Abstract
Objective. The purpose of this study was to evaluate body composition outcomes of foot-to-foot (FF) bioelectrical impedance (BIA) and dual-energy x-ray (DEXA) in 8 to 10-year-old African-American girls and, if different, to develop and cross-validate specific BIA prediction equations for this at-risk group. Methods. DEXA and FF-BIA body composition outcomes were analyzed in 183, 8–10-year-old African-American girls from the Memphis site of the Girls health Enrichment Multi-site Study (GEMS). Results. Mean body composition outcomes by FF-BIA and DEXA were significantly different (p<0.0001); therefore, population-specific equations were developed and cross-validated using split-sample, cross-validation methods. When equations were used, BIA and DEXA outcomes were significantly correlated (percent body fat [r=0.931], fat mass [r=0.985], and fat-free mass [r=0.944]). Mean predicted BIA measurements for body fat, fat mass, and fat-free mass were essentially equal to their counterpart DEXA measurements (t[182]=− 0.013, p = 0.897, t[182]=− 0.06, p=0.956, and t[182]=− 0.26, p=0.792, respectively). The Bland-Altman analysis revealed a significant slope for percent fat (p=0.009) and slopes approaching significance for fat mass (p=0.07) and fat-free mass (p=0.06). Conclusion. Although FF-BIA and DEXA are not directly interchangeable in young African-American girls, these equations accurately estimated average percent fat, fat mass, and fat-free mass of the cross-validation sample of African-American girls. However, the application of this equation may result in potential underestimation or overestimation of fat with respect to DEXA measures in some populations.