Reliability and Validity of Contemporary Bioelectrical Impedance Analysis Devices for Body Composition Assessment

Abstract

Introduction: The aim was to determine reliability and validity of bioelectrical impedance analysis (BIA) compared to dual energy x-ray absorptiometry (DEXA). Methods:  Participants (n=93) were fasted and euhydrated (confirmed with urine specific gravity, USG) and underwent anthropometrics, DEXA scan (GE Prodigy), and three repeated trials on each BIA device (i.e., InBody 770, SECA mBCA 514, and FitTrack). A subset of participants (n=36) re-tested 12-weeks later to investigate longitudinal changes. Results: All BIA devices had acceptable within-session reliability (coefficient of variation < 2%). Agreement with DEXA was unacceptable, poor, and moderate for FitTrack (Lin’s Concordance Correlation Coefficient, CCC=0.61), InBody (CCC=0.88), and SECA (CCC=0.91). FitTrack did not demonstrate systematic bias but had wide limits of agreement and larger underestimations occurring at higher BF%. InBody showed systematic underestimations with proportionate bias demonstrating more error at lower BF%. SECA demonstrated no bias but tended to underestimate BF% according to Bland-Altman Plots. Errors in BIA BF% estimates were not related with USG but were associated with DEXA measured BF%. The subtle 12-week changes in BF% did not agree between BIA and DEXA. Conclusions: Caution should be taken when using BIA devices to assess BF% as devices demonstrated unacceptable agreement compared to DEXA.