates of COVID-19 mortality [6]. These populations also have relatively high rates of health conditions associated with COVID-19 mortality, including overweight, obesity, diabetes, hypertension, and other morbidities [26, 33]. Thus, estimating the number and needs of children who lost a caregiver due to COVID-19 is a high priority for practitioners and policymakers interested in promoting resilience among these acutely vulnerable children and youth. Prior efforts have sought to estimate parental or grandparent loss attributable to COVID-19. Kidman and colleagues estimated that 40,000 children lost a parent to COVID-19 through February, 2021, accounting for a 17–20% increase in parental loss relative to a period without COVID-19 [19]. A second team produced two reports. The first, as part of an assessment of global orphanhood attributable to COVID-19, estimated 137,000 children in the U.S. had lost at least one primary or secondary caregiver to COVID-19 through April 2021 [14, 15]. In a follow-up analysis, they estimated that 143,000 children had lost a biological parent or caregiving grandparent due to COVID-19 through June of 2021 [14]. While the latter studies acknowledge non-parental caregivers, both efforts rely heavily on biological parenthood, with limited consideration given to alternate family and household structures. Changes in the structure of American households necessitate that we consider caregivers beyond biological parents in understanding the demographics and needs of bereaved children. Twenty-six percent of children and adolescents under 18, disproportionately low-income and Black or Hispanic, live with only a single caregiver [24]. In addition, 24% of children between birth and four years old, live in a multigenerational household; these children are, similarly, disproportionately non-White [25]. Many others live with a grandparent serving as a primary caregiver, whose deaths can have similar consequences as parental loss for a child [23]. Finally, a substantial proportion of children—disproportionately low-income, low-education, and Black or Hispanic—live in “doubled up” arrangements with non-family members, who may assume some caregiving role [9]. Merging COVID-19 death data with a representative survey of U.S. households, we estimate the counts and rates of COVID-19-related loss of any co-residing caregiver, allowing us to reflect current living situations. The resulting estimates are child-centric measures of caregiver loss based on demographic and geographic data for caregivers and children at the individual level. Method Data We use data from two sources to compute the number of children who lost a parent or other co-residing caregiver to COVID-19. We obtained data on COVID-19 deaths through May 9, 2022, from the National Center for Health Statistics (NCHS), which provides data disaggregated by state, race and ethnicity, and age band. Estimates of caregiver loss also rely on population data from the U.S. Census Bureau’s American Community Survey Public Use Microdata Sample (PUMS) for the survey year 2019 (U.S. [32]. PUMS includes individual and household level data for more than 3 million individuals living in the United States, representing the demographic and household characteristics of the entire U.S. population. We used the PUMS dataset to compute the number of households, including the number of children younger than 18 years old living in each household and their relationship to each adult in that household. We considered all adults living in a household with a child to be a caregiver, and any parent to be a primary caregiver. Grandparents or adults were primary caregivers only when a parent was not present. Analysis We calculate COVID-19 death rates by dividing the number of COVID-19 deaths by age group, race/ethnicity, and state by the corresponding estimated size of that population segment (i.e., the weighted sample) from PUMS. We exclude people living in institutional and non-institutional group settings (like prisons or student dormitories) because they were unlikely to be actively living in households with children. Children are any individual less than 18 years old. For households with at least one child, we first calculate a total household probability of death by summing the death rates of all adults. We subtract the death rate of the child, as the death of the child precludes the child from caregiver loss, as well as the probability of multiple caregiver deaths. We then multiply the probability of losing caregiver(s) in the households of children of each state, race/ethnicity, and age group by the corresponding weighted sample size, resulting in an estimate of the number of COVID-19 bereaved children. This allows us to calculate demographic characteristics associated with caregiver loss in relation to the children. Results We estimate that 216,617 U.S. children experienced the death of a caregiver with whom they lived, due to COVID-19 infection, between January 1, 2020, and May 9, 2022. This corresponds to about 1 out of every 336 children under the age of 18. Table 1 presents caregiver loss totals and rates by child’s age, race, ethnicity, and their relationship with the deceased caregiver. In fifty percent of cases (108,353), the child lost a primary caregiver to COVID-19; forty-five percent of cases (97,738) lost a parent, and an additional five percent (10,615) lost another primary caregiver, usually a grandparent. Nearly 85,000 (84,917) lost a grandparent, most frequently a grandparent in a multigenerational household that also included a parent. More than 17,000 children (17,042) lost their only co-residing caregiver. Table 1 Number and characteristics of children who lost a caregiver to COVID-19 Number of children losing a co-residing caregiver Rate of children losing a co-residing caregiver (Per 100,000) Estimate 95% confidence interval Estimate 95% confidence interval Total 216,617 (191,877, 241,357) 298 (264, 332) By caregiver role  Grandparent 84,917 (52,722, 117,111) 117 (72, 161)  Parent 97,738 (87,643, 107,833) 134 (121, 148)  Primary caregiver 108,353 (98,042, 118,664) 149 (135, 163)  Sole caregiver 17,042 (14,612, 19,473) 23 (20, 27) By child race and ethnicity  Hispanic 78,217 (69,939, 86,495) 427 (382, 472)  American Indian or Alaskan native 5,224 (3,517, 6,930) 710 (478, 942)  Non-Hispanic asian 9,849 (7,466, 12,232) 281 (213, 394)  Non-Hispanic black 40,437 (35,987, 44,886) 418 (372, 464)  Native Hawaiian or Pacific islander 1,017 (536, 1,498) 679 (358, 999)  Non-Hispanic white 72,996 (67,247, 78,745) 200 (185, 216)  Other or multiracial 8,878 (7,185, 10,571) 226 (183, 270) By Child Age  0–4 43,977 (37,372, 50,581) 228 (194, 262)  5–13 108,641 (96,631, 120,652) 295 (262,327) 14–17 63,999 (57,874, 70,124) 386 (349, 423) Caregiver loss occurs in every state, though counts and rates vary widely as shown in Fig. 1. California (32,843) and Texas (30,861) have by far the largest number of children who lost a caregiver, each more than double the caregiver loss counts in the next highest state of New York. Six states—California, Texas, New York, Florida, Arizona, and Georgia—account for half of national caregiver loss. There is less notable clustering when examining per capita rates of caregiver loss. New Mexico has the highest rate of caregiver loss (510 per 100,000), with four other states plus Washington, DC, also having caregiver loss rates exceeding 400 per 100,000. Vermont and New Hampshire have the lowest rates of caregiver loss while Vermont and Wyoming have the lowest counts.Fig. 1 Rates of co-residing caregiver loss to COVID-19 per 100,000 children COVID-19 caregiver loss occurred in every racial and ethnic group considered, though losses are concentrated among non-White households. Children from non-White racial and/or Hispanic ethnic groups make up 66.3% of those who have lost a caregiver but only 50% of children in the U.S. population. Native Hawaiian/Pacific Islander children (HPI; 679 per 100,000) and American Indian or Alaskan Native children (AIAN; 710 per 100,000) had the highest per capita rates of caregiver loss; rates 3.4 and 3.55 times, respectively, higher than the rates of caregiver loss among White children. Hispanic and Non-Hispanic Black children had rates of caregiver loss 2.1 times those of White children. Hispanic children (78,217) accounted for the largest cohort of children that lost a caregiver to COVID-19. Notably, racial and ethnic disparities in caregiver loss exceed disparities in overall COVID-19 deaths [5]. Twenty percent of children experiencing caregiver loss (43,977) were from birth through 4 years old; 50.2% (108,641) were primary school age, from 5 through 13 years old, and the remaining 29.5% were 14 through 17 years old. Risk of losing a caregiver increased with the child’s age, ranging from 228 per 100,000 birth through 4 year old children to 386 per 100,000 children 14 through 17 years old. Disparities based on race and ethnicity were starkest for the youngest cohorts, as seen in Table 2. AIAN and HPI children under 5 years old had caregiver loss rates of 4.6 and 4.96 times their White peers, respectively, compared to 2.9 (AIAN and HPI) for their 14 through 17-year-old counterparts; similar age-based disparities were apparent for Hispanic, Non-Hispanic Black, and Non-Hispanic Asian children, though the differences by age were notably smaller.Table 2 COVID-19 Caregiver Loss by Race, Ethnicity, and Age: Estimates, Rates, and 95% Confidence Intervals Non-Hispanic Asian Non-Hispanic Black Native Hawaiian or Pacific Islander Hispanic American Indian or Alaskan Native Non-Hispanic white # Rate Per 100,000 # Rate Per 100,000 # Rate Per 100,000 # Rate Per 100,000 # Rate Per 100,000 # Rate Per 100,000 0–4 1,881 213 8,508 342 280 675 17,154 348 1,112 629 13,055 136 (1,254, 2,508) (142, 283) (7,143, 9,872 (287, 397) (112, 448) (269, 1080) (14,996, 19,312) (304, 392) (626, 1,599) (354, 905) (11,693, 14,417) (122, 150) 5–13 5,078 285 20,378 408 475 629 39,343 420 2,704 708 36,099 198 (3,857, 6,298) (216, 353) (18,349, 22,407) (367, 449) (271, 679) (359, 898) (35,337, 43,350) (377, 463) (1,861, 3,547) (488, 929) (33,274, 38,924) (182, 213) 14–17 2,891 347 11,551 527 262 798 21,720 540 1,407 796 23,841 278 (2,355, 3,426) (283, 412) (10,494) (479, 576) (153, 371) (457, 1,130) (19,606, 23,834) (487, 593) (1,030, 1,784) (583, 1,009) (22,279, 25,404) (260, 297) Disparities in caregiver loss by the decedent’s relationship to child largely follow overall trends, though there are some noteworthy distinctions. Black (65 per 100,000) and AIAN (64 per 100,000) children were most likely to lose their only co-residing caregiver. Asian children had the lowest rate of parental or any primary caregiver loss and their rate of losing their sole co-residing caregiver was 39% that of White children, the next lowest category. Discussion An estimated 216,617 children lost a co-residing caregiver who died of COVID-19 in the U.S. as of May 9, 2022. This is about 1 out of every 336 Americans under 18 years old. Many (97,738) lost their parent, nearly 85,000 children experienced the death of a grandparent caregiver, and at least 33,962 lost another adult with whom they lived. More than 17,000 children lost their only co-residing caregiver and are thus especially vulnerable. The current findings are largely consistent with past efforts to estimate the number of caregivers who have died of COVID-19 [14]; [19]. We updated those estimates and expanded on them by using data from the PUMS, which allowed us to assess loss in relation to detailed child demographics including current household structure, age, race, ethnicity, and state of residence. Children and families who experience loss must adapt to life without the caregiver; this will include coping with grief and, for many, renegotiating family roles, responsibilities, and meeting basic physical, material, emotional, and cognitive needs. Although, most children who lose a caregiver will demonstrate resilience and healthy development, others will struggle with grief complicated by traumatic stress, other symptoms, or other developmental problems [17]. The context of the COVID-19 pandemic presents additional challenges because important resources to promote positive adaptation and financial stability may be more difficult to access due to remote education, social-distancing that impacts supportive relationships, or strained mental health, health, and social services. That COVID-19 deaths occur disproportionately among vulnerable communities exacerbates the challenge of accessing services for many bereaved children. A comprehensive response to caregiver loss will include resources to promote resilience for children, youth, and families who need different levels of support. Providing the necessary range of supports for more than 200,000 children and their families requires coordinated efforts from health and mental health systems, community-based organizations, philanthropies, and federal, state, and local policymakers whose attention and funding are critical to the success of a comprehensive, substantive, and sustained effort. They should leverage existing infrastructures like schools and pediatric health care settings that routinely interact with children, screen for social needs, and are putative service hubs. Early childhood education facilities, faith-based institutions, youth sports leagues, and other community-based settings in which adults routinely interface with children and youth provide additional opportunities to identify and serve bereaved children. A coordinated response to caregiver loss must target resources to the vulnerable communities bearing the disproportionate share of COVID-19 deaths. Caregiver death was concentrated in the most populous states (e.g., California, Florida, Georgia, New York, and Texas), among older children, and among racial and ethnic groups that have been systemically marginalized in the U.S. Non-White and/or Hispanic children constituted two-thirds of caregiver loss, with caregiver loss rates more than double those of White, non-Hispanic children. Disparities of caregiver loss by race and ethnicity exceed disparities in COVID-19 deaths; this is likely due to both the higher COVID-19 death rate among Non-White adults under 65 and the higher average number of adult caretakers among children identified as Hispanic, American Indian and Alaskan Native, and Native Hawaiian or Other Pacific Islander, compared to their Non-Hispanic White counterparts. Our ability to observe more specific geographic and demographic disparities is constrained by the limitations of available data, but we can infer from other analyses that caregiver loss is even more heavily concentrated in vulnerable populations than shown in our results. Findings from Seligman, Ferranna, and Bloom [28], Chen and colleagues [6], and Gross and colleagues [11] suggest that these losses are concentrated among low-income households and in lower-income communities that have been the subject of systemic disinvestment that may deprive children of resources critical for adapting to adversity. Such disadvantage usually spans generations and can cumulatively constrain the development of individual resources (e.g., educational attainment,emotional and behavioral regulation skills) and opportunities for external support (family functioning and other close relationships; effective health and human services) that are central to promoting and protecting positive outcomes [13]. The infrastructure to support affected households, especially in marginalized communities in which COVID-19 deaths are concentrated, must be buttressed through investments in capacity and grief competence. Professionals who interact with grieving children in normative settings like classrooms, community-based nonprofits, and primary care providers, are often poorly prepared to address their needs, and low-cost and free capacity building programs available to organizations already embedded within vulnerable communities may efficiently fill that gap for children whose needs are limited to preventative grief services [21]. For children whose needs include more clinical services like psychotherapy, mental health care—of which there is a national shortage—may be unavailable or unaffordable for communities hit hardest by COVID-19, and clinicians often lack training in grief-focused therapies [7, 27]. Grief-focused training programs may allow existing providers to offer appropriate care. Policymakers, health systems, and insurance providers can also explore options to expand access to mental health care where it is unavailable or unaffordable through new telehealth options and reductions in patients’ cost-sharing responsibilities for grief-focused care. In the long-term, mental health professional training programs may seek to increase training in grief-focused care as well as recruitment among rural, low-income, and non-White populations who may be more likely and more able to provide culturally competent care in underserved communities. Families in particularly vulnerable communities will also disproportionately need financial support to weather the costs a caregiver’s death and potential lost income; these challenges are especially salient given that COVID-19 deaths among working-age adults disproportionately include low-wage workers unlikely to have sufficient savings or life insurance policies to sustain their households for an extended period. Health and social service entities that interface with bereaved families should identify those that may be eligible for existing mechanisms of financial support, like Social Security Survivor’s Benefits, the Federal Emergency Management Agency’s Funeral Reimbursement Program, the Child Tax Credit, and means-tested programs like Temporary Assistance for Needy Families, the Supplemental Nutrition Assistance Program and Medicaid. In addition, there is precedent for ad hoc efforts from the public and philanthropic sectors to provide interim and supplementary financial assistance to victims of disasters [8, 29]. Limitations This study has limitations. First, estimates based on extrapolation are subject to inherent uncertainty, which we sought to minimize. Second, COVID-19 deaths among some racial and ethnic groups—people of Hispanic ethnicity or those identifying their race as American Indian or Alaska Native, Asian or Pacific Islander—may have been undercounted; thus, our estimates of their caregiver loss should be interpreted as lower bounds [3]. Third, we rely on 2019 population data, as pandemic-related complications rendered 2020 population estimates from the American Community Survey “experimental” [31]. More recent data may better reflect the distributions of demographic factors and COVID-19 deaths. Conclusions Many children in the U.S. have lost a caregiver to COVID-19. Most lost a parent, including a sizeable number of children who lost their only caregiver. Caregiver deaths are ubiquitous, spanning the entire country and all demographic categories, though they are concentrated among children and youth from marginalized racial and ethnic groups. These children and their families must confront the processes of grief while also meeting developmental needs required for long-term positive outcomes. Restrictions and other consequences of the COVID-19 pandemic present special challenges as children, families, and communities attempt to promote resilience. A comprehensive response to this phenomenon must be adaptable to the varying needs of children while targeting the vulnerable communities that faced the largest share of COVID-19 deaths with the fewest resources with which to cope. Author Contributions DT: guided the conceptualization, data analysis, and writing, wrote sections of the first draft of the manuscript, prepared tables, and critically reviewed and contributed to the full manuscript; JJC: contributed to the conceptualization of the manuscript, wrote sections of the first draft of the manuscript, and critically reviewed and contributed to the full manuscript; KA: conducted statistical analyses, verified the underlying data, wrote the first draft of the methods section, and critically reviewed and contributed to the full manuscript; JB: contributed to the conceptualization of the study and critically reviewed and contributed to the full manuscript. Funding Funding was provided through the COVID Collaborative from the Allstate Foundation, the Helmsley Charitable Trust, New York Life Foundation, the Skoll Foundation, and the Walton Family Foundation. Data Availability Data are from publicly available sources, though a compiled dataset has not been made publicly available. Interested parties can contact the corresponding author to access the data included in this analysis. Code Availability Interested parties may contact the corresponding author to obtain the code used to produce results included in this manuscript. Declarations Conflicts of interest Mr. Bridgeland is CEO of the COVID Collaborative, a national assembly of experts, leaders and institutions in health, education and the economy and associations representing the diversity of the country to turn the tide on the pandemic by supporting global, federal, state, and local COVID-19 response efforts. Drs. Treglia, Cutuli, and Arasteh received funding for this study through the COVID Collaborative and Social Policy Analytics. Dr. Treglia is CEO of Social Policy Analytics and an Expert Contributor to the COVID Collaborative, with whom he consults on COVID-19 related caregiver loss. Ethical Approval This study was reviewed by the Nemours Children’s Health Institutional Review Board and deemed not to be human subjects research. Consent to Participate Not applicable. Consent for Publication Not applicable. We have received expertise and general advice from many people whose collective contributions are critical to this work, and we thank them for their time, patience, and commitment to helping children suffering the loss of a parent or other caregiver. In particular, we want to thank Dr. Randall Kuhn for his methodological guidance, Dr. Bikki Tran Smith for creating the maps contained within our manuscript, and Delaney Michaelson for her general research and organizational efforts. We also wish to acknowledge funding from the Allstate Foundation, the Helmsley Charitable Trust, New York Life Foundation, the Skoll Foundation, and the Walton Family Foundation that made this work possible. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Berg L Rostila M Hjern A Parental death during childhood and depression in young adults – a national cohort study Journal of Child Psychology and Psychiatry and Allied Disciplines 2016 57 9 1092 1098 10.1111/jcpp.12560 27058980 2. Brent DA Melhem NM Masten AS Porta G Payne MW Longitudinal effects of parental bereavement on adolescent developmental competence Journal of Clinical Child and Adolescent Psychology 2012 41 6 778 791 10.1080/15374416.2012.717871 23009724 3. CDC. (2021). Health Disparities Provisional Death Counts for Coronavirus Disease 2019 (COVID-19). 4. Centers for Disease Control and Prevention. (2021). Distribution of COVID-19 Deaths and Populations, by Jurisdiction, Age, and Race and Hispanic Origin. Retrieved December 22, 2021, from https://data.cdc.gov/NCHS/Distribution-of-COVID-19-Deaths-and-Populations-by/jwta-jxbg 5. Centers for Disease Control and Prevention. (2022). Risk for COVID-19 Infection, Hospitalization, and Death By Race/Ethnicity. Retrieved February 15, 2022, from https://www.cdc.gov/coronavirus/2019-ncov/covid-data/investigations-discovery/hospitalization-death-by-race-ethnicity.html 6. Chen YH Glymour M Riley A Balmes J Duchowny K Harrison R Bibbins-Domingo K Excess mortality associated with the COVID-19 pandemic among Californians 18–65 years of age, by occupational sector and occupation: March through November 2020 PLoS ONE 2021 16 1 10 10.1371/journal.pone.0252454 7. Cicchetti RJ McArthur L Szirony GM Blum C Perceived competency in grief counseling: Implications for counselor education Journal of Social, Behavioral, and Health Sciences 2016 10 1 3 17 10.5590/jsbhs.2016.10.1.02 8. E4E Relief. (2021). Brave of Heart Fund. Retrieved July 2, 2022, from https://employeerelieffund.org/brave-of-heart-fund 9. Eggers FJ Moumen F Analysis of Trends in Household Composition Using American Housing Survey Data SSRN Electronic Journal 2014 10.2139/ssrn.2445476 10. Gimenez L Chou S-Y Liu J-T Liu J-L Parental loss and children ’ s well- being Journal of Human Resources 2013 48 1035 1071 10.1353/jhr.2013.0028 11. Gross CP Essien UR Pasha S Gross JR Wang SY Nunez-Smith M Racial and ethnic disparities in population-level Covid-19 mortality Journal of General Internal Medicine 2020 35 10 3097 3099 10.1007/s11606-020-06081-w 32754782 12. Hamdan S Melhem NM Porta G Song MS Brent DA Alcohol and substance abuse in parentally bereaved youth The Journal of Clinical Psychiatry 2013 10.4088/JCP.13m08391 13. Herbers JE Hayes KR Cutuli JJ Adaptive systems for student resilience in the context of COVID-19 School Psychology 2021 36 5 422 426 10.1037/spq0000471 34591590 14. Hillis SD Blenkinsop A Villaveces A Annor FB Liburd L Massetti GM Unwin HJT COVID-19-associated orphanhood and caregiver death in the United States Pediatrics 2021 148 e2021053760 10.1542/peds.2021-053760 15. Hillis SD Unwin HJT Chen Y Cluver L Sherr L Goldman PS Flaxman S Global minimum estimates of children affected by COVID-19-associated orphanhood and deaths of caregivers: A modelling study The Lancet 2021 398 10298 391 402 10.1016/S0140-6736(21)01253-8 16. Joint United Nations Programme on HIV/AIDS. (2009). National AIDS Spending Assessment (NASA): Classification and definitions. Geneva: UNAIDS. Available from https://www.unaids.org/sites/default/files/media_asset/jc1557_nasa_en_0.pdf 17. Kaplow JB Layne CM Pynoos RS Cohen JA Lieberman A DSM-V diagnostic criteria for Bereavement- Related Disorders in children and adolescents: Developmental considerations Psychiatry 2012 75 3 243 266 10.1521/psyc.2012.75.3.243 22913501 18. Keyes KM Pratt C Galea S Mclaughlin KA Koenen KC Shear MK The burden of loss: unexpected death of a loved one and psychiatric disorders across the life course in a national study American Journal of Psychiatry 2014 171 864 781 10.1176/appi.ajp.2014.13081132 24832609 19. Kidman R Margolis R Smith-Greenaway E Verdery AM Estimates and projections of COVID-19 and parental death in the U.S Journal of the American Medical Association 2021 175 7 745 746 10.1111/cen.12738 20. Li J Vestergaard M Cnattingius S Gissler M Bech BH Obel C Mortality after parental death in childhood: a nationwide cohort study from three Nordic countries National Library of Medicine. 2014 11 e1001679 21. New York Life Foundation, & American Federation of Teachers. (2020). Grief in School survey: key findings and topline results. 22. Parsons, S. (2011). Long-term impact of childhood bereavement, 1–22. 23. Pew Research Center. (2013). Grandparents living with or serving as primary caregivers for their grandchildren. 24. Pew Research Center. (2015). Parenting in America: Outlook, worries, aspirations are strongly linked to financial situation. Numbers, Facts and Trends Shaping the World. Retrieved from https://www.pewresearch.org/social-trends/2015/12/17/parenting-in-america/ 25. Pilkauskas NV Amorim M Dunifon RE Historical trends in children living in multigenerational households in the United States: 1870–2018 Demography 2020 57 6 2269 2296 10.1007/s13524-020-00920-5 33001418 26. Rosenstock S Whitman S West JF Balkin M Racial Disparities in Diabetes Mortality in the 50 Most Populous US Cities Journal of Urban Health 2014 91 5 873 885 10.1007/s11524-013-9861-4 24532483 27. SAMHSA. (2020). Double Jeopardy: COVID-19 and Behavioral Health Disparities for Black and Latino Communities in the U.S. Kaiser Family Foundation. 28. Seligman B Ferranna M Bloom DE Social determinants of mortality from COVID-19: A simulation study using NHANES PLoS Medicine 2021 18 1 1 13 10.1371/journal.pmed.1003490 29. September 11th Victims Compensation Fund. (2021). Annual Report 2020. 30. Simbi CMC Zhang Y Wang Z Early parental loss in childhood and depression in adults: A systematic review and meta-analysis of case-controlled studies Journal of Affective Disorders 2020 260 272 280 10.1016/j.jad.2019.07.087 31521863 31. U.S. Census Bureau. (2021a). 2020 ACS 1-Year Public Use Microdata Sample with Experimental Weights. Retrieved February 10, 2022, from https://www.census.gov/programs-surveys/acs/data/experimental-data/2020-1-year-pums.html 32. U.S. Census Bureau. (2021b). Accessing PUMS Data. Retrieved October 10, 2021, from https://www.census.gov/programs-surveys/acs/microdata/access.2019.html 33. Zhang H Rodriguez-Monguio R Racial disparities in the risk of developing obesity-related diseases: A cross-sectional study Ethnicity and Disease 2012 22 3 308 316 22870574

36515763

PMC9749637

NO-CC CODE

2022-12-15 23:23:21

no

J Community Health. 2022 Dec 14;:1-8

utf-8

J Community Health

10.1007/s10900-022-01160-x

oa_other

==== Front Int J Hematol Int J Hematol International Journal of Hematology 0925-5710 1865-3774 Springer Nature Singapore Singapore 3508 10.1007/s12185-022-03508-4 Original Article Effect of the COVID-19 pandemic on allogeneic stem cell transplantation in Japan http://orcid.org/0000-0003-1018-9508 Shimomura Yoshimitsu shimomura_0119@yahoo.co.jp 12 Kitamura Tetsuhisa 2 Nishikubo Masashi 1 Sobue Tomotaka 2 Uchida Naoyuki 3 Doki Noriko 4 Tanaka Masatsugu 5 Ito Ayumu 6 Ishikawa Jun 7 Ara Takahide 8 Ota Shuichi 9 Onizuka Makoto 10 Sawa Masashi 11 Ozawa Yukiyasu 12 Maruyama Yumiko 13 Ikegame Kazuhiro 14 Kanda Yoshinobu 15 Ichinohe Tatsuo 16 Fukuda Takahiro 6 Okamoto Shinichiro 17 Teshima Takanori 18 Atsuta Yoshiko 1920 1 grid.410843.a 0000 0004 0466 8016 Department of Hematology, Kobe City Hospital Organization Kobe City Medical Center General Hospital, Minamimati 2-1-1, Minatojima, Chuo-Ku, Kobe, 650-0047 Japan 2 grid.136593.b 0000 0004 0373 3971 Department of Environmental Medicine and Population Science, Graduate School of Medicine, Osaka University, Suita, Japan 3 grid.410813.f 0000 0004 1764 6940 Department of Hematology, Toranomon Hospital, Tokyo, Japan 4 grid.415479.a Hematology Division, Tokyo Metropolitan Cancer and Infectious Diseases Center, Komagome Hospital, Tokyo, Japan 5 grid.414944.8 0000 0004 0629 2905 Department of Hematology, Kanagawa Cancer Center, Kanagawa, Japan 6 grid.272242.3 0000 0001 2168 5385 Department of Hematopoietic Stem Cell Transplantation, National Cancer Center Hospital, Tokyo, Japan 7 grid.489169.b 0000 0004 8511 4444 Department of Hematology, Osaka International Cancer Institute, Osaka, Japan 8 grid.412167.7 0000 0004 0378 6088 Department of Hematology, Hokkaido University Hospital, Sapporo, Japan 9 grid.415262.6 0000 0004 0642 244X Department of Hematology, Sapporo Hokuyu Hospital, Sapporo, Japan 10 grid.265061.6 0000 0001 1516 6626 Department of Hematology, Tokai University School of Medicine, Kanagawa, Japan 11 grid.413779.f 0000 0004 0377 5215 Department of Hematology and Oncology, Anjo Kosei Hospital, Anjo, Japan 12 Department of Hematology, Japanese Red Cross Aichi Medical Center Nagoya Daiichi Hospital, Nagoya, Japan 13 grid.412814.a 0000 0004 0619 0044 Department of Hematology, University of Tsukuba Hospital, Tsukuba, Japan 14 grid.272264.7 0000 0000 9142 153X Department of Hematology, Hyogo Medical University Hospital, Nishinomiya, Japan 15 grid.415020.2 0000 0004 0467 0255 Division of Hematology, Jichi Medical University Saitama Medical Center, Saitama, Japan 16 grid.257022.0 0000 0000 8711 3200 Department of Hematology, Research Institute for Radiation Biology and Medicine, Hiroshima University, Hiroshima, Japan 17 grid.26091.3c 0000 0004 1936 9959 Division of Hematology, Department of Medicine, Keio University School of Medicine, Tokyo, Japan 18 grid.39158.36 0000 0001 2173 7691 Department of Hematology, Faculty of Medicine, Hokkaido University, Sapporo, Japan 19 grid.511247.4 Japanese Data Center for Hematopoietic Cell Transplantation, Nagakute, Japan 20 grid.411234.1 0000 0001 0727 1557 Department of Registry Science for Transplant and Cellular Therapy, Aichi Medical University School of Medicine, Nagakute, Japan 14 12 2022 18 13 10 2022 1 12 2022 2 12 2022 © Japanese Society of Hematology 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. The coronavirus disease 2019 (COVID-19) pandemic affected healthcare quality and access worldwide and may also have negatively affected the frequency and outcomes of allogeneic hematopoietic stem cell transplantation (HSCT). We evaluated the effect of the pandemic on allogeneic HSCT in Japan. Our subjects were patients who received allogeneic HSCT during January 2018–December 2020 in Japan. We assessed differences in yearly number of allogeneic HSCTs and 1-year outcomes in 2020 versus both 2019 and 2018. The total number of patients who received allogeneic HSCT increased from 3621 patients in 2018 and 3708 patients in 2019 to 3865 patients in 2020. Some following changes in allogeneic HSCT methods were observed: patients were older, fewer patients received bone marrow transplantation, fewer patients received transplants from unrelated donors, fewer patients received transplants from matched donors, more patients received reduced-intensity conditioning, and fewer patients received anti-thymocyte globulin in 2020 compared with previous years. HSCT outcomes were not affected, as 1-year overall survival was not significantly different (65.8% in 2020, vs. 66.5% in 2019 and 66.4% in 2018). Our results suggest that we can maintain transplant care during the pandemic by controlling the spread of COVID-19 and modifying HSCT methods. Keywords Coronavirus disease 2019 Pandemic Social restriction Allogeneic hematopoietic stem cell transplantation Transplant activity and outcome ==== Body pmcIntroduction Coronavirus disease 2019 (COVID-19) presents a global health threat and has caused substantial mortality [1]. Since the first outbreak at the end of 2019, COVID-19 has spread worldwide, causing a million deaths [2]. The World Health Organization declared the COVID-19 pandemic on March 11, 2020 [3]. The COVID-19 pandemic also profoundly affected healthcare quality and access worldwide because of restrictions based on social distancing and movement, aiming to mitigate and manage the spread of COVID-19 [4–9]. As a result of the pandemic, the Japanese government declared a state of emergency for the metropolitan areas on April 7, 2020, which subsequently expanded to a nationwide on April 16 and continued until May 25. There were three waves of COVID-19 outbreak in 2020, which peaked in April, August, and December. Japan more successfully controlled the COVID-19 pandemic than Western countries; the number of infected people and deaths in 2020 were 187 per 100,000 and 3492, respectively [2]. Even still, the COVID-19 pandemic affected healthcare quality and access in fields such as cancer and emergency medicine throughout Japan [10–13]. Allogeneic hematopoietic stem cell transplantation (HSCT) is an important treatment option to cure hematological disorders [14]. Patients diagnosed with hematological disorders and having undergone HSCT have been affected by COVID-19 with an initial dramatic mortality rate [15–17]. In addition, there were donor harvest cancellations, donor or patient illness, and operational restrictions [18]. With such concerns, it is expected that transplantation care was also affected to some degree by the COVID-19 pandemic. In the annual activity survey, the European Society for Blood and Marrow Transplantation (EBMT) group revealed that transplant activity decreased in 2020 compared to that of the pre-COVID-19 years [17]. However, there is little available evidence and remaining clinical questions, such as whether transplant activity was decreased in countries where the pandemic was controllable and the extent of change in transplantation outcomes amid the COVID-19 pandemic. Therefore, we aimed to evaluate the effect of the COVID-19 pandemic on transplant activity and its outcomes in Japan using the Second-Generation Transplant Registry Unified Management Program (TRUMP 2) from the Japanese Data Center for Hematopoietic Cell Transplantation (JDCHCT). Methods Data collection All data were obtained from the TRUMP2 database, which is a web-based HSCT registry in Japan established to collect clinical outcome data and evaluate the actual situation of HSCT [19]. Patient consent was obtained before the registration of the TRUMP2. This study was approved by the Data Management Committee of the JDCHCT and the Ethics Committee of Kobe City Medical Center General Hospital (#zn220802). Patient measures The patient and disease characteristics are summarized in Table 1. The hematopoietic cell transplantation-specific comorbidity index (HCT-CI) was defined as previously described [20]. The myeloablative conditioning regimen and reduced-intensity conditioning regimen (RIC) were defined as previously described [21, 22]. Human leukocyte antigen (HLA)-matched donor was defined as the same serologically identified HLA-A, HLA-B, and HLA-DRB1 between the donor and the recipient. Haplo-identical donors were defined as related donors mismatched at three HLA antigen levels of HLA-A, HLA-B, and HLA-DRB1. Other donors were defined as mismatched donor. Disease risk was assessed in patients with acute myeloid leukemia, acute lymphoblastic leukemia, myelodysplastic syndrome/myeloproliferative neoplasm, and malignant lymphoma. We defined AML and ALL without remission or above third complete remission and malignant lymphoma with stable and progressive disease as high risk and others as low risk.Table 1 Patient characteristics Characteristic Overall N = 11,194 2018 N = 3621 2019 N = 3708 2020 N = 3865 p value1 p value2 Age, years 50 (33, 60) 50 (33, 60) 49 (32, 60) 51 (35, 61) 0.026 0.016 Child (≤ 15 years) 1262 (11%) 409 (11%) 427 (12%) 426 (11%) 0.084 0.051 Adult (16–59 years) 7163 (64%) 2342 (65%) 2397 (65%) 2424 (63%) Older (60+ years) 2769 (25%) 870 (24%) 884 (24%) 1015 (26%) Sex 0.914 0.439  Female 4580 (41%) 1496 (41%) 1493 (40%) 1591 (41%)  Male 6614 (59%) 2125 (59%) 2215 (60%) 2274 (59%) HCT-CI 0 (0, 2) 0 (0, 2) 0 (0, 2) 0 (0, 2) 0.381 0.118  Missing 132 27 27 78  High (≥ 3) 1982 (18%) 644 (18%) 652 (18%) 686 (18%) 0.850 0.672  Low (0–2) 9080 (82%) 2950 (82%) 3029 (82%) 3101 (82%) Performance status 2 (1, 2) 2 (1, 2) 2 (1, 2) 2 (1, 2) 0.653 0.924  Missing 82 13 12 57  High (2–4) 1146 (10%) 399 (11%) 364 (10%) 383 (10%) 0.172 0.792  Low (0–1) 9966 (90%) 3209 (89%) 3332 (90%) 3425 (90%) Number of HSCT 0.416 0.648  Missing 111 36 29 46  1st HSCT 9456 (85%) 3037 (85%) 3157 (86%) 3262 (85%)  ≥ 2nd HSCT 1627 (15%) 548 (15%) 522 (14%) 557 (15%) Donor source < 0.001 < 0.001  Bone marrow 3544 (32%) 1241 (34%) 1237 (33%) 1066 (28%)  Peripheral blood 3522 (31%) 1089 (30%) 1118 (30%) 1,315 (34%)  Cord blood 4128 (37%) 1291 (36%) 1353 (36%) 1484 (38%) Donor type 0.046 0.003  Related 3549 (32%) 1132 (31%) 1123 (30%) 1294 (33%)  Unrelated 7645 (68%) 2489 (69%) 2585 (70%) 2571 (67%) HLA in patients with BM or PB transplant < 0.001 0.002  Missing 24 4 3 17  Matched 4523 (66%) 1539 (66%) 1541 (66%) 1443 (61%)  Mismatched 1842 (25%) 593 (26%) 600 (26%) 649 (28%)  Haplo-identical 677 (9%) 194 (8%) 211 (9%) 272 (12%) Conditioning 0.021 0.014  Missing 58 5 3 50  Myeloablative 6050 (54%) 1996 (55%) 2051 (55%) 2003 (53%)  Reduced intensity 5086 (46%) 1620 (45%) 1654 (45%) 1812 (47%) GVHD prophylaxis < 0.001 0.643  Missing 329 93 81 155  CSA base 1906 (18%) 690 (20%) 609 (17%) 607 (16%)  TAC base 8959 (82%) 2838 (80%) 3018 (83%) 3103 (84%) ATG < 0.001 < 0.001  Missing 93 19 14 60  Yes 1824 (16%) 637 (18%) 643 (17%) 544 (14%)  No 9277 (84%) 2965 (82%) 3051 (83%) 3261 (86%) Disease 0.222 0.255  AML 4353 (39%) 1397 (39%) 1428 (39%) 1528 (40%)  MDS/MPN 1932 (17%) 600 (17%) 624 (17%) 708 (18%)  ALL 1921 (17%) 636 (18%) 654 (18%) 631 (16%)  ML 1223 (11%) 399 (11%) 421 (11%) 403 (10%)  Other malignancies 1088 (10%) 358 (10%) 364 (10%) 366 (10%)  Non-malignant disease 677 (6%) 231 (6%) 217 (6%) 229 (6%) Disease risk in patients with AML, ALL, MDS and ML 0.002 0.076  Missing 386 124 114 148  High risk 2979 (33%) 1013 (35%) 998 (33%) 968 (31%)  Low risk 6064 (67%) 1895 (65%) 2015 (67%) 2154 (69%) Continuous variables are summarized as medians and interquartile ranges (quartiles 1–3), and categorical variables are summarized as numbers and percentages AML acute myeloid leukemia, ALL acute lymphoblastic leukemia, ATG antithymocyte globulin, CsA cyclosporine A, GVHD graft-versus-host disease, HCT-CI hematopoietic cell transplant comorbidity index, HSCT hematopoietic stem cell transplantation, MDS myelodysplastic syndrome, ML malignant lymphoma, MPN myeloproliferative neoplasm, Tac tacrolimus 1P value compared 2020 and 2018 2P value compared 2020 and 2019 Endpoints and statistical analyses To assess the difference in the total number of allogeneic HSCTs between 2020 and both 2018 and 2019, we showed the absolute number of patients who received HSCT and the relative ratio (RR), which was calculated considering the number of hematopoietic disorder cases per year as the denominator. Similarly, we showed the absolute number of HSCTs and RR per month to evaluate the effect of the waves of the COVID-19 pandemic and state of emergency. Continuous variables were summarized using medians and interquartile ranges (quartiles 1–3), and categorical variables were summarized as numbers and percentages. Data were compared using the Mann–Whitney U test for continuous variables and chi-square tests for categorical variables. We excluded patients without information on outcomes (n = 244) in the analysis of outcomes after HSCT. Then, we evaluated the following endpoints in regard to HSCT outcomes: 1-year overall survival rate (OS), 1-year cumulative incidence of relapse, and 1-year cumulative incidence of non-relapse mortality (NRM). If patients did not achieve complete remission after allogeneic HSCT, relapse was considered immediately after the allogeneic HSCT. Relapse was analyzed considering NRM as a competing risk factor. NRM was defined as death without relapse and analyzed considering relapse as a competing risk. Event rates were estimated with 95% confidence intervals (CI) using the Kaplan–Meier or Gray’s method for the OS and other endpoints. Gray’s method was employed to consider the competing risks. We also performed a univariate Cox proportional-hazards model for the OS and the Fine-Gray method for other endpoints. The endpoints were described as adjusted hazard ratios (HRs) and 95% CIs. The Fine-Gray method was employed to consider the competing risks. Statistical significance was set at P < 0.05. All the statistical analyses were performed using the R software package (version 4.0.2; R Development Core Team). Results HSCT activity The present study included 11,194 patients who received allogeneic HSCT between January 2018 and December 2020. The total number of patients who received allogeneic HSCT increased from 3621 patients in 2018 and 3708 patients in 2019 to 3865 patients in 2020 (Fig. 1, Table 1). There were no significant differences in the monthly numbers of HSCT between 2020 and 2018 or 2019 in corresponding months except for February, October, and December in 2018 (Fig. 1).Fig. 1 Number of hematopoietic stem cell transplantation in 2018, 2019, and 2020. In the upper panel, the points indicate relative risk, whereas the vertical lines indicate 95% confidence intervals. In the lower panel, the monthly numbers of overall patients who underwent hematopoietic stem cell transplantation are shown. *Indicates significant difference. **Total indicated average monthly number of patients who underwent hematopoietic stem cell transplantation. Patient characteristics Patient characteristics are shown in Table 1. The median age of patients was 50 years (interquartile range, 33–60 years), with 59% (n = 6614) being male. The median HCT-CI was zero (interquartile range, 0–2). Eighty-five percent of patients (n = 9456) received first allogeneic HSCT. Regarding the donor source, 32% (n = 3544), 31% (n = 3522), and 37% (n = 4128) of patients received bone marrow transplantation, peripheral blood transplantation, and cord blood transplantation, respectively. Among patients that received bone marrow or peripheral blood transplantation, 66% (n = 4523), 25% (n = 1842), and 9% (n = 677) of patients received transplants from HLA-matched donor, mismatched donor, and haploidentical donor, respectively. The majority of diseases requiring HSCT were acute myeloid leukemia (n = 4353, 39%), acute lymphoblastic leukemia (n = 1921, 17%), myelodysplastic syndrome/myeloproliferative neoplasm (n = 1932, 17%), and malignant lymphoma (n = 1223, 11%). The patient characteristics differed between patients who received HSCT in 2020 versus those in 2018 or 2019; the patient group transplanted in 2020 included more elderly patients, less patients who received bone marrow transplantation, fewer patients transplanted from unrelated and matched donors, more patients who received RIC, and less patients who received antithymocyte globulin compared with the patient groups transplanted in 2019 and 2018.Fig. 2 Transplant outcomes compared 2018, 2019 and 2020. A Overall survival, B cumulative incidence of relapse, and C cumulative incidence of non-relapse mortality Outcomes after HSCT In total, 10,950 patients were included in the outcome analysis after HSCT. The 1-year OS was 65.8% (95% CI 64.0–67.5%) in patients transplanted in 2020 versus 66.5% (95% CI 64.9–68.1%) and 66.4% (95% CI 64.8–67.9%) in patients transplanted in 2019 and 2018, respectively (P = 0.926) (Fig. 2A). The HR of OS in patients transplanted in 2020 was 1.02 (95% CI 0.93–1.11) compared with that in patients transplanted in 2019 (p = 0.639) and 1.00 (95% CI 0.96–1.05) compared with that in patients transplanted in 2018 (p = 0.900). The 1-year cumulative incidence of relapse was 27.1% (95% CI 25.6–28.7%) in patients transplanted in 2020 vs. 24.6% (95% CI 23.2–26.0%) and 24.9% (95% CI 23.5–26.4%) in patients transplanted in 2019 and 2018, respectively (p = 0.037) (Fig. 2B). The HR of relapse in patients transplanted in 2020 was significantly higher than that of patients transplanted in 2019 (HR 1.10, 95% CI 1.08–1.21, p = 0.033); it also tended to be higher than that of patients transplanted in 2018 (HR 1.04, 95% CI 1.00–1.09, p = 0.076). We performed the ad-hoc stratified analysis considering disease risk, which was a significant factor for relapse. The 1-year cumulative incidence of relapse was not significant among low-risk patients. In high-risk patients, the HR of relapse in patients that underwent transplantation in 2020 was significantly higher than that of patients who received transplants in 2019 (HR 1.26, 95% CI 1.09–1.46, p = 0.002) and in 2018 (HR 1.33, 95% CI 1.15–1.55, p < 0.001). In contrast, the cumulative incidence of NRM was not significantly different among the three groups (16.4% [95% CI 15.1–17.8] in 2020 vs. 17.1% [95% CI 15.9–18.4] in 2019 vs. 17.0% [95% CI 15.8–18.3] in 2018, p = 0.648) (Fig. 2C). Discussion Using the Japanese multicenter HSCT registry, we revealed that the total transplant activity in 2020 did not decrease compared with the previous two years. In a detailed comparison between 2020 and the previous two years, there were changes in the method of allogeneic HSCT, such as selection of donor and conditioning regimens. The transplant outcomes were similar in 2020 compared with the previous two years, except for a slight increase of cumulative incidence of relapse (HR 1.04–1.10). It can be inferred that the activity and outcomes were maintained as usual by modification of HSCT methods. Possessing several modifiable methods may maximize patient benefit in the event of a contingency. These findings are influenced by the fact that the pandemic was more successfully controlled in Japan than in Western countries [2]. Our results also suggest that we may be able to maintain transplant care as usual by controlling the pandemic to some degree. In the annual activity survey, the EBMT group revealed that the transplant activity was down 6.5% in 2020 compared with that of 2019 [17]. Regarding the methods of allogeneic HSCT, a change to a RIC regimen and a decrease in transplants from unrelated donors were observed due to restrictions associated with successive waves of COVID-19, including lockdown. In our study, the total transplant activity in 2020 was not decreased compared with that of the previous two years, which was inconsistent with the study from the EBMT. Similarly, the monthly trend of allogeneic HSCT did not decrease even after declaration of a state of emergency and outbreak waves. Social reasons for the decreased transplant activity in the COVID-19 pandemic year include lax restrictions, national success in infection control, and good access to medical facilities [23]. Additionally, cord blood, which is commonly used as a transplant source in Japan, was easy to use even within social restrictions and may have complemented the decrease of other donor sources [24, 25]. Similarly, cryopreservation of donor grafts from the Japan Marrow Donor Program might be another reason for maintenance of the transplant activity during the COVID-19 pandemic [26, 27]. The trend of changes in transplantation methods are similar to those reported by the EBMT, such as donor and conditioning selection [17]. The results indicate that the COVID-19 pandemic influenced transplant medicine even in Japan, although the total activity was maintained. There is little available evidence regarding the changes in transplant outcomes in the COVID-19 pandemic. Although it was expected to have a negative effect as in the results of other studies, the COVID-19 pandemic in Japan did not affect the OS and NRM of patients undergoing allogeneic HSCT in our study [28, 29]. The results indicate that there was no need to postpone transplantation out of concern for worsening patient outcomes, although it should be noted that only short-term outcomes could be evaluated. Conversely, the cumulative relapse incidence was slightly higher in 2020 than in 2018 and 2019 with a HR of 1.04–1.10. The increased relapse rate was remarkable in patients with a high-risk disease status. The increased relapse rate was due to changes in patient characteristics. More patients with poor disease conditions might have undergone allogeneic HSCT during the COVID-19 pandemic. Additionally, delayed diagnosis in the COVID-19 pandemic may be a contributing factor for the result. It is important to note this study is not without limitations. First, our study was conducted in Japan, one of the most successful countries in controlling the COVID-19 pandemic; therefore, researchers need to be careful regarding the extrapolation of the results. Second, there is currently no data regarding long-term outcomes. The COVID-19 pandemic may not only affect the early post-HSCT period; we need to collect and analyze long-term data to understand the holistic effects of the pandemic on HSCT outcomes. Third, we could not understand the reason behind some changes such as the increased number of elderly patients undergoing transplantation. Finally, we had no data regarding COVID-19 in study patients. Consecutive studies are needed to clarify the effect of the COVID-19 pandemic on transplant activity in other contexts, long-term prognosis, and effect of COVID-19 in pre- or post-HSCT patients. In conclusion, the findings of our study show that transplant activity was not decreased, and outcome of HSCT was not influenced, in the COVID-19 pandemic year compared with that of the two years prior. Acknowledgements The authors thank all the physicians and data managers of the centers who contributed to the collection of data on transplantation to the Japanese Data Center for Hematopoietic Cell Transplantation and the Second-Generation Transplant Registry Unified Management Program. We would like to express our gratitude to the Japan Society of Clinical Research for their support. Author contributions YS: designed the study, analyzed the data, performed the statistical analysis, and wrote the first draft of the manuscript. MN, TK, and TS: contributed to the critical review of the data analysis and the manuscript. All other authors contributed to data collection. All authors approved the final version of the manuscript. Funding None. Data availability The data used in this study are available from the corresponding author upon reasonable request. Declarations Conflict of interest The authors have no conflicts of interest to declare. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Banerjee A Pasea L Harris S Gonzalez-Izquierdo A Torralbo A Shallcross L Estimating excess 1-year mortality associated with the COVID-19 pandemic according to underlying conditions and age: a population-based cohort study Lancet 2020 395 1715 1725 10.1016/S0140-6736(20)30854-0 32405103 2. Ritchie H, Mathieu E, Rodés-Guirao L, Appel C, Giattino C, Ortiz-Ospina E, et al. (2020) Coronavirus Pandemic (COVID-19). Our World Data. https://ourworldindata.org/coronavirus. Accessed 15 July 2022 3. Coronavirus disease (COVID-19) pandemic. https://www.who.int/europe/emergencies/situations/covid-19. Accessed 15 July 2022 4. Patt D Gordan L Diaz M Okon T Grady L Harmison M Impact of COVID-19 on cancer care: how the pandemic is delaying cancer diagnosis and treatment for American seniors JCO Clin Cancer Inform 2020 10.1200/CCI.20.00134 5. Kuderer NM Choueiri TK Shah DP Shyr Y Rubinstein SM Rivera DR Clinical impact of COVID-19 on patients with cancer (CCC19): a cohort study Lancet 2020 395 1907 1918 10.1016/S0140-6736(20)31187-9 32473681 6. Lee LYW Cazier J-B Starkey T Briggs SEW Arnold R Bisht V COVID-19 prevalence and mortality in patients with cancer and the effect of primary tumour subtype and patient demographics: a prospective cohort study Lancet Oncol 2020 21 1309 1316 10.1016/S1470-2045(20)30442-3 32853557 7. Mafham MM Spata E Goldacre R Gair D Curnow P Bray M COVID-19 pandemic and admission rates for and management of acute coronary syndromes in England Lancet 2020 396 381 389 10.1016/S0140-6736(20)31356-8 32679111 8. Ranganathan P Sengar M Chinnaswamy G Agrawal G Arumugham R Bhatt R Impact of COVID-19 on cancer care in India: a cohort study Lancet Oncol 2021 22 970 976 10.1016/S1470-2045(21)00240-0 34051879 9. Grivas P Khaki AR Wise-Draper TM French B Hennessy C Hsu C-Y Association of clinical factors and recent anticancer therapy with COVID-19 severity among patients with cancer: a report from the COVID-19 and cancer consortium Ann Oncol 2021 32 787 800 10.1016/j.annonc.2021.02.024 33746047 10. Takashima J Kobayashi H Suzuki Y Koizumi A Shigehara F Yamazaki K Evaluating the burden of the COVID-19 pandemic on patients with colorectal cancer Oncol Lett 2022 24 263 10.3892/ol.2022.13383 35765278 11. Saeki H Shirabe K Miyazaki T Ogawa T Makita F Shitara Y Decreased numbers of gastric, colorectal, lung, and breast cancer surgeries performed in 17 cancer-designated hospitals in Gunma Prefecture of Japan during the COVID-19 pandemic Surg Today 2022 10.1007/s00595-022-02501-y 12. Terashima T Tsutsumi A Iwami E Kuroda A Nakajima T Eguchi K Delayed visit and treatment of lung cancer during the coronavirus disease 2019 pandemic in Japan: a retrospective study J Int Med Res 2022 50 030006052210973 10.1177/03000605221097375 13. Yamaji K Kohsaka S Inohara T Numasawa Y Ando H Wada H Percutaneous coronary intervention during the COVID-19 pandemic in Japan: Insights from the nationwide registration data Lancet Reg Health West Pac 2022 22 100434 10.1016/j.lanwpc.2022.100434 35330940 14. Bair SM Brandstadter JD Ayers EC Stadtmauer EA Hematopoietic stem cell transplantation for blood cancers in the era of precision medicine and immunotherapy Cancer 2020 126 1837 1855 10.1002/cncr.32659 32073653 15. Passamonti F Cattaneo C Arcaini L Bruna R Cavo M Merli F Clinical characteristics and risk factors associated with COVID-19 severity in patients with haematological malignancies in Italy: a retrospective, multicentre, cohort study Lancet Haematol 2020 7 e737 e745 10.1016/S2352-3026(20)30251-9 32798473 16. Ljungman P de la Camara R Mikulska M Tridello G Aguado B Zahrani MA COVID-19 and stem cell transplantation; results from an EBMT and GETH multicenter prospective survey Leukemia 2021 35 2885 2894 10.1038/s41375-021-01302-5 34079042 17. Passweg JR Labopin M Cornelissen J Volin L Socié G Huynh A Conditioning intensity in middle-aged patients with AML in first CR: no advantage for myeloablative regimens irrespective of the risk group–an observational analysis by the Acute Leukemia Working Party of the EBMT Bone Marrow Transplant 2015 50 1063 1068 10.1038/bmt.2015.121 26030052 18. Wolf J Lee J Pearce R Wilson M Snowden JA Orchard K The impact of COVID-19 on related-donor allogeneic stem cell harvest processes: a British Society of Blood and Marrow Transplantation and Cellular Therapy survey Br J Haematol 2022 10.1111/bjh.18299 19. Atsuta Y Introduction of Transplant Registry Unified Management Program 2 (TRUMP2): scripts for TRUMP data analyses, part I (variables other than HLA-related data) Int J Hematol 2016 103 3 10 10.1007/s12185-015-1894-x 26547570 20. Sorror ML Maris MB Storb R Baron F Sandmaier BM Maloney DG Hematopoietic cell transplantation (HCT)-specific comorbidity index: a new tool for risk assessment before allogeneic HCT Blood 2005 106 2912 2919 10.1182/blood-2005-05-2004 15994282 21. Giralt S Ballen K Rizzo D Bacigalupo A Horowitz M Pasquini M Reduced-intensity conditioning regimen workshop: defining the dose spectrum. Report of a workshop convened by the center for international blood and marrow transplant research Biol Blood Marrow Transplant J Am Soc Blood Marrow Transplant 2009 15 367 369 10.1016/j.bbmt.2008.12.497 22. Bacigalupo A Ballen K Rizzo D Giralt S Lazarus H Ho V Defining the intensity of conditioning regimens: working definitions Biol Blood Marrow Transplant J Am Soc Blood Marrow Transplant 2009 15 1628 1633 10.1016/j.bbmt.2009.07.004 23. Kinoshita R Jung S Kobayashi T Akhmetzhanov AR Nishiura H Epidemiology of coronavirus disease 2019 (COVID-19) in Japan during the first and second waves Math Biosci Eng 2022 19 6088 6101 10.3934/mbe.2022284 35603392 24. Yamamoto H Single cord blood transplantation in Japan; expanding the possibilities of CBT Int J Hematol 2019 110 39 49 10.1007/s12185-019-02672-4 31152417 25. Ballen KK Lazarus H Cord blood transplant for acute myeloid leukaemia Br J Haematol 2016 173 25 36 10.1111/bjh.13926 26766286 26. Kanda Y Inoue M Uchida N Onishi Y Kamata R Kotaki M Cryopreservation of unrelated hematopoietic stem cells from a blood and marrow donor bank during the COVID-19 pandemic: a nationwide survey by the Japan marrow donor program Transplant Cell Ther 2021 27 664.e1 664.e6 10.1016/j.jtct.2021.04.022 27. Kanda Y Doki N Kojima M Kako S Inoue M Uchida N Effect of cryopreservation in unrelated bone marrow and peripheral blood stem cell transplantation in the rra of the COVID-19 pandemic: an update from the Japan marrow donor program Transplant Cell Ther 2022 28 677.e1 677.e6 10.1016/j.jtct.2022.06.022 28. Morais S Antunes L Rodrigues J Fontes F Bento MJ Lunet N The impact of the COVID -19 pandemic on the short-term survival of patients with cancer in Northern Portugal Int J Cancer 2021 149 287 296 10.1002/ijc.33532 33634852 29. Martinez-Lopez J Hernandez-Ibarburu G Alonso R Sanchez-Pina JM Zamanillo I Lopez-Muñoz N Impact of COVID-19 in patients with multiple myeloma based on a global data network Blood Cancer J 2021 11 198 10.1038/s41408-021-00588-z 34893583

36515796

PMC9749640

NO-CC CODE

2022-12-15 23:23:21

no

Int J Hematol. 2022 Dec 14;:1-8

utf-8

Int J Hematol

10.1007/s12185-022-03508-4

oa_other

==== Front Ann Oper Res Ann Oper Res Annals of Operations Research 0254-5330 1572-9338 Springer US New York 5136 10.1007/s10479-022-05136-x Original Research Big data applications with theoretical models and social media in financial management Saito Taiga staiga@e.u-tokyo.ac.jp 1 http://orcid.org/0000-0002-2714-4958 Gupta Shivam shivam.gupta@neoma-bs.fr 2 1 grid.26999.3d 0000 0001 2151 536X Graduate School of Economics, The University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo, 1130033 Japan 2 grid.462778.8 0000 0001 0721 566X Department of Information Systems, Supply Chain Management & Decision Support, NEOMA Business School, 59 Rue Pierre Taittinger, 51100 Reims, France 14 12 2022 123 7 12 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This study presents big data applications with quantitative theoretical models in financial management and investigates possible incorporation of social media factors into the models. Specifically, we examine three models, a revenue management model, an interest rate model with market sentiments, and a high-frequency trading equity market model, and consider possible extensions of those models to include social media. Since social media plays a substantial role in promoting products and services, engaging with customers, and sharing sentiments among market participants, it is important to include social media factors in the stochastic optimization models for financial management. Moreover, we compare the three models from a qualitative and quantitative point of view and provide managerial implications on how these models are synthetically used along with social media in financial management with a concrete case of a hotel REIT. The contribution of this research is that we investigate the possible incorporation of social media factors into the three models whose objectives are revenue management and debt and equity financing, essential areas in financial management, which helps to estimate the effect and the impact of social media quantitatively if internal data necessary for parameter estimation are available, and provide managerial implications for the synthetic use of the three models from a higher viewpoint. The numerical experiment along with the proposition indicates that the model can be used in the revenue management of hotels, and by improving the social media factor, the hotel can work on maximizing its sales. Keywords Big data applications Revenue management Social media Financial management ==== Body pmcIntroduction Social media is important for business since social media includes valuable information that the other data that companies own normally do not contain. From social media such as Twitter, Instagram, and Facebook, companies can collect customers’ impressions and reviews and even make advertisements to the customers. Also, people’s views on financial markets and economics are observed on social media. The revenue management model by Saito et al. (2019, 2016), the interest rate model with sentiments by Nishimura et al. (2019) and Nakatani et al. (2020), and the high-frequency trading equity market model by Saito and Takahashi (2019), are quantitative theoretical models that express behaviors of customers, investors, and market participants. Moreover, these models incorporate big data, such as booking data, financial data with text data, and order and execution data into modeling. Since social media includes information that these data do not have, thus, the models become more valuable if they take into account the social media data that reflects people’s sentiments. In this study, we review these three theoretical models from an integrated perspective and consider possible extensions of the models to include social media factors. Also, we compare the three models to investigate what common features are and what the differences are by considering managerial implications about in which scenario each model works. As big data are used in our daily lives, it is becoming increasingly important to utilize big data and theoretical models that describe mechanisms of phenomenon. Kar and Dwivedi (2020) also point out the need for studies of theoretical modeling incorporating big data, which explains phenomena caused by the interaction of people. For example, the customers on the online hotel booking website in Saito et al. (2016, 2019) choose the hotel to book from a group of hotels in the same area by comparing the room charges and other characteristics of the hotels. In the interest rate model in Nishimura et al. (2019) and Nakatani et al. (2020), the investors maximize the expected utility of their wealth by choosing portfolio allocations under sentiments. In the high-frequency trading equity market model in Saito and Takahashi (2019), the market participants in a high-frequency trading market choose their trading strategies to maximize their expected profits. Those quantitative theoretical models describe human behaviors in social circumstances and explain patterns of the outcomes observed in reality. With applications of big data, the models become remarkably useful such that they can be utilized for determining optimal strategies in financial management. In detail, Saito et al. (2016, 2019) presented big data applications to revenue management for hotels, utilizing big data, which is online booking data collected from a hotel booking website. With the big data, they estimated the quantitative revenue management models that described booking behaviors of the customers in response to the room charges of hotels and showed how those models could be used in hotel revenue management. Nishimura et al. (2019) and Nakatani et al. (2020) proposed applications of big data, which is financial news, by text mining and estimated market sentiment factors in interest rate models. Nishimura et al. (2019) and Nakatani et al. (2020) assessed the sentiment factors by text mining in economic news, which is big data, and proposed interest rate models incorporating those factors, which can be used by financial managers in companies as well as financial institutions. In this study, we consider what further information helps describe the phenomena in detail and how it should be incorporated into the modeling using additional big data. Social media are essential in the utilization of big data since they contain information from customers, which can be utilized in marketing, and information on the market participants’ sentiments. We investigate how social media data can be added to the three quantitative theoretical models with big data for practical use. Accordingly, we set the first research question as (i) How can we add social media factors into the quantitative theoretical models incorporating big data? In the revenue management model in Saito et al. (2016, 2019), the importance of social media was indicated, but the social media factor was not incorporated into the modeling. Data from social media is helpful since it reflects the customers’ sentiments against the hotels for instance, which affect the customers’ choice behavior. We incorporate the social media factor into the revenue management model in Saito et al. (2016, 2019), which could be useful for revenue managers in hotels to maximize their expected sales. Also, sentiments of the investors on the financial market, which are also observed on social media, are important to the interest rate model with sentiments in Nishimura et al. (2019) and Nakatani et al. (2020) and the high-frequency trading equity market model in Saito and Takahashi (2019), which could be helpful for financial managers companies to predict when the best timing for financing with corporate bonds and equities, for central banks to control term structures of interest rates, and for financial authorities to regulate the market. We consider possible utilization of the social media information into the interest rate model and the high-frequency trading equity market model. Moreover, we investigate the three applications of big data with the quantitative theoretical optimization models in financial management from an integrated point of view. Particularly, we compare these three models and consider how these three models are utilized in financial management in different roles and scenarios. Thus, we set the second research question as (ii) How are those three theoretical models different from an integrated point of view and how can a company synthetically use the three models depending on a scenario? This study is new in discussing the applications of big data with theoretical models in financial management from an integrated point of view and their extensions to include social media from the stakeholders’ viewpoints. The contribution is the extension of the three quantitative theoretical models incorporating big data to include further social media factors, a comparison of the three models from an integrated point of view, and the managerial implication for the synthetic use of the models by companies. Also, this study connects social media with the quantitative theoretical models and provides how those models are utilized in financial management for companies, investors, and financial authorities. The organization of the paper is as follows. Section 2 proposes an extension of the revenue management model to include social media factors. Sections 3 and 4 investigate the interest rate model with market sentiments and the high-frequency trading equity market model by discussing possible extensions to include social media information. Section 5 compares the three theoretical models, and Sect. 6 investigates the managerial implications of the models and discusses the limitation of this study. Finally, Sect. 7 concludes. Revenue management model In this section, as an application of big data with a theoretical model in financial management, we extend the revenue management model for hotels in Saito et al. (2016, 2019), whose mathematical details are explained in “Appendix A.1”, to include social media information. In the hotel industry, quantitative revenue management models are used for day-to-day revenue control in online booking. The customers booking behavior is estimated, and the hotel revenue manager sets a price for booking to maximize the profit. By effectively utilizing the revenue management model, hotels can maximize their revenue by setting the room charge appropriately. Specifically, we propose a quantitative revenue management model for hotels incorporating social media factors estimated from internal information such as reviews on online booking websites, Twitter, and Facebook, which is an extension of the revenue management model in Saito et al. (2019). Since social media has valuable information for marketing and enhancement for the services of hotels, hotels can better use social media so that it affects the probability of booking more effectively. Data Along with the model described in “Appendix A.1”, Saito et al. (2016, 2019) use the online hotel booking data crawled from a Japanese hotel online booking website for the estimation of the parameters. In Saito et al. (2016, 2019), the collected data includes room charges and the number of available rooms for booking for a certain booking period for collective check-in dates for four major hotels in front of Kyoto station and two major luxurious hotels in the Shinjuku area, respectively. In Eqs. (A1)–(A3) in “Appendix A.1”, the big data that include the room charges and numbers of rooms booked, and the availability for booking of hotels in the same area are used for estimation. Particularly, the data are used in estimating the model for the sensitivity of the hotel’s score β and the characteristic part αi in equation (A3) through the expression of booking probabilities piγ in (A2). Specifically, in Saito et al. (2016), optimal room charges for the four hotels in front of Kyoto station are calculated, particularly the price competition among the four hotels when each hotel chooses its optimal room charge given the room charges of the other hotels is investigated. Extension to include social media information Although publicly available big data, online booking data crawled from a Japanese online booking website, are used in Saito et al. (2016, 2019) utilization of internally available information or information from social media, such as reviews of the hotels, Twitter, and Facebook, which reflect sentiments of the customers, is not considered. In this subsection, we extend the model to include the social media effect to Vi, the score of hotel i. In the constant term αi in Vi in (A3), which represents static characteristics of hotel i, some social media information of hotel i, such as the reputation of the hotel, is reflected. Thus, we decompose αi as1 αi=α¯i+νvi,ν>0, where vi∈R represents the social media factor of hotel i and ν expresses the sensitivity of the factor on αi, the static characteristics of hotel i. We remark that vi can take not only a positive value but also a negative value, which affects positively (when it is positive)/negatively (when it is negative) the hotel i’s score Vi compared with the case where αi=α¯i. This implies that if we identify the social factor vi and estimate the sensitivity of the social media factor ν in αi in (1), we can observe how the expected sales change if the social media factor vi shifts and how the optimal room charge and overbooking level vary. The social media factor vi, which is a real value, positively affects the booking probability piγ and the expected sales of the hotel in (A4) through Vi, the score of the hotel i, in (A3), and piγ, the booking probability, in (A2) if it increases. This is an extension of the static score inherent to hotel i, αi, to incorporate the time-varying effect on αi, the characteristic of the hotel, by social media. One way to identify the social media factor vi and estimate the sensitivity ν incorporated in αi is by utilizing social media data, such as frequencies of some specific words in online reviews or other social media, along with detailed results of the hotel’s revenue from online booking, which the hotel owns as private information. In detail, a hotel may collect and utilize customer reviews in TripAdvisor, where perceived review credibility, review usefulness, and ease of use predict customer satisfaction (Filieri et al., 2020), such as the scores rated by the customers and words in the customer reviews that affected the scores. (For other researches on online reviews, see Saumya et al. (2019) and Ismagilova et al. (2020) for instance). Particularly, a hotel may identify influential reviews from loyal customers who repeatedly book the hotel by those methods and utilize the reviews together with privately available detailed customer information. Moreover, other social media components, such as frequencies of positive or negative words in social media such as Twitter, Facebook, and Instagram, could be used. As output data, detailed information on sales through online bookings, attributes of customers who booked, and the hitting ratio in the booking website will help estimate the effect of the social media factor more precisely. Impact of social media factor changes In this subsection, using the estimation result for the model in Saito et al. (2019) and assuming the sensitivity of the social media factor ν in (1), we observe how the expected sales, optimal room charge, and overbooking level change, if the social media factor of hotel shifts. With the model parameters in Saito et al. (2019), whose attributes are described in “Appendix A.1”, assuming the social media factor, we consider the maximization of the expected sales with respect to the optimal room charge and the overbooking level. In short, the hotel aims to set the room charge and the overbooking level optimally to maximize its expected revenue, where there are trade-offs such that if the hotel sets the room charge high, the booking probability decreases, and if the hotel accepts more overbooking for last-minute cancellations, the hotel needs to repay the cost to decline the overbooked customers when the cancellation is less than expected. Specifically, we use the parameters originally estimated from the data collected from a Japanese online booking website for check-in dates ranging from March 1st to April 30th 2017 for standard nonsmoking twin rooms of two major luxurious hotels in Shinjuku area in Tokyo, Japan. We assume the two hotels’ case L=2 and name the hotels as hotel 1 and hotel 2. The parameters are as follows: T=14, λ=2.1429, μ=-8.161672, σ=1.053053, α¯1=0, α2=-2.094104, δ2=0.849365, q1=20, and x2=42,292. We remark that α2 was originally estimated with α1=0 in the estimation in Saito et al. (2019) due to the degree of freedom of the parameters since only the difference α2-α1 affects the booking probabilities in (A2). Then, we can calculate the expected sales, the optimal room charge, and the optimal overbooking. Particularly, the social media factor positively affects the expected sales when it increases. Here, we assume the high cancellation rate with the distribution P(r1H=100%)=P(r1H=50%)=P(r1H=30%)=13, the high over-sale cost c1H=100,000 for hotel 1, and the sensitivity of the social media factor ν=1.0. Firstly, the following proposition holds. Proposition 1 The booking probability of hotel i by a customer, piγ in (A2), increases for any availability of the hotels, γ, when the social media factor of hotel i, vi, increases. Also, the optimal expected sales max0<xi<∞E[xiRTi] in the expected sales maximization of E[xiRTi] in (A4) increases as the social media factor vi increases. Proof Since by (1), the hotel i’s characteristic factor αi is increasing with respect to the social media factor vi. Then, the hotel i’s score Vi in (A3) is increasing and thus the booking probability by a random customer piγ in (A2) is also increasing with respect to vi. Since the booking probability by a customer piγ increases in any availability of the hotels γ, the expected number of rooms booked E[RTi] increases with respect to vi for any 0<xi<∞, thus the expected revenue E[xiRT] also increases with respect to vi, for any 0<xi<∞. Therefore, the maximized value of E[xiRT] with respect to 0<xi<∞ is also increasing with respect to the social media factor vi. In the overbooking case, the maximization of the expected revenue in the optimal overbooking problem E[min(qi,(1-ri)RTi)x(i)-max((1-ri)RTi-qi,0)ci] in (A5), when the number of rooms booked RTi increases as the social media factor vi increases, the total oversale cost max((1-ri)RTi-qi,0)ci also increases. Thus, when the overbooking level Li(ob) is large, where the hotel accepts exceedingly more customers than the actual capacity, if the number of rooms booked RTi becomes large as the social media factor vi increases, the total revenue decreases due to the effect of the over-sale cost max((1-ri)RTi-qi,0)ci. In such a case, an increase in the social media factor vi does not necessarily mean an increase in the expected revenue for fixed overbooking level and room charge. However, in the optimization, the hotel i can set the overbooking level Li(ob) lower so that the over-sale cost effect does not exceed the revenue increase. □ Remark 1 The proposition indicates that the way the social media factor is incorporated into the revenue management model is plausible since the increase in the social media factor results in a rise in the booking probability and the expected sales of the hotel, which is compliant with the practical situation. Thus, if internal data necessary for parameter estimation are available, and the model is estimated, the quantitative model with the social media factor can be used to measure the sensitivity and the effect of the factor on the booking probability and expected sales quantitatively, which is the novelty of this proposition. As a managerial implication, hotels can first work on identifying what effort could increase the social media factor, then both the social media marketing team and the revenue management team cooperate to increase the revenue. Specifically, the proposition implies that the social media marketing team first works on identifying the social media factor and increasing the social media factor effect, and then the revenue management team conducts revenue maximization by optimally setting the room charge. Remark 2 The proposition partially answers the research questions. For the first research question, by showing that a social media factor is suitably incorporated in the revenue management model, this proposition indicates that the social media factor in the model can be estimated from booking probability, and with the obtained estimation result, the company can work on revenue management to maximize its sales by using the revenue management model. For the second research question, the proposition suggests a managerial implication that the hotels can set up a social media marketing team in the financial department, which works on the social media advertising strategy that can increase the booking probability of the hotel and monitors social media impact in the financial market. By setting up the team, the finance department could conduct the revenue management and financing strategies with debt and equity better in an integrated way. In this numerical example, the optimal overbooking level and room charge are solved, and we observe that the optimal expected revenue also increases as the social media factor increases. In this example, the optimal expected sales of hotel 1 is JPY 447, 155 when the overbooking level and the room charge are (L1(ob),x1)=(29,43,000) in the case of v1=1.0, while it is JPY 428, 082 when (L1(ob),x1)=(29,42,000) in the case of v1=0. Therefore, the expected revenue increases as the social media factor increase from v1=0 to v1=1.0 in this case. Figures  and  describe the expected sales of hotel 1 when the optimal room charge and the overbooking level vary, when the social media factor v1 is 0 and 1.0, respectively.Fig. 1 The expected sales in the case of the high cancellation rate and the high over-sale cost per room when v1=0. The maximized expected sales is JPY 428, 082 when (L1(ob),x1)=(29,42,000) Fig. 2 The expected sales in the case of the high cancellation rate and the high over-sale cost per room when v1=1.0. The maximized expected sales is JPY 447, 155 when (L1(ob),x1)=(29,43,000) Moreover, Fig.  shows the change in the expected sales when the social media factor v1 is increased from 0 to 1.0 for fixed overbooking level L1(ob)=29, which is optimal in both cases. We observe that by increasing the social media factor v1 from 0 to 1.0, the graph of the expected sales curve shifts to the right above, and the optimal room charge changes from 42,000 to 43,000, where the corresponding expected sales increase from JPY 428,082 to JPY 447,155.Fig. 3 The expected sales in the case of the high cancellation rate and the high over-sale cost per room when the room charge varies for v1=0 and 1.0 Interest rate model with market sentiments Next, as the second example of big data applications with models in financial management, we introduce an interest rate model with sentiment factors in Nishimura et al. (2019) and Nakatani et al. (2020). (For other studies on applications of interest rate models, see Menkveld et al. (2000), Morelli (2021), and Bali (2007), for instance.) After the global financial crisis, we have observed global monetary easing and the resultant low-interest-rate environment. In such an environment, market sentiments mainly affect asset prices, particularly in the government bond markets, and thus it is essential to incorporate the sentiment factors in financial modeling. In this section, we review the model in Nishimura et al. (2019) and Nakatani et al. (2020), whose mathematical details are explained in “Appendix A.2”, as an example in which the sentiment factors are estimated with big data, which are financial news, by text mining. The model exhibits impacts of the sentiments on the interest rate, which can be used by central banks that aim to control the market better to improve the economy. The model can also be useful for large traders, such as institutional investors and hedge funds, who invest money from pension funds in the financial markets. Moreover, we propose an extension of the model in Nishimura et al. (2019) and Nakatani et al. (2020), in which the sentiment factors are estimated not only by financial news but also by social media information. Data In Nishimura et al. (2019) and Nakatani et al. (2020), the three-factor model in “Appendix A.2” is estimated by a stochastic filtering method, in which the frequencies of words closely related to the steepening (pessimistic) or the flattening (optimistic) factor are used in the observation equations in the filtering. In detail, those words relevant to the steepening and flattening factors are specified with financial news text data from Reuter. The estimated three-factor interest model helps predict how the sentiment-related words affect the yield curve shape, which can be used in trading by hedge funds or the yield curve control by central banks. Possible extension to include social media information and estimation We extend the observation equations in the stochastic filtering in Nishimura et al. (2019) and Nakatani et al. (2020) to include social media information. In detail, we find words closely correlated with the steepening or flattening factor and incorporate the frequencies of the words in the observation equations. The estimated three-factor interest rate model implies how social media information, in addition to financial news, affects the yield curve shape. For instance, Grover et al. (2019) investigated social media impacts on voting behavior during an election through acculturation of ideologies and polarization of voter preferences. As social media information, we may consider specific words closely related to the steepening or flattening factor in Twitter for political or economic events, for example. Firstly, we assume the stochastic dynamics for x1, x2 and x3 in the system equations (A7). In Nishimura et al. (2019) and Nakatani et al. (2020), along with the observation equations for the bond yields and frequencies of specified words by text mining in financial news, the parameters in the system equations are estimated. Let Yt(n) be the yield of the n-year bond at time t and F(Ai),i=1,…,IA,F(Bi),i=1,…,IB, be the frequencies of the steepening related words and the flattening related words, respectively, specified by text mining in financial news. Then, the discretized system equations and the observation equations are as follows. System equations (discrete):2 xj,t=e-κjPΔtxj,t-Δt+σj21-e-2κjPΔt2κjPϵj,t,j=1,2,x3,t=x3,t-Δt+σ3Δtϵ3,t, where κjP=κjQ-σx,jσc,j, λ3Q=-σx,3σc,3, Δt=1250, ϵj,t∼i.i.d.N(0,1). Observation equations:3 Yt(10)-Yt(2)=∑l=13{Xl,t(10)-Xl,t(2)}+et,10-2y,Yt(20)-Yt(10)=∑l=13{Xl,t(20)-Xl,t(10)}+et,20-10y,Yt(20)=∑l=13Xl,t(20)+et,20y,Yt(30)=∑l=13Xl,t(30)+et,30y, 4 log∑i=1IAF(Ai)+1=ξ1,c+ξ1x1,t2+et,w1,log∑i=1IBF(Bi)+1=ξ2,c+ξ2x2,t2+et,w2, where Xj,j=1,2,3 are defined as in (A11), et,j∼i.i.d.N(0,γj2), j=10-2y, 20-10y, 20y, 30y, w1,w2. This indicates that as observable data in the observation equations in (3) and (4), we use the bond yield spreads for 10 year–2 year, 20 year–10 year, the yields of 20 year and 30 year bonds, frequencies of the steepening related words and the flattening related words and estimate the system equation parameters by the Monte Carlo filtering method. Here, et,j are called the observation noise, and ϵj is the system noise. For details of the Monte Carlo filtering method, see Nakatani et al. (2020). Since the investors monitor social media, such as online polls and Twitter, to observe the views of other investors and announcements from governors, social media affect the sentiments and views of investors. Thus, it is important to incorporate the effect of social media in addition to the financial news into the modeling. In addition to the observation equations, we further include5 log∑i=1ICF(Ci)+1=η1,c+η1x1,t2+et,s1,log∑i=1IDF(Di)+1=η2,c+η2x2,t2+et,s2, in the observation equations. Here, Ci,i=1,⋯,IC, and Di,i=1,⋯,ID, stand for the frequencies of the steepening related words and the flattening related words from social media, respectively, and et,j∼i.i.d.N(0,γj2),j=s1,s2. High-frequency trading equity market model Furthermore, as the third model, we review the high-frequency trading equity market model in Saito and Takahashi (2019), whose mathematical details are explained in “Appendix A.3”, and consider the possible estimation with big data. Saito and Takahashi (2019) analyzed how the parameter shifts affect the stock price movement and equilibrium trading strategies of the three types of players with the model described in “Appendix A.3”, which expresses the interactions among the trading of three types of players. Although the estimation of the parameters with data has not been done in Saito and Takahashi (2019), if the financial authorities or stock exchanges utilize their internal data, we can observe how changes in regulations affect the behaviors of the participants and the asset price movements. Specifically, the following internally available big data for financial authorities investigated in Saito et al. (2018) for Tokyo Stock Exchange, for example, could be used. To observe the transaction information, one needs high-frequency trading data, such as the trader ID, the order amount, the price, and the type of the order (the market order, the limit order, or the cancellation). If one has such data, one first analyzes the transaction data and classifies the traders into types by their trading patterns. Then, one estimates the parameters in the stochastic differential equations of the model by a stochastic filtering method and will be able to use the model to investigate the impacts of regulatory changes. Incorporation of social media in estimation Moreover, market sentiment-related words in social media could also be used in the estimation. Not only financial news but also views of the investors and announcements of politicians on social media can change sentiments in the market, which could affect the trading behaviors of the market participants. Particularly, since social media can affect elections and government policies, for example, monitoring social media and incorporating the sentiments are also important. By incorporating social media in estimating the model, we may estimate how the sentiments or announcements observed in social media affect the asset price. Comparison of the three theoretical models In this section, we compare the three quantitative theoretical models. The three models are common in that they are stochastic models for optimization, which utilize big data. In detail, the revenue management model maximizes the expected revenue of hotels, the high-frequency trading equity market model deals with the maximization of the expected revenue of the different types of traders, and the interest rate model is an equilibrium of agents who solve an optimal investment problem with sentiments, where the big data are online booking data, which include room charge and sales, interest data with sentiment-related words, and order and execution data of market participants, respectively. These three models are described and compared from a quantitative perspective as follows. Firstly, the hotel revenue management model is for optimization with a random choice model, the interest rate model is a model whose parameters are estimated with stochastic filtering, and the high-frequency market model is a stochastic differential game. Secondly, these models are common in that they are stochastic models, which aim to capture the random choice behavior of customers, interest rate movement, and trading behaviors of players in the high-frequency stock market. Specifically, the revenue management model deals with the score of each hotel as a function of the room charge and the hotel’s characteristic term, which includes the social media factor. The interest rate model expresses the short rate model with the steepening, flattening, and level factor, where the steepening and flattening factors could be estimated with financial news and social media-related information. The high-frequency stock market model deals with the stochastic trading behavior of the algorithmic trader, the market maker, and the general trader, and the stock price process where the expected return could be possibly estimated with a social media factor. Thus, they are common in dealing with optimization, maximization of expected revenue, minimization of error for estimation, and expected profit maximization. Moreover, the three models are qualitatively described as follows. The three models are used for financial management. The revenue management model is for the business department of hotels which aims to increase sales. The interest rate model and the high-frequency trading equity market model can be used in the finance department to plan the timing for debt financing and equity financing. Also, the three models describe human transactions in the environment with randomness. In the revenue management model, the customers choose a hotel to book depending on the room charge and the hotel-specific factor. In the interest rate model, the agents solve optimal investment problems, and the high-frequency trading equity market model deals with the investment activity of the algorithm traders, the market makers, and general traders. On the other hand, the three models are distinct in the following points. Firstly, they are different in usage. The revenue management model is for the maximization of the expected revenue. The interest rate model is for estimation of the model that describes the movement of the term structure of interest rates with sentiments, and the high-frequency trading equity market model solves for the equilibrium stock price under different types of traders. Also, the optimizing agents are dissimilar. In the revenue management model, optimization by a hotel revenue manager given the behavior of the customers is considered. In contrast, in the interest rate model and the high-frequency trading equity market model, the agents in the market maximize their objective functions. Furthermore, the big data types are different. The revenue management model utilizes online booking data, the interest rate model deals with interest rate data with the text news, and the high-frequency trading equity market model needs the order and execution data, respectively. Remark 3 We consider three models in financial management, particularly for the finance department of a company, which deals with revenue and expenditure from the business and access to the financial market for financing by issuing corporate bonds and stocks. These three areas are important in terms of revenue management, debt financing, and equity financing. Therefore, we use the three models from the important three areas, revenue management in business, financing by debt and equity, for the finance department. Also, the three quantitative models reflect the practical aspects of each market, the mechanism of online booking, the government bond market with different maturity and the financial news, and the interaction of market participants in a high-frequency market. In addition, we note that the basis of the three models is a mixed logit model for random choice behavior and its application to revenue management for hotels, a yield curve model with quadratic Gaussian factor processes, which reflect the steepening and the flattening effect of a yield curve, with stochastic filtering incorporating text mining, and a stochastic differential game that solves the Nash equilibrium of the strategies of three different types of players. Discussion Theoretical implications In this study, we have investigated three quantitative theoretical models for financial management utilizing big data and considered possible extensions to include social media factors. The contribution of this paper is the possible extensions of the quantitative modeling in financial management to include social media factors and the qualitative and quantitative comparison of the three models. Specifically, we have proposed the incorporation of the social media factor in a customer’s score on a hotel, the steepening and the flattening factor for the interest rate movement, and a stock price movement. There has been a vast of research on the utilization of big data and social media in management. Particularly, some research incorporates social media factors into quantitative modeling for optimization. For instance, Kumar et al. (2021a, 2021b) proposed a dynamic transmission model to investigate the impact of social media on the number of influenza and COVID-19 cases. Nilsang et al. (2019) investigated a model that considers real-time data from a social media application to minimize the response time and cost during emergencies and disasters. Zhu et al. (2021) developed the two-sided platform’s scalable decisions on when to cooperate and how to optimize the pricing and investment decisions. (For other utilization of big data in various aspects of information systems and information management, see Gupta et al. (2018, 2019), Kamboj and Gupta (2020), Kamboj et al. (2018), Modgil et al. (2021), Duan et al. (2019), Dwivedi et al. (2019), Kumar et al. (2021a, 2021b). For utilization of social media, see Giannakis et al. (2022), Rad et al. (2018), Grover et al. (2022), Wamba et al. (2019), Bogaert et al. (2018)). To the best of our knowledge, this study is the first attempt to investigate the comparison of the quantitative model using big data and social media in the field of financial management, which enables us to estimate the effect and the impact of social media quantitatively by introducing the social media factor as a new variable, if internal data necessary for parameter estimation are available, and consider the synthetic use of the three models in financial management with managerial implications from an integrated and higher viewpoint. Specifically, for the three models investigated in this study, Saito et al. (2019) worked on revenue management with the online booking data for two luxurious hotels in Shinjuku area in Tokyo, considering cancellation and overbooking strategies. Nakatani et al. (2020) estimated a yield curve model with the steepening and flattening sentiment factors using stochastic filtering with text mining for financial news. Saito and Takahashi (2019) considered a theoretical model that describes the different types of players in a high-frequency stock market where the trading by algorithmic traders, market makers, and general traders interact with each other and affect the stock price movement. Although these models implement practical aspects of financial management, the social media factor is not incorporated. Therefore, we have investigated possible social media extensions of the three models for revenue management, interest rate, and the stock market, which could be used in financial management in the measurement of the effect and the impact of social media quantitatively if internal data necessary for parameter estimation are available. Managerial implications Big data applications with theoretical models in financial management positively affect a wide range of stakeholders such as companies, investors, financial institutions, financial authorities, and the economy in the country as follows. Revenue management model The model incorporates the social media factor, and its increase affects the expected sales positively through the increase in the choice probability of the hotel. This revenue management model with social media has the following managerial implications. First of all, revenue managers in hotels, who set room charges of their hotel on online booking websites, can decide the room charges and take the overbooking strategy optimally so that the hotel can maximize its expected revenue. Hotels can analyze how social media affect the customers and utilize social media effectively by improving their services so that it can affect the booking from online customers positively. Moreover, as stakeholders, customers can benefit from the improved services and obtain valuable sales information through social media and book through online booking. Hotel investors can make the hotel introduce the revenue management model for better profitability, which leads to improvement of the investment performance. Also, in the revenue management model in “Appendix A.1”, hotel revenue managers can maximize the revenue of the hotel by optimally setting the room charge and managing the social media so that it can affect sales positively. Customers can enjoy the merit of the marketing efforts by the hotel and obtain valuable information through social media. Furthermore, the city and the companies in the area can earn revenue from tourism, where hotels make efforts to attract more visitors and transmit information through social media. Interest rate sentiment model The interest rate model with market sentiment could be estimated more precisely if we incorporate social media information in addition to the interest rate data and financial news. This interest rate model with sentiments and social media has the following managerial implications. Firstly, central banks can monitor how the words in financial news and social media affect the interest rate market through the sentiment model and effectively conduct monetary policies by making announcements strategically. Moreover, investors such as hedge funds and insurance companies trade effectively, monitoring how the words related to sentiments observed in the financial news and social media affect the bond prices. Furthermore, government liaises with central banks to make announcements, takes fiscal policies, and successfully controls the economy in the country. Secondly, in the interest rate model with sentiments in Sect. 3, investors such as institutional investors, hedge funds, and pension funds can trade, observing what financial news and words on social media would affect the yield curves. Moreover, central banks can effectively conduct monetary policies by monitoring the sentiment factors estimated from the financial news and social media in the model, which leads to a better economy in the country. High-frequency trading equity market model The high-frequency trading model incorporates the interactions among the different types of players. Moreover, with detailed internal transaction data and social media, the model becomes useful for financial authorities and large investors. If such internal data, all order and execution data with server IDs that the stock exchanges own and are shared with financial authorities by concluding a confidential agreement for research purposes, are available and the parameters are estimated with such data and social media as discussed in Sect. 4, by the estimated parameters in the model with the data that include the order and transaction data of algorithmic traders, market makers, and general traders, and data from social media, the market participants, which include the listed companies whose stocks are traded in the market, investors, and the financial authorities who regulate the high-frequency trading market, could predict how the stability of the current market by investigating the estimated parameters. In detail, this high-frequency trading market model has the following managerial implications. Firstly, the financial managers of a company, who needs to decide when to issue new stocks for equity financing, can predict how the algorithmic traders’ trading activities affect the price and predict the best timing for equity financing by conducting public offering for the new issuance. Secondly, financial authorities who need to regulate the rules in the financial markets can figure out how the trading of the algorithmic traders affects the stability of the market and set regulations so that the market becomes fair and stable. Stock exchanges can suitably maintain fees from all the participants for providing them a fair market, while they also satisfy the algorithmic traders by keeping them as good customers. Also, a central bank can consider the best timing for releasing the announcement, considering how the announcement affects the market. Moreover, although the data accessibility is limited compared to the financial authorities, institutional investors who care about price impacts caused by trading a large volume of stocks, with their internally available data, they can optimally execute their orders taking into account the algorithmic traders’ accelerations in the trades. Private investors can also predict when the market is unstable and about to crash and trade along with the large price moves. Furthermore, algorithmic traders can predict when the market becomes unstable by the parameters estimated with social media information and trade immediately when unexpected market news occurs. Furthermore, in the high-frequency trading market model in “Appendix A.3”, institutional investors can optimally execute large sizes of trades without causing price impacts. Private investors can follow the rapid price changes predicting the signals of instability. Additionally, financial authorities can set regulations for the stability of the market by observing what trading activities could affect the stability. Stock exchanges can set the trading fees for algorithmic traders properly to maintain a stable and fair trading environment for institutional and private investors. Synthetic use of the models by a company (Hotel REIT’s case) The finance department of a hotel conducts revenue management to increase sales and manage interest rate risk and the share price of the hotel for debt and equity financing. In addition to the hotel’s booking data, interest rate data, and stock market data, by incorporating social media factors, which affect the sentiment of the customers and the markets, the models become more useful in optimizing the whole business of the hotels. These three models can be used in the finance department of companies. For instance, hotel REITs (Real Estate Investment Trusts) can utilize the models to increase their financial results. A hotel REIT is a company that purchases hotel buildings and lends the buildings to hotels to earn rent fees as revenue. The REIT pays dividends from the revenue to REIT’s shareholders. The rent fee can be fixed or linked to the performance of the hotels. Then, the REIT can use the models as follows. Firstly, REITs make hotels that pay the performance-linked rent fee use the revenue management model to increase their sales, which results in an increase in the rent fee. Secondly, the REIT that issues corporate bonds for debt financing manages the risk of the interest rate hikes by the interest rate model with sentiments. Thirdly, the REIT, whose share is traded in the high-frequency trading equity market and subject to the trading of HFT traders, can increase its capital by public offering at the best timing by utilizing the high-frequency trading equity market model. After the outbreak of COVID-19, it has become more important for the hotel industry to conduct financial management optimally. According to Japan hotel REIT (2021), amid the COVID-19 pandemic, the domestic demand for hotels in Japan decreased, and the hotel management’s important metrics deteriorated. The hotel REIT introduced the variable rent fee scheme and also conducted third-party allocation of shares and debt financing with long maturities. As the COVID-19 situation calms down, it is expected that domestic and inbound demand will recover, and due to the changes in the market environment, hotels need to increase revenues by understanding customer needs. In such a situation, the hotel REIT can utilize the revenue management model to increase their revenue and predict long-term interest rates by the sentiment model and the best timing to conduct public offering with the high-frequency trading model. Limitations and future research direction There are some limitations to the extension of the revenue management model. Firstly, it is currently unable to collect new hotel booking data in this COVID-19 situation in which people are unable to travel. (For a collective insight on the impact of COVID-19 on information management research, see Dwivedi et al. (2020)). Secondly, even with the past booking data, as in Saito et al. (2019), it is still challenging to find a clear connection between the online reviews and the publicly available customers’ booking data since the data period is one month, which is relatively short compared with the gradual effect of the social media. To find a clear connection between social media and online booking, we need some additional internal information that hotels own, such as detailed sales data, attributes of the customers who booked rooms through the booking website, and answers to the questionnaires from customers. Thus, in this research, we limit ourselves to assuming parameters on the social media factors, conduct numerical experiments, and discuss possible methodologies to estimate the social media factors. Filieri et al. (2020) show that perceived review credibility is one of the most crucial determinants of travelers’ satisfaction and continued use of user-generated content (UGC) platforms. Particularly, Saumya et al. (2019) proposed a method to predict the most helpful online review, and Ismagilova et al. (2020) examined relations between emotions in the reviews and their perceived helpfulness. Identification of the social factor using influential reviews by those methods with internally available information and estimation for the sensitivity of the social media factor will be one of our future research topics. Concluding remarks In this study, we have investigated big data applications with theoretical models in financial management. Firstly, we have explored the hotel revenue management model, interest rate model with market sentiments, and high-frequency trading equity market model from an integrated perspective and discussed possible extensions to include social media factors. Moreover, in the extended revenue management model, we have conducted numerical experiments to observe how the revenue from online bookings changes and where the optimal room charges and overbooking levels are determined when the social media factor shifts. Finally, we have compared the models and discussed the managerial implications of those applications for stakeholders in financial management. Appendix A: Mathematical details of the models A.1 The revenue management model Firstly, we present the revenue management model in Saito et al. (2016, 2019). The model is described as follows. We assume that there are L hotels named hotels 1,⋯,L, in the same area with the same grade. We consider the revenue management model where customers, who visit an online booking website to book a room in the area, choose a hotel among those L hotels. We fix the room type that the customers are aiming to book and the check-in date. Let [0, T] be the booking period where time 0 is the first date of the checking period, and T is the last date of the check-in period, which is the same as the check-in date. Suppose that the customers visit the website at a frequency following a Poisson process {Nt}0≤t≤T with the intensity λ and choose a hotel to book among the L hotels. Hotel i, i=1,⋯,L has a limited capacity of qi rooms, but accept overbooking from the customers up to Li(ob) rooms, where qi≤Liob, for the last minute cancellation at the check-in date T. Moreover, let Rti(0≤Rti≤Li(ob)) be the number of rooms booked for hotel i by time t∈[0,T], where Ri,i=1,⋯,L satisfyA1 ∑i=1LRti=Nt. Furthermore, we suppose that a customer, who visits the website aiming to book a room among those L hotels at time t∈[0,T] at the random frequency following the Poisson process N, chooses hotel i, i=1,⋯,L with the probability pi given byA2 piγ=∫-∞∞exp(Vi)1{γi=1}∑j=1Lexp(Vj)1{γj=1}h(z)dz, where γ=(γ1,⋯,γL) stands for the availability of the hotels, in which γi=1 indicates that hotel i is available for booking (Rt-i<Li(ob), where t- indicates the time just before t), and γi=0 expresses that hotel i is fully booked (Rt-i=Li(ob)) and not available for booking, and each Vi expresses hotel i’s score, a linear combination of factors, namely, the room charge xi∈(0,∞), a holiday dummy variable y (y=0 if time t is a week day and y=1 if it is a day before a holiday), and a constant αi∈R that represents the static score inherent to hotel i asA3 Vi=αi-βxi+δiy,i=1,⋯,L,β>0,δi∈R,β=eμ+σz,h(z)=12πe-z22. Here -β<0 indicates that as the room charge xi increases, the score of hotel i decreases, and the hotel is less likely to be chosen by the customers. The equation (A3) describes that the coefficient follows a log-normal distribution which reflects the randomness of the sensitivity of the room charge for the customers. Then, Saito et al. (2016) consider the case where there is no overbooking, i.e. Li(ob)=qi,i=1,⋯,L is considered and the hotel i’s objective is set to maximize the expected revenue given byA4 E[xiRTi], by optimally choosing the room charge xi, where E[·] stands for the expectation of the random variable. Moreover, Saito et al. (2019) investigate an overbooking strategy for last minute cancellations, i.e., hotel i maximizesA5 E[min(qi,(1-ri)RTi)x(i)-max((1-ri)RTi-qi,0)ci], by optimally choosing the room charge xi and the overbooking level Li(ob), where the rooms are booked up to Li(ob)∈N rooms, instead of qi, qi is the actual capacity, ri∈[0,1] is the last minute cancellation ratio, which we assume to be a random variable independent of N and the hotel choices by the customers, namely, riRTi customers do not appear at the check-in data. (1-ri)RTi customers show up at the check-in date, however, if it exceeds the actual capacity qi, hotel i has to decline (1-ri)RTi-qi customers by compensating ci≥0 per room. In other words, this objective function describes that hotel i maximizes the expected revenue minus the overbooking cost by optimally setting the room charge and the overbooking level. A.2 The interest rate model with market sentiments Next, in this subsection, we introduce the interest rate model with market sentiments in Nishimura et al. (2019) and Nakatani et al. (2020). In the model, a steepening factor and a flattening factor for an interest rate curve, representing pessimistic and optimistic sentiment, respectively, are estimated by financial news text data along with interest rate data. The model is expressed as follows. Let {rt}0≤t<∞ be a short rate process expressed as a linear combination of three factors x12, x22, and x3,A6 rt=c1x1,t2+c2x2,t2+x3,t, where c1<0, c2>0, and xj,j=1,2,3, satisfy stochastic differential equations (SDEs)A7 dxj,t=-κjQxj,tdt+σx,jdBj,tQ,j=1,2,dx3,t=λ3Qdt+σx,3dB3,tQ, where κjQ>0,j=1,2,σx,j>0,j=1,2,3, and λ3Q∈R. Here, BQ is a three-dimensional Brownian motion under the risk neutral probability measure Q. Here, x12 and x22 represent the steepening factor and the flattening factor, respectively. c1x1,t2<0(c2x2,t2>0) affects the short rate rt negatively (positively), which fades away as time passes, since x1,t2(x2,t2) decreases to 0 due to the mean reversion of x1(x2) in SDE (A7). This implies that c1x12<0(c2x22>0) pushes down (up) the short rate rt first, but it fades away, which makes the yield curve shape at time 0 steepen (flatten), and thus we call it a bull-steepening (bull-flattening) effect. Thus, x12 and x22 control the slope of the yield curve, a collection of yields for bonds with different maturities. x12 and x22 also correspond to the pessimistic and the optimistic factor, respectively, since when the market is pessimistic (optimistic), the yields for the near future go lower (higher), which makes the upward sloping curve steepen (flatten). x3 is the level factor, a Gaussian process that controls the absolute level of the short rate r. Then, Pt(τ), the zero coupon bond price with time to maturity τ at time t and the zero coupon bond yield Yt(τ) for the time to maturity τ at time t are calculated asA8 Pt(τ)=EtQ[e-∫tt+τrudu] A9 =exp(-τ(X1,t(τ)+X2,t(τ)+X3,t(τ))), andA10 Yt(τ)=X1,t(τ)+X2,t(τ)+X3,t(τ), whereA11 Xj,t(τ)=-1τ[Aj(τ)+Cj(τ)xj,t2],j=1,2,X3,t(τ)=x3,t+λ3Q2τ-σ326τ2, withA12 Cj(τ)=C0,j+1zj(τ),Aj(τ)=σj22∫0τCj(s)ds, whereA13 C0,j=κjQ+(κjQ)2+cjσj2σj2,zj(τ)=σj2αj-eαjτ1C0,j+σj2αj,αj=2(κjQ-σj2C0,j),σj=2σx,j,j=1,2,σ3=σx,3. Here, EtQ[·] denotes the conditional expectation under Q with respect to the augmented filtration generated by BQ at time t. A.3 High-frequency trading equity market model Finally, as the third example of the application of big data along with theoretical models in financial management, we explain the high-frequency trading market model introduced by Saito and Takahashi (2019). The model describes the trading behaviors of three types of players and those interactions in high-frequency trading markets, where the players can trade in a millisecond interval, and the resultant price actions. This is important in the current financial markets, where algorithmic traders play a central role, and the price actions cause large economic effects. (For other studies on high-frequency trading markets, McGroarty et al. (2019) present an agent-based simulation for investigating algorithmic trading strategies, and Sun et al. (2014) propose a discrete optimization framework to describe how high-frequency trading can be utilized to supply liquidity and reduce execution cost, for example.) As a model to describe high-frequency trading markets, in which trading patterns of different types of players and the stock price dynamics are observed, Saito and Takahashi (2019) proposed a stochastic differential game model. Specifically, there are three types of players in the model, algorithmic traders, market makers, and general traders, who aim to maximize their objective functions, and equilibrium strategies of the three types of players, in which each type maximizes its objective given the others’ strategies, are obtained. The model indicates how the stock price moves, depending on the parameters in the model, which is useful in observing how the rapid price fall occurs and how the financial authorities should set appropriate regulations to prevent excessive price movements in the high-frequency trading environment. The model is described as follows. Firstly, suppose that there are three types of players, algorithmic traders (player 1), market makers (player 2), and general traders (player 3), who optimally trade the asset in the high-frequency trading market. Let [0, T] be the trading period, Xt0 be the mid-price of the asset at time t, and Xtj,j=1,2,3 be the positions of player j. Also, let αtjdt,j=1,2 be the units of the asset bought by player j in [t,t+dt] (if αtj<0, it indicates |αtj|dt units are sold by player j), and αt3 be the spread from the mid-price set by the market makers, player 3. Moreover, we assume the following dynamics for X0 and Xj,j=1,2,3, and objective functions for players 1,2,3. Let W be one-dimensional Brownian motion and χt=1{t≤t0} is an indicator function, which takes a value 1 until t=t0 and 0 thereafter. Here t0∈[0,T] is a time lag after which the general traders can respond and start trading.Mid-price process: A14 dXt0=(μ+(γ1αt1+γ2χtαt2+δαt3))dt+σtdWt,X00=x, where μ∈R,γ1,γ2,δ,σt>0. dXt0 stands for the change in the mid-price process of the asset in [t,t+dt). In addition to the drift term μdt and the diffusion term σtdWt, there is a market impact term (γ1αt1+γ2χtαt2+δαt3)dt caused by the trading of the three types of players. In detail, when the algorithmic traders or the general traders sell the asset (αt1<0 or αt2<0), it pushes down the mid-price by the amount proportional to the units of assets sold by them, and if the market makers set the spread from the mid-price negatively (αt3<0) to buy the amount sold by the algorithmic traders and the general traders, the price moves down by the amount proportional to the spread. The position of the algorithmic traders is A15 dXt1=αt1dt,X01=0, and the objective function of the algorithmic traders for maximization is A16 J1(α1,α2,α3)=E[-∫0Tαt1(Xt0+αt3+λ(αt1+χtαt2))dt+XT1XT0-12c1(XT1)2], where λ,c1>0. Here, dXt1 expresses the change in the position of the algorithmic traders (player 1), where X1 starts from 0, the flat position, at time 0. The algorithmic traders (player 1) start trading from time 0 and buy αt1dt units of the asset in [t,t+dt). The algorithmic traders can trade the asset at the price Xt0+αt3+λ(αt1+χtαt2), in which αt3 is the spread from the mid-price (e.g. αt3>0) set by the market makers (player 3), and λ(αt1+χtαt2) is the price slippage caused by the tradings from the algorithmic traders and the general traders (e.g. λ(αt1+χtαt2)>0, when αt1,αt2>0). Thus, -∫0Tαt1(Xt0+αt3+λ(αt1+χtαt2))dt indicates the cash paid for the trading in the period, XT1XT0 is the mark-to-market value of the asset position at time T, and -12c1(XT1)2 describes the liquidation cost paid for XT1 units of the asset at time T. Thus, the algorithmic traders aim to maximize J1(α1,α2,α3), the expectation of the mark-to-market value of the position at time T, by choosing α1 when α2,α3 are given. The position of the general traders is A17 dXt2=χtαt2dt,X02=x2>0, and the objective function of the general traders for maximization is A18 J2(α1,α2,α3)=E[-∫0Tχtαt2(Xt0+αt3+λ(αt1+χtαt2))dt-η2∫0Tχt(Xt2)2σt2dt+(XT2XT0-X02X00)-12c2(XT2)2], where η,c2>0. dXt2 describes the change in the position of the general traders (player 2), where X2 starts from a long position x2>0 at time 0. In contrast to the algorithmic traders, the general traders can start trading from time t0, the lag of the trading speed of the algorithmic traders. As in the algorithmic traders’ case, the general traders aim to maximize the trading profit with the risk aversion term -η2∫0Tχt(Xt2)2σt2dt, which implies that the general traders prefer to reduce the position size. The position of the market makers is A19 dXt3=(-(αt1+χtαt2)+kαt3)dt,X03=0, and the objective function of the market makers is A20 J3(α1,α2,α3)=E[-∫0T(-(αt1+χtαt2)+kαt3)(Xt0+αt3+λ(αt1+χtαt2))dt+XT3XT0-12c3(XT3)2], where k,c3>0. dXt3 represents the change in the position of the market makers (player 3). The market makers set the spread αt3 from the mid-price Xt0. For example, if αt1,αt2<0 and αt3<0, -(αt1+χtαt2)dt describes the amount the market makers accept against the selling orders from the algorithmic traders and the general traders. The market makers aim to buy at a lower level Xt0+αt3+λ(αt1+χtαt2), by setting the spread αt3<0. However, kαt3dt<0 indicates that if the market makers set the spread αt3<0 wide at a large negative value, they also miss the orders from the other two types by the amount proportional to the spread, where the fourth type of traders (others), who do not have any objective function, take the rest of the orders the market makers missed. Thus, the market makers aim to maximize the trading profit from the market making by setting the spread optimally. Funding This research was supported by CARF (Center for Advanced Research in Finance). Data Availability Data sharing is not applicable to this article as no datasets were generated or analyzed. Declarations Conflict of interest The authors have no relevant financial or non-financial interests to disclose. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Bali TG Modeling the dynamics of interest rate volatility with skewed fat-tailed distributions Annals of Operations Research 2007 151 1 151 178 10.1007/s10479-006-0116-6 Bogaert M Ballings M Van den Poel D Evaluating the importance of different communication types in romantic tie prediction on social media Annals of Operations Research 2018 263 1 501 527 10.1007/s10479-016-2295-0 Duan Y Edwards JS Dwivedi YK Artificial intelligence for decision making in the era of Big Data-evolution, challenges and research agenda International Journal of Information Management 2019 48 63 71 10.1016/j.ijinfomgt.2019.01.021 Dwivedi YK Hughes DL Coombs C Constantiou I Duan Y Edwards JS Upadhyay N Impact of COVID-19 pandemic on information management research and practice: Transforming education, work and life International Journal of Information Management 2020 55 102211 10.1016/j.ijinfomgt.2020.102211 Dwivedi YK Hughes L Ismagilova E Aarts G Coombs C Crick T Williams MD Artificial Intelligence (AI): Multidisciplinary perspectives on emerging challenges, opportunities, and agenda for research, practice and policy International Journal of Information Management 2019 57 101994 10.1016/j.ijinfomgt.2019.08.002 Filieri R Acikgoz F Ndou V Dwivedi Y Is TripAdvisor still relevant? The influence of review credibility, review usefulness, and ease of use on consumers’ continuance intention International Journal of Contemporary Hospitality Management 2020 33 1 199 223 10.1108/IJCHM-05-2020-0402 Giannakis M Dubey R Yan S Spanaki K Papadopoulos T Social media and sensemaking patterns in new product development: Demystifying the customer sentiment Annals of Operations Research 2022 308 145 175 10.1007/s10479-020-03775-6 Grover P Kar AK Dwivedi YK Understanding artificial intelligence adoption in operations management: Insights from the review of academic literature and social media discussions Annals of Operations Research 2022 308 177 213 10.1007/s10479-020-03683-9 Grover P Kar AK Dwivedi YK Janssen M Polarization and acculturation in US Election 2016 outcomes: Can twitter analytics predict changes in voting preferences ? Technological Forecasting and Social Change 2019 145 438 460 10.1016/j.techfore.2018.09.009 Gupta S Altay N Luo Z Big data in humanitarian supply chain management: A review and further research directions Annals of Operations Research 2019 283 1 1153 1173 10.1007/s10479-017-2671-4 Gupta S Kar AK Baabdullah A Al-Khowaiter WA Big data with cognitive computing: A review for the future International Journal of Information Management 2018 42 78 89 10.1016/j.ijinfomgt.2018.06.005 Ismagilova E Dwivedi YK Slade E Perceived helpfulness of eWOM: Emotions, fairness and rationality Journal of Retailing and Consumer Services 2020 53 101748 10.1016/j.jretconser.2019.02.002 Japan hotel REIT. (2021). Kessan Tanshin (Summary of financial results). https://www.jhrth.co.jp/file/term-089c9a0a7367506473473ba4a56136fa0a0693ef.pdf Kamboj S Gupta S Use of smart phone apps in co-creative hotel service innovation: An evidence from India Current Issues in Tourism 2020 23 3 323 344 10.1080/13683500.2018.1513459 Kamboj S Sarmah B Gupta S Dwivedi Y Examining branding co-creation in brand communities on social media: Applying the paradigm of Stimulus-Organism-Response International Journal of Information Management 2018 39 169 185 10.1016/j.ijinfomgt.2017.12.001 Kar AK Dwivedi YK Theory building with big data-driven research-Moving away from the “What” towards the “Why” International Journal of Information Management 2020 54 102205 10.1016/j.ijinfomgt.2020.102205 Kumar P Polonsky M Dwivedi YK Kar A Green information quality and green brand evaluation: The moderating effects of eco-label credibility and consumer knowledge European Journal of Marketing 2021 55 7 2037 2071 10.1108/EJM-10-2019-0808 Kumar S Xu C Ghildayal N Chandra C Yang M Social media effectiveness as a humanitarian response to mitigate influenza epidemic and COVID-19 pandemic Annals of Operations Research 2021 10.1007/s10479-021-03955-y McGroarty F Booth A Gerding E Chinthalapati VR High frequency trading strategies, market fragility and price spikes: An agent based model perspective Annals of Operations Research 2019 282 1 217 244 10.1007/s10479-018-3019-4 Menkveld AJ Vorst T A pricing model for American options with Gaussian interest rates Annals of Operations Research 2000 100 1 211 226 10.1023/A:1019275302878 Modgil S Gupta S Sivarajah U Bhushan B Big data-enabled large-scale group decision making for circular economy: An emerging market context Technological Forecasting and Social Change 2021 166 120607 10.1016/j.techfore.2021.120607 Morelli G Fair prices under a unified lattice approach for interest rate derivatives Annals of Operations Research 2021 299 429 441 10.1007/s10479-019-03209-y Nakatani S Nishimura KG Saito T Takahashi A Interest rate model with investor attitude and text mining IEEE Access 2020 8 86870 86885 10.1109/ACCESS.2020.2992477 Nilsang S Yuangyai C Cheng CY Janjarassuk U Locating an ambulance base by using social media: A case study in Bangkok Annals of Operations Research 2019 283 1 497 516 10.1007/s10479-018-2918-8 Nishimura KG Sato S Takahashi A Term structure models during the global financial crisis: A parsimonious text mining approach Asia-Pacific Financial Markets 2019 26 3 297 337 10.1007/s10690-018-09267-9 Rad AA Jalali MS Rahmandad H How exposure to different opinions impacts the life cycle of social media Annals of Operations Research 2018 268 1 63 91 10.1007/s10479-017-2554-8 Saito T Adachi T Nakatsuma T Takahashi A Tsuda H Yoshino N Trading and ordering patterns of market participants in high frequency trading environment: Empirical study in the Japanese stock market Asia-Pacific Financial Markets 2018 25 3 179 220 10.1007/s10690-018-9245-6 Saito T Takahashi A Stochastic differential game in high frequency market Automatica 2019 104 111 125 10.1016/j.automatica.2019.02.051 Saito T Takahashi A Koide N Ichifuji Y Application of online booking data to hotel revenue management International Journal of Information Management 2019 46 37 53 10.1016/j.ijinfomgt.2018.11.003 Saito T Takahashi A Tsuda H Optimal room charge and expected sales under discrete choice models with Limited capacity International Journal of Hospitality Management 2016 57 116 131 10.1016/j.ijhm.2016.06.006 Saumya S Singh JP Dwivedi YK Predicting the helpfulness score of online reviews using convolutional neural network Soft Computing 2019 24 1 17 Sun EW Kruse T Yu MT High frequency trading, liquidity, and execution cost Annals of Operations Research 2014 223 1 403 432 10.1007/s10479-013-1382-8 Wamba SF Edwards A Akter S Social media adoption and use for improved emergency services operations: The case of the NSW SES Annals of Operations Research 2019 283 1 225 245 10.1007/s10479-017-2545-9 Zhu X Yang C Liu K Zhang R Jiang Q Cooperation and decision making in a two-sided market motivated by the externality of a third-party social media platform Annals of Operations Research 2021 10.1007/s10479-021-04109-w

0

PMC9749645

NO-CC CODE

2022-12-15 23:23:21

no

Ann Oper Res. 2022 Dec 14;:1-23

utf-8

Ann Oper Res

10.1007/s10479-022-05136-x

oa_other

==== Front Biol Bull Rev Biology Bulletin Reviews 2079-0864 2079-0872 Pleiades Publishing Moscow 1222 10.1134/S207908642206010X Article Non-Specific Targets for Correction of Pneumonia Caused by Aerosols Containing Damaging Factors of Various Nature Yakovlev O. A. gniiivm_15@mil.ru 1 Yudin M. A. 12 Chepur S. V. 1 Vengerovich N. G. 13 Stepanov A. V. 1 Babkin A. A. 1 1 State Research Experimental Institute of Military Medicine, 198515 St. Petersburg, Russia 2 grid.445925.b 0000 0004 0386 244X North-Western State Medical University named after I.I. Mechnikov, 195067 St. Petersburg, Russia 3 grid.445902.e 0000 0001 0580 9341 Saint-Petersburg State Chemical Pharmaceutical University, 197376 St. Petersburg, Russia 14 12 2022 2022 12 6 649660 15 3 2022 22 3 2022 25 3 2022 © Pleiades Publishing, Ltd. 2022, ISSN 2079-0864, Biology Bulletin Reviews, 2022, Vol. 12, No. 6, pp. 649–660. © Pleiades Publishing, Ltd., 2022.Russian Text © The Author(s), 2022, published in Uspekhi Sovremennoi Biologii, 2022, Vol. 142, No. 4, pp. 390–403. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This review article provides data on the current state of the pathogenesis peculiarities of body and lung inflammation (pneumonia) under the influence of damaging factors of various nature: infectious agents, chemical toxicants, as well as incorporated radionuclides, etc. The peculiarities of inflammation itself, as a typical pathological process, are considered. Information on mediators that induce the so-called pro-resolving phase of inflammation manifestations is given. Approaches to the neuroimmune correction of non-specific inflammation are substantiated. Data on the following alternative approaches to the correction of nonspecific inflammation are summarized: factors of the coagulation system, modulators of the integrated stress response, and modulators of sigma-1 receptors. Based on the data presented, general directions for the treatment of nonspecific pneumonia are formulated, including reflexogenic and anti-inflammatory therapy in combination with multimodal drugs, as well as pro-resolving therapy in combination with drugs that prevent fibrosis. Keywords: inflammation lungs pulmonary protectors phlogogens aerosols non-specific pharmacological correction integrated stress response issue-copyright-statement© Pleiades Publishing, Ltd. 2022 ==== Body pmcINTRODUCTION Infectious agents, chemical toxicants, and incorporated radionuclides (hereinafter referred to as damaging factors) mediate specific and non-specific respiratory disorders as a result of aerogenic exposure. As this takes place, the formation of an inflammatory reaction acts as a key trigger mediating their damaging effect. The severity and intensity of this process depends on the characteristics of the bioavailability of the damaging factors and on the size of the inhaled aerosol particles: the larger the particles of the inhaled aerosol, the lower the probability of their deposition in the lower respiratory tract (Cargnello Roux, 2011). The geometry of various regions of the tracheobronchial tree plays an important role in the aerosol deposition profile. The diameter determines the necessary displacement of a particle before it contacts the epithelial lining, the cross-section determines the speed of air movement (Basharin et al., 2022), and the nature of bronchial branching affects the mixing between inhaled and reserve air (Chepur et al., 2019; Codagnone et al., 2018). In the case of aerosol intake, the basis for subsequent interactions of a damaging factor with the body is the level of the neuroimmunoendocrine reaction, which determines the intensity of nonspecific inflammatory processes, often resulting in death or pronounced morphological and functional changes in the lung tissue, significantly reducing the quality of life, the associated illnesses are: chronic obstructive pulmonary disease, diffuse pulmonary fibrosis, respiratory failure etc. Therefore, the tactics for prevention and treatment of injured individuals with this pathology should include emergency, if possible, inhalation administration of drugs that prevent the induction of trigger mechanisms of the inflammatory reaction, level the initial phase of the inflammation and accelerate the formation of an adaptive complex of structural and functional rearrangements. To determine the possible directions of therapy for inhalation lung injuries, one should take into account the molecular and cellular mechanisms of induction, maintenance, and resolution of inflammation inherent in the effect of a wide range of damaging factors. At the moment, the scheme for the correction of nonspecific pneumonia (pulmonoprotection) is based on the use of anti-inflammatory and pro-resolving therapy, as well as on blocking pathological reflex reactions (Fig. 1). Fig. 1. The main directions of therapy for nonspecific pneumonia caused by adverse environmental factors. CORRECTION OF THE INFLAMMATORY REACTION ITSELF It is known that the interaction of the body with a damaging factor initiates the release of pro-inflammatory mediators. Thus, direct damage to cells by chemical toxicants leads to the release of molecular fragments of DAMPs (damage-associated molecular patterns) associated with the damage, while damage caused by a biopathogen releases molecular fragments of PAMPs (pathogen-associated molecular patterns) associated with the pathogen. There is also a universal system of molecular signals associated with disruption of cellular homeostasis, gene expression, inflammation, allergy, and tissue repair. The above signal molecules are represented by proteins, RNA, DNA, lipopolysaccharides (LPSs), exo- and endotoxins, various glycans that activate pattern recognition receptors (toll-like receptors TLR1–11, NOD-like receptors NBS-LRR (nucleotide binding site leucine-rich repeat), RIG-I-like receptors (RLR)), membrane 2'‑5'-oligoadenylate synthetases, protein kinases activated by double-stranded RNA, etc. (Chepur et al., 2019; Di Salvo et al., 2021). The combined or isolated action of these molecular patterns determines the induction of inflammation. Some xenobiotics (phlogogens) can predetermine their activation without direct tissue damage, presumably through the stress molecules MICA, MICB, RAE-1, ULBP1, ULBP2, and ULBP3 with high affinity for killer-inhibiting receptors of NK cells (Zingoni et al., 2018). It is believed that the tissue- and stress-specific representation, concentration, and duration of the effect of these signals program the inflammatory response and determine its specificity. The circumstances mentioned above dictate the need to identify the most significant predictors of nonspecific inflammation. Regardless of the nature of the damaging factor, disruption of cell integrity leads to the release into the extracellular environment of a significant amount of cytosolic and intranuclear contents, including conservative hydrophobic domains, the alarmins. Alarmins are represented by defensins, cathelicidins, eosinophil neurotoxin, nuclear non-histone protein B1 (HMGB1), heat shock proteins HSP, nucleic acids, histones, and nucleosomes. Acting as ligands for specific receptors—TLRs, glycan-binding receptor RAGE (receptor for advanced glycation endproducts), phagocytosis receptors, including MerTK and MFG-E8, C-type lectin receptor (CLR), scavenger receptor (SR), etc.—alarmins induce an inflammatory response (Anfinogenova et al., 2020). Thus, massive tissue necrosis during surgical operations leads to a systemic inflammatory reaction, which ends in death in 4–7% of cases (Bakulina et al., 2017). Alarmin receptors act as an associating link in response to damaging stimuli, and the impact on them is justified from the viewpoint of the universality of the correction of nonspecific inflammation. The influence of DAMPs is mediated by a significant representation of receptors, among which TLR4 plays a significant role. Demonstrating the ability to receive a huge number of DAMPs, including bacterial LPS, TLR4 serves as a universal trigger for the development of nonspecific inflammation, the reinforcement of which is determined by cells of the human immune system. Existing in the form of homo- and heterodimeric complexes with receptors of other subtypes, which are located both on the cytolemma and intracellularly in the cytoplasm, TLR4 triggers the mechanisms of innate immunity. The ability to bind to all adapter proteins determines the main role of TLR4 in the formation of nonspecific inflammation, including cytokine secretion, the production of pro-inflammatory IL-13 and IL-18, as well as autophagy and apoptosis processes (Bakulina et al., 2017). TLR4s are involved in the pathogenesis of chronic inflammatory diseases, acute sepsis, and the life-threatening complications of dangerous infections (Perrin-Cocon et al., 2017). Thus, TLR4 gene knockout protects rodents from death as a result of impaired myocardial contractility caused by Bacillus anthracis toxin (Kandadi et al., 2012). It has been proven (Imai et al., 2008) that the severity of acute lung injury under the influence of damaging factors depends on the degree of activation of the innate immunity signaling pathway TLR4–TRIF–TRAF6, where TLR4 is toll-like receptor 4, TRIF is TIR-domain-containing adapter-inducing interferon-, and TRAF6 is TNF receptor-associated factor 6. In turn, the use of TLR4 antagonists prevents the secretion of pro-inflammatory cytokines, LPS-induced dendritic cell migration, and also reduces acute lung injury and mortality associated with viral infection. μ-Оpioid receptor antagonists (naloxone and naltrexone) can be used as TLR4 blockers. Their effectiveness against TLR4-dependent pathology, neuropathic pain or addiction, has been demonstrated in a number of studies (Watkins et al., 2014; Wang et al., 2016). Currently, the dextrorotatory isomer of naltrexone is considered the most promising means of correcting the hyperimmune response (Selfridge et al., 2015). The ability to block the cascade of PAMP-associated LPS-dependent stimuli in relation to TLR4 is associated with this isomer. The release of mediators into the bloodstream promotes the recruitment of neutrophils and monocytes of bone marrow origin, which migrate to the focus of inflammation, where they undergo differentiation. In the focus of inflammation, epithelial and immunocompetent cells begin to produce cytokines and chemokines: MCP-1 (monocyte chemoattractant protein 1), CCR2 (C-C motif chemokine receptor type 2), CXCL-1 (C-X-C motif chemokine ligand 1), among which the priority role in the development and progression of the inflammatory process belongs to TNF-α and IL-1β. As a result of the action of proteolytic enzymes, TNF-α is released from its association with the cell membrane and interacts with the receptor of the same name on the surface of macrophages, dendritic cells, and T lymphocytes and activates the apoptosis domain through the mitogen-activated protein kinase p38 MAPK and NF-κB. MAPK activates the Ets transcription factors and, as a result, the expression of the fas genes, resulting in an increase in the production of class II pro-inflammatory proteins (DP, DQ, DR) of HLA antigens of the T-cell receptor, AP-1, which regulates the transcription of cyclin 1D and the pro-inflammatory growth factors NGF, EGF, PDGF, etc. (Cargnello and Roux, 2011). The result of the activation of neutrophils and monocytes is the accumulation of reactive oxygen and nitrogen species. In terms of screening effective pulmonary protectors, it is possible to use a model of lung damage by cigarette smoke, based on the general pathogenetic mechanisms of lung damage inherent in many damaging factors (Fig. 2). Fig. 2. Pathogenetic links of lung damage by cigarette smoke. Among the molecular targets, the impact on which can contribute to the prevention and treatment of nonspecific pneumonia, it is advisable to mark out the following (Yao et al., 2008): (1) Cell signaling inhibitors: type B (PDE4B) and type 4 (roflumilast, cilomilast, GRC3886 and GSK842470) phosphodiesterase inhibitors; p38-MAPK inhibitors (SB 203580, SB 239063 and RWJ 67657, SD282, GSK-681323, GSK-85633); IKK-2 inhibitors (under development); and inhibitors of phosphoinositide-3-kinase PI3K of δ- and γ-types (LY294002). (2) Cytokine and chemokine inhibitors: CCL2 (MCP-1) and CCR2 antagonists (CCX915, INCB3284, antibodies ABN912, INCB8696, JNJ-27553292, SKL-2841, and INCB3344); CXCL1, CXCL8 antagonists (SB-265610, SCH 527123); TNF- inhibitors (infiximab, etanercept, adalumimab). (3) Modifiers of HDAC histone deacetylases (vorinostat and romidepsin). (4) Antiproteinases: neutrophil elastase inhibitors (ZD0892, AZD3342, α-1-antitrypsin); inhibitors of MMP-1, -2, -9, -12 (macrophage) types (marimastat, BMS-561392, and GW3333). PHARMACEUTICALS THAT INDUCE THE PRO-RESOLVING PHASE OF THE INFLAMMATORY REACTION Despite the fact that resolution is the outcome of inflammation of any etiology, this period is characterized by an active process regulated by special pro-resolving lipid mediators (SPMs). Their synthesis begins at the moment of neutrophil activation (minutes, hours) and continues until the functional and structural restoration of lung tissue (days, weeks). SPMs reduce the severity of the inflammatory response without depression of the immune system, protect tissues from damage, accelerate the removal of etiological factors of inflammation and apoptotic leukocytes, helping to restore homeostasis (Krishnamoorthy et al., 2018). SPMs are formed as a result of the fermentation of fatty acids: lipoxins are formed from arachidonic acid; E-series resolvins, from eicosapentaenoic acid; and D-series resolvins, protectins, and maresins, from docosahexaenoic acid. Under certain conditions, highly active conjugates can be formed, for example, SPM-sulfide or SPM-acetyl derivatives (Chiang and Serhan, 2020). A large number of recent publications devoted to resolvins are associated, among other things, with the prospects for their use in the treatment of COVID-19, since a hyperinflammatory reaction, up to a cytokine storm, is one of the main causes of high mortality during coronavirus infection (Panigrahy et al., 2020). The basis for this was the results of the use of pro-resolving therapy on models of lung tissue pathology, these results showed that the use of resolvin D1 is the most promising. Thus, it was shown (Xia et al., 2019) that resolvin D1, when administered intraperitoneally, reduced ventilation-associated damage to the lungs of mice, the severity of emphysema and chronic inflammation in the model of chronic (Hsiao et al., 2015) and acute (Hsiao et al., 2013) exposure to cigarette smoke. Resolvin D1 contributed to the resolution of pneumonia using a model of infection caused by Pseudomonas aeruginosa (Codagnone et al., 2018) or a model of LPS-induced lung injury (Wang et al., 2014). Resolvin D1 limited septic lung injury in mice induced by ligation and puncture of the caecum (Zhuo et al., 2018). The most important feature of the biochemical transformations of resolvins is associated with their sensitivity to the acetylating effect of aspirin. Thus, in a model of paraquat-induced lung injury in mice, aspirin-induced resolvin D1 had a protective effect by reducing oxidative stress, inflammatory response, and pulmonary edema (Hu et al., 2019), and in a model of pneumonia of mixed etiology, it reduced inflammation and infection of the lungs (Wang et al., 2017). In the resolution of inflammation, in addition to resolvins, lipoxins, and maresins, the following proteins are also involved: annexin A1, and TIM-4 (T-cell immunoglobulin and mucin domain 4), the action of which is realized due to the modification of the signaling pathway (PI3K associated with ATP) and the work of intracellular regulators of inflammatory reactions (the protein kinase B (PKB) family and serine-threonine specific protein kinase). It was demonstrated that by binding to the specific N-formyl peptide receptor type 2 (FPR2) of leukocyte membranes, annexin A1 reduces epithelial adhesion, processes of leukocyte migration, chemotaxis, and phagocytosis induced by N-formylmethionine-containing oligopeptides (including products of arachidonic acid) (Schloer et al., 2019). The above data indicate the perspectives of using resolvins, in particular, resolvin D1, as agents that induce the pro-resolving phase of nonspecific inflammation. NEUROIMMUNE CORRECTION OF NONSPECIFIC PNEUMONIA The acute impact of damaging factors and, most importantly, toxic substances on the lungs is caused not only by a violation of tissue integrity due to direct cytolytic action, but also by hyperactivation of the body’s protective functions, such as mucus secretion, cough, bronchiospasm, edema, and neurogenic inflammation. The intensity of reflex reactions can be reduced by opening the reflex arc represented by the afferent, central, and efferent phases. This approach is successfully implemented using β2-adrenergic agonists and M3-selective anticholinergic agents as means of urgent therapy for injuries with pulmonary toxicants (De Virgiliis and Di Giovanni, 2020). The respiratory epithelium contains pulmonary neuroendocrine cells, some of which are associated with the processes of nerve cells and are called NEBs (neuroepithelial bodies), which are capable of releasing immunotropic amines and peptides. The role of NEBs in the development of the lungs during inflammatory diseases, as well as the ability of NEBs to act as stem cells have been proven (Yeger et al., 2019). It was shown that the innervation of NEBs, by analogy with the vascular endothelium, is carried out by sensory fibers of the nodose and jugular ganglia of n. vagus, which express membrane receptors for neurotropic factors and purines, respectively. Activation of fibers by endogenous and exogenous mediators causes depolarization of neurons, followed by the release of neuropeptides both in the lungs and near the secondary neurons of the nucleus of the solitary pathway, triggering a reflex cough. Low conduction (~1 m/s) unmyelinated C fibers that express TRPV1 receptors are also involved in the development of cough (Nassenstein et al., 2018). Parasympathetic efferent regulation of neuroimmune processes in the lungs is anatomically and histologically provided by preganglionic neurons of the dorsal motor nucleus of n. vagus reaching the intramural ganglia of the trachea containing acetylcholine-producing neurons and non-cholinergic neurons of Auerbach’s plexus near the outer longitudinal muscular layer of the esophagus. The latter release vasoactive intestinal peptide (VIP) and NO as neurotransmitters (Yildiz-Pekoz and Ozsoy, 2017). Unlike neuronal acetylcholine, non-neuronal acetylcholine release is non-quantum. Thus, in an experimental model of bronchial asthma in animals, the use of anticholinergic drugs provided a bronchodilatation effect, a decrease in mucus and inflammation, and remodeling through the process of blocking the M3R. The pro-inflammatory effect of acetylcholine, including an increase in neutrophil chemotaxis and stimulation of the production of reactive oxygen species, is probably also mediated by M3R activation, as confirmed using pharmacological probes (Shen et al., 2020). VIP-ergic parasympathetic nerve fibers are extensively co-localized with a peptide that activates pituitary adenylate cyclase. Both mediators increase vasodilation and bronchodilatation. In addition, mast cells and Th2 lymphocytes synthesize VIP which has an anti-inflammatory effect that has been confirmed in various models of pneumonia. Thus, prophylactic intratracheal injection of a VIP analogue reduced the recruitment of inflammatory cells by 70% in the bronchoalveolar lavage fluid (Prescott et al., 2020). Late interaction between sensory neurons and eosinophils, as well as antigen-presenting cells, was confirmed histologically. Eosinophils influence the branching of sensory neurons, presumably through excessive secretion of nerve growth factor. In turn, it has been proven that nerve fibers stimulate the migration of eosinophils due to the secretion of eotaxin-1. A significant role in the organization of neuroimmune interactions belongs to mast cells, which can activate the TRPV1 receptor through the neuronal H1 receptor. The clinical significance of neuroimmune interaction in providing inflammatory processes is confirmed by the clinical success of refractory therapy of bronchial asthma by bronchial thermoplasty and vagus nerve stimulation (Caravaca et al., 2019). Neuroimmune interactions at the anatomical level, acting as targets for the pharmacological correction of inflammation are schematically shown in Fig. 3. Fig. 3. Schematic representation of neuroimmune interactions that act as targets for the pharmacological correction of inflammation. P2XR—ATP-gated P2X receptor cation channel family; P2YR—ATP-gated P2Y receptor cation channel family; AR—adrenoreceptors; M2R—muscarinic acetylcholine receptor M2; M3R—cholinergic/acetylcholine receptor M3; CGRP—calcitonin gene-related peptide; β2AR—β2 adrenoreceptor; VIP—vasoactive intestinal peptide; NO—nitric oxide; TRPV1R—transient receptor potential vanilloid subtype 1; TRPA1R—transient receptor potential cation channel subfamily V member 1; TRPM8R—transient receptor potential cation channel subfamily M (melastatin) member 8; NKA—neurokinin A; SP—substance P; 5-HT—serotonin. Stimulation of parasympathetic fibers induces bronchiospasm, mucus hypersecretion, and vasodilation of pulmonary vessels through activation of acetylcholine muscarinic M3 receptors (CHRM3). Muscarinic acetylcholine receptors M2 (CHRM2) are present on presynaptic terminals, and their activation reduces the release of acetylcholine and prevents bronchiospasm. Many biological factors, such as parainfluenza viruses, disrupt the function of CHRM2, which leads to an increase in the release of acetylcholine and an increase in smooth muscle tone. Taking into account the ability of pulmonary macrophages to express CHRM2 and CHRM3, it can be concluded that there is a correlation between the activity of the acetylcholine system and the severity of inflammation (Koarai et al., 2012). The positive feedback is confirmed by the ability of IFNγ, TNF-α, and IL-1β to increase CHRM2 expression, which has been shown in models of viral infections, including SARS-CoV-2 (Rynko et al., 2014). Activation of C-fibers containing calcitonin gene-related peptide CGRP, tachykinins (substance P and neurokinin A) causes spasm of airway smooth muscles, vasodilation, excessive mucus formation, which together form the clinical manifestations of nonspecific inflammation syndrome. It has been shown that P2RY1+ neurons of the vagus nerve trigger the inflammatory reaction cascade in response to inhalation intake of various irritants, while Piezo2 airway stretch receptors integrate and coordinate protective reflexes, in particular, laryngospasm (Nonomura et al., 2017). Sensory cells rich in NEBs with bioactive mediators—bombesin, serotonin, and CGRP participate in the implementation of non-specific inflammation under the influence of damaging factors (Noguchi et al., 2020). Promising targets for pharmacological correction of reflex-dependent inflammation may be cationic channels acting through a transient receptor potential channel (TRP channel): vanilloid (TRPV1), ankyrin (TRPA1), and melastatin (TRPM8), as well as receptors for mediators of efferent fibers of the lung tissue (substance P and CGRP). TRPA1, TRPV1, and TRPM8, localized on the nociceptive peripheral neurons are responsible for the perception of acute and chronic pain, the initiation of the reflex cough, and the development of asthma attacks, lung damage, and aseptic inflammation (Grace et al., 2013). Their peculiarity is the possibility of activating inflammatory processes without connection with a specific pulmotoxicant, but only through ingress of particles with a diameter of less than 2.5 mm, the so-called fine particles (FP 2.5). It has been experimentally shown that prophylactic treatment with antagonists of TRPV1- and TRPA1 receptors effectively prevents the pneumonia and bronchial hypersensitivity caused by intranasal instillation of mice with FP 2.5 a dose of 7.8 mg/kg (Xu et al., 2019). Another member of the TRP receptor family, TRPC6, is expressed on eosinophils, neutrophils, mast cells, and CD4+ lymphocytes, and its activation is involved in the pathogenesis of asthma and allergic inflammation. In turn, activation of TRPV1 on the C fibers of sensory neurons leads to the release of tachykinins and increased recruitment of immune cells (Jia and Lee, 2007). TRPA1 receptors are activated by lachrymators (CS, CN, CR), drugs (paracetamol, diphenhydramine), combustion products, chemotherapeutic agents, cigarette smoke, and other pollutants (Belvisi and Birrell, 2017). Currently, the following are considered as the most likely candidates capable of blocking TRP channels: AG489 isolated from the venom of the Agelenopsis aperta spider, HCRG21 isolated from the sea anemone Heteractis crispa, JYL-1421, and AMG8562 (Kvetkina et al., 2019). ALTERNATIVE APPROACHES TO THE CORRECTION OF NONSPECIFIC INFLAMMATION Blood Coagulation Factors Blood coagulation factors are classified as pro-inflammatory factors, while substances that have the opposite effect have an anti-inflammatory action, as shown in various models of lung injury (Choi et al., 2008). Thus, in a model of LPS-induced lung damage in rats, the administration of heparin by a nebulizer at a dose of 1000 UI/kg provided a decrease in the expression of the plasminogen gene, as well as the effectors TGFβ—Smad 2, Smad 3—NF-κB—P-selectin, and CCL2. Preventive inhalation of heparin had a pronounced protective effect (Chimenti et al., 2017). In addition, heparin had an anti-asthma effect in models of allergen-, adenosine-, and exercise-induced asthma by preventing mast cell degranulation (Mousavi et al., 2015). Inhalation of heparin in conditions accompanied by inflammation of the lung tissue and systemic hypercoagulability is becoming a mandatory indication for patients with severe forms of COVID-19 (van Haren et al., 2020). It has been proven that inhalations of unfractionated heparin provide an anti-inflammatory effect due to the following: — block of heparan sulfate-containing receptors of the cytolemma, considered as a translocation cofactor of the peplomer spike protein to the receptor-binding domain of the angiotensin-converting enzyme 2 receptor (ACE2) (Clausen et al., 2020); — decreased expression of pro-inflammatory mediators and inhibition of the complement system due to interaction with the C1 component (Shi et al., 2021); — disaggregation of DNA and actin followed by activation of endogenous DNase and a decrease in electrostatic interactions between mucin molecules (Broughton-Head et al., 2007); — inactivation of thrombin, factor Xa, kallikrein, serine proteases and reduction of excessive fibrin deposition (Camprubí-Rimblas et al., 2018). In addition to heparin, antithrombin is of practical interest, which, with a similar anti-inflammatory activity to heparin, is devoid of the side effects of the latter (osteoporosis and thrombocytopenia). In a model of pneumonia in rats caused by Streptococcus pneumoniae, intravenous administration of antithrombin led to a decrease in the number of neutrophils in bronchoalveolar lavage, and after 48 hours it led to a significant drop in the levels of TNF-α, IL-6, and cytokine-induced neutrophil chemoattractant 3, CINC-3, also in the model of acute lung injury by endotoxin it led to a decrease in the expression of ERK1/2 and p38-MAPK (Sun et al., 2009). Modulators of the Integrated Stress Response Each eukaryotic cell, when exposed to damaging factors, reacts in an integrated way by reducing protein synthesis with two goals: both preventing the accumulation of defective forms and preventing the replication of the bioagent, as well as increasing gene expression, which makes it possible to specifically adapt either to damaging factors or to factors that trigger apoptosis (Emanuelli et al., 2020). For various stress factors, there are intracellular sensor kinases that trigger the integrated stress response (ISR). In the presence of double-stranded DNA (dsDNA) in the cytosol, cytosolic protein kinase R (PKR) and PKR-like endoplasmic reticulum kinase activate kinases of the eukaryotic initiation factor 2α (eIF-2α), in the case of oxidative stress and Fe2+ deficiency, the heme-regulated inhibitor (HRI) is activated, in the event of a deficiency of amino acids, the kinase that controls the general control nonderepressible 2 (GCN2) is activated. After phosphorylation of the α subunit of translation initiation factor 2 (eIF-2), the latter interacts with methionine transfer RNA (tRNA) on a ribosome and starts translation. Normally, eIF-2α hydrolyzes bound GTP, which is replenished by the guanine nucleotide exchange factor eIF-2β. However, under the influence of damaging factors, eIF-2α is mainly represented by a phosphorylated form which easily binds to eIF-2β and inhibits further GTP metabolism. Therefore protein synthesis can slow down significantly, which contributes to the activation of the transcription factors ATF4 and CHOP and an increase in the expression of the damage-induced protein (DIP) by the GADD34 gene. DIP forms a complex with protein phosphatase PP1 and G-actin, ensuring the functioning of eIF-2α-specific phosphatases and maintenance of translation. This mechanism allows cells to temporarily activate the ISR (whereas in the absence of such interaction, cell death can be observed) (Pakos-Zebrucka et al., 2016). The function of ISR is viewed from opposite positions relating to viral invasion. This is due, as a rule, to the very nature of pathogens. PKR activation of viral DNA is considered part of antiviral immunity, but many viruses have improved systems to counteract this mechanism. For viruses of the families Togaviridae, Reoviridae, and hepatitis C virus, eIF-2α phosphorylation can promote translation of their mRNA (Fusade-Boyer et al., 2019). For other rapidly replicating viruses, on the contrary, the implementation of such a mechanism significantly slows down the intracellular assembly of the protein capsid of viral particles. ISR is a universal nonspecific defense mechanism in response to the action of damaging factors of various nature. However, excessive ISR may be unfavorable for the cell and may not be adaptive. In this regard, pharmacological tools are being developed to reduce the intensity of ISR in the form of so-called inhibitors of the integrated stress response, ISRIB. Their mechanism of action is to bind by the allosteric site to eIF-2β, the main target of phosphorylated eIF-2α, and to enhance the recruitment of guanidine residues in the presence of phosphorylated eIF-2α (Zyryanova et al., 2021). Therefore, ISRIB converts an inactive complex involved in protein synthesis into an active one, restoring the cell’s potential to synthesize protective proteins. ISR dysfunction can underlie conditions associated with insufficient function of the endogenous defense system against damaging factors of various nature, which requires the study of the issue of targeted pharmacological correction of dysfunctional ISR. To date, several ISR-modulating compounds have been synthesized that are structurally similar to ISRIB. Although ISR represents a novel target in pneumonia protection (Van’t Wout et al., 2014), there is already evidence for the effectiveness of ISR modulators in various respiratory pathologies. Thus, ISRIB at a dose of 2.5 mg/kg, when administered intraperitoneally, reduced the severity of bleomycin- and asbestosis-induced pulmonary fibrosis, including that caused by a decrease in excess collagen formation. In addition, ISRIB promotes the transformation of type 2 alveolar macrophages into type 1 alveolar macrophages, which is necessary to maintain gas exchange in the lungs (Watanabe et al., 2021). ISRIB, like salubrinal and Sal003, inhibits eIF-2α phosphatase. Sephin1 (a derivative of guanabenz) selectively blocks GADD34-phosphatase and stress-induced protein phosphatase 1 (PPP1R15A), as a result, the released serine/threonine phosphatase PP1 dephosphorylates eIF-2α. Apparently, due to the additional effect on protein phosphatase, the latter, after a course of administration for 11 days at a dose of 5 mg/kg, showed antiviral activity in the model of infection of rabbits with the myxoma virus. Thus, the use of ISR modulators to prevent fibrotic processes in the lungs after inhalation exposure to pathological factors may be promising . The anti-inflammatory potential of this class of compounds is well realized, especially in diseases associated with protein misassembly (folding), in storage diseases, however, there is still no unequivocal opinion regarding bacterial and viral lesions (Pierre, 2019). Multimodal Agents Since the inflammatory process is characterized by the involvement of a huge number of effector cells and their signaling pathways, it may be justified to use anti-inflammatory drugs of a multimodal type of action, for example, plant-derived substances related to flavonoids: quercetin and its analogue, dihydroquercetin (Adhikari et al., 2021). Using a model of LPS-induced damage to the TC-1 cell line, it was shown that taxifolin reduces the cytotoxic effect and the level of lethality through modulation of NF-κB signaling (Liu et al., 2020). Using computer simulation, the potential of taxifolin and rhamnetin to act as inhibitors of SARS-CoV-2 main protease (Mpro) with satisfactory tolerability characteristics has been proven (Fischer et al., 2020). Taxifolin showed the ability to suppress oxidative stress and pneumonia caused by benzo[a]pyrene (125 mg/kg) when administered to mice at a dose of 20–40 mg/kg for 14 days. The corrective effect was due to the high level of expression of NF-E2-related factor 2 (Nrf2), NAD(P)H quinone dehydrogenase 1 (NQO1), heme oxygenase 1 (HO-1), and superoxide dismutase (SOD), where Nrf2 plays the leading role in the suppression of inflammation through inhibition of the NF-κB signaling pathway (Islam et al., 2021). Quercetin exhibits anti-inflammatory, antioxidant properties, and also blocks lipid peroxidation, platelet aggregation, and vascular permeability. Its efficacy has been demonstrated in the model of LPS-induced TNF-α production in macrophages, LPS-induced IL-9 production in A54 lung cells, LPS-induced increase in mRNA and TNF-α in glial cells, FcεRI-mediated release of pro-inflammatory cytokines, tryptases, and histamine from mast cell culture (Jafarinia et al., 2020). Quercetin has a bronchodilatory effect, presumably by enhancing downstream signals from β-adrenergic receptors of bronchial smooth muscle and by inhibiting PDE4. Exposure was increased by introduction via inhalation with PBS buffer solution using a nebulizer and prevented methacholine induced increase in airway resistance. The noted ability of the substance to increase isoprenaline-induced relaxation of the bronchi may be required in the development of tachyphylaxis to β-adrenergic agonists, which complicates the treatment of not only an attack of bronchial asthma, but also treatment of damage by air pollutants. Antidepressants, Sigma-1 Receptor Modulators Fluvoxamine, a selective serotonin reuptake inhibitor (SSRI), has a high affinity for sigma-1 receptors at therapeutically relevant doses. The mechanisms of its anti-inflammatory and immunomodulatory properties are not fully understood, however, the ability to block the transfer of SARS-CoV-2 through endolysosomes and hypercoagulability in COVID-19 indicates a potential impact upon lysosome membrane formation (Marčec and Likić, 2021). In a double-blind, randomized study of adult patients with COVID-19, individuals treated with fluvoxamine had a lower likelihood of clinical deterioration within 15 days. Later, it was suggested that the positive therapeutic effect may also be associated with the inhibition of acid sphingomyelinase and the ability to influence membrane protonation and thereby promote the retention of the viral particle in the lysosome (Sukhatme et al., 2021). It is known that SSRIs affect platelet aggregation by reducing their serotonin content. After 12 weeks of therapy with fluvoxamine at a dose of 100–150 mg/day, the concentration of serotonin in platelets decreases by 86%, and in blood plasma, by 60% (Celada et al., 1992). By reducing the concentration of serotonin in platelets, SSRIs can reduce their potential for aggregation and prevent the state of hypercoagulability against the background of inflammation caused by damaging factors. The additional anti-inflammatory effect of fluoxamine is associated with the ability to inhibit the breakdown of melatonin in the liver, which leads to an increase in its concentration in blood plasma. By blocking mast cell degranulation, it reduces histamine-mediated tissue edema and hypercoagulability. On the other hand, fluvoxamine reduces the manifestations of hypercytokinemia in a model of LPS-induced sepsis. This effect may be mediated by the impact of fluvoxamine upon the endoplasmic reticulum resident protein, the sigma-1 receptor, which is considered a natural factor in inhibition of cytokine production (Rosen et al., 2019). The ability of fluvoxamine to lower levels of pro-inflammatory cytokines explains the reduction in excess IL-6 in patients who have had COVID-19. Thus, the order of application of therapeutic agents aimed at reducing the activity of pro-inflammatory cytokines should be adjusted based on specific periods of potential growth in their level after exposure to damaging factors. CONCLUSION Acute and chronic effects of damaging factors of chemical and biological origin on the respiratory system and the body as a whole, are ubiquitous and represent a significant proportion in the structure of overall morbidity and mortality (Fig. 4). The use of standardized treatment regimens does not always lead to complete recovery of affected individuals. For example, those who have recovered from COVID-19 may display long-term consequences that reduce their quality of life. In this regard, there is a need to study and develop universal (non-specific) means of treating pneumonia, providing for the blockade of pathogenetic cascades, and targeted therapeutic and prophylactic effects on the key stages of inflammation in order to accelerate the solution of inflammation and interrupt maladaptive reflex reactions. Fig. 4. Chronology of molecular and cellular changes during pneumonia. ABB—air-blood barrier; damaging factors displayed from top to bottom: bacteria, cigarette smoke, industrial air pollutants, pulmonotoxicants like paraquat, viruses. General directions for the treatment of nonspecific inflammation should include: reflexogenic therapy immediately after exposure to damaging factors; then, anti-inflammatory therapy in combination with multimodal drugs; then, pro-resolving therapy in combination with drugs that prevent fibrosis. COMPLIANCE WITH ETHICAL STANDARDS Conflict of interest. The authors declare that they have no conflicts of interest. In carrying out this work, all ethical standards were observed. This article does not contain any studies involving humans and animals as objects. Translated by A. Ostyak ==== Refs REFERENCES 1 Adhikari B. Marasini B.P. Rayamajhee B. Potential roles of medicinal plants for the treatment of viral diseases focusing on COVID–19: a review Phytother. Res. 2021 35 1298 1312 10.1002/ptr.6893 33037698 2 Anfinogenova N.D. Quinn M.T. Schepetkin I.A. Atochin D.N. Alarmins and c-Jun N-terminal kinase (JNK) signaling in neuroinflammation Cells 2020 9 2350 10.3390/cells9112350 33114371 3 Bakulina, L.S., Litvinenko, I.V., Nakatis, Ya.A., et al., Sepsis: Pozhar i bunt na tonushchem v shtorm korable. Chast’ 1. Triggery vospaleniya. Retseptsiya triggerov vospaleniya i signal’naya transduktsiya (Sepsis: Fire and Riot on a Ship Sinking in a Storm. Part 1. Triggers of Inflammation. Reception of Inflammatory Triggers and Signal Transduction), St. Petersburg: Severo-Zapad. Gos. Med. Univ. im. I.I. Mechnikova, 2017. 4 Basharin, V.A., Chepur, S.V., Tolkach, P.G., et al., Toksikologiya produktov goreniya polimernykh materialov (Toxicology of Combustion Products of Polymeric Materials), St. Petersbufg: Levsha, 2022. 5 Belvisi M.G. Birrell M.A. The emerging role of transient receptor potential channels in chronic lung disease Eur. Respir. J. 2017 50 1601357 10.1183/13993003.01357-2016 28775042 6 Broughton-Head V.J. Shur J. Carroll M.P. Unfractionated heparin reduces the elasticity of sputum from patients with cystic fibrosis Am. J. Physiol.: Lung Cell. Mol. Physiol. 2007 293 L1240 L1249 17827252 7 Camprubí-Rimblas Anticoagulant therapy in acute respiratory distress syndrome Ann. Transl. Med. 2018 6 36 10.21037/atm.2018.01.08 29430453 8 Caravaca A.S. Gallina A.L. Tarnawski L. An effective method for acute vagus nerve stimulation in experimental inflammation Front. Neurosci. 2019 13 877 10.3389/fnins.2019.00877 31551672 9 Cargnello M. Roux P.P. Activation and function of the mapks and their substrates, the MAPK-activated protein kinases Microbiol. Mol. Biol. Rev. 2011 75 50 83 10.1128/MMBR.00031-10 21372320 10 Celada P. Dolera M. Alvarez E. Artigas F. Effects of acute and chronic treatment with fluvoxamine on extracellular and platelet serotonin in the blood of major depressive patients. Relationship to clinical improvement J. Affective Disord. 1992 25 243 249 10.1016/0165-0327(92)90082-H 11 Chepur S.V. Pluzhnikov N.N. Saiganov S.A. The hypothesis of the aperiodic polysaccharides matrix synthesis Usp. Sovrem. Biol. 2019 139 583 593 12 Chiang N. Serhan C.N. Specialized pro-resolving mediator network: an update on production and actions Essays Biochem. 2020 64 443 462 10.1042/EBC20200018 32885825 13 Chimenti L. Camprubí-Rimblas M. Guillamat-Prats R. Nebulized heparin attenuates pulmonary coagulopathy and inflammation through alveolar macrophages in a rat model of acute lung injury Thromb. Haemostasis 2017 117 2125 2134 10.1160/TH17-05-0347 29202212 14 Choi G. Hofstra J.J. Roelofs J.J. Antithrombin inhibits bronchoalveolar activation of coagulation and limits lung injury during Streptococcus pneumoniae pneumonia in rats Crit. Care Med. 2008 36 204 210 10.1097/01.CCM.0000292012.87482.F4 18090375 15 Clausen T.M. Sandoval D.R. Spliid C.B. SARS-CoV-2 infection depends on cellular heparan sulfate and ACE2 Cell 2020 183 1043 1057 10.1016/j.cell.2020.09.033 32970989 16 Codagnone M. Cianci E. Lamolinara A. Resolvin D1 enhances the resolution of lung inflammation caused by long-term Pseudomonas aeruginosa infection Mucosal Immunol. 2018 11 35 49 10.1038/mi.2017.36 28422188 17 De Virgiliis F. Di Giovanni S. Lung innervation in the eye of a cytokine storm: neuroimmune interactions and COVID-19 Nat. Rev. Neurol. 2020 16 645 652 10.1038/s41582-020-0402-y 32843733 18 Di Salvo E. Di Gioacchino M. Tonacci A. Alarmins, COVID-19 and comorbidities Ann. Med. 2021 53 777 785 10.1080/07853890.2021.1921252 34042528 19 Emanuelli G. Nassehzadeh-Tabriz N. Morrell N.W. Marciniak S.J. The integrated stress response in pulmonary disease European Respiratory Review 2020 29 200184 10.1183/16000617.0184-2020 33004527 20 Fischer A. Sellner M. Neranjan S. Potential inhibitors for novel coronavirus protease identified by virtual screening of 606 million compounds Int. J. Mol. Sci. 2020 21 3626 10.3390/ijms21103626 32455534 21 Fusade-Boyer M. Dupré G. Bessière P. Evaluation of the antiviral activity of Sephin1 treatment and its consequences on eIF2α phosphorylation in response to viral infections Front. Immunol. 2019 10 134 10.3389/fimmu.2019.00134 30809223 22 Grace M.S. Dubuis E. Birrell M.A. Belvisi M.G. Pre-clinical studies in cough research: role of transient receptor potential (TRP) channels Pulm. Pharmacol. Ther. 2013 26 498 507 10.1016/j.pupt.2013.02.007 23474212 23 Hsiao H.M. Sapinoro R.E. Thatcher T.H. A novel anti-inflammatory and pro-resolving role for resolvin D1 in acute cigarette smoke-induced lung inflammation PLoS One 2013 8 e58258 10.1371/journal.pone.0058258 23484005 24 Hsiao H.M. Thatcher T.H. Colas R.A. Resolvin D1 reduces emphysema and chronic inflammation Am. J. Pathol. 2015 185 3189 3201 10.1016/j.ajpath.2015.08.008 26468975 25 Hu X. Shen H. Wang Y. Aspirin-triggered resolvin D1 alleviates paraquat-induced acute lung injury in mice Life Sci. 2019 218 38 46 10.1016/j.lfs.2018.12.028 30571954 26 Imai Y. Kuba K. Neely G.G. Identification of oxidative stress and Toll-like receptor 4 signaling as a key pathway of acute lung injury Cell 2008 133 235 249 10.1016/j.cell.2008.02.043 18423196 27 Islam J. Shree A. Vafa A. Taxifolin ameliorates benzo[a]pyrene-induced lung injury possibly via stimulating the Nrf2 signalling pathway Int. Immunopharmacol. 2021 96 107566 10.1016/j.intimp.2021.107566 33813368 28 Jafarinia M. Hosseini M.S. Kasiri N. Quercetin with the potential effect on allergic diseases Allergy, Asthma, Clin. Immunol. 2020 16 36 10.1186/s13223-020-00434-0 32467711 29 Jia Y. Lee L.-Y. Role of TRPV receptors in respiratory diseases, Biochim. Biophys. Acta Mol. Basis Dis. 2007 1772 915 927 10.1016/j.bbadis.2007.01.013 30 Kandadi M.R. Frankel A.E. Ren J. Toll-like receptor 4 knockout protects against anthrax lethal toxin-induced cardiac contractile dysfunction: role of autophagy Br. J. Pharmacol. 2012 167 612 626 10.1111/j.1476-5381.2012.02040.x 22612289 31 Koarai A. Traves S.L. Fenwick P.S. Expression of muscarinic receptors by human macrophages Eur. Respir. J. 2012 39 698 704 10.1183/09031936.00136710 21885397 32 Krishnamoorthy N. Abdulnour R.E. Walker K.H. Specialized proresolving mediators in innate and adaptive immune responses in airway diseases Physiol. Rev. 2018 98 1335 1370 10.1152/physrev.00026.2017 29717929 33 Kvetkina A.N. Kaluzhskiy L.A. Leychenko E.V. New targets of Kunitz-type peptide from sea anemone Heteractis magnifica Dokl. Biochem. Biophys. 2019 487 260 263 10.1134/S1607672919040033 31559593 34 Liu J.H. Cao L. Zhang C.H. Dihydroquercetin attenuates lipopolysaccharide-induced acute lung injury through modulating FOXO3-mediated NF-κB signaling via miR-132-3p Pulm. Pharmacol. Ther. 2020 64 101934 10.1016/j.pupt.2020.101934 32805387 35 Marčec R. Likić R. Could fluvoxamine keep COVID-19 patients out of hospitals and intensive care units? Croat. Med. J. 2021 62 95 100 10.3325/cmj.2021.62.95 33660966 36 Mousavi S. Moradi M. Khorshidahmad T. Motamedi M. Anti-inflammatory effects of heparin and its derivatives: a systematic review Adv. Pharmacol. Sci. 2015 2015 507151 26064103 37 Nassenstein C. Krasteva-Christ G. Renz H. New aspects of neuroinflammation and neuroimmune crosstalk in the airways J. Allergy Clin. Immunol. 2018 142 1415 1422 10.1016/j.jaci.2018.09.011 30409249 38 Noguchi M. Furukawa K.T. Morimoto M. Pulmonary neuroendocrine cells: physiology, tissue homeostasis and disease Dis. Models Mech. 2020 13 dmm046920 10.1242/dmm.046920 39 Nonomura K. Woo S.H. Chang R.B. Piezo2 senses airway stretch and mediates lung inflation-induced apnoea Nature 2017 541 176 181 10.1038/nature20793 28002412 40 Pakos-Zebrucka K. Koryga I. Mnich K. The integrated stress response EMBO Rep. 2016 17 1374 1395 10.15252/embr.201642195 27629041 41 Panigrahy D. Gilligan M.M. Huang S. Inflammation resolution: a dual-pronged approach to averting cytokine storms in COVID-19? Cancer Metastasis Rev. 2020 39 337 340 10.1007/s10555-020-09889-4 32385712 42 Perrin-Cocon L. Aublin-Gex A. Sestito S.E. TLR4 antagonist FP7 inhibits LPS-induced cytokine production and glycolytic reprogramming in dendritic cells, and protects mice from lethal influenza infection Sci. Rep. 2017 7 40791 10.1038/srep40791 28106157 43 Pierre P. Integrating stress responses and immunity Science 2019 365 28 29 10.1126/science.aay0987 31273112 44 Prescott S.L. Umans B.D. Williams E.K. An airway protection program revealed by sweeping genetic control of vagal afferents Cell 2020 181 574 589 10.1016/j.cell.2020.03.004 32259485 45 Rosen, D.A., Seki, S.M., Fernández-Castañeda, A., et al., Modulation of the sigma-1 receptor-IRE1 pathway is beneficial in preclinical models of inflammation and sepsis, Sci. Transl. Med., 2019, vol. 11, no. 478, p. eaau5266. 46 Rynko A.E. Fryer A.D. Jacoby D.B. Interleukin-1β mediates virus-induced M2 muscarinic receptor dysfunction and airway hyperreactivity Am. J. Respir. Cell Mol. Biol. 2014 51 494 501 10.1165/rcmb.2014-0009OC 24735073 47 Schloer S. Hübel N. Masemann D. The annexin A1/FPR2 signaling axis expands alveolar macrophages, limits viral replication, and attenuates pathogenesis in the murine influenza A virus infection model FASEB J. 2019 33 12188 12199 10.1096/fj.201901265R 31398292 48 Selfridge B.R. Wang X. Zhang Y. Structure-activity relationships of (+)-naltrexone-inspired Toll-like receptor 4 (TLR4) antagonists J. Med. Chem. 2015 58 5038 5052 10.1021/acs.jmedchem.5b00426 26010811 49 Shen X. Wang Q. Lin Y. Voltage-dependent modulation of TRPA1 currents by diphenhydramine Cell Calcium 2020 90 102245 10.1016/j.ceca.2020.102245 32634675 50 Shi C. Tingting W. Li J.P. Comprehensive landscape of heparin therapy for COVID-19 Carbohydr. Res. 2021 254 117232 10.1016/j.carbpol.2020.117232 51 Sukhatme V.P. Reiersen A.M. Vayttaden S.J. Sukhatme V.V. Fluvoxamine: a review of its mechanism of action and its role in COVID-19 Front. Pharmacol. 2021 12 652688 10.3389/fphar.2021.652688 33959018 52 Sun H.M. Hong L.Z. Shen X.K. Antithrombin-III without concomitant heparin improves endotoxin-induced acute lung injury rats by inhibiting the activation of mitogen-activated protein kinase Chin. Med. J. (Engl.) 2009 122 2466 2471 20079161 53 Van Haren F.M.P. Page C. Laffey J.G. Nebulised heparin as a treatment for COVID-19: scientific rationale and a call for randomized evidence Crit. Care 2020 24 454 10.1186/s13054-020-03148-2 32698853 54 Wang L. Yuan R. Yao C. Effects of resolvin D1 on inflammatory responses and oxidative stress of lipopolysaccharide-induced acute lung injury in mice Chin. Med. J. (Engl.) 2014 127 803 809 24571866 55 Wang X. Zhang Y. Peng Y. Pharmacological characterization of the opioid inactive isomers (+)-naltrexone and (+)-naloxone as antagonists of toll-like receptor 4 Br. J. Pharmacol. 2016 173 856 869 10.1111/bph.13394 26603732 56 Wang H. Anthony D. Yatmaz S. Aspirin-triggered resolvin D1 reduces pneumococcal lung infection and inflammation in a viral and bacterial coinfection pneumonia model Clin. Sci. (Lond.) 2017 131 2347 2362 10.1042/CS20171006 28779028 57 Watanabe S. Markov N.S. Lu Z. Resetting proteostasis with ISRIB promotes epithelial differentiation to attenuate pulmonary fibrosis Proc. Natl. Acad. Sci. USA 2021 118 e2101100118 10.1073/pnas.2101100118 33972447 58 Watkins L.R. Wang X. Mustafa S. Hutchinson M.R. In vivo veritas: (+)-naltrexone’s actions define translational importance: a letter in response to Skolnick et al., “Translational” potential of naloxone and naltrexone as tlr4 “antagonists,” Trends Pharmacol. Sci. 2014 35 432 433 10.1016/j.tips.2014.07.002 25109571 59 Van’t Wout E.F. Hiemstra P.S. Marciniak S.J. The integrated stress response in lung disease Am. J. Respir. Cell Mol. Biol. 2014 50 1005 1009 10.1165/rcmb.2014-0019TR 24605820 60 Xia H. Wang J. Sun S. Resolvin D1 alleviates ventilator-induced lung injury in mice by activating PPARγ/NF-κB signaling pathway BioMed. Res. Int. 2019 2019 6254587 10.1155/2019/6254587 31275979 61 Xu M. Zhang Y. Wang M. TRPV1 and TRPA1 in lung inflammation and airway hyperresponsiveness induced by fine particulate matter (PM2.5) Oxid. Med. Cell. Longevity 2019 2019 7450151 10.1155/2019/7450151 62 Yao H. De Boer W.I. Rahman I. Targeting lung inflammation: novel therapies for the treatment of copd Curr. Respir. Med. Rev. 2008 4 57 68 10.2174/157339808783497873 19672474 63 Yeger H. Pan J. Cutz E. Airway Chemoreceptors in the Vertebrates. Structure, Evolution and Function 2019 Boca Raton, FL CRC Press 64 Yildiz-Pekoz A. Ozsoy Y. Inhaled heparin: therapeutic efficacy and recent formulations J. Aerosol Med. Pulm. Drug Delivery 2017 30 143 156 10.1089/jamp.2015.1273 65 Zhuo Y. Zhang S. Li C. Resolvin D1 promotes SIRT1 expression to counteract the activation of STAT3 and NF-κB in mice with septic-associated lung injury Inflammation 2018 41 1762 1771 10.1007/s10753-018-0819-2 30014231 66 Zingoni A. Molfetta R. Fionda C. NKG2D and its ligands: “One for all, all for one, Front. Immunol. 2018 9 476 10.3389/fimmu.2018.00476 29662484 67 Zyryanova A.F. Kashiwagi K. Rato C. ISRIB blunts the integrated stress response by allosterically antagonising the inhibitory effect of phosphorylated eIF2 on eIF2B Mol. Cell 2021 81 88 103 10.1016/j.molcel.2020.10.031 33220178

0

PMC9749646

NO-CC CODE

2022-12-15 23:23:21

no

Biol Bull Rev. 2022 Dec 14; 12(6):649-660

utf-8

oa_other

==== Front Econ Change Restruct Economic Change and Restructuring 1573-9414 1574-0277 Springer US New York 9468 10.1007/s10644-022-09468-3 Article The arbitrage strategy in the crude oil futures market of shanghai international energy exchange Niu Jing 1 Ma Chao 1 http://orcid.org/0000-0003-1782-4551 Chang Chun-Ping cpchang@g2.usc.edu.tw 2 1 grid.464491.a 0000 0004 1755 0877 School of Economics, Xi’an University of Finance and Economics, Xi’an, China 2 grid.412566.2 0000 0004 0596 5274 Shih Chien University, Kaohsiung, Taiwan 14 12 2022 123 30 6 2022 29 11 2022 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This research conducts an empirical study on arbitrage opportunities in the crude oil futures market of Shanghai International Energy Exchange in the period of China’s economic change, 2020–2022. We use the daily closing price data of crude oil futures sc2303 and sc2212 to test whether there is a statistical arbitrage opportunity in China’s crude oil futures market. Taking the most commonly used 12 + 6 rolling window mode in statistical arbitrage, we select several one-year periods that pass the cointegration test during the formation period and apply the optimal opening, closing, and stop loss thresholds based on the highest yield during the formation period to the trading period. Finally, we draw the following conclusions: (i) the pair trading strategy based on the cointegration model is profitable in China’s crude oil futures market; (ii) the 12 + 6 window model can be applied to the pair trading strategy based on China’s crude oil futures. Our research proves the effectiveness of pair trading strategy in China’s crude oil futures market for institutional and individual investors. Keywords Crude oil futures Pair trade arbitrage Cointegration relationship JEL Classification C4 G1 G2 ==== Body pmcIntroduction As an important energy source for human survival, the price of crude oil has been constantly fluctuating. Since the COVID-19 epidemic swept the world in 2020, its price fluctuation has intensified, especially the price of international crude oil futures. The continued spread of the COVID-19 pandemic around the world and the Russian–Ukrainian war are undoubtedly the main reasons for the recent sharp fluctuations in China’s crude oil futures prices (Yousef and Shehadeh, 2020; Yu et al. 2020; Farhad et al. 2021;). The Russian –Ukraine war has affection on the global commodity and stock markets seriously,especially in financial markets. During these time, oil is the safe haven for investors ( Belucio et al. 2022; Diaconaşu et al. 2022).The global economy may even suffer multiple shocks, because of repeated outbreaks (Zaremba et al. 2020; Zhang et al. 2020; Chinazzi et al. 2020; Barbier and Burgess, 2020). Under the continuous influence of the COVID-19 pandemic and the Russian–Ukrainian war, crude oil futures prices may continue to fluctuate, bringing opportunities for statistical arbitrage. With the volatility of global financial markets during the epidemic and the continued decline in investor risk appetite, more and more institutional and individual investors are looking for an investment strategy that can avoid risk and obtain stable returns. The severe volatility of crude oil futures and its derivatives prices has attracted widespread attention from institutional and individual investors (Dutta et al. 2020; Adekoya et al., 2021). Under the influence of COVID-19 and the Russian–Ukrainian war, the trading volume and open interest of China’s crude oil futures not only did not decrease, but instead achieved stable growth (Corbet et al. 2020; Fu et al. 2020; Mensi et al. 2020; Wen et al. 2021; Hu et al. 2022; Wang et al. 2022a; Zhao et al. 2022; Zheng et al. 2022a). Data show from January to February 2022 that the trading volume of medium-quality sour crude oil futures at the Shanghai International Energy Exchange was 6,780,600 lots, or an increase of 654,100 lots compared to the same period in 2021 and a year-on-year increase of 10.68%. At the same time, on May 20, 2022, the single-day open interest of SC2707, the main force of domestic crude oil futures, increased by 0.714%. In a market environment where crude oil futures prices fluctuate violently, a slight misjudgment by both the seller and the buyer can result in huge losses (Wang et al. 2021a; Wang et al. 2022b; Long et al. 2022; Yin et al. 2022a; Zheng et al. 2022c). Therefore, both institutional and individual investors are reluctant to go all out when the market is unstable. Although the price of crude oil futures fluctuates violently, the same characteristics of the subject matter make the price of crude oil futures fluctuate for the same reason, and the occasional price deviation between crude oil futures will be corrected in a timely manner. Based on this, this study aims to explore the neutral investment strategies of individual and institutional investors based on crude oil futures. Paired trading is a commonly used method in statistical arbitrage strategies. After 1978, it has been often used by investment banks and hedge funds and is a market-neutral strategy with weaker risk appetite (Gatev et al. 2006). It is based on the high correlation between two assets or two stocks during market movements (Gloukhov et al. 2014; Miao, 2014). When this correlation is broken and a deviation occurs, this deviation from historical experience will return to its original equilibrium, and so one can sell overvalued assets and buy undervalued assets to try and gain excess profits. Pair trading relies on the deviation of the relative price of certain securities in the short term. Since the relative price will eventually return to a reasonable range in the long run, investors will open positions when the price deviates, hoping to close the position and make a profit when it returns to a reasonable range (see Table 1).Table 1 Summary of the literature review Author (s) Model Period Sample Key-related findings Quan Gu, Xinghui Lei (2018) Statistical arbitrage method: crush arbitrage January 4, 2013, to December 2016 Soybean, soybean meal and soybean oil futures rapeseed, rapeseed meal and rapeseed oil futures Compare with soybean rapeseed crushing arbitrage strategy portfolio returns better Hoffman(2021) Partial co-integration January 1990 to November 2020 Johannesburg Stock Exchange Partial co-integration trading strategy made higher returns during bear cycles compared with bull cycles Li Chen, Guang Zhang(2022) Parametric pairs trading model January 1, 2015, to December 5, 2021 Energy-related securities, including futures, stocks, and ETFs traded in the USA The strategy performed well before COVID-19 but yielded poor results in the pandemic era Nicolas Huck (2009) Paired trading 1999 to 2006 S&P 100 index stocks Paired trading captures positive excess returns Hanxiong Zhang,Andrew Urquhart (2019) Pairs trading January 1996 to July 2017 Mainland China and Hong Kong on highly liquid large‐cap and midcap stocks If investors can trade across Mainland China and Hong Kong, pairs trading is profitable after adjusting for risk and transaction costs, where the annualized abnormal return is 9% Quite a few scholars have studied the profitability of pair trading based on different markets. Based on the US stock market, Gatev et al. (2006) selected 40 years of stock data to study pair trading strategies. Their findings showed that pair trading strategies generate an average annual return of 11%, are low risk relative to just buying stocks, and have limited exposure to known stock risk factors. Since this original paper, the related literature has expanded significantly. Do and Faff (2010) conducted a further study using the pair trading model constructed by Gatev et al. and the same research subjects, extending the analysis interval to 2009, and came to similar conclusions as Gatev et al. Bowen and Hutchinson (2016) proved that the strategy of pair trading got the maximum monthly income in October 1987, and the first five pairs and the first twenty pairs of pair trading returned 46% and 36%, respectively, between 2007 and 2008. They also show that unconditional returns of the strategy do not relate to recognized systematic risk factors. Zhang and Urquhart (2019) switched the research subjects to stocks in the China market and the Hong Kong market, applied the pair trading strategy to both markets, and found that the strategy produced statistically and economically significant net monthly excess returns and net abnormal returns. Based on previous research, this study applies pair trading strategy to China's crude oil futures market and makes the following contributions. First, it proves the effectiveness of the pair trading strategy in China’s crude oil futures market. Second, this paper verifies the effectiveness of the cointegration model based on a 12 + 6 rolling window. The original and pioneering studies of pair trading strategies began from Gatev et al. (2006), whose GGR model has been adopted by most scholars studying pair trading strategies, especially the dynamic rolling window mode of 12 + 6 (that is, the formation period is 12 months and the trading period is 6 months). The 12 + 6 rolling window mode has not been verified in China’s crude oil futures market, and the trading time of China’s crude oil futures is usually more than one year or more, which can be used to test the validity of the 12+6 window mode. Therefore, this paper determines the rolling window is 12+6. Based on the 12+6 rolling window, we apply the daily closing price data of crude oil futures sc2303 and sc2212 from March 2, 2020, to April 29, 2022. During the 12-month formation period, the paired combinations are screened and a pairing trading strategy based on the cointegration model is constructed. Finally, through the verification of the six month trading period of the screening matching, the earnings of the matching trading strategy in the China crude oil futures market of the Shanghai International Energy Exchange were obtained. Verify the effectiveness of China's crude oil futures market, and make a qualitative analysis of all investors' expectations of the market, so that investors can better grasp the market and improve the effectiveness of the market. The guidance of futures on spot prices should be achieved, so as to make the economic operation more stable. Scholars in previous literature have used pair trading strategies to find arbitrage space in the fund market, stock market, stock index futures market, and even the Shanghai 50ETF market. However, few studies have studied the effectiveness of pair trading strategies in China’s crude oil futures market. Considering this situation, this paper takes crude oil futures whose prices have fluctuated violently in the financial market recently as the research object and explores the effectiveness of the pair trading strategy in this market. The purpose of this article is to provide a practical approach to individual and institutional investors in China’s crude oil futures market. More and more investors use the method of pair trading to compare transactions, which could eliminate arbitrage opportunities and improve market efficiency. It is thus beneficial for market makers to provide more stable and reasonable bilateral quotations. Furthermore, this paper has a certain reference value for regulatory authorities to improve supervision efficiency and maintain market order. Finally, with its large number of participants, the improvement in market efficiency, and the improvement in regulatory laws and regulations, China’s crude oil futures trading products can be built into Brent crude oil futures. The remainder of this paper runs as follows. Section 2 provides the literature review. Section 3 introduces the construction of a crude oil futures pair trading model. Section 4 presents the empirical analysis. Section 5 concludes and discusses the policy implications. Literature review Pair trading is divided into three steps: the first is the construction of the pairing pool; the second is to screen out the appropriate paired combination from the pairing pool; the third is to develop appropriate trading strategies. On the study of pairing pool selection, Zhang (2012) selects six industries with high homogeneity to build a matching pool, including real estate, steel, coal, electricity, banking, and automobile. The author chooses homogeneous industries, because of the high degree of similarity in internal operations, products, or services in homogeneous industries, and hence, the price sequence of all enterprises in the industry is more similar. Therefore, it is more conducive to the screening of the stock pair with long-term cointegration. Hu (2013) selects commodity index futures and Shanghai commodity ETF as specific research objects. Allowing short selling on ETF Shanghai commodity stocks, the article proves the effectiveness of the two-span arbitrage strategy in China’s futures market. Most domestic scholars choose two specific research objects in the study of pair trading strategies. According to the brokerage research report, Huang (2015) takes the real estate stocks with high homogeneity in the current industry classification as the research object. She selects 12 large estate stocks to build a matching pool. Based on the assumption that the spread follows the O-U process, it is then proved that an arbitrage strategy based on this process has the advantages of low cost, high income, and low risk compared with the traditional arbitrage strategy based on the cointegration theory. By using the strong correlation between some A stocks and US stocks, statistical arbitrage is carried out. Gatev (2006) studies the performance of pair trading strategies in the US stock market, using all stocks in the market as paired pools, and then, the minimum distance method defined by them is used to screen paired combinations. Perlin (2009) selects 100 stocks with the highest liquidity in Brazil between 2000 and 2006 to construct the matching pool according to the principle of liquidity, which finally proves the effectiveness of the pair trading strategy based on the stocks with better liquidity. Do and Faff (2010) expand the research scope of Gatev et al. (2006) by studying the performance of a pair trading strategy in the US market from 1962 to 2009, selecting all the stocks in the US market to construct the matching pool. Since stocks in the same index generally have many common properties, Bolgun (2010) finds a good matching effect when Turkey ISE-300 index component stocks are chosen as an alternative stock pool. In contrast with the study by Gatev et al. David A. Bowen (2014) selects 767 stocks included in the FTSE full stock index to build a matching pool to remove illiquid stocks and finds that the performance of pair trading in the UK market does not decline over time. As far as the method of selecting paired combinations in the pairing pool is concerned, scholars have developed many different methods in academic or practical terms, including the minimum distance method, cointegration, the random price difference method, and others. In recent years, with the development of science and technology and the extension of theory, some algorithms in machine learning have also been applied to the screening of paired combinations. Han and Chen (2007) select 50 component stocks of the Shanghai Stock Exchange Index to build a matching pool. In the screening of stock pairs, they draw on the idea of cointegration and employ a stepwise regression method to determine the portfolio and pricing subspace. Cui et al. (2011) select the Shanghai Stock Exchange 50 index component shares to construct the matching pool. They use the minimum distance method as the technical means of empirical analysis to select the paired combination part. Considering the cost, there is still a lot of profit from pairing and the profit is less affected by market risk. Wang (2013) conducts a pair study of stocks according to the gap between different stock prices. Via a thorough analysis of the final results, he notes that the method could obtain approximately 1% of the proceeds, and the final data will not fluctuate too much. In addition, the trading method will not be affected by the market environment, so that the safety factor is high. Hu (2016) combines cointegration with the minimum distance method and explores in depth the matching behavior in stocks. Through the analysis of the final conclusion, it is found that the effect of the combination of the two methods is much higher than that of any method. In addition, this method is largely affected by the parameters involved. The trading time should not be too short, and the threshold must fluctuate within the specified range. Whistler (2004) adopts the correlation coefficient method to screen the paired combination based on whether the correlation coefficient is close to positive or negative. The statistical data such as price difference or price ratio and cumulative probability are used to determine the opening or closing of the position. Nath (2003) conducts an in-depth study of the use of pair trading schemes in British bonds. He mainly used the minimum distance method instead of the cointegration method when screening stocks, and explored the opening and closing time in the paper, and finally verified and evaluated the final conclusion through the omega function. Elliott et al. (2005) points out that the price difference can be used to verify the post-pair situation. He arranges the gap between stock prices and constructs a linear space model based on this. After that, if the real spread deviates from the theoretical spread calculated by the model beyond a certain range, then a trading opportunity arises. A relatively perfect theory of distance methods was first proposed by Gatev et al. (2006), which can be called the GGR method. They consider n stocks with data for the first 12 months selected as a training set. The sum of the euclidean distance squared (SSD) of n*(n-1)/2 possible combinations is then calculated in the training set. In the process, the combinations are sorted by SSD from small to large with only the first 20 combinations. Binh (2006) uses the random spread model in the selection of stock pairs to carry out further research. Researchers such as Chang (2009), Perlin (2009), and Faff (2010) have explored pair trading with minimal distance. Through a lot of research and a series of improvement work, Bertram (2010) obtains the optimal solution of price difference from the O-U random process. For the design of a pair trading model, Agarwal (2004) studies the trading signal and sets it according to the degree of the price difference in the paired stocks’ deviation from the historical average level, and the result is that the research object has higher annualized returns. But its shortcomings are also more obvious, that is, the spread range is not static, but changes over time. Gatev et al. (2006) carries out a thorough exploration of statistical arbitrage. The concrete approach first takes the processing of the stock price and then selects the optimal stock pair through the model. On this basis, the sum of squares of the price difference between the two stocks is obtained, and the stock portfolio with the least sum of squares is kept. If the price difference is 2 times higher than the standard deviation, you can start to prepare to establish a position. Through the above conclusions, this method can bring more profits. Many researchers have perfected GGR strategies. For example, Papadakis and Wysocki (2007) reformulates the standard of building a position and finds that the absolute value of the price difference being greater than the pre-set threshold is the new standard of building a position. Engelberg (2009) establishes a ream-skimming closing strategy that closes the position when the profit is greater. Mai and Wang (2014) combine the GGR method with the Herlemont method to construct the FTBD strategy to set up the trading position. The above three schemes are used to pair stocks listed on the exchanges in Shanghai, Shenzhen, and Hong Kong. As fully reflected in the final results, the profits obtained through the FTBD scheme are the largest, and the profits obtained in the stock pairings of the Shenzhen Stock Exchange are much higher than those of the Hong Kong Stock Exchange. In a better economic development market, the effect of matching exchanges is not obvious. Because China has developed its financial sector in just a short period of time, the market has not yet formed a sound system, and relevant policies are not perfect, thus limiting the role of pair trading to a large extent there. Ouyang and Li (2015) study the pair trading strategy in depth, hoping to find an optimal threshold. They apply it to Chinese A + H stocks on the basis of this threshold. Their final exploration results fully show that when the parameters no longer change, the profits obtained through pair trading can be greater. If the relevant parameter values are not stable, then the profits will decrease under the influence of the threshold. Multiple trading activities mean greater costs, which have a great negative impact on profitability. Hu et al. (2017) combine a pair trading strategy with reinforcement learning to build a new trading model with dynamic optimization parameters and apply it to China’s bond market. Compared with the traditional pair trading strategy with fixed parameters, their findings offer a better return. Fu et al. (2019) use the O-U process modeling to design specific transaction models when studying dynamic management of fund assets. Wang et al. (2019) focus on checking the SC crude oil futures intertemporal arbitrage strategy. This paper takes the heteroscedasticity of the spread series into account. We compare EARCH(1,1) with GARCH(1,1) and choose the latter to fit the conditional heteroskedasticity of spread series. Robert B, Pavel T (2020) took the 10 US commodity futures, and by capturing the dynamics of the futures volatility terms structure with three factors, the paper shows that in most markets the slope factor is strongly negative in certain periods and at best only weakly negative in other periods. High inventory levels are found to correspond to flatter volatility term structures in seven futures. As mentioned above, pair trading is divided into three main steps. For the selection of the pairing pool, we use fundamental analysis, or select all securities contained in a securities index, or select related indicators such as liquidity indicators to build the pairing pool. However, from the literature, the selection methods of the pairing pool are much the same and have not been studied. For the screening of paired combinations, the methods are divided into the minimum distance method and cointegration method. The minimum distance method and cointegration method are adopted by most scholars, and their effectiveness is widely proved. The cointegration method is an innovation of the minimum distance method which avoids the disadvantages of a similar trend of paired combination selected by the distance method and cointegration method, such as fewer number of transactions caused by infrequent price fluctuation, long-term position holding, etc (Feng et al. 2021; Wang et al. 2021b; Peng et al. 2022; Wang et al. 2022b; Yang et al. 2022; Yin et al. 2022b; Zheng et al. 2022b, c). Therefore, the screening of paired combination has improved and is more innovative. For the design of a pair trading model, the key lies in seeking the best transaction threshold and trading opportunity. On the basis of determining the threshold, the random price difference method makes use of its good predictive property and takes the initiative to choose the best time to build and close positions, thus maximizing arbitrage profit. By the stochastic control method, the HJB equation is established and the optimal threshold is obtained. However, the shortcomings of these two methods are obvious—that is, the assumptions based on them are harsh and different from reality, and so it is difficult to apply them directly to actual transactions. Therefore, the innovation of this step mainly lies in seeking a more approximate process for price difference or the solution of optimal threshold. In general, a pair trading strategy is a very mature strategy system, and the research covered by the relevant literature has been extremely rich and comprehensive. There are several problems in the existing literature on pair trading. First, when selecting the research target, most of them choose stocks, stock indices, ETFs, and futures, and no literature studies the effectiveness of a pair trading strategy in the crude oil futures market. Based on this, this research studies the effectiveness of a pair trading strategy in the crude oil futures market in order to further improve the relevant literature on pair trading. Second, when selecting the optimal transaction threshold, the traditional research sets the fixed transaction threshold based on experience. In recent years, some scholars have used the random control method to model asset pricing dynamically and obtain the optimal transaction threshold. The disadvantage is that the assumption of price difference is harsh, different from reality, and not universal. Therefore, it is difficult to apply it to practical trading to guide statistical arbitrage. Therefore, this study adopts the dynamic trading threshold. During the formation of the pairing combination, it traverses different positions and stop loss combinations, selects the optimal threshold, and applies it to the trading period, in order to explore whether the dynamic rolling arbitrage strategy of 12+6 in the crude oil futures market has a basis. Construction of a crude oil futures pair trading model As mentioned above, the main steps of pair trading are divided into three steps. For the selection of pairing pool, we can use fundamental analysis or select all securities contained in a securities index or select related indicators such as liquidity indicators to build a pairing pool. For the screening of paired combinations, the methods are divided into minimum distance method and cointegration method. For the design of a pair trading model, the key lies in seeking the best transaction threshold and trading opportunity. Trading pair screening The screening of transaction pairs is during the formation period of pair trading. Pair trading is divided into the formation period and trading period. The formation period refers to the stage of screening paired combination. First, the original futures price sequence is logarithmic in the formation period. Second, the cointegration theory is used to find the paired combination, which passes the cointegration test. Correlation analysis The pair trading should first screen out the futures pairs with the long-term equilibrium relationship of the price trend in the formation period—that is, the price time series of the paired futures should have a high correlation. The famous statistician Pearson has proposed correlation coefficients to measure the close relationship between variables. This study uses the correlation coefficient to measure the correlation degree between the time series of future price. Assuming that the price variables of the two futures are Zt and Yt ,respectively, the correlation between the two futures is expressed by γ with the following calculation method:1 γ=∑i=1n(Zi-Z¯)(Yi-Y¯)∑i=1n(Zi-Z¯)2∑i=1n(Yi-Y¯)2 The correlation coefficient γ between price time series Zt and price time series Yt can be calculated by formula (1). The value is between 0 and 1, and the closer the absolute value γ is to 1, the higher is the correlation between variables. Cointegration test Vidyamurthy pioneered the introduction of the cointegration method into the pair trading strategy to screen initially qualified pair combinations. After the correlation analysis, the cointegration test is conducted on the matching combination with the correlation coefficient above the set critical value, and the eligible paired combination is finally selected. The steps of the cointegration test are as follows. The first step is to test the single integer number of time series Y1t,Y2t,...Ykt. If only two variable sequences are included, then the single integer number of two variables should be the same. The second step is to use the OLS method for the covariant regression and the OLS method for the regression Eq. (2) (also known as the cointegration regression equation).2 Y1t=α+β2Y2t+...+βkYkt+μt The residual sequence (3) is now obtained.3 et=Y1t-α^+β2^Y2t+...+βk^Ykt The third step is to test the smoothness of et If it is smooth, then Y1t,Y2t,...Ykt. pass the cointegration test, and vice versa. Trading strategy Setup of trading signal The basic idea of a pair trading strategy is to use the mean recovery characteristic of a price difference to capture the short-term deviation of such price difference by setting the threshold of trading operation such as opening, closing, and stop loss, so as to obtain the arbitrage income. This study determines the trading signal by the deviation of the logarithmic price difference between the two futures contracts from its long-term mean. Most studies use this method when determining trading signals (see, for example, Gatev et al. 2006; Perlin 2009; Do and Faff 2010; David A. Bowen and Mark C. Hutchinson 2014; Hanxiong Zhang and Andrew Urquhart 2019). Most studies use traditional trading signals, such as the multiple value of the fixed standard deviation of the valence difference sequence. This multiple is often determined by empirical values, usually when the price difference sequence exceeds 0.75 times the standard deviation. According Vidyamurthy (2004), when the residual sequence is normal and the trigger condition is 0.75 times the optimal value of the transaction signal, the earnings can be maximized. In practice, the residual sequence does not conform to the normal distribution, and the financial time series has the characteristics of high peak and thick tail distribution. It would thus be inappropriate to use a zero-fold optimal transaction signal for normal distribution. There is no guarantee that the earnings will be maximized. Scholars choose these parameters according to experience, and most of them take stocks as the research target. This study chooses crude oil futures as the research object. Therefore, the parameters selected according to experience may not be applicable to crude oil futures. The paper adopt dynamic opening and stop loss threshold and use the fluctuation information reflected in the formation period of crude oil futures contract pairing. This study calculates the optimal open position threshold and stop loss threshold to guide the trading of the futures. The method of traversal is used to calculate the optimal parameters by traversing different open positions and stop loss threshold combinations during the formation period. The opening line of position is set between 1.1 and 1.5, the closing line of position is set between 2.1 and 2.5, and the stop loss of position is set between is 0.3 and 0.5. The step size is 0.2. Traversing the operation process is implemented with Python. This article also makes an innovation in trading signals—that is, when the spread suddenly expands from the closing line to above the opening line, the trading strategy will not be to open an position. This is because the “leapfrogging” or sudden change in the spread is likely to be caused by non-systematic risk and the change is inertial. Pair trading is a neutral strategy, which minimizes transaction risk and is more appropriate with pair trading strategies. Long-short position ratio In pair trading there are two commonly used long-short positions: coefficient neutral strategy and capital neutral strategy. A capital neutral strategy means that the initial capital can be 0 without considering the transaction cost. The funds obtained from short selling futures are used to buy a short, and the market risk is small during the holding period. At the same time, the income obtained by closing positions is relatively small. A coefficient neutral strategy refers to the coefficient obtained by using cointegration pairing to construct a linear model as the ratio of a long-short position, and the coefficient neutral strategy is used to construct the ratio of a long-short position strictly according to the regression coefficient. This study uses the cointegration method to construct the matching transaction. Considering that the margin should be paid when the future is bought, the ideal situation of initial fund 0 cannot exist, and so the coefficient neutral strategy is adopted to build the long-short position ratio according to the regression coefficient:4 share\,A∗price\,Ashare\,B∗price\,B=beta Share A and share B are the position quantity of the paired futures contract in Eqs. (4). Price A and price B are the price when the paired futures contract reaches the opening condition. Beta is the coefficient of the model independent variable of the paired combination through a cointegration test in the formation period. Transaction costs The cost in the course of trading is mainly composed of several parts: transaction handling fee is 2 yuan per contract, exercise right handling fee is 2 yuan per contract, exercise right pre-hedging fee is 2 yuan per contract, and trading margin (futures seller). However, futures trading has capital requirements, which are specified as follows: the balance of funds available in the margin account for five consecutive trading days before applying for the opening of the transaction code or trading authority is not less than RMB 100,000 yuan, and so the initial fund of this study is 100,000 yuan. Selling crude oil futures requires a certain margin, and this study also takes the margin into account. The margin calculation formula is the larger value of the following two: Futures contract settlement price * underlying futures contract trading unit + underlying futures contract trading margin—futures contract value *0.5(equal value and real value futures contract has a value of 0). Futures contract settlement price * underlying futures contract trading unit + underlying futures contract trading margin *0.5 In this study, the margin is sufficient—that is, the margin required to sell crude oil futures is always sufficient. In practice, when the position pair expands exponentially, the initial capital also expands accordingly. Evaluation indicators for trading models This research selects the monthly average rate of return and the Sharp ratio to measure the return and risk of the strategy. Pair trading is a neutral trading strategy. Investors who use pair trading often have a low appetite of risk—that is, risk is an important consideration when they choose their portfolio. The Sharp ratio is a classical index to consider benefit and risk as a whole. Therefore, the monthly average yield and Sharp ratio are selected as the model evaluation index. Monthly rate of return Because the paper set up a window mode of 12 + 6, the rate of return calculated by the trading period is the monthly yield rate. After the pair trading is completed throughout the trading period, the monthly yield of the strategy is calculated. Assuming that the total amount after the completion of the trading period is Q1, the initial capital is C, and the formula(5) for calculating the monthly rate of return is:5 r=Q1-CQ1 (2) Sharp ratio This paper uses the Sharp ratio to measure the risk factors of a pair trading strategy. This indicator is considered to be a standardized indicator of performance evaluation in the fund industry, and it considers both the return of the portfolio and the risk of the portfolio. For rational investors in the securities market, when considering investment objects, they will use the Sharpe ratio as a measure, that is, to choose the investment object with the highest return expectation under the same risk level or the investment object with the lowest risk under the same return expectation. Therefore, it is reasonable to believe that when investors choose investment portfolios, they will require that the expected return of the investment object is greater than the return of risk-free assets. The Sharp ratio is designed to calculate the excess return per unit risk of the portfolio. The larger the value of the Sharpe ratio, the higher the return on the selected investment object. The Sharp ratio(6) is calculated as follows:6 Sharp\,Ratio=ERp-Rf∗t250σp Here, ERp is the expected rate of return of the venture portfolio. This study uses the cumulative rate of return of the pair trading portfolio, where t is the trading days of the formation period or trading period and Rf is the risk-free rate of return. In this paper, the risk-free rate is 3.75, and σp is the standard deviation of the return rate. Herein, the standard deviation of the rate of return on previous trades in a pair trading portfolio is used. Empirical analysis Data When selecting data in this paper, the paper first excludes crude oil futures contracts with a duration of less than one and a half years, because when designing the pair trading strategy, the paper selects the 12+6 window mode most used by scholars in the past. The period can characterize the numerical characteristics of the price difference of the future pairing portfolio, and in the 6-month trading period, the price difference will not deviate significantly from this range. Based on this, the paper eliminates a series of crude oil futures contracts with a duration of less than one and a half years, such as sc2003 and sc2109. Second, according to the principle of liquidity, 30% of the crude oil futures contracts after the open interest are excluded, because the pairing trading strategy can only describe the range of the spread well when the liquidity is strong. When the liquidity is poor, the price often cannot reflect the real price. The market situation is highly contingent. Third, excluding the abnormal point of crude oil futures price, the paper stipulate that when the crude oil price changes by more than 50% at some time, it is considered that its price is exhibiting an excessive response to market information, and so it is kicked out. The selected data come from the official website of Shanghai International Energy Exchange, and its annual data are updated to April 29, 2022. This study therefore uses the crude oil futures corresponding to sc2303 and sc2212 when establishing a two-way position and selects the daily closing price data of two crude oil futures contracts for empirical research. Due to some missing data, only data with the same time in the two futures contracts are retained. Matching combination screening Before the correlation analysis, the crude oil futures sc2303 and sc2112 are first divided into 8 pairs of different time periods according to the 12-month formation period, and the last pair is from October 2020 to October 2021. The annual data package of Shanghai International Energy Center is only updated to April 29, 2022 for the time being. In the data processing process, the paper eliminates 30% of the crude oil futures contract data after liquidity and only selects the data of the two crude oil futures contracts with the best liquidity. Table 2 shows different formation and trading periods of the two futures contracts.Table 2 Different formation and trading periods SC2303 and SC2212 Formative period Trading period 2020.3–2021.3 2021.3–2021.9 2020.4–2021.4 2021.4–2021.10 2020.5–2021.5 2021.5–2021.11 2020.6–2021.6 2021.6–2022.12 2020.7–2021.7 2021.7–2022.1 2020.8–2021.8 2021.8–2022.2 2020.9–2021.9 2021.9–2022.3 2020.10–2021.10 2021.10–2022.4 Correlation analysis First of all, in pair trading, crude oil futures contracts whose price trends have a long-term equilibrium relationship should be screened out—that is, the price time series of paired futures should have high correlation. This study chooses the Pearson correlation coefficient to test the correlation between contracts. The paired combinations selected in this paper are sc2303 and sc2212, which are divided into 8 time periods as shown in Table 2. Calculating by Formula (1), correlation tests are carried out for different formation periods, and the time period with the correlation coefficient higher than 0.9 is selected for the next cointegration test. Table 3 shows the correlation coefficient of the price of the paired combination formation period in different time periods. As shown in Table 3, the price correlation coefficients in the formation period of the 8 time periods are all higher than 0.9, which can be tested in the next step.Table 3 Correlation coefficients of different formation periods Ranking Formative period Correlation coefficient 1 2020.3–2021.3 0.999 2 2020.4–2021.4 0.998 3 2020.5–2021.5 0.998 4 2020.6–2021.6 0.997 5 2020.7–2021.7 0.997 6 2020.8–2021.8 0.996 7 2020.9–2021.9 0.995 8 2020.10–2021.10 0.995 It can be seen from Table 3 that the correlation between the prices in the formation period of 8 different time periods is very high, all above 0.99, and so the 8 groups of paired combinations all pass the correlation test. At the same time, Table 3 sorts different time periods according to the size of the correlation coefficient. Therefore, the paired combinations of 8 different time periods pass the correlation analysis. Stationarity and cointegration tests After screening out the matching combination of 8 pairs of futures contracts with correlation coefficient above 0.9, it is necessary to carry out the cointegration test. After that, the paper selects the paired combination that passes the cointegration test during the formation period of 12 months. Taking the first pairing combination as an example, the paper selects the daily closing price data with the formation period from March 2, 2020 to March 2, 2021. The Augmented Dickey–Fuller test (ADF test) results of the first group of original sequences and the first-order difference sequences are displayed in Table 3. It can be seen from Table 4 that the logarithmization of the two futures contracts is a first-order single integer, and so they are tested by the EG two-step method. First, the regression results are sorted as shown in Table 5. According to the regression results, the residual expression is obtained:ui=lnSC2303-0.9405lnSC2112-0.3697 Table 4 Test results of original sequence and first-order differential ADF Variable (after taking logarithm) ADF value 1% critical mass 5% critical mass 10% critical mass P value Stability Log-SC2303 − 2.163 − 3.46 − 2.87 − 2.57 0.220 Not smooth Log-SC2212 − 2.240 − 3.46 − 2.87 − 2.57 0.192 Not smooth First-order difference of SC2303 − 13.554 − 3.46 − 2.87 − 2.57 0.000 Stable First-order difference ofc344 − 18.040 − 3.46 − 2.87 − 2.57 0.000 Stable Table 5 EG first-step results of the two-step approach Variable Coefficient Standard deviation T value P value const 0.3697 0.091 4.057 0 Log-SC2212 0.9405 0.015 61.825 0 Next, EG is the second step of the two-step method, and the residual sequence is ADF tested in Table 6.Table 6 EG results of the second step of the two-step approach Variable ADF value 1% critical mass 5% critical mass 10% critical mass P value Stability ui − 4.423 − 2.58 − 1.94 − 1.62 0.000 Stable Table 6 shows that the ADF value of the residual sequence is less than the critical value at the level of 1%. It is considered that the residual sequence satisfies the stationary test and the two variables have a long-term equilibrium and stability relationship, which adheres with the precondition of paired transaction. Setting up the trading strategy After correlation analysis and cointegration test, the price difference sequence is:Spread=lnSC2303-0.9405lnSC2112 The paper then standardizes the spread and make trading decisions according to the transaction signal. Before this, the optimal parameters are obtained by traversing different combinations of values of open and close positions in Python, taking the yield rate in the formation period as the criterion. Different from the dynamic open position and stop loss threshold in stock pair trading, this study only needs to determine one pair in the formation period. Because all futures contract subjects in the pairing pool are crude oil, the reason for fluctuation is the same. Table 7 shows the statistics of the top five in terms of yield rate in the first pair according to traversal results.Table 7 Searching for optimal parameters Opening threshold Clearance threshold Stop loss threshold Forming period rate of return 1.1 0.5 2.1 30.2% 1.1 0.5 2.3 26.4% 1.1 0.5 2.5 29.7% 1.1 0.3 2.1 28.6% 1.1 0.3 2.3 29.4% The specific trading strategy is as follows. When the spread scale (standardized price difference) is larger than 1.1, the paper shorts the price difference, selling SC2303 and buying SC2212 according to the calculated proportion in a neutral strategy; when the spread scale is smaller than − 1.1, the paper longs the price difference, buying SC2303 and selling SC2212; when the absolute value of the spread scale is larger than 2.1, the paper reverses the operation and stops the loss in time; when the absolute value of the spread scale is less than 0.5, the paper closes the position and obtains earnings from the arbitrage. After obtaining the optimal parameters according to the traversal method, the paper draws the price difference sequences between the formation period and the trading period of the first group by Python. According to Fig. 2, the trading model constructed according to the high frequency data of the formation period performs well. In Figs. 1 and 2, the blue solid line is the closing line, the green dashed line is the building line, and the red dotted line is the stop line Figs. 3 and 4.Fig. 1 Standardized spread sequence of au2006c344 au2006c340 in the formation period Fig. 2 Standardized spread sequence of au2006c344 and au2006c340 in the trading periods Fig. 3 Standardized spread sequence of au2012c432 au2012c440 in the formation period Fig. 4 Standardized spread sequence of au2012c432 and au2012c440 in the trading period Empirical results of pair trading According to the pair trading strategy, the paper back-tests the data of the formation period and trading period of SC2303 and SC2212. Taking the first group of paired combinations as an example, 168 transaction signals are captured in the formation period, and all 168 transactions have achieved positive returns. On the premise of considering the transaction cost and the transaction margin, the yield obtained by the transaction is 65.76%, and the Sharp ratio is 5.96. When the spread deviates from the mean value, it can return to the mean value at a fast rate. The model captures two transactions in the trading period data, and both transactions have positive returns. Considering the transaction cost and the trading margin, the yield or monthly gain rate in one month obtained by the two exchanges is 5.24, and the Sharp ratio is 0.65. Table 8 shows the statistical information on the formation and trading periods of 8 pairs of paired combinations screened. Table 8 shows that, of the 8 pairs, all pass the cointegration test, which is conducted by the ADF test in the Python arch package. The average monthly rate of return for the 8 pairs screened is 37.13%, and the Sharp ratio is 3.72. For each pair of the 8 pairs, they at least once open a position. Of the 8 pairs, one pair generates negative returns during the trading period, and 7 pairs produce positive returns. Table 8 shows the statistical results.Table 8 Rate of return and Sharp ratio of portfolio SC2302 and SC2212 Pair group Forming period rate of return Sharp ratio of formation Trading rate Sharp trading rate ***1 65.76% 5.96 5.24% 0.65 **2 48.26% 7.84 2.68% 4.32 ***3 29.71% 10.09 − 3.49% − 7.11 **4 69.26% 21.65 7.20% 8.33 **5 25.32% 2.02 9.62% 5.77 **6 62.08% 20.01 20.87% 5.33 *7 52.86% 9.21 17.26% 9.45 **8 41.37% 8.72 24.57% 2.09 ***Indicates stability at the 1% level,** indicates stability at the 5% level,* indicates stability at the 10% level Conclusion and policy implications Scholars in previous literature have used pair trading strategies to find arbitrage space in the fund market, stock market, stock index futures market, and even the Shanghai 50ETF market. However, few studies have studied the effectiveness of pair trading strategies in China’s crude oil futures market. Considering this situation, this paper takes crude oil futures whose prices have fluctuated violently in the financial market recently as the research object and explores the effectiveness of the pair trading strategy in this market. The paper selects the futures pool according to the liquidity principle and select 8 pairs of future contracts with a correlation coefficient greater than 0.9 through correlation analysis in period of Chinese economic change 2020–2022. The paper then empirically analyzes the 8 pairs that meet the pair trading conditions as empirical objects. Finally, the average monthly return of the 8-pair portfolio far exceeds the average monthly return of the stock market, reaching as high as 10.51%, with a Sharpe ratio of 3.61. This paper provides the first evidence of the profitability of pair trading strategies in China’s crude oil futures market. Most pair trading with futures adopts the rolling window model of 12 + 6—that is, 12 months to characterize the price range of the research object, and 6 months to implement the trade. The duration of crude oil futures contracts is usually greater than two years, and so the 12+6 rolling window model can be applied. On this basis, this paper selects the paired combination with a cointegration relationship, uses its statistical information to build a pairing trading model with a 12-month formation period, applies it to the trading period, and achieves better returns. It is proved that the 12+6 rolling window model is effective in China’s crude oil futures market. We traverse the formation period of transaction signal setting to obtain the optimal parameters and then apply it to the transaction period. Because the reasons for the changes of crude oil futures in different periods are different, the optimal threshold needs to be determined for the specific formation period. Overall, with full consideration of trading margin, the pair trading strategy achieves an average yield of 49.32% in the formation period, a Sharp ratio of 10.68, a monthly average yield of 10.51% in the trading period, and a Sharp ratio of 3.61. The decline of the Sharp ratio is because the model is built on the formation period and is expected. This study overall confirms the effectiveness of pair trading and the 12 + 6 window model in China’s crude oil futures market. China’s crude oil futures contracts have a natural high correlation, because of the same underlying objects. At the same time, its T+0 trading mechanism makes it possible to quickly open or close positions when the price deviates and returns, compared with the mechanism that can only be operated again on the next day in several stock markets, including A shares. Combined with the strategic asset allocation of crude oil futures and its derivatives under the epidemic situation, this study provides operational guidance for institutional and individual investors. It confirms that pair trading strategies also have arbitrage space in the crude oil futures market. These studies combined indirectly prove that the financial market in China is not efficient. Quantitative trading strategies represented by pair trading have matured in recent years and are expected to become the mainstream risk-neutral trading strategies in the future against the backdrop of large fluctuations in the global financial market. Pair trading not only brings higher investment returns to investors in practice, but also improves the value discovery function and operation efficiency of the market through a large number of arbitrage behaviors in the market. In this paper, we suppose that the initial fund is 100,000 yuan. Actually, in the world of arbitrage one needs to borrow the money from a bank. There exists an interest cost. In our study, we do not consider the interest cost. Interest will be a limitation and future recommendation. Acknowledgements Jing Niu thanks financial support from Shaanxi Provincial Soft Science research project of China (No. 2021KRM088). Declarations Conflict of interest The authors declared that they have no conflicts of interest to this work. We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Adekoya OB Oliyide JA Oduyemi GO How COVID-19 upturns the hedging potentials of gold against oil and stock markets risks: Nonlinear evidences through threshold regression and markov-regime switching models Resour Policy 2021 70 101926 10.1016/j.resourpol.2020.101926 Barbier EB Burgess JC Sustainability and development after COVID-19 World Dev 2020 135 105082 10.1016/j.worlddev.2020.105082 32834381 Belucio M Santiago R Fuinhas JA Braun L Antunes J The impact of natural gas, oil, and renewables consumption on carbon dioxide emissions: european evidence Energies 2022 15 14 5263 10.3390/en15145263 Bowen DA Hutchinson MC Pairs trading in the UK equity market: risk and return Eur J Financ 2016 22 14 1363 1387 10.1080/1351847X.2014.953698 Chen L Zhang G COVID-19 effects on arbitrage trading in the energy market Energies 2022 15 13 4584 10.3390/en15134584 Chinazzi M Davis JT Ajelli M Gioannini C Litvinova M Merler S Vespignani A The effect of travel restrictions on the spread of the 2019 novel coronavirus (COVID-19) outbreak Science 2020 368 6489 395 400 10.1126/science.aba9757 32144116 Corbet S Larkin C Lucey B The contagion effects of the COVID-19 pandemic: Evidence from gold and cryptocurrencies Financ Res Lett 2020 35 101554 10.1016/j.frl.2020.101554 Diaconaşu DE Mehdian SM Stoica O The reaction of financial markets to Russia’s invasion of Ukraine: evidence from gold oil bitcoin and major stock markets Appl Econ Lett 2022 10.1080/13504851.2022.2107608 Do B Faff R Does simple pairs trading still work? Financ Anal J 2010 66 4 83 95 10.2469/faj.v66.n4.1 Dutta A Das D Jana RK Vo XV COVID-19 and oil market crash: revisiting the safe haven property of gold and Bitcoin Resour Policy 2020 69 101816 10.1016/j.resourpol.2020.101816 Elliott RJ Van Der Hoek J Malcolm WP Pairs trading Quant Financ 2005 5 3 271 276 10.1080/14697680500149370 Feng GF Yang HC Gong Q Chang CP What is the exchange rate volatility response to COVID-19 and government interventions? Econ Anal Policy 2021 69 705 719 10.1016/j.eap.2021.01.018 Fu Q Chen YE Jang CL Chang CP The impact of international sanctions on environmental performance Sci Total Environ 2020 745 141007 10.1016/j.scitotenv.2020.141007 32731073 Gatev E Goetzmann WN Rouwenhorst KG Pairs trading: performance of a relative-value arbitrage rule Rev Financ Stud 2006 19 3 797 827 10.1093/rfs/hhj020 Gloukhov VV Ilin IV Koposov VI Levina AI Market risk neutral strategies: modeling and algorithmization Asian Soc Sci 2014 10 24 209 10.5539/ass.v10n24p209 Gu Q, Lei X (2018) The profitability of crush spread: statistical arbitrage method. In: 2018 6th International Conference on Machinery, Materials and Computing Technology (ICMMCT 2018). Atlantis Press, p 249–252 Hoffman AJ Statistical arbitrage on the JSE based on partial co-integration Invest Anal J 2021 50 2 110 132 10.1080/10293523.2021.1886723 Hu H Chen D Fu Q Does a government response to COVID-19 hurt the stock price of an energy enterprise? Emerg Markets Finance Trade 2022 58 1 1 10 10.1080/1540496X.2021.1911803 Hu WW Hu JQ Li Z Zhou JF Adaptive paired transaction model based on reinforcement learning algorithm Manage Sci 2017 02 148 160 Huck N Pairs selection and outranking: an application to the S&P 100 index Eur J Oper Res 2009 196 2 819 825 10.1016/j.ejor.2008.03.025 Long H Chang CP Jegajeevan S Tang K Can Central Bank Mitigate the Effects of the COVID-19 Pandemic on the Macroeconomy? Emerg Markets Finance Trade 2022 58 9 2652 2669 10.1080/1540496X.2021.2007880 Mensi W Sensoy A Vo XV Kang SH Impact of COVID-19 outbreak on asymmetric multifractality of gold and oil prices Resour Policy 2020 69 101829 10.1016/j.resourpol.2020.101829 34173419 Miao GJ High frequency and dynamic pairs trading based on statistical arbitrage using a two-stage correlation and cointegration approach Int J Econ Financ 2014 6 3 96 110 10.5539/ijef.v6n3p96 Ouyang HB Li J A study on optimal thresholding of pairing trading strategy based on co-integration technique Invest Res 2015 34 11 79 90 Papadakis, G., & Wysocki, P (2007) Pairs trading and accounting information. Boston university and mit working paper. Peng XY, Zou XY, Zhao XX, Chang CP (2022) How does economic policy uncertainty affect green innovation? Technol Econ Dev Econ (forthcoming) Perlin MS Evaluation of pairs-trading strategy at the Brazilian financial market Journal of Derivatives & Hedge Funds 2009 15 2 122 136 10.1057/jdhf.2009.4 Robert B Pavel T Environmental impact assessment analysis in the Kahak’s wind farm J Environ Assess Policy Manag 2020 22 2250006 10.1142/S1464333222500065 Wang CF Lin BB Zhu L Paired trading strategies based on stock price differences J Beijing Inst Technol (soc Sci Edition) 2013 01 71 75 Wang (2019) The Test of intertemporal strategy of sc crude oil futures based on GARCH model. Huazhong University of Science and Technology. Wang HJ an K Zheng M Who has done a better job in fighting the COVID-19 epidemic? Left or Right? Emerg Markets Finance Trade 2021 57 8 2415 2425 10.1080/1540496X.2021.1908259 Wang QJ Feng GF Wang HJ Chang CP The impacts of democracy on innovation: Revisited evidence Technovation 2021 108 102333 10.1016/j.technovation.2021.102333 Wang QJ Feng GF Wang HJ Chang CP The influence of political ideology on greenhouse gas emissions Global Environmental Change 2022 74 102496 10.1016/j.gloenvcha.2022.102496 Wang QJ Wang HJ Chang CP Environmental performance, green finance and green innovation: What's the long-run relationships among variables? Energy Econ 2022 110 106004 10.1016/j.eneco.2022.106004 Wen J Zhao XX Chang CP The impact of extreme events on energy price risk Energy Econ 2021 99 105308 10.1016/j.eneco.2021.105308 Yang HC Feng GF Zhao XX Chang CP The impacts of energy insecurity on green innovation: A multi-country study Econ Anal Policy 2022 74 139 154 10.1016/j.eap.2022.01.017 Yin HT, Chang CP, Wang HJ (2022a) The impact of monetary policy on green innovation: Global evidence. Technol Econ Dev Econ (forthcoming) Yin HT Wen J Chang CP Science-technology intermediary and innovation in China: Evidence from State Administration for Market Regulation, 2000–2019 Technol Soc 2022 68 101864 10.1016/j.techsoc.2022.101864 Yousef I Shehadeh E The impact of COVID-19 on gold price volatility Int J Econ Bus Adm 2020 8 4 353 364 Zaremba A Kizys R Aharon DY Demir E Infected markets: novel coronavirus, government interventions, and stock return volatility around the globe Financ Res Lett 2020 35 101597 10.1016/j.frl.2020.101597 32550842 Zhang H Urquhart A Pairs trading across Mainland China and Hong Kong stock markets Int J Financ Econ 2019 24 2 698 726 10.1002/ijfe.1687 Zhang D Hu M Ji Q Financial markets under the global pandemic of COVID-19 Financ Res Lett 2020 36 101528 10.1016/j.frl.2020.101528 32837360 Zhao XX Zheng M Fu Q How natural disasters affect energy innovation? The perspective of environmental sustainability Energy Economics 2022 109 105992 10.1016/j.eneco.2022.105992 Zheng M Feng GF Jiang RA Chang CP Does environmental social and governance performance move together with corporate green innovation in China? Bus Strategy Environ 2022 10.1002/bse.3211 Zheng M Feng GF Wang QJ Chang CP Financial globalization and technological innovation: International evidence Economic Systems 2022 10.1016/j.ecosys.2022.101048 Zheng M, Du Q, Wang QJ (2022c) Nexus between green finance and renewable energy development in China. Emerg Markets Finance Trade (forthcoming)

0

PMC9749647

NO-CC CODE

2022-12-15 23:23:21

no

Econ Change Restruct. 2022 Dec 14;:1-23

utf-8

oa_other

==== Front Glob J Flex Syst Manag Global Journal of Flexible Systems Management 0972-2696 0974-0198 Springer India New Delhi 327 10.1007/s40171-022-00327-8 Original Research IT Capabilities, Strategic Flexibility and Organizational Resilience in SMEs Post-COVID-19: A Mediating and Moderating Role of Big Data Analytics Capabilities http://orcid.org/0000-0002-8782-6782 Wided Ragmoun w.ragmoun@qu.edu.sa 12Ragmoun Wided is an associates professor in College of Business and Economics – Qassim University in Saudi Arabia and affiliated to Faculty of Economic Sciences and Management in Nabeul- Tunisia on Business department. Currently, she has been teaching managerial courses and has engaged in the supervision of research projects for the students in the same department. Her research interests focus on strategic management, entrepreneurship and innovation, with the particular interest to resilience and open innovation in actual circumstances. She published many previous studies in indexed journal as well as some development project. 1 grid.412602.3 0000 0000 9421 8094 Department of Business Administration, College of Business and Economics, Qassim University, P.O. Box: 6640, Buraidah, 51452 Saudi Arabia 2 grid.419508.1 0000 0001 2295 3249 Department of Business Administration, Faculty of Economics and Management of Nabeul, University of Carthage, Carthage, Tunisia 14 12 2022 120 25 5 2022 21 11 2022 © The Author(s) under exclusive licence to Global Institute of Flexible Systems Management 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. This research provides a novel progression to the existing research about big data analytics capabilities (BDAC) by investigating and measuring its influence on organizational resilience and strategic flexibility. Toward that end, 400 different SMEs in Saudi Arabia were approached. Data were collected via questionnaire. Results confirm that the ability to handle big data analytics totally mediates the relationship between IT capabilities and strategic flexibility. Big data infrastructure flexibility has a negative effect on strategic flexibility. Big data personal expertise not only negatively affects the relationship between IT capabilities and strategic flexibility but also stimulates and reinforces the relationship between strategic flexibility and organizational resilience. The critical pathway developed and tested the trend to make the organization as an immune system able to make the best of the worst. This implies the urgent need for policymakers and managers to adopt and comprehend the concept of BDAC instead of IT capabilities to define oriented plans specifically formulated for stimulating strategic flexibility and organizational resilience. By adopting the proposed model, SMEs can interact more effectively internally and externally. Keywords Big data analytics capabilities IT capacity Organizational resilience Strategic flexibility ==== Body pmcIntroduction The impact of COVID-19 pandemic is cruel, and it shakes the whole business activity, especially Small and Medium-sized Enterprises (SMEs), with a fatal impact (Lee & Trimi, 2021). In the post-pandemic world, businesses try to use all their possible and available effort to bring back their prosperity (Ufua et al.,et al. 2022; Zutshi et al.,et al. 2021), especially for SMEs in which corresponding outcomes are still relevant to their abilities to strategically cope and react to turn negative crisis effect into opportunities (Kraus et al. 2020; Wenzel et al. 2020). Many actions are required to get benefits (Angeles et al.,et al. 2022; Davidsson et al.,et al. 2020). Specialized researchers on SMEs continually explore ways to help SMEs relaunch themselves. Based on a systematic literature review, Zutshi et al. (2021) identify three main groups dealing with strategic recommendations for SMEs after COVID-19. The first group focuses on the economic impact of the pandemic on small firms (Hoorens et al.,et al. 2020) and how managers can be prepared to cope with an eventual crisis (Jain et al. 2019). The second group focuses on the current impact of pandemic on SMEs with their different impacts (Bartik et al. 2020) as well as the long-term impact of COVID-19 restrictions on survivability (Bartik et al.,et al. 2020). The third and last group treated various topics, such as resilience in SMEs and its relative importance in anticipating, facing and responding to business challenges (Mokline & Ben Abdallah, 2022). In fact, there is an urgent need for empirical research dealing with a systemic approach to building resilience based on specific components and characteristics of SMEs (Gerald et al.,et al. 2020; Hadi et al.,et al. 2020; Martinez-Lozada & Espinosa, 2020) to maintain productivity and resolve financial difficulties (Nyanga & Zirima, 2020). Organizational resilience is considered the first and ultimate option to survive in a turbulent and uncertain environment (Duchek, 2020; Mallak & Yildiz, 2016; Utz, 2020). It can be considered as a proactive attribute (Conz et al. 2017; Ragmoun & Almoshaigeh, 2020), absorptive -adaptative capability (Gray & Jones, 2016), reactive attribute (Branicki et al. 2018) or a dynamic attribute (Asamoah et al. 2020; Halkos and Skouloudis, 2019), which is developed before the event (or crisis) and occurs during the event and after the event. Independently, there is a consensus that is proactively thinking about organizational resilience is still the only best way to survive and face future and present crises in such a turbulent environment (Gorjian Khanzad & Gooyabadi, 2021; Mokline & Ben Abdallah, 2021; Settembre-Blundo et al. 2021). Proportionally, being resilient for SMEs, especially during and after COVID-19, represent a great challenge due to their limited internal resources compared to large organizations (Del Vecchio et al. 2018). In fact, many factors  inhibit the development of organizational resilience for SMEs such as the in adequation of staff training, weak planning, and inconsistent relationships between control mechanisms, management and operating units also can cause a low ability to scan the general context, all of this reduces SMEs’ capacity to change and to react to changes in turbulent environments (Figueiredo et al. 2020). Previous studies have pointed out some determinants of new practices to manage environmental uncertainty and change (Chesbrough, 2020), such as stakeholders’ involvement, co-creation activities, acquisition of dynamic capabilities or establishment of inter-organizational knowledge exchanges. However, the definition of principal key levers that generate a continuous adaptation and support resilience is still changing in the context of SMEs (Gorjian Khanzad & Gooyabadi, 2021). In this line of ideas, Martinez-Lozada and Espinosa (2020) argue that there is a need for additional empirical studies to understand and deal with different relationships between specific components to generate asset resilience in SMEs according to a systemic approach. Also, the emerging literature on resilience for SMEs is still limited and lacks knowledge about identifying an appropriate set of strategic responses to survive during and post-pandemic (Fitriasari, 2020; Liguori & Pittz, 2020). A critical and pragmatic pathway for resilience in SMEs is still inexistent. Therefore, this research aims to fill this gap identified in previous studies and enrich knowledge in this field of organizational resilience by investigating and appreciating the impact of IT capability on organizational resilience while examining the mediating and moderating effect of big data analytics capabilities (BDAC) on strategic flexibility as determinants of organizational resilience. The research also outlines the critical and pragmatic setting for OR enabled by different factors according to an integrative and dynamic approach to shaping organizational resilience. The research’s purpose is important, and its contribution depends on the critical importance of SMEs on economic structure in different countries across the globe as a stimulator for socioeconomic development at global and local levels (OECD, 2017). Concretely, the paper clarifies, defines and appreciates conditions that enhance SMEs’ chances to survive based on a dynamic capability approach (DC). On the one hand, the theoretical paradigm of DC, as defined by Teece and Pisano (1994), can facilitate an understanding of how organizations acquire competitive capabilities by adopting new technologies, such as big data analytics capabilities (Ragmoun, 2022). Therefore, we answer the urgent call for future and additional research on the opportunities associated with using digital technologies for SMEs and their benefits in this case (Ragmoun and Alwehabie, 2020; Zutshi et al. 2021). On the other hand, under such conditions, strategic flexibility seems important and can maintain survival. It is admitted that organizations which adopt strategic flexibility may stay resilient and agile during turbulent moments (Uzoma Ebubechukwu and Edwinah, 2022). Resilient organizations should be able to inculcate flexibility when defining their strategy to adapt to change and to move ahead. Shimizu and Hitt (2004) state that strategic flexibility signifies identifying the most important environmental shifts and responding immediately to those changes using actual resources to define a new course of action. In this sense, strategic flexibility represents an organizational capability allowing a quick understanding of what is happening to move quickly and respond to the environment using the most opportune sources (Dehghan-Dehnavi and Nadafi, 2010). In this context, Information and communication technology (ICT) can make the collecting data process easier for managers and policymakers to gather efficiently and quickly data, but, with the greater amount of data needed for a decision, analyzing, managing and using information seem to be not easy and big data analytics becomes more appropriate (Jum’a et al. 2022). According to the research question dealing with how SMEs can maintain and develop resilience when faced with a crisis, a mediating analysis was conducted to measure and investigate the interrelationships among IT capabilities, BDAC and strategic flexibility. A moderating analysis using BDAC was also adopted to identify how adopting such capabilities can reinforce or slow down the development of resilience capabilities. Firstly, the present research has developed a theoretical framework for studying and testing the association between information technology capabilities (ITC), BDAC, strategic flexibility and organizational resilience. This integrative approach has not been evaluated in previous literary articles dealing with their effects on SMEs via a dynamic capabilities perspective. Secondly, this research expressively seeks to enrich IT literature by defining guidelines for OR on SMEs, especially in such turbulent environments and avoid the negative scenario of COVID-19. Thirdly, this research assists information technology managers in mater OR with its different dynamic aspects and capabilities within their organizations and from empirical perspectives. Based on the above, research prevents answering three main questions related to OR via BDAC on SMEs:Is there an eventual direct relationship between ITC and OR? Is there an eventual direct relationship between ITC and strategic flexibility using BDAC? Does BDAC moderate the effect of strategic flexibility on OR? This means that the proposed model comprises two main loops: The first deals with a mediating effect of BDAC, and the second is related to the moderating effect of this variable to amplify and stimulate the development of OR on SMEs. The manuscript is organized as follows. The authors define the theoretical perspective and detail the literature in the first section of the manuscript. Then, the literature on four main variables, IT capabilities, big data analytics capabilities, strategic flexibility and organizational resilience, are discussed in the second section. Consequently, Sect. 3 deals with hypotheses construction based on the existent literature. Section 4 presents methodology and data collecting process. Before concluding in Sect. 6, the findings are discussed in terms of theoretical and empirical implications added to limitations in Sect. 5. Theoretical Background A Dynamic Capability View on SMEs Capability-based View of Resilience In this research, the main objective remains the definition of a pragmatic and operational way to achieve and define resilience. As mentioned in the beginning, there is a consensus according to which how and to what resilience can be designed remains unclear (Ulz et al.,et al. 2020). For this purpose, it is admitted that resilience is represented by a dynamic interaction between environment and organization (Williams et al.,et al. 2017). It is not a curative process adopted if needed but a preventive one, and resilience capabilities are associated with a toolkit available for organization survival if needed. It must be developed, maintained, managed and updated. The adoption of such an approach seems to be more operational. Resilience is considered a meta-capability (Duchek, 2020) that depends on organizational capabilities. Such a conception allows us to underline the dynamism of resilience (Burnard & Bhamra, 2011) pressed by the dynamic interaction of organization (internal) with the environment (external) to identify and adapt to changes and be more flexible. Although, as a meta-competence, resilience is represented first as a process of three main stages, the interconnection and the continuity between them require information flow. Second, it considers the internal working of merged, associated and combined capabilities to develop resilience. IT capabilities regain importance in this state because it facilitates communication, integration and alignment of capabilities and resources (Bharadwaj, 2000). Organizational Resilience (OR) Organizational resilience is, at the same time, an ability to adapt and respond (Distel, 2017; Kahn et al.,et al. 2018), a capacity to react adequately to unexpected events or threats to survive (Lengnick-Hall et al.,et al. 2011; Ortiz-de-Mandojana & Bansal, 2016) and a meta-capability (Duchek, 2020). Despite its importance, supported by most studies, there is little consensus on its essence (Utz, 2020) and its determinants. To summarize these different conceptions, we can refer to the dimensions enumerated by Ramezani and Camarinha-Matos (2020). They assimilate resilience into an umbrella, which contains the same time: the capacity to rebound and recover, the capability to sustain a desirable and positive state and the capacity to focus on persistence. Resilient organizations can always take advantage and chances under any circumstances (Aldianto et al. 2021). Translated to our context of research (SMEs), resilience will be considered a successful adaptation to maintain business by generating, acquiring and combining external and internal knowledge as resources to explore, perceive and adopt rapid changes in its environments. Drawing from these investigations, resilience can be considered a dynamic capability that guarantees that it can respond effectively and rapidly to environmental change and define in three dimensions or three main capabilities: anticipation, coping and adaptation capabilities (Duchek, 2020). Given the variety of existent theoretical perspectives on resilience, the authors chose these aspects that are commonly shared and respond to the main objective of the research. Strategic Flexibility (SF) With reference to Holweg (2005), flexibility represents the capacity to adapt to external and/or internal stimuli. Escrig-Tena et al. (2011) associate flexibility with the capacity to respond effectively and quickly to different challenges to satisfy environmental changes and demands. For Osita-Ejikeme and Amah (2022), flexibility is an innate ability to rethink and rearrange to accommodate and adapt to the environment successfully. Strategic flexibility (SF) is recognizing and accepting environmental dynamics to define new effective responses to these dynamics (external/ internal) (Dehghan-Dehnavi and Nadafi, 2010). For Zahra et al. (2008), strategic flexibility is the degree of change a business can adopt to adjust its strategy according to opportunities, changes and external threats (Zahra et al. 2008). The association between proactivity and reactivity deserves our attention because we must remember here that SF outmoded a simple reaction to prevent subsequent action and reduce risks. In this sense, strategic flexibility can be considered one of the most determinant critical assets of a successful organization by maintaining competitive advantage (Arshad et al. 2018; Johnson et al. 2003; Xiu et al. 2017), establishing success (Wadstrom, 2019; Xiu et al. 2017), surpassing inactiveness (Zhou & Wu, 2010), reallocating resources as needed and required (Sanchez, 1995) to bring creativity and innovation (Li et al. 2010). All of these will positively affect organizational performance (Brozovic, 2016). Information Technology Capabilities (ITC) IT capabilities are defined as the ability to use and explore IT-based resources combined with other capabilities and resources to enhance a variety of key performance indicators (Bharadwaj, 2000). Other researchers consider that ITCs are the ability to implement a set of a variety of common platforms and manage them (Lu & Ramamurthy, 2011). There is consensus on which ITCs are developed and adopted to process, collect, retrieve and store information (Basheer et al.,et al. 2016; Galliers et al.,et al. 2020; Zhen and Hu, 2008). Three main dimensions of IT capabilities are detailed in the existing literature: IT infrastructure capability (ITIC), proactive IT capability (PITC) and capability to align IT (AITC) (Lu & Ramamurthy, 2011). Information technology infrastructure capability represents the base to share and process information. It is a collection of human and technical services coordinated and budged by the management (Weill & Ross, 2004). This capability is directedly related to operational processes, such as company applications and sharing services and products in different locations, to take advantage of eventual synergetic opportunities on business lines (Bharadwaj, 2000; Lioukas et al. 2016). The IT alignment capability supports the integrative processes between IT and other functional departments. This can help organizations to exploit and visualize IT resources that can contribute to achieving organizational strategic objectives. All this process is based on managing and planning using technological architectures to face future and current challenges (Bharadwaj, 2000; Chen & Tsou, 2012; Wade & Hulland, 2004). Proactive IT capability is related to exploring technological resources to maximize business opportunities created in the market (Cepeda & Arias-Pérez, 2018). This enables organizations to anticipate new trends raised from technological developments and to exploit all opportunities created by emerging technologies (Weill & Ross, 2004). With proactive IT, the organization can operate and establish innovations through quick restructuring and reconfiguration of functional processes (Agarwal & Sambamurthy, 2002; Lu & Ramamurthy, 2011). Big Data Analytics Capability (BDAC) Recently, Big data as a concept has been at the forefront of most recent discussions in management research (Lombardi, 2019). It permits the broad manipulation of a significant amount of data. It offers a heterogeneous and large amount of information that it simply approaches (Yin & Kaynak, 2015) and is distinct in terms of volume, value, variety, velocity and veracity (5 V) (Wamba et al.,et al. 2017). This research focuses on the analytical capabilities of big data that focus on the incorporation and management of big data rather than on its technological and computational infrastructure aspects (Gandomi & Haider, 2015; Lozada et al. 2019). BDA capability refers to an organizational management’s ability to deploy and use big data resources for strategic aims, develop competitive advantage and create value (AlNuaimi et al. 2021; Wamba et al. 2017). The existing literature identifies three main axes of BDAC (managerial, personal and infrastructure), such as tangible and intangible (Gupta & George, 2016). Tangible is related to infrastructure and resources, as well as human resources represented by managerial and technical skills for big data. Intangible resources include organizational learning and a data-driven culture (Gupta & George, 2016). At the same time, BDA is technology, application, practices, methodology and techniques, with the ultimate objective of treating and analyzing data to make the decision (Gandomi & Haider, 2015). Rialti et al. (2020) argue that BD management capabilities fix the right BDA infrastructure, execution, and selection. Provost and Fawcett (2013) explain that the decision-maker must use specific skills to extract the best technical solution and manage information. This mechanism can stimulate strategic flexibility and resilience because it guarantees adaptability and survival. BDA personnel are considered essential to the organization despite its position on the organizational hierarchy (Rialti et al. 2020). It is provided by analytic skills related to the data collection and treatment to maintain integrity while changing or adapting (Wamba et al. 2017) and is still relevant to scientific, analytic, and architectural skills dealing with technological infrastructure and datasets (De Mauro et al. 2018). BDA infrastructures represent technical Information system (IS) available and used to collect, store, process, and analyze big data with its different types to facilitate data flow in every situation (Wamba et al. 2017). It is supposed that this capability is flexible because it should be adaptative to handle much more data or storage capacity (Wang et al. 2018). Hypotheses Development The Mediation Role of BDA Capabilities IT Capabilities and BDAC GUPTA and George (2016) identified some specific factors for the development of BDAC qualified as tangible and intangible. In this state, technical skills are considered a tangible human resource to build BD, defined as specific know-how to use new and emerging forms of technology to explore and extract intelligence from BD. As defined at the beginning of this paper, IT capability refers to an organizational ability to deploy and mobilize IT resources combined with other capabilities and resources (Bharadwaj, 2000). The literature presents BDA as a new generation of architectures and technologies conceived to economically extract added value through a large volume of data (Mikalef et al.,et al. 2018). Based on Resources Based View (RBV) and recent studies on BDA, BDAC can be defined as an organizational ability to deploy technology (Mikalef et al. 2020). On a practical level, IT strategists are frequently admitted to being concerned with the availability and quality of the data analyzed (Brinkhues et al. 2014). If data, traditionally provided by IT and managed by ITC, are a core resource, it will be important to provide an infrastructure able to store, share and analyze data (Mikalef et al. 2020). Besides, this can be guaranteed by using BDAC. Some research considers BD novel technologies that are able to handle many fast-moving and diverse data (GUPTA & George, 2016). Therefore, BDA is derived from IT resources (Agrawal, 2013), which are uncertain about the adoption and application of BDA (Rahman and Zhao, 2020). Based on this brief analysis, the interdependence can be admitted between ITC and BDAC on theoretical and empirical levels. H1 IT capacities have a positive impact on big data analytics capacity. IT Capacities and Strategic Flexibility Strategic flexibility can be defined as the ability to respond to uncertainties based on skills and information for continual development (Eryesil et al. 2015). When there is information, we will need information technologies to collect it. Chen et al. (2017) support that IT is a determinant element of strategic flexibility. It permits the organization to be automatized and improve its operational efficiency (Bhatt & Grover, 2005). IT associated with strategic flexibility ensures not only an operational and tactical impact (Chen et al.,et al. 2017) but also an instrumental role in supporting strategy and the organization’s relationships with its partners and customers (Bharadwaj et al.,et al. 2013). Through the development of computing capacity and information processes, IT can assist the organization in entering new markets and satisfying consumers rapidly and adequately. With reference to the different dimensions of ITC detailed below, it is admitted that IT alignment refinement strategies seem important for profitability and competitive advantage (Aydıner et al. 2018) and for avoiding wasted resources (Ravishankar et al. 2011). With IT, coordination and knowledge acquisition becomes easier and more reliable. IT capabilities as an ability can help managers assimilate knowledge developed via information and, consequently, be able to survive (Grover & Saeed, 2007). IT, in the organization, provides analytics and information to help them innovate and enter new markets (Chen et al. 2017). It increases coordination and assists in the dissemination of operational information between organizations and suppliers with efficiency (Kotabe et al. 2011). IT can also enhance collaborative spirit among all organization units to support R&D and respond effectively and rapidly to customer’s needs (Nabeel-Rehman & Nazri, 2019). H2 IT capacities have a positive impact on strategic flexibility. BDAC, Strategic Flexibility, and IT Capacities WAMBA et al. (2017) argue that BDA is associated with strategic flexibility and impact organization. This contribution can be represented by its capacity to enable and support the capacity of managers to monitor data use and its related process to plan performance and workflow appropriately (Akter et al.,et al. 2016). Using BDA, an organization can monitor not only competitors as a determinant of their performance and operations (Erevelles et al.,et al. 2016) but behavioral customer patterns are also analyzed and managed in real time, on different levels (Hofacker et al.,et al. 2016). Overall, the majority of emerging research on BDA agrees that, as an analytic tool, it assures coordination between many large socio-economic databases (George et al.,et al. 2014) and permits the organization to navigate, manage and adapt to the business environment (Wamba et al.,et al. 2017) being more dynamic and reactive to change to enhance agility, market responsiveness and dynamic capabilities (Rialti et al.,et al. 2020; Ryabchikov & Ryabchikova, 2022). Wamba and Mishra (2017) demonstrate that BDA is also analytically related to the capacity to analyze reality or existence and automatic to reduce the time allocated to decision-making and adaptative. In addition, ITC helps identify weaknesses and strengths of business strategy (Awamleh & Ertugan, 2021; Rajesh, 2017). Enabling ITC makes sense of what is happening in an external environment to define the appropriate process based on incoming to improve the external and internal environment (Chu et al. 2019). Also, IT capability literature admits that the ability to deploy and mobilize IT resources distinguishes an organization from its competitors and is considered a source of competitive advantage (Ravichandran & Lertwongsatien, 2005). Undependably, Awwad et al. (2022) demonstrate that IT capabilities positively and significantly affect organizational agility via a dynamic capability approach. An additional need for strategic flexibility requires more internal and/or external information; BDAC seems to be more appropriate to deal with a significant amount of information in real time to unsure flexibility. H3 Big data capability mediates the impact of IT capacities on strategic flexibility. The Moderation Role of BDA Capabilities The capacity of an organization to change technology and adapt consumer orientation with dynamism according to environmental demands determines its ability to predict continuous, systematic and rapid evolutionary adaptation to maintain and gain competitive advantage and survive (Onyokoko & Needorn, 2021). The adoption of integrated platforms, which are a form of BDA, contributes to providing means and forecast change for efficient and effective organizational responses by stimulating the development of flexible processes, the definition of real-time connectivity, and collaboration between all external and internal stakeholders (Ashrafi et al.,et al. 2019; Xie et al.,et al. 2022). O'Leary (2013) supports that big data provide a considerable and enormous amount of data that can be unstructured or structured but also available immediately, every time, and everywhere. This largely meets our needs for resilience. In this case, the generated data are distinctive, generated with speed and characterized by their dimensions and potential to provide valuable information (Wamba et al.,et al. 2017). Consequently, this huge amount of information flow in big data makes the decision more real, based on facts and evidence rather than a simple managerial intuition (Ferraris et al.,et al. 2019; Rialti et al.,et al. 2020), to generate flexibility. Proportionally, many scholars have recently supported that Big Data (BD) is one of the main sources of competitive advantage and performance (Côrte-Real et al. 2016; LaValle et al. 2011; Morabito, 2015). By managing and collecting determinant and efficient market-related information, big data support opportunities to fit customer needs and sustain competitive advantage (Côrte-Real et al. 2016). The process of adopting Big Data Analytics (BDA) provides resilience and has become more and more determinant. It is the best alternative to traditional information systems, which can limit resilience due to their rigid structure, as admitted by previous research (Ciampi et al. 2018). Nabeel-Rehman and Nazri (2019) demonstrate that heavy investment in IT capabilities contributes to the sustainability of the firm’s competitive advantage, improving information and knowledge flow in organization and inter-organization. Rathina et al (2019) argue that resilience depends on the awareness of factors and resources that can impact organizations. Mohamed and Singh (2012) demonstrate that IT is fundamental to sustaining, growing and supporting business. Mazini (2014) insists on the relative importance of aligning strategic objectives, information and technological resources to face adversity and controlling data and information to assist decision-making. In a recent report published by the World Bank (2019), ICT is considered a critical factor for resilience based on the development of a specific framework to manage disasters in Japan in two areas: Disaster Information Management System (DIMS) and Early Warning System (EWS). In conclusion, this report presents ICT as a solution to resilience and invites practitioners from other countries to find the best way to explore ICT for resilience. In this state, it must be remembered that BDAC is an advanced version of ITC, providing different kinds of information with significant volume and realism. H4 Big data capability moderates the link between strategic flexibility and organizational resilience capacities. The research model represented in Fig. 1 shows that Big Data Analytics capabilities created from IT capacities (alignment and integration) positively impact strategic flexibility. The model also shows a moderating effect of BDAC between strategic flexibility and resilience because it is supposed that more reliable and valid information at the right time (or specifically in the brief time) can accelerate the reaction process to guarantee the adequacy between SF and resilience. So, the organization can adapt and survive appropriately (within its strategic goals). As shown in Fig. 1, BDAC is appreciated in three dimensions, IT capabilities in two dimensions and resilience capabilities in three dimensions. Each dimension can ensure a specific role in developing resilience capabilities. The authors adopted such a subdivision of concept to break down in detail the underlying mechanism for resilience and bring clear responses to its mystery.Fig. 1 Research model Methodology Procedure and Sample To collect data, a questionnaire was performed and sent by e-mail due to the need for social distancing imposed by COVID-19. A list of industrial firms in Riyadh and Qassim was selected based on data extracted from https://modon.gov.sa/. The authors tried to resend the e-mail as needed to get the answer. This process took two months. Four hundred responses were collected, representing 85% of the total questionnaires. The e-mail was addressed to HRM direction, general managers or the IT department. A random sample technique was adopted to select SMEs object of the research. According to previous specialized research in structural equation modeling, the appropriate range of the adequate sample size is from 30 to 460 to be meaningful compared to the number of associations between sample size and parameters (Hoyle & Gottfredson, 2015). The definition of sample size when using SEM depends on many factors such as model complexity, normality and missing patterns (Wolf et al.,et al. 2013a, 2013b). But, the majority of researcher recommend at least 200 which means 5 or 10 cases per parameters (Kline, 2011). In our case, 10 parameters are tested. Most recent studies recommend that small sample sizes are enough. It can be ranged from 30 (four indicators or latent variables and loadings level around 0.80) up to 350 for mediating model (Sideridis et al. 2014; wolf et al. 2013a, 2013b). According to Kline (2011), a typical sample size when using SEM is about 200 cases. So, we can admit that the size adopted here can be representative. Measurement Instruments The scales used were extracted from existing literature. In cases where scales are unavailable, the authors use dimensions, descriptions or domains provided in previous studies to establish scales. Appendix A is used to summarize the adopted scales and supporting literature. To measure the construct of IT capabilities, the researchers adopted two dimensions: IT integration and IT alignment. Four items were used to measure IT integration adopted from the scale developed by Rai and Tang (2010), which insists on partners' relative importance in data and information. The dimension of IT business alignment was appreciated by five items from the works of Kearns and Lederer (2003) and oriented planification on both internal and external. Strategic flexibility was assessed on a six-item scale defined by Zhou and Wu (2010). Resilience capabilities were measured by five items extracted from previous research and considered the most common indicators for resilience capacity as a variable (Lengnick-Hall et al. 2011; Siebert & Gaskin, 2006; Sila, 2007): employee empowerment (EE), employee resilience training (ERT), employee involvement (EI), employee capacity to adapt changes (ECC) and teamwork employee capacity (TEC). It is argued that an organization is less or more resilient depending on its internal capacity to treat, analyze and make a decision as needed and communicated by the external environment. The main scale of BDAC within the existing literature considers it as a multidimensional concept appreciated through three main dimensions as defined by Wamba et al. (2017) in their BDAC model developed based on information technology and information system through the resource-based view. The present research adopted the scale developed by Ramadan et al. (2020) because it fits the main research interest and corresponds to the definition of BDAC adopted in this case. It is a composite scale extracted from different research. Four items are presented; each represents the most commune aspect mentioned in the existent literature (Kim et al. 2012; Upadhyay et al. 2020; Wamba et al. 2017). The majority of indicators were appreciated by a Likert scale ranging between 1 and 7 (1 = disagree completely and 7 = agree completely). The complete instrument is presented in Appendix A. The constructed survey was pre-tested with 15 respondents from faculty members and doctoral students. Feedback was used to refine items if needed, and corrections were made according to different recommendations in the instrument. Analytical Tools Structural equation modeling using AMOS 24 was employed to interpret and analyze the proposed model among the research variables (IT capabilities, BDAC, strategic flexibility and organizational resilience). This choice is related to this function’s ability to estimate, specify, validate and assist the research model. Initially, the goodness of fit was appreciated via the validation and development of the associations among corresponding observable variables as well as heir measurement (indicators and factors). Consequently, data were submitted to the structural equation modeling (SEM) to examine and appreciate interrelations among different endogenous variables after the descriptive statistics used to identify general charachteristics of our sample (Table 1).Table 1 Descriptive statistics Factors Proportion of the sample (N = 400) (%) Business industry sector Construction industry 38.5 Energy industry 32 Pharmaceutic industry 10 Food Industry 12.5 Manufacturing industry 7 Profile HRM direction 42.2 General manager 11.5 IT department 28.2 Others 9.8 Non-specified 8.2 IS size  < 50 42 51–70 39.2 71–90 8.8 91–110 5.8 111–200 4.2 Data Analysis and Results Confirmatory Factor Analysis As a first step, the scale was validated based on a factor analysis by SPSS 16. For each construct, the authors tested the loadings and reliability. The convergent and discriminant validity were calculated too. An adequate convergent validity signifies that the items are highly loaded onto the construct. The corresponding research model was constructed, and scales and items were adopted. Loading’s value and correlations issued from this analysis were used to verify convergent validity, discriminant validity, and internal consistency of scales and items. Table 2 summarizes the main results related to scale validation. To be accepted, the minimum required for item loading must be higher or equal to 0.50. As shown, most items were accepted, and the rest with a low loading were deleted to obtain a purified scale. Moreover, t values are revised to verify the significant of loadings. It is admitted that lodgings are accepted at p < 0.001; on this level, a high convergent validity is confirmed. Values on the diagonal indicate the average variance extracted (AVE) level between scale item and its relative construct.Table 2 Validation of scales and descriptive analysis Constructs and loadings a Number of items Means SD ICR 1 2 3 4 5 6 IT capabilities 9 4.981 1.231 0.811 0.843 Big data analytics management capabilities 4 4.654 1.266 0.975 0.232 0.765 Big data analytics infrastructure flexible capabilities 3 4.214 1.376 0.884 0.254 0.354 0.750 Big data analytics personal capabilities 4 4.132 1.054 0.876 0.228 0.378 0.434 0.876 Strategic Flexibility 6 4.665 1.298 0.966 0.298 0.401 0.399 0.334 0.771 Resilience capabilities 6 4.287 1.116 0.877 0.266 0.411 0.433 0.441 0.601 0.813 aThe significance of the item loadings was assessed using bootstrapping. The t values for all item loadings were significant, at least at the p < 0.001 level Correlations between constructs are indicated. The variance shared and the correlations are used to appreciate the validity of the discriminants. In this case, the square root of the AVE must be larger than the correlations between the constructs. As can be seen, all diagonal values are greater than all values on the offline diagonal, and this confirms the acceptance of discriminant validity for all constructs. As mentioned at the beginning of this part, reliability related to scale validation was also appreciated, and the internal consistency ratio (ICR) was calculated. To accept the reliability of constructs, ICR has to be greater than 0.70. In this case, all values satisfy this level, which means that all constructs are accepted. At the end of this step, the scales are tested and purified to be used in structural model analysis. Structural Model Analysis The structural model was tested using AMOS 24. In the proposed model, resilience three constructs were appreciated by dimensions due to their complexity and multidimensionality as defined in the literature. Resilience capabilities are modeled as a dependent variable and appreciated by three dimensions based on Duchek (2020): anticipation, coping, and adaptation capabilities. BDAC is modeled as a mediating-moderating construct with three main sub-constructs. ITC as the independent variable with two main dimensions and finally SF as a unidimensional construct issued from ITC affect directedly resilience capabilities. Table 3 details loadings and weights for the constructs; as can be seen, all values are significant. Table 4 presents the square root of average variance between indicators and constructs, as well as correlations. It is clear that all indicators are accepted: ICR, convergent validity, and discriminant validity. It is argued that BDAC mediates the relationship between ITC and SF to develop resilience capabilities, and in this case, only two dimensions are used. The third dimension of this construct, BDA related to infrastructure, moderates the link between SF and RC. According to the recommendation of Baron and Kenny (1986), the authors calculated an interaction value to test the moderating effect and integrated it into the proposed model as a latent variable.Table 3 Loadings and weights Latent variables Dimensions Items Weights Loadings T-value KMO IT capabilities IT integration ITI1 1.00 1.00 – 0.66 ITI2 0.344 0.701 58.443 ITI3 0.401 0.676 38.676 ITI4 0.422 0.649 24.667 IT alignment ITA1 1.00 1.00 – 0.69 ITA2 0.377 0.562 13.342 ITA3 0.398 0.642 12.568 ITA4 0.378 0.589 11.564 ITA5 0.403 0.679 9.766 Big data analytics capabilities BDA management capabilities BDAM1 1.00 1.00 – 0.84 BDAM2 0.278 0.766 16.876 BDAM3 0.289 0.707 15.433 BDAM4 0.302 0.728 15.401 BDA infrastructure flexible capabilities BDAI1 1.00 1.00 – 0.89 BDAI2 0.377 0.758 22.453 BDAI3 0.387 0.774 21.657 0.546 8.772 BDA personal expertise capabilities BDAC1 1.00 1.00 – 0.85 BDAC2 0.661 0.728 11.767 BDAC3 0.650 0.858 11.988 BDAC4 0.643 0.738 9.967 Strategic flexibility SF1 1.00 1.00 – 0.79 SF2 0.466 0.872 3.767 SF3 0.454 0.777 3.987 SF4 0.473 0.795 6.878 SF5 0.481 0.814 6.899 SF6 0.440 0.628 4.541 Resilience capabilities Anticipation capabilities AC1 1.00 1.00 – 0.71 AC2 0.336 0.801 6.989 AC3 0.388 0.798 7.056 Copying capabilities COP1 1.00 1.00 – 0.69 COP2 0.232 0.743 15.765 COP3 0.269 0.703 15.877 Adaptation capabilities ANT1 1.00 1.00 – 0.80 ANT2 0.331 0.805 4.891 ANT3 0.353 0.823 4.766 Table 4 Correlation, AVE (Average variance extracted) and internal consistency ratio Constructs ICR 1 2 3 4 5 6 1 IT capabilities – 1.00 2 BDA management capabilities 0.899 0.429 0.885 3 BDA infrastructure capabilities 0.961 0.344 0.644 1.00 4 BDA personal capabilities – 0.289 0.289 0.865 0.848 5 Strategic flexibility 0.862 0.331 0.531 0.664 0.234 1.00 6 Resilience capabilities – 0.120 0.553 0.432 0.443 0.653 0.766 In the collected sample, one of the most important results is the multidimensionality of strategic flexibility. Two dimensions emerge from the analysis with a respective variance of 36.15% and 27.22%, contrary to the definition adopted here. It is supposed that those two dimensions can be represented by the strategic level and flexibility. This concept requires more attention and must be appreciated differently as a composite latent variable. Figure 2 represents results related to the mediating effect. All fit index of the structural model is detailed in Table 5. The model tested explains 34.4% of the variance in IT capabilities, 43.4% of the variance in big data analytics, 38% of strategic flexibility and only 28% of resilience capabilities. Besides, this supposes the existence of other factors that can support resilience or mechanisms to maximize joint effects reached in this research.Fig. 2 Structural model results Table 5 Fit index of structural model (mediating hypothesis) Fit index RMSEA NFI NNFI CFI IFI RFI GFI AGFI Value requires  ≤ 0.08  ≥ 0.9  ≥ 0.9  ≥ 0.9  ≥ 0.9  ≥ 0.9  ≥ 0.9  ≥ 0.9 Value 0.071 0.971 0.911 0.921 0.933 0.917 0.974 0.937 Table 6 Significance for structural model and path coefficients Hypothesized paths Path coefficient and significatively Hypothesis Base model Mediated model ITC → BDAM 1.00 (7.965) 0.932 (6.544) H4 supported ITC → BDAI – – H5 rejected—Non-significative ITC → BDAC 0.36 (3.342) 0.22 (2.458) H6 supported BDAM → BDAI 0.90 (5.132) BDAI cannot be supported directedly by ITC ITC → SF 0.63 (3.878) 0.54 (3.773) H1 supported ITC → RC 0.21 (2.982) 0.19 (2.913) H3 supported BDAM → SF 0.11 (2.061) 0.09 (1.877) H7 supported BDAI → SF 0.68 (7.043) 0.87 (5.772) H8 rejected BDAM → RC 0.20 (2.879) This value will be reported to the moderating hypothesis SF → RC 0.76 (8.113) 0.63 (5.564) H2 supported BDAI → RC 0.19 (2.011) 0.11 (2.322) H7 supported The path coefficients indicate that the majority of hypotheses are supported. As hypothesized, ITC has a strong and positive impact on strategic flexibility (0.63, t = 3.878), which supports the majority of researchers cited below. Examining the impact of ITC on resilience capabilities, it is evident that there is a positive effect, but it seems to be less important than the impact of ITC on strategic flexibility. This result also confirms the interest in the mediating and moderating effects. Proportionally, the effect of SF on RC is positive and strong enough (0.76). This supposes that to be resilient, organizations must develop strategic flexibility that depends on ITC. There is a joined effect of SF and ITC on resilience capabilities. According to the sample, it is confirmed thatThe direct effect of ITC on resilience was positive and significant but reduced with the integration of SF The same direct effect ITC-RC is ameliorated by the integration of BDAM and reduced by BDAI with a negative effect BDAI stimulate the link between ITC and RC but not the link ITC-SF BDAM support the effect of ITC on SF BDAC reduce the effect of ITC on SF Independently, the positive effect of ITC on BDAC is positive and significant. The biggest data analytic capabilities developed by information technology are related to the management capabilities, which demonstrate the friability of such technology to treat, analyze and collect data for real-time decision-making. Moderating Effect To test hypotheses, the authors ran a new model in which a new variable named interaction is added after standardizing all variables. Using SPSS 16, the researchers computed values of SF, RC and BDACI and extracted the new database on AMOS 24. The findings confirm the moderation hypotheses. BDACI moderates the effect of SF on CR. The relationship between the three variables was significant and positive. A partial moderating effect was detected here. This supposes the existence of other possible variables which can contribute or stimulate this effect, such as organizational cultural. Figure 3 represents results related to the moderating effect. The model after the integration of the new variable appreciated by the interaction between SF and BDACI seems to be representative, and all fit indexes are acceptable.Fig. 3 Moderating effect Table 7 summarizes the different fit indices of the new structural model, and as can be seen, all values are significant. The structural model is well represented with the interaction value introduction. So, it confirms the moderating effect of BDAC on the link between strategic flexibility and resilience capabilities.Table 7 Fit index of the structural model Fit index RMSEA NFI NNFI CFI IFI RFI GFI AGFI Value requires  ≤ 0.08  ≥ 0.9  ≥ 0.9  ≥ 0.9  ≥ 0.9  ≥ 0.9  ≥ 0.9  ≥ 0.9 Value 0.078 0.901 0.925 0.930 0.908 0.941 0.921 0.910 Discussion Synthesis of Findings With reference to the first question of this study, it is found that BDA capability is a catalyst which can enhance resilient organizational activities and awareness of possible changes in the business and strategic management by anticipating, adapting and coping. Findings indicated that the effect of ITC on strategic flexibility in the actual circumstance of post-COVID-19 was less than the effect of BDAC. Dynamic capabilities approach adopted for IT and BDA was positively associated with resilience, but their corresponding impact seems indirect. Based on the second research question, the findings demonstrate that BDAC was a fundamental variable in recreating the effect of strategic flexibility on organizational resilience with a total moderating effect. Finally, in the third question pertaining to the direct effect of ITC on strategic flexibility, the findings indicated that IT capabilities can stimulate strategic flexibility under specific conditions (Oberoi et al.,et al. 2007; Tallon, 2008). Therefore, organizations adopting and defining BDAC on management, personal and infrastructure levels should stimulate strategic flexibility, which generates and guarantees the development of organizational resilience automatically as well as enterprise performance management (Akhtar & Mittal, 2010; Dey et al.,et al. 2019). The empirical implications section details the managerial and practical implications of interventions stemming from the findings. In an attempt to answer research questions regarding the mediating and moderating effect of BDAC, this research has presented both practical and theoretical implications. First, it confirmed viewpoints on how BDAC, ITC and strategic flexibility improve and influence organizational resilience if developed under a dynamic capability perspective. This converges with Singh et al. (2021) who detailed this approach using a bibliometric analysis. BDAC amplify the impact of ITC on strategic flexibility. Thus, strategic flexibility enables organizational resilience, and as stronger the use of BDAC is high and adequate, the level of organizational resilience generated is higher. Theoretical Implications As mentioned earlier, the dynamic capability view (DCV) has gained great importance in the IS field as a conceptual perspective to explain competitive advantage development in complex and turbulent environments (Steininger et al.,et al. 2021). In alignment with that, and based on such a theoretical approach, the findings of this research explain how and when organizational capabilities can be considered an organizational competence to manage, understand and prevent turbulence (Li and Chan, 2019). Further, this research expands the main scope of DCV by including implicit and new underlying forces (the mediation – moderation of BDA capability) to the resilience path of SMEs. This amalgamation of BDA capabilities with dynamic capabilities will guide and assist SMEs in harmonizing and aligning their external environment with threats and opportunities. The main challenge of contemporary SMEs lies in the perfect and speed alignment of dynamic and operational capabilities (Zighan et al.,et al. 2021). The core role of DCV is to study organizational competitive advantage development in increasingly turbulent environments (Teece & Pisano, 1997). Transposed to IT, DCV introduces some new concepts that are considered fruitful for IT business value because it can help to explain how organizations can renew and develop some value-generating mechanisms within the means of information’s technology (Schryen, 2013). This can complement and enrich existing knowledge in the field of IT business by identifying specific abilities or processes enabled by the use, mobilization and deployment of IT, adding to the generation of many performance outcomes (Melville et al. 2004; Schryen, 2013). Based on this, DCV seems helpful in explaining how IT can be considered a privileged strategic driver for organizational change in high-velocity environments (Galliers et al. 2012). Therefore, this research belongs to the few empirical studies which devoted attention to the significance of IT dynamic capabilities to sustain and confirm that IT can assist an organization in increasing its strategic values and improving organizational resilience (Arunima et al. 2016; Sumant, 2005). This is still also available for BDA (Grover & Kar, 2017; Kushwaha et al. 2021). This research provides additional opportunities to adopt further empirical research about IT dynamic capabilities, big data analytic capabilities and organizational resilience such as Singh et al. (2019). It is still one of the pioneering types of research that details and investigates the impact of DCV from an organizational resilience perspective on SMEs and points out its importance as well as the urgent need for further research in this field to understand, apply, use and handle IT dynamic capabilities on organizational resilience if supported by a DCV of BDA. Further, this research recommends that BDAC should be viewed and considered a primordial strategic element due to its positive impact on strategic flexibility, the main source of organizational resilience. It is claimed that a BDA management capability is an important aspect of reinforcing the effect of ITC and strategic flexibility. Based on findings, it is obvious that BDA management capability is the major factor in developing organizational resilience in direct and indirect aspects. Practical Implications This research makes several practical and pragmatic contributions. Empirical evidence argues that IT capabilities coupled with BDA capabilities could enable resilience and witness the importance of investing in the development of ITC as well as the introduction of BDA. The findings support the viewpoint related to the positive effect of BDA on resilience (Rialti et al.,et al. 2020) and the fact that BDA can constitute a competitive advantage for achieving organizational resilience (Vossen, 1998). Many senior executives admit the strategic value of IT tends to consider IS activities as basic and suppose that it can reduce costs (Ravichandran, 2018), and it must be revisited or planned to achieve cost-cutting. The result supposes that IT capabilities and BDA are valuable for strategic flexibility and resilience, but developing such capabilities must be revisited. Organizations must consider investing in training programs to acquire and improve IT capabilities. Such capabilities must be maintained, and it is a cumulative learned process that can take time and must be managed. This point addresses another critical aspect compared to the findings: the internal process of resilience. Coordination, complementarity and sustainability of the development of IT capabilities are not independent or separated processes. It is a collective learning process which means that a systemic approach must be considered to maintain resilience. Knowledge, information, time, and IT develop and sustain resilience. There is a synergetic effect that must be maintained as long as possible to achieve organizational resilience. It is admitted that there is not a one-way process but an interactive one. IT capabilities by integration and alignment, alimented by BDA on the term of data and information in the right time by the right way, facilitate the strategic decision-making through an equilibrium intern/extern to establish strategic flexibility and consequently resilience because in such case, the organization will be able to adapt and continue according to the external need but, also, create an ideal internal solution. In addition, this process can be repeated as much as needed to provide organization expertise, and every time, its IT capabilities are reinforced and can be automized by BDA to develop an immune system at the end. An adequate managerial agenda can make this dream reality through an appropriate strategic plan defined in parallel with the organizational strategic plan. Limitations and Future Research Independently, several limitations can be identified in this research. Initially, SMEs used in this study belong to different sectors, which might define some limitations in terms of time effect or contextual factors affecting causality associations between research variables. Further, data collected and used were gathered throughout only one country, this can limit the significance of results, and a multicultural approach can be more benefic in generalizing findings. The findings revealed that an IT dynamic capability added to BDAC could develop strategic and resilience capabilities. Hence, they call for much more empirical and theoretical investigation in this research domain. Further, the combination of IT capabilities, BDA capabilities, strategic flexibility and organizational resilience is complex. It connotes linear and nonlinear mechanisms that have to be identified, defined and measured to reinforce the critical pathway for organizational resilience in SMEs. That is said, the current model identified and tested can be expanded and enriched to include additional moderating factors such as organizational culture or size that can enrich our understanding of mechanisms that foster organizational resilience. Conclusion This research details a new progression to the existing research effort exploring the impact of ITC on organizational resilience through strategic resilience and BDAC in SMEs in Saudi Arabia. More specifically, this study has adopted the perspective of DC to explain the significance of OR on SMEs. In fact, the study supposes that the dynamic capabilities of ITC play a capital role in identifying and amplifying the value of strategic flexibility, which needs to be assisted by BDAC. In the contemporary environment, SMEs have to be equipped with ITC and BDAC required to develop strategic flexibility (Rialti et al.,et al. 2020). Toward that end, a sample of SMEs in Saudi Arabia was obtained using a questionnaire to collect data. The research findings indicated significant and positive associations among information technology capabilities, big data analytic capabilities, strategic flexibility and organizational resilience. More precisely, strategic flexibility positively and significantly mediated the relationship between IT capabilities and organizational resilience. This implies an additional need for managers to define and understand the concept of strategic flexibility in order to design appropriate plans to boost organizational resilience (Onyokoko & Needorn, 2021; Uzoma Ebubechukwu and Edwinah, 2022). This also involves adaptation, copying and anticipation capabilities (Duchek, 2020). With big data analytics capabilities, organizations can share internal and external knowledge, especially through integration and analysis processes (Merhi & Bregu, 2020), to improve environmental response and optimize adaptation processes to achieve dynamism (Božic and Dimovski, 2019). Appendix A Constructs Dimensions Items Supporting literature Information technology capabilities (ITC) IT Integration Our firm transfers data with our partners Rai and Tang (2010) Our firm provides a seamless connection between our partner systems and our systems Our firm easily aggregates relevant information from our partner databases Our firm easily accesses data from our partners IT Alignment Our firm’s IT plans to reflect the business plan goals Kearns and Lederer (2003) Our firm’s IT plans to support business strategies Our firm’s IT plans recognize external business environment forces Our firm’s plans refer to IT Plans Our firm’s plans refer to specific information technologies Strategic flexibility Respond to changes in aggregate consumer demand Thomas (2014) React to new product launches by competitors Introduce a new pricing schedule in response to changes in competitor’s prices Expand into new regional or international markets Change (i.e., expand or reduce) the variety of products available for sale. Adopt new technologies to produce better, faster and cheaper products Wamba et al. (2017) Big Data Analytics Capabilities BDA management capabilities BDA planning BDA decision-making BDA coordination BDA control BDA Infrastructure flexibility BDA connectivity Wamba et al. (2017) BDA compatibility BDA modularity BDA personal capabilities expertise BDA technical knowledge Wamba et al. (2017) BDA Technology management BDA business knowledge BDA relational knowledge Resilience capabilities (RC) Adaptation capabilities Observation Duchek (2020) Identification Preparation Copying capabilities Accepting Duchek (2020) Developing Implementation solutions Anticipation capabilities Reflection Duchek (2020) Learning Organizational change Acknowledgements The authors are very thankful to all the associated personnel in any reference that contributed to the purpose of this research. Funding This research was not funded by any person neither by any public nor private body. Declarations Conflict of interest The authors declare no competing interest. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References Agarwal R Sambamurthy V Principles and models for organizing the IT function MIS Quarterly Executive. 2002 1 1 1 25 Agrawal S Paleoclimatic, paleovegetational and provenance change in the Ganga Plain during the late Quaternary Journal of Earth System Science. 2013 122 4 1141 1152 10.1007/s12040-013-0332-9 Akhtar M Mittal RK Enterprise performance management and flexibility: Factors of effectiveness in upstream oil industry in India Global Journal of Flexible System Management 2010 11 4 1 10 10.1007/BF03396589 Akter S Wamba SF Gunasekaran A Dubey R Childe SJ How to improve firm performance using big data analytics capability and business strategy alignment? International Journal. of Production Economy. 2016 182 113 131 10.1016/j.ijpe.2016.08.018 Aldianto L Anggadwita G Permatasari A Mirzanti IR Williamson IO Toward a business resilience framework for startups Sustainability 2021 13 6 1 19 10.3390/su13063132 AlNuaimi B Khan K Mian M Ajmal M The role of big data analytics capabilities in greening e-procurement: A higher order PLS-SEM analysis Technological Forecasting and Social Change 2021 169 120808 10.1016/j.techfore.2021.120808 Angeles A Perez-Encinas A Villanueva CE Characterizing organizational lifecycle through strategic and structural flexibility: Insights from MSMEs in Mexico Global Journal of Flexible Systems Management 2022 23 2 271 290 10.1007/s40171-022-00301-4 Arshad, D., Zakaria, N., Abdul-kadir, K., & Ahmad, N., (2018). Linking between flexibility and SMEs performance: does improvisation matter?, Paper Presented at the 2nd International Conference on Technology, Innovation, Society and Science-to Business (ICTIS), Asamoah JKK Owusu MA Jin Z Oduro FT Abidemi A Gyasi EO Global stability and cost-effectiveness analysis of COVID-19 considering the impact of the environment: Using data from Ghana Chaos, Solitons & Fractals 2020 140 110103 10.1016/j.chaos.2020.110103 32834629 Ashrafi M Acciaro M Walker T Magnan G Adams M Corporate sustainability in Canadian and US maritime ports Journal of Cleaner Production. 2019 220 386 397 10.1016/j.jclepro.2019.02.098 Awamleh F Ertugan A The relationship between information technology capabilities, organizational intelligence, and competitive advantage SAGE Open 2021 11 2 1 14 10.1177/21582440211015201 Awwad AS Ababneh OMA Karasneh M The mediating impact of IT capabilities on the association between dynamic capabilities and organizational agility: The case of the Jordanian IT Sector Global Journal of Flexible Systems Management 2022 23 3 315 330 10.1007/s40171-022-00303-2 Aydıner A Tatoglu E Bayraktar E Zaim S Delen D Business analytics and firm performance: The mediating role of business process performance Journal of Business Research. 2018 96 228 237 10.1016/j.jbusres.2018.11.028 Baron RM Kenny DA The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations Journal of Personality and Social Psychology 1986 51 6 1173 1182 10.1037/0022-3514.51.6.1173 3806354 Bartik, A. W. , Bertrand, M. , Cullen, Z. B. , Glaeser, E. L. , Luca, M. , & Stanton, C. T. (2020). How are small businesses adjusting to COVID‐19? Early evidence from a survey. NBER Working Paper No. w26989. Basheer A Hugerat M Kortam N Hofstein A The Effectiveness of Teachers' Use of Demonstrations for Enhancing Students' Understanding of and Attitudes to Learning the Oxidation-Reduction Concept Eurasia Journal of Mathematics Science and Technology Education 2016 13 3 555 570 10.12973/eurasia.2017.00632a Bharadwaj A A resource-based perspective on information technology capability and firm performance: An empirical investigation MIS Quarterly. 2000 24 1 169 196 10.2307/3250983 Bharadwaj A Sawy OAE Pavlou PA Venkatraman N Digital business strategy: Toward a next generation of insights MIS Quarterly. 2013 37 2 471 482 10.25300/MISQ/2013/37:2.3 Bhatt GD Grover V Types of information technology capabilities and their role in competitive advantage: An empirical study Journal of Management Information Systems 2005 22 253 277 10.1080/07421222.2005.11045844 Božič K Dimovski V Business intelligence and analytics use, innovation ambidexterity, and firm performance: A dynamic capabilities perspective The Journal of Strategic Information Systems. 2019 28 4 101578 10.1016/j.jsis.2019.101578 Branicki L Sullivan-Taylor B Livschitz S How entrepreneurial resilience generates resilient SMEs International Journal of Entrepreneurial Behavior & Research. 2018 24 7 1244 1263 10.1108/IJEBR-11-2016-0396 Brinkhues, R., Maçada, Antonio C., Casalinho, G., (2014). Information Management Capabilities: Antecedents And Consequences Completed Research Paper. 20th Americas Conference on Information Systems, AMCIS 2014. Brozovic D Strategic flexibility: a review of the literature International Journal of Management Reviews 2016 20 1 1 29 Burnard K Bhamra R Organisational resilience: Development of a conceptual framework for organisational responses International Journal of Production Research. 2011 49 18 5581 5599 10.1080/00207543.2011.563827 Cepeda J Arias-Pérez J Making an organization more flexible and agile: The significance of information technology and open innovation Strategic Direction 2018 34 11 23 25 10.1108/SD-08-2018-0175 Chen JS Tsou HT Performance effects of IT capability, service process innovation, and the mediating role of customer service Journal of Engineering and Technology Management - Journal of Engineering and Technology Management. 2012 29 1 71 94 Chen Y Wang Y Nevo S Improving strategic flexibility with information technologies: Insights for firm performance in an emerging economy Journal of Information Technology 2017 32 10 25 10.1057/jit.2015.26 Chesbrough HW To recover faster from COVID-19, open up: Managerial implications from an open innovation perspective Industrial Marketing Management 2020 88 410 413 10.1016/j.indmarman.2020.04.010 Chu Y Chi M Wang W Luo B The impact of information technology capabilities of manufacturing enterprises on innovation performance: Evidences from SEM and fsQCA Sustainability 2019 11 21 5946 10.3390/su11215946 Ciampi F Cillo V Fiano F Combining Kohonen maps and prior payment behavior for small enterprise default prediction Small Business Economics 2018 54 4 1007 1039 10.1007/s11187-018-0117-2 Conz E Denicolai S Zucchella A The resilience strategies of SMEs in mature clusters Journal of Enterprising Communities: People and Places in the Global Economy 2017 11 1 186 210 10.1108/JEC-02-2015-0015 Côrte-Real N Oliveira T Ruivo P Assessing business value of Big Data Analytics in European firms Journal of Business Research. 2016 70 379 390 10.1016/j.jbusres.2016.08.011 Davidsson P Recker J von Briel F COVID-19 as external enabler of entrepreneurship practice and research BRQ Business Research Quarterly 2020 24 3 214 223 10.1177/23409444211008902 De Mauro A Greco M Grimaldi M Ritala P Human resources for Big Data professions: A systematic classification of job roles and required skill sets Information Processing & Management. 2018 54 5 807 817 10.1016/j.ipm.2017.05.004 Dehghan-Dehvani H Nadafi Q Can strategic flexibility effective in firms through product innovation? Modern Economy and Business 2010 4 1 4 Dey S Sharma RRK Pandey BK Relationship of manufacturing flexibility with organizational strategy Global Journal of Flexible System Management 2019 20 3 237 256 10.1007/s40171-019-00212-x Distel AP Unveiling the Microfoundations of Absorptive Capacity: A Study of Coleman's Bathtub Model Journal of Management 2017 45 5 2014 2044 10.1177/0149206317741963 Duchek S Organizational resilience: A capability-based conceptualization Business Research. 2020 13 215 246 10.1007/s40685-019-0085-7 Erevelles S Fukawa N Swayne L Big data consumer analytics and the transformation of marketing Journal of Business Research. 2016 69 2 897 904 10.1016/j.jbusres.2015.07.001 Eryeşil K Esmen O Bedük A The role of strategic flexibility for achieving sustainable competition advantage and its effect on business performance World Academy of Science, Engineering and Technology 2015 9 10 587 593 26388980 Escrig-Tena AB Segarra-Ciprés M García-Juan B Beltrán-Martín I The impact of hard and soft quality management and proactive behavior in determining innovation performance International Journal of Production Economics 2011 200 1 14 10.1016/j.ijpe.2018.03.011 Ferraris A Mazzoleni A Devalle A Couturier J Big data analytics capabilities and knowledge management: Impact on firm performance Management Decision. 2019 57 8 1923 1936 10.1108/MD-07-2018-0825 Figueiredo R Ferreira JJM Silveira RG Villarinho AT Innovation and co-creation in knowledge intensive business services: The spinner model Business Process Management Journal 2020 26 4 909 923 10.1108/BPMJ-10-2019-0424 Fitriasari F How do small and medium-sized enterprises (SME) survive the COVID-19 outbreak? Jurnal Inovasi Ekonomi. 2020 5 2 53 62 10.22219/jiko.v5i3.11838 Galliers RD Leidner DE Simeonova B Agarwal R Sambamurthy V Principles and models for organizing the IT function Strategic Information Management 2020 10.4324/9780429286797-11 Gandomi A Haider B Beyond the hype: Big data concepts, methods, and analytics International Journal of Information Management 2015 35 137 144 10.1016/j.ijinfomgt.2014.10.007 George G Haas MR Pentland A Big data and management Academy of Management Journal. 2014 57 2 321 326 10.5465/amj.2014.4002 Gerald E Obianuju A Chukwunonso N Strategic agility and performance of small and medium enterprises in the phase of Covid-19 pandemic International Journal of Financial, Accounting, and Management 2020 2 1 41 50 10.35912/ijfam.v2i1.163 Gorjian Khanzad Z Gooyabadi A Development of strategic resilience framework for small businesses Post-COVID-19 Businesses. 2021 1 127 141 10.3390/businesses1030010 Gray G Jones MD A qualitative narrative policy framework? Examining the policy narratives of US campaign finance regulatory reform Public Policy and Administration 2016 31 3 193 220 10.1177/0952076715623356 Grover P Kar AK Big data analytics: A review on theoretical contributions and tools used in literature Global Journal of Flexible Systems Management 2017 18 3 203 229 10.1007/s40171-017-0159-3 Grover V Saeed K The impact of product, market, and relationship characteristics on interorganizational system integration in manufacturer-supplier dyads Journal of Management Information Systems. 2007 23 185 216 10.2753/MIS0742-1222230409 Gupta M George JF Toward the development of a big data analytics capability Information and Management 2016 53 8 1049 1064 10.1016/j.im.2016.07.004 Hadi MS Lawey AQ El-Gorashi TE Elmirghani JM Patient-centric HetNets powered by machine learning and big data analytics for 6G networks IEEE Access 2020 8 85639 85655 10.1109/ACCESS.2020.2992555 Hofacker CF Malthouse EC Sultan F Big data and consumer behavior: Imminent opportunities Journal of Consumer Marketing. 2016 33 2 89 97 10.1108/JCM-04-2015-1399 Holweg M The three dimensions of responsiveness International Journal of Operations & Production Management. 2005 25 7 603 622 10.1108/01443570510605063 Hoorens S Lucy H Clement F How small businesses are coping with the impact of COVID-19: Results from a survey in Turkey and the Middle East 2020 RAND Corporation Hoyle RH Gottfredson NC Sample size considerations in prevention research applications of multilevel modeling and structural equation modeling Prevention Science 2015 16 987 996 10.1007/s11121-014-0489-8 24752569 Johnson JL Lee RPW Saini A Grohmann B Market-focused strategic flexibility: Conceptual advances and an integrative model Journal of the Academy of Marketing Science. 2003 31 74 89 10.1177/0092070302238603 Jum’a L Ikram M Alkalha Z Do companies adopt big data as determinants of sustainability: Evidence from manufacturing companies in Jordan Global Journal of Flexible Systems Management 2022 23 3 1 17 10.1007/s40171-022-00313-0 Kahn WA Barton MA Fisher CM Heaphy ED Reid EM Rouse ED The geography of strain: Organizational resilience as a function of intergroup relations The Academy of Management Review 2018 43 3 509 529 10.5465/amr.2016.0004 Kearns G Lederer A A resource-based view of strategic IT Alignment: How knowledge sharing creates competitive advantage Decision Sciences. 2003 34 1 29 10.1111/1540-5915.02289 Kim G Shin B Kwon O Investigating the value of sociomaterialism in conceptualizing IT capability of a firm Journal of Information System. 2012 29 327 362 Kline RB Principles and practice of structural equation modeling 2011 Guilford Press Kotabe M Jiang C Murray J Managerial ties, knowledge acquisition, realized absorptive capacity and new product market performance of emerging multinational companies: A case of China Journal of World Business. 2011 46 2 166 176 10.1016/j.jwb.2010.05.005 Kraus S Breier M Dasí-Rodríguez S The art of crafting a systematic literature review in entrepreneurship research International Entrepreneurship and Management Journal. 2020 16 3 1023 1042 10.1007/s11365-020-00635-4 Kushwaha AK Kar AK Dwivedi YK Applications of big data in emerging management disciplines: A literature review using text mining International Journal of Information Management Data Insights 2021 1 2 100017 10.1016/j.jjimei.2021.100017 Lavalle S Lesser E Shockley R Hopkins M Kruschwitz N Big data, analytics and the path from insights to value MIT Sloan Management Review. 2011 52 2 21 32 Lee SM Trimi S Convergence Innovation in the Digital Age and in the COVID-19 Pandemic Crisis Journal of Business Research 2021 123 14 22 10.1016/j.jbusres.2020.09.041 33012897 Lengnick-Hall CA Beck TE Lengnick-Hall ML Developing a capacity for organisational resilience through strategic human resource management Human. Resource Management Review. 2011 21 243 255 10.1016/j.hrmr.2010.07.001 Li, W. & C. A. Chan (2019). Big Data Driven Predictive Caching at the Wireless Edge," 2019 IEEE International Conference on Communications Workshops (ICC Workshops), 1-6, doi: 10.1109/ICCW.2019.8756663 Liguori E Pittz T Strategies for small business: Surviving and thriving in the era of COVID-19 Journal of the International Council for Small Business. 2020 1 4 1 5 Lioukas C Reuer J Zollo M Effects of information technology capabilities on strategic alliances: Implications for the resource-based view Journal of Management Studies. 2016 53 2 1 24 10.1111/joms.12179 Lombardi R Knowledge transfer and organizational performance and business process: Past, present and future researches Business Process Management Journal. 2019 25 1 2 9 10.1108/BPMJ-02-2019-368 Lozada DN Ward BP Carter AH Gains through selection for grain yield in a winter wheat breeding program PLoS ONE 2019 15 4 e0221603 10.1371/journal.pone.0221603 Lu Y Ramamurthy K Understanding the Link between information technology capability and organizational agility: An empirical examination MIS Quarterly. 2011 35 4 931 954 10.2307/41409967 Mallak LA Yildiz M The development of a workplace resilience A Journal of Prevention, Assessment & Rehabilitation. 2016 54 241 253 Martinez-Lozada AC Espinosa A Corporate viability and sustainability: A case study in a Mexican corporation Systems Research and Behavioral Science 2020 39 1 143 158 10.1002/sres.2748 Mazini SR The contribution of information and information technology in building organizational resilience 2014 IGI Global Melville N Kraemer K Gurbaxani V Review: Information technology and organizational performance: An integrative model of IT business value MIS Quarterly. 2004 28 2 283 322 10.2307/25148636 Merhi M Bregu K Effective and efficient usage of big data analytics in public sector Transforming Government: People, Process and Policy. 2020 10.1108/TG-08-2019-0083 Mikalef P Pappas I Krogstie J Giannakos M Big data analytics capabilities: a systematic literature review and research agenda Information Systems and e-Business Management. 2018 16 2 547 578 10.1007/s10257-017-0362-y Mikalef P Boura M Lekakos G Krogstie J Big Data Analytics capabilities and innovation: The mediating role of dynamic capabilities and moderating effect of the environment British Journal of Management. 2020 30 2 272 298 10.1111/1467-8551.12343 Mohamed N Singh JKG A conceptual framework for information technology governance effectiveness in private organizations Information Management & Computer Security 2012 20 2 88 106 10.1108/09685221211235616 Mokline B Ben Abdallah MA The mechanisms of collective resilience in a crisis context: The case of the ‘COVID-19’ crisis Global Journal of Flexible Systems Management 2022 23 1 151 163 10.1007/s40171-021-00293-7 Mokline B Ben Abdallah MA Individual resilience in the organization in the face of crisis: Study of the concept in the context of COVID-19 Global Journal of Flexible Systems Management 2021 22 3 219 231 10.1007/s40171-021-00273-x Morabito V Big data and analytics: Strategic and organizational impacts 2015 Springer Nabeel-Rehman R Nazri M Information technology capabilities and SMEs performance: An understanding of a multi-mediation model for the manufacturing sector Interdisciplinary Journal of Information, Knowledge, and Management 2019 14 253 276 10.28945/4429 Nyanga T Zirima H Reactions of small to medium enterprises in Masvingo Zimbabwe to Covid 19 implications on productivity Business Excellence and Management. 2020 1 22 32 10.24818/beman/2020.S.I.1-02 Oberoi JS Khamba JS Kiran R Impact of new technology and sourcing practices in managing tactical and strategic manufacturing flexibilities - an empirical study Global Journal of Flexible System Management. 2007 8 3 1 14 10.1007/BF03396523 OECD Enhancing the contributions of SMEs in a global and digitalised economy 2017 OECD O'Leary DE Artificial Intelligence and Big Data IEEE Intelligent Systems 2013 28 2 96 99 10.1109/MIS.2013.39 25505373 Onyokoko IO Needorn RS Operational flexibility and adaptive capability of manufacturing firms in south-south Nigeria African Journal of Business and Economic Development 2021 1 6 27 48 Ortiz-de-Mandojana N Bansal P The long-term benefits of organizational resilience through sustainable business practices Strategic Management Journal 2016 37 8 1615 1631 10.1002/smj.2410 Osita-Ejikeme, U. & Amah, E. (2022). Strategic Flexibility and corporate resilience of manufacturing firms in South-South, Nigeria. 9. 50-73 Provost F Fawcett T Data science and its relationship to big data and data-driven decision making Big Data. 2013 1 1 51 59 10.1089/big.2013.1508 27447038 Ragmoun W Achieving sustainable tourism with dynamic capabilities and resilience factors: A post disaster perspective case of the tourism industry in Saudi Arabia Cogent Social Sciences 2022 8 1 1 24 Ragmoun W Almoshaigeh MSS Sustainable motivation as a driver of organizational resilience on crisis period Journal of Entrepreneurship Education 2020 23 S2 1 18 Ragmoun W Alwehabie A Sustainable human resource management (SHRM) and corporate social responsibility (CSR): An integrated mediated moderation model of dynamic capabilities (DC) on family business industry Management Science Letters 2020 10 10 2259 2268 10.5267/j.msl.2020.3.010 Rahman MJ Zhaoa N Information technology capability and firm performance in the era of big data analytics: Evidence from the United States Accounting and Management Information Systems. 2020 19 3 595 621 Rai A Tang X Leveraging IT capabilities and competitive process capabilities for the management of interorganizational relationship portfolios Information Systems Research. 2010 21 3 516 542 10.1287/isre.1100.0299 Rajesh R Technological capabilities and supply chain resilience of firms: A relational analysis using Total Interpretive Structural Modeling (TISM) Technological Forecasting and Social Change 2017 118 161 169 10.1016/j.techfore.2017.02.017 Ramadan M Shuqqo H Qtaishat L Asmar H Salah B Sustainable competitive advantage driven by big data analytics and innovation Applied Sciences. 2020 10 19 6784 10.3390/app10196784 Ramezani J Camarinha-Matos LM Approaches for resilience and antifragility in collaborative business ecosystems Technological Forecasting and Social Change 2020 151 119846 10.1016/j.techfore.2019.119846 RathinaMamunKanesanHayatGopinathan VSATNS Effect of Information system artifacts on organizational resilience: A study among Malaysian SMEs Sustainability. 2019 11 11 1 23 Ravichandran T Exploring the relationships between IT competence, innovation capacity and organizational agility The Journal of Strategic Information Systems. 2018 27 1 22 42 10.1016/j.jsis.2017.07.002 Ravichandran T Lertwongsatien C Effect of information systems resources and capabilities on firm performance: A resource-based perspective Journal of Management Information Systems. 2005 21 4 237 276 10.1080/07421222.2005.11045820 Ravishankar M Pan SL Leidner DE Examining the strategic alignment and implementation success of a KMS: A subculture-based multilevel analysis Information Systems Research 2011 22 1 39 59 10.1287/isre.1080.0214 Rialti R Zollo L Ferraris A Alon I Big data analytics capabilities and performance: Evidence from a moderated multi-mediation model Technological Forecasting and Social Change. 2020 149 119781 10.1016/j.techfore.2019.119781 Ryabchikov MY Ryabchikova ES Big data-driven assessment of proposals to improve enterprise flexibility through control options untested in practice Global Journal of Flexible Systems Management 2022 23 1 43 74 10.1007/s40171-021-00287-5 Sanchez R Strategic flexibility in product competition Strategic Management Journal 1995 16 1 135 159 10.1002/smj.4250160921 Schryen G Revisiting is business value research: what we already know, what we still need to know, and how we can get there European Journal of Information Systems. 2013 22 2 139 169 10.1057/ejis.2012.45 Settembre-Blundo D González-Sánchez R Medina-Salgado S Flexibility and resilience in corporate decision making: A new sustainability-based risk management system in uncertain times Global Journal of Flexible Systems Management 2021 22 Suppl 2 S107 S132 10.1007/s40171-021-00277-7 Shimizu K Hitt M Strategic flexibility: Organizational preparedness to reverse ineffective strategic decisions Academy of Management Executive. 2004 18 4 44 59 Sideridis G Simos P Papanicolaou A Fletcher J Using structural equation modeling to assess functional connectivity in the brain power and sample size considerations Educational and Psychological Measurement 2014 74 5 733 758 10.1177/0013164414525397 25435589 Siebert D Gaskin N Creating, naming and justifying fractions Teaching Children Mathematics 2006 12 394 400 Sila I Examining the effects of contextual factors on TQM and performance through the lens of organizational theories: An empirical study Quality Engineering 2007 52 507 508 Singh RK Modgil S Acharya P Assessment of supply chain flexibility using system dynamics modeling Global Journal of Flexible System Management 2019 20 Suppl 1 S39 S63 10.1007/s40171-019-00224-7 Singh S Dhir S Evans S The trajectory of two decades of global journal of flexible systems management and flexibility research: A bibliometric analysis Global Journal of Flexible System Management 2021 22 4 377 401 10.1007/s40171-021-00286-6 Steininger DM Mikalef P Pateli AG Ortiz de Guinea A Dynamic capabilities in information systems research: A critical review, synthesis of current knowledge, and recommendations for future research Journal of the Association for Information Systems, Forthcoming. 2021 23 2 1 72 Tallon PP Stuck in the middle: Overcoming strategic complexity through IT flexibility Global Journal of Flexible System Management 2008 9 4 1 9 10.1007/BF03396546 Teece D Pisano G The dynamic capabilities of firms: An introduction Industrial and Corporate Change. 1994 3 3 537 556 10.1093/icc/3.3.537-a Teece DJ Pisano G The dynamic capabilities of the firm Industrial and Corporate Change 1997 3 3 537 556 10.1093/icc/3.3.537-a Ufua DE Olujobi OJ Tahir H Lean Entrepreneurship and SME Practice in a Post COVID-19 Pandemic Era: A Conceptual Discourse from Nigeria Global Journal of Flexible Systems Management 2022 23 3 331 344 10.1007/s40171-022-00304-1 Ulz, T., Pieber, T. W., & Steger, C. (2020). Model-Based Design of Secured Power Aware Smart Sensors. Sensor Systems Simulations (pp. 227–251). Springer, Cham. Upadhyay N Dwivedi YK Hughes DL Coombs C Constantiou I Duan Y Edwards JS Impact of COVID-19 pandemic on information management research and practice: Transforming education, work and life International Journal of Information Management 2020 55 102211 10.1016/j.ijinfomgt.2020.102211 Utz S Levers of Organizational Resilience Controlling & Management Review. 2020 64 6 9 18 Vecchio D NegroFallaBazzucchiFarinaFelici FDIDD Higher muscle fiber conduction velocity and early rate of torque development in chronically strength trained individuals Journal of Applied Physiology. 2018 125 4 1218 1226 10.1152/japplphysiol.00025.2018 Vossen RW Relative strengths and weaknesses of small firms in innovation International Small Business Journal 1998 16 3 88 94 10.1177/0266242698163005 Wade MR Hulland JS The resource-based view and information systems research: Review, extension, and suggestions for future research MIS Quartely. 2004 28 107 142 10.2307/25148626 Wadstrom P Aligning corporate and business strategy: managing the balance Journal of Business Strategy. 2019 40 3 44 52 10.1108/JBS-06-2018-0099 Wamba S MishraBhatt D Big data integration with business processes: a literature review Business Process Management Journal 2017 10.1108/BPMJ-02-2017-0047 Wamba SF Gunasekaran A Akter S Ren SJF Dubey R Childe SJ Big data analytics and firm performance: Effects of dynamic capabilities Journal of Business Research 2017 70 1 356 365 10.1016/j.jbusres.2016.08.009 Weill P Ross J IT governance: How top performers manage IT decision rights for superior results 2004 Harvard Business School Press Wenzel M Stanske S Lieberman M Strategic responses to crisis Strategic Management Journal. 2020 42 2 1 13 Williams TA Gruber DA Sutcliffe KM Shepherd DA Zhao EY Organizational response to adversity: Fusing crisis management and resilience research streams Academy of Management Annals 2017 11 2 733 769 10.5465/annals.2015.0134 Wolf EJ Harrington KM Clark SL Miller MW Sample size requirements for structural equation models: An evaluation of power, bias, and solution propriety Educational and Psychological Measurement 2013 73 6 913 934 10.1177/0013164413495237 Wolf EJ Harrington KM Clark SL Miller MW Sample size requirements for structural equation models an evaluation of power, bias, and solution propriety Educational and Psychological Measurement 2013 73 6 913 934 10.1177/0013164413495237 World Bank, (2019). Information and Communication Technology for Disaster Risk Management in Japan: How Digital Solutions Are Leveraged to Increase Resilience through Improving Early Warnings and Disaster Information Sharing. Other Infrastructure Study. World Bank. doi:10.1596/32797 Xie Y Bowe B Al-Aly Z Burdens of post-acute sequelae of COVID-19 by severity of acute infection, demographics and health status Nature Communications 2022 12 1 6571 10.1038/s41467-021-26513-3 Xiu L Liang X Chen Z Xu W Strategic flexibility, innovative HR practices, and firm performance: A moderated mediation model Personnel Review. 2017 46 7 1335 1357 10.1108/PR-09-2016-0252 Yin S Kaynak O Big data for modern industry: Challenges and trends [point of view] Proceedings of the IEEE 2015 103 2 143 146 10.1109/JPROC.2015.2388958 Zahra S Rawhouser H Bhawe N Neubaum D Hayton J Globalization of Social Entrepreneurship Opportunities Strategic Entrepreneurship Journal. 2008 2 2 117 131 10.1002/sej.43 Zhou KZ Wu F Technological capability, strategic flexibility, and product innovation Strategic Management Journal 2010 31 547 561 Zighan S Abualqumboz M Dwaikat N Alkalha Z The role of entrepreneurial orientation in developing SMEs resilience capabilities throughout COVID-19 The International Journal of Entrepreneurship and Innovation 2021 10.1177/14657503211046849 Zutshi A Mendy J Sharma GD Thomas A Sarker T From challenges to creativity: Enhancing SMEs’ resilience in the context of COVID-19 Sustainability 2021 13 12 1 16 10.3390/su13126542

0

PMC9749648

NO-CC CODE

2022-12-15 23:23:21

no

Glob J Flex Syst Manag. 2022 Dec 14;:1-20

utf-8

oa_other

==== Front Nature Nature Nature 0028-0836 1476-4687 Nature Publishing Group UK London 5542 10.1038/s41586-022-05542-y Article SARS-CoV-2 infection and persistence in the human body and brain at autopsy http://orcid.org/0000-0002-0259-4485 Stein Sydney R. 12 Ramelli Sabrina C. 3 Grazioli Alison 4 http://orcid.org/0000-0001-5041-5982 Chung Joon-Yong 5 Singh Manmeet 6 Yinda Claude Kwe 6 http://orcid.org/0000-0002-4460-3378 Winkler Clayton W. 7 Sun Junfeng 3 Dickey James M. 12 Ylaya Kris 5 http://orcid.org/0000-0002-4678-7449 Ko Sung Hee 8 http://orcid.org/0000-0001-8481-114X Platt Andrew P. 12 http://orcid.org/0000-0003-1717-048X Burbelo Peter D. 9 Quezado Martha 5 http://orcid.org/0000-0001-7688-1439 Pittaluga Stefania 5 Purcell Madeleine 10 http://orcid.org/0000-0002-2288-3196 Munster Vincent J. 6 Belinky Frida 8 Ramos-Benitez Marcos J. 1211 Boritz Eli A. 8 Lach Izabella A. 12 Herr Daniel L. 12 Rabin Joseph 13 http://orcid.org/0000-0002-5116-0042 Saharia Kapil K. 1415 Madathil Ronson J. 16 Tabatabai Ali 17 Soherwardi Shahabuddin 18 http://orcid.org/0000-0001-5319-0475 McCurdy Michael T. 1719 NIH COVID-19 Autopsy ConsortiumBabyak Ashley L. 12 Perez Valencia Luis J. 12 Curran Shelly J. 3 Richert Mary E. 3 Young Willie J. 5 Young Sarah P. 5 Gasmi Billel 5 Sampaio De Melo Michelly 5 Desar Sabina 5 Tadros Saber 5 Nasir Nadia 5 Jin Xueting 5 Rajan Sharika 5 Dikoglu Esra 5 Ozkaya Neval 5 Smith Grace 5 Emanuel Elizabeth R. 21 Kelsall Brian L. 21 Olivera Justin A. 22 Blawas Megan 22 Star Robert A. 22 Hays Nicole 10 Singireddy Shreya 10 Wu Jocelyn 10 Raja Katherine 10 Curto Ryan 10 Chung Jean E. 23 Borth Amy J. 23 Bowers Kimberly A. 23 Weichold Anne M. 23 Minor Paula A. 23 Moshref Mir Ahmad N. 23 Kelly Emily E. 23 Sajadi Mohammad M. 1415 Scalea Thomas M. 24 Tran Douglas 16 Dahi Siamak 16 Deatrick Kristopher B. 16 Krause Eric M. 25 Herrold Joseph A. 17 Hochberg Eric S. 17 Cornachione Christopher R. 17 Levine Andrea R. 17 Richards Justin E. 26 Elder John 27 Burke Allen P. 27 Mazzeffi Michael A. 28 Christenson Robert H. 29 Chancer Zackary A. 30 Abdulmahdi Mustafa 31 Sopha Sabrina 31 Goldberg Tyler 31 Sangwan Yashvir 32 Sudano Kristen 19 Blume Diane 19 Radin Bethany 19 Arnouk Madhat 19 Eagan James W. Jr 33 Palermo Robert 34 Harris Anthony D. 35 Pohida Thomas 36 Garmendia-Cedillos Marcial 36 Dold George 37 Saglio Eric 37 Pham Phuoc 37 Peterson Karin E. 7 Cohen Jeffrey I. 20 http://orcid.org/0000-0002-9763-7758 de Wit Emmie 6 http://orcid.org/0000-0002-0972-5011 Vannella Kevin M. 12 http://orcid.org/0000-0001-8283-1788 Hewitt Stephen M. 5 http://orcid.org/0000-0003-3442-4453 Kleiner David E. 5 http://orcid.org/0000-0002-1675-1728 Chertow Daniel S. chertowd@cc.nih.gov 12 1 grid.410305.3 0000 0001 2194 5650 Emerging Pathogens Section, Critical Care Medicine Department, Clinical Center, National Institutes of Health, Bethesda, MD USA 2 grid.419681.3 0000 0001 2164 9667 Laboratory of Immunoregulation, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD USA 3 grid.410305.3 0000 0001 2194 5650 Critical Care Medicine Department, Clinical Center, National Institutes of Health, Bethesda, MD USA 4 grid.419635.c 0000 0001 2203 7304 Kidney Disease Section, Kidney Diseases Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD USA 5 grid.417768.b 0000 0004 0483 9129 Laboratory of Pathology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD USA 6 grid.94365.3d 0000 0001 2297 5165 Laboratory of Virology, Division of Intramural Research, National Institute of Allergy and Infectious Diseases, National Institute of Health, Hamilton, MT USA 7 grid.94365.3d 0000 0001 2297 5165 Laboratory of Persistent Viral Diseases, Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, National Institute of Health, Hamilton, MT USA 8 grid.419681.3 0000 0001 2164 9667 Vaccine Research Center, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD USA 9 grid.419633.a 0000 0001 2205 0568 National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, MD USA 10 grid.411024.2 0000 0001 2175 4264 University of Maryland School of Medicine, Baltimore, MD USA 11 grid.280785.0 0000 0004 0533 7286 Postdoctoral Research Associate Training Program, National Institute of General Medical Sciences, National Institutes of Health, Bethesda, MD USA 12 grid.411024.2 0000 0001 2175 4264 R Adams Cowley Shock Trauma Center, Department of Medicine and Program in Trauma, University of Maryland School of Medicine, Baltimore, MD USA 13 grid.411024.2 0000 0001 2175 4264 R Adams Cowley Shock Trauma Center, Department of Surgery and Program in Trauma, University of Maryland School of Medicine, Baltimore, MD USA 14 grid.411024.2 0000 0001 2175 4264 Department of Medicine, Division of Infectious Disease, University of Maryland School of Medicine, Baltimore, MD USA 15 grid.411024.2 0000 0001 2175 4264 Institute of Human Virology, University of Maryland School of Medicine, Baltimore, MD USA 16 grid.411024.2 0000 0001 2175 4264 Department of Surgery, Division of Cardiac Surgery, University of Maryland School of Medicine, Baltimore, MD USA 17 grid.411024.2 0000 0001 2175 4264 Department of Medicine, Division of Pulmonary and Critical Care Medicine, University of Maryland School of Medicine, Baltimore, MD USA 18 grid.417209.9 0000 0004 0429 3816 Hospitalist Department, TidalHealth Peninsula Regional, Salisbury, MD USA 19 grid.416700.4 0000 0004 0440 9540 Division of Critical Care Medicine, Department of Medicine, University of Maryland St. Joseph Medical Center, Towson, MD USA 20 grid.419681.3 0000 0001 2164 9667 Medical Virology Section, Laboratory of Infectious Diseases, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD USA 21 grid.419681.3 0000 0001 2164 9667 Mucosal Immunobiology Section, Laboratory of Molecular Immunology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD USA 22 grid.419635.c 0000 0001 2203 7304 Renal Diagnostics and Therapeutics Unit, Kidney Diseases Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD USA 23 grid.413036.3 0000 0004 0434 0002 University of Maryland Medical Center, Baltimore, MD USA 24 grid.411024.2 0000 0001 2175 4264 Department of Shock Trauma Critical Care, University of Maryland School of Medicine, Baltimore, MD USA 25 grid.411024.2 0000 0001 2175 4264 Department of Surgery, Division of Thoracic Surgery, University of Maryland School of Medicine, Baltimore, MD USA 26 grid.411024.2 0000 0001 2175 4264 Department of Anesthesiology, Division of Critical Care Medicine, University of Maryland School of Medicine, Baltimore, MD USA 27 grid.411024.2 0000 0001 2175 4264 Department of Autopsy and Thoracic Pathology, University of Maryland School of Medicine, Baltimore, MD USA 28 grid.253615.6 0000 0004 1936 9510 Department of Anesthesiology and Critical Care Medicine, George Washington University School of Medicine and Health Sciences, Washington, D.C., USA 29 grid.411024.2 0000 0001 2175 4264 Department of Laboratory Science, University of Maryland School of Medicine, Baltimore, MD USA 30 grid.42505.36 0000 0001 2156 6853 Department of Anesthesiology, Keck School of Medicine, University of Southern California, Los Angeles, CA USA 31 grid.449876.0 0000 0004 0433 9888 Critical Care Medicine, University of Maryland Baltimore Washington Medical Center, Glen Burnie, MD USA 32 grid.417209.9 0000 0004 0429 3816 Department of Interventional Pulmonology, TidalHealth Peninsula Regional, Salisbury, MD USA 33 grid.416700.4 0000 0004 0440 9540 Department of Pathology, University of Maryland St. Joseph Medical Center, Towson, MD USA 34 grid.413287.b 0000 0004 0373 8692 Department of Pathology, Greater Baltimore Medical Center, Townson, MD USA 35 grid.411024.2 0000 0001 2175 4264 Department of Epidemiology and Public Health, University of Maryland School of Medicine, Baltimore, MD USA 36 grid.280347.a 0000 0004 0533 5934 Instrumentation Development and Engineering Application Solutions, National Institute of Biomedical Imaging and Bioengineering, National Institutes of Health, Bethesda, MD USA 37 grid.416868.5 0000 0004 0464 0574 Section on Instrumentation, National Institute of Mental Health, National Institutes of Health, Bethesda, MD USA 14 12 2022 16 3 12 2021 8 11 2022 © This is a U.S. Government work and not under copyright protection in the US; foreign copyright protection may apply 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. Coronavirus disease 2019 (COVID-19) is known to cause multi-organ dysfunction1–3 during acute infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), with some patients experiencing prolonged symptoms, termed post-acute sequelae of SARS-CoV-2 (refs. 4,5). However, the burden of infection outside the respiratory tract and time to viral clearance are not well characterized, particularly in the brain3,6–14. Here we carried out complete autopsies on 44 patients who died with COVID-19, with extensive sampling of the central nervous system in 11 of these patients, to map and quantify the distribution, replication and cell-type specificity of SARS-CoV-2 across the human body, including the brain, from acute infection to more than seven months following symptom onset. We show that SARS-CoV-2 is widely distributed, predominantly among patients who died with severe COVID-19, and that virus replication is present in multiple respiratory and non-respiratory tissues, including the brain, early in infection. Further, we detected persistent SARS-CoV-2 RNA in multiple anatomic sites, including throughout the brain, as late as 230 days following symptom onset in one case. Despite extensive distribution of SARS-CoV-2 RNA throughout the body, we observed little evidence of inflammation or direct viral cytopathology outside the respiratory tract. Our data indicate that in some patients SARS-CoV-2 can cause systemic infection and persist in the body for months. A study reports the distribution, replication and persistence of SARS-CoV-2 throughout the human body including in the brain at autopsy from acute infection to more than seven months following symptom onset. Subject terms SARS-CoV-2 Viral infection Infection ==== Body pmcMain COVID-19 has respiratory and non-respiratory manifestations1–3, including multi-organ failure and shock among patients with severe and fatal disease. Some individuals who survive experience post-acute sequelae of SARS-CoV-2, also known as long COVID4,5. Although autopsy studies of fatal COVID-19 cases support the ability of SARS-CoV-2 to infect multiple organs3,7–12, extrapulmonary organs often lack histopathological evidence of virally mediated injury or inflammation10–14. The paradox of infection outside the respiratory tract without injury or inflammation raises many pathogen- and host-related questions. To investigate the cellular tropism, replication competence, persistence and evolution of SARS-CoV-2 in humans, and to look for associated histopathology in infected tissues, we carried out autopsies on 44 COVID-19 cases. Our approach focused on timely, systematic and comprehensive tissue sampling and preservation for complementary analyses. We carried out droplet digital polymerase chain reaction (ddPCR) for detection and quantification of SARS-CoV-2 nucleocapsid (N) gene targets and in situ hybridization (ISH) to validate the ddPCR findings and determine the cellular tropism of SARS-CoV-2. Immunofluorescence (IF) and chromogenic immunohistochemistry (IHC) were used to further validate the presence of SARS-CoV-2 in the brain. We carried out quantitative real-time PCR with reverse transcription (RT–qPCR) to detect subgenomic RNA, a marker suggestive of recent virus replication15, and demonstrated replication-competent SARS-CoV-2 in selected respiratory and non-respiratory tissues, including the brain, by virus isolation in traditional and modified Vero E6 cell culture. In six individuals, we measured the diversity and anatomic distribution of intra-individual SARS-CoV-2 spike gene variants using high-throughput, single-genome amplification and sequencing (HT-SGS). We categorized autopsy cases as early (n = 17), mid (n = 13) or late (n = 14) by illness day (d) at the time of death, being ≤d14, d15–30 or ≥d31, respectively. We defined persistence as the presence of SARS-CoV-2 RNA among late cases. We analysed and described our results in terms of respiratory and non-respiratory tissues to quantify and statistically compare SARS-CoV-2 RNA levels across tissues and cases. Autopsy cohort overview Between 26 April 2020 and 2 March 2021, we carried out 44 autopsies, all among unvaccinated individuals who had died with COVID-19. SARS-CoV-2 PCR positivity was confirmed premortem in 42 cases and postmortem in 2 cases (P3 and P17; Extended Data Fig. 1). A total of 38 cases were SARS-CoV-2 seropositive (Supplementary Data 1a), 3 were seronegative (P27, P36 and P37), and plasma was unavailable for 3 cases (P3, P4 and P15). Brain sampling was accomplished in 11 cases (Fig. 1). The cohort was racially and ethnically diverse. Thirty per cent were female, and the median age was 62.5 years (interquartile range (IQR): 47.3–71.0; Extended Data Table 1a). A total of 61.4% had three or more comorbidities. The median interval from symptom onset to final hospitalization and subsequently death was 6 days (IQR: 3–10) and 18.5 days (IQR: 11.25–37.5), respectively (Extended Data Table 1b). The median postmortem interval was 22.2 h (IQR: 18.2–33.9). Individual-level case data can be found in Supplementary Data 2a.Fig. 1 Distribution, quantification and replication of SARS-CoV-2 across the human body and brain. The heat map depicts the highest mean quantification of SARS-CoV-2 RNA (N) through ddPCR present in the autopsy tissues of 11 patients who died with COVID-19 and underwent whole-body and brain sampling. Patients are aligned from shortest to longest duration of illness (DOI) before death, listed at the bottom of the figure, and grouped into early (≤14 days), mid (15–30 days) and late (≥31 days) duration of illness. Tissues are organized by tissue group beginning with the respiratory tissues at the top and CNS at the bottom. Viral RNA levels range from 0.002 to 500,000 N gene copies per nanogram of RNA input, depicted as a gradient from dark blue at the lowest level to dark red at the highest level. Tissues that were also positive for subgenomic RNA (sgRNA+) through real-time RT–qPCR are shaded with black vertical bars. O, other; PNS, peripheral nervous system; SM, skeletal muscle. Widespread infection and persistence SARS-CoV-2 RNA was detected in 84 distinct anatomical locations and body fluids (Supplementary Data 1b–d), with a significantly (P < 0.0001 for all) higher burden detected in respiratory compared with non-respiratory tissues among early (2.04 ± 0.10 log10[N gene copies] per nanogram of RNA), mid (1.36 ± 0.12 log10[N gene copies] per nanogram of RNA) and late (0.67 ± 0.11 log10[N gene copies] per nanogram of RNA; Extended Data Fig. 2a) cases. We compared linear trends in SARS-CoV-2 RNA levels by illness day, as a continuous variable, and observed a significantly steeper negative slope of SARS-CoV-2 RNA levels in respiratory (−3.14, s.e. 0.39) compared with non-respiratory (−1.62, s.e. 0.38; P < 0.0001) tissues (Extended Data Fig. 2b,c). We detected SARS-CoV-2 RNA in perimortem plasma of 11 early and 1 mid case (Supplementary Data 1b,d). SARS-CoV-2 RNA was undetectable or just above the limit of detection in peripheral blood mononuclear cells from select early and mid cases (Supplementary Data 1a). The median and IQR of SARS-CoV-2 N gene copies per nanogram of RNA and proportion of cases with RNA detected in each tissue group and fluids are summarized in Extended Data Table 2. SARS-CoV-2 RNA persistence was detected across multiple tissue groups among all late cases despite being undetectable in plasma in any (Supplementary Data 1b–d). SARS-CoV-2 RNA was detected in central nervous system (CNS) tissue in 10/11 cases (90.9%), including across most brain regions evaluated in 5/6 late cases, including P42 who died at D230 (Fig. 1). We detected SARS-CoV-2 subgenomic RNA across all tissue groups and multiple fluid types, including plasma, pleural fluid and vitreous humour (Supplementary Data 1a–c). ddPCR and subgenomic RNA RT-qPCR results closely correlated among 1,025 jointly tested samples (ρ = 0.76; 95% confidence interval (CI): 0.73–0.78), particularly among respiratory samples (n = 369, ρ = 0.86; 95% CI: 0.84–0.89), early cases (n = 496, ρ = 0.88; 95% CI: 0.85–0.89) and samples that tested positive by both assays (n = 302, ρ = 0.91; 95% CI: 0.88–0.93; Extended Data Fig. 2d,e). With sensitivity and specificity weighted equally, a ddPCR value of ≥1.47 N copies per nanogram of RNA predicted a positive subgenomic RNA result with 93.0% sensitivity and 91.6% specificity, with a receiver operating characteristic (ROC) area under the curve (AUC) of 0.965 (95% CI: 0.953–0.977; Extended Data Fig. 2f). We isolated SARS-CoV-2 in Vero E6 cell culture from diverse tissues in and outside the respiratory tract including heart, lymph node, gastrointestinal tract, adrenal gland and eye from early cases (Extended Data Fig. 3a). In total, we isolated virus from 25/55 (45%) specimens tested across four SARS-CoV-2 subgenomic RNA quantification cycle (Cq) value intervals, with decreasing yield with rising Cq interval. Among the 55 samples tested for virus isolation, with sensitivity and specificity weighted equally, a ddPCR value of ≥758 N copies per nanogram of RNA predicted replication-competent virus with 76% sensitivity and 90% specificity (ROC AUC = 0.887; 95% CI: 0.795–0.978), and a subgenomic RNA value of ≥25,069 copies per microlitre of RNA (about Cq 22.40) predicted replication-competent virus with 72% sensitivity and 100% specificity (ROC AUC = 0.915; 95% CI: 0.843–0.987; Extended Data Fig. 3b,c). We reattempted virus isolation from thalamus and hypothalamus of P38 on Vero E6-TMPRSS2-T2A-ACE2 cells16 and observed a cytopathic effect 48 h after inoculation with thalamus tissue homogenate from P38. RT–qPCR for SARS-CoV-2 envelope (E) genomic RNA was carried out on the tissue homogenate and the supernatant of the virus isolation process at the time of the cytopathic effect, and yielded Ct values of 27.33 and 13.24, respectively. Viral genome sequencing We used HT-SGS to analyse SARS-CoV-2 spike gene variant sequences from a total of 46 tissues in 6 individuals (Extended Data Fig. 4). In P27 (D1), P19 (D7) and P18 (D9), no nonsynonymous virus genetic diversity was detected in respiratory and non-respiratory sites despite a high depth of single-molecule sampling. In P27, two virus haplotypes, each with a single synonymous substitution, were preferentially detected in non-respiratory sites including the right and left ventricles and mediastinal lymph node. In P38 (D13), a D80F residue was identified in 31/31 pulmonary but 0/490 brain sequences, and a G1219V residue was restricted to brain variants. SARS-CoV-2 virus isolated from thalamus of P38 through Vero E6-TEMPRSS2-TA2-ACE2 cell culture and subjected to short-read, whole-genome sequencing matched the minor haplotype detected from P38 RNAlater-preserved thalamus and the major haplotype of P38 RNAlater-preserved hypothalamus. A nonsynonymous substitution was also detected in P36 (D4) dura mater, albeit at very low sampling depth (n = 2 sequences), compared with non-CNS tissue. ISH reveals the cellular tropism of SARS-CoV-2 We validated our ddPCR results by ISH for SARS-CoV-2 spike RNA in respiratory and non-respiratory tissues in selected early, mid and late cases across >35 cell types and hyaline membranes (Figs. 2 and 3, Extended Data Table 3 and Supplementary Data 3). Detailed annotation of SARS-CoV-2 spike RNA ISH-positive cells by tissue, including across multiple brain regions, is provided in Figs. 2 and 3 and Supplementary Data 3.Fig. 2 RNA in situ (RNAscope) detection of SARS-CoV-2 in extrapulmonary tissues. a–h, SARS-CoV-2 virus is localized to the Golgi and endoplasmic reticulum, perinuclear in appearance, in the following organs and cell types (×500 magnifications, scale bars, 2 μm, all panels): thyroid of P19, demonstrating the presence of virus in follicular cells (a), oesophagus of P18, demonstrating the presence of virus in the stratified squamous epithelium (asterisk), as well as signal in capillaries within the stroma (hash) (b), spleen of P19, demonstrating the presence of virus in mononuclear leukocytes in the white pulp (c), appendix of P19, demonstrating the presence of virus in both colonic epithelium (asterisk) and mononuclear leukocytes in the stroma (hash) (d), adrenal gland of P19, demonstrating the presence of virus in endocrine secretory cells (e), ovary of P18, demonstrating the presence of virus in stromal cells of the ovary in a post-menopausal ovary (f), testis of P20, demonstrating the presence of virus in both Sertoli cells (asterisk) and maturing germ cells in the seminiferous tubules of the testis (hash) (g), endometrium of P35, demonstrating the presence of virus in endometrial gland epithelium (asterisk) and stromal cells (hash), in a pre-menopausal endometrial sample (h). The images are exemplars of extrapulmonary tissues that were positive for SARS-CoV-2 N RNA during 20 batches of ISH staining. Fig. 3 SARS-CoV-2 protein and RNA expression in human CNS tissues. a, High-magnification visualization of hypothalamus from P38 labelled for SARS-CoV-2 N protein (green) and neuronal nuclear (NeuN) protein (magenta), demonstrating viral-specific protein expression in neurons (white arrowheads) by IF. The z-stack orthogonal views to the right and bottom of a demonstrate NeuN labelling in the nucleus and SARS-CoV-2 protein in the cytoplasm of the cell (red arrowhead). b–d, SARS-CoV-2 spike (S) (b) and N (c) RNA (brown) by ISH and SARS-CoV-2 N (d) protein (brown) by chromogenic IHC of hypothalamus of P38. e, Infected neurons were found in the cervical spinal cord of P42 by IF, for which white arrowheads indicate NeuN-positive neurons with associated virus protein. Viral protein labelling was also identified in linear structures radiating away from neuronal cell bodies suggestive of neuronal projections (yellow arrowheads). f, A higher magnification of neuron-associated viral protein labelling. g, Viral protein was also detected by IF in neurons of the spinal ganglia at the level of the cervical spinal cord of P42 (white arrowheads). h–j, SARS-CoV-2 S (h) and N (i) RNA by ISH and SARS-CoV-2 N (j) protein by chromogenic IHC of cervical spinal cord of P42. k–m, SARS-CoV-2 S (k) and N (l) RNA by ISH and SARS-CoV-2 N (m) protein by chromogenic IHC, all of which are predominantly found in the granular layer (GL) as compared to the molecular level (ML) of cerebellum of P38. WM, white matter. n,o, SARS-CoV-2 S (n) and N (o) RNA by ISH and SARS-CoV-2 N (p) viral protein by chromogenic IHC of basal ganglia of P40. Hoechst 33342 was used to identify nuclei (blue) in all IF images, and IF images were obtained by confocal microscopy. Haematoxylin was used as a counterstain and all ISH and chromogenic IHC images were obtained by bright-field microscopy. Scale bars, 15 μm (a), 10 μm (b–d,g–p) and 25 μm (e,f). To determine the relationship between SARS-CoV-2 N RNA detected by ddPCR and SARS-CoV-2 spike RNA detected by ISH, we carried out image analysis on interventricular septal tissue from 16 cases covering a four-log range of SARS-CoV-2 N gene copies per nanogram of RNA ddPCR values. Interventricular septum was selected for this analysis owing to consistent histomorphology. Mean SARS-CoV-2 N gene copies per nanogram of RNA significantly correlated with the median SARS-CoV-2 spike RNA-positive cells over thirty ×40 fields (ρ = 0.704, 95% CI: 0.320–0.889, P = 0.002; Supplementary Data 3). SARS-CoV-2 N RNA and protein in CNS To further validate detection and distribution of SARS-CoV-2 in CNS tissue, we used a second ISH assay targeting N RNA, and IF and chromogenic IHC-based assays targeting N protein. We confirmed the specificity of these assays with appropriate positive and negative controls (Supplementary Data 3, panels yy–bbb) and applied them to selected CNS tissues that were SARS-CoV-2 positive by ddPCR. We observed SARS-CoV-2 RNA and protein in hypothalamus and cerebellum of an early case (P38) and cervical spinal cord and basal ganglia of late cases (P42 and P40, respectively), with a pattern consistent with neuronal staining (Fig. 3). COVID-19 histological findings The histopathology findings from our cohort were similar to those reported in other case series (Extended Data Fig. 5). Of 44 cases, 38 were determined to have died from COVID-19, and of these, 35 (92.1%) had either acute pneumonia or diffuse alveolar damage at the time of death (Supplementary Data 2). Phases of diffuse alveolar damage showed a clear temporal progression (Extended Data Fig. 6). Pulmonary thromboembolic complications were noted in 10 (23%) cases and myocardial infiltrates were observed in four cases, including one case of substantial myocarditis17 (P3). In the lymph nodes and spleen, we observed both lymphodepletion and follicular and paracortical hyperplasia. Other non-respiratory histological changes were mainly related to complications of therapy or pre-existing comorbidities. Five cases had old ischaemic myocardial scars and three had coronary artery bypass grafts in place. Diabetic nephropathy and steatohepatitis were observed in ten cases (23%) and five cases (12%), respectively. One case had known hepatitis C with cirrhosis, but the other cases of advanced hepatic fibrosis were probably related to fatty liver disease. Hepatic necrosis (13 cases, 30%) and changes consistent with acute kidney injury (17 cases, 39%) were probably related to hypoxic–ischaemic injury in these very ill patients. In the examination of 11 brains, we found few histopathologic changes, despite substantial viral burden. Vascular congestion was an unusual finding that had an unclear aetiology and could be related to the haemodynamic changes incurred with infection. Global hypoxic–ischaemic change was seen in two cases, one of which was a juvenile (P36) with a seizure disorder who was found to be SARS-CoV-2 positive on hospital admission, but who probably died of seizure complications unrelated to viral infection. Discussion Here we provide, to our knowledge, the most comprehensive analysis to date of the cellular tropism, quantification and persistence of SARS-CoV-2 across the human body including the brain. Our focus on short postmortem intervals, a comprehensive standardized approach to tissue collection, dissecting the brain before fixation, preserving tissue in RNAlater and flash freezing of fresh tissue allowed us to detect and quantify SARS-CoV-2 RNA levels with high sensitivity by ddPCR and ISH, as well as isolate virus in cell culture from multiple non-respiratory tissues including the brain, which are notable differences compared to other studies. We show that SARS-CoV-2 disseminates early in infection in some patients, with a significantly higher viral burden in respiratory than non-respiratory tissues. We demonstrated virus replication in multiple non-respiratory sites during the first two weeks following symptom onset and detected subgenomic RNA in at least one tissue in 14 of 27 cases beyond D14, indicating that viral replication may occur in non-respiratory tissues for several months. Whereas others have proposed that the detection of SARS-CoV-2 in non-respiratory tissues might be due to residual blood within tissues8,18 or cross-contamination from the lungs during tissue procurement8, our data indicate otherwise. Specifically, only 12 of our cases had detectable SARS-CoV-2 RNA in a perimortem plasma sample, only 2 cases had SARS-CoV-2 subgenomic RNA detected in plasma, and negligible, if any, RNA was detected in banked peripheral blood mononuclear cells from representative cases. Further, we validated detection of SARS-CoV-2 outside the respiratory tract by direct cellular identification of virus in cells through ISH, IHC and IF, isolation of SARS-CoV-2 by cell culture, and detection of distinct SARS-CoV-2 spike sequence variants in non-respiratory sites. Others have previously reported SARS-CoV-2 RNA within the heart, lymph node, small intestine and adrenal gland6,8–12,18. We replicate these findings and conclusively demonstrate that SARS-CoV-2 is capable of infecting and replicating within these and many other tissues, including brain. Specifically, we report the recovery of replication-competent SARS-CoV-2 from thalamus of P38 at D13 using a modified Vero E6 cell line that stably expresses ACE2 and TMPRSS2. This along with detection of genomic RNA and subgenomic RNA through PCR, multiple imaging modalities showing SARS-CoV-2 RNA and protein within cells of the CNS, and distinct minor variants detected through sequencing in the CNS prove definitively that SARS-CoV-2 is capable of infecting and replicating within the human brain. HT-SGS of SARS-CoV-2 spike demonstrates homogeneous virus populations in many tissues, while also revealing informative virus variants in others. Low intra-individual diversity of SARS-CoV-2 sequences has been observed frequently in previous studies19–21, and probably relates to the intrinsic mutation rate of the virus as well as lack of early immune pressure to drive virus evolution. It is important to note that our HT-SGS approach has both a high accuracy and a high sensitivity for minor variants within each sample, making findings of low virus diversity highly reliable22. Genetic compartmentalization of SARS-CoV-2 between respiratory and non-respiratory tissues in several individuals supports independent replication of the virus at these sites, although lack of compartmentalization between sites does not rule out independent virus replication. We note several cases in which brain-derived virus spike sequences showed nonsynonymous changes relative to sequences from other non-CNS tissues. Further studies will be needed to understand whether these cases might represent stochastic seeding of the CNS or differential selective pressure on spike by antiviral antibodies in the CNS, as others have suggested23–25. Our results show that although the highest burden of SARS-CoV-2 is in respiratory tissues, the virus can disseminate throughout the entire body. Whereas others have posited that this viral dissemination occurs through cell trafficking11 due to a reported failure to culture SARS-CoV-2 from blood3,26, our data support an early viraemic phase, which seeds the virus throughout the body following infection of the respiratory tract. Recent work26 in which SARS-CoV-2 virions were pelleted and imaged from plasma of patients with acute COVID-19 supports this mechanism of viral dissemination. Our cohort is predominantly composed of severe and ultimately fatal COVID-19 cases. However, two cases (P36 and P42) reported only mild or no respiratory symptoms and died with, not from, COVID-19, yet had SARS-CoV-2 RNA widely detected across the body and brain. Additionally, P36 was a juvenile with an underlying neurological condition, but without evidence of multisystem inflammatory syndrome in children, suggesting that children may develop systemic infection with SARS-CoV-2 without developing a generalized inflammatory response. Finally, our work begins to elucidate the duration and locations at which SARS-CoV-2 RNA can persist. Although the respiratory tract was the most common location in which SARS-CoV-2 RNA persisted, ≥50% of late cases also had persistent RNA in the myocardium, lymph nodes from the head and neck and from the thorax, sciatic nerve, ocular tissue, and in all sampled regions of the CNS, except the dura mater. Notably, despite having more than 100 times higher SARS-CoV-2 RNA in respiratory compared to non-respiratory tissues in early cases, this difference greatly diminished in late cases. Less efficient viral clearance in non-respiratory tissues may be related to tissue-specific differences in the ability of SARS-CoV-2 to alter cellular detection of viral mRNA, interfere with interferon signalling, or disrupt viral antigen processing and presentation27–29. Understanding mechanisms by which SARS-CoV-2 evades immune detection is essential to guide future therapeutic approaches to facilitate viral clearance. We detected subgenomic RNA in tissue from more than 60% of the cohort, including in multiple tissues of a case at D99. Although subgenomic RNA is generated during active viral replication, it is less definitive than cell culture at demonstrating replication-competent virus because subgenomic RNA is protected by double-membrane vesicles that contribute to nuclease resistance and longevity beyond immediate viral replication30–33. However, nonhuman primates exposed to γ-irradiated SARS-CoV-2 inoculum with high subgenomic RNA copy numbers through multiple mucosal routes had detectable SARS-CoV-2 genomic RNA but undetectable subgenomic RNA levels in respiratory samples by day 1 post-inoculation15. These data suggest that detection of SARS-CoV-2 subgenomic RNA probably reflects recent viral replication. Prolonged detection of subgenomic RNA in a subset of our cases may, however, represent defective rather than productive viral replication, which has been described in persistent infection with measles virus—another single-strand enveloped RNA virus—in cases of subacute sclerosing panencephalitis34. Our study has several important limitations. First, our cohort largely represents older unvaccinated individuals with pre-existing medical conditions who died from severe COVID-19, limiting our ability to extrapolate findings to younger, healthier or vaccinated individuals. Second, our cases occurred during the first year of the pandemic, before widespread circulation of variants of concern, and thus findings might not be generalizable to current and future SARS-CoV-2 variants. Finally, although it is tempting to attribute clinical findings observed in post-acute sequelae of SARS-CoV-2 to viral persistence, our study was not designed to address this question. Despite these limitations, our findings fundamentally improve the understanding of SARS-CoV-2 cellular distribution and persistence in the human body and brain and provide a strong rationale for pursuing future similar studies to define mechanisms of SARS-CoV-2 persistence and contribution to post-acute sequelae of SARS-CoV-2. Methods Autopsies Autopsies were carried out and tissues were collected as previously described35 in the National Cancer Institute’s Laboratory of Pathology at the National Institutes of Health Clinical Center following consent of the legal next of kin. Autopsy patients in this cohort were unvaccinated against SARS-CoV-2. Tissues preserved for histopathologic analysis and special staining were dissected fresh at the time of autopsy, placed into tissue cassettes, fixed for 24 h in neutral-buffered formalin, and then transferred to 70% ethanol for 48 h before impregnation with paraffin. Measurement of IgG and IgM antibodies to N and spike protein of SARS-CoV-2 Fluid-phase luciferase immunoprecipitation system assays were used to study IgG and IgM antibody response to SARS-CoV-2. For IgG luciferase immunoprecipitation system measurements, Renilla luciferase–nucleocapsid and Gaussia luciferase–spike protein extracts were used with protein A/G beads (Protein A/G UltraLink Resin, Thermo Fisher) as the IgG capture reagent as previously described with microtitre filter plates36. For IgM measurements, anti-human IgM goat agarose beads (Sigma) were substituted as the capture reagent using both the microfilter plate and microtube format37. The IgM immunoprecipitation assays were carried out in 1.5-ml microfuge tube format containing 1 µl serum or plasma, Renilla luciferase–N (10 million light unit input per tube) or Gaussia luciferase–spike protein (40 million light input per tube) and buffer A (20 mM Tris, pH 7.5, 150 mM NaCl, 5 mM MgCl2, 0.1% Triton X-100) to a total volume of 100 µl. After mixing, the tubes were incubated at room temperature for 1 h. Next, 10 µl of the anti-human IgM agarose bead suspension was added to each tube for a further 60 min, and tubes were placed on a rotating wheel at 4 °C. The samples were then washed by brief centrifugation to collect the bead pellet at room temperature three times with 1.5 ml buffer A and once with 1.5 ml PBS. After the final wash, the beads were mixed with coelenterazine substrate (100 µl) and light units were measured in a tube luminometer. Known seronegative and seropositive samples for IgG and IgM antibodies to the N and spike proteins were used for assigning seropositive cutoff values and for standardization. SARS-CoV-2 RNA quantification of tissues and body fluids Total RNA was extracted from RNAlater (Invitrogen)-preserved tissues and body fluids collected at autopsy using the RNeasy Mini, RNeasy Fibrous Tissue Mini, RNeasy Lipid Tissue Mini and QIAamp Viral RNA Mini Kits (Qiagen) according to the manufacturer’s protocols. Upstream tissue processing and subsequent RNA quantification have been described previously35. The QX200 AutoDG Droplet Digital PCR System (Bio-Rad) was used to detect and quantify SARS-CoV-2 RNA in technical replicates of 5.5 µl RNA for fluids and up to 550 ng RNA for tissues. Raw data were collected using QuantaSoft version 1.7.4.0917 and analysed using QuantaSoft Analysis Pro version 1.0.596. Results were then normalized to copies of N1, N2 and RP per millilitre of sample input for fluids and per nanogram of RNA concentration input for tissues. Samples had to be positive for the human RNase P (RP) gene at the manufacturer’s limit of detection (LOD) of ≥0.2 copies per microlitre and ≥4 positive droplets per well to ensure RNA extraction was successful and be reported. For samples to be considered positive for SARS-CoV-2 N1 or N2 genes, the technical replicates needed to have an average at or above the manufacturer’s LOD of ≥0.1 copies per microlitre and ≥2 positive droplets per well. More than 60 control autopsy tissues from uninfected individuals, representing all organs collected for COVID-19 autopsy cases, were used to validate the manufacturer’s emergency use authorization published LOD for nasopharyngeal swabs for tissues (Supplementary Data 1e). ddPCR data for P3 (ref. 17) as well as a portion of the oral cavity35 have been reported previously. Subgenomic RNA analysis of ddPCR positive tissues Tissues that tested positive for one or both SARS-CoV-2 N gene targets through ddPCR had RNA submitted for subgenomic RNA analysis. Briefly, 5 µl of sample RNA was added to a one-step real-time RT–qPCR assay targeting subgenomic RNA of the envelope (E) gene (forward primer 5′-CGATCTCTTGTAGATCTGTTCTC-3′; reverse primer 5′-ATATTGCAGCAGTACGCACACA-3′; probe 5′-FAM-ACACTAGCCATCCTTACTGCGCTTCG-ZEN-IBHQ-3′)38 using the Rotor-Gene probe kit (Qiagen) according to instructions of the manufacturer. In each run, standard dilutions of counted RNA standards were run in parallel to calculate copy numbers in the samples. The LOD for this assay was determined to be <40 Cq (Supplementary Data 1) using 40 control autopsy tissues from uninfected individuals, representing all organs collected for COVID-19 autopsy cases. Virus isolation from select postmortem tissues Select tissues with high viral RNA levels through ddPCR and subgenomic RNA RT-qPCR measuring across a broad range of 16 to <35 Cq underwent virus isolation to prove the presence of infectious virus. Virus isolation was carried out on tissues by homogenizing the tissue in 1 ml DMEM and inoculating Vero E6 cells in a 24-well plate with 250 µl of cleared homogenate and a 1:10 dilution thereof. Plates were centrifuged for 30 min at 1,000 r.p.m. and incubated for 30 min at 37 °C and 5% CO2. The inoculum was then removed and replaced with 500 µl DMEM containing 2% FBS, 50 U ml−1 penicillin and 50 μg ml−1 streptomycin. Six days after inoculation, the cytopathic effect was scored. A blind passage of samples in which no cytopathic effect was present was carried out according to the same method. Additional virus isolation from P38 thalamus and hypothalamus was carried out using Vero E6-TMPRSS2-T2A-ACE2 (catalogue no. NR-54970, BEI Resources) grown in DMEM containing 10% FBS, 50 U ml−1 penicillin, 50 μg ml−1 streptomycin and 10 μg ml−1 puromycin. Virus isolation was carried out as described for other tissues on Vero E6 cells, without selection antibiotics. Tissue homogenate from flash-frozen specimens and supernatants from plates were analysed using RT–qPCR for SARS-CoV-2 E gene subgenomic RNA (described above) or genomic RNA as previously described39 to rule out other causes for the cytopathic effect. Cell lines were not authenticated in house, but were confirmed to be free of mycoplasma contamination. Virus sequencing Five early cases (P18, P19, P27 and P38) and one late case (P33) with multiple body site tissues containing subgenomic RNA levels ≤31 Cq were selected for HT-SGS as previously described22. Presence of variants of SARS-CoV-2 was analysed within and between tissues. Supernatant from the virus isolation plates of thalamus of P38 were sequenced using short-read, whole-genome sequencing. Total RNA was depleted of rRNA using Ribo-Zero+ following the manufacturer’s protocol (Illumina). Cleaned RNA was eluted in water and sequencing libraries were prepared following the Kapa RNA HyperPrep kit according to the manufacturer’s protocol (Roche Sequencing Solutions). Briefly, 10 µl of depleted RNA was used as a template for fragmentation (65 °C for 1 min) and first-strand synthesis. To facilitate multiplexing, adapter ligation was carried out with KAPA Unique Dual-Indexed Adapters, and samples were enriched for adapter-ligated product using KAPA HiFi HotStart Ready mix and a range of 9–19 PCR amplification cycles based on SARS-CoV-2 Ct values and total RNA starting inputs. Pools consisting of 1–6 sample libraries were used for hybrid-capture virus enrichment using myBaits Expert Virus SARS-CoV-2 panel following the manufacturer’s manual, version 5.01, with a range of 12–18 cycles of post-capture PCR amplification (Arbor Biosciences). Purified, enriched libraries were quantified on a CFX96 Real-Time System (Bio-Rad) using Kapa Library Quantification kit (Roche Sequencing Solutions). Libraries were diluted to 2 nM stock, pooled together as needed in equimolar concentrations and sequenced on the MiSeq (Illumina) generating 2 × 150-bp paired-end reads. Raw sequence reads were trimmed of Illumina adapter sequence using Cutadapt version 1.12 (ref. 40) and then trimmed and filtered for quality using the fastq_quality_trimmer and fastq_quality_filter tools from the FASTX-Toolkit 0.0.14 (Hannon Lab, CSHL). Reads were then mapped to the SARS-CoV-2 2019-nCoV/USA-WA1/2020 genome (MN985325.1) using Bowtie2 version 2.2.9 (ref. 41) with parameters -local -no-mixed -X 1500. PCR duplicates were removed using picard MarkDuplicates, version 2.26.10 (Broad Institute). SARS-CoV-2 RNA ISH Chromogenic ISH detection was carried out using the manual RNAScope 2.5 HD assay (catalogue no. 322310, Advanced Cell Diagnostics) with a modified pretreatment protocol. Briefly, formalin-fixed and paraffin-embedded (FFPE) tissue sections were cut at 7 μm, air dried overnight, and baked for 1–2 h at 60 °C. The FFPE tissue sections were deparaffinized, dehydrated and then treated with pretreat 1 for 15 min at room temperature. The slides were boiled with pretreatment reagent for 15 min, digested with protease at 40 °C for 20 min, and then hybridized for 2 h at 40 °C with probe-V-nCov2019-S (catalogue no. 848561, Advanced Cell Diagnostics) or probe-V-nCoV-N (catalogue no. 846081, Advanced Cell Diagnostics)42. In addition, probe-Hs-PPIB (peptidylprolyl isomerase B, catalogue no. 313901, Advanced Cell Diagnostics) and probe-dapB (catalogue no. 310043, Advanced Cell Diagnostics) were used as a positive and negative control, respectively. Subsequent amplification was carried out according to the original protocol. Detection of specific probe-binding sites was visualized with RNAScope 2.5 HD Reagent Kit-BROWN chromogenic labels (Advanced Cell Diagnostics). The slides were counterstained with haematoxylin and coverslipped. To correlate viral load detected between ddPCR and ISH, the interventricular septum of 16 cases spanning a four-log range of SARS-CoV-2 N copies per nanogram of RNA through ddPCR underwent ISH and subsequent quantification using image analysis. The interventricular septum stained slides were digitalized using a NanoZoomer XR Digital Pathology system (Hamamatsu, Hamamatsu City, Japan) at 40× magnification. Digitalized images were automatically analysed using Visiopharm software v2021.09.02 (Visiopharm, Hørsholm, Denmark). A training set was used to configure the algorithm and identify SARS-CoV-2 RNA signals. In brief, 3,3′-diaminobenzidine (DAB) dots of positive signals were identified using a Bayesian classifier trained on pre-processing steps. We randomly selected 30 regions of interest per slide and calculated the median of positive cells. SARS-CoV-2 multiplex IF FFPE CNS sections were deparaffinized, rehydrated and subjected to 0.01 M citrate buffer antigen retrieval for 20 min at 120 °C. Slides were then rinsed briefly in deionized water, washed with PBS and subsequently incubated in a 5% milk (catalogue no. 1706404, Bio-Rad), 5% normal donkey serum, 0.3 M glycine and 0.1% Triton X-100 blocking solution made up in PBS for 30 min. Primary antibodies to SARS-CoV-2 N protein 1 (NP1, 1:1,000, custom made GenScript U864YFA140-4/CB2093)43–45 and neuronal nuclear protein (NeuN, 1:200, catalogue no. MAB377, Chemicon) or transmembrane protein 119 (TMEM119, 1:1,000, catalogue no. MAB130313, R&D Systems) were diluted in blocking serum and applied to slides overnight at 4 °C. The following day, slides were washed extensively with PBS to remove any detergent and freshly made True Black Plus solution (1:40 in PBS, catalogue no. 23014, Biotium) was applied for 14 min. Slides were again extensively washed and then species-specific secondary conjugates (1:500, catalogue nos. A-21206 and A-21203, Thermo Fisher) were applied for 1 h at room temperature. Following PBS wash, Hoechst 33342 was applied for 10 min (1:2,000, catalogue no. H3570, Thermo Fisher) to label nuclei. Slides were coverslipped with Prolong Gold (catalogue no. P36930, Thermo Fisher). SARS-CoV-2 chromogenic IHC Chromogenic IHC was carried out on various ddPCR positive CNS and lung tissues and negative pre-pandemic control cases demonstrating relative expression of the target protein between infected and control samples. Briefly, 5 µm FFPE tissue sections were incubated at 60 °C for 2 h, deparaffinized in xylene, and hydrated in serial alcohol solutions to distilled water. Heat antigen retrieval was carried out using a pressure cooker (DAKO) by submerging slides in 1× pH 6 citrate buffer for 20 min. Endogenous enzyme activity was quenched with 3% hydrogen peroxide containing sodium azide for 10 min with additional 10% non-fat dry milk (Bio-Rad) for 20 min to prevent nonspecific binding. Tissue sections were then incubated with polyclonal SARS or SARS-CoV-2 N antibody (1:500, custom made, GenScript U864YFA140-4/CB2093, 0.447 mg ml−1)43–45 for 1 h at room temperature. Negative controls were congruently stained on subsequent sections following the same protocol replacing the primary antibody with a rabbit IgG control antibody (0.5 mg ml−1, catalogue no. I-1000-5, Vector Laboratories). The antigen–antibody reaction was detected with Dako Envision+Rb polymer detection system (DAKO) and visualized with DAB chromogen. Sections were lightly counterstained with haematoxylin, dehydrated in graded alcohols, cleared in xylene, mounted and coverslipped. Statistical analysis Correlations between two continuous variables were assessed using Spearman’s rank correlation coefficient (ρ). Fisher’s z-transformation was used for the calculations of 95% CI and P values. To compare ddPCR levels between tissue types (respiratory versus non-respiratory), we used linear mixed models with compound symmetry correlation structure to account for repeated measures within each subject. Standard residual diagnoses were used to check model assumptions. Log-transformations were used when needed. To log10-transform ddPCR values, 0 values were replaced with a small positive random number according to the detection limits. Logistic regression models were used to generate ROC curves. Optimal cutoff values were selected by treating sensitivity and specificity as equally important. SAS version 9.4 was used for all analyses. All P values are two-sided and reported without adjustment for multiple comparisons. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Online content Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41586-022-05542-y. Supplementary information Reporting Summary Peer Review File Supplementary Data 1 Seroconversion and peripheral blood mononuclear cell ddPCR; ddPCR, subgenomic RNA, virus isolation; tissue heat map; fluid heat map; and ddPCR and subgenomic RNA LOD validation. Supplementary Data 2 Demographic and clinical data; and histopathology findings. Supplementary Data 3 Table detailing the cell-type specificity of SARS-CoV-2 S RNA by ISH staining throughout the human body and brain shown in subsequent image panels a–xx. Extended data figures and tables Extended Data Fig. 1 Autopsy procurement relative to Maryland COVID-19 cases, March 19th, 2020 to March 9th, 2021. Daily COVID-19 reported cases for Maryland (light blue bars) with 7-day average (dark blue line) with timing of autopsies (red arrows). Extended Data Fig. 2 Analysis of ddPCR quantification. (a) Linear mixed model analysis of estimated difference in log10ddPCR SARS-CoV-2 N copies/ng RNA between all respiratory and all non-respiratory tissues among early, mid, and late cases with SE and p-values from relevant contrasts, (b) graph of linear mixed model analysis comparing linear trends of log10ddPCR SARS-CoV-2 N copies/ng RNA by log10DOI of respiratory and non-respiratory tissues, (c) linear mixed model analysis estimating the linear trends of log10ddPCR SARS-CoV-2 N copies/ng RNA by log10DOI of individual tissue groups with intercept with standard error (SE), and slope with SE and p-values from relevant contrasts (a signification p-value indicates a non-zero slope), (d) Spearman correlation between ddPCR (N copies/ng RNA) and sgRNA (copies/µL RNA) for all tissues tested for sgRNA, with subset analyses of these tissues from early, mid and late cases and all non-respiratory and respiratory samples with 95% confidence intervals (CI), (e) Spearman correlation between ddPCR (N copies/ng RNA) and sgRNA (copies/µL RNA) for tissues jointly positive by both assays, with subset analyses of these tissues from early, mid, and late cases and all non-respiratory and respiratory samples with 95% CI, (f) receiver operating characteristic (ROC) curve of logistic regression using log10ddPCR in tissues to predict the detection of sgRNA in tissues, area under the curve is 0.965 (95% CI 0.953, 0.977), optimal cut-off for ddPCR is 1.47 N copies/ng RNA (sensitivity 93.0%, specificity 91.6%). All p-values were two-sided without adjustment for multiple comparisons. Extended Data Fig. 3 Virus isolation summary and correlation between ddPCR and sgRNA. (a) Summary of 55 tissues selected for virus isolation organized by the sgRNA qPCR quantification cycle (Cq) for the RNAlater preserved tissue, (b) Receiver operating characteristic (ROC) curve of logistic regression using log10ddPCR to predict the presence of cytopathic effect (CPE), area under the curve 0.887 (95% CI 0.795, 0.978), optimal cut-off for ddPCR is 758 N copies/ng RNA (sensitivity 76%, specificity 90%), (c) ROC curve of logistic regression using log10sgRNA to predict presence of CPE, area under the curve 0.915 (95% CI 0.843, 0.987), optimal cut-off for sgRNA is 25,069 copies/µL RNA (sensitivity 72%, specificity 100%). sgRNA qPCR was additionally performed on the flash frozen tissue homogenate and the supernatant from the least diluted tissue culture wells with CPE in order to rule out CPE from other causes; if both wells at that dilution showed CPE the samples were pooled. Extended Data Fig. 4 Analysis of SARS-CoV-2 genetic diversity across body compartments in patients. (a) P18, (b) P19, (c) P27, (d) P33, (e) P36, (f) P38. Haplotype diagrams (left) show SARS-CoV-2 spike single genome sequences detected in multiple organs. Spike NH2-terminal domain (NTD), receptor-binding domain (RBD), and furin cleavage site (F) regions are shaded grey, and remaining regions of the spike are shaded white. Ticks with different colors indicate mutations relative to the WA-1 reference sequence; green indicates non-synonymous differences from WA-1 detected in all sequences in the individual; blue indicates synonymous mutations detected variably within the individual, and pink indicates non-synonymous mutations detected variably within the individual. Bar graphs (right) show the percentage of all single genome sequences in the sample matching each haplotype. The spike region of the consensus sequence generated from short read, whole genome sequencing (WGS) of the supernatant of P38 thalamus frozen tissue on Vero E6-TMPRSS2-T2A-ACE2 cells is additional shown at the bottom of (f) for comparison. Extended Data Fig. 5 Representative histopathologic findings in COVID-19 autopsy patients. (a) Lung, Subject P22, exudative phase diffuse alveolar damage with hyaline membranes and mild interstitial inflammation (H&E, 100x), (b) Lung, Subject P26, proliferative phase diffuse alveolar damage and sparse inflammation (H&E, 200x), (c) Lung, Subject P22, organizing thrombus in medium sized pulmonary artery (H&E, 40x), (d) Lung, Subject P28. Diffuse pulmonary hemorrhage (H&E, 100x), (e) Heart, Subject P3, active lymphocytic myocarditis with cardiomyocyte necrosis (H&E, 400x), (f) Heart, Subject P38, microscopic focus of bland myocardial contraction band necrosis (H&E, 400x), (g) Liver, Subject P41, steatohepatitis with mild steatosis and scattered ballooned hepatocytes (H&E, 400x), (h) Liver, Subject P41, focal bridging fibrosis involving central hepatic veins (Masson trichrome, 40x), (i) Kidney, Subject P16, nodular glomerulosclerosis (Masson trichrome, 600x), (j) Spleen, Subject P16, preservation of white pulp and congestion (H&E, 40x), (k) Spleen, Subject P14, lymphoid depletion of white pulp with proteinaceous material and red pulp congestion (H&E, 100x), (l) Spleen, Subject P34, relative preservation of white pulp with extramedullary hematopoiesis (inset) in red pulp (H&E, 200x), (m) Lymph node, Subject P25, follicular hyperplasia with well-defined follicles (H&E), (n) Lymph node, Subject P25, marked plasmacytosis in the medullary cord (H&E, 400x), (o) Lymph node, Subject P25, marked plasmacytosis and sinus histiocytosis (H&E, 400x), (p) Brain, Subject P35, focal subarachnoid and intraparenchymal hemorrhage (H&E, 40x), (q) Brain, Subject P44, vascular congestion (H&E, 40x), (r) Brain, Subject P43, intravascular platelet aggregates (anti-CD61 stain, 100x). All H&E (and Masson trichrome) photomicrographs are exemplars of histopathology observed across a diversity of patients within the cohort, see Extended Data Table 4 for a summary of histopathology observed across the autopsy cohort and Supplemental data 2b for individual case-level data. The histopathologic observations were validated by a minimum of two board certified anatomic pathologist. Extended Data Fig. 6 Temporal association of diffuse alveolar damage in patients dying from COVID-19. Number of autopsy cases with diagnosed phase of diffuse alveolar damage (DAD) via histopathologic analysis by duration of illness. Early time points mainly show the initial exudative phase of diffuse alveolar damage, while patients dying after prolonged illness are more likely to have proliferative or fibrosing phases of DAD. Extended Data Table 1 Autopsy cohort demographics, comorbidities, and clinical intervention summary Autopsy cohort demographics, comorbidities, and clinical intervention summary (a) Summary of demographics and known comorbidities for autopsy cases, (b) Summary of illness course and clinical care for autopsy cases. Data compiled from available patient medical records. ECMO/extracorporeal membrane oxygenation. *In reference to final hospitalization prior to death if hospitalized multiple times following COVID-19 diagnosis. IQR/interquartile range. Extended Data Table 2 Summary of SARS-CoV-2 RNA and subgenomic RNA by tissue group over time Summary of SARS-CoV-2 RNA and subgenomic RNA by tissue group over time (a). Summary of the median and interquartile range of Nucleocapsid gene copies/ng RNA across by tissue group and duration of illness (days), (b) summary of the number and percentage of cases with SARS-CoV-2 RNA detected via droplet digital (dd)PCR by tissue group for all cases and by tissue and duration of illness (days). The number and percentage of tissues positive for ddPCR that were additionally positive for subgenomic (sg)RNA PCR is listed in the right most column. *A tissue positive via ddPCR was not tested via subgenomic RNA PCR. CNS/central nervous system, LN/lymph node. Extended Data Table 3 SARS-CoV-2 cellular tropism SARS-CoV-2 cellular tropism Summary of tissues with cell types that were identified as SARS-CoV-2 positive by in situ hybridization (ISH) with the associated panels demonstrating the cellular tropism within Fig. 2, Fig. 3, and Supplementary Data 3. NOS/not otherwise specified. Extended Data Table 4 Histopathologic findings of COVID-19 autopsy cases Histopathologic findings of COVID-19 autopsy cases Summary of histopathologic findings across organ system across 44 autopsy cases. Central nervous system findings are reported for the 11 cases in which consent for sampling was obtained. 1Includes one case in which the COVID lungs were transplanted and data from explanted lungs used in table. 2Individual lung weights were missing in 4 cases. 3Findings missing on 1 case due to extreme autolysis. 4Weight missing on one case. 5Lymph node findings missing in 4 cases. Extended data is available for this paper at 10.1038/s41586-022-05542-y. Supplementary information The online version contains supplementary material available at 10.1038/s41586-022-05542-y. Acknowledgements This study was financed and supported by the Intramural Research Program of the National Institutes of Health, Clinical Center, the Center for Cancer Research within the National Cancer Institute, the National Institute of Dental and Craniofacial Research and the National Institute of Allergy and Infectious Diseases. This research was made possible through the National Institutes of Health (NIH) Medical Research Scholars Program, a public–private partnership supported jointly by the NIH and contributions to the Foundation for the NIH from the Doris Duke Charitable Foundation, Genentech, the American Association for Dental Research and the Colgate-Palmolive Company. The following reagent was obtained through BEI Resources, National Institute of Allergy and Infectious Diseases, NIH: Cercopithecus aethiops kidney epithelial cells expressing transmembrane protease, serine 2 and human angiotensin-converting Enzyme 2 (Vero E6-TMPRSS2-T2A-ACE2), NR-54970. We thank C. Martens, S. Anzick and K. Barbian for whole-genome sequencing and related analysis. Author contributions D.S.C., K.M.V., S.R.S., M.J.R.-B., A.L.B., L.J.P.V., A.G., D.L.H., S.M.H. and D.E.K. contributed to the study design and protocols for autopsy procurement. A.P.P., J.M.D., M.E.R., A.G., N.H., M.P., S. Singireddy, J.W., K.R., R.C., J.E.C., A.J.B., K.A.B., A.M.W., P.A.M., M.A.N.M., E.E.K., M.M.S., K.K.S., D.L.H., T.M.S., D.T., R.J.M., S. Dahi, K.B.D., E.M.K., J.R., J.A.H., A.T., E.S.H., C.R.C., A.R.L., J.E.R., J.E., A.P.B., M.A.M., R.H.C., Z.A.C., M. Abdulmahdi, S. Sopha, T.G., S. Soherwardi, Y.S., M.T.M., K.S., D.B., B.R., M. Arnouk, J.W.E., R.P. and A.D.H. provided care for, recruited, collected samples from and/or procured medical records for the patients in this study. T.P., G.D., M.G.-C., E.S. and P.P. designed and produced the box used to safely collect CNS samples during autopsy. D.E.K., S.M.H., M.Q., W.J.Y., S.P.Y., B.G., M.S.D.M., S. Desar, S.T., N.N., X.J., S.R., E.D., N.O., K.Y., J.-Y.C., S.P. and G.S. conducted the autopsies and/or histological analysis. S.R.S., M.J.R.-B., A.P.P., J.M.D., A.L.B., L.J.P.V., S.C.R., S.J.C., E.R.E., B.L.K., J.A.O., M.B. and R.A.S. assisted with procurement and preservation of autopsy specimens. S.R.S. with assistance from S.C.R., J.M.D., A.P.P. and I.A.L. carried out RNA extraction, ddPCR and data analysis. M.S., C.K.Y., V.J.M. and E.d.W. carried out and analysed data for subgenomic RNA RT–PCR. C.W.W. and K.E.P. carried out IF. K.Y., J.-Y.C., S.C.R. and S.M.H. carried out ISH and chromogenic IHC. P.D.B. and J.I.C. measured antibody responses to SARS-CoV-2 in perimortem plasma samples. S.H.K., F.B. and E.A.B. carried out viral sequencing. J.S. carried out all statistical analyses. S.R.S. drafted the manuscript with critical input from D.S.C., K.M.V., S.M.H., D.E.K., S.C.R., A.P.P., M.J.R.-B., E.d.W., V.J.M., A.G., D.L.H., K.K.S., M.M.S., M.T.M., P.D.B., J.I.C., C.W.W., K.E.P. and S.J.C. All authors approved the submitted version of the manuscript. Peer review Peer review information Nature thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available. Data availability The datasets that support the findings of this study are available in Supplementary Data 1, 2 and 3. Positive and negative controls for ISH, IF and IHC are available in Supplementary Data 3. The sequencing data of SARS-CoV-2 isolated from Vero E6-TMPRSS2-T2A-ACE2 cell culture of thalamus of P38 have been deposited to GenBank (OP125352). Code availability The SAS code for statistical analysis has been deposited at https://github.com/niaid/COVID-19-Autopsy-SAS-Code. The bioinformatic pipeline for HT-SGS data analysis has been deposited at https://github.com/niaid/UMI-pacbio-pipeline. Competing interests The authors declare no competing interests. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. A list of authors and their affiliations appears at the end of the paper ==== Refs References 1. Bourgonje AR Angiotensin-converting enzyme 2 (ACE2), SARS-CoV-2 and the pathophysiology of coronavirus disease 2019 (COVID-19) J. Pathol. 2020 251 228 248 10.1002/path.5471 32418199 2. Salamanna F Maglio M Landini MP Fini M Body localization of ACE-2: on the trail of the keyhole of SARS-CoV-2 Front. Med. 2021 7 594495 10.3389/fmed.2020.594495 3. Sridhar S Nicholls J Pathophysiology of infection with SARS-CoV-2—what is known and what remains a mystery Respirology 2021 26 652 665 10.1111/resp.14091 34041821 4. Al-Aly Z Xie Y Bowe B High-dimensional characterization of post-acute sequelae of COVID-19 Nature 2021 594 259 264 10.1038/s41586-021-03553-9 33887749 5. Crook H Raza S Nowell J Young M Edison P Long covid-mechanisms, risk factors, and management Brit. Med. J. 2021 374 n1648 10.1136/bmj.n1648 34312178 6. Puelles VG Multiorgan and renal tropism of SARS-CoV-2 N. Engl. J. Med. 2020 383 590 592 10.1056/NEJMc2011400 32402155 7. Martines RB Pathology and pathogenesis of SARS-CoV-2 associated with fatal coronavirus disease, United States Emerg. Infect. Dis. 2020 26 2005 2015 10.3201/eid2609.202095 32437316 8. Bhatnagar J Evidence of severe acute respiratory syndrome coronavirus 2 replication and tropism in the lungs, airways, and vascular endothelium of patients with fatal coronavirus disease 2019: an autopsy case series J. Infect. Dis. 2021 223 752 764 10.1093/infdis/jiab039 33502471 9. Dorward DA Tissue-specific immunopathology in fatal COVID-19 Am. J. Respir. Crit. Care Med. 2021 203 192 201 10.1164/rccm.202008-3265OC 33217246 10. Schurink B Viral presence and immunopathology in patients with lethal COVID-19: a prospective autopsy cohort study Lancet Microbe 2020 1 e290 e299 10.1016/S2666-5247(20)30144-0 33015653 11. Yao XH A cohort autopsy study defines COVID-19 systemic pathogenesis Cell Res. 2021 31 836 846 10.1038/s41422-021-00523-8 34135479 12. Remmelink M Unspecific post-mortem findings despite multiorgan viral spread in COVID-19 patients Crit. Care 2020 24 495 10.1186/s13054-020-03218-5 32787909 13. Mukerji SS Solomon IH What can we learn from brain autopsies in COVID-19 Neurosci. Lett. 2021 742 135528 10.1016/j.neulet.2020.135528 33248159 14. Matschke J Neuropathology of patients with COVID-19 in Germany: a post-mortem case series Lancet Neurol. 2020 19 919 929 10.1016/S1474-4422(20)30308-2 33031735 15. Speranza E Single-cell RNA sequencing reveals SARS-CoV-2 infection dynamics in lungs of African green monkeys Sci. Transl. Med. 2021 13 eabe8146 10.1126/scitranslmed.abe8146 33431511 16. Matsuyama S Enhanced isolation of SARS-CoV-2 by TMPRSS2-expressing cells Proc. Natl Acad. Sci. USA 2020 117 7001 7003 10.1073/pnas.2002589117 32165541 17. Vannella KM Evidence of SARS-CoV-2-specific T-cell-mediated myocarditis in a MIS-A case Front. Imunol. 2021 12 779026 10.3389/fimmu.2021.779026 18. Desai N Temporal and spatial heterogeneity of host response to SARS-CoV-2 pulmonary infection Nat. Commun. 2020 11 6319 10.1038/s41467-020-20139-7 33298930 19. Tonkin-Hill G Patterns of within-host genetic diversity in SARS-CoV-2 Elife 2021 10 e66857 10.7554/eLife.66857 34387545 20. Lythgoe KA SARS-CoV-2 within-host diversity and transmission Science 2021 372 eabg0821 10.1126/science.abg0821 33688063 21. Valesano AL Temporal dynamics of SARS-CoV-2 mutation accumulation within and across infected hosts PLoS Pathog. 2021 17 e1009499 10.1371/journal.ppat.1009499 33826681 22. Ko SH High-throughput, single-copy sequencing reveals SARS-CoV-2 spike variants coincident with mounting humoral immunity during acute COVID-19 PLoS Pathog. 2021 17 e1009431 10.1371/journal.ppat.1009431 33831133 23. Brioschi S Heterogeneity of meningeal B cells reveals a lymphopoietic niche at the CNS border Science 2021 373 eabf9277 10.1126/science.abf9277 34083450 24. Wang Y Early developing B cells undergo negative selection by central nervous system-specific antigens in the meninges Immunity. 2021 54 2784 2794 10.1016/j.immuni.2021.09.016 34626548 25. Song E Divergent and self-reactive immune responses in the CNS of COVID-19 patients with neurological symptoms Cell Rep. Med. 2021 2 100288 10.1016/j.xcrm.2021.100288 33969321 26. Jacobs JL SARS-CoV-2 viremia is associated with COVID-19 severity and predicts clinical outcomes Clin. Infect. Dis. 2021 10 ciab686 27. Wong LR Perlman S Immune dysregulation and immunopathology induced by SARS-CoV-2 and related coronaviruses – are we our own worst enemy? Nat. Rev. Immunol. 2022 22 47 56 10.1038/s41577-021-00656-2 34837062 28. Diamond MS Kanneganti TD Innate immunity: the first line of defense against SARS-CoV-2 Nat. Immunol. 2022 23 165 176 10.1038/s41590-021-01091-0 35105981 29. Beyer DK Forero A Mechanisms of antiviral immune evasion of SARS-CoV-2 J. Mol. Biol. 2022 434 167265 10.1016/j.jmb.2021.167265 34562466 30. Binnicker MJ Can testing predict SARS-CoV-2 infectivity? The potential for certain methods to be surrogates for replication-competent virus J. Clin. Microbiol. 2021 59 e0046921 10.1128/JCM.00469-21 34346713 31. Alexandersen S Chamings A Bhatta TR SARS-CoV-2 genomic and subgenomic RNAs in diagnostic samples are not an indicator of active replication Nat. Commun. 2020 11 6059 10.1038/s41467-020-19883-7 33247099 32. van Kampen J Duration and key determinants of infectious virus shedding in hospitalized patients with coronavirus disease-2019 Nat. Commun. 2021 12 267 10.1038/s41467-020-20568-4 33431879 33. Dimcheff DE Severe acute respiratory syndrome coronavirus 2 total and subgenomic RNA viral load in hospitalized patients J. Infect. Dis. 2021 224 1287 1293 10.1093/infdis/jiab215 33870434 34. Sidhu MS Defective measles virus in human subacute sclerosing panencephalitis brain Virology 1994 202 631 641 10.1006/viro.1994.1384 8030228 35. Huang N SARS-CoV-2 infection of the oral cavity and saliva Nat. Med. 2021 27 892 903 10.1038/s41591-021-01296-8 33767405 36. Burbelo PD Sensitivity in detection of antibodies to nucleocapsid and spike proteins of severe acute respiratory syndrome coronavirus 2 in patients with coronavirus disease 2019 J. Infect. Dis. 2020 222 206 213 10.1093/infdis/jiaa273 32427334 37. Burbelo PD Goldman R Mattson TL A simplified immunoprecipitation method for quantitatively measuring antibody responses in clinical sera samples by using mammalian-produced Renilla luciferase-antigen fusion proteins BMC Biotechnol. 2005 5 22 10.1186/1472-6750-5-22 16109166 38. Wölfel R Virological assessment of hospitalized patients with COVID-19 Nature 2020 581 465 469 10.1038/s41586-020-2196-x 32235945 39. Corman VM Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR Euro Surveill. 2020 25 2000045 10.2807/1560-7917.ES.2020.25.3.2000045 31992387 40. Martin M Cutadapt removes adapter sequences from high-throughput sequence reads EMBnet J. 2011 17 10 12 10.14806/ej.17.1.200 41. Langmead B Salzberg SL Fast gapped-read alignment with Bowtie 2 Nat. Methods 2012 9 357 359 10.1038/nmeth.1923 22388286 42. Carossino M Detection of SARS-CoV-2 by RNAscope® in situ hybridization and immunohistochemistry techniques Arch. Virol. 2020 165 2373 2377 10.1007/s00705-020-04737-w 32761270 43. Williamson BN Clinical benefit of Remdesivir in rhesus macaques infected with SARS-CoV-2 Nature 2020 585 273 276 10.1038/s41586-020-2423-5 32516797 44. Munster VJ Respiratory disease in rhesus macaques inoculated with SARS-CoV-2 Nature 2020 585 268 272 10.1038/s41586-020-2324-7 32396922 45. Rosenke K Hydroxychloroquine prophylaxis and treatment is ineffective in macaque and hamster SARS-CoV-2 disease models JCI Insight 2020 5 e143174 10.1172/jci.insight.143174 33090972

36517603

PMC9749650

NO-CC CODE

2022-12-15 23:23:21

no

Nature. 2022 Dec 14;:1-6

utf-8

Nature

10.1038/s41586-022-05542-y

oa_other

==== Front J Appl Math Comput J Appl Math Comput Journal of Applied Mathematics & Computing 1598-5865 1865-2085 Springer Berlin Heidelberg Berlin/Heidelberg 1828 10.1007/s12190-022-01828-6 Original Research Dynamical analysis of a stochastic non-autonomous SVIR model with multiple stages of vaccination http://orcid.org/0000-0002-4751-4987 Mehdaoui Mohamed m.mehdaoui@edu.umi.ac.ma Alaoui Abdesslem Lamrani abdesslemalaoui@gmail.com Tilioua Mouhcine m.tilioua@umi.ac.ma grid.10412.36 0000 0001 2303 077X MAIS Laboratory, MAMCS Group, Moulay Ismail University of Meknes, P.O. Box 509, 52000 Boutalamine, Errachidia, Morocco 14 12 2022 130 17 9 2022 7 12 2022 9 12 2022 © The Author(s) under exclusive licence to Korean Society for Informatics and Computational Applied Mathematics 2022, Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. In this paper, we analyze the dynamics of a new proposed stochastic non-autonomous SVIR model, with an emphasis on multiple stages of vaccination, due to the vaccine ineffectiveness. The parameters of the model are allowed to depend on time, to incorporate the seasonal variation. Furthermore, the vaccinated population is divided into three subpopulations, each one representing a different stage. For the proposed model, we prove the mathematical and biological well-posedness. That is, the existence of a unique global almost surely positive solution. Moreover, we establish conditions under which the disease vanishes or persists. Furthermore, based on stochastic stability theory and by constructing a suitable new Lyapunov function, we provide a condition under which the model admits a non-trivial periodic solution. The established theoretical results along with the performed numerical simulations exhibit the effect of the different stages of vaccination along with the stochastic Gaussian noise on the dynamics of the studied population. Keywords Epidemic model Extinction Persistence in the mean Stochastic differential equations Periodic solution Mathematics Subject Classification 34K50 65C30 92B05 ==== Body pmcIntroduction Throughout history, researchers from different disciplines have developed scientific knowledge that played a major role in the advancement of Epidemiology. In the mathematical framework, the contribution of mathematicians consists of developing adequate models, based on a good understanding of the modeled disease, which allows to describe the evolution of the latter within the studied population, predict the worst outcome by performing virtual numerical simulations and even propose control strategies that can help reduce the severity of the situation, especially when it comes to disease outbreaks. Kermack–McKendrick theory [1] has been a cornerstone to the mathematical modeling of epidemics. The basic idea is to divide the studied population into so called compartments, based on the number of clinical states induced by the modeled disease. Then, to incorporate the transition of individuals from one clinical state to another, to each compartment, a set of parameters describing all the possible transitions are considered. Once the epidemic model is derived, it takes the form of a dynamical system, which then can be interpreted from two points of view. The first one is the deterministic point of view, which assumes that the output of the system is a time-dependent function that is entirely determined by the initial conditions and the input parameters, while the second is the stochastic point of view, which assumes that the same initial conditions and input parameters can lead to different outputs due to the random effect present in the environment. Consequently, the output, in this case, takes the form of a stochastic process. In the deterministic framework, numerous pioneering results in term of the dynamical and numerical analysis of epidemic and ecological models have been established by many authors [2–12], while several other works were done in the aim of extending the deterministic results to the stochastic case [13–20]. In further work, non-autonomous stochastic models have gained the attention of several researchers, due to their ability to incorporate the seasonal variation of diseases [21–23]. We briefly outline some of the existing literature in this sense for the stochastic case. For instance, in [24], Qi et al. analyzed an SEIS model and were able to prove that it admits a non-trivial periodic solution. Additionally, conditions under which the model admits an ergodic stationary distribution were obtained. The same results were proved by Shangguan et al. [25] for an SEIR model and by Liu et al. [26] for an SIR model. In [27], Lin et al. considered an SIR model and were able to derive a threshold characterizing the persistence and extinction of the disease. Furthermore, in the case of persistence, they proved the existence of a non-trivial periodic solution. However, to the best of the authors’ knowledge, the extension of these types of results to SVIR-type models, incorporating vaccination, has not yet been done. When it comes to stochastic epidemic models incorporating the ineffectiveness of vaccination, most of the current research works neglect the dynamics of the vaccinated population, and make use of time delays to take into consideration the duration elapsed before the effectiveness of the vaccine wears off. In this context, we mention for instance the results presented in [28, 29]. Another limitation of the aforementioned works is assuming that the immunity can be gained solely after one stage of vaccination. These assumptions can be considered in order to simplify the formulation of the model. However, for some new emerging diseases such as COVID-19 and its variants, not taking these characteristics into account in the formulation of the model can reduce the amount of information acquired from the numerical simulation. To highlight the crucial role of the multiple stages of vaccination in the acquisition of immunity, we refer the reader to the recent studies presented in [30, 31]. Hence, the main contributions of our work is to address the previous limitations by providing a different approach, allowing to incorporate the multiple stages of vaccination as well as the ineffectiveness of the first stages. More precisely, we propose a new non-autonomous stochastic model extending the standard SVIR model [2], on one hand by considering time-varying parameters, incorporating the seasonal variation, and on the other, by dividing the vaccinated population V into three sub-populations V1,V2 and V3, such that V1 and V2 stand for the vaccinated sub-population of individuals in the first and second stages of vaccination, respectively, and are not supposed to develop immunity against the disease. Consequently, they become infected. While V3 stands for the vaccinated sub-population of individuals who complete the third stage of vaccination and are supposed to develop immunity against the disease, for a large period of time. The model in question is expressed by the following system of coupled nonlinear stochastic differential equations.1 dS(t)=Λ(t)-βS(t)S(t)I(t)-μ(t)S(t)-κ1(t)S(t)dt-σ1(t)S(t)I(t)dB1(t),dV1(t)=-βV1(t)V1(t)I(t)+κ1(t)S(t)-μ(t)V1(t)-κ2(t)V1(t)dt-σ2(t)V1(t)I(t)dB2(t),dV2(t)=-βV2(t)V2(t)I(t)+κ2(t)V1(t)-μ(t)V2(t)-κ3(t)V2(t)dt-σ3(t)V2(t)I(t)dB3(t),dV3(t)=κ3(t)V2(t)-μ(t)V3(t)-γV3(t)V3(t)dt,dI(t)=βS(t)S(t)+βV1(t)V1(t)+βV2(t)V2(t)-γ(t)-μ(t)I(t)dt+σ1(t)S(t)I(t)dB1(t)+σ2(t)V1(t)I(t)dB2(t)+σ3(t)V2(t)I(t)dB3(t),dR(t)=γ(t)I(t)-μ(t)R(t)+γV3(t)V3(t)dt, equipped with the following initial conditionsS(0):=S0≥0,V1(0):=V10≥0,V2(0):=V20≥0,V3(0):=V30≥0,I(0):=I0≥0andR(0):=R0≥0, where (B1(t))t≥0,(B2(t))t≥0 and (B3(t))t≥0 are mutually independent Brownian motions defined on a probabilistic space (Ω,F,{Ft}t≥0,P) with a filtration {Ft}t≥0 which is increasing, right-continuous and such that F0 contains the null sets, while σ1(t),σ2(t) and σ3(t) denote the time-dependent intensities of the environmental Gaussian noise present in the disease transmission rates (Fig. 1, 1). Table 1 Signification of the model parameters Parameter Biological signification Λ(t) Natural birth rate at time t μ(t) Natural death rate at time t βS(t) Rate in which a susceptible individual at time t becomes infected βVi(t) Rate in which an individual at time t and in the ith stage of vaccination (i∈{1,2}) becomes infected γ(t) Natural recovery rate at time t γV3(t) Rate in which an individual at time t and in the third stage of vaccination possesses immunity κi(t) Rate in which a susceptible individual at time t reaches the ith stage of vaccination (i∈{1,2,3}) Fig. 1 Flow diagram of the model (1) in the deterministic case In order to unify the notations, we setu(t)=Δ(S(t),V1(t),V2(t),V3(t),I(t),R(t))⊤,u0=Δ(S0,V10,V20,V30,I0,R0)⊤,dB(t)=Δ(dB1(t),dB2(t),dB3(t),dB4(t),dB5(t),dB6(t))⊤,θ(t)=Δ(Λ(t),βS(t),μ(t),κ1(t),κ2(t),κ3(t),βV1(t),βV2(t),γ(t),γV3(t))⊤,f(t,u(t))=Δ(f1(t,u(t)),f2(t,u(t)),f3(t,u(t)),f4(t,u(t)),f5(t,u(t)),f6(t,u(t)))⊤, wheref1(t,u(t)):=Λ(t)-βS(t)S(t)I(t)-μ(t)S(t)-κ1(t)S(t),f2(t,u(t)):=-βV1(t)V1(t)I(t)+κ1(t)S(t)-μ(t)V1(t)-κ2(t)V1(t),f3(t,u(t)):=-βV2(t)V2(t)I(t)+κ2(t)V1(t)-μ(t)V2(t)-κ3(t)V2(t),f4(t,u(t)):=κ3(t)V2(t)-μ(t)V3(t)-γV3(t)V3(t),f5(t,u(t)):=βS(t)S(t)I(t)+βV1(t)V1(t)I(t)+βV2(t)V2(t)I(t)-γ(t)I(t)-μ(t)I(t),f6(t,u(t)):=γ(t)I(t)-μ(t)R(t)+γV3(t)V3(t), andg(t,u(t)):=-σ1(t)S(t)I(t)000000-σ2(t)V1(t)I(t)000000-σ3(t)V2(t)I(t)000000000[1ex]σ1(t)S(t)I(t)σ2(t)V1(t)I(t)σ3(t)V2(t)I(t)000000000. Then, the model (1) can be rewritten in the following abstract compact form2 du(t)=f(t,u(t))dt+g(t,u(t))dB(t),u(0)=u0≥0. When no confusion occurs, the value of a given function h at time t∈(0,T) will occasionally be denoted h and we shall omit the explicit notation. Given a function V∈C1,2(R+×R6,R). The differential operator associated with (2) is defined as followsLV(t,u)=∂V(t,u)∂t+∇uV(t,u).f(t,u)+12trg⊤(t,u)Hessu(V(t,u))g(t,u), where ∇u:=∂∂u1,⋯,∂∂u6, tr denotes the trace operator, ⊤ stands for the transpose operation, while Hessu is the Hessian matrix with respect to u. Itô’s formula [32] states thatdV(t,u)=LV(t,u)dt+∇uf(t,u(t)).g(t,u(t))dB(t). We now announce some definitions and notations that will be used throughout the paper.For T>0, denote by C([0, T]) the Banach space of real-valued continuous functions defined on [0, T]. Given f∈C([0,T]), we define f¯:=supt∈[0,T]|f(t)|andf_:=inft∈[0,T]|f(t)|. For an integrable function f:(0,T)→R, we set ⟨f⟩t:=1t∫0tf(s)ds∀t∈(0,T). Given a,b∈R, we set a∨b=Δsup{a,b}anda∧b=Δinf{a,b}. Consider the following open bounded set U:=u∈(0,+∞)6,∑i=16ui<Λ¯μ_. Hereafter, T is a strictly positive real number and it is assumed thatθi,σj∈C([0,T])andθi_,θi¯,σj_,σj¯>0,∀(i,j)∈{1,⋯,10}×{1,2,3}. The rest of this paper is organized as follows: In Sect. 2, we study the mathematical and biological well-posedness of the model (1). We devote Sect. 3 to establish conditions under which the infected population becomes extinct or persistent in the mean. While in Sect. 4, we provide a condition under which the model (1) admits a non-trivial periodic solution. Additionally, in order to support the theoretical results, in Sect. 5, we present the outcome of the performed numerical simulations. Finally, we leave Sect. 6 to state some conclusions and future works. Mathematical and biological well-posedness We begin this section by stating a remark, which will be useful overall throughout the paper. Remark 1 It can be seen that the set I:={u∈R+6,∑i=16ui≤Λ¯μ_}, is positively invariant for the stochastic system (1). Indeed, define the total population at time t∈(0,T) by N(t):=∑i=16ui(t). Direct application of the comparison principle yieldsN(t)≤N(0)exp(-μ_t)+Λ¯μ_1-exp(-μ_t). Then if u0∈I, it follows that u(t)∈I∀t∈(0,T). Additionally,limt→+∞N(t)≤Λ¯μ_almost surely. Theorem 1 For every initial condition u0∈I, the stochastic system (2) admits a unique global, almost surely positive solution. Proof Since the coefficients of the stochastic system (2) satisfy the local Lipschitz condition, by the standard theory of stochastic differential equations [32], there exists a unique local solution u defined up to a maximal time of existence that we denote Tmax. In order to prove that the local solution is a global one that remains almost surely positive, let n~∈N∗ be sufficiently large such that u0∈1n~,Λ¯μ_)6. Then, for n≥n~ define the following stopping timeτn:=inft∈[0,Tmax)∃i0∈{1,⋯,6}ui0(t)≤1n, with the usual convention inf∅=+∞, where ∅ denotes the empty set. It is clear that the sequence (τn)n≥n~ is increasing and τn≤Tmax. Hence, there exists τl such that limn→+∞τn=τl≤Tmax. Thus, it suffices to prove that τl=+∞. We argue by contradiction and suppose that there exist ϵ∈(0,1), T>0 and n0≥n~ such that∀n≥n0,P(τn≤T)≥ϵ. Now, consider the following function F:U⟶R+ defined by F(u):=-∑i=16lnμ_uiΛ¯. By Itô’s formula, it holds thatdF=-1SΛ-βSSI-μS-κ1S-1V1-βV1V1I+κ1S-μV1-κ2V1-1V2-βV2V2I+κ2V1-μV2-κ3V2-1V3κ3V2-μV3-γV3V3-1IβSSI+βV1V1I+βV2V2I-γI-μI-1RγI-μR+γV3V3+12σ12+σ22+σ32I2+12σ12S2+σ22V12+σ32V22dt+σ1(I-S)dB1+σ2(I-V1)dB2+σ3(I-V2)dB3. Thereby, by using Remark 1, it follows that3 dF≤Cdt+σ1(I-S)dB1+σ2(I-V1)dB2+σ3(I-V2)dB3, whereC:=Λ¯μ_βS¯+βV1¯+βV2¯+6μ¯+κ1¯+κ2¯+κ3¯+γ¯+γV3¯+Λ¯2μ_2σ1¯2∨σ2¯2∨σ3¯2. By integrating both sides of inequality (3) from 0 to T∧τn and evaluating the expectation, we obtainE(F(u(T∧τn)))≤F(u(0))+CT. On the other hand, by definition of τn, there exists i0∈{1,⋯,6} such that ui0(τn)≤1n. Consequently, -lnμ_ui0(τn)Λ¯≥-lnμ_Λ¯n. Therefore, F(u(τn))≥-lnμ_Λ¯n. Hence, due to the positiveness of F, it holds that4 -lnμ_Λ¯n≤E(F(u(τn)1τn≤T))≤E(F(u(T∧τn)))≤F(u(0))+CT. where 1 stands for the indicator function. Letting n→+∞ in inequality (4) leads to the contradiction +∞≤F(u(0))+CT<+∞. Thus, Tmax=+∞ and the solution is global and remains almost surely positive. □ Analysis of the disease extinction and persistence In this section, we are interested in establishing conditions under which the disease vanishes or persists. To this end, we define the following parametersR1s(t):=Λ¯μ_βS(t)+βV1(t)+βV2(t)μ(t)+γ(t)-Λ¯2μ_2(μ(t)+γ(t))12σ12(t)+12σ22(t)+12σ32(t), ∀t∈(0,T), andR2s:=Λ_βS_μ¯+κ2¯μ¯+κ3¯+Λ_κ1_βV1_μ¯+κ3¯+Λ_κ1_κ2_βV2_μ¯+κ1¯μ¯+κ2¯μ¯+κ3¯μ¯+γ¯+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2. Theorem 2 Let u be the solution of the system (2) with the initial value u0∈I. If one of the following conditions lim supt→+∞ϱ(t)<0, where ϱ(t):=βS2σ12t+βV12σ22t+βV22σ32t-2⟨μ⟩t+⟨γ⟩t, ⟨R1s⟩T<1, and ∀t∈(0,T), μ_Λ¯βS(t)>σ12(t), μ_Λ¯βV1(t)>σ22(t), μ_Λ¯βV2(t)>σ32(t), is satisfied, then the infected population goes to extinction. That is, lim supt→+∞I(t)=0almost surely. Proof By using Itô’s formula, it holds that5 d(ln(I))=βSS+βV1V1+βV2V2-μ-γ-12σ12S2-12σ22V12-12σ32V22dt+σ1SdB1+σ2V1dB2+σ3V2dB3=-βS2σ1-σ1S22-βV12σ2-σ2V122-βV22σ3-σ3V222+βS22σ12+βV122σ22+βV222σ32-μ-γdt+σ1SdB1+σ2V1dB2+σ3V2dB3≤βS22σ12+βV122σ22+βV222σ32-μ-γdt+σ1SdB1+σ2V1dB2+σ3V2dB3. Dividing inequality (5) by t>0, then integrating from 0 to t yields6 ln(I(t))t≤ln(I(0))t+12βS2σ12t+12βV12σ22t+12βV22σ32t-⟨μ⟩t-⟨γ⟩t+1tM1(t), where M1(t) is a local continuous martingale satisfying M1(0)=0, and is defined byM1(t):=∫0tσ1(s)S(s)dB1(s)+σ2(s)V1(s)dB2(s)+σ3(s)V2(s)dB3(s). By evaluating the supremum limit on both sides of inequality (6) and by the law of large numbers for local martingales [32], we have limt→+∞M1(t)t=0, almost surely. Consequently, we obtainlim supt→+∞ln(I(t))t≤lim supt→+∞12βS2σ12t+βV12σ22t+βV22σ32t-⟨μ⟩t-⟨γ⟩t. Hence, if condition (1) is satisfied. Then lim supt→+∞I(t)=0 almost surely. Now, we suppose that μ_Λ¯βS(t)>σ12(t),μ_Λ¯βV1(t)>σ22(t)andμ_Λ¯βV2(t)>σ32(t),∀t∈(0,T). By using Itô’s formula and taking Remark 1 into account, we obtain7 d(ln(I))=βSS-12σ12S2+βV1V1-12σ22V12+βV2V2-12σ32V22-μ-γdt+σ1SdB1+σ2V1dB2+σ3V2dB3≤βSΛ¯μ_-12σ12Λ¯2μ_2+βV1Λ¯μ_-12σ22Λ¯2μ_2+βV2Λ¯μ_-12σ32Λ¯2μ_2-μ-γdt+σ1SdB1+σ2V1dB2+σ3V2dB3=μ+γΛ¯μ_βS+βV1+βV2μ+γ-Λ¯2μ_2(μ+γ)12σ12+12σ22+12σ32-1dt+σ1SdB1+σ2V1dB2+σ3V2dB3. By dividing inequality (7) by t>0 and integrating from 0 to t, we acquire that8 ln(I(t))t≤ln(I(0))t+⟨μ⟩t+⟨γ⟩t⟨R1s⟩t-1+1tM1(t). By applying the supremum limit on both sides of inequality (8), it follows thatlim supt→+∞ln(I(t))t≤⟨R1s⟩T-1lim supt→+∞⟨μ⟩t+⟨γ⟩t. Hence, if ⟨R1s⟩T<1, it follows that lim supt→+∞I(t)=0 almost surely. □ We now proceed to derive the condition under which the infected population becomes persistent in the mean. Namely, under a suitable condition, we prove that: ∃α>0,lim inft→+∞⟨I⟩t≥α almost surely. Theorem 3 Let u be the solution of the system (2) with the initial value u0∈I. Under the following condition9 R2s>1, the infected population is persistent in the mean. More precisely, lim inft→+∞⟨I⟩t≥λλ0 almost surely, whereλ:=μ¯+γ¯+12σ1¯2Λ¯2μ2_+12σ2¯2Λ¯2μ2_+12σ3¯2Λ¯2μ2_R2s-1, λ0:=α3βS_βS¯Λ¯α2+βV1_βV1¯Λ¯α1+Λ¯βS¯κ1_βV1_μ_α1α2α3+α1βV2¯βV2_Λ¯α2+Λ¯κ2_βV1¯βV2_+Λ¯βS¯κ1_κ2_βV2_μ_α1α2α3, α1:=μ¯+κ1¯,α2:=μ¯+κ2¯andα3:=μ¯+κ3¯. Proof By using Itô’s formula, it holds that10 d(ln(I))=βSS+βV1V1+βV2V2-μ-γ-12σ12S2-12σ22V12-12σ32V22dt+σ1SdB1+σ2V1dB2+σ3V2dB3≥βS_S+βV1_V1+βV2_V2-μ¯-γ¯-12σ1¯2Λ¯2μ2¯-12σ2¯2Λ¯2μ2_-12σ3¯2Λ¯2μ2_dt+σ1SdB1+σ2V1dB2+σ3V2dB3. An integration of inequality (10) from 0 to t and a divison by t>0 lead to11 ln(I(t))-ln(I(0))t≥βS_⟨S⟩t+βV1_⟨V1⟩t+βV2_⟨V2⟩t-μ¯-γ¯-12σ1¯2Λ¯2μ2_-12σ2¯2Λ¯2μ2_-12σ3¯2Λ¯2μ2_+M1(t)t, where M1(t) is the local continuous martingale defined in the proof of Theorem 2. Now, by taking Remark 1 into account, an integration of the first three equations of the stochastic system (1) from 0 to t and a division by t>0 yield12 ⟨S⟩t≥1μ¯+κ1¯-βS¯Λ¯μ_⟨I⟩t+Λ_-S(t)-S(0)t+M2(t)t,⟨V1⟩t≥1μ¯+κ2¯-βV1¯Λ¯μ_⟨I⟩t+κ1_μ¯+κ1¯Λ_-βS¯Λ¯μ_⟨I⟩t-S(t)-S(0)t+M2(t)t-V1(t)-V1(0)t+M3(t)t,⟨V2⟩t≥1μ¯+κ3¯-βV2¯Λ¯μ_⟨I⟩t+κ2_μ¯+κ2¯-βV1¯Λ¯μ_⟨I⟩t+κ1_μ¯+κ1¯×Λ_-βS¯Λ¯μ_⟨I⟩t-S(t)-S(0)t+M2(t)t-V1(t)-V1(0)t+M3(t)t-V2(t)-V2(0)t+M4(t)t, where M2(t),M3(t) and M4(t) are continuous local martingales, satisfying M2(0)=M3(0)=M4(0)=0, and are defined by M2(t):=∫0t-σ1(s)S(s)I(s)dB1(s), M3(t):=∫0t-σ2(s)V1(s)I(s)dB2(s), and M4(t):=∫0t-σ3(s)V2(s)I(s)dB3(s). By injecting the inequalities of (12) into the inequality (11) and rearranging the terms, we obtainln(I(t))-ln(I(0))t≥μ¯+γ¯+12σ1¯2Λ¯2μ2_+12σ2¯2Λ¯2μ2_+12σ3¯2Λ¯2μ2_R2s-1+M1(t)t+βS_μ¯+κ1¯+κ1_βV1_μ¯+κ1¯μ¯+κ2¯+κ1_κ2_βV2_μ¯+κ1¯μ¯+κ2¯μ¯+κ3¯×M2(t)t+βV1_μ¯+κ2¯+κ2_βV2_μ¯+κ2¯μ¯+κ3¯M3(t)t+βV2_μ¯+κ3¯M4(t)t-Λ¯βS¯βS_μ_μ¯+κ1¯+Λ¯βV1¯βV1_μ_μ¯+κ2¯+κ1_βV1_βS¯Λ¯μ_μ¯+κ1¯μ¯+κ2¯+κ2_βV2_βV1¯Λ¯μ_μ¯+κ2¯μ¯+κ3¯+Λ¯βV2¯βV2_μ_μ¯+κ3¯+κ1_κ2_βV2_βS¯Λ¯μ_μ¯+κ1¯μ¯+κ2¯μ¯+κ3¯⟨I⟩t-βS_μ¯+κ1¯+κ1_βV1_μ¯+κ1¯μ¯+κ2¯+κ1_κ2_βV2_μ¯+κ1¯μ¯+κ2¯μ¯+κ3¯S(t)-S(0)t-βV1_μ¯+κ2¯+κ2_βV2_μ¯+κ2¯μ¯+κ3¯V1(t)-V1(0)t-βV2_μ¯+κ3¯×V2(t)-V2(0)t. Consequently13 ln(I(t))t⩾λ-λ0⟨I⟩t+H(t)talmostsurely,∀t⩾0, where λ and λ0 are as defined in Theorem 3, andH(t):=M1(t)+βS_μ¯+κ1¯+κ1_βV1_μ¯+κ1¯μ¯+κ2¯+κ1_κ2_βV2_μ¯+κ1¯μ¯+κ2¯μ¯+κ3¯M2(t)+βV1_μ¯+κ2¯+κ2_βV2_μ¯+κ2¯μ¯+κ3¯M3(t)+βV2_μ¯+κ3¯M4(t)-βS_μ¯+κ1¯+κ1_βV1_μ¯+κ1¯μ¯+κ2¯+κ1_κ2_βV2_μ¯+κ1¯μ¯+κ2¯μ¯+κ3¯S(t)-S(0)-βV1_μ¯+κ2¯+κ2_βV2_μ¯+κ2¯μ¯+κ3¯V1(t)-V1(0)-βV2_μ¯+κ3¯V2(t)-V2(0)+ln(I(0)). By the law of large numbers for local martingales and by taking Remark 1 into account, it follows that limt→+∞H(t)t=0, almost surely. The result follows by letting t⟶+∞ in (13). □ Remark 2 We emphasize that in the case of non-autonomous epidemic models with Gaussian noise in the disease transmission, the characterization of the disease extinction and persistence in terms of one stochastic threshold has not been done, due to major difficulties caused by the considered type of noise as well as the time varying parameters, prohibiting to define a unified stochastic threshold. Such a characterization can be obtained for the autonomous case (see e.g. [19]). On the other hand, for the model (1), considered in this paper, the characterization of the disease extinction and persistence is given independently, in terms of the two stochastic parameters R1s and R2s. However, for the autonomous counterpart of the model, that is, when the model parameters don’t depend on time, following the approach used in Theorem 3, it can be proved that when R2s<1, the infected population goes to extinction. Consequently, R2s can be seen as a stochastic threshold characterizing the disease persistence and extinction, in the stochastic case. Furthermore, in the absence of Gaussain noise, R2s coincides with the basic reproduction number corresponding to the deterministic counterpart of the model. Existence of a non-trivial periodic solution In this section, we investigate the condition under which the system (2) admits a non-trivial periodic solution. From the biological point of view, the existence of such a solution means that the susceptible, vaccinated, infected and recovered populations are persistent. Meaning that their corresponding densities remain strictly positive throughout time. Hence, for diseases with seasonal characteristics, by analyzing the existence of such solutions, one can obtain additional conditions under which, the disease persists within the studied population. In order to achieve the main result of this section, we recall the definition of a periodic stochastic process. Definition 1 (See [33]) A stochastic process (η(t))t∈R is said to be periodic with period ν if for every finite sequence of numbers t1,t2,…,tn, the joint distribution of random variables ηt1+h,ηt2+h,⋯,ηtn+h, is independent of h, where h:=kν(k=±1,±2,⋯). Lemma 1 (See [33]) Let (X(t))t≥t0 be an l-dimensional stochastic process, consider the following system dX(t)=b(t,X(t))dt+σ(t,X(t))dB(t), such that the corresponding coefficients are ν-periodic in t and satisfy the local Lipschitz condition with respect to X. If there exists a function V∈C1,2((0,+∞)×Rl,R) such that V is ν-periodic with respect to t∈(0,+∞). inf|x|>RV(t,x)⟶+∞asR⟶+∞∀t∈(0,+∞). LV(t,x)≤-1 outside some compact set. Then, there exists a solution of the above system, which is a ν-periodic Markov process. Theorem 4 Suppose that (θi)i=110 and (σ)i=13 are periodic functions and denote by ν>0 their corresponding period. Moreover, let u be the solution of the system (2) with the initial value u0∈I. Define the following parametersR1:=⟨ΛβS12⟩ν2⟨μ+γ⟩ν+12⟨σ12+σ22+σ32⟩νΛ¯2μ2_⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_, R2:=⟨Λκ1βV113⟩ν3⟨μ+γ⟩ν+12⟨σ12+σ22+σ32⟩νΛ¯2μ2_⟨μ+κ2⟩ν+12⟨σ22⟩νΛ¯2μ2_×1⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_, andR3:=⟨Λκ1κ2βV214⟩ν4⟨μ+γ⟩ν+12⟨σ12+σ22+σ32⟩νΛ¯2μ2_⟨μ+κ3⟩ν+12⟨σ32⟩νΛ¯2μ2_×1⟨μ+κ2⟩ν+12⟨σ22⟩νΛ¯2μ2_. Set R:=R1+R2+R3. If the following condition14 R>1, is satisfied, then the stochastic system (2) admits a ν-periodic solution. Proof We consider the following function V:(0,+∞)×U⟶R defined byV(t,u):=M(-b1+b2+b3lnμ_Λ¯S-b4+b5lnμ_Λ¯V1-b6lnμ_Λ¯V2-lnμ_Λ¯I+ω(t))+S+V1+V2+V3+I+R-lnμ_Λ¯S-lnμ_Λ¯V1-lnμ_Λ¯V2-lnμ_Λ¯V3-lnμ_Λ¯R, such thatb1:=⟨ΛβS12⟩ν2⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_2,b2:=⟨Λκ1βV113⟩ν3⟨μ+κ2⟩ν+12⟨σ22⟩νΛ¯2μ2_⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_2,b3:=⟨Λκ1κ2βV214⟩ν4⟨μ+κ3⟩ν+12⟨σ32⟩νΛ¯2μ2_⟨μ+κ2⟩ν+12⟨σ22⟩νΛ¯2μ2_⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_2,b4:=⟨Λκ1βV113⟩ν3⟨μ+κ2⟩ν+12⟨σ22⟩νΛ¯2μ2_2⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_,b5:=⟨Λκ1κ2βV214⟩ν4⟨μ+κ3⟩ν+12⟨σ32⟩νΛ¯2μ2_⟨μ+κ2⟩ν+12⟨σ22⟩νΛ¯2μ2_2⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_, andb6:=⟨Λκ1κ2βV214⟩ν4⟨μ+κ3⟩ν+12⟨σ32⟩νΛ¯2μ2_2⟨μ+κ2⟩ν+12⟨σ22⟩νΛ¯2μ2_⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_, while M∈R+∗andω:(0,T)⟶Ris a periodic function, which will both be chosen thereafter accordingly. For simplicity, setV1(u):=-b1+b2+b3lnμ_Λ¯S-b4+b5lnμ_Λ¯V1-b6lnμ_Λ¯V2-lnμ_Λ¯I. By applying Itô’s formula and rearranging the terms, we obtainL(V1(u))=-b1ΛS-βSS+b1μ+κ1+12σ12I2-b2ΛS-βV1V1-b4κ1SV1+b2μ+κ1+12σ12I2+b4μ+κ2+12σ22I2-b3ΛS-b5κ1SV1-b6κ2V1V2-βV2V2+b3μ+κ1+12σ1I2+b5μ+κ2+12σ22I2+b6μ+κ3+12σ32I2+b1+b2+b3βS+b4+b5βV1+b6βV2I+μ+γ+12σ12S2+σ22V12+σ32V22. Owing to the inequality of arithmetic and geometric means, we acquire thatL(V1(u))≤-2ΛβSb112+b1μ+κ1+12σ12Λ¯2μ_2-3Λκ1βV1b2b413+b2μ+κ1+12σ12Λ¯2μ_2+b4μ+κ2+12σ22Λ¯2μ_2-4Λκ1κ2βV2b3b5b614+b3μ+κ1+12σ12Λ¯2μ_2+b6μ+κ3+12σ32Λ¯2μ_2+b5μ+κ2+12σ22I2+b1+b2+b3βS+b4+b5βV1+b6βV2I+μ+γ+12Λ¯2μ_2σ12+σ22+σ32. For t∈(0,T), setζ(t):=-2Λ(t)βS(t)b112+b1μ(t)+κ1(t)+12σ12(t)Λ¯2μ_2-3Λ(t)κ1(t)βV1(t)b2b413+b2μ(t)+κ1(t)+12σ12(t)Λ¯2μ_2+b4μ(t)+κ2(t)+12σ22(t)Λ¯2μ_2+b5μ(t)+κ2(t)+12σ22(t)Λ¯2μ_2-4Λ(t)κ1(t)κ2(t)βV2(t)b3b5b614+b3μ(t)+κ1(t)+12σ12(t)Λ¯2μ_2+b6μ(t)+κ3(t)+12σ32(t)Λ¯2μ_2+μ(t)+γ(t)+12Λ¯2μ_2σ12(t)+σ22(t)+σ32(t), andξ:=b1+b2+b3βS¯+b4+b5βV1¯+b6βV2¯, so thatL(V1(u(t)))≤ζ(t)+ξI. Let ω satisfyω′(t)=⟨ζ⟩ν-ζ(t),ω(0)=0. Then, clearly the ν-periodicity of ζ implies that of ω. Indeed, taking into account thatω(ν)=ν⟨ζ⟩ν-∫0νζ(s)ds=ν⟨ζ⟩ν-1ν∫0νζ(s)ds=ν(⟨ζ⟩ν-⟨ζ⟩ν)=0, we obtainω(t+ν)=ω(ν)+∫νt+ν⟨ζ⟩ν-ζ(s)ds=∫0t⟨ζ⟩ν-ζ(u+ν)du=∫0t⟨ζ⟩ν-ζ(u)du=ω(t). ThusL(V1(u)+ω(t))≤⟨ζ⟩ν+ξI. Since-2⟨ΛβSb112⟩ν=-2⟨ΛβS12⟩ν2⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_,-3⟨Λκ1βV1b2b413⟩ν=-3⟨Λκ1βV113⟩ν3⟨μ⟩ν+12⟨σ22⟩νΛ¯2μ2_⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_, and-4⟨Λκ1κ2βV2b3b5b614⟩ν=-4⟨Λκ1κ2βV214⟩ν4⟨μ+κ3⟩ν+12⟨σ32⟩νΛ¯2μ2_⟨μ+κ2⟩ν+12⟨σ22⟩νΛ¯2μ2_×1⟨μ+κ1⟩ν+12⟨σ12⟩νΛ¯2μ2_, it follows thatL(V1(u)+ω(t))≤-⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_⟨σ12+σ22+σ32⟩νR-1+ξI. On the other hand, by Itô’s formula, one can obtainL-lnμ_Λ¯S≤-Λ_S+βS¯I+μ¯+κ1¯+12σ1¯2Λ¯2μ_2,L-lnμ_Λ¯V1≤βV1¯I-κ1_SV1+μ¯+κ2¯+12σ2¯2Λ¯2μ_2,L-lnμ_Λ¯V2≤βV2¯I-κ2_V1V2+μ¯+κ3¯+12σ3¯2Λ¯2μ_2,L-lnμ_Λ¯V3≤-κ3_V2V3+μ¯+γV3¯,L-lnμ_Λ¯R≤μ¯-γV3_V3R, and LS+V1+V2+V3+I+R≤Λ¯-μ_S+V1+V2+V3+I+R. Thereby,L(V(t,u))≤M-⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_σ12+σ22+σ32R-1+ξI+Λ¯-μ_S+V1+V2+V3+I+R-Λ_S+βS¯+βV1¯+βV2¯I+κ1¯+κ2¯+κ3¯-κ1_SV1-κ2_V1V2-κ3_V2V3+5μ¯+γV3¯-γV3_V3R+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2. Now, consider the following compact setK:=u∈U,ϵi≤ui≤1ϵi∀i∈{1,⋯,6}, such that ϵ1,⋯,ϵ6>0 will be chosen later. Let u∈U\K. To verify that L(V(t,.))≤-1 in U\K, it suffices to investigate the following distinguished seven cases Case 1 : u∈u∈U,I<ϵ5. Case 2 : u∈u∈U,I≥ϵ5,S<ϵ1. Case 3 : u∈u∈U,I≥ϵ5,S≥ϵ1,V1<ϵ2. Case 4 : u∈u∈U,I≥ϵ5,S≥ϵ1,V1≥ϵ2,V2<ϵ3. Case 5 : u∈u∈U,I≥ϵ5,S≥ϵ1,V1≥ϵ2,V2≥ϵ3,V3<ϵ4. Case 6 : u∈u∈U,I≥ϵ5,S≥ϵ1,V1≥ϵ2,V2≥ϵ3,V3≥ϵ4R<ϵ6. Case 7 : u∈u∈U,∃j∈{1,⋯,6},uj>1ϵj. For case 1, we obtainL(V(t,u))≤M-⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_⟨σ12+σ22+σ32⟩νR-1+ξϵ5+Λ¯+βS¯+βV1¯+βV2¯ϵ5+κ1¯+κ2¯+κ3¯+5μ¯+γV3¯+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2. For case 2, we obtainL(V(t,u))≤M-⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_⟨σ12+σ22+σ32⟩νR-1+ξΛ¯μ_+Λ¯-Λ_ϵ1+βS¯+βV1¯+βV2¯Λ¯μ_+κ1¯+κ2¯+κ3¯+5μ¯+γV3¯+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2. For case 3, we obtainL(V(t,u))≤M-⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_⟨σ12+σ22+σ32⟩νR-1+ξΛ¯μ_+Λ¯+βS¯+βV1¯+βV2¯Λ¯μ_+κ1¯+κ2¯+κ3¯-κ1_ϵ1ϵ2+5μ¯+γV3¯+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2. For case 4, we obtainL(V(t,u))≤M-⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_⟨σ12+σ22+σ32⟩νR-1+ξΛ¯μ_+Λ¯+βS¯+βV1¯+βV2¯Λ¯μ_+κ1¯+κ2¯+κ3¯-κ2_ϵ2ϵ3+5μ¯+γV3¯+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2. For case 5, we obtainL(V(t,u))≤M-⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_⟨σ12+σ22+σ32⟩νR-1+ξΛ¯μ_+Λ¯+βS¯+βV1¯+βV2¯Λ¯μ_+κ1¯+κ2¯+κ3¯-κ3_ϵ3ϵ4+5μ¯+γV3¯+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2. For case 6, we obtainL(V(t,u))≤M-⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_⟨σ12+σ22+σ32⟩νR-1+ξΛ¯μ_+Λ¯+βS¯+βV1¯+βV2¯Λ¯μ_+κ1¯+κ2¯+κ3¯-γV3_ϵ4ϵ6+5μ¯+γV3¯+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2. For case 7, we obtainL(V(t,u))≤M-⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_⟨σ12+σ22+σ32⟩νR-1+ξΛ¯μ_+Λ¯+βS¯+βV1¯+βV2¯Λ¯μ_+κ1¯+κ2¯+κ3¯-μ_1ϵj+5μ¯+γV3¯+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2. Now, setM≥2+Λ¯+βS¯+βV1¯+βV2¯Λ¯μ_+κ1¯+κ2¯+κ3¯+5μ¯+γV3¯+12Λ¯2μ_2σ1¯2+σ2¯2+σ3¯2⟨μ⟩ν+⟨γ⟩ν+12Λ¯μ_⟨σ12+σ22+σ32⟩νR-1. Then, for case 1 and j=5 in case 7, chooseϵ5:=min1,1Mξ+βS¯+βV1¯+βV2¯,μ_2Λ¯Mξ. For case 2, case 3, case 4, case 5, case 6 and j≠5 in case 7, chooseϵ1=min1,μ_Λ_MξΛ¯,μ_κ1_MξΛ¯,κ2_μ_MΛ¯ξ,κ3_μ_MΛ¯ξ,γV3_μ_MΛ¯ξ,μ_2λ¯Mξ,μ_2λ¯Mξ12,μ_2λ¯Mξ13,μ_2λ¯Mξ14,μ_2λ¯Mξ15, ϵ2=ϵ12,ϵ3=ϵ13,ϵ4=ϵ14andϵ6=ϵ15. Consequently, L(V(t,.))≤-1inU\K. On the other hand, since inf|u|>RV(t,u)⟶+∞asR⟶+∞∀t∈(0,+∞), and taking into consideration the ν-periodicity of V with respect to t,  the assumptions of Lemma 1 are verified. Conclusively, the stochastic system (2) admits a ν-periodic solution. □ Fig. 2 Verification of the first disease extinction condition Fig. 3 Paths of S, V1, V2, V3, I and R when the first condition of the disease extinction holds Fig. 4 Verification of the second disease extinction condition Fig. 5 Paths of S, V1, V2, V3, I and R when the second condition of the disease extinction holds Fig. 6 Paths of S, V1, V2, V3, I and R when the condition of the disease persistence in the mean holds Fig. 7 Persistence in the mean of the infected population Fig. 8 Probability density functions of S, V1, V2, V3, I and R at time t=200 when the condition of the disease persistence in the mean holds Fig. 9 Paths of S, V1, V2, V3, I and R in the deterministic non-autonomous case and under the condition R>1 Fig. 10 Paths of S, V1, V2, V3, I and R in the stochastic non-autonomous case and under the condition R>1 Fig. 11 Probability density functions of S,V1,V2,V3,I and R at time t=200 in the stochastic non-autonomous case and under the condition R>1 Remark 3 For the deterministic autonomous counterpart of the model (1), that is σ1(t)=σ2(t)=σ3(t)=0andθ(t)=θ∈(0,+∞)10∀t∈(0,T). One can use the Next Generation Method [34] to compute the basic reproduction number R0, which yields that R0=ΛβSμ+κ2μ+κ3+Λκ1βV1μ+κ3+Λκ1κ2βV2μ+κ1μ+κ2μ+κ3μ+γ. Hence, the value of R0 coincides with that of R2s and R stated in Sects. 3 and 4, respectively. Numerical simulations We consider the time horizon (0, 200) and we choose the following initial condition u0=0.8,0.1,0.01,0.04,0.03,0.02. We simulate the model (1) numerically by relying on Matlab software [35] to develop a script implementing the Milstein method presented in [36], which was chosen due to its accuracy. The resulting numerical scheme of the model (1) is the same one presented in [37, 38] and hence is omitted here for brevity. To support all the established theoretical results, five cases are numerically simulated. In the first and second cases, the parameters are chosen such that the conditions (1) and (2) stated in Theorem 2 are verified, respectively. In the third case, the parameters are chosen such that the condition (9) of Theorem 3 is verified. While the fourth and fifth cases correspond to choices of parameters in which the condition (14) of Theorem 4 is verified in both deterministic and stochastic non autonomous cases. For the first case, Fig. 2 shows that the condition (1) of Theorem 2 is satisfied. For the second case, by using Simpson’s method, we have ⟨R1s⟩200≈0.7812<1 (Fig. 3). Moreover, from Fig. 4, it can be deduced that the condition (2) of Theorem 2 is satisfied. The numerical outcomes are shown in Figs. 3 and 5 and exhibit in both cases that the disease goes to extinction. For the third case, by calculation, we have R2s=1.2192>1, thereby, the condition (4) of Theorem 3 holds (Fig. 6). Consequently, lim inft→+∞⟨I⟩t≥0.0029, which is illustrated by Fig. 7. Figures 6 and 8 show the obtained solution. For the fourth case, by using Simpson’s method, we have R≈1.4697>1. Similarly, for the fifth case, we have R≈1.3340>1. Consequently, the condition (14) of Theorem 4 is verified. Figures 9 and 10 illustrate the deterministic and stochastic periodicity of the obtained solution (Fig. 11). The assigned values to the parameters in each case are as follows Case 1 : ∀t∈[0,200],Λ(t)=0.3+0.02sin(t),βS(t)=0.2+0.06sin(t),βV1(t)=0.1+0.02sin(t),βV2(t)=0.1+0.05sin(t),γ(t)=0.3+0.001sin(t),γV3(t)=0.2+0.001sin(t),κ1(t)=0.1+0.02sin(t),κ2(t)=0.1+0.02sin(t),κ3(t)=0.2+0.02sin(t),μ(t)=0.3+0.02sin(t). Case 2 : ∀t∈[0,200]σ1(t)=0.3+0.1sin(t),σ2(t)=0.2+0.1sin(t),σ3(t)=0.1+0.05sin(t),Λ(t)=0.3+0.02sin(t),βS(t)=0.2+0.1sin(t),βV1(t)=0.2+0.05sin(t),βV2(t)=0.3+0.02sin(t),γ(t)=0.3+0.001sin(t),γV3(t)=0.2+0.001sin(t),κ1(t)=0.3+0.02sin(t),κ2(t)=0.2+0.02sin(t),κ3(t)=0.3+0.02sin(t),μ(t)=0.3+0.02sin(t). Case 3 : ∀t∈[0,200]σ1(t)=0.1+0.01sin(t),σ2(t)=0.05+0.01sin(t),σ3(t)=0.04+0.01sin(t),Λ(t)=0.1+0.02sin(t),βS(t)=0.6+0.3sin(t),βV1(t)=0.7+0.2sin(t),βV2(t)=0.8+0.4sin(t),γ(t)=0.01+0.001sin(t),γV3(t)=0.2+0.001sin(t),κ1(t)=0.03+0.01sin(t),κ2(t)=0.02+0.01sin(t),κ3(t)=0.05+0.02sin(t),μ(t)=0.1+0.02sin(t). Case 4 : ∀t∈[0,200]σ1(t)=0,σ2(t)=0,σ3(t)=0,Λ(t)=0.1+0.02sin(t),βS(t)=0.3+0.2sin(t),βV1(t)=0.4+0.2sin(t),βV2(t)=0.3+0.2sin(t),γ(t)=0.01+0.001sin(t),γV3(t)=0.01+0.001sin(t),κ1(t)=0.3+0.02sin(t),κ2(t)=0.2+0.02sin(t),κ3(t)=0.3+0.02sin(t),μ(t)=0.1+0.02sin(t). Case 5 : ∀t∈[0,200]σ1(t)=0.06+0.02sin(t),σ2(t)=0.03+0.02sin(t),σ3(t)=0.05+0.02sin(t),Λ(t)=0.1+0.02sin(t),βS(t)=0.3+0.2sin(t),βV1(t)=0.4+0.2sin(t),βV2(t)=0.3+0.2sin(t),γ(t)=0.01+0.001sin(t),γV3(t)=0.01+0.001sin(t),κ1(t)=0.3+0.02sin(t),κ2(t)=0.2+0.02sin(t),κ3(t)=0.3+0.02sin(t),μ(t)=0.1+0.02sin(t). Conclusions and future work The appearance of new emerging diseases requires the enhancement of existing epidemic models, in order to have a more pertinent interpretation of reality [39–43]. In this context, inspired by the characteristics of new emerging diseases such as COVID-19, in this paper, we have conducted a dynamical study of a new-proposed stochastic SVIR model, in the aim of studying the effect of the multiple stages of vaccination, required to gain immunity, along with the environmental noise on the dynamics of the studied population. Our results are briefly outlined as follows.For a large values of the Gaussian noise intensities, the infected population goes to extinction if lim supt→+∞ϱ(t)<0. For sufficiently small values of the Gaussian noise intensities, a sufficient condition guaranteeing that the infected population goes to extinction is ⟨R1s⟩T<1, and ∀t∈(0,T), μ_Λ¯βS(t)>σ12(t), μ_Λ¯βV1(t)>σ22(t), and μ_Λ¯βV2(t)>σ32(t). Under the condition R2s>1, the infected population becomes persistent in the mean. For diseases with seasonal patterns, under the condition R>1, the susceptible, infected, vaccinated and recovered subpopulations become persistent. It is worth mentioning that while our primal focus in this work resided in the dynamical analysis, this paper brings about other interesting questions that need to be investigated. Case in point, we can think of dealing with the identification problem for COVID-19 in Morocco, due to the availability of the data [44], which will permit us to identify the stochastic thresholds characterizing the disease extinction and persistence and then test the effectiveness of the vaccination strategy adopted by the authorities. On the other hand, the model (1) can be further generalized. For instance, taking into account that a certain amount of time is necessary between each stage of vaccination as well as the mean time in which the effectiveness of each stage wears off, we can add delay variables to the model and analyze the changes induced in the dynamics. Finally, by taking into account that the population may suffer from sudden environmental shocks. Precisely, ones exhibited by socio-cultural changes such as anti-vaccination movements, adding Lévy jumps to the model can increase its pertinence. All these questions will be the subjects of future work. Acknowledgements The authors would like to express their gratitude to the editor and anonymous reviewers for their careful reading and valuable suggestions. Code availability The Matlab code used for the numerical simulations is available from the corresponding author upon request. Declarations Conflicts of interest The authors declare that they have no conflicts of interest. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. ==== Refs References 1. Kermack WO McKendrick AG A contribution to the mathematical theory of epidemics Proc. R. Soc. Lond. Ser. A Contain Papers Math. Phys. Charact. 1927 115 700 721 10.1098/rspa.1927.0118 2. Liu X Takeuchi Y Iwami S SVIR epidemic models with vaccination strategies J. Theor. Biol. 2008 253 1 1 11 10.1016/j.jtbi.2007.10.014 18023819 3. Kumar S Ahmadian A Kumar R Kumar D Singh J Baleanu D Salimi M An efficient numerical method for fractional sir epidemic model of infectious disease by using Bernstein wavelets Mathematics 2020 8 4 558 10.3390/math8040558 4. Khan ZA Lamrani Alaoui A Zeb A Tilioua M Djilali S Global dynamics of a SEI epidemic model with immigration and generalized nonlinear incidence functional Results Phys. 2021 27 104477 10.1016/j.rinp.2021.104477 5. Kumar S Kumar A Samet B Gómez-Aguilar J Osman M A chaos study of tumor and effector cells in fractional tumor-immune model for cancer treatment Chaos Solitons Fractals 2020 141 110321 10.1016/j.chaos.2020.110321 6. Kumar S Kumar R Osman M Samet B A wavelet based numerical scheme for fractional order SEIR epidemic of measles by using Genocchi polynomials Numer. Methods Partial Differ. Equ. 2021 37 2 1250 1268 10.1002/num.22577 7. Mohammadi H Kumar S Rezapour S Etemad S A theoretical study of the Caputo-Fabrizio fractional modeling for hearing loss due to mumps virus with optimal control Chaos Solitons Fractals 2021 144 110668 10.1016/j.chaos.2021.110668 8. Ghanbari B Kumar S Kumar R A study of behaviour for immune and tumor cells in immunogenetic tumour model with non-singular fractional derivative Chaos Solitons Fractals 2020 133 109619 10.1016/j.chaos.2020.109619 9. Mehdaoui M Alaoui AL Tilioua M Optimal control for a multi-group reaction-diffusion SIR model with heterogeneous incidence rates Int. J. Dyn. Control 2022 10.1007/s40435-022-01030-3 10. Mezouaghi A Djilali S Bentout S Biroud K Bifurcation analysis of a diffusive predator-prey model with prey social behavior and predator harvesting Math. Methods Appl. Sci. 2022 45 2 718 731 10.1002/mma.7807 11. Djilali S Bentout S Pattern formations of a delayed diffusive predator-prey model with predator harvesting and prey social behavior Math. Methods Appl. Sci. 2021 44 11 9128 9142 10.1002/mma.7340 12. Djilali S Bentout S Spatiotemporal patterns in a diffusive predator-prey model with prey social behavior Acta Appl. Math. 2020 169 1 125 143 10.1007/s10440-019-00291-z 13. Kiouach, D., Sabbar, Y.: Ergodic stationary distribution of a stochastic hepatitis B epidemic model with interval-valued parameters and compensated poisson process. Comput. Math. Methods Med. 2020 (2020) 14. Kiouach D Sabbar Y Global dynamics analysis of a stochastic SIRS epidemic model with vertical transmission and different periods of immunity Int. J. Dyn. Syst. Differ. Equ. 2020 10 5 468 491 15. Kiouach D Sabbar Y Dynamic characterization of a stochastic SIR infectious disease model with dual perturbation Int. J. Biomath. 2021 14 04 2150016 10.1142/S1793524521500169 16. Liu Q Jiang D Hayat T Alsaedi A Ahmad B Dynamical behavior of a higher order stochastically perturbed SIRI epidemic model with relapse and media coverage Chaos Solit. Fractals 2020 139 110013 10.1016/j.chaos.2020.110013 17. Kiouach D Sabbar Y El Azami El-idrissi S New results on the asymptotic behavior of an SIS epidemiological model with quarantine strategy, stochastic transmission, and Lévy disturbance Math. Methods Appl. Sci. 2021 44 17 13468 13492 10.1002/mma.7638 18. Shangguan D Liu Z Wang L Tan R A stochastic epidemic model with infectivity in incubation period and homestead-isolation on the susceptible J. Appl. Math. Comput. 2021 67 1 785 805 10.1007/s12190-021-01504-1 33613144 19. Jiang D Yu J Ji C Shi N Asymptotic behavior of global positive solution to a stochastic SIR model Math. Comput. Model. 2011 54 1–2 221 232 10.1016/j.mcm.2011.02.004 20. Sabbar Y Khan A Din A Kiouach D Rajasekar S Determining the global threshold of an epidemic model with general interference function and high-order perturbation AIMS Math. 2022 7 11 19865 19890 10.3934/math.20221088 21. Keeling MJ Rohani P Grenfell BT Seasonally forced disease dynamics explored as switching between attractors Phys. D Nonlinear Phenom. 2001 148 3–4 317 335 10.1016/S0167-2789(00)00187-1 22. Weber A Weber M Milligan P Modeling epidemics caused by respiratory syncytial virus (RSV) Math. Biosci. 2001 172 2 95 113 10.1016/S0025-5564(01)00066-9 11520501 23. Greenhalgh D Moneim IA SIRS epidemic model and simulations using different types of seasonal contact rate Syst. Anal. Model. Simul. 2003 43 5 573 600 10.1080/023929021000008813 24. Qi H Leng X Meng X Zhang T Periodic solution and ergodic stationary distribution of SEIS dynamical systems with active and latent patients Qual. Theory Dyn. Syst. 2019 18 2 347 369 10.1007/s12346-018-0289-9 25. Shangguan, D., Liu, Z., Wang, L., Tan, R.: Periodicity and stationary distribution of two novel stochastic epidemic models with infectivity in the latent period and household quarantine. J. Appl. Math. Comput. 1–20 (2021) 26. Liu Q Jiang D Hayat T Ahmad B Periodic solution and stationary distribution of stochastic SIR epidemic models with higher order perturbation Phys. A Stat. Mech. Appl. 2017 482 209 217 10.1016/j.physa.2017.04.056 27. Lin Y Jiang D Liu T Nontrivial periodic solution of a stochastic epidemic model with seasonal variation Appl. Math. Lett. 2015 45 103 107 10.1016/j.aml.2015.01.021 28. El Fatini M Pettersson R Sekkak I Taki R A stochastic analysis for a triple delayed SIQR epidemic model with vaccination and elimination strategies J. Appl. Math. Comput. 2020 64 1 781 805 10.1007/s12190-020-01380-1 32837464 29. Zhang X Liu M Dynamical analysis of a stochastic delayed sir epidemic model with vertical transmission and vaccination Adv. Contin. Discrete Models 2022 2022 1 1 18 10.1186/s13662-022-03707-7 30. Burki TK Omicron variant and booster COVID-19 vaccines Lancet Respir. Med. 2022 10 2 17 10.1016/S2213-2600(21)00559-2 31. Mattiuzzi, C., Lippi, G.: Primary COVID-19 vaccine cycle and booster doses efficacy: analysis of Italian nationwide vaccination campaign. Eur. J. Public Health (2022) 32. Mao X Stochastic Differential Equations and Applications 2007 New York Elsevier 33. Khasminskii R Stochastic Stability of Differential Equations 2011 Berlin Springer 34. Van den Driessche P Watmough J Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmission Math. Biosci. 2002 180 1–2 29 48 10.1016/S0025-5564(02)00108-6 12387915 35. MATLAB: Version 9.4.0 (R2018a). The MathWorks Inc., Natick, Massachusetts (2010) 36. Higham DJ An algorithmic introduction to numerical simulation of stochastic differential equations SIAM Rev. 2001 43 3 525 546 10.1137/S0036144500378302 37. Bao K Zhang Q Stationary distribution and extinction of a stochastic sirs epidemic model with information intervention Adv. Differ. Equ. 2017 2017 1 1 19 10.1186/s13662-017-1406-9 38. Rao, F.: Dynamics analysis of a stochastic sir epidemic model. In: Abstract and Applied Analysis, vol. 2014 (2014). Hindawi 39. Soufiane B Touaoula TM Global analysis of an infection age model with a class of nonlinear incidence rates J. Math. Anal. Appl. 2016 434 2 1211 1239 10.1016/j.jmaa.2015.09.066 40. Bentout S Tridane A Djilali S Touaoula TM Age-structured modeling of Covid-19 epidemic in the USA, UAE and Algeria Alex. Eng. J. 2021 60 1 401 411 10.1016/j.aej.2020.08.053 41. Bentout S Chekroun A Kuniya T Parameter estimation and prediction for coronavirus disease outbreak 2019 (Covid-19) in Algeria AIMS Public Health 2020 7 2 306 10.3934/publichealth.2020026 32617358 42. Soufiane B Touaoula TM Global analysis of an infection age model with a class of nonlinear incidence rates J. Math. Anal. Appl. 2016 434 2 1211 1239 10.1016/j.jmaa.2015.09.066 43. Bentout S Chen Y Djilali S Global dynamics of an SEIR model with two age structures and a nonlinear incidence Acta Appl. Math. 2021 171 1 1 27 10.1007/s10440-020-00369-z 44. The Moroccan Ministry of Public Health: COVID-19 Platform. http://www.covidmaroc.ma/pages/Accueilfr.aspx (2022)

0

PMC9749651

NO-CC CODE

2022-12-15 23:23:21

no

J Appl Math Comput. 2022 Dec 14;:1-30

utf-8

J Appl Math Comput

10.1007/s12190-022-01828-6

oa_other

==== Front Exp Neurol Exp Neurol Experimental Neurology 0014-4886 1090-2430 Published by Elsevier Inc. S0014-4886(22)00181-9 10.1016/j.expneurol.2022.114156 114156 Editorial Gene modification after spinal cord injury: Mechanisms and therapeutics Smith George M. a⁎ Steward Oswald b Bradbury Elizabeth J. c a Department of Neural Sciences, Lewis Katz School of Medicine, Temple University, 3500 North Broad Street, Philadelphia, PA 19140, United States of America b Reeve-Irvine Research Center, 837 Health Sciences Dr., University of California Irvine School of Medicine, Irvine, CA 92697, USA c King's College London, Regeneration Group, The Wolfson Centre for Age-Related Diseases, Guy's Campus, London Bridge, London SE1 1UL, UK ⁎ Corresponding author. 28 6 2022 10 2022 28 6 2022 356 114156114156 © 2022 Published by Elsevier Inc. 2022 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcThe concept for this special issue on “Viral Vectors for Gene Modification after Spinal Cord Injury” was initially proposed in 2019 at the last in person meeting of the Experimental Neurology Editorial Board before the COVID-19 pandemic, and was approved by the Editor-in-Chief with Elizabeth Bradbury, George Smith and Oswald Steward as Co-Guest Editors. The guest editors identified and contacted potential contributing authors, and the rate of acceptance of our invitations was unusually high, confirming that this was an opportune time for a special issue in this area. We are delighted in the way that the Special Issue has come together, and grateful for the enthusiastic participation of the contributors. This Special Issue was designed to provide an overview of genetic engineering approaches to design and test therapeutic candidates for spinal cord injury (SCI) as well as to identify mechanisms and post-injury changes in spinal circuits after injury and/or recovery. Some articles in this issue review the literature and discuss the advantages and potential disadvantages of some of these approaches. The first article by Van Steenbergen and Bareyre provides a topical review on the use of chemogenetic approaches to functionally and anatomically dissect neural circuits within the spinal cord, while providing practical insights into advantages and general concerns using these designer compounds. The article by Sidney-Smith et al. discusses the use of peripherally delivered gene therapies as potential treatments for spinal cord disorders and viral modifications to increase the utility and function of this approach. In efforts to improve the safety of gene therapy approaches, De Winter et al. describe recent advances in the use of inducible viral vectors, which enable controlled expression of the therapeutic transgene. The design of a gene switch that evades immune cell detection ensures protection of the inducible gene therapy. An overview of anterograde and retrograde transsynaptic viral tracer approaches to map spinal cord circuits connecting onto peripheral targets is reviewed in the article by Fortino et al. They further discuss the utility of this approach to map the integration of transplanted neuronal precursor cells with host circuitry within the injured spinal cord. Viral systems have multiple advantages and several limitations in their ability to transfer genetic materials into neuronal and non-neuronal cells within the CNS. The paper by Islam and Tom highlights the advantages and use of various vector types, serotypes and promotors in targeting specific cell populations and the capacity and limitation of these viruses for therapeutic transgene expression. Sherrington referred to motor neurons as the final common pathway, however, these motor neurons receive extensive innervation from the supraspinal regions that change after spinal cord injury or with functional recovery. Retrograde transportable adeno-associated viruses (AAVretro) are useful in identifying presynaptic supraspinal neuronal targets innervating specific regions of the spinal cord. Blackmore et al., discuss the use of AAVretro viral approaches to map and evaluate changes in the supraspinal connectome and the importance of these alterations in both the functional deficits after injury and recovery mediated by restorative therapies. This article also highlights the important perspective and priorities of those living with SCI and the importance of developing therapies to restore multiple functions that are lost, some of which are apparent only to those who are living with SCI. Likewise, the article by Metcalfe et al. examines differences in the supraspinal connectome dependent on regional locations within the spinal cord and the use of this method to simultaneously target multiple spinal pathways for gene therapy. Schrank and Satkunendrarajah provide a brief review of approaches using different types of retrogradely-transported vectors and clarify some seeming discrepancies between studies. This paper also highlights other strategies for multiple supraspinal pathway targeting, and highlight their potential uses in mapping, modulating and treating these neural networks. To better understand the functional roles of specific neural circuits after SCI, several papers discuss the use of intersectional genetics involving selective transcriptional regulation mediated by either two viral systems or viral/transgenic mice models. Nicola et al., discuss the supraspinal and spinal circuitry involved in skilled forelimb patterning and the use of combining Cre-dependent tools with optogenetic or pharmacogenetic perturbation in transgenic mice to functionally dissect these circuits. Using similar 2-viral systems to selectively induce Tetanus toxin to silence specific neuronal populations, the article by Isa examines the circuitry involved in functional recovery of non-human primate reaching and grasping maneuvers mediated by C3/C4 propriospinal and supraspinal circuits. In general, plasticity associated with propriospinal neurons is thought to enable bypass relays around the lesion contributing to functional recovery. The article by Deng et al. further discusses the roles of propriospinal neurons in recovery and viral methods to induce plasticity and the functional mapping of the circuits to identify their contribution to recovery. Although several articles in this collection highlight the therapeutic potential of viral vector-based candidates to enhance axonal regeneration, sprouting, and recovery of function, the paper by Campion et al. provides a cautionary tale of how overexpression of Akt leads to extensive regeneration and sprouting, but also the development of epilepsy due to hemimegalencephaly. This paper highlights the need for vigilance as we deploy powerful new technologies, and also hints that unanticipated pathophysiologies may limit recovery even when the long-sought goal of axon regeneration is achieved. This discovery has the potential of changing our thinking from “Our interventions aren't powerful enough” to “Our interventions may be too powerful because competing deleterious processes reduce recovery that might otherwise be achieved”. It is clear from the scope of the articles in this Special Issue that tools and technologies to genetically modify cells and circuits of the spinal cord are advancing at a rapid pace. Gene modification after SCI is a burgeoning field, and if harnessed correctly has the potential to deliver new therapeutics as well as provide mechanistic insight into circuits and functions involved in injury and repair.

35792382

PMC9749756

NO-CC CODE

2022-12-15 23:23:21

no

Exp Neurol. 2022 Oct 28; 356:114156

utf-8

Exp Neurol

10.1016/j.expneurol.2022.114156

oa_other

==== Front J Psychosom Res J Psychosom Res Journal of Psychosomatic Research 0022-3999 1879-1360 Elsevier Inc. S0022-3999(21)00229-4 10.1016/j.jpsychores.2021.110584 110584 Article Perception of stress and cognitive efficiency in older adults with mild and moderate dementia during the COVID-19-related lockdown Paolini Simone a Devita Maria a⁎ Epifania Ottavia M. b Anselmi Pasquale b Sergi Giuseppe c Mapelli Daniela a Coin Alessandra c a Department of General Psychology (DPG), University of Padua, Italy b Department of Philosophy, Sociology, Education and Applied Psychology – FISPPA, University of Padua, Italy c Geriatrics Division, Department of Medicine – DIMED, University of Padua, Italy ⁎ Corresponding author. 27 7 2021 10 2021 27 7 2021 149 110584110584 10 2 2021 20 7 2021 22 7 2021 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Aim Many studies have been carried out with the aim of understanding the manifold effects of the novel coronavirus (Covid-19) on individuals' clinical and psychological states. This paper deals with perceived stress (PS) and cognitive efficiency (CE) in older adults with dementia during the first wave of the pandemic. The study also investigated the potential effects of PS and CE on participants' cognitive functioning. The modulating effect of cognitive reserve (CR) on these variables was also considered, given its well-known role in the onset and evolution of neurodegenerative diseases. Method Thirty-eight older adults with mild/moderate dementia (mean age: 81.47 ± 5.05; mean MMSE pre-lockdown: 24.03 ± 3.14) were recruited for this study from March to May [4]. Two questionnaires, the Perceived Stress Scale (PSS) and the FLEI Mental Ability Questionnaire (FLEI), were administered to all participants by telephone every two weeks during lockdown (T1: early April, T2: mid-April, T3: early May). After lockdown, their neuropsychological and psychological profiles were assessed. Linear mixed-effects models were used to investigate changes over time. Results The level of PS increased at both Time 2 and Time 3 (f2 = 0.10). Cognitive functioning worsened during lockdown, resulting in lower scores at the post-lockdown evaluation (f2s = 0.09 and 0.06 for MMSE and ENB-2, respectively). The decrease in these scores was not associated with either PS or CE. Although the size of these effects was rather small, their clinical relevance is not negligible. Conclusion Individuals with dementia seem to have experienced stress (S) during the first-wave of lockdown related to Covid-19. Cognition worsened during the pandemic, in accordance with the neurodegenerative nature of the disease, but it was unrelated to PS and CE. Keywords Covid-19 Lockdown Older adults Dementia Perceived stress Cognitive efficiency ==== Body pmc1 Introduction The spread of the novel coronavirus SARS-CoV-2 (Covid-19) throughout Italy from February [4] required, as in the rest of the world, the implementation of timely strategies and interventions in order to limit infection and guarantee assistance to as many people as possible. Although other pandemics have occurred throughout history, none has had an impact like that of Covid-19, an event of exceptional magnitude. A meta-analysis by Sepúlveda-Loyola and colleagues [4] drew attention to the consequences of the pandemics for mental health that have occurred so far around the world. The results show an increased risk of developing depression [8], emotional disturbances [30], stress [11], and deflection of mood, irritability or insomnia [19]. Alarmingly, Yip et al. [29] also showed that these disorders are associated with higher suicide rates during pandemics, particularly in older adult populations. These data were corroborated by further studies investigating the psychological effects of lockdown during the SARS, H1N1, Ebola, MERS and Equine Influenza epidemics ([3, 26]). Several studies have shown that people obliged to quarantine generally have a high prevalence of post-traumatic and depressive symptoms, stress and anxiety [11,[14], [15], [16], [17],19,25]. Increasing amounts of data on the psychological and social consequences of the Covid-19 pandemic in the general population and on the interventions devised to deal with its effects [24] are becoming available. Particular consideration should be given to older adults, who represent the section of the population with the highest rate of mortality linked to this virus [28]. As also confirmed by Devita and colleagues [10], older individuals constitute an extremely frail population at risk of contagion from Covid-19. Among these individuals, particular attention needs to be focussed on those with neurocognitive disorders, described as “the frailest among the frailest”. Living with the Covid-19 pandemic can, in fact, be extremely difficult for them, as they may not remember to keep an adequately safe distance from other people or may forget to observe the strict health and hygiene guidelines and regulations. Added to these problems, and aggravating the overall picture, are altered behaviours, such as wandering or disinhibition, which are very frequent in individuals suffering from dementia. Due to their pathology, these older individuals may not have acquired a full awareness of the health emergency around them, and because of this they may not have experienced a high level of perceived stress (PS). As a result, they may also have been unaffected by the cognitive burden produced by the pandemic and may not have perceived important changes in their cognitive efficiency (CE) during lockdown. The aims of this study are twofold. The first is to evaluate the above-mentioned issues by investigating the trend and evolution over time of PS and CE in older individuals with mild/moderate dementia during the first-wave of lockdown. The second is to investigate the potential effects of PS and CE on participants' cognitive functioning and whether these variables are modulated by cognitive reserve (CR), which is known to play an important role in the onset and evolution of neurodegenerative diseases [27]. 2 Materials and methods Thirty-eight participants were recruited from patients routinely attending the Geriatric Clinic (Department of Medicine - DIMED, University of Padua) and for whom a complete neuropsychological assessment, performed immediately prior to (maximum of 2 months) the spread of Covid-19, was available. The characteristics of participants and their caregivers are reported in Table 1 . At the end of lockdown, 29 individuals underwent a follow-up full neuropsychological evaluation.Table 1 Descriptive socio-demographic and clinical characteristics of the sample. Table 1Variables Participants Demographic and cognitive characteristics Total sample (n = 38) MMSE ≥ 24 (n = 21) MMSE < 24 (n = 17) Age 81.47 ± 5.05 82.00 ± 4.52 80.82 ± 5.71 Gender, Female 20 (52.6) 9 (42.9) 11 (64.7) Education 10.37 ± 5.20 10.00 ± 5.45 10.82 ± 5.01 CRIq 106.89 ± 24.13 110.52 ± 24.51 102.41 ± 23.60 MMSE pre- lockdown 24.03 ± 3.14 26.30 ± 1.35 21.22 ± 2.33 Semi-structured Interview Before Covid-19 related lockdown During Covid-19 related lockdown No one Spouses Adult Children Other Caregiver Caregiver + Children Others No one Spouses Adult Children Other Caregiver Caregiver + Children Others Who did/do you live with? 9 (23.7) 24 (63.2) 1 (2.6) 3 (7.9) 1 (2.6) 8 (21.1) 25 (65.8) 1 (2.6) 4 (10.5) Who helped/helps you with shopping? 5 (13.2) 21 (55.3) 8 (21.1) 2 (5.3) 2 (5.3) 1 (2.6) 9 (23.7) 23 (60.5) 2 (5.3) 3 (7.9) Who oversaw/oversees the purchase of medicines? 5 (13.2) 17 (44.7) 11 (28.9) 3 (7.9) 2 (5.3) 2 (5.3) 12 (31.6) 19 (50) 3 (7.9) 2 (5.3) Never Max once a week Max twice a week More than twice a week Never Max once a week Max twice a week More than twice a week How often did/do you exercise (e.g., walking, cycling, etc.)? 16 (42.1) 2 (5.3) 20 (52.6) 33 (86.8) 1 (2.6) 4 (10.5) How often did/do you have social contacts with friends and acquaintances before lockdown? 12 (31.6) 6 (15.8) 5 (13.2) 15 (39.5) 28 (73.7) 1 (2.6) 1 (2.6) 8 (21.1) Continuous variables are showed as Mean ± Standard Deviation, while frequencies as numerosity (%). CRIq: Cognitive Reserve Index questionnaire; MMSE: Mini-Mental State Examination; M = mean of the scores; SD = Standard Deviation; n = sample size. Other caregiver: nephews, volunteers, social and health workers, etc. Others: home shopping, friends. The participants' diagnoses, received in the period 2018–2020, ranged from mild to moderate Alzheimer's Dementia (AD) according to the NINCDS-ADRDA standard clinical diagnostic criteria (mean MMSE pre-lockdown: 24.03 ± 3.14). Participants had to be able to independently provide informed consent to be eligible for the study. All individuals with pre-lockdown MMSE scores <18 were excluded from the study, as were those unable to perform the tests due to advanced cognitive impairment, severe sensory deficits or pathologies of purely psychiatric interest. Participants were contacted by telephone in April–May 2020 (the Italian lockdown started 9th March 2020 and ended 18th May 2020) and were invited to undergo a telephone interview, during which two questionnaires designed to investigate PS and CE were administered. The level of care provided by family members and participants' living arrangements in the period before and during lockdown were also established through two ad-hoc semi-structured interviews. The presence of a caregiver or appointed legal guardian (e.g., a support administrator) was always required during telephone interviews. 2.1 Study design and data collection As illustrated in Fig. 1 , each participant underwent a total of three interviews during the Covid-19 lockdown period at intervals of two weeks. At a maximum of two months after the end of lockdown, participants were called back for a post-lockdown follow-up. Participants were first administered the same questionnaires as in the previous telephone interviews, and secondly were cognitively assessed using the MMSE and the ENB-2 (Esame Neuropsicologico Breve-2 - Brief Neuropsychological Examination-2), which they had also undertaken before the pandemic.Fig. 1 Study design. Fig. 1 MMSE: Mini-Mental State Examination; ENB-2: Esame Neuropsicologico Breve-2 [Brief Neuropsychological Examination-2]; PSS: Perceived Stress Scale; FLEI: FLEI Mental Ability Questionnaire; T = assessment time. 2.2 Measures Perceived stress (PS; [7]). The Perceived Stress Scale (PSS) is the most frequently used psychological measure to assess perceptions of S [6]. The degree to which the situations in a person's life are rated as stressful are evaluated by items (n = 10) constructed to capture the level at which respondents perceive their lives as unpredictable, uncontrollable, or overloaded. The scale also contains a series of direct questions about current levels of perceived S. The PSS was designed to be used in samples drawn from the general population with an educational level at least equal to lower middle school. The items and the response alternatives are easy to understand: for each item, respondents are asked to indicate how often they felt a certain way in the last month (“0 = Never”, “4 = Very often”). The PSS scores are obtained by reverse-scoring the responses to the four positively formulated items (items 4, 5, 7 and 8), then adding together the scores for each and every item. A short 4-item scale can be obtained using questions 2, 4, 5 and 10 of the 10 items in the PSS scale. For our study, we adapted the questionnaire to the context by reducing the reference period from “the last month” to “the last 2 weeks”. Cohen and Williamson [7] have shown that PSS scores of healthy individuals correlate with S measures, self-administered measures relating to health and health services, measures of healthy behaviours, smoking habits and help-seeking behaviour. Cognitive efficiency (Vienna Test System; www.Schuhfried.at). The FLEI Mental Ability Questionnaire (FLEI) evaluates the level of subjectively perceived cognitive efficiency. Patients with mental disorders may exhibit considerable discrepancy between their subjectively experienced neuropsychological disorder and their objective test results. This test provides a subjective assessment of the respondent's performance in the areas of memory, attention and executive functions. Respondents state how often over the past six months they have experienced various daily problems on a 5-point Likert-type scale (from 0 = ‘Never’ to 5 = ‘Very often’). There is also a control scale (to check the validity of the answers) that indicates whether the respondent has a response bias or certain impairments (for example, neglect or problems with understanding), although it was not used in the analysis presented here. This choice was made because of some important weaknesses, underlined by the FLEI authors themselves, of the control scale (www.schuhfried.at). For our study, we adapted the questionnaire to the context and the participants by shortening the reference period from “the last 6 months” to “the last 2 weeks” and removing one item deemed not relevant for the intended purposes (i.e., “In important interviews, I have it in mind to address certain points. In the end, I realise I have forgotten to address some points.”) thus reducing the scale from 32 to 31 items. Mini-Mental State Examination (MMSE; [13,20]). The MMSE is a widely-used test to assess the presence of cognitive impairment in older people. It takes only 10–15 min to administer, although it is not timed, and provides a reliable measure of cognitive impairment and the progression of dementia. The maximum total score is 30 with a cut-off <24 indicating impairment. Scores are adjusted for age and education. Brief Neuropsychological Examination - 2 (ENB-2; [21]). The ENB-2 is a protocol for neuropsychological assessment. It comprises three parts: (i) neuropsychological history, (ii) neuropsychological interview with the patient and family members, (iii) administration of cognitive tests. It consists of sixteen items for evaluating lexical access and selection, verbal understanding, verbal abstraction, selective attention, divided attention, sustained and alternating attention, executive control functions, short- and long-term memory, working memory, motor skills training, visual recognition and logical reasoning. The ENB-2 allows the patient's performance to be assessed through both quantitative and qualitative analyses. The scores for individual items are adjusted for age and education and the cut-offs differ according to the respondent's decade of life and number of years of schooling (and extracurricular courses). A cut-off ≤8 is considered a “low” level of education, while >8 is considered “high”. The entire protocol takes approximately one hour to administer. Some tests have a time limit, others require the researcher to time the patient while s/he is executing the task, while other tests have no time constraints nor should they be timed. Cognitive Reserve Index questionnaire (CRIq; [23]). The CRIq is used to assess the level of CR. It consists of 20 items that evaluate educational level (i.e., total numbers of years in education and on other training courses, CRI-Education), working activity (i.e., all the occupations held from the age of 18, CRI-Working activity) and leisure time (i.e., cognitively stimulating leisure activities, CRI-Leisure time). The CRI-Total score is calculated as the average of the three sub-scores, standardised and transposed to a scale with a mean of 100 and a standard deviation of 15. In order to reduce possible and involuntary errors (e.g. confabulations or lack of memory due to the clinical characteristics of the sample itself), the CRIq was administered to caregivers, as allowed by the questionnaire instructions [23]. Informed consent was given by all participants when collecting their clinical data. The protocol was submitted to the local ethic committee (protocol N. 15,228). 2.3 Statistical analyses A linear mixed-effects model (LMM) approach was used to investigate the changes over time in the variables of interest, namely the level of perceived S (PSS), the level of perceived CE (FLEI), and cognitive functioning as indicated by the MMSE and ENB-2 scores. LMMs are used to analyse the inter-dependencies among the observations in a repeated-measures study, such as the present one, which increases the reliability of the results and reduces the probability of committing a type I error. Moreover, LMMs can easily handle missing data, a problem with repeated-measure studies, without discarding the entire set of observations in which the missing data are observed. In a typical repeated-measure analysis of variance (ANOVA), missing data are dealt with by listwise deletion (i.e., respondents with incomplete observations are discarded entirely from the analysis), which comes at a very high price, especially in the case of small sample sizes, as in the present study. LMMs, on the other hand, can retain all the observations with a valid value, even at only one time point. To estimate the mean on which comparisons are based, LMMs use the data from respondents with complete observations at all time points and from respondents whose data are complete only at a specific time point. This avoids completely eliminating a respondent with missing observations at only one of the time points. Four models were specified to investigate of the effects of time, the level of CR, and their interaction on the level of perceived S (PSS). Four models were also created to investigate the effects of the same variables on the level of perceived CE (FLEI). In all models, respondents were specified as random intercepts to account for the within-respondents / between-assessments variability over time. Given that there was only one observation per respondent at each assessment time (i.e., the PSS or FLEI score), it was not possible to specify the random slopes of the subjects at each time point, a specification that would have provided information on the adjustment of each individual throughout time. The first model (Model 1) included only the fixed intercept and the random intercepts of the respondents. In the second model, assessment time was added as a fixed effect (Model 2). The CR score was added in the third model as an additive effect, hence only the main effect of CR on the PSS or FLEI score was investigated (Model 3). Finally, in the fourth model, the effect of the interaction between assessment time and CR was added (Model 4). A similar approach was taken to investigate the pre- and post-lockdown changes in cognitive functioning (ENB-2 and MMSE). The effects of the level of perceived S and the level of perceived CE on cognitive functioning were also investigated. The levels of perceived S and CE were considered at each of the three assessment times in terms of their interaction with the pre- and post-lockdown assessment. This resulted in five models being specified for investigating the changes in cognitive functioning as indicated by MMSE scores, and five models for investigating the changes in cognitive functioning as indicated by ENB-2 scores. The first model (Model 1) included only the intercept, representing the expected MMSE or ENB-2 score. The second and third models also included the main effects of time (Model 2), and of perceived S and perceived CE (Model 3). The effects of the interaction between Time T2 and the levels of perceived S and CE (Model 4) and the effects of the interaction between Time T3 and the levels of perceived S and CE (Model 5) were added to the fourth and fifth models. To test the significance of each of the variables of interest (i.e., the time of assessment, the level of CR as indicated by the CRIq score, and their interaction) on the perceived level of S (PSS score) and on CE (FLEI score), a constrained model and a full model were compared and a Kenward-Roger's adjusted F-test [18] was performed, as illustrated in Faraway [12]. If the F statistics resulting from the comparison between the constrained model (i.e., without the effect of interest) and the full model (i.e., including the effect of interest) are significant, the addition of the new variable, and hence its contribution in explaining the dependent variable, can be considered significant. The same approach was taken to test the significance of the effects of the time point and the perceived levels of S and CE on cognitive functioning (i.e., the MMSE and ENB-2 scores). The marginal R 2 of the models (i.e., the proportion of variance explained by the fixed effects) were computed according to Nakagawa et al. [22]. The f 2 statistic was computed as an effect size measure for the fixed effects [1]. This statistic expresses the proportion of variance explained by the model including the fixed effect(s) of interest (R target 2) and the proportion of variance explained by the model without the fixed effect(s) of interest (R 2). Values of f 2 close to 0.02, 0.15, and 0.35 denote small, medium, and large effect sizes, respectively [6]. 3 Results The main descriptive characteristics of the sample are showed below (see Table 1). As emerged from the semi-structured interview, some differences can be observed before and after the Covid-19 related lockdown in some specific domain of life and daily routine. In particular, before the lockdown, the majority of participants were mainly helped by their spouses for shopping and purchase of medicines; during lockdown, instead, a more central role was given to the adult children that supported our participants in these tasks more than other caregivers, spouses included. Furthermore, and more predictably because of the social restrictions associated with the lockdown, physical activity and social contacts significantly decreased during the lockdown. Thirty-eight participants (F = 52.63%, age = 81.47 ± 5.05 years, range = 70–91 years) completed PSS and FLEI data at all three assessments carried out during lockdown. Analyses of the trends in these variables were therefore performed on the entire data set from these 38 respondents. The means and standard deviations of the PSS, FLEI, MMSE and ENB-2 scores at each assessment time are reported in Table 2 .Table 2 Means and standard deviations of the PSS, FLEI, MMSE and ENB-2 scores. Table 2Time PSS FLEI T1 11.29 ± 7.18 77.13 ± 17.55 T2 13.95 ± 8.57 75.47 ± 12.82 T3 13.79 ± 8.67 75.26 ± 14.20 MMSE ENB-2 Pre-lockdown 24.18 ± 3.47 54.08 ± 10.09 Post-lockdown 22.19 ± 3.73 47.70 ± 12.36 Notes: T1: first assessment during lockdown; T2: second assessment during lockdown; T3: third assessment during lockdown; PSS: Perceived Stress Scale; FLEI: FLEI Mental Ability Questionnaire; MMSE: Mini-Mental State Examination; ENB-2: Esame Neuropsicologico Breve-2 [Brief Neuropsychological Examination-2]. Nine respondents with invalid data at both the pre- and post-lockdown ENB-2 assessments had to be discarded from the analysis. All the remaining 29 respondents (F = 58.62%, age = 81.21 ± 5.30 years, range = 70–91 years) had at least one valid observation at either the pre-lockdown or the post-lockdown ENB-2 assessment, and at least one valid observation at either the pre- or the post-lockdown MMSE assessment. The results of the effects of time and CR on the levels of perceived S as assessed by the PSS are reported in Table 3 .Table 3 Changes in PSS scores throughout lockdown. Table 3 Model 1 Model 2 Model 3 Model 4 B (SE) B (SE) B (SE) B (SE) Intercept 13.01 (1.19) 11.29 (1.31) 11.29 (1.25) 11.29 (1.24) T2 2.66 (0.95) 2.66 (0.95) 2.66 (0.93) T3 2.50 (0.95) 2.50 (0.95) 2.50 (0.93) CRIq 0.10 (0.05) 0.06 (0.05) T2 × CRIq 0.08 (0.04) T3 × CRIq 0.05 (0.04) Observations (n) 114 114 114 114 Respondents σ 6.97 7.01 6.65 6.66 Log Likelihood −370.99 −366.37 −364.07 −362.01 AIC 747.99 742.75 740.13 740.01 BIC 756.20 756.43 756.55 761.90 Kenward-Roger's F-test F(2,74) = 4.78* F(1,36) = 4.65* F(2,72) =2.00 R2 0.00 0.02 0.11 0.12 f2 0.02 0.10 0.01 Notes: *: p < 0.05; CRIq: Cognitive Reserve Index questionnaire; T1: first assessment during lockdown; T2: second assessment during lockdown; T3: third assessment during lockdown; AIC: Akaike Information Criterion; BIC: Bayesian Information Criterion; R2: Proportion of variance explained by the fixed effects [22]: f2: Effect size measure expressing the proportion of variance explained by the fixed effect relative to the proportion of outcome variance unexplained [1]. The first assessment during lockdown (T1) is described by the intercept in each model. As such, T1 is the reference level of the time variable against which the other two levels (i.e., T2 and T3) are compared. The Kenward-Roger's F-test revealed the significance of the main effects of both the time of assessment and the level of CR on the level of PS, although their effect size was small (f 2 = 0.10). Specifically, the level of perceived S increased at both T2 and T3, but slightly more so at T2 than at T3. CR positively predicted the level of perceived S throughout time, such that higher levels of CR tended to be associated with higher levels of S, regardless of the time of assessment. Further corroborating this result, both the Kenward-Roger's test and the entropy indices (AIC and BIC, the smaller the better) favoured the model without the interaction term between CR and time of assessment. The influence of CR on the level of perceived S did not change as a function of time. The results of the effects of time and CR on the level of perceived CE as assessed by FLEI scores are reported in Table 4 .Table 4 Changes in FLEI scores throughout lockdown. Table 4 Model 1 Model 2 Model 3 Model 4 B (SE) B (SE) B (SE) B (SE) Intercept 75.96 (2.23) 77.13 (2.40) 77.13 (2.37) 77.13 (2.36) T2 −1.66 (1.55) −1.66 (1.55) −1.66 (1.50) T3 −1.87 (1.55) −1.87 (1.55) −1.87 (1.50) CRIq 0.10 (0.09) 0.18 (0.10) T2 × CRIq −0.11 (0.06) T3 × CRIq −0.13 (0.06) Observations (n) 114 114 114 114 Respondents σ 13.14 13.16 12.93 12.97 Log Likelihood −428.23 −427.37 −426.76 −424.23 AIC 862.47 864.74 865.53 864.47 BIC 870.68 878.42 881.94 886.36 Kenward-Roger's F-test F(2,74) = 0.85 F(1,36) = 1.17 F(2,72) = 2.48 R2 0.00 0.00 0.03 0.04 f2 0.00 0.03 0.01 Notes: CRIq: Cognitive Reserve Index questionnaire; T1: first assessment during lockdown; T2: second assessment during lockdown; T3: third assessment during lockdown; AIC: Akaike Information Criterion; BIC: Bayesian Information Criterion; R2: Proportion of variance explained by the fixed effects [22]: f2: Effect size measure expressing the proportion of variance explained by the fixed effect relative to the proportion of outcome variance unexplained [1]. Both the Kenward-Roger's test and the entropy indices revealed the absence of effects of time, CR and their interaction on the level of perceived CE. Nonetheless, the trends in the FLEI scores indicated a decrease in the level of perceived CE (i.e., the respondents' level of perceived CE was higher as time passed during lockdown). The results of the trends in cognitive functioning pre- and post-lockdown as indicated by MMSE scores are reported in Table 5 .Table 5 Pre- and post-lockdown changes in MMSE scores. Table 5 Model 1 Model 2 Model 3 Model 4 Model 5 B (SE) B (SE) B (SE) B (SE) B (SE) Intercept 23.19 (0.64) 24.18 (0.67) 24.18 (0.67) 24.18 (0.67) 24.18 (0.66) Post −1.99 (0.42) −1.99 (0.39) −1.99 (0.42) −1.99 (0.41) PSS1 (T1) 0.07 (0.10) FLEI (T1) −0.02 (0.05) Post × PSS (T1) 0.02 (0.06) Post × FLEI (T1) −0.07 (0.03) PSS (T2) 0.03 (0.09) FLEI (T2) −0.05 (0.05) Post × PSS (T2) −0.03 (0.06) Post × FLEI (T2) −0.04 (0.03) PSS (T3) 0.06 (0.09) FLEI (T3) −0.06 (0.05) Post × PSS (T3) −0.08 (0.06) Post × FLEI (T3) −0.03 (0.03) Observations (n) 58 58 58 58 58 Respondents σ 3.08 3.23 3.28 3.25 3.21 Log Likelihood −148.73 −140.17 −144.38 −146.34 −145.56 AIC 303.45 288.33 304.77 308.68 307.12 BIC 309.63 296.57 321.25 325.16 323.60 Kenward-Roger's F-test F(1,28) = 22.38*** F(4,37) = 2.17 F(4,37) = 1.00 F(4,37) = 1.41 R2 0.00 0.08 0.11 0.11 0.13 f2 0.09 0.03 0.00 0.02 Notes: PSS: Perceived S scale; FLEI: FLEI Mental Ability Questionnaire; Post: Post-lockdown assessment; T2: second assessment during lockdown; T3: third assessment during lockdown; AIC: Akaike Information Criterion; BIC: Bayesian Information Criterion; R2: Proportion of variance explained by the fixed effects [22]: f2: Effect size measure expressing the proportion of variance explained by the fixed effect relative to the proportion of outcome variance unexplained [1]. The intercept in all models represents the expected pre-lockdown average MMSE score. Only time of assessment had a significant effect on MMSE scores, although its effect size was small (f 2 = 0.09). Regardless of the specific time at which they were administered, FLEI and PSS had no effect on MMSE scores. No effects of the interaction between either FLEI or PSS and the time of assessment on MMSE scores was found. The results of the trends in cognitive functioning pre- and post-lockdown as indicated by ENB-2 scores are reported in Table 6 .Table 6 Pre- and post-lockdown changes in ENB-2 scores. Table 6 Model 1 Model 2 Model 3 Model 4 Model 5 B (SE) B (SE) B (SE) B (SE) B (SE) Intercept 50.21 (2.05) 53.76 (2.26) 53.99 (2.38) 53.91 (2.27) 54.15 (2.26) Post −6.06 (1.63) −6.29 (1.77) −6.21 (1.74) −6.45 (1.69) PSS (T1) −0.02 (0.36) FLEI (T1) 0.03 (0.16) Post × PSS (T1) −0.06 (0.26) Post × FLEI (T1) −0.05 (0.12) PSS (T2) 0.48 (0.29) FLEI (T2) −0.13 (0.20) Post × PSS (T2) −0.08 (0.22) Post × FLEI (T2) −0.07 (0.16) PSS (T3) 0.43 (0.31) FLEI (T3) −0.12 (0.18) Post × PSS (T3) −0.14 (0.22) Post × FLEI (T3) −0.12 (0.14) Observations (n) 52 52 52 52 52 Respondents σ 9.68 10.08 10.42 9.77 9.9 Log Likelihood −193.94 −186.86 −189.77 −188.10 −187.83 AIC 393.88 381.72 395.54 392.20 391.67 BIC 399.73 389.52 411.15 407.81 407.28 Kenward-Roger's F-test F(1,23) = 13.65** F(4,33) = 0.09 F(4,33) = 0.89 F(4,33) = 1.02 R2 0.00 0.06 0.06 0.14 0.15 f2 0.06 0.00 0.09 0.01 Notes: ***: p < 0.001; *: p < 0.05; PSS: Perceived Stress scale; FLEI: FLEI Mental Ability Questionnaire; Post: Post-lockdown assessment; T2: second assessment during lockdown; T3: third assessment during lockdown; AIC: Akaike Information Criterion; BIC: Bayesian Information Criterion; R2: Proportion of variance explained by the fixed effects [22]: f2: Effect size measure expressing the proportion of variance explained by the fixed effect relative to the proportion of outcome variance unexplained [1]. The ENB-2 scores showed the same trend as the MMSE scores. Specifically, time of assessment was the only significant effect, although its effect size was small (f2 = 0.06). Post-lockdown scores tended to be lower than pre-lockdown scores, indicating a decrease in the cognitive functioning of respondents after lockdown. Neither the perceived levels of S nor of CE nor their interaction with time had an effect on the respondents' cognitive functioning. 4 Discussion This study investigated the stress experienced by older individuals with mild/moderate dementia during the first-wave of lockdown and their perceptions of cognitive efficiency in that period. As shown in the Results section, the perceived level of stress increased during lockdown, at both T2 and T3. The highest increase was observed at T2, the assessment time coinciding with the peak of the pandemic (i.e., around mid-April). Despite a lower awareness of surrounding events and the fragile cognition which characterises people with neurocognitive disorders [9], these results suggest that older individuals with mild/moderate dementia are still able to perceive events taking place around them, hence causing them to experience stress. Since the major peak of psychological malaise coincided exactly with the pandemic reaching its most critical point, we may speculate that the impact of Covid-19 was general and pervasive and also affected people with neurodegenerative diseases. The significant association between cognitive reserve and stress further corroborates this supposition. Regardless of the time of assessment, the higher the level of cognitive reserve, the higher the level of perceived stress. As such, we may hypothesise that individuals with higher levels of cognitive reserve also have more cognitive resources allowing them a better understanding of the world and the circumstances around them. Taken together, the results of the MMSE and the ENB-2 indicate that cognitive functioning worsened during lockdown, and this deterioration was not associated with either the perceived level of stress or perceived cognitive efficiency. Although the effects sizes observed in the present study were rather small, their clinical relevance is not negligible, also in light of other studies in literature that report a worsening of cognitive and behavioural symptoms like memory deficits, apathy/aggressivity and stress in individuals with dementia [5]. Interestingly, FLEI scores exhibited a decreasing trend, which indicates a higher level of perceived cognitive efficiency at the end of lockdown than that at the beginning (higher scores correspond to a lower perceived cognitive efficiency while lower scores correspond to a higher level of cognitive efficiency). However, this result was not statistically significant and should therefore be treated with caution. The data collected qualitatively contain additional information about participants' routines before and during the pandemic and reveal that the majority of them were assisted in their daily activities (e.g., in purchasing medicines and necessities) from close caregivers (mainly sons or daughters). According to other studies (i.e., [2]), it could be hypothesised that caregivers also perceived a general malaise due to lockdown (for example, stress, anxiety and depression), emphasizing, in turn and indirectly, the stress perceived by individuals with dementia. Furthermore, the ‘negative’ influence of the caregivers' burden could induce (or exacerbate) behavioural and psychological symptoms in this clinical population [4]. In this way, one may argue that the PS observed in our sample was increased also by the PS expressed by caregivers. However, the “stay-at-home” directives for reducing the risk of contagion prevented caregivers from having contact with the study participants, who therefore had to fend for themselves day by day. We may hypothesise that in a context such as that of lockdown, individuals with dementia did indeed experience stress but that the circumstances gave them the chance to “roll up their sleeves”, which, in turn, gave them a greater feeling of efficiency in their everyday lives, including at the cognitive level. Notwithstanding these last points, a significant cognitive decline emerged at the post-lockdown neuropsychological follow-up, in line with the neurodegenerative nature of the disease, but it was not associated with PS and CE. 4.1 Limitations Some limitations must certainly be acknowledged. Firstly, the small sample size may weaken the generalisability of the results. In addition, we were unable to conduct a post-lockdown follow-up. Secondly, baseline PS and CE data (i.e., before Covid-19) for the population studied were not available. Furthermore, we had not investigated the affective component that probably had a significative influence on the PS and CE of these individuals. If additional data on these variables were collected at the end of the pandemic they might be helpful in clarifying the participants' routine self-evaluations. Lastly, there was no control group. Given these important limitations, the results reported here should be considered preliminary and treated with caution. 5 Conclusions Individuals with dementia seem to have experienced stress during the first-wave of lockdown related to Covid-19. This result suggests that, in particular situations, these individuals may still have consciousness and perceive stress even if not everyone has consciousness of the events happening around them. Consequently, it is reasonable to think that the clinicians could attend these persons with ad hoc ‘remote’ interventions (because of social distancing) to calm down the Behavioural and Psychological Symptoms of Dementia (BPSD) or cognitive impairment (for example, through psycho-educational support for caregivers or the Cognitive Stimulation Therapy; [10]). These individuals' level of perceived stress was significantly associated with their cognitive reserve, while their cognitive functioning was lower after lockdown than in the pre-lockdown period. This worsening seems to be due to the neurodegenerative nature of the disorder as the cognitive functioning scores were not associated with perceived stress nor with perceived cognitive efficiency. Nevertheless, we did not know how much the affective component influenced our participants, so these data should be taken cautiously. Disclosure statement No potential conflict of interest was reported by the authors. Funding This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors. Statement of human and animal rights This study was carried out with respect for human and animal rights. Informed consent Informed consent to collect their clinical data was given by all patient. ==== Refs References 1 Aiken L.S. West S.G. Multiple Regression: Testing and Interpreting Interactions 1991 Sage Newbury Park 2 Altieri M. Santangelo G. The psychological impact of COVID-19 pandemic and lockdown on caregivers of people with dementia Am. J. Geriatr. Psychiatry 29 1 2021 27 34 33153872 3 Brooks S.K. Webster R.K. Smith L.E. Woodland L. Wessely S. Greenberg N. Rubin G.J. The psychological impact of quarantine and how to reduce it: rapid review of the evidence Lancet 395 10227 2020 912 920 32112714 4 Cagnin A. Di Lorenzo R. Marra C. Bonanni L. Cupidi C. Laganà V. …SINdem COVID-19 Study Group Behavioral and psychological effects of coronavirus disease-19 quarantine in patients with dementia Front. Psychiatry 11 2020 916 5 Canevelli M. Valletta M. Blasi M.T. Remoli G. Sarti G. Nuti F. …Bruno G. Facing dementia during the COVID-19 outbreak J. Am. Geriatr. Soc. 2020 6 Cohen J. A power primer Psychol. Bull. 112 1992 155 159 19565683 7 Cohen S. Williamson G. Perceived stress in a probability sample of the United States Spacapan S. Oskamp S. The Claremont Symposium on Applied Social Psychology The Social Psychology of Health 1988 Sage Publications Inc. 31 67 8 Courtin E. Knapp M. Social isolation, loneliness and health in old age: a scoping review Health Social Care Commun. 25 3 2017 799 812 9 De Roeck E.E. van der Vorst A. Engelborghs S. Zijlstra G.R. Dierckx E. D-SCOPE Consortium Exploring cognitive frailty: prevalence and associations with other frailty domains in older people with different degrees of cognitive impairment Gerontology 66 1 2020 55 64 31330515 10 Devita M. Bordignon A. Sergi G. Coin A. The psychological and cognitive impact of Covid-19 on individuals with neurocognitive impairments: research topics and remote intervention proposals Aging Clin. Exp. Res. 2020 1 4 31721096 11 DiGiovanni C. Conley J. Chiu D. Zaborski J. Factors influencing compliance with quarantine in Toronto during the 2003 SARS outbreak Biosecur. Bioterr. 2 4 2004 265 272 12 Faraway J.J. Extending the Linear Model with R: Generalized Linear, Mixed Effects and Nonparametric Regression Models 2016 CRC Press 13 Folstein M.F. Folstein S.E. McHugh P.R. “Mini-mental state”: a practical method for grading the cognitive state of patients for the clinician J. Psychiatr. Res. 12 3 1975 189 198 1202204 14 Hawryluck L. Gold W.L. Robinson S. Pogorski S. Galea S. Styra R. SARS control and psychological effects of quarantine, Toronto, Canada Emerg. Infect. Dis. 10 7 2004 1206 15324539 15 Holmes E.A. O'Connor R.C. Perry V.H. Tracey I. Wessely S. Arseneault L. …Ford T. Multidisciplinary research priorities for the COVID-19 pandemic: a call for action for mental health science Lancet Psychiatry 10 2020 30168-1.ISO 690 16 Jalloh M.F. Li W. Bunnell R.E. Ethier K.A. O'Leary A. Hageman K.M. …Marston B.J. Impact of Ebola experiences and risk perceptions on mental health in Sierra Leone, July 2015 BMJ Glob. Health 3 2 2018 e000471 17 Jeong H. Yim H.W. Song Y.J. Ki M. Min J.A. Cho J. Chae J.H. Mental health status of people isolated due to Middle East Respiratory Syndrome Epidemiol. Health 2016 38 18 Kenward M.G. Roger J.H. Small sample inference for fixed effects from restricted maximum likelihood Biometrics 53 1997 983 997 9333350 19 Lee S. Chan L.Y. Chau A.M. Kwok K.P. Kleinman A. The experience of SARS-related stigma at Amoy gardens Soc. Sci. Med. 61 9 2005 2038 2046 15913861 20 Magni E. Binetti G. Bianchetti A. Rozzini R. Trabucchi M. Mini-mental state examination: a normative study in Italian elderly population Eur. J. Neurol. 3 3 1996 198 202 21284770 21 Mondini S. Mapelli D. Vestri A. Arcara G. Bisiacchi P.S. L’Esame Neuropsicologico Breve-2 2011 Raffaello Cortina Editore Milano 22 Nakagawa S. Johnson P.C. Schielzeth H. The coefficient of determination R2 and intra-class correlation coefficient from generalized linear mixed-effects models revisited and expanded J. R. Soc. Interface 14 134 2017 23 Nucci M. Mapelli D. Mondini S. Cognitive Reserve Index questionnaire (CRIq): a new instrument for measuring cognitive reserve Aging Clin. Exp. Res. 24 3 2012 218 226 21691143 24 Orrù G. Ciacchini R. Gemignani A. Conversano C. Psychological intervention measures during the COVID-19 pandemic Clin. Neuropsychiatry 17 2 2020 76 79 34908972 25 Saadatian-Elahi M. Facy F. Del Signore C. Vanhems P. Perception of epidemic's related anxiety in the General French Population: a cross-sectional study in the Rhone-Alpes region BMC Public Health 10 1 2010 191 20385030 26 Mazza C. Ricci E. Biondi S. Colasanti M. Ferracuti S. Napoli C. Roma P. A nationwide survey of psychological distress among Italian people during the COVID-19 pandemic: immediate psychological responses and associated factors International journal of environmental research and public health 17 9 2020 3165 32370116 27 Stern Y. Cognitive reserve in ageing and Alzheimer's disease Lancet Neurol. 11 11 2012 1006 1012 23079557 28 Yanez N.D. Weiss N.S. Romand J.A. Treggiari M.M. COVID-19 mortality risk for older men and women BMC Public Health 20 1 2020 1 7 31898494 29 Yip P.S. Cheung Y.T. Chau P.H. Law Y.W. The impact of epidemic outbreak: the case of severe acute respiratory syndrome (SARS) and suicide among older adults in Hong Kong Crisis 31 2 2010 86 20418214 30 Yoon M.K. Kim S.Y. Ko H.S. Lee M.S. System effectiveness of detection, brief intervention and refer to treatment for the people with post-traumatic emotional distress by MERS: a case report of community-based proactive intervention in South Korea Int. J. Ment. Heal. Syst. 10 1 2016 51

34340137

PMC9749757

NO-CC CODE

2022-12-15 23:23:21

no

J Psychosom Res. 2021 Oct 27; 149:110584

utf-8

J Psychosom Res

10.1016/j.jpsychores.2021.110584

oa_other

==== Front JACC Cardiovasc Imaging JACC Cardiovasc Imaging Jacc. Cardiovascular Imaging 1936-878X 1876-7591 by the American College of Cardiology Foundation. Published by Elsevier. S1936-878X(22)00613-1 10.1016/j.jcmg.2022.10.008 Imail Cardiac Injury Before and After COVID-19 A Longitudinal Cardiac Magnetic Resonance Study González Jan Elliot BEng Doltra Adelina MD Perea Rosario J. MD, PhD Lapeña Pau Garcia-Ribas Cora MD Reventos Jana MSc Caixal Gala MD Tolosana Jose Maria MD, PhD Guasch Eduard MD, PhD Roca-Luque Ivo MD, PhD Arbelo Elena MD, PhD Sitges Marta MD, PhD Prat-Gonzalez Susanna MD, PhD Mont Lluís MD, PhD ∗ Althoff Till F. MD ∗ ∗ Cardiovascular Institute, Clínic—University Hospital Barcelona, C/ Villarroel 170, 08036 Barcelona, Spain 14 12 2022 14 12 2022 © 2022 by the American College of Cardiology Foundation. Published by Elsevier. 2022 American College of Cardiology Foundation Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcRecent studies based on late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR) have raised great concern about frequent residual myocardial injury even after mild or asymptomatic COVID-19.1 , 2 However, although signs of myocardial injury were found in large proportions of patients after COVID-19, all studies published to date lack baseline imaging and are therefore unable to discriminate between preexisting and COVID-19-induced injury. Against this background, we performed a longitudinal study to assess the individual cardiac impact of COVID-19. A prospective registry of patients undergoing LGE CMR in the context of atrial fibrillation was screened for patients with documented SARS-CoV-2 infection subsequent to LGE CMR. Eligible patients then underwent post-COVID-19 LGE CMR using the same scanner and sequence as for pre-COVID-19 LGE CMR. T1-weighted inversion recovery gradient-echo sequences were acquired in sinus rhythm using electrocardiographic gating and a free-breathing 3-dimensional navigator 15 to 20 minutes after administering an intravenous bolus of 0.2 mmol/kg gadobutrol. An inversion time scout sequence was used to determine the optimal inversion time that nulled the left ventricular myocardial signal. LGE was independently assessed qualitatively by 2 experienced investigators blinded to patient information. For quantitative analyses, a 3-dimensional reconstruction of the left ventricle was performed using dedicated software (Adas3D Medical). LGE was then automatically quantified on the basis of a prespecified signal intensity threshold of ≥3 SDs above the mean of a remote nonenhanced myocardial region. Approval was obtained from the local research ethics committee, and written informed consent was obtained from each patient. Thirty-one patients with confirmed COVID-19 between March 2020 and February 2021 were included. Seven patients (23%) had been hospitalized at the time of acute presentation with COVID-19, of whom 2 (6%) required intensive care. Most patients (29 [96%]) had been symptomatic, but none reported cardiac symptoms. At a median of 5 months post-COVID-19, LGE lesions indicative of residual myocardial injury were encountered in 15 of the 31 patients (48%; not considering LGE lesions at the right ventricular insertion points), which is in line with previous reports.1 , 2 The majority of lesions were located midmyocardially (65%), with only a few isolated subendocardial (18%) or transmural (18%) lesions. However, intraindividual comparison with the pre-COVID-19 CMR revealed all of these lesions as preexisting with identical localization, pattern, and transmural distribution and thus not COVID-19-related (Figure 1 ). Quantitative analyses, performed independently, detected no increase in the size of individual LGE lesions nor in the global left ventricular LGE extent. Comparison of pre- and post-COVID-19 cine imaging sequences did not show any differences in ventricular functional or structural parameters.Figure 1 Corresponding LGE Lesions Pre- and Post-COVID-19 (A to C) Representative examples of corresponding lesions in pre- and post-COVID-19 scans from 3 different patients. (Left) Three-dimensional reconstructions of the left ventricle with late gadolinium enhancement (LGE)–based color coding. (Middle and right) T1-weighted short-axis slices with and without LGE color coding of the respective layer. Of course, these findings are not generalizable and in particular may not apply to patients with severe COVID-19 and/or evident cardiac involvement. Unfortunately, although none of the patients displayed elevated troponin or C-reactive protein levels at the time of post-COVID-19 CMR, troponin levels during acute COVID-19 as a surrogate for cardiac involvement were not acquired systematically, because patients were included only after recovery. Moreover, although T1/T2 mapping would allow for better discrimination of ongoing inflammatory processes and edema from chronic scarring, such CMR sequences were not performed in the pre-COVID-19 scans and were therefore not available for intraindividual comparisons. Because patients were recruited from our prospective atrial fibrillation registry, all patients had a history of atrial fibrillation, and consequently a large proportion displayed cardiovascular risk factors such as hypertension (58%), dyslipidemia (45%), diabetes (26%), coronary artery disease (10%), and history of heart failure (6%). On the basis of available data linking atrial fibrillation and cardiovascular risk factors to COVID-19-associated complications, this selected cohort would be considered at particularly high risk not only for preexisting cardiac injury but also for cardiac sequelae of COVID-19. Against this background, the complete absence of de novo LGE lesions is specifically noteworthy. To the best of our knowledge this is the first CMR study to assess myocardial injury pre- and post-COVID-19. Although with only 31 patients longitudinally studied we cannot rule out the possibility of rare events of COVID-19-induced myocardial injury, the complete absence of de novo LGE lesions after COVID-19 in this cohort indicates that outside special circumstances, COVID-19-induced myocardial injury may be much less common than suggested by previous studies. This work is supported in part by grants from Instituto de Salud Carlos III, Spanish Government, Madrid, Spain (FIS_PI16/00435 – FIS_CIBER16), and Fundació la Marató de TV3, Catalonia, Spain (20152730). Dr Althoff has received research grants for investigator-initiated trials from Biosense Webster. Dr Mont has received honoraria as a lecturer and consultant and has received research grants from Abbott Medical, Biosense Webster, Boston Scientific, and Medtronic; and is a shareholder of Galgo Medical. Dr Marta Sitges has received grants, consulting honoraria, and speaker fees from GE, Edwards Lifesciences, Abbott Medical, and Medtronic. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the Author Center. ==== Refs References 1 Puntmann V.O. Carerj M.L. Wieters I. Outcomes of cardiovascular magnetic resonance imaging in patients recently recovered from coronavirus disease 2019 (COVID-19) JAMA Cardiol 5 2020 1265 1273 32730619 2 Rajpal S. Tong M.S. Borchers J. Cardiovascular magnetic resonance findings in competitive athletes recovering from COVID-19 infection JAMA Cardiol 6 2021 116 118 32915194

0

PMC9749793

NO-CC CODE

2022-12-15 23:23:21

no

JACC Cardiovasc Imaging. 2022 Dec 14; doi: 10.1016/j.jcmg.2022.10.008

utf-8

JACC Cardiovasc Imaging

10.1016/j.jcmg.2022.10.008

oa_other

==== Front Injury Injury Injury 0020-1383 1879-0267 Elsevier Ltd. S0020-1383(21)00127-3 10.1016/j.injury.2021.02.024 Article Epidemiology of hand traumas during the COVID-19 confinement period Fortané Thibaut a⁎ Bouyer Michael a Le Hanneur Malo b Belvisi Baptiste d Courtiol Guillaume c Chevalier Kevin c Dainotto Caroline c Loret Marie a Kling Agathe a Bentejac Antonin a Lafosse Thibault a a Hand, Upper Limb, Brachial Plexus, and Microsurgery Unit (PBMA), Clinique Générale d'Annecy, 4 Chemin de la Tour la Reine, 74000 Annecy, France b Department of Orthopedics and Traumatology – Service of Hand, Upper Limb and Peripheral Nerve Surgery, Georges Pompidou European Hospital (HEGP), Assistance Publique Hôpitaux de Paris - Paris Descartes University, 20 rue Leblanc, 75015 Paris, France. c Emergency department, Clinique Générale d'Annecy, 4 Chemin de la Tour la Reine, 74000 Annecy, France d Orthopedic Surgery Department, Centre Hospitalier Annecy Gennevois (CHANGE), Metz-Tessy, France ⁎ Corresponding author: Thibaut Fortané, Postal address: Hand, Upper Limb, Brachial Plexus, and Microsurgery Unit, Clinique Générale d'Annecy; 4 Chemin de la Tour la Reine, 74000 Annecy, France. 17 2 2021 4 2021 17 2 2021 52 4 679685 12 2 2021 © 2021 Elsevier Ltd. All rights reserved. 2021 Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Introduction hand injuries are a common emergency mainly caused by domestic accidents or sport injuries. During the COVID-19 pandemic confinement period, with a cut off in transportation as well as in occupational and physical activities, we observed a decrease in medical and elective surgical activities but emergency cases of upper limb and hand surgery increased. Materials and methods we conducted a retrospective epidemiological study to analyze two periods between the same dates in 2019 and 2020, for all the duration of the confinement period. We compared the numbers of consultations in the emergency department, elective surgeries, hand and upper limb emergency cases in our center and urgent limb surgeries in the nearby hospital. Then we compared the mechanisms and severity of injuries and the type of surgery. Results between 2019 and 2020 there was a decrease of consultations in the emergency department in our institution of 52%, a decrease of total elective surgeries of 75%, a decrease in surgeries for urgent peripheral limb injuries of 50%, whereas the hand and upper limb emergency remained stable or even increased by 4% regard to occupational and domestic accidents. There was a significant difference in the mechanism of injury with an increase of domestic accident and a decrease of occupational, road traffic and sport accidents. Severity of the injuries increased, with augmentation of the number of tissues involved and longer expected time of recovery. Conclusion during the confinement period of the COVID-19 pandemic, despite an important reduction of medical activities, the amount and severity of hand emergency cases increased. A specific plan regarding duty shift organization for hand trauma should be maintained regardless of the sanitary situation. Keywords Epidemiology Hand injury Confinement Emergency Hand surgery COVID-19 ==== Body pmcIntroduction Hand injuries are common emergency situations, requiring surgical management in specialized centers on a daily basis [1,2]. More than a quarter of all unintentional injuries are hand injuries [3,4]. They are mainly caused by accidents in daily life situations while other traumatic injuries of the upper limb are often observed in the context of sports injuries or road traffic accident [3,5]. There are two main types of hand injuries requiring surgical treatment. Wounds with potential complex lesions of major structures in the deep tissues (e.g., tendons, nerves, arteries, ligaments and joint, bones), which require emergency management in order to save the hand function (e.g., devascularization, open fractures, flexor tendons lesions…) [6,7]. The second category includes all closed injuries involving bones and joints but without skin lesions, such as closed fractures and sprains, which require an urgent care but can be deferred for 7 to 10 days. The orthopedic surgical activity finds a majority of wounds on hand trauma, while on the rest of the upper limb we reported a greater number of closed fractures [8]. In 2020, during the confinement period secondary to the COVID-19 pandemic, there was a significant decrease in professional activities, physical activities and transportation. We then hypothesized that despite the confinement of the population and the decrease in most of the medical and elective surgical activities, traumas of the hand operated in emergency would continue [9]. In addition, we expected a significant reduction in closed trauma of the peripheral limb treated with osteosynthesis, and an increase of number and severity of complex cases. Materials and methods We conducted a retrospective epidemiological study in our center to analyze the medical and surgical activities, along with hand and upper limb injuries within the period of lockdown of the population related to the COVID-19 pandemic inducing a cutback in all professional work activities, physical activities and transportation. Our center is a provincial town with a population of about 210,000 inhabitants. It drains a large population base from the northern French Alpine massif, south of Geneva, north of Grenoble and east of Lyon. Two periods of eight weeks were compared. The first from Wednesday the 18th of March in 2020 to Tuesday the 10th of May in 2020. The time of inclusion was the same as the confinement period as it corresponds to a period during which information emanating from the local health authorities (i.e., hospital direction comity, regional medical council and medical insurance companies) were unanimous (i.e. surgical activity restricted to critical emergencies, trauma, infections, tumors). After the 10th of May 2020, information regarding the authorized activity in hospitals and health care centers authorized a partial resumption of medical activity. The second was a control period of two months from the 18th of March, 2019 to the 10th of May, 2019.For these two periods, we retrospectively collected the number of patients coming to the emergency department of our hospital for other motives than COVID-19-related symptoms, the number of elective cases managed in our hospital and the number of patients operated for upper limb and hand injuries; regarding those cases, we looked more precisely at the energy of the accidents. In addition to our private clinic in Annecy, Rhônes-Alpes, France, we collected the data regarding orthopedics traumas, which were operated at the trauma center of the local public hospital, Annecy, Rhônes-Alpes, France, as the regional health organization (Agence Régionale de Santé) of our area distributes the traumas management between our private center (i.e., hand traumas and microsurgical reconstruction) and the public hospital (i.e., polytrauma and peripheral traumatology). In France, during the two month of lockdown, social policies were defined with few authorized activities. Only authorized movements were those: in the professional context only if essential and that cannot be carried out by telework; to make purchases of basic necessities; for medical reasons, administrative or judicial summons; recreational and individual activity within the limit of 1 kilometer and 1 hour around the home. Analysis criteria The primary analysis criterion was the comparison of the number of patients between these two periods. The secondary analysis criteria were the difference between the type, mechanism and severity of the hand and upper limb injuries operated at our institution. We also analyzed the influence of the age in the group of patients operated for peripheral limb trauma in the public hospital. Definitions A hand injury was defined as any closed or open injury to the wrist and/or the hand, substantially involving skin, muscle, tendon, bone and joint, nerve and/or vessels [10]. An upper limb injury was defined as an injury to the forearm, elbow, arm and shoulder. A high-energy injury was defined by road, ski and mountain accidents, industrials machine and “do it yourself” tools accidents. Peripheral limb trauma was defined as the traumas to the musculoskeletal system, excluding spine and skull and could include the upper limb, excluding hands. Combined injury criteria was analyzed by looking at the number of structures damaged. The skin was considered only in case of a skin loss requiring a flap. Nerves and arteries from a collateral pedicle were considered as one tissue (i.e., neurovascular bundle) on fingers. We described a severity score as the mean of 3 criteria: high-energy injury, combined injury and expected delay before resuming manual daily activity longer or equal to 3 months, marking 1 point for each criteria. Inclusion and exclusion criteria For the primary analysis criteria, we included all patients coming to the emergency department of our institution for a motive of consultation other than COVID-19-related symptoms, and all patients who underwent surgery at our institution regardless of the specialties. For the analysis of the epidemiological data about upper limb and hand emergencies, inclusion criteria were all upper limb injuries that happened during the period of confinement and required surgical management within the 10 days following the initial trauma. The same criteria were applied to the group of patients managed in the public hospital for peripheral limb injuries. The epidemiological data collected were gender, age, job, type of surgery, mechanism of injury and time between the date of injury, the first consultation and surgery. The type of surgery was separated into three categories: osteosynthesis and ligaments suture; wound with soft tissues lesions (skin, muscle, tendon, bone and joint, ligaments, nerves, vessels); infections and hematoma. The mechanism of injury was separated into four categories: occupational accidents; home accidents; motor vehicle accident (MVA) and sports traumatisms. Statistic analysis R for Mac OS X (R Foundation for Statistical Computing, CRAN) and Stata (StataCorp LP, College Station, TX, USA) were used for the statistical analysis. Data were presented as mean and standard deviation (SD). We performed paired Student t-tests if the sample was normally distributed and the homogeneity of variance was given. Otherwise, non-parametric Mann–Whitney tests were used. Normality was verified using Shapiro–Wilk tests and Gaussian aspects of variable distribution on histograms. Homogeneity of variances was verified using Bartlett tests. Chi-squared analyses with individual comparisons were made with Fisher exact tests and χ2 tests, as appropriate. A P-value less than .05 was considered statistically significant. The study was approved by the local review board (number 2217606 v 0), and conducted according to good clinical practice and applicable laws, and the 1964 Declaration of Helsinki. Written consent was given by each patient. Results Between the two periods, we observed in our private institution a decrease of only 10% of hand or upper limb injuries operated in emergency. More specifically, within the hand and upper limb injuries, when high energy injuries were excluded (i.e. transport and sport injuries) we observed an increase by 4%. Meanwhile, a decrease of 52% of the consultations in the emergency department, 75% of elective cases operated in our institution was observed. A decrease of 50% of peripheral limb injuries operated in the nearby public hospital, was observed. Table 1 and Figure 1 describes the repartition of epidemiologic and surgical data between the two periods and the two medical centers.Table 1 Modifications of medical and surgical activities between 2019 and 2020. Table 1 2019 2020 Increase / decrease Hand or upper limb emergencies 1 136 123 -10%  Hand emergencies without road and sport injuries 112 117 +4% Elective surgeries 1 2964 742 -75% Consultations in emergency unit 2 2742 1309 -52% Peripheral limb injury 3 245 122 -50% 1 Orthopaedic department in private clinic. 2 Emergency department in private clinic. 3 Orthopaedic department in public hospital. Data are presented as number of patients in absolutes values. Fig. 1 Modifications of medical and surgical activities between 2019 and 2020. Fig. 1 Epidemiological data The main epidemiological data are summarized in Table 2 .Table 2 Epidemiologic data. Table 2 2019 2020 Increase / decrease POPULATION Age* (years) 41±21 44±21 Sex ratio (=Men / Women) 2,5 2,6 SURGERY Period* from consultation to surgery (days) 1,5±2,1 0,7±1,2 -50% p<0,01 Period* from injury to surgery (days) 3,0±5,3 2,0±4,9 -32% p>0.05 Hand osteosynthesis (No) 41 28 -34% Upper limb osteosynthesis (No) 12 5 -58% Soft tissue lesions (No) 58 73 +26% Infections / hematoma (No) 25 19 +24% ⁎ Data are presented as mean ± standard deviation, unless otherwise stated. No: number of cases/patients in absolutes values The mechanism of injury is presented in Figure 2 . There was a significant difference in repartition of trauma etiology between 2019 and 2020 with a decrease by 50% of occupational accident, a decrease of sport accident by 73% and an increase of domestic injury by 40%, (fisher test, p < 0.01).Fig. 2 Mechanism of injury. Fig. 2 Severity score We observed significant augmentation of severity of hand and upper limb trauma in 2020 in comparison to 2019. The severity score increased from 0.8±0.9 to 1.1±0;9 between 2019 and 2020 (p<0,01). Severity of trauma is presented in Table 3 and Figure 3 . There was a significant augmentation of combined injury and expected delay before resuming manual daily activity from 2019 to 2020.Table 3 Severity of injuries. Table 3 2019 2020 Increase / decrease Expected delay1 before resuming manual daily activity (months) 2,0±1,5 2,9±1,9 +45% p<0,01 Combined injury (No) 26 39 +50% High energy injury (No) 30 29 -3% Severity score value1,2 0,79±0,89 1,14±0,92 +44% p<0,01 1 Data are presented as mean ± standard deviation, unless otherwise stated. No: number of cases/patients in absolutes values. 2 The severity score is calculated by scoring 1 point for each of the three previous criteria in the table (patient with expected delay before resuming manual daily activity ≥3 months = 1 point). Fig. 3 Severity of injury. Fig. 3 Age analysis In the group of patients operated from peripheral limb injuries at the local public hospital, the mean age was not different in 2020 (62±28 years) and 2019 (57±28 years), p > 0.05. When this population was separated into three groups (<65, 65-80 and >80 years old) the decrease in surgical activity were respectively by 60%, 55% and 31%. In our private clinic, the mean age was not different between 2020 (44±21 years) and 2019 (41±21 years). We didn't observe any difference when the population was separated into the three same groups. Discussion In this study, we highlighted an augmentation of surgeries for hand emergencies, in opposition with a global medical and surgical decrease in activity. Despite a massive reduction of all surgical and medical activity in our institution and the local public hospital, with the number of patients coming to the emergency department for another reason of consultation than COVID-19, the activity of the elective surgery, and the surgical trauma activity all decreased by more than half (respectively 52%, 75% and 50%). On the contrary the amount of upper limb injuries and hand injuries requiring an urgent surgical management was almost the same (decrease of only 10%). Furthermore, the amount of hand and upper limb injuries, not related to sports and transportation, has increased. We could therefore answer to our hypothesis, stating that hand emergency activity remains equivalent independently from an overall cutback into any other activity. As a center to which patients are referred from nearby hospitals, (being the only hand and microsurgery unit in the area), and knowing that general hospitals focused on managing COVID-19, a greater number of upper limb injuries could have been expected during this period. In our area, sports practiced take place in mountains, and accidents may be serious with upper limb injuries (proximal humerus, clavicle and elbow fractures). Indeed, many of these injuries are secondary to sports trauma or motor vehicle accidents. During this period of the year, people are usually outdoor, playing sports, which increases our upper limb trauma surgery activity. During the confinement period, people stayed at home. As a result, we have seen an increase in domestic injuries such as simple hand soft tissues injuries with knives or glass. On the contrary, there is a decrease in upper limb injuries linked to the prohibition of mountain sports (especially skiing) and the decrease in road traffic. The discrepancy established between the great decrease of activity in the emergency department and the elective cases and the less significant decrease in the amount of upper limb injuries managed in our institution can also be explained by the fact that while many healthcare professionals had to manage COVID-19 infection related cases (explaining the general decline in medical activities), our institution focused on the management of hand and upper limb injuries which became the main reason for consultation in our department, and in the emergency department of our institution [11,12]. During the confinement, the trend of the amount of major structure lesions, and the severity of the damaged caused by domestic accidents seemed to highlight an increase compared to the previous year. Consequently, we designed a severity score showing an increase of combined injury and an augmentation of the expected delay before resuming manual daily activity. We believe, since people were confined at home, there were more accidents as they performed handy or “do it yourself” work, manipulating tools they were not used to, such as chain or circular saws. The most severe lesions affecting many tissues (tendon, artery, nerve, and skin-loss lesion mainly) were indeed mainly due to gardening activities (wounds by chainsaw or hedge trimmer) and do-it-yourself activities (circular saw). The distribution of the mechanism of accident in 2019 is similar to those reported in other studies particularly with regard to occupational activities (between 25 and 30%) [4,5,13,14]. Sports injuries are less represented compared to the literature, because in this study we only evaluated the number of injuries requiring surgery, whereas many are managed in the emergency department or in our clinic. Domestic accidents are comparable to the literature and as shown by Campbell, falling and punching were the commonest mechanisms of fracture whereas glass, knives and "do-it-yourself" materials were most frequently implicated in wounds [15]. Domestic accidents were responsible for the majority of the traumas observed during this period. We have observed a decrease of work accident during this time. This is consistent with the cutback in professional activity linked to confinement. Sorock showed that hand injuries during occupational activities are mainly due to industries [16]. As they almost all closed, we observed a decrease in occupational hand injuries. Rettig showed that most sport injuries at the hand appear to be associated with competition as opposed to the practice arena [17]. This explains the great decrease in sport injuries during confinement. Despite the fact that sports injuries are mostly closed injuries (sprain or fractures) [18,19], and that the number of sport injury decreased, we did not observe any decrease on closed injury of the hand. This is explained by the fact that closed hand injuries are mainly caused by domestic accidents (falls, punching) and are only few sports injuries [15]. When we separated the population of the public hospital into three groups (<65, 65-80 and >80 years old) there seems to be a tense to a greater decrease in general orthopedic surgery activity in the youths than in the elderly. Even though we could not statistically demonstrate this constatation, it is easily explained since the injuries of the elderly are mainly home accidents. Conversely, many sports injuries and motor vehicle accident (MVA) of young people have decreased. In our institution we did not highlight such a population difference because in hand injuries the population remains the same. A strong point of our study is to take into account the entire duration of confinement. Even if this duration is only two months and on only two center in Annecy, it was long enough to highlight a specific trend regarding the augmentation of hand traumas compared to the decrease of the rest of the surgical and medical activity. Still we acknowledged that in order to increase the power of our conclusions, a further retrospective, multicentric study, with a larger population, more representative of the country should be performed [20,21]. We agree with Wilson that traumatic emergencies are linked to weather conditions [22]. Consequently, we chose the same period than the previous year in order to obtain similar weather conditions over weather and season. In order to avoid any selection bias, we chose a duration of inclusion long enough to have a comparable weather conditions between the two periods. Selection bias could also appear if a different population had been referred to us over the period of inclusion. However, as our institution remains the only hand emergency center in the area, this parameter could not influence the hand injury population. Indeed, we usually receive thoroughly all hand emergencies. Regarding the emergencies of peripheral limbs which could have been referred to our institution, as the nearby hospital kept part of its surgical activity, our population was not changed. Conclusion During the confinement period caused by the COVID-19 pandemia, most elective medical and surgical activities sustained a severe drawdown. However, the amount of upper limb injury related surgeries declined less than the rest of the other hospital activities, and the amount of hand traumas which required surgical care increased. Despite the decrease of all medical activities more than half, the amount of upper limb and hand injuries remained stable or even increased regarding to occupational and domestic accidents. Since hand injuries must be managed in specialized centers and do not decrease even when half the population of the world is confined in their homes, and despite a cutback in the activity of the emergency room and in elective or general trauma cases, medical care continuity should always be maintained for hand trauma, whatever the sanitary situation becomes. When healthcare providers focus on managing the COVID-19 pandemia, sometimes despite their area of expertise, hand and orthopedic surgeons should keep providing an on-duty organization to be able to operate on hand injuries and trauma. Funding No funding was received for this study. Ethical approval Since this study was a retrospective chart review, no formal approval from the IRB of our institution was required. All investigations were conducted in accordance with the 1964 Declaration of Helsinki ethical standards and the MR-003 reference methodology*; the study was registered in the National Committee for the Computer Sciences and Liberties (Commission Nationale de l'Informatique et des Libertés – CNIL) database register (number 2217606 v 0) and all patients were individually informed and gave his/her consent before any data collection and analysis. *Journal Officiel de la République Française n°0189 du 14 août 2016. Texte N°77. legifrance.gouv.fr. https://www.legifrance.gouv.fr/affichTexte.do?cidTexte=JORFTEXT000033028290&dateTexte=&categorieLien=id. Access date: 01/01/2020. Potential reviewers Pr Thiery Begué, PHD, Department of Orthopaedics and Traumatology, Antoine-Béclère Hospital - AP-HP, 157 Rue de la Porte de Trivaux, 92140 Clamart E-mail address: tcbegue@free.fr Tel: +33 6 09 44 49 23 Dr Marion Burnier, M.D. Upper Limb and Hand surgery, Medipole Lyon Villeurbanne, 17 avenue Condorcet, 69100 Villeurbanne E-mail address: marion.burnier@wanadoo.fr Tel: +33 6 83 49 19 79 Declaration of Competing Interest The authors declare that they have no conflict of interests. None of them has a financial interest in any of the products, devices, or drugs mentioned in this manuscript. They have not received or will receive any financial aid, in any form, for this study, from any of the following organizations: National Institutes of Health (NIH); Welcome Trust; Howard Hughes Medical Institute (HHMI); or other(s). Appendix Supplementary materials Image, application 1 Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.injury.2021.02.024. ==== Refs References 1 Emmett JE Breck LW. A review and analysis of 11,000 fractures seen in a private practice of orthopaedic surgery, 1937-1956 J Bone Joint Surg Am 40-A 1958 1169 1175 13587591 2 Chung KC Spilson SV. The frequency and epidemiology of hand and forearm fractures in the United States J Hand Surg Am 26 2001 908 915 10.1053/jhsu.2001.26322 11561245 3 Larsen CF Mulder S Johansen AMT Stam C. The epidemiology of hand injuries in The Netherlands and Denmark Eur J Epidemiol 19 2004 323 327 10.1023/b:ejep.0000024662.32024.e3 15180102 4 Smith ME Auchincloss JM Ali MS. Causes and consequences of hand injury J Hand Surg Br 10 1985 288 292 10.1016/s0266-7681(85)80045-0 4078453 5 Angermann P Lohmann M. Injuries to the hand and wrist. A study of 50,272 injuries J Hand Surg Br 18 1993 642 644 10.1016/0266-7681(93)90024-a 8294834 6 Altman RS Harris GD Knuth CJ. Initial management of hand injuries in the emergency patient Am J Emerg Med 5 1987 400 403 10.1016/0735-6757(87)90392-5 3304319 7 Ghosh S Sinha RK Datta S Chaudhuri A Dey C Singh A. A study of hand injury and emergency management in a developing country Int J Crit Illn Inj Sci 3 2013 229 234 10.4103/2229-5151.124101 24459618 8 Crowe CS Massenburg BB Morrison SD Chang J Friedrich JB Abady GG Global trends of hand and wrist trauma: a systematic analysis of fracture and digit amputation using the Global Burden of Disease 2017 Study Inj Prev 2020 10.1136/injuryprev-2019-043495 9 Ducournau F Arianni M Awwad S Baur E-M Beaulieu J-Y Bouloudhnine M COVID-19: Initial experience of an international group of hand surgeons Hand Surg Rehabil 2020 10.1016/j.hansur.2020.04.001 10 Czarnecki P, Dailiana Z, Golubev I, Houpt P, Fernandez S, Megerle K. CRITERIA FOR INCLUSION IN A EUROPEAN EMERGENCY HAND TRAUMA NETWORK n.d.:1. 11 pubmeddev, BL D-KS and H. Considerations for Obstetric Care during the COVID-19 Pandemic. - PubMed - NCBI n.d. https://www.ncbi.nlm.nih.gov/pubmed/32303077 (accessed April 21, 2020). 12 Mak ST Yuen HK. Oculoplastic surgery practice during the COVID-19 novel coronavirus pandemic: experience sharing from Hong Kong Orbit 2020 1 3 10.1080/01676830.2020.1754435 13 Nieminen S Nurmi M Isberg U. Hand injuries in Finland Scand J Plast Reconstr Surg 15 1981 57 60 10.3109/02844318109103413 7268315 14 Marty J Porcher B Autissier R. [Hand injuries and occupational accidents. Statistics and prevention] Ann Chir Main 2 1983 368 370 10.1016/s0753-9053(83)80049-0 9382655 15 Campbell AS. Hand injuries at leisure J Hand Surg Br 10 1985 300 302 10.1016/s0266-7681(85)80048-6 4078455 16 Sorock GS Lombardi DA Courtney TK Cotnam JP Mittleman MA. Epidemiology of occupational acute traumatic hand injuries: a literature review Safety Science 38 2001 241 256 10.1016/S0925-7535(01)00004-2 17 Rettig AC. Epidemiology of hand and wrist injuries in sports Clin Sports Med 17 1998 401 406 10.1016/s0278-5919(05)70092-2 9700410 18 Bergfeld JA Weiker GG Andrish JT Hall R. Soft playing splint for protection of significant hand and wrist injuries in sports Am J Sports Med 10 1982 293 296 10.1177/036354658201000506 7137450 19 DeHaven KE Lintner DM. Athletic injuries: comparison by age, sport, and gender Am J Sports Med 14 1986 218 224 10.1177/036354658601400307 3752362 20 pubmeddev, al BI et. Surgery in the time of Ebola: how events impacted on a single surgical institution in Sierra Leone. - PubMed - NCBI n.d. https://www.ncbi.nlm.nih.gov/pubmed/26787775 (accessed April 21, 2020). 21 Brolin Ribacke KJ van Duinen AJ Nordenstedt H Höijer J Molnes R Froseth TW The Impact of the West Africa Ebola Outbreak on Obstetric Health Care in Sierra Leone PLoS One 11 2016 10.1371/journal.pone.0150080 22 Wilson JM Staley CA Boden AL Boissonneault AR Schwartz AM Schenker ML. The Effect of Season and Weather on Orthopaedic Trauma: Consult Volume Is Significantly Correlated with Daily Weather Adv Orthop 2018 2018 10.1155/2018/6057357

33622592

PMC9749794

NO-CC CODE

2022-12-15 23:23:21

no

Injury. 2021 Apr 17; 52(4):679-685

utf-8

Injury

10.1016/j.injury.2021.02.024

oa_other

==== Front J Pediatr J Pediatr The Journal of Pediatrics 0022-3476 1097-6833 Elsevier Inc. S0022-3476(21)01052-0 10.1016/j.jpeds.2021.10.052 European Paediatric Association Global Emergencies in Child Health: Challenges and Solutions—Viewpoint and Recommendations from the European Paediatric Association and the International Pediatric Association Thacker Naveen MD 123 Hasanoglu Enver MD 14 Dipesalema Joel MD 156 Namazova-Baranova Leyla MD 1789 Pulungan Aman MD 1310 Alden Errol MD 111 Abu-Libdeh Abdulsalam MD 112 Díaz Juan José MD 113 Hoey Hilary MD 17814 Kyne Louise MD 1714 Vural Mehmet MD 17815 Riestra Sergio MD 116 Camcıoğlu Yıldız MD 14 Mujkic Aida MD 17817 Carrasco-Sanz Angel MD 781318 Pettoello-Mantovani Massimo MD, PhD 17819∗ 1 International Pediatric Association, Marengo, IL 2 Indian Academy of Pediatrics, Mumbai, India 3 Asia Pacific Pediatric Association, Kuala Lumpur, Malaysia 4 Turkish National Pediatric Society, Ankara, Turkey 5 Union of National African Pediatric Societies and Associations, Nairobi, Kenya 6 Diabetes Association of Botswana, Gaborone, Botswana 7 European Paediatric Association, Union of National European Paediatric Societies and Associations, Berlin, Germany 8 Association for Scientific Activity and Research, Nouchatel, Switzerland 9 Russian Academy of Pediatrics, Moscow, Russia 10 Indonesian Pediatric Society, Jakarta, Indonesia 11 American Academy of Pediatrics, Itasca, IL 12 Pediatric Society Palestine, Al-Quds University, Palestine 13 Spanish Association of Pediatrics, Madrid, Spain 14 Faculty of Pediatrics of the Royal College of Physicians of Ireland, Dublin, Ireland 15 Turkish Pediatric Association, Istanbul, Turkey 16 National Pediatric Confederation of Mexico, Mexico City, Mexico 17 Croatia Pediatric Society, Zagreb, Croatia 18 European Confederation of Primary Care Pediatricians, Lyon, France 19 Italian Academy of Pediatrics, Milan, Italy ∗ Reprint requests: Massimo Pettoello-Mantovani, MD, PhD, Department of Pediatrics, University of Foggia, Foggia, Italy. 29 10 2021 2 2022 29 10 2021 241 266266.e3 © 2021 Elsevier Inc. All rights reserved. 2021 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. ==== Body pmcGlobal emergencies, including natural disasters, epidemics, drought, armed conflicts, and the SARS-CoV-2 pandemic, have affected populations on five continents, causing devastating socioeconomic effects.1 Children are a most vulnerable and defenseless group.2 In tense situations, they feel overwhelmed and insecure and are often left to their own devices. Their rights to protection and integrity are threatened.3 The European Paediatric Association, Union of the National European Paediatrics Societies and Associations and the International Pediatric Association, representing the national pediatric societies from 149 countries, held a joint conference on October 9, 2021, in Zagreb, Croatia, to discuss the issue of global emergencies in pediatrics and their significant economic and social impact.4 , 5 Delegates discussed the challenges posed by a wide variety of emergencies in the context of increasing global complexities, while also exploring possible solutions. This commentary, prepared by the European Paediatric Association, Union of the National European Paediatrics societies and Associations–International Pediatric Association working group, includes the viewpoint and recommendations from the conference (Table; available at www.jpeds.com). The group strongly recommends that countries develop, reorganize, and strengthen their health systems to address the social and environmental issues caused by global emergencies to enable more efficient and effective allocation and use of available resources for disaster preparedness and emergency response resources in communities. The statement emphasizes the importance of active involvement of all stakeholders, including governments and healthcare professionals. Their joint effort should focus on developing strategic partnerships with key international constituencies, such as the diplomatic, nongovernmental, legal, and academic communities, and the media with the aim of providing adequate support to infants, children, and adolescents affected by adversities.6 Pediatricians and their professional societies across the world must collaborate and share their experiences with new and emerging challenges in child health and together mitigate against inequalities within and between countries and between continents and together strongly and courageously advocate for the health and well-being of children.7 , 8 Appendix Table Global emergencies in child health: challenges and solutions. Viewpoint and recommendations from the 2021 joint conference of EPA/UNEPSA-ECPCP-IPA Diversities, variations, and heterogeneities characterize healthcare services throughout the 5 continents. Cultural and economic complexity and a large disparity in availability, affordability, and accessibility of pediatric care has been shown by pediatric services and community care across the world. The adoption of international aims and objectives relating to services for children including standards of care, education, training, quality improvement projects, research programs, robust outcome measures and socioeconomic goals, are required in order to ensure the effective management of healthcare and the well-being of infants, children, adolescents and their families. Child health andwell-beingchallenged by emergencies  New emergencies, including the effects of a devastating global economic crisis, random cyberattacks and the SARS-CoV-2 pandemic, were added to the long list of issues that chronically afflict children and impact their health and well-being.  The consequences of such socioeconomic turbulence raise serious concerns regarding the sustainability of healthcare systems and their ability to provide effective childcare to ensure that essential and optimum medical attention and health care services are provided from infancy, through childhood and adolescence. Turning crisis into opportunities  In a world of increasing economic interconnections the challenges are greater, but so too are the opportunities. Crisis may become an opportunity if interventions for recovery are well-planned and managed. It is of fundamental importance to carefully explore a number of possible solutions to counteract the negative effects of global emergencies and assist countries and their administrators in their efforts to develop cost-effective solutions, while ensuring that the goal of balancing budgets does not decrease the basic quality standards for public health. Cultural heritage as a strength for cooperation  Cultural heritage is a fundamental category of tangible and intangible values characterizing all aspects of a community, which are continuously remodeled by the political, economic and social concerns. The concept of diversity, as a direct expression of cultural heritage, has become progressively perceived as a resource, helping societies to identify the best solutions to confront and respond to challenges encountered by the different communities throughout the world. The importance of dialogue within diversity and between diverse cultures has emerged as a strong element of cohesion between cultures, in addition to the contribution that diversity can provide for solutions of common problems. Establishing effective cooperation platforms among countries  The presidents of the national pediatric societies, were convened by the EPA/UNEPSA, in collaboration with the IPA and the European Confederation of Primary Care Pediatricians (ECPCP) in Zagreb on October 9, 2021. One conclusion—effective cooperation between countries throughout—is based on the acknowledgement that diversity is a factor of strength and not of weakness. This key factor may create the basis of an effective cooperation in all fields of public interest. Contribution to the efforts of creating an effective platform for cooperation and a multidisciplinary approach to common issues in public health may reduce fragmentation of pediatrics and tackle the legal, economic, and organizational challenges of child healthcare throughout the world. Natural, economic disasters and public health emergencies  Natural and economic adversities and public health crises have often revealed a low degree of self-sufficiency and a high degree of unpreparedness by nations. Natural, economic disasters and public health emergencies are interconnected phenomena.  Disasters and distressing natural events need to be met with rescue and recovery interventions, including adequate health approaches. The nature and effects of these disasters are progressively more complex because they are influenced by several factors including climate change, population movement, economic interdependence, and the general phenomenon of globalization. Tackling emerging andre-emerginginfections of major importance in child health  A complex interplay of environmental and human factors, including ecological, genetic, political and socioeconomic factors, interact to result in the emergence of infectious diseases with unique impacts on children as the most vulnerable members of our society These emerging infections can have unique impacts on younger populations in terms of both physical and mental health, as well as social well-being. In order to best protect them from the impact of emerging and re-emerging infectious diseases, it is imperative to understanding how factors that determine disease emergence and emerging diseases themselves can affect the young. Mitigating harm to children directly and indirectly involved in armed conflicts and gun violence  All children should be guaranteed the right to live, learn and grow up safely—free from violence and fear. Each year, thousands of children are either killed or severely injured in armed conflicts around the world or involved in episodes of domestic and community violence characterized by the use of guns. Although concerns are raised worldwide about the use of explosives, the protection of children often lacks practical solutions. Focus on challenges and practical steps on how to strengthen the protection of children in such conflicts should be further implemented at national and international level. Meeting minor refugees' basic needs  Around the world, millions of families and their children are fleeing their homes owing to adverse events. Protracted conflicts, persistent violence, extreme poverty, and disadvantage press for action to protect children from conflict and to address the root causes of violence and poverty that displace children from their homes. Unaccompanied minors traveling across the world seeking protection is a related increasing phenomenon, which poses a significant challenge to the authorities and the social and healthcare systems worldwide. Both short- and long-term solutions are essential in order for children to escape conflict, persecution and poverty. These must include increasing access to education, strengthening health and child protection systems and social safety nets, expanding opportunities for family income and youth employment, and facilitating peaceful conflict resolution and tolerance.  A multitude of children currently face danger, detention, deprivation, and discrimination. The global pediatric community must stand up for them, work collectively in order to identify issues of concern and strongly advocate with local and international stakeholders. The importance of developing reliable and effective coordinated strategies  A nation's ability to prepare for, respond to, and recover from disaster/emergencies, especially in regard to children, should not depend on a single level or agency of government, and cannot be tackled with fragmented approaches, but by integrated strategic plans. RECOMMENDATIONS  An effective system for disaster management should depend on well-planned, coordinated, interactive strategies and reliable methodologies, based on a shared responsibility, centered on each team member doing what it does best and leveraging the expertise and strengths of others. Most importantly, it must be relevant and applicable to the needs of the country and community.  The many threats posed to global health have emphasized the importance for countries to accelerate the development of guidelines for short-, medium-, and long-term preparedness, to be applicable to different situations, and to enhance the ability to develop adequate strategies and target resources.  Effective strategies should guide states and their local authorities to better identify impediments, which at any level may delay timely distribution of funds, identify best practices, and make recommendations to overcome these complications.  Effective strategies established to tackle emergencies should include an integrated competent communication system that is able to reach both local administrators and populations, in order to keep them informed as to program requirements and opportunities for assistance. Sharing knowledge, expertise and recourses will be key to success.  Caring for children after an emergency event should be a priority, as the amount of trauma/damage caused by a disaster can be overwhelming and affect children physically and mentally. Separation from school, family, and lack of peer support from friends can create additional stress and anxiety for children. Final statement  Investment in services for the health and well-being of children and adolescents will improve the physical, mental and emotional development of children and the subsequent health and well-being of adults. These services must have the ability to respond effectively to emergencies and to any form of challenge to preserving the health and well-being of all children.  It is now more important than ever that pediatricians and their professional societies across the world collaborate and share their experiences as new and emerging challenges in child health continue. Together they can mitigate against inequalities within child health, between countries, between continents and together strongly and courageously advocate for the health and well-being of children. Zagreb, Croatia, October 9, 2021  Internaional Pediatric Associaion (IPA)  European Paediatric Association-Union of National European Paediatric Societies and Associations  (EPA-UNEPSA)  European Confederation of Primary Care Pediatricians (ECPCP) Working Group  Errol Alden (USA), Donjeta Bali (Albania), Shimon Barak (Israel), Alexander Baranov (Russia), Yackov Berkun (Israel), Yıldız Camcıoğlu (Turkey), Angel Carrasco-Sanz (Spain), Robert Cohen (France), Joel Dipesalema (Bothswana), Juan José Díaz (Spain), Zachi Grossman (Israel), Enver Hasanoglu (Turkey), Hilary Hoey (Ireland), Georgios Konstantinidis (Serbia), Giorgina Kuli-Lito (Albania), Louise Kyne (Ireland), Corinne Levy (France), Abdulsalam Abu-Libdeh (Palestine), Arnaud G. L’Huillier (Switzerland), Julije Mestrovic (Croatia), Mohammed Mugar Resen (Iraq), Aida Mujkic (Croatia), Leyla Namazova-Baranova (Russia), Luigi Nigri (Italy), Mónica Oliva (Portugal), Massimo Pettoello-Mantovani (Italy), Doina Plesca (Romania), Tudor Lucian Pop (Romania), Aman Pulungan (Indonesia), Sergio Riestra (Mexico), Sergey Sargysan (Armenia), Eli Somekh (Israel), Neveen Thacker (India), Mehmet Vural (Turkey) EPA/UNEPSA, European Paediatric Association, Union of the National European Paediatrics societies and Associations; IPA, International Pediatric Association; European Confederation of Primary Care Pediatricians (ECPCP). The authors declare no conflict of interest. ==== Refs References 1 Pettoello-Mantovani M. Namazova-Baranova L. Ehrich J. Integrating and rationalizing public healthcare services as a source of cost containment in times of economic crises Ital J Pediatr 42 2016 18 26911573 2 Ferrara P. Guadagno C. Sbordone A. Amato M. Spina G. Perrone G. Child abuse and neglect and its psycho-physical and social consequences: a review of the literature Curr Pediatr Rev 12 2016 301 310 27634538 3 Pettoello-Mantovani M. Pop T.L. Mestrovic J. Ferrara P. Giardino I. Carrasco-Sanz A. Fostering resilience in children: the essential role of healthcare professionals and families J Pediatr 205 2019 298 299.e1 30684982 4 Pettoello-Mantovani M. Mestrovic J. Mehmet Vural M. Namazova-Baranova L. Looking at the future, learning from the past: current activities and upcoming goals of the European Paediatric Association, the Union of National European Paediatric Societies and Associations J Pediatr 220 2020 272 274.e1 32151391 5 International Pediatric Association Every child, every age, everywhere https://www.ipa-world.org/ 6 Parikh P.A. Shah B.V. Phatak A.G. Vadnerkar A.C. Uttekar S. Thacker N. COVID-19 pandemic: knowledge and perceptions of the public and healthcare professionals Cureus 12 2020 e8144 32550063 7 Ferrara P. Corsello G. Ianniello F. Sbordone A. Ehrich J. Giardino I. Internet addiction: starting the debate on health and well-being of children overexposed to digital media J Pediatr 191 2017 280 281.e1 29637892 8 Pettoello-Mantovani M. Carrasco-Sanz A. Pop T.L. Mestrovic J. Somekh E. Giardino I. Plan for the worst, but hope for the best: investing in pediatric services J Pediatr 232 2021 314 315.e1 33548263

34756940

PMC9749795

NO-CC CODE

2022-12-15 23:23:21

no

J Pediatr. 2022 Feb 29; 241:266-266.e3

utf-8

J Pediatr

10.1016/j.jpeds.2021.10.052

oa_other

==== Front Inf Sci (N Y) Inf Sci (N Y) Information Sciences 0020-0255 1872-6291 Elsevier Inc. S0020-0255(22)00100-1 10.1016/j.ins.2022.01.062 Article Deep features to detect pulmonary abnormalities in chest X-rays due to infectious diseaseX: Covid-19, pneumonia, and tuberculosis Mahbub Md. Kawsher a Biswas Milon a Gaur Loveleen b Alenezi Fayadh c Santosh KC d⁎ a Bangladesh University of Business and Technology, Rupnagar, Mirpur-2, Dhaka 1216, Bangladesh b Amity University, Gautam Buddha Nagar, 201313 Uttar Pradesh, India c Department of Electrical Engineering, College of Engineering, Jouf University, 72238, Saudi Arabia d 2AI: Applied Artificial Intelligence Research Lab – Computer Science, University of South Dakota, 414 E Clark St, Vermillion, SD 57069, USA ⁎ Corresponding author. 4 2 2022 5 2022 4 2 2022 592 389401 16 11 2021 28 1 2022 30 1 2022 © 2022 Elsevier Inc. All rights reserved. 2022 Elsevier Inc. Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre - including this research content - immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active. Chest X-ray (CXR) imaging is a low-cost, easy-to-use imaging alternative that can be used to diagnose/screen pulmonary abnormalities due to infectious diseaseX: Covid-19, Pneumonia and Tuberculosis (TB). Not limited to binary decisions (with respect to healthy cases) that are reported in the state-of-the-art literature, we also consider non–healthy CXR screening using a lightweight deep neural network (DNN) with a reduced number of epochs and parameters. On three diverse publicly accessible and fully categorized datasets, for non–healthy versus healthy CXR screening, the proposed DNN produced the following accuracies: 99.87% on Covid-19 versus healthy, 99.55% on Pneumonia versus healthy, and 99.76% on TB versus healthy datasets. On the other hand, when considering non–healthy CXR screening, we received the following accuracies: 98.89% on Covid-19 versus Pneumonia, 98.99% on Covid-19 versus TB, and 100% on Pneumonia versus TB. To further precisely analyze how well the proposed DNN worked, we considered well-known DNNs such as ResNet50, ResNet152V2, MobileNetV2, and InceptionV3. Our results are comparable with the current state-of-the-art, and as the proposed CNN is light, it could potentially be used for mass screening in resource-constraint regions. Keywords Chest X-ray DNN Medical imaging Infectious DiseaseX Covid-19 Pneumonia Tuberculosis Abbreviations CXR, Chest X-ray TB, Tuberculosis DNN, Deep Neural Network MTB, Mycobacterium Tuberculosis CT, Computed Tomography AI, Artificial Intelligence NN, Neural Network DL, Deep Learning ML, Machine Learning ACC, Accuracy AUC, Area Under the Curve SPEC, Specificity SEN, Sensitivity CADx, Computer-Aided Diagnosis CNN, Convolutional Neural Network WHO, World Health Organization ==== Body pmc1 Introduction Infectious diseaseX is global threat. Covid-19 is caused by the SARS-CoV-2 virus, Pneumonia is caused by viral infections, and Tuberculosis (TB) is caused by Mycobacterium Tuberculosis (MTB) bacteria. All of these affect the lungs. According to the World Health Organization (WHO) [1] 213 million people were sick at the time of this study, and Covid-19 was responsible for 4.45 million deaths. Pneumonia took the lives of around 8,08,000 people in 2017, with children under the age of five accounting for 15% of all deaths. The death rate of the elderly has remained steady since 1990. TB infected 10 million individuals and killed 1.4 million people in 2019. With these data, it is critical to concentrate on understanding pulmonary disorders/abnormalities. Artificial intelligence (AI) has prompted numerous improvements in medical imaging, and X-ray imaging technology is fairly common and less expensive as compared to Computed Tomography (CT) scans. The nature of the screening processes in the Covid-19 period has been enhanced by health tools [2], [3], [4]. Researchers [5] employed deep learning (DL) based algorithms to detect the evidence of Covid-19 infection in lung region, reducing prognosis time and highlighting the need for RT-PCR tests. Custom Neural Networks (NNs) have been suggested for chest CT scans and chest X-rays (CXRs) with the help of Computer-aided Diagnosis (CADx) transfer learning to diagnose pulmonary disease. Other studies also used convolutional neural networks (CNNs) to separately detect Covid-19, Pneumonia, and TB infection in patients [6], [7], [8]. In this paper, we proposed a custom designed Deep Neural Network (DNN) model to detect Covid-19, Pneumonia, and TB in CXRs. We conducted cross validation to analyze and evaluated performance using Covid-19, Pneumonia, and TB CXRs datasets, excluding healthy cases. Our findings/results are comparable with the state-of-the-art results for Covid-19, Pneumonia, and TB positive cases. Overall, let us itemize our research contributions:1. Lightweight CNN: We aimed at building a lightweight (less number of layers, parameters, and epochs) DNN model to detect pulmonary abnormalities in CXRs due to infectious diseaseX: Covid-19, Pneumonia, and TB. 2. Comprehensive experiments: Unlike state-of-the-art literature, we were not limited to classify non–healthy cases from healthy ones, but also to classify one type of non–healthy cases from another non–healthy ones. To be precise, we produced an accuracy of 99.87% on Covid-19 versus healthy, 99.55% on Pneumonia versus healthy, 99.76% on TB versus healthy datasets. When considering non–healthy X-ray screening, we received an accuracy of 98.89% on Covid-19 versus Pneumonia, 98.99% on Covid-19 versus TB, and 100% on Pneumonia versus TB. 3. Genericity, scalability, and comparison: As experiments were done on three publicly accessible diverse datasets (excluding healthy cases): Covid-19 (1,200 CXRs), Pneumonia (3,875 CXRs), and TB (3,500 CXRs) without changing/modifying parameters, our lightweight CNN is robust enough to be compared with state-the-art results. Our results are also compared with other well-known DNN algorithms. The rest of the manuscript is organized as follows. Section 2 presents background study of Covid-19, Pneumonia, and TB. Section 3 describes the proposed DNN. Section 4 describes the experimental setup and outcomes in full. Section 5 provides results analysis. We discuss our results (including previous results) in Section 6. Section 7 presents comparison with popular DNNs. The work is concluded in Section 8. 2 Related works AI contributed a lot to healthcare, and pulmonary screening/diagnosis using CXRs is no exception. DL models made it more possible, reliable, and have a significant impact on biomedical research. In this section, we review up-to-date (recently published, peer-reviewed) articles. Covid-19:Covid-19’s primary symptoms include headaches, muscular discomfort, cough, frequent colds, periodic fevers, and breathing difficulties in various susceptible instances [9]. Machine learning (ML) and/or DL models are used to prevent potential human life hazards [10], [11]. Researchers continue working on Covid-19 in 2021 on a large scale as compared to 2020. We refer authors few books on Covid-19 screening for further detailed information [12]. Authors, in [13], worked on a new definition of cluster-based effective one-shot learning to detect Covid-19 cases. Das et al.[14] employed pre-trained CNN model to detect Covid-19 positive patients. Mukherjee et al. [15], [16] developed a CNN model to detect Covid-19 and reported an overall accuracies of 96.28% (CT scans + CXRs) and 99.69% (CXRs) respectively. Authors, in [17], [18], used fuzzy color and stacking approaches with DL models. Deep transfer learning has been always common [19], [20], where authors reported an accuracy of more than 98%. DL models namely ‘DarkCovidNet’ and ‘CoroDet’ were proposed [21], [22]. Pre-trained DL models are no exception [23], [24] to detect Covid-19 cases. Ismael and Sengur [25] pre-trained ImageNet, ResNet50, and SVM classifier, and their reported accuracy was 94.74%. Because of the diverse data sets, the results for each study may vary. For further reading, we refer readers to follow up-to-date research article – how big data is big (for medical imaging: Covid-19)? [26]. Pneumonia: In [27], authors examined DL approaches and automated CXR analysis for pneumonia detection so medical practitioners could accurately diagnose Covid-19. Authors, in [28], employed transfer learning technique with a pre-trained ImageNet model for diagnosis of Pneumonia based on lung segmentation (U-Net architecture). Authors employed largely fine-tuned version of MobileNet V2, InceptionResNet V2, and ResNet50 to see how effective single and combined model can be made to diagnose Pneumonia [29]. Using their combined model, they achieved an accuracy of 93.52%. The COVID-DeepNet system [30] is a popular hybrid multimodal deep learning system that helps radiologists in precise and efficient image interpretation with a precision rate of 99.93%. They reported 100 % precision and F1 score 99.93 %. A.K. Jaiswal et al. [31] employed a DL method to diagnose Pneumonia using CXRs. Similarly, authors [32] used deep transfer learning using pre-trained models such as MobileNetV3, InceptionV3, ResNet18, Xception, DenseNet121, and InceptionV3 to detect Pneumonia, and they achieved an accuracy of 98.43%. Pre-trained models are common. In [33], AlexNet was employed to classify between Covid-19 and normal, and viral and bacterial pneumonia. Authors, in [34], [35], employed deep transfer learning using pre-trained models: Inception-V3, VGG16, ResNet18, DenseNet201, Xception, and SqueezeNet to detect Pneumonia. Similarly, pre-trained models such as GoogLeNet, LeNet, and ResNet-50, ResNet152, DenseNet121, and ResNet18 were used [36], [37]. In [38], transfer learning techniques were used to identify treatable diseases like Pneumonia and reported an accuracy of 100.0%. They have not, however, analyzed whether results were biased. Tuberculosis (TB): TB – a fatal infectious disease – caused by MTB bacteria. In medical imaging, feature selection is crucial, and in [39], authors addressed the importance of correct use of features to achieve optimal performance. Another work [40] evaluates on the impact of image enhancement. In [41], ensemble learning based TB detection in CXRs using hybrid feature descriptors. Another work, in [42], was to eliminate attainable sources of bias in computer assisted CXR analysis for pulmonary TB. In [43], authors suggested CNN models for TB diagnosis based on voting and pre-processing variation ensemble. Their findings show that 97.500% and 97.699% of data accuracy in Montgomery and Shenzhen datasets were achieved with the suggested approach, respectively. Authors of [44] used DL techniques to identify TB using CXRs, and received 94.73% and 98.6% accuracy, respectively. For TB, DCNNs are used [45] for detecting TB using CXR images. Three DNNs, namely CAD4TB, Lunit INSIGHT, and qXR were used [46]. Interestingly, authors used thoracic edge map of CXRs for automatically screening pulmonary abnormalities [47], [48]. Karargyris et al. [49] combined features (local and global), aiming to detect pulmonary abnormalities caused by pneumonia or (TB). Qin et al. [50] explore five different AI algorithms to detect TB from CXRs. Overall, in all three infectious disease types, more often, we observed that transfer learning techniques with pre-trained models such as VGG16, GoogLeNet, LeNet, ResNet-50, ResNet152, DenseNet121, ResNet18, MobileNetV3, InceptionV3, ResNet18, and Xception were employed. Also, they (most of them) were limited to binary classification: non–healthy versus healthy. The proposed tools’ primary objective, whether on Covid-19, Pneumonia, and TB, was to train a single DNN architecture and test accordingly. Additional objective is not only to consider non–healthy versus healthy CXR screening but also to check whether non–healthy CXR screening does work using the exact same DNN model. As before, our aim is to develop one DNN architecture so Covid-19, Pneumonia, and TB positive patients (in CXRs) can be detected. On three publicly available datasets, we evaluated the model. 3 DNN architecture There are various reasons why proposed architecture is a DNN architecture. The first thing we found is that DNNs are quite good at lowering the amount of parameters while maintaining model quality. DNN does not require human feature engineering because it can extract features from an image automatically. We also noted that several of the researchers employed DNN and achieved good image classification and recognition accuracy. Our suggested DNN architecture comprises of three layers: a convolutional layer, a pooling layer, and a fully connected layer to detect Covid-19, Pneumonia, and TB positive patients. To accomplish operations successfully, the layers are fully integrated. We have made our suggested DNN model open to the public1 . The architecture of proposed model is shown in Fig. 1 . The first layer of the architecture is the input layer, with the input shape (224, 224, 1) with strides 2. The second layer is a convolution layer with 256 filters. The filter size of this layer is 3×3 followed by activation function ReLU and 2×2 max pooling layer. The third layer is convolution layer with 128 filters. The filter size of this layer is 5×5 followed by activation function ReLU and 2×2 maxpooling layer. Fourth layer is convolution layer with 256 filters. The filter size of this layer is 7×7 followed by activation function ReLU and 2×2 maxpooling layer. The fifth layer of the architecture is a convolution layer with 128 filters. The filter size of this layer is 3×3 followed by activation function ReLU and 2×2 maxpooling layer. Sixth layer of the architecture is a convolution layer with 64 filters. The filter size of this layer is 1×1, followed by activation function ReLU and the 1×1 maxpooling layer. The seventh layer of the architecture is a flatten or fully connected layer with 0.5 or 50% dropout. The subsequent two layers are dense layers with 256 and 128 neurons with 0.5 or 50% dropout. The eighth and last layer is output layers with a sigmoid activation function. The output shape and number of learning parameters of the proposed DNN architecture are shown in Table 1 .Fig. 1 Architecture of the proposed DNN model for abnormality screening (Covid-19, Pneumonia, and TB). Table 1 Learning parameters for our proposed architecture (image size: 224×224). No Layer (type) Output shape Parameters 1 conv2d (Conv2D) (None, 112, 112, 256) 256 2 max_pooling2d (MaxPooling2) (None, 56, 56, 256) 0 3 conv2d_1 (Conv2D) (None, 52, 52, 128) 819,328 4 max_pooling2d_1 (MaxPooling2) (None, 26, 26, 128) 0 5 conv2d_2 (Conv2D) (None, 20, 20, 256) 1,605,888 6 max_pooling2d_2 (MaxPooling2) (None, 10, 10, 256) 0 7 conv2d_3 (Conv2D) (None, 8, 8, 128) 2,95,040 8 max_pooling2d_3 (MaxPooling2) (None, 4, 4, 128) 0 9 conv2d_4 (Conv2D) (None, 4, 4, 64) 8,256 10 max_pooling2d_4 (MaxPooling2) (None, 4, 4, 64) 0 11 flatten (Flatten) (None, 1024) 0 12 dropout (Dropout) (None, 1024) 0 13 dense (Dense) (None, 256) 2,62,400 14 dropout_1 (Dropout) (None, 256) 0 15 dense_1 (Dense) (None, 128) 32,896 16 dropout_2 (Dropout) (None, 128) 0 17 dense_2 (Dense) (None, 1) 129 Total parameters 3,026,497 4 Experimental setup 4.1 Datasets DNN requires a large amount of data. Using multiple resources, we created six different data combinations to train and evaluate the proposed architecture. Table 2 shows all the detailed information about a collection of all images for this study. Few samples from the above-mentioned collections are provided in Fig. 2 . The combinations of six different dataset (D1 to D6) are listed below (also in Table 3 ):1. [C1:] Tawsifur Rahman’s Covid-19 collection [23], [24] is publicly available. It contains 1,200 Covid-19 positive CXRs (date: October, 2021). 2. [C2:] A publicly available collection [38] (Paul Mooney and his team) is composed of 5,856 CXRs, where 3,875 of them were used in our study. 3. [C3:] We used Kaggle data (Tawsifur Rahman and his team) [44]. It contains 7,000 CXRs (date: October, 2021). This dataset contains 3500 images of TB positive CXRs. 4. [C4:] We collected healthy images from several publicly available sources [23], [24], [38], [44], and 6,182 CXRs were used in our study. Table 2 Data collections (open source). Data collections # of CXRs C1: Covid-19 1,200 C2: Pneumonia 3,875 C3: Tuberculosis 3,500 C4: Healthy 6,182 Fig. 2 CXR samples (see Table 2): Covid-19 (left-most), Pneumonia (middle-left), TB (middle-right) and healthy (right-most). Table 3 Details of dataset used our experiment (Table 1). Dataset Covid-19 Pneumonia Tuberculosis +ve −ve +ve −ve +ve −ve D1 1,200 1,341 – – – – D2 – – 3,875 1,341 – – D3 – – – – 3,500 3,500 D4 1,200 – 3,875 – – – D5 1,200 – – – 3,500 – D6 – – 3,875 – 3,500 – With these collections, we built six different datasets (D1 to D6) for our experiment:1. [D1:] It contains a total of 1,200 Covid-19 and 1,341 healthy cases. 2. [D2:] It is composed of 3,875 Pneumonia and 1,341 healthy cases. 3. [D3:] It includes 3,500 TB and 3,500 healthy cases. 4. [D4:] In D4, we have 1,200 Covid-19 and 3,875 Pneumonia cases. 5. [D5:] In D5, 1,200 Covid-19 and 3,500 TB cases were considered. 6. [D6:] In D6, 3,875 Pneumonia and 3,500 TB cases are considered. The purpose of building the different data organizations (D1 to D3) is to show that our suggested DNN can detect Covid-19, Pneumonia, and TB cases with respect to healthy cases. As Covid-19 could cause Pneumonia, a new dataset (D4) was created to classify Covid-19 positive individuals and those having conventional Pneumonia. TB manifestation were also included in D5 so that we can separate Covid-19 infected patients from TB patients, and vice versa. In addition, D6 was constructed to evaluate whether our model can classify between Pneumonia and TB cases. CXRs images were scaled down to 224×224×1 (grayscale) for this study to match the input dimensions of the proposed DNN model as an input to our architecture. It is also possible to reduce the computational complexity of such a resizing. 4.2 Evaluation protocol and performance metrics We followed 10-fold cross-validation technique to evaluate our approach on all six different data sets: D1 to D6. To measure the performance, six distinct assessment metrics: accuracy (ACC), sensitivity (SEN), specificity (SEPC), precision (PREC), F1 score, and area under the curve (AUC) were used for all 10 folds, and these are computed as follows:ACC=tp+tntp+tn+fp+fn,SEN=tptp+fn,SPEC=tntn+fp,PREC=tptp+fp,andF1score=2PREC×SENPREC+SEN, where tp,fp,tn ,fn refer to true positive, false positive, true negative, and false negative respectively. For evaluation, in addition to straightforward classification accuracy, we emphasize other important metrics such as precision, specificity, sensitivity, and F1 score. Sensitivity refers to the probability of a positive test provided that the individual has the disease. And, specificity refers to the probability of a negative test provided that the individual is healthy. A model’s combined performance score, the symphonic mean of its accuracy and sensitivity, is calculated using the F1 score. 4.3 Model validation The vision was to build a simple, low-computing, low-epoch model that could identify three different lung abnormalities using the exact same DNN model. We utilized few epochs to train our model, and we achieved an optimal performance (twenty epochs). The entire dataset was partitioned into 70:30 ratios for training and validation purpose. The training and validation accuracy of six data sets are shown in Fig. 3 .Fig. 3 Training accuracy versus validation accuracy and training loss versus validation loss on all datasets (see Table 3): a) D1, b) D2, c) D3, d) D4, e.) D5, and f) D6. The training datasets D1, D2, and D3 contain 807, 27,12, 2,478 positive and 968, 939, 2,422 negative Covid-19, pneumonia, and TB CXRs, respectively. Validation datasets D1, D2 and D3 comprise 393, 1,163, 1,022 positive and 373, 402, 1,078 negative Covid-19, Pneumonia and TB CXRs, respectively. To see how our proposed DNN performs for non–healthy CXR screening, we have trained our DNN with non–healthy CXR screening. The training set of non–healthy CXR screening dataset D4 contains 852 Covid-19 positive and 2,797 Pneumonia positive; D5 contains 8,41 Covid-19 positive and 2,451 TB positive CXRs, and D6 2,711 Pneumonia positive and 2,451 TB positive CXRs. The validation set includes 348 Covid-19 positive and 1,078 Pneumonia positive (D4); 359 Covid-19 positive and 1,049 TB positive CXRs (D5); and 1,164 Pneumonia positive and 1,049 TB positive CXRs (D6). In Table 4 , we provide accuracies and corresponding loss from both training validation of six datasets.Table 4 Training and validation performance (after 20 epochs, in %): Training Accuracy (TA), Validation Accuracy (VA), Training Loss (TL), and Validation Loss (VL). Dataset TA VA TL VL D1 99.03 98.88 0.0168 0.0229 D2 99.48 98.36 0.0223 0.0621 D3 99.12 99.18 0.0251 0.0200 D4 99.46 98.31 0.0170 0.1280 D5 98.81 96.06 0.0314 0.1559 D6 99.68 99.61 0.0165 0.0178 5 Results and analysis In this section, we provide both results and analysis of our experiments. In our experiments, we considered not only healthy versus non–healthy CXR screening but also within non–healthy CXRs. 5.1 Healthy versus non–healthy CXR screening To provide a quantifiable result, we present the mean findings on each of healthy versus non–healthy CXR screening using the 10-fold cross-validation. In Table 5 , we provide experimental outcomes. Furthermore, detailed performance scores are provided in the form of confusion matrix in Table 6 . Confusion matrix depicts the distribution of correct and inaccurate predictions by class. Note that, in datasets (D1 to D3), we considered non–healthy versus healthy CXR screening.Table 5 Performance: 10-fold cross-validation accuracy (in %). Dataset k1 k2 k3 k4 k5 k6 k7 k8 k9 k10 Avg (μ±σ) D1 99.61 98.82 99.22 99.61 100.0 99.61 100.0 99.22 99.61 99.61 99.53 ± 0.34 D2 99.62 99.81 99.23 99.81 99.04 98.85 99.42 99.23 99.62 99.42 99.41 ± 0.30 D3 99.29 99.57 99.43 99.86 99.14 99.57 99.57 98.86 99.57 99.57 99.44 ± 0.27 Table 6 Confusion matrix table for healthy versus non–healthy CXR screening. Dataset ap tp fp an tn fn D1 393 393 0 373 372 1 D2 1,163 1,163 0 402 395 7 D3 1,022 1,021 1 1,078 1,074 4 Overall, the proposed DNN correctly detected all 393 Covid-19 CXRs from D1 test set, and one of 372 healthy cases was misclassified as Covid-19 (see confusion matrix in Table 6). In D2 dataset, seven (0f 402) normal instances were misclassified as Pneumonia and 1,163 Pneumonia cases were correctly detected. In D3, 1,021 TB cases were successfully detected and 1 TB case was misclassified as healthy, and 4 out of 1,078 healthy cases were misclassified as TB. Following Table 6, we calculated accuracy, sensitivity, specificity, precision, and F1 score for datasets (D1 to D3) to better understand the performance of the proposed model (see Table 7 ).Table 7 Performance (in %) on healthy versus non–healthy CXR screening: accuracy, sensitivity, specificity, precision, and F1 score. Dataset ACC AUC SEN SPEC PREC F1 score D1 99.87 100.00 99.75 100.00 100.00 99.87 D2 99.55 100.00 99.40 100.00 100.00 99.70 D3 99.76 100.00 99.61 99.91 99.90 99.76 μ 99.72 100.00 99.59 99.97 99.97 99.78 σ ± 0.133 0.00 ± 0.181 ± 0.051 ± 0.058 ± 0.086 5.2 Non–healthy CXR screening To provide quantifiable outcome, we present the mean findings on each of the non–healthy CXR datasets (D3 to D6) using 10-fold cross-validation. Table 8 shows experimental results. As before, we provide confusion matrix in Table 9 for better results analysis and/or understanding.Table 8 Performance: 10-fold cross-validation accuracy (in %). Dataset k1 k2 k3 k4 k5 k6 k7 k8 k9 k10 Avg (μ±σ) D4 99.61 98.82 99.41 99.02 98.04 99.41 99.02 99.21 98.43 99.02 99.00 ±0.45 D5 98.73 97.88 97.45 98.51 97.88 97.66 97.66 98.30 98.73 99.36 98.22 ± 0.58 D6 99.73 99.73 99.46 99.46 99.73 99.59 99.73 99.86 99.86 99.73 99.69 ± 0.14 Table 9 Confusion matrix for non–healthy CXR screening. Dataset ap tp fp an tn fn D4 1,078 1,068 10 348 341 7 D5 1,049 1,043 6 359 349 10 D6 1,049 1,049 0 1,164 1,164 0 For non–healthy CXR screening (D4 to D6 in Table 9), we observe the following. The proposed DNN in D4 misclassified 7 Covid-19 cases as Pneumonia and 10 Pneumonia cases were misidentified as Covid-19. From D5, 10 Covid-19 cases were misidentified as TB and 6 TB cases were misidentified as Covid-19. In D6, both TB and Pneumonia CXRs were correctly classified. With these, Table 10 shows accuracy, sensitivity, specificity, precision, and F1 score. In addition, we provide standard deviation (σ) and mean (μ) that help us check statistical stability of the proposed DNN model. Interestingly, we achieved AUC of 1 across all datasets (D4 to D6).Table 10 Performance (in %) on non–healthy CXR screening: accuracy, sensitivity, specificity, precision, and F1 score. Dataset ACC AUC SEN SPEC PREC F1 score D4 98.89 100.00 99.40 97.15 99.1 99.28 D5 98.99 100.00 99.15 98.31 99.43 99.29 D6 100.00 100.00 100.00 100.00 100.00 100.00 μ 99.29 100.00 99.52 98.49 99.53 99.52 σ ± 0.614 0 ± 0.437 ± 1.433 ± 0.433 ± 0.413 6 Previous studies In this section, we provide a quick overview of our experiments as well as comparison study with several existing approaches. In the literature, majority of the researchers used deep transfer learning approaches [19], [16], [21] using pre-trained models to detect Covid-19, Pneumonia, and TB using either CXRs. Of all, few authors [14], [21] presented DL models that took into account non–healthy CXR screening for Covid-19 and Pneumonia detection. No significant research works took into account non–healthy CXR screening for multiple disease types: Covid-19, TB, and Pneumonia. In Table 11, Table 12, Table 13 , we provide comparative studies. Table 11 shows the findings of some of the most recent methodologies reported for Covid-19 detection. We observed that most of the studies at the beginning of the coronavirus pandemic did not have enough Covid-19 positive CXRs, but later in the year 2020, we there exist relatively large data. In this category, the proposed DNN model achieved 99.87% accuracy, which is the third highest accuracy among them. Note that, our comparison may not be fair as we do not have exact same dataset as well as size. As before, Table 12 shows findings of some of the most recent methodologies reported on Pneumonia detection. In contrast, the proposed model reported an accuracy of 99.55%, which outperforms the existing models. However, it is important to note that our comparison may not be fair as results were produced from using different datasets. In a similar fashion, several modern approaches were found for TB detection; and few recent works are provided in Table 13. As before, our results (e.g., 99.76% accuracy) are comparable with the state-of-the-art results.Table 11 Performance comparison for Covid-19 detection. Also, Covid-19 positive cases are provided. Authors Dataset size Performance (in %) ACC AUC SPEC SEN Das et al. (2020) [14] 18,524 CXRs: Covid-19 (972) 98.77 99.00 99.00 95.00 Togaçar et al. (2020) [17] 458 CXRs: Covid-19 (295) 97.78 – 95.74 98.86 Minaee et al. (2020) [19] 5,420 CXRs: Covid-19 (420) 98.00 – 95.80 91.00 Ozturk et al. (2020) [21] 1,127 CXRs: Covid-19 (127) 98.08 – 95.30 95.13 Chowdhury et al. (2020) [23] 3,487 CXRs: Covid-19 (423) 99.70 100.00 99.55 99.70 Aradhya et al. (2021) [13] 306 CXRs: Covid-19 (69) 79.76 – – – Ismael et al. (2021) [25] 380 CXRs: Covid-19 (180) 92.63 – – – Narin et al. (2021) [18] 14,194 CXRs: Covid-19 (341) 96.01 – 96.60 91.80 Mukherjee et al. (2021) [15] 336 CXRs: Covid-19 (168) 95.83 97.31 98.21 93.45 Mukherjee et al. (2021) [16] 260 CXRs: Covid-19 (130) 96.92 99.22 100.00 94.20 Bassi & Attux (2021) [20] 2,064 CXRs: Covid-19 (439) 100.00 – 99.98 99.99 Hussain et al. (2021) [22] 7,390 CXRs: Covid-19 (2,843) 99.12 – 97.36 95.36 Rahman et al. (2021) [24] 18,479 CXRs: Covid-19 (3,616) 99.65 100.00 95.59 94.56 Proposed DNN 2,541 CXRs: Covid-19 (1,200) 99.87 100.00 100.00 100.00 Table 12 Performance comparison for Pneumonia detection. Also, Pneumonia positive cases are provided. Authors Dataset size Performance (in %) ACC AUC SPEC SEN Kermany, et al. (2018) [38] 5,232 CXRs: Pneumonia (3,883) 93.40 98.80 94.00 96.60 Jaiswal et al. (2019) [31] 2,5684 CXRs: Pneumonia (11,500) 98.18 – – – Stephen et al. (2019) [36] 5,856 CXRs: Pneumonia (4,273) 95.31 – – – Hashmi et al. (2020) [32] 5,856 CXRs: Pneumonia (4,273) 98.43 99.76 98.65 99.00 Jain et al. (2020) [34] 5,856 CXRs: Pneumonia(4,273) 92.31 – – 98.00 Chouhan et al. (2020) [35] 5,229 CXRs: Pneumonia (3,883) 96.40 99.34 – 99.62 Hammoudi et al. (2021) [27] 5,232 CXRs: Pneumonia (3,883) 97.97 – – – Manickam et al. (2021) [28] 5,229 CXRs: Pneumonia (3,883) 93.06 – – 96.78 Asnaoui et al. (2021) [29] 6,087 CXRs: Pneumonia (4,273) 95.09 – 98.31 94.43 Waisy et al. (2021) [30] 800 CXRs: Pneumonia (400) 99.93 – 100.00 99.90 Ibrahim et al. (2021) [33] 11,568 CXRs: Pneumonia(4,450) 94.43 – 100.00 97.44 Cha et al. (2021)[37] 5,856 CXRs: Pneumonia (4,273) 96.63 96.03 – 98.46 Proposed DNN 5,216 CXRs: Pneumonia (3,875) 99.55 100.00 100.00 99.40 Table 13 Performance comparison for TB detection. Also, TB positive cases are provided. Authors Dataset size Performance (in %) ACC AUC SPEC SEN Karargyris et al. (2016) [49] 615 CXRs: TB (275) 89.60 93.00 79.10 89.60 Santosh et al. (2016) [47] 682 CXRs: TB (400) 86.36 94.00 – – Santosh et al. (2017) [48] 1,160 CXRs:: TB (478) 91.00 96.00 – – Lakhani et al. (2017) [45] 1,007 CXRs: TB (492) 99.00 99.00 100.00 97.30 Vajda et al. (2018) [39] 814 CXRs: TB (392) 97.03 99.00 – – Qin et al. (2019) [46] 1,196 CXRs: TB (218) 96.00 94.00 95.00 95.00 Munadi et al. (2020) [40] 662 CXRs: TB (336) 67.55 – – 94.08 Rahman et al. (2020) [44] 7,000 CXRs: TB (3,500) 96.47 – 96.40 Khan et al. (2020) [42] 2,198 CXRs: TB (272) – – 75.00 93.00 Ayaz et al. (2021) [41] 662 CXRs: TB (336) 97.59 99.00 – – Qin et al. (2021) [50] 23,954 CXRs: TB (10,837) 91.29 – 95.00 95.00 Proposed DNN 7,000 CXRs: TB (3,500) 99.76 100.00 99.91 99.61 7 Popular DNNs To provide a fair comparison with the popular DNNs, we used exact same datasets (D1 to D3) and evaluation protocol. In our study, we were limited to the following popular DNNs: ResNet50, ResNet152V2, MobileNetV2, and InceptionNetV3. With these, we computed ROC curves and provided in Fig. 4 . ROC curve (the area beneath the receiver operating characteristics) is an important statistical assessment. As compared to ResNet50, ResNet152V2, MobileNetV2, and InceptionV3, our proposed DNN (Covtben) performed better for all three different infectious diseases types: Covid-19 (D1 dataset), Pneumonia (D2 dataset), and TB (D3 dataset).Fig. 4 AUC comparison: the proposed DNN (CovTbPnNet), ResNet50, ResNet152V2, MobileNetV2, and InceptionNetV3 a) Covid-19, b) Pneumonia, and c) Tuberculosis. For a statistical significance test, we employed the Friedman statistics on three different datasets that follows ROC curve in Fig. 4. With this, we have k(=5) DNN models (InceptionNetV3, CovTbPnNet (proposed model), ResNet152V2, ResNet52, and MobileNetV2) applied on N(=3) different datasets. For this test, we used AUC scores. To know what models perform the best, let us consider rji be the rank of jth model on ith dataset. We then computed mean of the ranks of all jth models on all datasets as,Rj=1N∑i=1Nrji. In Table 14 , we provide detalied ranking information of all models we employed. Our proposed model (CovTbPnNet) ranked 1.333 as opposed to 1.667 (InceptionNetV3). Using null hypothesis, we observed that models do not show significant difference even though our model ranked first. We then computed the Friedman statistics using a chi-squared score,Xk-12=12Nk(k+1)∑jRj2-k(k+1)24. Table 14 Statistical significance test using AUC scores of five different DNN models on three different datasets Dataset/Model InceptionNetV3 CovTbPnNet ResNet152V2 ResNet52 MobileNetV2 D1 1.00 (1.5) 1.00 (1.5) 0.99 (3) 0.98 (4) 0.50 (5) D2 1.00 (1.5) 1.00 (1.5) 0.99 (3) 0.81 (4) 0.50 (5) D3 0.97 (2) 1.00 (1) 0.93 (3) 0.92 (4) 0.61 (5) Mean rank 1.667 1.333 3 4 5 With k-1(=4) degrees of freedom in our test, Xk-12 was 4.83. For = 0.05, upper-tail critical value of chi-square distribution was 5.435. This means that we observed no significant difference. 8 Conclusion In this paper, we have presented a lightweight (9-layered) deep neural network (DNN) to detect pulmonary abnormalities in chest x-rays (CXRs) due to infectious diseaseX: Covid-19, Pneumonia, and Tuberculosis (TB). In our experiments, we were not just limited to healthy versus non–healthy CXR screening, we also extended to non–healthy CXR screening. The latter part of the experiments helped us analyze how well multiple disease types can be used for classification. In all scenarios, performance scores can be compared with existing models (for Covid-19, Pneumonia, and TB). Further, popular DNNs were compared as previous studies used different dataset sizes. As we have received the highest possible accuracy of more than 99.50%, we could see possible screening tool for infectious diseaseX detection. Note that such a tool could help in assisting radiologists to make clinical decisions. Further, we are encouraged to work on cross-population train/test models under the scope of activities as well as federated learning. Ethics declarations Funding: NA. Ethical approval: This article does not contain any studies with human participants performed by any of the authors. CRediT authorship contribution statement Md. Kawsher Mahbub: Methodology, Writing – original draft. Milon Biswas: Methodology, Writing – original draft, Writing – review & editing. Loveleen Gaur: Dicsussion. Fayadh Alenezi: Writing – review & editing. KC Santosh: Methodology, Conceptualization, Writing – original draft, Writing – review & editing. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Md. Kawsher Mahbub is currently working as a Lecturer in the department of Computer Science & Engineering, BUBT. He had received his Bachelor of Engineering degree from Bangladesh University of Business and Technology (BUBT). He enjoy learning about the new advancement in technology, especially concerning IT, Artificial Intelligence, Machine learning, and robotics. His aim is to work innovatively for the enhancement and betterment of education. He enjoys learning about new advancements in technology. He has experienced working in Python, Keras, TensorFlow, Sklearn, and Scipy. His aim is to work innovatively for the enhancement and betterment of education. His particular research interests include health informatics, deep learning, computer vision, and robotics. Assistant Professor Milon Biswas, is Coordinator of Diploma holders of the Department of Computer Science and Engineering at the Bangladesh University of BUsiness and Technology (BUBT). He also serves Bangladesh University of Professionals as an Adjunct Assistant Professor. Artificial intelligence, machine learning, pattern recognition, computer vision, medical image processing, and data mining with applications are some of his research interests. He was awarded the National Science and Technology Fellowship by ICT Ministry of People’s Republic of Bangladesh for his research work during graduate studies. Professor Loveleen Gaur is the Professor and Program Director (Artificial Intelligence Data Analytics of the Amity international Business School, Amity University, Noida, India. She is a senior IEEE member and Series Editor with CRC and Wiley Her expertise in Artificial intelligence and its applications in the business and healthcare domain. Prof Gaur has significantly contributed to enhancing scientific understanding of Artificial Intelligence by participating in over three hundred scientific conferences, symposia, and seminars, by chairing technical sessions and delivering plenary and invited talks. She has chaired various positions in international Conferences of repute and is a reviewer with top rated journals of IEEE, SCI and ABDX Journals. She has been honored with prestigious National and international awards. Dr. Fayadh Alenezi is an Assistant Professor in the Department of Electrical Engineering at Jouf University, Sakaka, Saudi Arabia. He received the B.Sc. degree (Hons.) in Electrical Engineering Electronics and Communications Track from Jouf University, Saudi Arabia in 2012, and the M.S. degree Electrical Engineering from Southern Illinois University Carbondale, Illinois, the USA in 2015, and the Ph.D. degree in Electrical Engineering from the University of Toledo, Ohio, the USA in 2019. Alenezi has authored many journal and conference papers. His research interests include artificial intelligence, image processing, signal processing, image enchantment, machine learning, neural networks, and facial recognition Professor KC Santosh is Chair of the Department of Computer Science at the University of South Dakota (USD). He also serves International Medical University as an Adjunct Professor (Full). Before joining USD, he worked as Research Fellow at the US National Library of Medicine (NLM), National Institutes of Health (NIH). He was Postdoctoral Research Scientist at the Loria Research Centre (with industrial partner, ITESOFT (France)). He has demonstrated expertise in artificial intelligence, machine learning, pattern recognition, computer vision, image processing, and data mining with applications such as medical imaging informatics, document imaging, biometrics, forensics, and speech analysis. His research projects are funded (of more than $2 m) by multiple agencies, such as SDCRGP, Department of Education, National Science Foundation, and Asian Office of Aerospace Research and Development. He is the proud recipient of the Cutler Award for Teaching and Research Excellence (USD, 2021), the President’s Research Excellence Award (USD, 2019), and the Ignite from the U.S. Department of Health & Human Services (2014). Acknowledgement NA 1 GitHub: https://github.com/Kawsher/A-unified-deep-learning-model.git ==== Refs References 1 Fan Wu. Zhao Su. Bin Yu. Chen Yan-Mei Wang Wen Song Zhi-Gang Yi Hu. Tao Zhao-Wu Tian Jun-Hua Pei Yuan-Yuan A new coronavirus associated with human respiratory disease in china Nature 579 7798 2020 265 269 32015508 2 Shi Feng Wang Jun Shi Jun Ziyan Wu. Wang Qian Tang Zhenyu He Kelei Shi Yinghuan Shen Dinggang Review of artificial intelligence techniques in imaging data acquisition, segmentation, and diagnosis for covid-19 IEEE Rev. Biomed. Eng. 14 2020 4 15 3 McCall Becky Covid-19 and artificial intelligence: protecting health-care workers and curbing the spread Lancet Digital Health 2 4 2020 e166 e167 32289116 4 Vaishya Raju Javaid Mohd Khan Ibrahim Haleem Haleem Abid Artificial intelligence (ai) applications for covid-19 pandemic Diabetes Metabolic Syndrome: Clin. Res. Rev. 14 4 2020 337 339 5 Albahri Ahmed Shihab Hamid Rula A Alwan Jwan K Al-Qays Z.T. Zaidan A.A. Zaidan B.B. Albahri A.O.S. AlAmoodi A.H. Khlaf Jamal Mawlood Almahdi E.M. Role of biological data mining and machine learning techniques in detecting and diagnosing the novel coronavirus (covid-19): a systematic review J. Med. Syst. 44 2020 1 11 6 Xueyan Mei, Hao-Chih Lee, Kai-yue Diao, Mingqian Huang, Bin Lin, Chenyu Liu, Zongyu Xie, Yixuan Ma, Philip M Robson, Michael Chung, et al., Artificial intelligence–enabled rapid diagnosis of patients with covid-19, Nat. Med. 26(8) (2020) 1224–1228. 7 Yujin Oh. Park Sangjoon Ye Jong Chul Deep learning covid-19 features on cxr using limited training data sets IEEE Trans. Med. Imaging 39 8 2020 2688 2700 32396075 8 Fan Deng-Ping Zhou Tao Ji Ge-Peng Zhou Yi Chen Geng Fu Huazhu Shen Jianbing Shao Ling Inf-net: Automatic covid-19 lung infection segmentation from ct images IEEE Trans. Med. Imaging 39 8 2020 2626 2637 32730213 9 Huang Chaolin Wang Yeming Li Xingwang Ren Lili Zhao Jianping Hu Yi Zhang Li Fan Guohui Xu Jiuyang Gu Xiaoying Clinical features of patients infected with 2019 novel coronavirus in wuhan, china Lancet 395 10223 2020 497 506 31986264 10 Santosh K.C. Ai-driven tools for coronavirus outbreak: need of active learning and cross-population train/test models on multitudinal/multimodal data J. Med. Syst. 44 5 2020 1 5 11 Santosh K.C. Covid-19 prediction models and unexploited data J. Med. Syst. 44 9 2020 1 4 12 Santosh K.C. Joshi Amit COVID-19: prediction, decision-making, and its impacts 2021 Springer 13 V.N. Manjunath Aradhya, Mufti Mahmud, D.S. Guru, Basant Agarwal, M. Shamim Kaiser, One-shot cluster-based approach for the detection of covid–19 from chest x–ray images, Cogn. Comput. (2021) 1–9. 14 Das Dipayan Santosh K.C. Pal Umapada Truncated inception net: Covid-19 outbreak screening using chest x-rays Phys. Eng. Sci. Med. 43 3 2020 915 925 32588200 15 Mukherjee Himadri Ghosh Subhankar Dhar Ankita Obaidullah Sk Md Santosh K.C. Roy Kaushik Deep neural network to detect covid-19: one architecture for both ct scans and chest x-rays Appl. Intell. 51 5 2021 2777 2789 16 Mukherjee Himadri Ghosh Subhankar Dhar Ankita Obaidullah Sk Md Santosh K.C. Roy Kaushik Shallow convolutional neural network for covid-19 outbreak screening using chest x-rays Cogn. Comput. 2021 1 14 17 Toğaçar Mesut Ergen Burhan Cömert Zafer Covid-19 detection using deep learning models to exploit social mimic optimization and structured chest x-ray images using fuzzy color and stacking approaches Comput. Biol. Med. 121 2020 103805 18 Narin Ali Kaya Ceren Pamuk Ziynet Automatic detection of coronavirus disease (covid-19) using x-ray images and deep convolutional neural networks Pattern Anal. Appl. 2021 1 14 19 Minaee Shervin Kafieh Rahele Sonka Milan Yazdani Shakib Soufi Ghazaleh Jamalipour Deep-covid: Predicting covid-19 from chest x-ray images using deep transfer learning Med. Image Anal. 65 2020 101794 20 Bassi Pedro R.A.S. Attux Romis A deep convolutional neural network for covid-19 detection using chest x-rays. Research on Biomed. Eng. 2021 1 10 21 Ozturk Tulin Talo Muhammed Yildirim Eylul Azra Baloglu Ulas Baran Yildirim Ozal Acharya U Rajendra Automated detection of covid-19 cases using deep neural networks with x-ray images Comput. Biol. Med. 121 2020 103792 32568675 22 Emtiaz Hussain, Mahmudul Hasan, Md Anisur Rahman, Ickjai Lee, Tasmi Tamanna, and Mohammad Zavid Parvez. Corodet: A deep learning based classification for covid-19 detection using chest x-ray images. Chaos, Solitons & Fractals, 142:110495, 2021. 23 Muhammad EH Chowdhury, Tawsifur Rahman, Amith Khandakar, Rashid Mazhar, Muhammad Abdul Kadir, Zaid Bin Mahbub, Khandakar Reajul Islam, Muhammad Salman Khan, Atif Iqbal, Nasser Al Emadi, et al. Can ai help in screening viral and covid-19 pneumonia? IEEE Access, 8:132665–132676, 2020. 24 Tawsifur Rahman, Amith Khandakar, Yazan Qiblawey, Anas Tahir, Serkan Kiranyaz, Saad Bin Abul Kashem, Mohammad Tariqul Islam, Somaya Al Maadeed, Susu M Zughaier, Muhammad Salman Khan, et al. Exploring the effect of image enhancement techniques on covid-19 detection using chest x-ray images. Computers in biology and medicine, 132:104319, 2021. 25 Ismael Aras M Şengür Abdulkadir Deep learning approaches for covid-19 detection based on chest x-ray images Expert Syst. Appl. 164 2021 114054 26 KC Santosh and Sourodip Ghosh Covid-19 imaging tools: How big data is big? J. Med. Syst. 45 7 2021 1 8 27 Hammoudi Karim Benhabiles Halim Melkemi Mahmoud Dornaika Fadi Arganda-Carreras Ignacio Collard Dominique Scherpereel Arnaud Deep learning on chest x-ray images to detect and evaluate pneumonia cases at the era of covid-19 J. Med. Syst. 45 7 2021 1 10 28 Adhiyaman Manickam, Jianmin Jiang, Yu Zhou, Abhinav Sagar, Rajkumar Soundrapandiyan, and R Dinesh Jackson Samuel. Automated pneumonia detection on chest x-ray images: A deep learning approach with different optimizers and transfer learning architectures. Measurement, page 109953, 2021. 29 El Asnaoui Khalid Design ensemble deep learning model for pneumonia disease classification International Journal of Multimedia Information Retrieval 10 1 2021 55 68 33643764 30 AS Al-Waisy, Mazin Abed Mohammed, Shumoos Al-Fahdawi, MS Maashi, Begonya Garcia-Zapirain, Karrar Hameed Abdulkareem, SA Mostafa, Nallapaneni Manoj Kumar, and Dac Nhuong Le. Covid-deepnet: hybrid multimodal deep learning system for improving covid-19 pneumonia detection in chest x-ray images. Computers, Materials and Continua, 67(2), 2021. 31 Amit Kumar Jaiswal, Prayag Tiwari, Sachin Kumar, Deepak Gupta, Ashish Khanna, and Joel JPC Rodrigues. Identifying pneumonia in chest x-rays: a deep learning approach. Measurement, 145:511–518, 2019. 32 Mohammad Farukh Hashmi, Satyarth Katiyar, Avinash G Keskar, Neeraj Dhanraj Bokde, and Zong Woo Geem. Efficient pneumonia detection in chest xray images using deep transfer learning. Diagnostics, 10(6):417, 2020. 33 Abdullahi Umar Ibrahim, Mehmet Ozsoz, Sertan Serte, Fadi Al-Turjman, and Polycarp Shizawaliyi Yakoi. Pneumonia classification using deep learning from chest x-ray images during covid-19. Cognitive Computation, pages 1–13, 2021. 34 Rachna Jain, Preeti Nagrath, Gaurav Kataria, V Sirish Kaushik, and D Jude Hemanth. Pneumonia detection in chest x-ray images using convolutional neural networks and transfer learning. Measurement, 165:108046, 2020. 35 Vikash Chouhan, Sanjay Kumar Singh, Aditya Khamparia, Deepak Gupta, Prayag Tiwari, Catarina Moreira, Robertas Damaševičius, and Victor Hugo C De Albuquerque. A novel transfer learning based approach for pneumonia detection in chest x-ray images. Applied Sciences, 10(2):559, 2020. 36 Stephen Okeke Sain Mangal Maduh Uchenna Joseph Jeong Do-Un An efficient deep learning approach to pneumonia classification in healthcare Journal of healthcare engineering 2019, 2019. 37 Cha So-Mi Lee Seung-Seok Ko Bonggyun Attention-based transfer learning for efficient pneumonia detection in chest x-ray images Applied Sciences 11 3 2021 1242 38 Daniel S Kermany, Michael Goldbaum, Wenjia Cai, Carolina CS Valentim, Huiying Liang, Sally L Baxter, Alex McKeown, Ge Yang, Xiaokang Wu, Fangbing Yan, et al. Identifying medical diagnoses and treatable diseases by image-based deep learning. Cell, 172(5), 1122–1131, 2018. 39 Szilárd Vajda, Alexandros Karargyris, Stefan Jaeger, KC Santosh, Sema Candemir, Zhiyun Xue, Sameer Antani, and George Thoma. Feature selection for automatic tuberculosis screening in frontal chest radiographs. Journal of medical systems, 42(8):1–11, 2018. 40 Munadi Khairul Muchtar Kahlil Maulina Novi Pradhan Biswajeet Image enhancement for tuberculosis detection using deep learning IEEE Access 8 2020 217897 217907 41 Ayaz Muhammad Shaukat Furqan Raja Gulistan Ensemble learning based automatic detection of tuberculosis in chest x-ray images using hybrid feature descriptors Physical and Engineering Sciences in Medicine 44 1 2021 183 194 33459996 42 Faiz Ahmad Khan, Arman Majidulla, Gamuchirai Tavaziva, Ahsana Nazish, Syed Kumail Abidi, Andrea Benedetti, Dick Menzies, James C Johnston, Aamir Javed Khan, and Saima Saeed. Chest x-ray analysis with deep learning-based software as a triage test for pulmonary tuberculosis: a prospective study of diagnostic accuracy for culture-confirmed disease. The Lancet Digital Health, 2(11), e573–e581, 2020. 43 Tasci Erdal Uluturk Caner Ugur Aybars A voting-based ensemble deep learning method focusing on image augmentation and preprocessing variations for tuberculosis detection. Neural Computing and Applications 2021 1 15 44 Tawsifur Rahman, Amith Khandakar, Muhammad Abdul Kadir, Khandaker Rejaul Islam, Khandakar F Islam, Rashid Mazhar, Tahir Hamid, Mohammad Tariqul Islam, Saad Kashem, Zaid Bin Mahbub, et al. Reliable tuberculosis detection using chest x-ray with deep learning, segmentation and visualization. IEEE Access, 8:191586–191601, 2020. 45 Lakhani Paras Sundaram Baskaran Deep learning at chest radiography: automated classification of pulmonary tuberculosis by using convolutional neural networks Radiology 284 2 2017 574 582 28436741 46 Zhi Zhen Qin, Melissa S Sander, Bishwa Rai, Collins N Titahong, Santat Sudrungrot, Sylvain N Laah, Lal Mani Adhikari, E Jane Carter, Lekha Puri, Andrew J Codlin, et al. Using artificial intelligence to read chest radiographs for tuberculosis detection: A multi-site evaluation of the diagnostic accuracy of three deep learning systems. Scientific reports, 9(1):1–10, 2019. 47 KC Santosh, Szilárd Vajda, Sameer Antani, and George R Thoma. Edge map analysis in chest x-rays for automatic pulmonary abnormality screening. International journal of computer assisted radiology and surgery, 11(9):1637–1646, 2016. 48 KC Santosh and Sameer Antani Automated chest x-ray screening: Can lung region symmetry help detect pulmonary abnormalities? IEEE transactions on medical imaging 37 5 2017 1168 1177 49 Alexandros Karargyris, Jenifer Siegelman, Dimitris Tzortzis, Stefan Jaeger, Sema Candemir, Zhiyun Xue, KC Santosh, Szilárd Vajda, Sameer Antani, Les Folio, et al. Combination of texture and shape features to detect pulmonary abnormalities in digital chest x-rays. International journal of computer assisted radiology and surgery, 11(1):99–106, 2016. 50 Zhi Zhen Qin, Shahriar Ahmed, Mohammad Shahnewaz Sarker, Kishor Paul, Ahammad Shafiq Sikder Adel, Tasneem Naheyan, Rachael Barrett, Sayera Banu, and Jacob Creswell. Tuberculosis detection from chest x-rays for triaging in a high tuberculosis-burden setting: an evaluation of five artificial intelligence algorithms. The Lancet Digital Health, 3(9), e543–e554, 2021.

0

PMC9749798

NO-CC CODE

2022-12-15 23:23:21

no

Inf Sci (N Y). 2022 May 4; 592:389-401

utf-8

Inf Sci (N Y)

10.1016/j.ins.2022.01.062

oa_other