Bioelectrical impedance is an accurate method to assess body composition in obese but not severely obese adolescents

The aim of this study was to compare total and segmental body composition results between bioimpedance analysis (BIA) and dual x-ray absorptiometry (DXA) scan and to test the reproducibility of BIA in obese adolescents. We hypothesized that BIA offers an accurate and reproducible method to assess body composition in adolescents with obesity. Whole-body and segmental body compositions were assessed by BIA (Tanita MC-780) and DXA (Hologic) among 138 (110 girls and 28 boys) obese adolescents (Tanner stage 3-5) aged 14±1.5years. The BIA analysis was replicated on 3 identical occasions in 32 participants to test the reproducibility of the methods. Whole-body fat mass percentage was significantly higher using the BIA method compared with DXA (40.6±7.8 vs 38.8±4.9%, P<.001), which represents a 4.8% overestimation of the BIA technique compared with DXA. Similarly, fat mass expressed in kilograms is overestimated by 2.8% using BIA (35.8±11.7kg) compared with the DXA measure (34.3±8.7kg) (P<.001), and fat-free mass is underestimated by -6.1% using BIA (P<.001). Except for the right arm and leg percentage of fat mass, all the segmental measures of body composition are significantly different between the 2 methods. Intraclass correlation coefficient and Lin coefficient showed great agreement and concordance between both methods in assessing whole-body composition. Intraclass correlation coefficient between the 3 BIA measures ranged from 0.99 to 1 for body weight, body fat, and fat-free mass. Bioimpedance analysis offers an acceptable and reproducible alternative to assess body composition in obese adolescents, with however a loss of correlation between BIA and DXA with increasing body fat; its validity remains uncertain for segmental analysis among obese youth.