Agreement Between Bia And Dxa In Estimating Body Composition Changes In Obese Adolescents

Bioelectrical Impedance (BIA) and Dual Energy X-Ray Absorptiometry (DXA) are commonly employed techniques for assessing body composition variables. While DXA is thought to be a more valid and reliable technique, it is more costly and availability is often limited. PURPOSE: To compare the level of agreement between BIA and DXA in detecting changes in body composition following a 6-month weight management program. METHODS: Twenty-four adolescents (12.9±1.8 yrs.; 35.2±4.4 kg/m2) voluntarily participated in the program. Body fat percentage (%BF), fat mass (FM), and fat-free mass (FFM) were assessed upon enrollment (PRE) and after 6-months. RESULTS: %BF, FM, and FFM between BIA and DXA were not significantly different (p≥0.075) at enrollment. BIA and DXA PRE %BF was not significantly correlated (r2=0.262, p=0.217), however estimates of Pre FM (r2=0.796, p<0.001) and Pre FFM (r2=0.497, p=0.014) demonstrated good relative agreement. However, there was poor absolute agreement between BIA and DXA as indicated by a large bias and wide level of agreement for each measured variable (bias ± 1.96 S.D.; %BF -4.0±18.9%, FM -4.2±16.9 kg, FFM 4.2±16.9 kg). The changes in body composition as assessed by BIA and DXA were significantly correlated (%BF, r2=0.671, p<0.001; FM, r2=0.912, p<0.001; FFM, r2=0.668, p<0.001). Additionally, due to a small bias and narrow limits of agreement for BIA and DXA changes (%BF 0.4±3.6%, FM 0.1±3.5 kg, FFM -0.1±3.5 kg), there were no significant differences between BIA and DXA when evaluating the changes in %BF, FM, and FFM (p≥0.595). CONCLUSIONS: The similar detected changes in %BF, FM, and FFM between the two techniques suggests that although BIA may not accurately estimate body composition variables, BIA may be a viable alternative to DXA in assessing body composition changes in obese adolescents following 6-months of a weight management program. Financial support was provided by Virginia Premier Health Plan, Inc. and the VCU General Clinical Research Center MO1 RR00065