Objective: To evaluate the diagnostic accuracy of body mass index (BMI, kg/m(2)), waist-circumference (WC) and waist-hip-ratio (WHR) as diagnostic tests for detecting fatness in adolescents.
Design: A cross-sectional analysis of 474 healthy adolescents aged 17 y was used. Measurements of height, weight, WC, hip-circumference and body fat percentage (%BF) were obtained. The diagnostic accuracy for detecting excess fatness was evaluated through receiver operating characteristics (ROC) analyses with %BF, measured by densitometry (air-displacement plethysmography), as reference test.
Results: BMI and WC showed strong positive correlation (r=0.68-0.73; P<0.0001) with %BF in both sexes, but the correlation was weaker for WHR (r=0.30-0.41; P<0.0001). For overweight and obesity in boys and obesity in girls, the area under the ROC curve was high (0.96-0.99) for BMI and WC. WHR was not significantly better than chance as diagnostic test for obesity in girls. For BMI and WC, highly sensitive and specific cutoffs for obesity could be derived, while larger trade-offs were needed for detecting overweight in girls. The cutoffs producing equal sensitivity and specificity were lower than the ones minimizing the absolute number of misclassifications. The latter approached internationally recommended reference values, but were still several units lower for BMI in girls and several centimeters lower for WC in boys.
Conclusion: BMI and WC were found to perform well as diagnostic tests for fatness, while WHR was less useful. The discrepancies between cutoffs producing equal sensitivity and specificity, cutoffs minimizing the absolute number of misclassifications and internationally recommended reference values for overweight and obesity highlight the importance of specifying the characteristics of classification systems for different settings.