Comparison of three bioelectrical impedance methods with DXA in overweight and obese men.

Abstract

To compare bioelectrical impedance analysis (BIA) of body composition using three different methods against DXA in overweight and obese men. Forty-three healthy overweight or obese men (ages 25 to 60 years; BMI, 28 to 43 kg/m(2)) underwent BIA assessment of body composition using the ImpediMed SFB7 (version 6; ImpediMed, Ltd., Eight Mile Plains, Queensland, Australia) in multifrequency mode (Imp-MF) and DF50 single-frequency mode (Imp-SF) and the Tanita UltimateScale (Tanita Corp., Tokyo, Japan). Validity was assessed by comparison against DXA using linear regression and limits of agreement analysis. All three BIA methods showed good relative agreement with DXA [Imp-MF: fat mass (FM), r(2) = 0.81; fat-free mass (FFM), r(2) = 0.81; percentage body fat (BF%), r(2) = 0.69; Imp-SF: FM, r(2) = 0.65; FFM, r(2) = 0.76; BF%, r(2) = 0.40; Tanita: BF%, r(2) = 0.44; all p < 0.001]. Absolute agreement between DXA and Imp-MF was poor, as indicated by a large bias and wide limits of agreement (bias, +/-1.96 standard deviation; FM, -6.6 +/- 7.7 kg; FFM, 8.0 +/- 7.1 kg; BF%, -7.0 +/- 6.6%). Imp-SF and Tanita exhibited a smaller bias but wide limits of agreement (Imp-SF: FM, -1.1 +/- 8.5 kg; FFM, 2.5 +/- 7.9 kg; BF%, -1.7 +/- 7.3%; Tanita: BF%, 1.2 +/- 9.5%). Compared with DXA, Imp-MF produced large bias and wide limits of agreement, and its accuracy estimating body composition in overweight or obese men was poor. Imp-SF and Tanita demonstrated little bias and may be useful for group comparisons, but their utility for assessment of body composition in individuals is limited.