Making the choice between bioelectrical impedance measures for body hydration status assessment

Abstract

Situational or persistent body fluid deficit (i.e., de- or hypo-hydration) is considered a significant health risk factor. Bioimpedance analysis (BIA) has been suggested as an alternative to less reliable subjective and biochemical indicators of hydration status. The present study aimed to compare various BIA models in the prediction of direct measures of body compartments associated with hydration/osmolality. Fish (n = 20) was selected as a biological model for physicochemically measuring proximate body compartments associated with hydration such as water, dissolved proteins, and non-osseous minerals as the references or criterion points. Whole-body and segmental/local impedance measures were used to investigate a pool of BIA models, which were compared by Akaike Information Criterion in their ability to accurately predict the body components. Statistical models showed that 'volumetric-based' BIA measures obtained in parallel, such as distance²/R_p, could be the best approach in predicting percent of body moisture, proteins, and minerals in the whole-body schema. However, serially-obtained BIA measures, such as the ratio of the reactance to resistance and the resistance adjusted for distance between electrodes, were the best fitting in predicting the compartments in the segmental schema. Validity of these results should be confirmed on humans before implementation in practice.

Figure 1

Derivations from four dorsal (by contact gold electrodes, 5 × 10 mm) and four ventral (by stainless steel subdermal needle electrodes, d = 0.35 mm and length = 10 mm) points of two-side segments of the body of fish at dorsal and ventral rows. Space on vertical line between paired on one side electrodes, including signal-emitting (higher) and signal-detecting (lower) electrodes, was kept constant (3 mm between closest edges for contact and 7.5 mm for needle electrodes). Distances between each paired sets of electrodes (i.e., width between sides of the segments) were individually varied for every segment within and between subjects (see Table 1).