Background: Diabetes is an issue in public health and clinical medicine. Many studies have been conducted to pinpoint the predictors of diabetes in older adults, but the predictive ability of BMI and waist circumference has not been carefully evaluated in older Taiwanese.Objectives: The objective of this study was to determine the association of BMI and waist circumference (WC) with diabetes, and to evaluate the abilities of these two indicators to predict new cases of diabetes in older Taiwanese.Methods: This study analyzed the 2003 and 2007 database of "the Survey of Health and Living Status of the Elderly in Taiwan". We performed a cross-sectional analysis involving 4987 subjects (?50 years old) to determine the association in 2003 and a longitudinal analysis involving 3,626 subjects to determine the ability of BMI and WC in predicting diabetes four years later. The study also analyzed the cutoffs using Receiver Operating Characteristics.Results: In regression models controlled for demographic, lifestyle and health status variables, cross-sectional analysis showed that BMI and waist circumference were significantly associated with diabetes. The study also showed that both BMI and waist circumference can predict the risk of new diabetes in four years. WC showed a stronger association in both cases. The BMI cutoffs for diabetes decreased with aging in both men and women, while WC cutoffs increased with aging in both men and women. Conclusion: Although both BMI and waist circumference are significantly associated with the risk of diabetes, or the risk of new diabetes. Waist circumference is a better predictor of new diabetes in older Taiwanese. The predicted cutoffs for diabetes are age-related, an observation inconsistent with the cutoffs suggested by the Bureau of Health Promotion. The recommended cutoffs for diabetes may deserve more careful consideration.
