The observation that started it
Take two men with an identical body-mass index, one South Asian and one European, and measure their body fat properly rather than inferring it. They differ — substantially — with the South Asian man typically carrying a higher proportion of fat at the same index. That comparison was published in The Lancet in 2004 and it is the cleanest single statement of why an index built as a proxy for fatness behaves differently between populations.
The index itself is only weight over height squared. It has no way to distinguish muscle from fat, or fat under the skin from fat around the organs. When the relationship between the index and what it is standing in for shifts, every threshold laid on top of it shifts with it, and nobody reprints the chart.
What the WHO consultation concluded
A World Health Organization expert consultation examined the evidence and reported that in Asian populations the risk of type 2 diabetes and cardiovascular disease begins to rise at body-mass index values below the conventional overweight cut-off of 25. It proposed lower public health action points rather than a wholesale redefinition, and it was explicit that the appropriate cut-off varies between Asian populations rather than being one figure for the continent.
Indian consensus statements went further for domestic use, setting lower body-mass index and waist circumference action points for Asian Indians than those in general international use. That is why an Indian clinician may treat a figure as significant that a chart printed from an international template describes as unremarkable.
The consultation itself noted the evidence did not support a single universal Asian cut-off, and recommended action points rather than new diagnostic thresholds. Pages that quote one number as 'the Indian BMI limit' are firmer than the source is.
It is about where the fat is, not only how much
The pattern most consistently described is a greater proportion of fat stored around and inside the abdominal organs relative to fat stored under the skin, at a given total weight. That distinction matters because the two depots behave differently: fat around the organs is more strongly associated with the metabolic abnormalities that a panel picks up.
This is also why waist circumference earns its place beside the index rather than being a cruder version of it. A waist measurement, taken properly, carries information about distribution that weight and height together cannot, and the thresholds for it are explicitly population-specific.
The pattern appears remarkably early
The Pune Maternal Nutrition Study measured newborns and found Indian babies lighter than white Caucasian babies overall while preserving body fat — lighter in muscle and viscera, comparatively fat-preserving. That finding, published as the thin-fat Indian baby, indicates a body-composition pattern present at birth rather than one assembled in adulthood from office chairs and restaurant food.
It is worth being careful about what that does and does not imply. It describes a group-level pattern with a partly developmental origin. It does not mean an individual's metabolic future is fixed at birth, and the intervention evidence from adults with impaired glucose handling argues plainly against that reading.
Sources: [3]
What this changes in practice
Mainly the threshold for looking. If risk begins to rise at an index below the conventional cut-off, then 'your BMI is normal' is a weaker reassurance here than the same sentence would be elsewhere, and the sensible response to symptoms in somebody of unremarkable weight is to measure rather than to dismiss.
India's national metabolic survey supports that operationally: prediabetes, abdominal obesity and lipid abnormalities were found at a scale, and in combinations, that a screening habit built around weight alone will systematically under-detect. The argument is not that the index is useless. It is that it was never sufficient, and here it is less sufficient than usual.
What we do with it on this site
Our own calculator at /check returns a body-mass index and then spends most of its space explaining what the number cannot tell you, and it deliberately refuses to sort anybody into a category. That behaviour is a direct consequence of the evidence on this page: a categorisation drawn from the wrong population, presented confidently, is worse than no categorisation at all.
The honest version is that the index is a screening convenience with a known and quantified blind spot in this population, and that the way past the blind spot is a waist measurement and a blood panel rather than a better chart.
Sources: [4]