This cross-sectional analysis pooled individual-level data from nationally representative surveys in 82 low- and middle-income countries, restricted to 598,883 adults aged 25 or over whose BMI fell between 18.5 and 29.9 kg/m2 — that is, deliberately excluding both the underweight and the frankly obese, and testing the band where clinicians are least certain. Pooled diabetes prevalence was 9.1% (95% CI 8.51–9.68).
Waist circumference and relative fat mass both classified diabetes status better than BMI. In men the area under the curve was 0.69 for waist circumference and 0.69 for relative fat mass, against 0.64 for BMI; in women, 0.69 and 0.69 against 0.63. Relative fat mass was equal to or better than waist circumference across most regions and both sexes, and risk was highest in the top quintile of the waist-based measures.
This is the statistical version of something Indian clinicians already say out loud: a patient with a BMI of 24 and central adiposity is not low risk. The finding gives that intuition a number and, more usefully, an argument for changing what gets measured. A tape measure costs nothing, takes twenty seconds, and adds more discriminatory information than the weighing scale it would sit beside.
The limits are real. An AUC of 0.69 is modest — better than 0.63, but nowhere near a diagnostic test. This is about deciding who to screen with a glucose test, not about diagnosing anyone. And a cross-sectional design cannot tell you whether measuring waists changes outcomes; it tells you the measurement carries information the current one misses.
- Add a waist measurement to routine review for any adult in the normal-to-overweight BMI band, not only the obese
- Measure at the midpoint between the lowest rib and the iliac crest, on bare skin, at the end of a normal breath out
- Use a raised waist in a normal-BMI patient as a trigger for HbA1c or a fasting glucose, not as a diagnosis in itself
- Where a screening programme uses BMI thresholds, expect it to miss central adiposity in leaner South Asian patients
- Record the number so change over time is visible; a rising waist at stable weight is a meaningful signal
The statistics, in plain English
Area under the curve is the probability that the measure ranks a person with diabetes above a person without. At 0.50 the measure is a coin toss; at 1.00 it is perfect. Moving from 0.63 to 0.69 is a genuine but modest improvement — enough to justify changing which measurement you take routinely, not enough to make waist circumference a test that rules diabetes in or out. Note also that this is a cross-sectional comparison, so it shows association with existing diabetes, not prediction of future diabetes.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for diabetes & endocrinology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free