Body composition estimates from multicomponent models using BIA to determine body water

The purpose of this study was to compare estimates of body fat (%BF) from three- and four-component models with total body water (TBW) determined by single-frequency bioelectrical impedance analysis (BIA; %BF3C-BIA and %BF4C-BIA) to %BF estimates from densitometry (%BF2C-D) and from three- and four-component models with TBW determined using deuterium dilution (%BF3C-D2O and %BF4C-D2O), the criterion methods.

Measures of body density by hydrostatic weighing, TBW by BIA and D2O dilution, and bone mineral by dual energy x-ray absorptiometry (DXA) were obtained in 40 men and 93 women, 18-42 yr. TBW was estimated from BIA resistance (RJL analyzer) using an equation developed and cross-validated in two independent samples. Body fat was estimated using the three-component model of Siri (1961) and a four-component model modified from Lohman (1986).

There was a strong relation and no significant difference between TBW estimated by BIA and D2O [r = 0.94, SEE = 2.4; xDiff = 0.0 +/- 2.4 L (SD), P > 0.05]. There were strong relations between methods for estimating %BF, with deviations from %BF4C-D2O (errors) for %BF3C-BIA [r = 0.99, SEE = 2.4% BF, xDiff = -0.4 +/- 2.4% BF (SD)] and %BF4C-BIA [r = 0.99, SEE = 2.3% BF, xDiff = 0.2 +/- 2.3% BF (SD)] being nonsignificant (P > 0.05) although greater than for %BF3C-D2O [r = 1.00, SEE = 0.5% BF, xDiff = -0.6 +/- 0.5% BF (SD)], and comparable or slightly worse than for %BF2C-D [r = 0.99, SEE = 2.3% BF, xDiff = 0.4 +/- 2.3% BF (SD)].

We conclude that because estimates of %BF from multicomponent models with TBW estimated from BIA are not more accurate than from body density alone using a two-component model, estimates of %BF from three- and four-component models using TBWBIA are not acceptable substitutes for estimates from the same models using TBWD2O.