Which Obesity Index Best Explains Prevalence Differences in Type 2 Diabetes Mellitus?*

Abstract

Obesity drives the diabetes epidemic. However, it is not known which obesity index best explains variations in type 2 diabetes mellitus prevalence across populations. We analyzed three cross-sectional studies from San Antonio, TX, (Mexican-Americans and non-Hispanic whites, n = 2839), Mexico City (n = 2233), and Spain (n = 2161) (age range, 35 to 64 years). We used the area under the receiver operating characteristic curve (AUC) to assess performance for identifying diabetic subjects and logistic regression analysis to examine differences in diabetes prevalence. AUCs for waist circumference and BMI were similar in white subjects, but the AUC for waist circumference was greater in Mexican-origin subjects (Mexican men, 0.594 vs. 0.549, p = 0.008; and women, 0.605 vs. 0.557, p = 0.002; Mexican-American men, 0.648 vs. 0.600, p < 0.001; and women, 0.744 vs. 0.693, p < 0.001). The AUC for waist-to-height ratio tended to be greater than that for waist circumference, but statistical significance was demonstrated only in Mexican women (0.628 vs. 0.613, p = 0.044), Mexican-American women (0.774 vs. 0.758, p < 0.001), and Spanish women (0.734 vs. 0.715, p = 0.039). No obesity index was consistently superior to the others for explaining differences in diabetes prevalence among populations. In white and Mexican-origin men, waist circumference may be the preferred marker for identifying diabetic subjects on account of its simplicity; in women, waist-to-height ratio may be better. Differences in diabetes prevalence among these populations cannot be attributed to a single measure of obesity.