Assessing obesity: are ethnic differences in body mass index and waist classification criteria justified?

When the World Health Organization (WHO) first considered the issue of the overall impact of obesity in 1997, it was challenged to consider how best to relate the problem of weight gain with excess body fat to health throughout the world (1). The earlier 1993 WHO evaluation (2) of the value of anthropometry included both adult and childhood overweight but was mainly concerned with childhood malnutrition and the need to apply new body mass index (BMI) criteria to adult underweight with a classification of adult malnutrition. This was originally specified by us as ‘chronic energy deficiency’ and identified a lower BMI limit for normal weights as 18.5 (3) rather than the previous value of 20 used in Europe (4). At that stage, in the early 1990s, we were conscious that the upper normal cut-off of BMI 25.0 was based on a simplification of the relationship between BMI and mortality statistics. These were obtained for the most part in the US with some analyses suggesting that women should perhaps have a lower range of normal BMIs than men (5). At the 1997 WHO meeting we had the benefit of working with a draft produced over a period of a year by the Council and 11 sub-committees of the International Obesity TaskForce (IOTF). It was already obvious that the risks of dyslipidaemia, type 2 diabetes and hypertension rose in Caucasians with increasing BMIs from a BMI of about 20, but the focus was on total mortality and trying to produce a pragmatic classification which would be clinically useful and universally applicable. At the WHO meeting a BMI of 25 seemed a reasonable figure but even this was a compromise because many US investigators considered BMIs up to 27 or 28 as normal, particularly for the middle aged and elderly. Japanese representatives, however, were already concerned about the great risk of diabetes at much lower BMI values and sought an upper BMI cut-off point of 23.0. Given the paucity of data the compromise was not only to stick with a universal BMI value of 25.0. Furthermore a different nomenclature was used for excess BMIs so that all grades could be designated as overweight with grade 1 obesity as an alternative description for BMIs of 30–34.9. This attempt to limit any pejorative labelling of patients seemed justified when it became uncertain a year late whether the new National Institute of Health (NIH) report would be published using the same cut-off points because they dramatically escalated the perceived problem of obesity within the US (6). European data on cardiovascular risk had already shown that abdominal obesity was a far better predictor of cardiovascular disease than BMIs (7) so the WHO report adopted (for Caucasians only) the proposed waist circumferences cut-off points originally used in the Scottish Intercollegiate Guidelines Network (SIGN) guidelines for the management of obesity (8). These were based on a single study from the Netherlands where, in an attempt to find some validity for the use of a simple measure of waist circumferences (WCs) in the SIGN scheme, Lean and his colleagues had deliberately chosen WCs which corresponded to BMIs of 25 and 30 (9). WHO then simply reproduced these values whilst recognizing that waist/hip ratios had been used in epidemiological studies but were not being used clinically and would be less amenable for use in patient education and in busy doctors’ practices. The ethnic issues could not, however, be ignored and in a conjoint WHO/IASO/IOTF meeting in Hong Kong a different set of criteria were proposed, with lower BMI and WC measures (10). This meeting was not part of the formalized system within WHO so did not become official WHO policy for transmission to governments. Nevertheless, the IOTF began to collect survey results to assess whether comparable data with valid measurements could allow direct comparisons of the BMI–morbidity relationships in different national and ethnic groups (11). This initiative also contributed data to the WHO Singapore meeting where again it was inferred that Asians were at greater risk than Caucasians at equivalent BMIs (12). Rachel Huxley’s article in this issue (page 193) reflects the next phase of the analyses needed to obtain a more appropriate view of how to handle the issue of weight gain in the majority of the world’s population who live in Asia, Africa and Latin America. So far obesity criteria have been dominated by a Western perspective, despite our recognition that there are nearly twice as many deaths from cardiovascular disease in the developing countries than in affluent developed societies (13), and that Asia is the dominant region for diabetes in the world (14).