Predictive accuracy of bioelectrical impedance in estimating body composition of Native American women.

The predictive accuracy of race-specific and fatness-specific bioelectrical impedance analysis (BIA) equations for estimating criterion fat-free mass (FFM) derived from two-component (2C) and multicomponent (MC) models was examined. Body density (Db) of Native American women (n = 151) aged 18-60 y was measured by hydrostatic weighing at residual volume. Total body bone ash was obtained by dual-energy, x-ray absorptiometry. Cross-validation of the Rising (5), Segal (3), and Gray (4) equations against FFM2C yielded high correlation coefficients (0.86-0.95) and acceptable SEEs (1.47-2.72 kg). Cross-validation of these equations against criterion FFMMC, with Db adjusted for total body mineral, yielded similar correlation coefficients (0.82-0.94) and SEEs (1.69-2.80 kg). However, each BIA equation significantly overestimated FFMMC. A new race-specific BIA equation based on an MC model was developed: FFMMC = 0.001254(HT2)-0.04904(R) + 0.1555(WT) + 0.1417(Xc) - 0.0833(AGE) + 20.05 (R = 0.864, and SEE = 2.63 kg).