LAUSR

Background: The predictive power of obesity measures varies according to the presence of coexistent measures. The present study aimed to determine the predictive power of combinations of obesity measures for diabetes by calculation of a linear risk score. Methods: Data from a population-based cross-sectional study of 994 representative samples of Iranian adults in Babol, Iran were analyzed. Measures of obesity including waist circumference (WC), body mass index (BMI), waist–to–height ratio (WHtR), and waist to hip ratio (WHR) were calculated, and diabetes was diagnosed by fasting blood sugar >126 mg/dl or taking antidiabetic medication. Multiple logistic regression model was used to develop a logit risk score based on BMI, WC, WHtR, and WHR. The ROC analysis was applied to determine the priority of every single index and combined logit score for the prediction of diabetes. Results: All four measures of general and abdominal obesity were predictors of diabetes individually in both sexes (P=0.0001). Calculation of risk score for a combination of all measures use full model improved predictive power. Adjustment for age resulted in further improvement in diagnostic power and combined novel risk score differentiated individuals with and without diabetes with an accuracy of 0.747 (95%CI: 0.690-0.808) in men and 0.789 (95%CI: 0.740, 0.837) in women. Conclusion: These findings indicate that the simultaneous calculation of age-adjusted risk score for all measures provides stronger diagnostic accuracy in both sexes. This issue suggests the calculation of combined risk scores for all obesity indices especially in a population at borderline risk.