Changes in body composition during weight loss in obese subjects in the NUGENOB study: Comparison of bioelectrical impedance vs. dual-energy X-ray absorptiometry

Abstract

Aim We studied the accuracy of bioelectrical impedance analysis (BIA) to assess changes in body composition during moderate weight loss in obese subjects. Methods Estimates of changes in fat mass (FM) and fat-free mass (FFM) by BIA were compared with those by dual-energy X-ray absorptiometry (DXA) as the reference method during a 10-week standardized weight-loss intervention. In obese women (age: 20–50 years, mean BMI: 33.8 kg/m 2) participating in a European multicentre trial (nutrient–gene interactions in human obesity [NUGENOB]), body composition was assessed by BIA (Bodystat QuadScan 4000) and DXA (Lunar DPX-IQ at two centres, Hologic QDR 2000 at another centre) at baseline ( n = 131) and at week 10 ( n = 105) after a mean weight loss of −5.7 kg. Results At baseline, BIA significantly overestimated FFM and underestimated FM (by 1–3 kg on average) compared with DXA, and the limits of agreement were wide (mean ± 7–8.5 kg). For body-composition changes, although biases were generally non-significant, the limits of agreement were also wide (mean ± 3.7–4.6 kg). An FFM prediction equation for BIA data was developed in subjects scanned with Lunar instruments and cross-validated in an independent sample of 31 obese women undergoing similar weight loss. However, no major improvement in limits of agreement was found. Conclusion During moderate diet-induced weight loss, the use of BIA leads to estimates of changes in body composition at the individual level that can differ substantially from those assessed by DXA, indicating that BIA and DXA cannot be used interchangeably. However, BIA in this context may be used for assessing changes in body composition at group level.