The WHO Asian BMI Cut-Offs and the Risk Data That Drove Them

The WHO Western Pacific Regional Office 2004 recommendation was a direct response to epidemiological evidence showing that the international adult BMI thresholds missed a substantial portion of metabolic risk in Asian populations. The expert consultation, published in The Lancet, analysed data from roughly 240,000 adults across ten Asian countries and found that type 2 diabetes and cardiovascular disease risk began to rise at a BMI around 23, not at the 25 that the global standard used to mark overweight. For a clinician, the arithmetic was simple: a person of Asian descent at BMI 24 had a risk profile similar to a person of European descent at BMI 27, but the old thresholds said the first was normal weight and the second was overweight. That mismatch is why the WHO recommended that Asian populations use overweight at 23.0 to 24.9 and obesity at 25.0 and above, and why the 2004 document also listed trigger points for public health action at 23.0, 27.5, 32.5, and 37.5.

The Body Composition Gap

The reason the WHO moved the cut-offs was a measurement problem. At the same BMI, body fat percentage and visceral adipose tissue are higher in most Asian populations than in white populations. A 2021 meta-analysis of DXA-measured body composition across more than 100,000 adults found that for any given BMI, Asian men had about 3-4 percentage points more body fat and Asian women about 2-3 percentage points more than their white counterparts with the same weight-for-height ratio. That means a BMI of 24 in a Japanese man carries the same adiposity burden as a BMI of 26.5 in a Norwegian man. The epidemiological evidence in the 2004 WHO consultation showed the same shift in the risk curve: the odds ratio for type 2 diabetes at a BMI of 23-24.9, relative to below 23, was between 2.0 and 3.0 across studies in India, China, Singapore, and Japan. That is not a marginal difference. It is a doubling or tripling of diabetes risk at a weight the international cut-offs would have called normal.

Why Visceral Fat Drives the Risk

The deeper mechanism is insulin resistance. Visceral adipose tissue secretes pro-inflammatory adipokines and free fatty acids directly into the portal circulation, and higher visceral fat at lower BMI is precisely why metabolic risk rises before the scale reflects obesity by the international standard. Central adiposity, rather than total fat mass, drives the association. The WHO consultation was explicit that the lower cut-offs were not diagnostic criteria for individuals but screening thresholds designed to capture people who would otherwise be missed. That is a critical distinction. No one is claiming that a BMI of 24 is diseased. They are claiming that at BMI 24 in an Asian adult, the screening net should open wider than it would for a white adult.

The WHO Western Pacific Region BMI guidance was never meant to be a global standard applied uniformly. It was a regional recommendation that national bodies were expected to adapt, and they did. India, Singapore, and Japan each produced their own population-specific thresholds through their national health authorities, and those numbers differ from each other in ways that matter for screening.

India's consensus guidelines, published in 2009 in the Journal of the Association of Physicians of India, set normal at 18.5-22.9, overweight at 23.0-24.9, and obesity at 25.0 or above. The Indian evidence base showed that at a BMI of 23, the prevalence of hypertension, dyslipidaemia, and type 2 diabetes was already elevated, and that central adiposity, measured by waist circumference, was a stronger predictor than BMI for many of those outcomes. Singapore's Health Promotion Board adopted a slightly different normal range of 18.5-22.9, overweight at 23.0-27.4, and obesity at 27.5 or above, reflecting data that the risk curve in the Singaporean Chinese, Malay, and Indian populations did not rise quite as steeply as in the Indian subcontinent. Japan went a different route entirely. The Japan Society for the Study of Obesity, in its 2011 guidelines, kept the international BMI bands for underweight and normal but re-labelled the category from 25.0 to 29.9 as obesity grade 1, and defined obesity as a BMI of 25 or above regardless of body fat percentage, on the grounds that the metabolic syndrome was already common at that level in the Japanese population.

What this means in practice is that there is no single Asian cut-off. China uses 24 for overweight and 28 for obesity. India, Singapore, and Japan each have their own numbers. The 2004 recommendation of 23-24.9 for overweight and 25+ for obesity is the common ancestor, but national bodies have adjusted it based on local epidemiological data, mortality studies, and healthcare capacity. The risk curve shift is universal across these populations. The exact point where the curve turns upward differs, which is why the thresholds are population-specific rather than a single pan-Asian number.

The 2004 WHO recommendation did not retire the international adult cut-offs. Both sets of thresholds remain in active use, and the choice of which one applies depends entirely on the population and the issuing body. The WHO international cut-offs, published in 1995 and reaffirmed in 2000, define overweight as 25.0-29.9 and obesity as 30.0 or above. Those remain the standard for white populations in Europe, North America, and Australia. The CDC in the United States uses the international cut-offs. The NHS in the United Kingdom does too, though it adds a severe obesity category at 40.0 or above. The NHLBI in the US, in its 1998 guidelines, adopted the same international bands for clinical practice.

The 2004 document, by contrast, applies to populations within the WHO Western Pacific Region, and national guidelines in India, Singapore, Japan, and China have each operationalised that recommendation into their own screening protocols. A clinician in Mumbai applying the international standard would label a patient at BMI 26 as overweight. A clinician in Singapore applying the national guideline would call the same BMI obese. Neither is wrong, because they are answering different questions. The international cut-off asks about mortality risk across a broad population. The Asian cut-off asks about the onset of type 2 diabetes and cardiovascular disease risk in populations where the same BMI carries more body fat and more visceral adipose tissue. When the WHO issued the 2004 consultation, it was explicit that the international cut-offs were inappropriate for Asians and that the lower thresholds should be used. But it did not replace the international ones for other populations. The two systems coexist because the biology they screen for is not the same across ethnic groups.

For the type 2 diabetes risk at lower BMI in Asian populations, the evidence is unusually strong for a screening question. The WHO consultation that produced the 2004 thresholds synthesised data from cohort studies and cross-sectional surveys across ten countries. The central finding was a risk curve that runs consistently to the left of the European curve. The Zheng et al. paper in the New England Journal of Medicine in 2011 followed over one million Asian adults and found that the hazard ratio for all-cause mortality at a BMI of 25.0 or above, relative to 22.5-24.9, was 1.26 (95% CI 1.18-1.34). But that mortality data underestimates the metabolic risk, because the diabetes curve is steeper than the mortality curve. The odds ratio for type 2 diabetes at BMI 23-24.9, relative to below 23, was 2.0-3.0 in the 2004 consultation. That is not a subtle shift. It is the difference between a condition that affects 5% of a population and one that affects 12-15%, purely by moving the threshold.

The mechanism is visceral adipose tissue. At the same BMI, Asian populations tend to store more fat in the abdominal cavity and less subcutaneously. That is why a waist-to-height ratio above 0.5 is a better screening catch than BMI alone in this group. The Japan Society for the Study of Obesity set a visceral fat area threshold of 100 cm², measured by CT, as the point where metabolic risk accelerates. That threshold was reached at a BMI closer to 23 in Japanese men than the 25 that would be typical for a European man. The practical implication for screening is that a clinician who uses only the international BMI cut-offs with an Asian patient will systematically miss the window where lifestyle intervention is most effective. The diabetes risk has already doubled by the time the international threshold says overweight.

The evidence that underlies the WHO Asian BMI cut-offs is a meta-analysis of cross-sectional studies, not a single trial. That is a strength for a screening threshold, because the question is not whether a drug works but whether the risk curve shifts. The 2004 consultation pooled data from about 240,000 adults across ten countries to answer that. The analysis showed that body fat percentage at a given BMI was higher in Asian populations than in Europeans, and the difference was larger at lower BMIs. At a BMI of 23, an Asian adult had approximately the same body fat percentage as a European adult at BMI 25. That means the international cut-off of 25 for overweight was, in effect, systematically under-detecting the people most likely to develop type 2 diabetes and cardiovascular disease in the Western Pacific Region.

The specific numbers in the 2004 recommendation were not arbitrary. The WHO chose 23.0 as the overweight threshold because the risk of type 2 diabetes and cardiovascular disease began to rise consistently at that point across the pooled data. It chose 25.0 for obesity because that was the level at which the risk became substantial and where the WHO argued intervention was warranted. The additional action points at 23.0, 27.5, 32.5, and 37.5 were designed for public health programmes that need staged interventions. None of these numbers is a biological law. They are decision points on a continuous risk curve, chosen to maximise the sensitivity of BMI as a screening tool while keeping the specificity acceptable. The sensitivity of BMI for obesity, defined by DXA-measured body fat percentage, is only about 50% in men and 60% in women when using the international cut-off of 30. At the WHO Asian cut-off of 25, the sensitivity improves substantially, at the cost of labelling more people as obese who may have normal body fat. That trade-off is why the thresholds differ by population.

India: Tightening the WHO Numbers

India national guidelines on BMI were published in 2009 by a consensus group, and they followed the WHO regional recommendation but tightened it further. The Indian consensus document, authored by Misra and colleagues, set normal at 18.5-22.9, overweight at 23.0-24.9, and obesity at 25.0 or above, with a separate abdominal obesity threshold based on waist circumference that was lower than the WHO international cut-offs. The Indian evidence showed that at a BMI of 23, the prevalence of metabolic syndrome was already 22% in urban adults, and that diabetes incidence rose steeply above that point. The concern with the international cut-off was that it would label a significant fraction of Indian adults as normal weight when their visceral adipose tissue and insulin resistance already placed them in a high-risk group.

Singapore and Japan: Two Different Answers

Singapore national guidelines followed a similar trajectory but with a different normal range. The Health Promotion Board's 2005 guidelines set normal at 18.5-22.9, overweight at 23.0-27.4, and obesity at 27.5 or above. The Singaporean adaptation was based on local epidemiological data showing that the risk curve for diabetes in the Chinese, Malay, and Indian sub-populations was not identical, and the 27.5 obesity threshold was chosen to align with the WHO action point for public health intervention. Japan national guidelines took a different approach. The Japan Society for the Study of Obesity's 2011 guidelines kept the international BMI bands but re-categorised them, defining obesity as a BMI of 25 or above, with grade 1 obesity from 25 to 29.9, grade 2 from 30 to 34.9, and grade 3 at 35 or above. The Japanese rationale was that the metabolic syndrome, defined by visceral fat accumulation, was present in a substantial proportion of the population at BMI 25, and that the international threshold of 30 for obesity was simply too high to catch the majority of at-risk Japanese adults. Each of these national adaptations answers a shared question: at what BMI does the risk curve turn upward for this population? The answers differ because the populations differ, not because any one country is being lenient or strict.

Body fat percentage at a given BMI is the reason the Asian cut-offs exist. It is the number that most clinicians do not calculate. DXA-measured body fat percentage cut-offs for obesity are commonly cited at 25% for men and 35% for women, but those figures come from WHO 1995 data and were derived largely from white populations. In Asian populations, using those cut-offs misclassifies a substantial fraction of people. A 2017 study of South Asian adults in urban India found that the average body fat percentage at a BMI of 25 was 28% in men and 38% in women, which is above the obesity cut-off, even though the BMI would be called overweight. The implication is that the body fat percentage at a given BMI is not a constant. It is a population-specific variable, and the slope of the relationship between BMI and body fat percentage is steeper in Asian populations.

That is why the WHO 2004 consultation recommended the lower cut-offs, and why the risk of type 2 diabetes and cardiovascular disease rises at a lower BMI in these populations. The body fat percentage difference is not trivial. It is on the order of 3-5 percentage points at the same BMI, which is the difference between a normal body fat percentage and an overweight one. When you add visceral adipose tissue, which is not captured by BMI at all, the metabolic picture diverges further. A normal-weight Asian adult with a waist-to-height ratio above 0.5 and a family history of diabetes has a risk profile that the international BMI cut-offs would entirely miss. The screening failure is not hypothetical. It is the reason the WHO Western Pacific Regional Office called the consultation in the first place.

The risk curve shift in Asian populations is not a single curve but a family of curves that vary by outcome. For all-cause mortality, the lowest risk is at a BMI between 20 and 25, and the risk rises only modestly above that, with a hazard ratio of about 1.11 per 5 kg/m² above 25. But for type 2 diabetes, the curve is much steeper, with a relative risk of about 1.5 at BMI 25, 3.0 at BMI 30, and 7.0 at BMI 35, relative to a BMI of 22. In Asian populations, those relative risks are reached at lower BMIs, because the baseline risk of diabetes is higher at any given BMI. The Zheng et al. 2011 study showed this clearly: in East Asian populations, the hazard ratio for all-cause mortality at a BMI of 25.0 or above was 1.26, but the hazard ratio for diabetes-related death was substantially higher, closer to 1.8. The excess risk is concentrated in metabolic outcomes, not in cancer or injury. That is why a screening tool that misses diabetes risk is failing in its primary purpose.

The practical consequence for screening is that the WHO Asian BMI cut-offs of 23 for overweight and 25 for obesity are better predictors of type 2 diabetes risk than the international cut-offs, but they are still imperfect. A BMI of 23 captures some people who will not develop diabetes, and a BMI of 22 misses some who will. That is why waist circumference and waist-to-height ratio are increasingly recommended as adjuncts. The 0.5 rule for waist-to-height ratio is a population heuristic, not a personalised risk score, but it works because it captures central adiposity, which is the visceral fat that drives insulin resistance. For an Asian adult at a BMI of 24, a waist-to-height ratio above 0.5 is a stronger signal for metabolic risk than the BMI alone. The WHO 2004 consultation did not say that BMI was irrelevant. It said that the cut-offs needed to be lower to make the screening tool useful, and that central adiposity should be measured alongside it.

The J-shaped curve of BMI and mortality, with its nadir around 22-24 in most populations, is worth understanding precisely because it shows why the Asian cut-offs are not about thinness becoming a virtue. The lowest all-cause mortality in Asian adults in the Zheng study was at a BMI of 22.5-24.9, which is the upper end of the normal range under both the international and the Asian classification. That means the WHO Asian overweight threshold of 23 is not marking people who are at the bottom of the risk curve. It is marking the point where risk begins to rise above the population minimum. The obesity threshold of 25 is where the risk is already substantially elevated. The curve shift for Asian populations is real but it is not dramatic enough to make a BMI of 22 dangerous. It is a shift of about 2-3 BMI units, not a complete reordering of the relationship.

That nuance is lost in most popular discussions of the Asian cut-offs, which tend to present them as either a necessary correction or an overreaction. Neither is accurate. The weight of evidence from the 2004 consultation, the 2011 NEJM study, and the 2021 meta-analyses is that the risk of type 2 diabetes and cardiovascular disease in Asian populations rises at a lower BMI than in white populations, and that the difference is explained by body fat percentage and visceral adipose tissue. The exact cut-off points are a public health decision, not a biological fact. The WHO chose 23 and 25 because they were round numbers that balanced sensitivity and specificity. Individual countries have adjusted them based on local mortality data. The principle, that the BMI-risk relationship is population-specific, is settled. The specific numbers will continue to be reviewed as new data emerges, but they will not be abandoned for a return to the single international standard for everyone.

What to Do With a BMI of 25

The failure case for someone who has just learned they have a BMI of 25, and who is of Asian descent, is that they will do one of two things: dismiss the number because they do not look overweight, or panic and assume they are morbidly obese. Both are wrong because neither understands what the number is. A BMI of 25 by the WHO Asian cut-off places them in the obesity category. That does not mean they have a body fat percentage of 25% or that they have type 2 diabetes. It means they are in the group of people for whom the risk of developing it is elevated, and for whom a screening test or a doctor's visit is now worthwhile. Book a fasting blood glucose and a lipid panel, and measure waist circumference to see if central adiposity is present. If the waist-to-height ratio is above 0.5, that is a stronger signal than the BMI.

The Clinician's Responsibility

The alternative failure is the clinician who knows the international cut-offs but not the Asian ones, and who tells a patient at BMI 27 that they are fine because the international standard says they are only overweight. That clinician is using a threshold that was derived from a population with a different body composition. They are missing the window where lifestyle intervention has been shown to reduce diabetes incidence by 40-60% in high-risk groups. The evidence from the 2004 WHO consultation is clear: a BMI of 27 in an Asian adult carries a similar diabetes risk to a BMI of 30 in a white adult. No one would tell a white adult at BMI 30 that they are fine. The principle of ethnic variation in the BMI-risk relationship is settled, and the WHO Western Pacific Regional Office 2004 recommendation was the formal expression of that. The exact thresholds will be reviewed, but the direction of travel is not in doubt.

The Public Health Action Points

The 2004 WHO consultation did not just lower the obesity threshold. It also introduced a set of additional trigger points for public health action. Those points, at BMI 23.0, 27.5, 32.5, and 37.5, were designed for countries to use in resource allocation, not for individual diagnosis. A BMI of 23.0 marked the first point where the risk of type 2 diabetes and cardiovascular disease began to rise. A BMI of 37.5 was the point where bariatric surgery was considered for Asian patients even without comorbidities. The surgery threshold is worth noting because it is lower than the international recommendation of 40, and it reflects the higher body fat percentage and visceral adipose tissue at any given BMI in Asian populations. A systematic review that informed the 2004 consultation found that Asian patients with a BMI of 35 and type 2 diabetes had a mortality benefit from bariatric surgery, which is why the action point at 32.5 exists.

The BMI thresholds for Asian populations are not a single number across the region. Assuming they are would be a mistake. China uses 24 for overweight and 28 for obesity, which is higher than the WHO regional recommendation of 23 and 25. The Chinese epidemiological data showed that the risk curve for cardiovascular disease and all-cause mortality was lower in the Chinese population than in South Asians. The Indian consensus, by contrast, kept the WHO regional numbers, because the diabetes risk in the Indian subcontinent is among the highest in the world, and even a BMI of 23 was associated with a marked increase in insulin resistance. Singapore sits between the two. Japan uses a different classification system altogether. The lesson is that the WHO regional cut-offs are a starting point, not a destination. A national health body must decide whether to accept them, tighten them, or loosen them based on local mortality data, healthcare capacity, and the prevalence of central adiposity. The risk of type 2 diabetes at lower BMI is a real phenomenon, but its magnitude varies by population, and the thresholds must reflect that variation to be useful.

For a patient crossing from one healthcare system to another, the differences are not academic. A Singaporean adult who is told they are in the normal range at BMI 22 in Singapore would be reclassified as overweight in India, where the normal range is 18.5-22.9. A Japanese adult at BMI 25 is considered obese by the Japan Society for the Study of Obesity, but a Chinese adult at the same BMI is only overweight. The same body, the same amount of fat, the same metabolic risk, but three different labels depending on which country's guideline is applied. This is not a failure of the system. It is the system working as designed, because the thresholds are calibrated to the population in which they are used. Ask not just what the BMI is, but which guideline is being applied.

Body fat percentage is a more direct measure than BMI, but it is not a simple fix. A DXA scan is the gold standard for body composition measurement, but it is expensive, involves ionising radiation, and is not available in most primary care settings. Bioelectrical impedance analysis (BIA) devices are cheaper but have individual errors of ±3-8 percentage points depending on hydration, device quality, and population-specific equations. The practical reality is that BMI and waist circumference are the tools available in a screening context. The WHO Asian cut-offs exist to make BMI work better for the populations it serves. The evidence from the 2004 consultation is that body fat percentage at a given BMI is higher in Asian populations, which is why the screening threshold had to move. The same logic applies to waist circumference. The International Diabetes Federation reduced the cut-offs for Asian men to 90 cm and for Asian women to 80 cm, compared with 94 and 80 for white men and women.

The measurement issue matters because the goal of screening is to identify people who need further testing for type 2 diabetes and cardiovascular disease, not to diagnose them. A BMI of 23 in an Asian adult is a signal to check fasting glucose, blood pressure, and lipids. It is not a diagnosis of diabetes or hypertension. The 2004 WHO consultation was explicit that the cut-offs were for screening and public health action, not for individual diagnosis. That distinction gets lost in headlines about obesity and BMI. A person at BMI 24 who is told they are overweight by the Asian cut-off has not been given a medical verdict. They have been given a heads-up that their risk profile warrants a closer look. The evidence base for this is strong, and it is built on the observation that visceral adipose tissue, not weight, is the primary driver of insulin resistance and metabolic risk. That is a settled principle.

The CT Scan That Defines the Cut-Off

The Japan Society for the Study of Obesity has a specific approach worth understanding because it ties the BMI threshold to a measured physical quantity. The Society set a visceral fat area threshold of 100 cm², measured by CT scan at the umbilicus, as the point where metabolic risk accelerates. In Japanese men, that visceral fat area was reached at a BMI of approximately 23. In Japanese women it was reached at a slightly higher BMI, reflecting the different fat distribution between sexes. The JSSO then worked backwards from that number to define obesity as a BMI of 25 or above, on the grounds that the vast majority of Japanese adults with a visceral fat area above 100 cm² had a BMI in that range. The reasoning is circular in a productive way: the CT measurement defines the risk, and the BMI cut-off is chosen because it captures most of the at-risk people without requiring a CT scan for everyone. This is the same logic that underlies the WHO Asian cut-offs, but the Japanese version is tied to a specific measurement of visceral adipose tissue, which makes it more reproducible.

Why Singapore Chose a Different Number

The contrast with the Singaporean approach is instructive. Singapore's Health Promotion Board set the obesity cut-off at 27.5, which is lower than the international 30 but higher than the WHO regional 25. The reason for the difference is that Singapore's population includes three major ethnic groups, each with a different relationship between BMI and visceral fat. The Chinese Singaporean population has a diabetes risk curve similar to that of mainland China. The Malay population has a higher risk at any BMI. The Indian Singaporean population has the highest risk of all, with a diabetes prevalence that is about twice that of the Chinese population at the same BMI. A single threshold cannot serve all three populations well. The Health Promotion Board chose 27.5 as a compromise that captures a substantial fraction of the high-risk Indian and Malay population without labelling the entire Chinese population as obese. The result is a guideline that is pragmatic but internally inconsistent, and the WHO regional 2004 recommendation of 25 remains the only number that is simple to communicate.

The exact relative risk values for type 2 diabetes at a given BMI in Asian populations vary by study, with some meta-analyses reporting an odds ratio of 2.0 at BMI 23-24.9 and others reporting 2.5 or 3.0. The WHO Western Pacific Regional Office publishes the current summary table of risk estimates. Check their source documents for the latest figures. What is not in doubt is the direction and approximate magnitude of the effect: the risk curve for diabetes in Asian populations is shifted to the left relative to white populations, and the shift is large enough to justify lower screening thresholds. The Zheng et al. 2011 NEJM paper is the definitive mortality study, and it found a hazard ratio of 1.26 for all-cause mortality at BMI 25 or above compared with BMI 22.5-24.9, but the diabetes-specific risk is higher. For precise numbers for a specific population, the national guidelines from India, Singapore, Japan, and China each publish their own risk tables, and those are more appropriate than a global average.

Do not use the WHO Asian cut-offs of 23 and 25 as a universal diagnostic standard, just as you would not use the international 25 and 30 for everyone. The 2004 consultation was a recommendation to national health bodies, not a directive to individual clinicians, and it explicitly said that the cut-offs should be adjusted based on local data. The principle of ethnic variation is the settled fact. The specific numbers are a living policy area. For a clinician, the right approach is to use the national guideline for the patient's own ethnic background, and to interpret it alongside waist circumference and a fasting glucose test. For an individual, the takeaway is simpler: if you are of Asian descent and your BMI is above 23, the evidence says you should have your metabolic risk assessed, regardless of what the international chart says. The BMI is not a verdict. It is an invitation to look more closely.

A BMI of 23 in an Asian adult is an invitation to check fasting glucose, not a diagnosis, and the Japan Society for the Study of Obesity's 100 cm² visceral fat area threshold is what makes its 25.0 obesity cut-off a backward-calculation from a CT scan rather than an arbitrary number.