To the Editor: The recent report by the DECODE-DECODA Study Group [1] on the effect of ethnicity on the association between body mass index (BMI) and prevalence of diabetes is extremely important. I wish to support the group’s finding that the association between BMI and diabetes is greatly modified by ethnicity and consequent recommendation of a redefinition of obesity based on geographic region and ethnicity suggested by the World Health Organization.

The Chinese have a lower baseline BMI to begin with [2] (baseline value=21 [3], mean=18.5–23.9 [4]), and it takes less increment to reach an obese level [5], so that in China a BMI over 23 [6] is considered as overweight and a BMI over 28 as obese [4]. It takes smaller increments to increase the risk of hypertension, coronary artery disease and Type 2 diabetes in the Chinese population [3, 7, 8].

Furthermore, it is not the obesity by itself that is an important risk factor; it is where the obesity is that matters. For example, postmenopausal Chinese women with abdominal obesity carries a higher metabolic cardiovascular risk than those without it and it is the waist circumference rather than the BMI that predicts the risk in those women [9]. Just like comparing apples and oranges, one should not compare apples (abdominal obesity) and pears (ordinary feminine body build).

Of course, some 250 years ago, Joannes Baptista Morgagni with the help of only a knife for anatomical dissection, an acute mind and an observational skillfulness was able to identify the intra-abdominal fat accumulation in android obesity [10]. He clearly described the association between visceral obesity, hypertension and atherosclerosis, long before the modern recognition of this connection [10].

Finally, no man should ever attempt to adopt the rotund figure of Luciano Pavarotti [11], unless he thinks he can sing as well as Pavarotti. Even Pavarotti is aware of the risk of obesity and is doing something about it.