Objective: We report the effects of several different measures of body size at baseline on the subsequent development of diabetes. High levels of body fat predict the onset of diabetes, but this association has not been previously reported in a large multiethnic population of overweight or obese people with impaired glucose tolerance.
Research methods and procedures: Height, weight, waist circumference, hip circumference, and skinfolds were measured at baseline in 3234 participants enrolled in a randomized clinical trial to treat individuals with impaired glucose tolerance with placebo, metformin, or a lifestyle modification program. Cox proportional hazards models were used to assess the effect of baseline body size variables on the development of diabetes.
Results: Over an average of 3.2 years in both the placebo and lifestyle groups, baseline waist circumference had the highest or second highest R(2) value for predicting diabetes in both sexes. Cox hazard ratios per 1 standard deviation were 1.43 and 1.49 for men in the placebo or lifestyle groups, respectively, and 1.29 and 1.53 for women in the placebo and lifestyle groups, respectively, adjusted for age and self-reported race/ethnicity. The c-statistic from the receiver operating characteristic curves also favored the waist circumference in men and women in the lifestyle group and men in the placebo group. No components of body size were predictive in the metformin-treated group, and metformin compared with the placebo group was effective in preventing diabetes only in individuals with a BMI > or =35 kg/m(2) or a waist circumference > or =98.0 cm.
Discussion: Large waist circumference was a better predictor of risk for developing diabetes than most other measures in the placebo and lifestyle groups. No baseline measure of body size or shape predicted risk of diabetes in the metformin-treated group.