Equations to estimate body composition using bioelectrical impedance in Chilean adults

Background: Equations for the evaluation of fat-free mass (FFM) and fat mass (FM) with Bioelectrical impedance analysis (BIA) were formulated in Caucasian populations. International recommendations suggest that population-specific equations should be formulated. Aim: To validate an equation previously formulated in Chileans adults and compare it to a new equation generated on an independent sample. Material and Methods: In108 adult volunteers aged 38.1 +/- 14.1 years (44% males), with a body mass index (BMI) of 25.1 +/- 4.1 kg/m(2), body composition was measured by BIA (Bodystat) and dual X-ray absorptiometry (DXA: Lunar Prodigy). Body composition estimated using Schifferli equation and BIA were compared with DEXA, by the Bland-Altman method and simple linear regression. Results: FFM and FM measured by DXA were 45.2 +/- 9.8 kg and 29.6 +/- 11.7 % respectively. Resistance was 467.7 +/- 76.3 ohm. Schifferli equation and BIA significantly overestimated FFM by 7.3 and 7.4 kg, respectively. The error was higher for high levels of FFM (slope beta < 1, p < 0.01). Both equations underestimated FM measured by DXA (averages of 7.5 and 7.8%, respectively, p < 0.01), without a differential bias for Schifferli equation, but with a bias in low levels of FM measured with BIA (slope beta < 1, p < 0.01). Estimation biases could be eliminated using the regression coefficients. Conclusions: Both equations behave similarly and have biases, although less with Schifferli. Statistically correcting for biases, the new adjusted equations provide clinically valid estimates of FFM and FM. Equations should not only be population-specific, but also device-specific.