Abstract

We formally evaluated waist circumference (WC) percentile cutoffs for predicting insulin resistance (IR) and whether different cutoffs should be used for adolescents of different race/ethnicities. Analysis was performed for 1575 adolescents aged 12-18 yr from the National Health and Nutrition Examination Survey 1999-2002. Adolescents were classified as having IR if they had a homeostasis model assessment-insulin resistance level, a validated measure of IR, of >4.39, and WC percentile was classified according to previously published universal (all races combined) and race/ethnicity-specific WC percentile cutoffs. Receiver operating characteristic curves for predicting IR were constructed comparing the race/ethnicity-specific vs. universal WC percentile cutoffs, and area under the curve (AUC) was calculated. Comparing universal with race/ethnicity-specific WC percentiles, there were no significant differences in AUC for Black, Mexican-American, or White adolescents. Because race/ethnicity-specific thresholds did not discriminate better than universal WC thresholds, universal WC thresholds may be used effectively to identify adolescents with IR in primary care practices. A WC > or =75th or > or =90th percentile for all race/ethnicities combined would be appropriate to apply in clinical practice for identification of adolescents with IR, a risk factor for development of type 2 diabetes.