Differences in the associations of anthropometric measures with insulin resistance and type 2 diabetes mellitus between Korean and US populations: Comparisons of representative nationwide sample data

Backgrounds Variation among ethnic groups in the association between obesity and insulin resistance (IR)/diabetes has been suggested, but studies reported inconsistent results. We evaluated ethnic differences in the association between obesity and insulin resistance (IR)/diabetes. Methods We conducted a cross-sectional analysis using Korea (n = 18,845) and the USA (n = 4657) National Health and Nutrition Examination Survey(NHANES) 2007–2010. We performed statistical comparisons of AUC-ROC (area under the curve in a receiver operating characteristic curve) values for body mass index (BMI), waist circumference (WC) and homeostasis model assessment of insulin resistance (HOMA-IR) to predict IR or diabetes among different ethnic groups. Results AUC-ROC values for BMI and WC for predicting IR were highest in Whites (0.8324 and 0.8468) and lowest in Koreans (0.7422 and 0.7367). Whites showed the highest AUC-ROC values for BMI (0.6869) and WC (0.7421) for predicting diabetes, while the AUC-ROC for HOMA-IR was highest in Koreans (0.8861). Linear regression showed significant interactions between ethnicity and the main effects (all P < 0.0001). Increases in BMI were associated with a larger increase in HOMA-IR in Whites (β = 0.0719) and WC in Hispanics (β = 0.0324), while BMI was associated with a larger increase in fasting glucose in Koreans (β = 0.8279) and WC in Blacks (β = 0.4037). In addition, the slope for fasting glucose with increasing HOMA-IR was steeper in Koreans (β = 16.5952, P < 0.001) than in other groups. Conclusion The ability of BMI and WC to predict IR and diabetes was highest in Whites, while the ability of HOMA-IR to predict diabetes was highest in Koreans.