The levels of BMI associated with increased mortality are based on epidemiological studies of primarily white populations. The interest in confirming the association between BMI and mortality in other racial/ethnic groups stems partly from observations that lower-than-average total mortality has been observed among some populations with a high BMI level (299), and partly from observations that within certain populations there appears to be no effect of obesity at all or at the BMI levels that are associated with higher mortality in whites.
Three small cohort studies of narrowly defined populations of African Americans failed to show the expected association of BMI and mortality based on data from white populations (300-303). Although the shape of the association of BMI and mortality in two large, representative U.S. data sets (the National Health and Nutrition Examination Follow-up Study and the National Health Interview Survey) is similar for black and white males and females (304), the BMI-related increase in risk begins at a 1 to 3 kg/m2 higher BMI level for blacks than for whites. For example, in the National Health and Nutrition Examination Follow-up Survey, the estimated BMI associated with minimum mortality was 27.1 for black men and 26.8 for black women, compared with 24.8 and 24.3, respectively, for white men and women. On the basis of these data, the use of the cutpoint of BMI > 30 kg/m2 for defining obesity is clearly applicable to African Americans as well as to whites.
Limited data relating obesity to mortality in American Indians were identified, but no data were found relating obesity to mortality in Hispanic Americans, Asian Americans, or Pacific Islanders (305). The lowest mortality rate among Pima men is observed at a BMI range of 35 to 40 kg/m2 for men, and no relationship between BMI and mortality is observed among Pima women (306, 307). Based on mortality data alone, it would be hard to justify using the same standard for defining obesity in populations, such as America Indians, among whom the mean BMI is much higher than in the general U.S. population. However, diabetes-related morbity among obese American Indians is extremely high (154), and the overall age-specific mortality among American Indians is generally higher than in the U.S. general population (306, 307). Thus, obesity in American Indians is associated with a compromised overall survival of the population.
Although the data on mortality are still fragmentary for many minority populations, there are no studies that would support the exclusion of any racial/ethnic group from the current definitions of obesity. Secular trends in many polulations in the United States and throughout the world have demonstrated that longstanding overweight and obesity eventually leads to the emergence of chronic diseases. Therefore, prevalent overweight and obesity cannot be ignored even where the associated health problems have not reached the level that would be expected on the basis of data for white populations.