The aim of this study was to determine the level of agreement between body composition measurements by dual-energy X-ray absorptiometry (DXA), single-frequency bioelectrical impedance analysis (BIA) and multifrequency bioelectrical impedance spectroscopy (BIS). Fat-free mass (FFM), body fat mass and body fatness (percentage fat) were measured by DXA, BIA and BIS in 61 healthy children (37M, 24F, aged 10.9-13.9 y). Estimates of FFM, body fat mass and body fatness were highly correlated (r = 0.73-0.96, p < 0.0001) between the different methods. However, a Bland-Altman comparison showed wide limits of agreement between the methods. The mean differences between methods for FFM ranged from -2.31 +/- 7.76 kg to 0.48 +/- 7.58 kg. Mean differences for body fat mass ranged from 0.16 +/- 5.06 kg to 2.95 +/- 5.65 kg and for body fatness from -2.3 +/- 7.8% to 0.8 +/- 9.3%. Calculations of body composition with BIS were not superior to BIA. However, BIA overestimated fat mass in lean, subjects and underestimated fat mass in overweight subjects more than BIS, compared with DXA.
Conclusion: The methods used provided estimates of FFM, body fat mass and body fatness that were highly correlated in a population of healthy children. However, the large limits of agreement derived from the Bland-Altman procedure suggest that the methods should not be used interchangeably.