Cross-validation of generalised body composition equations with diverse young men and women: the Training Intervention and Genetics of Exercise Response (TIGER) Study

Abstract

Generalised skinfold equations developed in the 1970s are commonly used to estimate laboratory-measured percentage fat (BF%). The equations were developed on predominately white individuals using Siri's two-component percentage fat equation (BF%-GEN). We cross-validated the Jackson–Pollock (JP) generalised equations with samples of young white, Hispanic and African–American men and women using dual-energy X-ray absorptiometry (DXA) as the BF% referent criterion (BF%-DXA). The cross-sectional sample included 1129 women and men (aged 17–35 years). The correlations between BF%-GEN and BF%-DXA were 0·85 for women and 0·93 for men. Analysis of measurement error showed that BF%-GEN underestimated BF%-DXA of men and women by 1·3 and 3·0 %. General linear models (GLM) confirmed that BF%-GEN systematically underestimated BF%-DXA of Hispanic men and women, and overestimated BF%-DXA of African–American men. GLM were used to estimate BF%-DXA from the JP sum of skinfolds and to account for race/ethnic group bias. The fit statistics (R and standard error of the estimate; see) of the men's calibration model were: white, R 0·92, see 3·0 %; Hispanic, R 0·91, see 3·0 %; African–American, R 0·95, see 2·6 %. The women's statistics were: white and African–American, R 0·86, see 3·8 %; Hispanic, R 0·83, see 3·4 %. These results showed that BF%-GEN and BF%-DXA were highly correlated, but the error analyses documented that the generalised equations lacked accuracy when applied to these racially and ethnically diverse men and women. The inaccuracy was linked to the body composition and race/ethnic differences between these Training Intervention and Genetics of Exercise Response (TIGER) study subjects and the men and women used to develop the generalised equations in the 1970s and using BF%-DXA as the referent criterion.