Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
12
33
analyte
stringlengths
3
66
loinc
stringlengths
6
14
unit
stringlengths
1
13
population
stringclasses
2 values
sex
stringclasses
3 values
age_min
int64
5
40
age_max
float64
18
50
measure_type
stringclasses
6 values
sa_value
stringlengths
14
278
general_value
stringlengths
8
67
direction_of_risk
stringclasses
2 values
provenance_tier
stringclasses
3 values
evidence_grade
stringclasses
2 values
source_labels
stringlengths
39
224
source_urls
stringlengths
32
162
source_refs
stringlengths
13
80
source_kinds
stringlengths
9
36
sources_in_registry
stringclasses
6 values
overclaim_guard
stringlengths
90
261
notes
stringlengths
21
203
bmi-overweight-category
BMI
PRAJNA-BMI
kg/m2
south_asian
any
18
null
reference_range
overweight_at=23; obese_at=27.5
overweight_at=25; obese_at=30
high
guideline-endorsed
null
WHO Expert Consultation — appropriate BMI for Asian populations (Lancet 2004)
https://doi.org/10.1016/S0140-6736(03)15268-3
DOI:10.1016/S0140-6736(03)15268-3; PMID:14726171
guideline
true
Lower cutoffs reflect earlier metabolic risk, not a different disease; do not restate as 'South Asians are unhealthy at normal weight'.
Encoded in loinc.yaml as PRAJNA-BMI ref_high 23. Source in evidence registry as who-asian-bmi-2004 (grade A, approved). obese_at corrected 25->27.5 (WHO 2004 Asian high-risk action point; 2026-07-11).
bmi-diabetes-screening-threshold
BMI (diabetes screening trigger)
PRAJNA-BMI
kg/m2
south_asian
any
30
null
screening_threshold
comparator=>=; value=23
comparator=>=; value=25
high
guideline-endorsed
A
Diagnosis and Classification of Diabetes: Standards of Care in Diabetes—2024 | Optimum BMI cut points to screen Asian Americans for type 2 diabetes (Araneta/Kanaya, Diabetes Care 2015)
https://doi.org/10.2337/dc24-S002 | https://pubmed.ncbi.nlm.nih.gov/25665815/
DOI:10.2337/dc24-S002 | PMID:25665815
guideline | primaryStudy
true | true
A screening trigger is not a diagnosis; a positive trigger means 'test glucose', not 'has diabetes'.
SouthAsianScreeningPack.diabetesScreenBMI23 (v2024.2), fires from age 30.
diabetes-screening-age
Diabetes screening start age
1558-6
year
south_asian
any
30
null
screening_age
comparator=>=; value=30
comparator=>=; value=35
null
guideline-endorsed
A
ADA Standards of Care 2024 | Araneta/Kanaya, Diabetes Care 2015
https://doi.org/10.2337/dc24-S002 | https://pubmed.ncbi.nlm.nih.gov/25665815/
DOI:10.2337/dc24-S002 | PMID:25665815
guideline | primaryStudy
true | true
null
South Asians develop dysglycaemia ~a decade earlier; screen earlier.
waist-central-obesity-male
Waist circumference (central obesity)
56115-9
cm
south_asian
male
18
null
screening_threshold
comparator=>=; value=90
comparator=>=; value=102
high
guideline-endorsed
null
IDF consensus worldwide definition of the metabolic syndrome — ethnic-specific waist cutoffs
https://www.idf.org/e-library/consensus-statements/60-idfconsensus-worldwide-definitionof-the-metabolic-syndrome.html
null
consensus
true
null
loinc.yaml 56115-9 ref_high 90 encodes the male SA cut.
waist-central-obesity-female
Waist circumference (central obesity)
56115-9
cm
south_asian
female
18
null
screening_threshold
comparator=>=; value=80
comparator=>=; value=88
high
guideline-endorsed
null
IDF consensus worldwide definition of the metabolic syndrome
https://www.idf.org/e-library/consensus-statements/60-idfconsensus-worldwide-definitionof-the-metabolic-syndrome.html
null
consensus
true
null
SouthAsianScreeningPack.isCentralObesity female cut = 80 cm.
waist-to-height-ratio
Waist-to-height ratio
null
ratio
south_asian
any
18
null
screening_threshold
comparator=>=; value=0.5
comparator=>=; value=0.5
high
study-derived
null
Comparison of obesity indices for CVD risk classification in South Asian adults — The CARRS Study (Patel et al., PLoS One 2017)
https://doi.org/10.1371/journal.pone.0174251
DOI:10.1371/journal.pone.0174251; PMID:28448582
primaryStudy
true
WtHR >=0.5 is a general adiposity rule; the South-Asian-specific claim is that it outperforms BMI here, not a distinct cutoff.
Used in SouthAsianScreeningPack.isCentralObesity as a fallback when waist+height on file. Source in evidence registry as carrs-patel-2017 (review_status: proposed — awaiting clinician sign-off).
hdl-low-female
HDL-C (low threshold)
2085-9
mg/dL
south_asian
female
18
null
risk_flag
comparator=<; value=50
comparator=<; value=40
low
guideline-endorsed
null
IDF metabolic syndrome consensus (2006) | Prevention of ASCVD in South Asians — NLA clinical perspective (2021)
https://www.idf.org/e-library/consensus-statements/60-idfconsensus-worldwide-definitionof-the-metabolic-syndrome.html | https://doi.org/10.1016/j.jacl.2021.03.007
| DOI:10.1016/j.jacl.2021.03.007
consensus | consensus
true | true
null
IDF sex-specific low-HDL cut; women flagged at <50.
hdl-low-male
HDL-C (low threshold)
2085-9
mg/dL
south_asian
male
18
null
risk_flag
comparator=<; value=40
comparator=<; value=40
low
guideline-endorsed
null
IDF metabolic syndrome consensus (2006)
https://www.idf.org/e-library/consensus-statements/60-idfconsensus-worldwide-definitionof-the-metabolic-syndrome.html
null
consensus
true
Male cut equals the general threshold; the SA-specific piece is the female cut and the pattern's weight, not this value.
Included for completeness so the male/female contrast is explicit.
atherogenic-dyslipidemia-pattern
Atherogenic dyslipidemia (TG-high + HDL-low)
2571-8
composite
south_asian
any
18
null
risk_flag
triglycerides_mg_dl=>=150; with_low_hdl=True
null
high
study-derived
null
AHA Scientific Statement — ASCVD in South Asians (Volgman 2018) | Lipid Association of India SA lipid targets (Enas 2013) | Risk factors for early MI in South Asians (INTERHEART, JAMA 2007)
https://doi.org/10.1161/CIR.0000000000000580 | https://pmc.ncbi.nlm.nih.gov/articles/PMC3868060/ | https://doi.org/10.1001/jama.297.3.286
DOI:10.1161/CIR.0000000000000580 | PMCID:PMC3868060 | DOI:10.1001/jama.297.3.286
guideline | consensus | primaryStudy
true | true | true
A normal LDL-C does not rule out risk when TG is high and HDL low; do not present LDL-C as the sole lipid target for SA users.
The SA-characteristic pattern (small-dense LDL, high ApoB/ApoA-I) that drives risk at normal LDL-C. Component values match general thresholds; the differentiation is the emphasis.
apob-preferred-marker
ApoB (preferred atherogenic marker)
1884-6
mg/dL
south_asian
any
18
null
reference_range
low=60; high=90
low=60; high=90
high
study-derived
null
NLA — Prevention of ASCVD in South Asians (2021) | AHA Volgman 2018
https://doi.org/10.1016/j.jacl.2021.03.007 | https://doi.org/10.1161/CIR.0000000000000580
DOI:10.1016/j.jacl.2021.03.007 | DOI:10.1161/CIR.0000000000000580
consensus | guideline
true | true
The range is not SA-specific; the SA-specific claim is preference of ApoB over LDL-C, not a different cutoff.
SouthAsianScreeningPack.apobPreferred nudges ApoB when a lipid panel exists but ApoB doesn't.
lpa-elevated-and-test-once
Lp(a)
10835-7
mg/dL
south_asian
any
18
null
risk_flag
elevated_at=50; test_frequency=once_in_adulthood
elevated_at=50; test_frequency=once_in_adulthood
high
guideline-endorsed
A
NLA 2024 Lp(a) focused update / AHA 2024 Lp(a) toolkit | 2018 AHA/ACC/Multi-Society Blood Cholesterol Guideline
https://professional.heart.org/en/-/media/PHD-Files/Education/AHA2024_Lpa_Toolkit_Healthcare_Professionals.pdf | https://doi.org/10.1161/CIR.0000000000000625
| DOI:10.1161/CIR.0000000000000625
guideline | guideline
true | true
Do NOT claim South Asians universally run high Lp(a): MASALA (Huffman 2019) found no Lp(a)-atherosclerosis signal within the cohort. The rationale is 'genetically fixed + test-once is cheap', plus higher elevated-prevalence, not a proven SA outcome association.
The 'test once' guidance is guideline-endorsed and not SA-specific; it appears here because it is under-done in this population. loinc.yaml Lp(a) mass ref_high 30 mg/dL, molar 43583-7 ref_high 75 nmol/L.
non-hdl-aggressive-target
Non-HDL-C (treatment target)
PRAJNA-NONHDL
mg/dL
south_asian
any
18
null
reference_range
high=130; note=LAI recommends more aggressive non-HDL targets by risk stratum
high=130
high
study-derived
null
Lipid Association of India SA lipid targets (Enas 2013)
https://pmc.ncbi.nlm.nih.gov/articles/PMC3868060/
PMCID:PMC3868060
consensus
true
LAI targets are risk-stratified; do not present a single non-HDL cutoff as applying to all SA users.
Default non-HDL ref_high 130 from loinc.yaml; SA-specific aggressiveness is stratum-dependent — represented qualitatively pending clinician review.
hba1c-under-read-caveat
HbA1c (interpretation caveat)
4548-4
%
south_asian
any
18
null
interpretation_caveat
confirm_when=5.7-6.4; confirm_with=fasting glucose or OGTT
reference_range_low=4.0; reference_range_high=5.6
high
study-derived
null
Isolated HbA1c identifies a different T2D subgroup vs glucose in Asian Indians — CARRS & MASALA (Gujral 2019) | ADA Standards of Care 2024
https://pubmed.ncbi.nlm.nih.gov/31150721/ | https://doi.org/10.2337/dc24-S002
PMID:31150721 | DOI:10.2337/dc24-S002
primaryStudy | guideline
true | true
HbA1c is not universally 'wrong' in SA users; the claim is that it can under- or over-read (iron deficiency, glycation rate), so borderline values warrant glucose confirmation.
The reference range is not changed; this is an interpretation flag on borderline values.
cac-baseline-age
Coronary artery calcium — baseline discussion age
PRAJNA-CAC
year
south_asian
any
35
null
screening_age
comparator=>=; value=35
comparator=>=; value=40
null
guideline-endorsed
null
AHA Scientific Statement — ASCVD in South Asians (Volgman 2018)
https://doi.org/10.1161/CIR.0000000000000580
DOI:10.1161/CIR.0000000000000580
guideline
true
null
SA develop CAD ~10 years earlier; earlier baseline CAC discussion. In-app, CAC nudges tie to PREVENT 5-<20% band.
vitamin-d-screening-emphasis
25-OH Vitamin D (screening emphasis)
1989-3
ng/mL
south_asian
any
18
null
reference_range
low=30; high=100
low=30; high=100
low
study-derived
null
Endocrine Society — Evaluation, Treatment, and Prevention of Vitamin D Deficiency (2011)
https://doi.org/10.1210/jc.2011-0385
DOI:10.1210/jc.2011-0385
guideline
true
The reference range is not SA-specific; the SA-relevant point is high deficiency prevalence, warranting screening — not a different normal range.
SouthAsianScreeningPack.vitaminDScreen prompts if not measured within a year.
fasting-insulin-optional-marker
Fasting insulin (optional early marker)
1554-5
uIU/mL
south_asian
any
30
null
reference_range
low=2; high=15
low=2; high=15
high
study-derived
B
MASALA cardiometabolic profile / HOMA-IR vs MESA (Kanaya 2014)
https://masalastudy.org/publications/
null
primaryStudy
true
Explicitly NOT guideline-backed; present as an optional/exploratory marker, never as recommended screening.
In-app copy flags this as not guideline-backed. Good example of an honest study-derived row.
egfr-race-adjustment-deprecated
eGFR (race/ancestry adjustment)
33914-3
mL/min/1.73m2
general
any
18
null
deprecated_adjustment
adjust=False; guidance=do NOT apply a race/ancestry coefficient
low=60; high=120
low
contested-deprecated
null
New Creatinine- and Cystatin C–Based Equations without Race (CKD-EPI 2021, Inker et al., NEJM)
https://doi.org/10.1056/NEJMoa2102953
DOI:10.1056/NEJMoa2102953; PMID:34554658
primaryStudy
true
Do NOT ship an ancestry-adjusted eGFR. The 2021 refit removed the race term as a social-not-biological construct that delayed care. This row exists to instruct consumers to NOT stratify eGFR by ancestry.
The credibility showcase row: an adjustment the field is moving away from, included precisely to warn against it. Consistent with loinc.yaml keeping eGFR at real LOINC 33914-3 with no ancestry term.
prevent-risk-underestimate-caveat
10-year ASCVD risk (PREVENT)
PRAJNA-PREVENT
%
south_asian
any
30
null
interpretation_caveat
adjustment=none available; guidance=treat estimate as a floor; it underestimates SA risk
low=0; high=5
high
study-derived
null
Development and Validation of the AHA PREVENT Equations (Circulation 2024)
https://doi.org/10.1161/CIRCULATIONAHA.123.067626
DOI:10.1161/CIRCULATIONAHA.123.067626
guideline
true
There is no validated SA multiplier in PREVENT; do not invent one. State the limitation ('treat as a floor'), do not apply a made-up adjustment.
Contrast with the eGFR row: here we flag under-estimation but still refuse to fabricate an adjustment. loinc.yaml PRAJNA-PREVENT plain_language already says 'treat it as a floor'.
triglycerides-risk-enhancer
Triglycerides (risk-enhancer threshold)
2571-8
mg/dL
south_asian
any
18
null
risk_flag
comparator=>=; value=175
comparator=>=; value=175
high
guideline-endorsed
null
2018 AHA/ACC/Multi-Society Blood Cholesterol Guideline
https://doi.org/10.1161/CIR.0000000000000625
DOI:10.1161/CIR.0000000000000625
guideline
true
The 175 mg/dL cut is not SA-specific; the SA-relevant point is that atherogenic dyslipidemia (high TG + low HDL) reaches it often even at a normal LDL-C.
Completes the atherogenic-dyslipidemia composite inputs. TG LOINC 2571-8.
hscrp-risk-enhancer
hs-CRP (risk-enhancer threshold)
30522-7
mg/L
south_asian
any
18
null
risk_flag
comparator=>=; value=2
comparator=>=; value=2
high
guideline-endorsed
null
2018 AHA/ACC/Multi-Society Blood Cholesterol Guideline
https://doi.org/10.1161/CIR.0000000000000625
DOI:10.1161/CIR.0000000000000625
guideline
true
The >=2 mg/L cut is not SA-specific; SA users more often exceed it due to higher visceral adiposity at the same BMI.
hs-CRP LOINC 30522-7.
sa-ancestry-ascvd-risk-enhancer
South Asian ancestry (ASCVD risk-enhancing factor)
null
null
south_asian
any
18
null
interpretation_caveat
ancestry_is_ascvd_risk_enhancer=True; guidance=The 2018 AHA/ACC guideline lists South Asian ancestry as a formal ASCVD risk-enhancing factor; use it to move borderline/intermediate-risk patients toward the treat decision.
null
high
guideline-endorsed
null
2018 AHA/ACC/Multi-Society Blood Cholesterol Guideline
https://doi.org/10.1161/CIR.0000000000000625
DOI:10.1161/CIR.0000000000000625
guideline
true
A risk-enhancer is not a risk multiplier; it does not by itself diagnose disease or mandate a statin — it shifts the borderline decision.
Explicitly named in the 2018 AHA/ACC Cholesterol Guideline risk-enhancer list.
sa-very-high-risk-lipid-targets
LDL-C / non-HDL-C (SA very-high-risk targets)
PRAJNA-NONHDL
mg/dL
south_asian
any
18
null
interpretation_caveat
ldl_c_target_very_high_risk=<70; non_hdl_target_very_high_risk=<100; guidance=LAI recommends more aggressive LDL-C/non-HDL targets for SA very-high-risk patients.
ldl_c_target_very_high_risk=<70; non_hdl_target_very_high_risk=<130
high
study-derived
null
Lipid Association of India SA lipid targets (Enas 2013) | NLA — Prevention of ASCVD in South Asians (2021)
https://pmc.ncbi.nlm.nih.gov/articles/PMC3868060/ | https://doi.org/10.1016/j.jacl.2021.03.007
PMCID:PMC3868060 | DOI:10.1016/j.jacl.2021.03.007
consensus | consensus
true | true
LAI targets are risk-stratified; do not present a single cutoff as applying to all SA users. LDL-C <70 matches general very-high-risk guidance; the SA-specific piece is the tighter non-HDL target.
Tightens the non-HDL-130 reference row for the very-high-risk stratum. review pending clinician sign-off.
bmi-masld-screening-threshold
BMI (MASLD / fatty-liver screening trigger)
PRAJNA-BMI
kg/m2
south_asian
any
18
null
screening_threshold
comparator=>=; value=23
comparator=>=; value=25
high
guideline-endorsed
null
Multi-society Delphi consensus statement on new fatty liver disease nomenclature (MASLD) (Rinella et al., 2023) | Ectopic fat depots and coronary artery calcium in South Asians vs other ethnic groups (Garg et al., JAHA 2016)
https://doi.org/10.1097/HEP.0000000000000520 | https://pubmed.ncbi.nlm.nih.gov/27856485/
DOI:10.1097/HEP.0000000000000520 | PMID:27856485
consensus | primaryStudy
false | false
A lower screening BMI reflects earlier ectopic-fat accumulation, not a different disease; MASLD is diagnosed by imaging/enzymes, not by BMI alone.
SA carry the highest intrahepatic fat of MESA ethnic groups at lower BMI (Garg 2016). MASLD nomenclature (2023) uses BMI >=23 for Asian populations.
score2-south-asian-multiplier
SCORE2 10-year CVD risk (South Asian multiplier)
PRAJNA-PREVENT
%
south_asian
any
40
null
interpretation_caveat
multiplier_indian_bangladeshi=1.3; multiplier_pakistani=1.7; guidance=Multiply the SCORE2 estimate by the ancestry-specific factor before applying risk thresholds.
multiplier=1.0
high
guideline-endorsed
null
2021 ESC Guidelines on cardiovascular disease prevention in clinical practice
https://doi.org/10.1093/eurheartj/ehab484
DOI:10.1093/eurheartj/ehab484
guideline
false
The multiplier applies to SCORE2 specifically (derived from QRISK3 data); do not transplant it onto PREVENT or the Pooled Cohort Equations, which have no validated SA multiplier.
The only major guideline with a numeric SA correction factor (2021 ESC, Class IIa LOE B). Companion to the PREVENT 'treat as a floor' caveat row.
cac-south-asian-percentiles
Coronary artery calcium (South Asian age/sex percentiles)
PRAJNA-CAC
Agatston
south_asian
any
40
null
reference_range
prevalence_cac_gt0={'age50': {'men': 0.45, 'women': 0.2}, 'age60': {'men': 0.7, 'women': 0.4}, 'age70': {'men': 0.9, 'women': 0.7}}; p75_age60={'men': 186, 'women': 26}; guidance=Use age/sex-specific percentiles, not the raw 100 cutoff, for younger SA patients.
null
high
study-derived
null
South Asian coronary artery calcium age/sex percentiles — MASALA + DILWALE (Tasdighi et al., JACC Advances 2025)
https://pubmed.ncbi.nlm.nih.gov/40402122/
PMID:40402122
primaryStudy
false
These are cohort percentiles (MASALA+DILWALE), not a guideline threshold; CAC=0 remains reassuring at age 50 (55% of men, 80% of women) but less so at 70.
First SA-specific CAC percentile table to exist (n=2,743, ASCVD-free). review pending clinician sign-off.
remnant-cholesterol-sa-marker
Remnant cholesterol (optional atherogenic marker)
PRAJNA-REMNANT
mg/dL
south_asian
any
18
null
risk_flag
comparator=>=; value=33
null
high
study-derived
null
Remnant cholesterol and incident coronary artery calcium in South Asians — MASALA (Manghis et al., J Clin Lipidol 2025)
https://pubmed.ncbi.nlm.nih.gov/41168049/
PMID:41168049
primaryStudy
false
Explicitly not a guideline cutoff; the top MASALA quartile (>=~33 mg/dL) was associated with incident CAC, not CAC progression. Present as exploratory. Calculate as total cholesterol - HDL-C - LDL-C.
MASALA: Q4 vs Q1 OR 2.57 for incident CAC over 5y, independent of LDL-C. review pending clinician sign-off.
apol1-not-south-asian
APOL1 kidney-risk variants (ancestry misattribution)
null
null
general
any
18
null
deprecated_adjustment
adjust=False; guidance=Do NOT apply APOL1 G1/G2 kidney-risk stratification to South Asians — these variants are high-frequency only in West-African ancestry (<1% in non-Africans).
null
null
contested-deprecated
null
APOL1 kidney-risk variants are West-African ancestry–specific (review)
https://pmc.ncbi.nlm.nih.gov/articles/PMC5495568/
PMCID:PMC5495568
primaryStudy
false
This row exists to instruct consumers NOT to use APOL1 in a South Asian risk model; including it would be a false-ancestry-attribution error.
Credibility-showcase 'do-not-apply' row, in the spirit of the deprecated eGFR row.
aldh2-adh1b-not-south-asian
ALDH2*2 / ADH1B*2 alcohol-flush variants (ancestry misattribution)
null
null
general
any
18
null
deprecated_adjustment
adjust=False; guidance=Do NOT surface alcohol-flush / fast-metabolizer warnings for South Asians based on ALDH2*2 or ADH1B*2 — these are predominantly East Asian (low single-digit % in SA).
null
null
contested-deprecated
null
ALDH2*2 / ADH1B*2 alcohol-metabolism variants are predominantly East Asian (review)
https://pmc.ncbi.nlm.nih.gov/articles/PMC10594868/
PMCID:PMC10594868
primaryStudy
false
This row exists to instruct consumers NOT to apply East-Asian alcohol-metabolism genetics to South Asians.
Credibility-showcase 'do-not-apply' row. South Asians more often carry the slower ADH1B*1.
vitamin-b12-screening-emphasis
Vitamin B12 (screening emphasis)
2132-9
pg/mL
south_asian
any
18
null
reference_range
low=200; high=900
low=200; high=900
low
study-derived
null
Hyperhomocysteinemia & cobalamin (B12) deficiency in Asian Indians (Refsum et al., Am J Clin Nutr 2001) | Vitamin B12 deficiency & hyperhomocysteinemia in rural and urban Indians (Yajnik et al., JAPI 2006)
https://doi.org/10.1093/ajcn/74.2.233 | https://pubmed.ncbi.nlm.nih.gov/17214273/
DOI:10.1093/ajcn/74.2.233; PMID:11470726 | PMID:17214273
primaryStudy | primaryStudy
false | false
The reference range is not SA-specific; the SA-relevant point is high deficiency prevalence in vegetarian South Asians, warranting screening — not a different normal range.
Exact parallel to the vitamin-D screening-emphasis row. B12 LOINC 2132-9.
thalassemia-carrier-screening
Thalassemia / hemoglobinopathy carrier screening
30428-7
fL
south_asian
any
18
null
interpretation_caveat
mcv_below=80; confirm_with=hemoglobin electrophoresis / HbA2; guidance=Offer carrier screening in South Asian ancestry (higher β-thalassemia, HbE, HbD-Punjab prevalence), especially preconception; low MCV with normal iron warrants electrophoresis.
null
low
guideline-endorsed
null
ACOG Practice Bulletin No. 78: Hemoglobinopathies in Pregnancy (Obstet Gynecol 2007)
https://doi.org/10.1097/00006250-200701000-00055
DOI:10.1097/00006250-200701000-00055; PMID:17197616
guideline
false
A carrier trait is not a disease; screening informs reproductive counseling and explains a spuriously altered HbA1c — it is not a diagnosis.
Also the mechanism behind the HbA1c interpretation caveat. MCV LOINC 30428-7.
normal-alt-masld-caveat
ALT (MASLD interpretation caveat)
1742-6
U/L
south_asian
any
18
null
interpretation_caveat
guidance=A normal ALT does not exclude MASLD; in SA with metabolic risk, use imaging (ultrasound / hepatic-fat) rather than relying on enzymes.
null
high
study-derived
null
Proportion of NAFLD/MASLD patients with normal ALT: systematic review & meta-analysis (BMC Gastroenterol 2020)
https://doi.org/10.1186/s12876-020-1165-z
DOI:10.1186/s12876-020-1165-z; PMID:31937252; PMCID:PMC6961232
primaryStudy
false
ALT is not universally unreliable; the point is that a normal value is insufficient to rule out fatty liver in a high-metabolic-risk SA profile.
Pairs with the BMI-23 MASLD screening trigger. ALT LOINC 1742-6.
benign-ethnic-neutropenia
WBC / neutrophils (benign ethnic neutropenia)
6690-2
10*3/uL
south_asian
any
18
null
interpretation_caveat
guidance=A mildly low WBC / absolute neutrophil count can be a normal ancestry-linked variant; do not over-investigate isolated mild neutropenia without other features.
null
low
study-derived
null
Benign ethnic neutropenia (Atallah-Yunes et al., Blood Rev 2019)
https://doi.org/10.1016/j.blre.2019.06.003
DOI:10.1016/j.blre.2019.06.003; PMID:31255364; PMCID:PMC6702066
primaryStudy
false
Benign ethnic neutropenia is best-established for African ancestry; the South Asian evidence base is thinner — flag it as a possibility, not a rule.
Included with an explicit evidence-strength caveat. WBC LOINC 6690-2.
slco1b1-statin-pgx
SLCO1B1 rs4149056 (statin pharmacogenomics)
null
null
south_asian
any
18
null
interpretation_caveat
variant=SLCO1B1 rs4149056 (c.521T>C); action=In decreased-function carriers, limit simvastatin dose and prefer rosuvastatin or pravastatin.; frequency_note=C-allele frequency in SA is broadly comparable to White populations (~15-20%).
null
null
guideline-endorsed
null
CPIC Guideline for SLCO1B1/ABCG2/CYP2C9 & statin-associated musculoskeletal symptoms (Clin Pharmacol Ther 2022)
https://doi.org/10.1002/cpt.2557
DOI:10.1002/cpt.2557; PMID:35152405; PMCID:PMC9035072
guideline
false
The recommendation is genotype-guided statin selection, not SA-specific dosing; there is no SA-specific dosing table.
CPIC 2022 statin/SLCO1B1 guideline. Default move in SA with statin myalgia: consider genotyping and switch off simvastatin before declaring intolerance.
pediatric-bmi-iap
Pediatric BMI (IAP South Asian cut-offs)
PRAJNA-BMI
percentile
south_asian
any
5
18
reference_range
overweight_percentile=75; obese_percentile=95; adult_equivalent=overweight = adult BMI 23, obese = adult BMI 27 at age 18
overweight_percentile=85; obese_percentile=95
high
guideline-endorsed
null
Revised IAP 2015 growth charts / BMI cut-offs for 5-18y Indian children (Khadilkar et al., Indian J Endocrinol Metab 2015)
https://doi.org/10.4103/2230-8210.159028
DOI:10.4103/2230-8210.159028; PMID:26180761; PMCID:PMC4481652
guideline
false
These are cardiometabolic-risk-adjusted centiles on IAP 2015 charts; under-5 growth still follows WHO international standards regardless of ethnicity.
Plot SA children on CDC and IAP charts; when CDC says normal but IAP says at-risk, treat as at-risk. IAP 2015 (Khadilkar).
lean-diabetes-phenotype
Diabetes risk at low BMI (lean-diabetes phenotype)
PRAJNA-BMI
kg/m2
south_asian
any
18
null
interpretation_caveat
guidance=Do not gate diabetes-risk screening on BMI>=25 — T2D at BMI<25 is 2-4x more common in Asian Indians. Screen from BMI 23 and use waist-to-height >=0.5.
null
high
study-derived
null
Young-onset diabetes in Asian Indians linked to lower beta-cell function (INSPIRED; Diabetologia 2022)
https://doi.org/10.1007/s00125-022-05671-z
DOI:10.1007/s00125-022-05671-z; PMID:35247066; PMCID:PMC9076730
primaryStudy
false
Lean diabetes reflects reduced β-cell function / 'thin-fat' body composition, not the absence of metabolic risk; a normal BMI is not reassurance.
INSPIRED study. Reinforces the BMI-23 diabetes-screening trigger.
gdm-early-screening
Gestational diabetes (early screening in SA)
null
null
south_asian
female
18
50
interpretation_caveat
early_screen=first-trimester screening for high-risk groups including South Asian ancestry; iadpsg_75g_ogtt_cutoffs=FPG>=92, 1h>=180, 2h>=153 mg/dL (numeric cutoffs unchanged); guidance=SA women develop GDM at lower BMI and earlier in pregnancy; screen earlier, same thresholds.
screen=24-28 weeks
high
guideline-endorsed
null
IADPSG recommendations on diagnosis & classification of hyperglycemia in pregnancy (Diabetes Care 2010) | Diagnosis and Classification of Diabetes: Standards of Care in Diabetes—2024
https://doi.org/10.2337/dc09-1848 | https://doi.org/10.2337/dc24-S002
DOI:10.2337/dc09-1848; PMID:20190296; PMCID:PMC2827530 | DOI:10.2337/dc24-S002
consensus | guideline
false | true
The diagnostic thresholds are unchanged; the SA-specific action is earlier screening, not a different cutoff.
IADPSG 2010; ADA 2024 §15 (Pregnancy); FIGO 2015.
menopause-age-sa
Age at natural menopause (South Asian women)
null
year
south_asian
female
40
null
reference_range
median_age=49; early_or_premature_le45_pct=0.3; guidance=~30% of US SA women experience early/premature menopause (<=45y) vs <7% in other US groups.
median_age=50-51 (US White/Black)
null
study-derived
null
Menopausal experience (hot flashes, urinary incontinence, mood) among South Asian American women — MASALA (Lyu et al., Climacteric 2025)
https://doi.org/10.1080/13697137.2025.2480584
DOI:10.1080/13697137.2025.2480584; PMID:40177961; PMCID:PMC12353937
primaryStudy
false
Cohort estimate (n=405); earlier menopause raises cardiometabolic risk earlier but is not itself a disease.
Lyu, Climacteric 2025. New domain: women's health.
parity-cardiometabolic-sa
Parity (cardiometabolic effect, SA women)
null
null
south_asian
female
18
null
interpretation_caveat
high_parity=>=2 live births; effect_over_5y_vs_0_1=SBP +8.9 mmHg, DBP +4.1 mmHg, LDL-C +8.9 mg/dL
null
high
study-derived
null
Association of Parity With ASCVD Risk Factors Among South Asian Women — MASALA (Mehta et al., JACC Advances 2026)
https://doi.org/10.1016/j.jacadv.2026.102647
DOI:10.1016/j.jacadv.2026.102647
primaryStudy
false
Association from MASALA (n~415); parity is a risk-context flag, not a screening threshold.
Mehta, JACC Advances 2026. Surface on women's cardiometabolic panels.
mthfr-not-sa-variant
MTHFR C677T as an 'SA variant' (misattribution)
null
null
general
any
18
null
deprecated_adjustment
adjust=False; guidance=Do NOT treat MTHFR C677T as a South Asian homocysteine variant — it is not enriched in SA. Elevated homocysteine here is driven by dietary (vegetarian) B12 deficiency; treat the B12, not the SNP.
null
null
contested-deprecated
null
Hyperhomocysteinemia & cobalamin (B12) deficiency in Asian Indians (Refsum et al., Am J Clin Nutr 2001)
https://doi.org/10.1093/ajcn/74.2.233
DOI:10.1093/ajcn/74.2.233; PMID:11470726
primaryStudy
false
Credibility-showcase 'do-not-apply' row: the actionable target is B12 status, not MTHFR genotyping.
Driver evidence: Refsum 2001 (B12 deficiency → hyperhomocysteinemia in Indians).
apob-apoa1-ratio
ApoB/ApoA-I ratio (lead SA atherogenic metric)
PRAJNA-APOBA1
ratio
south_asian
any
18
null
interpretation_caveat
note=The ApoB/ApoA-I ratio was the strongest single modifiable MI predictor in South Asians (INTERHEART; population-attributable risk ~47%).; guidance=Where available, weight the ApoB/ApoA-I ratio over LDL-C alone for SA risk discussions.
null
high
study-derived
null
Risk factors for early MI in South Asians (INTERHEART, JAMA 2007)
https://doi.org/10.1001/jama.297.3.286
DOI:10.1001/jama.297.3.286
primaryStudy
true
No SA-specific numeric ratio cutoff is defined here; the claim is the ratio's predictive weight, not a threshold.
Reuses the INTERHEART citation already in the dataset. PRAJNA-APOBA1.
hypothyroidism-screening-emphasis
TSH / hypothyroidism (screening emphasis)
3016-3
mIU/L
south_asian
any
18
null
reference_range
low=0.4; high=4.0; guidance=Reference range unchanged; the SA-relevant point is high hypothyroidism prevalence in Indian adults, supporting a lower threshold to check TSH.
low=0.4; high=4.0
high
study-derived
null
Prevalence of hypothyroidism in adults: epidemiological study in eight cities of India (Unnikrishnan et al., Indian J Endocrinol Metab 2013)
https://doi.org/10.4103/2230-8210.113755
DOI:10.4103/2230-8210.113755; PMID:23961480; PMCID:PMC3743364
primaryStudy
false
Not a redefined range — a prevalence-driven screening emphasis; do not present a different TSH normal range for SA.
Unnikrishnan, Indian thyroid epidemiology. Soft/borderline candidate. TSH LOINC 3016-3.

Open Masala — Ancestry-Adjusted Health Reference Ranges

The open, machine-readable, cited dataset of South Asian biomarker reference ranges and screening thresholds.

South Asians develop cardiovascular disease and type-2 diabetes earlier, at lower BMI, through different metabolic pathways. The screening thresholds that reflect this — WHO Asian BMI cutoffs, IDF waist limits, earlier diabetes and coronary-calcium screening, Lp(a) prompts — exist, but they're scattered across guideline PDFs and hundreds of papers. Nobody has ever assembled them into one structured, cited, downloadable table. This is that table.

⚠️ v0 / early release. A curation and synthesis of published guidelines and studies — not a primary-data release, and not yet clinician-reviewed for clinical use. Not a medical device or clinical advice. See Disclaimer.

What this is (and isn't)

  • Is: a synthesis of published, citable thresholds into a machine-readable table — each row LOINC-coded, evidence-graded, provenance-tiered, and linked to its source.
  • Isn't: raw cohort data. The studies these numbers come from (MASALA, GenomeAsia, UK Biobank) are access-controlled for good reasons; we don't re-host them. We encode the published thresholds they produced. This is a curation/synthesis contribution.

Dataset at a glance

  • 41 rows (v0.1) spanning cardiometabolic (BMI, waist, HDL, triglycerides / atherogenic pattern, ApoB, ApoB/ApoA-I ratio, Lp(a), non-HDL, remnant cholesterol, hs-CRP, HbA1c, fasting insulin, SCORE2 SA multiplier, CAC screening + South-Asian CAC percentiles, 10-year risk, eGFR), plus women's health (GDM early screening, menopause, parity), pediatric (IAP BMI), hepatic (BMI-23 MASLD, normal-ALT caveat), nutrition (vitamin D, vitamin B12), hematology (thalassemia carrier screening, benign ethnic neutropenia), endocrine (hypothyroidism), and pharmacogenomics (SLCO1B1 statins).
  • Each South Asian value is paired with its general-population counterpart, so the ancestry delta is visible in a single row.
  • Provenance tiers: 18 guideline-endorsed · 19 study-derived · 4 contested-deprecated (do-not-apply showcases: eGFR race adjustment, APOL1, ALDH2/ADH1B, MTHFR-as-SA-variant).
  • Citation completeness: 100% — all 35 unique sources carry a resolvable identifier (DOI / PMID / PMCID / URL).
  • ⚠️ 20 rows added in v0.1 are review_status: proposed — pending clinician sign-off; full provenance + overclaim guards, not yet promoted to approved.

Files

Path Format For
data/ancestry-reference-ranges.v0.csv CSV (powers the viewer above) Tabular / ML use.
data/ancestry-reference-ranges.v0.json JSON (canonical) Full nested sources + overclaim guards.
data/ancestry-reference-ranges.fhir.json FHIR R4 Bundle Health systems — ObservationDefinition.qualifiedInterval per row.

The provenance model (why you can trust a row)

Every row carries a provenance_tier — the single most important field:

Tier Meaning
guideline-endorsed A named body recommends this exact cutoff for this population (WHO Asian BMI, IDF waist, ADA SA screening). Highest confidence.
study-derived A real, cited effect estimate not yet codified into a guideline cutoff (e.g. the HbA1c under-read). Use with care; not settled practice.
contested-deprecated An adjustment the field is moving away from (the eGFR race coefficient, retired 2021). Included precisely so you can tell you must not apply it.

Every row also carries an overclaim_guard: the claim you must not make from it. If you're training or building on this data, honor those guards — they're designed to stop confidently-wrong ancestry statements.

Columns

Column Meaning
id Stable row key.
analyte Biomarker / measure name.
loinc LOINC code (synthetic PRAJNA-* for computed metrics).
unit Unit (UCUM-preferred).
population south_asian or general — the stratification dimension.
sex, age_min, age_max Strata the row applies to.
measure_type reference_range / screening_threshold / screening_age / interpretation_caveat / risk_flag / deprecated_adjustment.
sa_value, general_value The South Asian value and the general-population value it replaces.
direction_of_risk Which way is concerning (high / low / n/a).
provenance_tier, evidence_grade Credibility fields.
source_labels, source_urls, source_refs, source_kinds Citations (pipe-separated where multiple).
overclaim_guard The claim you must not make from the row.
notes Anything else needed.

Usage

from datasets import load_dataset

ds = load_dataset("masalahealthco/open-masala", split="reference_ranges")
row = next(r for r in ds if r["id"] == "bmi-diabetes-screening-threshold")
print(row["sa_value"], "vs general", row["general_value"])
# comparator=>=; value=23  vs general  comparator=>=; value=25

The JSON (with nested sources) and the FHIR Bundle are in data/ for consumers that need the full structure.

Citation

See CITATION.cff. Please also cite the primary sources named in each row — the original authors did the underlying work. A DOI-minted release (Zenodo) is on the roadmap.

License

CC-BY-4.0 — use freely, with attribution and the per-row citations preserved.

Disclaimer

Open Masala is a reference dataset for research and product development. It is not a medical device, not clinical advice, and not a substitute for a clinician's judgment. Reference ranges vary by lab and method; screening decisions require individual clinical context. This is a v0 release and has not yet undergone clinical sign-off.

Downloads last month
48