Cardiometabolic Risk and Diagnostic-Laboratory Reference-Range Disagreement in 794,811 Indian Insurance Applicants

This study of nearly 800,000 Indian insurance applicants reveals a high prevalence of cardiometabolic risk factors while highlighting that significant heterogeneity in laboratory reference ranges across thousands of diagnostic centers materially distorts patient classification and disease tracking, thereby underscoring the urgent need for national standardization of laboratory reporting in India.

The Gist

Imagine a massive library containing nearly 800,000 health check-up reports from people in India who are applying for life insurance. This isn't a random group of people; they are mostly working-age, living in cities, and buying insurance (which means they are likely employed and have a certain income level).

The author of this paper, Saif Lakhani, acted like a detective sifting through these digital files to find two big stories: one about the health of these people, and a second, surprising story about confusion in the rules used to read the test results.

Here is the breakdown in simple terms:

Story 1: The Health Picture (The "What")

The author looked at the health data and found that this group of people has a very specific "South Asian" health profile, which is well-known in medical science but now confirmed with huge numbers.

• The "Silent" Risks: A huge chunk of these people have high blood pressure and unhealthy blood fats (cholesterol), even if they feel fine.
  • Blood Pressure: If you use the strict, modern rules (AHA 2017), nearly 61% of these people have high blood pressure. If you use the older, looser rules, only about 18% do. The paper notes that India hasn't officially switched to the strict rules yet, so this number depends entirely on which "rulebook" you pick.
  • Blood Fats: About 42% of the people had at least one type of bad blood fat. The most common issue wasn't high "bad" cholesterol (LDL), but rather low "good" cholesterol (HDL) and high triglycerides (fats in the blood). This is a classic pattern seen in South Asians that puts them at higher risk for heart issues.
• The BMI Twist: The average person in this group is slightly overweight. However, the paper points out that for South Asians, the standard "overweight" line might be too high. They might get sick with heart problems at a lower weight than Europeans do, but the current global rules might miss that risk.

Story 2: The "Rulebook" Chaos (The "Why it's messy")

This is the most surprising part of the paper. The author realized that while the numbers on the lab reports were real, the rules for deciding if those numbers are "bad" or "good" were all over the place.

Imagine you are playing a game where the goal is to jump over a wall.

• Lab A says the wall is 110 cm high. If you jump 111 cm, they say, "Great job! You're normal."
• Lab B says the wall is 100 cm high. If you jump 111 cm, they say, "Oh no! You jumped too high! You have a problem."

The person jumped the exact same height, but one lab says they are healthy, and the other says they are sick.

The Findings on Confusion:
The paper looked at 33,000 different labs. They found that for common tests like blood sugar and LDL cholesterol, the "cut-off" lines were so different that if you took the exact same blood sample to two different labs, there was a 50/50 chance they would give you opposite diagnoses (one saying "normal," the other "abnormal").

• The "Confused" Tests: Blood sugar, LDL cholesterol, and liver enzymes (SGOT/SGPT) had the most confusion.
• The "Agreed" Tests: Total cholesterol and triglycerides were the only ones where most labs agreed on the rules.

Why This Matters (According to the Paper)

The author argues that this "Rulebook Chaos" is a big problem for three reasons:

1. For Doctors: If a doctor sees a "high" result, they shouldn't panic immediately. They need to know which lab did the test and what their specific rules were. A single "abnormal" flag might just be a difference in the rulebook, not a real disease.
2. For Insurance: Insurance companies use these tests to decide who is a "risky" customer. If the rules are inconsistent, two people with the exact same biology could get different insurance rates just because they went to different labs.
3. For Public Health: If we try to count how many people in India have diabetes or high cholesterol, our numbers will be messy. We can't get an accurate national count until all the labs agree on the same "wall height."

The Bottom Line

The paper confirms that working-age Indians in cities have a high risk of heart-related issues, specifically a pattern of low "good" cholesterol and high triglycerides.

However, the biggest takeaway is a warning: We cannot trust the "Yes/No" labels on lab reports until India standardizes its rules. Right now, the same number can mean "healthy" in one city and "sick" in another, creating a lot of confusion in how we track and treat disease.

Note: The paper explicitly states it does not make claims about individual patients, does not change insurance policies, and does not offer new medical treatments. It simply reports what the data shows and highlights the inconsistency in how labs report results.

Technical Summary

Technical Summary: Cardiometabolic Risk and Diagnostic-Laboratory Reference-Range Disagreement in 794,811 Indian Insurance Applicants

Problem Statement
The study addresses two primary challenges in assessing cardiometabolic health within the Indian population. First, there is a demographic gap in existing literature: national surveys like NFHS-5 undersample the working-age, urban, and employed population, a segment critical for understanding the specific cardiometabolic risk profile of South Asians. Second, and more structurally significant, is the lack of standardization in laboratory reporting. The authors identified that diagnostic laboratories in India utilize heterogeneous reference ranges, creating "classification noise" where identical numeric values receive different clinical interpretations (normal vs. abnormal) depending on the testing facility. This inconsistency threatens the validity of prevalence tracking and patient classification.

Methodology
The research is an aggregate analysis of 794,811 digitized medical examinations collected from Indian life-insurance applicants between January 2023 and May 2026. The data was processed through an automated pipeline involving:

• Extraction: Mistral OCR for text extraction and Google Gemini 2.5 Pro for structured data extraction into a fixed JSON schema.
• Validation: Records were filtered for physiological plausibility (e.g., age 0–120, valid blood pressure ranges), resulting in the exclusion of 0.3% of records.
• Anonymization: All personally identifiable information was removed; laboratories were assigned deterministic pseudonyms.
• Canonicalization: A critical step involved mapping thousands of variable test names (e.g., "SGPT," "ALT," "Alanine Aminotransferase") and units to a standardized set of ten high-volume tests.
• Disagreement Metric: The authors defined a "pairwise reference-range disagreement rate." For a given clinical decision value (e.g., 110 mg/dL for fasting blood sugar), this metric calculates the fraction of record pairs where the laboratories' respective upper limits would classify that value differently (one as normal, one as abnormal).

Key Contributions

1. Large-Scale Demographic Insight: The study provides the largest single-source analysis of cardiometabolic risk in the working-age, urban Indian demographic, confirming the "South-Asian atherogenic-dyslipidemia phenotype" (low HDL, high triglycerides) at scale.
2. Quantification of Structural Inconsistency: The paper introduces a formal metric to quantify reference-range heterogeneity across 33,244 diagnostic centers, demonstrating that this is not merely assay variance but a classification inconsistency.
3. Open-Source Canonicalization: The authors released a canonicalization map and analysis code, addressing a "vocabulary-control problem" where 9,165 distinct test names existed for a domain of only a few hundred distinct tests.

Key Results

• Cardiometabolic Profile:

  • Hypertension: Under AHA 2017 criteria, 61.4% of the cohort met the threshold for Stage 1 or higher hypertension. Under older JNC 7 criteria, this figure drops to 18.0%, highlighting the impact of cutoff selection.
  • Dyslipidemia: 41.9% of valid records met criteria for dyslipidemia. The profile was driven heavily by low HDL (35.2%) and elevated triglycerides (20.6%), while LDL and total cholesterol elevations were less prevalent.
  • Demographics: The cohort had a median age of 38, was 68.6% male, and showed a mean BMI of 26.1 kg/m².

• Reference-Range Disagreement:

  • Significant heterogeneity was found across laboratories. At the clinical decision value of 110 mg/dL for fasting blood sugar, the disagreement rate was 49.7%.
  • For LDL at 130 mg/dL, the disagreement rate was 49.9%.
  • Other high-disagreement tests included SGOT (46.3%) and GGT (46.3%).
  • Conversely, Total Cholesterol (200 mg/dL) and Triglycerides (150 mg/dL) showed near-zero disagreement, indicating that Indian laboratories have effectively converged on ATP III thresholds for these specific markers.
  • The mean disagreement across ten common tests was 24.8%.

Significance and Claims
The paper asserts that the observed reference-range heterogeneity is not a residual OCR error but a genuine structural inconsistency in Indian diagnostic reporting. This has three specific implications:

1. Clinical Practice: Abnormal flags on high-disagreement tests (LDL, FBS, SGOT, GGT) should be treated as preliminary rather than definitive, ideally re-evaluated against a patient's historical laboratory baseline.
2. Insurance Underwriting: Risk models incorporating laboratory flags must account for "invisible" classification noise, as an applicant's risk profile may change based solely on the laboratory they visit.
3. Public Health Surveillance: Aggregated prevalence rates for laboratory-flagged abnormalities carry an additive bias that cannot be corrected without national standardization.

The authors conclude that while the cardiometabolic findings confirm established South-Asian risk patterns, the magnitude of reference-range disagreement necessitates an accelerated national standardization of laboratory reporting in India to ensure accurate disease tracking and patient classification. The study explicitly limits its scope to aggregate statistics, avoiding individual-level claims or causal inferences.