OBJECTIVE

To evaluate the performance of the body mass index (BMI), waist circumference, waist-to-hip ratio (WHR), and waist-to-height ratio (WHTR) in predicting incident diabetes in Jamaica.

RESEARCH METHODS AND PROCEDURES

A cohort of 728 nondiabetic adults (290 men and 438 women), ages 25 to 74 years and residents of Spanish Town, Jamaica, were followed for a mean of 4 years. Participants had fasting and 2-hour postchallenge glucose concentrations measured at baseline and follow-up.

RESULTS

There were 51 cases of incident diabetes (17 men and 34 women). All indices were independent predictors of diabetes, and none was clearly superior. The area under the receiver operating characteristics curves (95% confidence interval) for BMI was 0.74 (0.59 to 0.88) for men and 0.62 (0.51 to 0.72) for women. For waist circumference, these values were 0.78 (0.65 to 0.91) in men and 0.61 (0.50 to 0.71) in women. Similar results were obtained for WHR and WHTR. "Optimal" cut-off points for BMI were 24.8 kg/m(2) (men) and 29.3 kg/m(2) (women). For waist circumference, these were 88 cm and 84.5 cm for men and women, respectively. Corresponding values for WHR were 0.87 and 0.80 and for WHTR were 0.51 and 0.54, respectively.

DISCUSSION

Cut-off points for waist circumference and WHR were similar to those proposed in developed countries for women but lower in men. Waist circumference could be useful in health promotion as an alternative to BMI.