Can anthropometric field measures be used to improve the predictive value of body mass index for chronic disease risk in multi-ethnic populations

Background. Overweight and obesity are associated with increases in chronic disease risk and all-cause mortality. Although BMI is widely used as an indicator of adiposity, it is an imprecise measure of excess body fat across multi-ethnic populations.;Objective. To develop anthropometric prediction equations for DXA-derived percent body fat and biomarkers of chronic disease risk (serum insulin and HDL cholesterol), and validate across race/ethnic groups.;Study design. Cross-sectional secondary analyses of Mexican American, Non-Hispanic White, and Non-Hispanic Black men and women 18 years of age and older (N=7, 109), examined in the National Health and Nutrition Examination Survey 2003--2004.;Methods. Equations were developed to predict percent body fat and serum insulin and HDL cholesterol from anthropometric measurements, using multiple linear regression analyses. Anthropometric measurements included body mass index, waist circumference, arm circumference, calf circumference, thigh circumference, subscapular skinfold, triceps skinfold, upper arm length, and upper leg length. The equations were assessed for validity across race/ethnic groups.;Results. Several equations combining anthropometric measurements explained greater than 80% (men) and 70% (women) of the variance in percent body fat. The most powerful predictions across race/ethnic groups included waist circumference, triceps skinfold, and arm circumference (men only), which explained 80.0% and 70.0% of the variance in percent body fat among men and women, respectively. Percent body fat for men = 2.142 + 0.310 waist circumference (cm) - 0.361 arm circumference (cm) + 0.403 triceps skinfold (mm) + 0.027 age (yr); percent body fat for women = 7.168 + 0.221 waist circumference (cm) + 0.338 triceps skinfold (mm) + 0.071 age (yr). Anthropometric measurements did not give good predictions of serum insulin and HDL cholesterol.;Conclusion. These anthropometric-derived prediction equations can serve as useful tools for identifying excess body fat among multi-ethnic populations in field and healthcare settings.