RESUMO
OBJECTIVES: The aim of the study was to examine the agreement of multifrequency bioelectric impedance analysis (BIA) and anthropometry with reference methods for body composition assessment in children with intestinal failure (IF). METHODS: We conducted a prospective pilot study in children 14 years or younger with IF resulting from either short bowel syndrome or motility disorders. Bland-Altman analysis was used to examine the agreement between BIA and deuterium dilution in measuring total body water (TBW) and lean body mass (LBM), and between BIA and dual-energy x-ray absorptiometry (DXA) techniques in measuring LBM and fat mass (FM). FM and percent body fat (%BF) measurements by BIA and anthropometry were also compared in relation to those measured by deuterium dilution. RESULTS: Fifteen children with IF, median (interquartile range) age 7.2 (5.0, 10.0) years, and 10 (67%) boys, were studied. BIA and deuterium dilution were in good agreement with a mean bias (limits of agreement) of 0.9 (-3.2 to 5.0) for TBW (L) and 0.1 (-5.4 to 5.6) for LBM (kg) measurements. The mean bias (limits) for FM (kg) and %BF measurements were 0.4 (-3.8 to 4.6) kg and 1.7 (-16.9 to 20.3)%, respectively. The limits of agreement were within 1 standard deviation of the mean bias in 12 of 14 (86%) subjects for TBW and LBM, and in 11 of 14 (79%) for FM and %BF measurements. Mean bias (limits) for LBM (kg) and FM (kg) between BIA and DXA were 1.6 (-3.0 to 6.3) kg and -0.1 (-3.2 to 3.1) kg, respectively. Mean bias (limits) for FM (kg) and %BF between anthropometry and deuterium dilution were 0.2 (-4.2 to 4.6) and -0.2 (-19.5 to 19.1), respectively. The limits of agreement were within 1 standard deviation of the mean bias in 10 of 14 (71%) subjects. CONCLUSIONS: In children with IF, TBW and LBM measurements by multifrequency BIA method were in agreement with isotope dilution and DXA methods, with small mean bias and clinically acceptable limits of agreement. In comparison with deuterium dilution, BIA was comparable to anthropometry for FM and %BF assessments with small mean bias, but the limits of agreement were large. BIA is a reliable method for TBW and LBM assessments in population studies; however, its reliability in individual patients, especially for FM assessments, cannot be guaranteed.
