Serial measurements of body composition in obese subjects during a very-low-energy diet (VLED) comparing bioelectrical impedance with hydrodensitometry.

Bioelectrical impedance analysis (BIA) is a convenient, inexpensive, and noninvasive technique for measuring body composition. BIA has been strongly correlated with total body water (TBW) and also has been validated against hydrodensitometry (HD). The accuracy and clinical utility of BIA and HD during periods of substantial weight loss remain controversial. We measured body composition in moderately and severely obese patients serially using both methods during a very-low-energy diet (VLED). Mean initial weight in these patients was 116 (+/-30) kg (range, 74-196 kg). Mean weight loss was 24 (+/-13) kg with a decrease in fat mass (FM) by HD of kg (p < 0.001) and a decrease in fat-free mass (FFM) of 3.6 kg (p < 0.05). Loss of FFM is best predicted by the rate (kg/wk) of weight loss (r2 = 0.86, p < 0.0001). FFM, as predicted from BIA equations, was highly correlated with FFM as estimated by HD during all testing sessions (r = 0.92-0.98). Although highly correlated, BIA overestimated FFM relative to HD and this difference appeared to be more pronounced for taller patients with greater truncal obesity. Although the discrepancy was no greater during weight-loss treatment, the level of disagreement was considerable. Therefore, the two methods cannot be used interchangeably to monitor relative changes in body composition in patients with obesity during treatment with VLED. The discrepancy between BIA and HD may be caused by body mass distribution considerations and by perturbations in TBW which affect the hydration quotient for FFM (BIA) and/or which affect the density constants for FFM and FM (HD).