Assessment of fat mass and fat-free mass in Portuguese adults: calibration of single-frequency leg-to-leg bioelectrical impedance with dual-energy x-ray absorptiometry

Abstract

Objectives: Bioelectrical impedance analysis (BIA) is more commonly employed in outpatient evaluations and epidemiological studies as it is cheaper and less time-consuming. Therefore, we aimed to compare fat mass (FM) and fat-free mass (FFM) estimated by BIA and dual-energy x-ray absorptiometry (DXA) in Portuguese adults, and to calibrate BIA estimates.

Methods: We analyzed data from two population-based cohorts: EPIPorto (n = 391; <65 y) and EPITeen (n = 973; 27 y). Both completed single-frequency leg-to-leg BIA (Tanita TBF-300) and DXA (QDR 4500A Hologic). Agreement between BIA and DXA (FM/FFM) was evaluated by Bland and Altman. To calibrate the BIA estimates, the coefficients were estimated by linear regression using univariate (FM or FFM, separately) and multivariate models (FM and FFM in the same model).

Results: The correlation between BIA and DXA was high for FFM and for FM (FFM: r = 0.946/r = 0.954; FM: r = 0.926/r = 0.921, for EPIPorto and EPITeen, respectively). However, BIA underestimated FM (19.4% EPIPorto; 24.3% EPITeen) and overestimated FFM (11.3% EPIPorto; 14.4% EPITeen). Using the multivariate constrained model to have an absolute equal intercept for FFM and FM, the following calibration equations for BIA measurements were obtained for EPIPorto: FFM_calibrated = -5.119 + 0.995 × FFM_BIA; FM_calibrated = +5.119 + 0.980 × FM_BIA; and for EPITeen: FFM_calibrated = -5.707 + 0.984 × FFM_BIA; FM_calibrated = +5.707 + 0.975 × FM_BIA.

Conclusions: FM assessed by single-frequency leg-to-leg BIA was largely underestimated and FFM overestimated in comparison to DXA, suggesting the use of BIA model-specific calibration equations to calibrate FM and FFM estimates in clinical settings and research studies.

Keywords: Absorptiometry; Adults; Bioelectrical impedance; Body composition; Calibration.