Background/objectives: The validity of bioelectrical impedance analysis (BIA) for body composition analysis is limited by assumptions relating to body shape. Improvement in BIA technology could overcome these limitations and reduce the population specificity of the BIA algorithm.
Subjects/methods: BIA equations for the prediction of fat-free mass (FFM), total body water (TBW) and extracellular water (ECW) were generated from data obtained on 124 Caucasians (body mass index 18.5-35 kg/m(2)) using a four-compartment model and dilution techniques as references. The algorithms were validated in an independent multiethnic population (n=130). The validity of BIA results was compared (i) between ethnic groups and (ii) with results from the four-compartment model and two-compartment methods (air-displacement plethysmography, dual-energy X-ray absorptiometry and deuterium dilution).
Results: Indices were developed from segmental R and Xc values to represent the relative contribution of trunk and limbs to total body conductivity. The coefficient of determination for all prediction equations was high (R(2): 0.94 for ECW, 0.98 for FFM and 0.98 for TBW) and root mean square error was low (1.9 kg for FFM, 0.8 l for ECW and 1.3 kg for TBW). The bias between BIA results and different reference methods was not statistically different between Afro-American, Hispanic, Asian or Caucasian populations and showed a similar difference (-0.2-0.2 kg FFM) when compared with the bias between different two-compartment reference methods (-0.2-0.3 kg FFM).
Conclusions: An eight-electrode, segmental multifrequency BIA is a valid tool to estimate body composition in healthy euvolemic adults compared with the validity and precision of other two-compartment reference methods. Population specificity is of minor importance when compared with discrepancies between different reference methods.