Abstract
The purpose of this study was to compare the relationships of anthropometric, bioelectrical impedance analysis (BIA), and near infrared interactance (NIR) measurements with a multiple-component (MC) criterion estimate of body composition derived from body density (D), body water (W), and bone mineral (B) in 48 white adults aged 49-80 yr. Relative errors of predicting the MC criterion from the practical measurements were determined by simple regressions within gender and calculated as the SEE divided by the criterion mean and expressed as a percentage. Relative errors were lowest for the BIA variable, height2/resistance (4.8-5.0%), higher for body mass index and the sum of 10 skinfold thicknesses (7.0-14.5%), and highest for NIR-derived optical density readings at the biceps and the sum of 10 sites (10.8-15.8%). Due to the low relative prediction error for height2/resistance, sex-specific BIA formulas for estimating fat-free mass from D, W, and B (FFM-DWB) were developed. The SEEs for predicting FFM-DWB from BIA, weight, and age were both 1.5 kg in women and men. Because BIA is not limited to ambulatory subjects, it is concluded that BIA may be a particularly useful, practical technique for estimating body composition in older adults.