Abstract

Three commercially available bioimpedance spectroscopy analysers were compared for technical performance and for their design purpose of prediction of body composition. All three analysers were electronically stable, remaining in calibration over a year, and provided highly reproducible (coefficients of variation < 0.5%) and accurate (within 0.5% of component values) measurements of impedances of a test circuit. Whole body impedances in humans were highly correlated between all three instruments although significant biases between instruments were observed, particularly for the measurement of intracellular resistance. However, when the measured impedances, and using instrument-specific resistivity coefficients, were used to predict fat-free mass of the subjects, the difference between instruments was approximately 1.7 kg fat-free mass, a value comparable with that observed for the error associated with reference methods such as multi-compartment models of body composition. It is concluded that, with appropriate regard to standardisation of measurement protocol, all three analysers are suitable for their design purpose of estimating body composition in humans.