Abstract
BACKGROUND AND OBJECTIVES: Studies have shown a strong association between excess weight and risk of incident diabetes in Iranian women. Therefore, we investigated anthropometric indices in the prediction of diabetes in Iranian women. SUBJECTS AND METHODS: We examined 2801 females aged > or =20 years (mean [SD] age, 45.2 [12.9] years) in an Iranian urban population who were non-diabetic or had abnormal glucose tolerance at baseline. We estimated the predictive value of central obesity parameters (waist circumference [WC], waist-to-hip ratio [WHR], waist-to-height ratio [WHtR], body mass index [BMI]) in the prediction of diabetes. We classified each parameter in quartiles and compared the lowest with the highest quartile after adjusting for confounding variables, including age, hypertension, triglyceride levels, HDL-cholesterol, family history of diabetes, and abnormal glucose tolerance in a multivariate model. Receiver operator characteristic (ROC) curves were used to determine the predictive power of each variable. RESULTS: Over a median follow up of 3.5 years (11 months-6.3 years), 114 individuals developed diabetes (4.1%). The risk for developing diabetes was significantly higher for the highest quartile of BMI, WC, WHR and WHtR, respectively, compared to the lowest quartile, and the risk decreased but remained statistically significant when abnormal glucose tolerance was included in the multivariate model. WHtR had the highest area under the ROC curve. CONCLUSIONS: In Iranian women, BMI, WC, WHR, WHtR were predictive of development of type 2 diabetes, but WHtR was a better predictor than BMI.