Accuracy of bioimpedance equations for measuring body composition in a cohort of 2134 patients with obesity

Abstract

Measuring body composition is an important issue to phenotype patients with obesity and to follow the nutritional care efficiency. Bioelectrical Impedance Analysis (BIA) is a simple and rapid technique. However, validity of BIA in patients with obesity remains controversial. Thus, we aimed to evaluate the validity of several BIA equations to assess body composition in a large cohort of patients with obesity by using dual X ray absorptiometry (DXA) as reference. Seven BIA equations have been retrospectively applied on electrical data measured by BIA in patients with obesity with BMI equal or higher than 30kg/m2 and results were compared to DXA-derived fat mass (FM) and fat-free mass (FFM). BIA and DXA were done the same day after an overnight fasting. Results were compared with Bland-Altman method and Pearson correlation. We also calculated the accuracy defined as the percentage of patients with DXA-BIA difference within ± 10% of DXA measures for FFM and FM. Data from 2134 patients with class I and II obesity (ob1/2, n=1452, 47.4±14.2y; 35.0±2.7kg.m-2) and class III obesity (ob3, n=682, 48.2±13.9y; 44.1±3.5kg.m-2) were analyzed. The best results to evaluate FFM both in ob1/2 and ob3 groups were obtained with Roubenoff's equation: Bland Altman bias at-1.96 and-0.82kg, Pearson correlation r at 0.93 and 0.87, accuracy at 75.7% and 83.3%, respectively. However, limits of agreements at 95% were high: [-9.42; 5.49kg] and [-8.16; 6.52kg]. For FM evaluation, Roubenoff's equation also showed best results for ob1/2 group (bias at-1.17kg; correlation r at 0.89 and accuracy at 79.1%) but not for ob3 group. In this latter group, Deurenberg's equation exhibited the best results (bias at 2.09kg; correlation r at 0.81 and accuracy at 76.8%). However, limits of agreements remained high. In patients with obesity, Roubenoff BIA equation should be recommended to assess fat free mass, even if limits of agreements remain high.