Concordance between muscle mass assessed by bioelectrical impedance analysis and by dual energy X-ray absorptiometry: a cross-sectional study

Background: Besides magnetic resonance imaging, dual energy X-ray absorptiometry (DXA) seems the most reliable tool to evaluate body composition and is often considered as the gold standard in clinical practice. Bioelectrical impedance analysis (BIA) could provide a simpler, portative, and less expensive alternative. Because the body composition assessment by BIA is device-dependent, the aim of this study was to appraise the concordance between the specific bioelectrical impedance device InBody S10 and DXA for the body composition evaluation. Methods: Body composition, included appendicular lean mass divided by height squared (ALM/ht(2)) was measured by DXA (Hologic QDR Discovery device) and by BIA (InBody S10 Biospace device). Agreement between tools was assessed by means of the Bland Altman method and reliability was determined using the IntraClass Coefficient (ICC). ICC was also computed to assess the reliability of the test-retest performed by the same operator or by two different ones. Results: A total of 219 subjects were enrolled in this study (mean age: 43.7 +/- 19.1 years old, 51.6% of women). For the ALM/ht(2), reliability of the test-retest of the BIA was high with an ICC of 0.89 (95% CI: 0.86-0.92) when performed by the same operator and an ICC of 0.77 (95% CI: 0.72-0.82) when performed by two different operators. Agreement between ALM/ht(2) assessed by DXA and BIA was low (ICC = 0.37 (95% CI: 0.25-0.48)). Mean ALM/ht(2) was 9.19 +/- 1.39 kg/m(2) with BIA and 7.34 +/- 1.34 kg/m(2) with DXA, (p < 0001). A formula developed using a multiple regression analysis, and taking into account muscle mass assessed by BIA, as well as sex and body mass index, explains 89% of the ALM/ht(2) assessed by DXA. Conclusions: Although our results show that the measure of ALM/ht(2) by BIA is reliable, the agreement between DXA and BIA is low. Indeed, BIA seems to overestimate ALM/ht(2) compared to DXA and, consequently, it is important to use an adapted formula to obtain measurement of the appendicular lean mass by BIA close to that measured by DXA.