Estimation of Body Composition Depends on Applied Device in Patients Undergoing Major Abdominal Surgery

Background: Bioelectrical impedance analysis (BIA) is a method used to estimate body compartments such as fat-free mass (FFM) and fat mass (FM). Two BIA devices, a single-frequency BIA (SF-BIA) device and a bioimpedance spectroscopy (BIS) approach, were compared to evaluate their reliability and to study whether their estimations resulted in similar classifications of body composition. Materials and Methods: In a prospective observational study, body composition was estimated by SF-BIA and BIS in 123 adult patients scheduled for major abdominal surgery. Measurement agreement for the continuous variables FFM and FM were analyzed by using the intraclass correlation coefficient (ICC), the mean differences, and their limits of agreement. Measurement differences were also visualized by Bland-Altman plots. For the dichotomized FFM index (FFMI) and FM index (FMI), interobserver agreement was calculated using Cohen kappa statistics; the McNemar test was performed to compare the paired proportions. Results: Agreement for the continuous variables was almost perfect for FM (.86; 95% confidence interval [CI],.80-.90) and substantial for FFM (.78; 95% CI, .70-.84). For the dichotomous variables, the agreement was substantial for FMI (.67; 95% CI,.51-.83) and slight for FFMI (.19; 95% CI, .01-.37). BIS classified a larger proportion having a low FFMI and a high FMI. Conclusion: There were good ICCs between SF-BIA and BIS for FFM and FM. However, the mean differences were substantial, whereas the classification of body composition based on FFMI and FMI was influenced by the device. Therefore, BIA devices are not interchangeable.