Multiple frequency bioelectrical impedance analysis: a cross-validation study of the inductor circuit and Cole models

It has been proposed that multiple frequency bioelectrical impedance models of the human body should include an inductive property for the circulatory system, the inductor circuit model (ICM), and that such a model, when coupled with a new method of data analysis, can improve the predictive power of multiple frequency bioelectrical impedance analysis (MFBIA). This hypothesis was tested using MFBIA measurements and gold standard measures of total body and extracellular water volumes in a cross-validation study in two subject groups (viz. controls and HIV). The MFBIA measurements were analysed using the current, widely accepted Cole model and the alternative ICM model which includes an inductive component. Correlations in the range 0.75 to 0.92 (for TBW) and 0.46 to 0.79 (for ECW) for impedance quotients versus gold standard measures within the subject groups were observed. These decreased, to as low as r = 0.50 for TBW and r = 0.29 for ECW, when the derived algorithms were applied to the alternative subject group. These results suggest that lack of portability of MFBIA algorithms between subject groups is not due to an inadequacy of the analogue circuit model per se but is possibly due more to fundamental flaws in the principles associated with its application. These include assuming a constant proportionality of body segment geometries and tissue fluid resistivities. This study has also demonstrated that this inadequacy cannot be overcome by simply introducing an inductive component into the analogue electrical circuit.