Bioelectrical impedance analysis for body composition: clinical evaluation in general surgical patients.

Body composition was measured in 112 surgical patients and eight normal subjects by bioelectrical impedance analysis (BIA), and the results compared with our standard five-compartment method utilizing in vivo neutron activation analysis and tritium dilution space to measure fat-free mass and total body water. In a population with a mean fat-free mass of 49.2 kg by the latter method, BIA underestimated fat-free mass by 1.3 kg, with a standard deviation (SD) of 4.0 kg. Total body water was overestimated by 0.5 liter, with SD of 0.47 liter, when compared to a mean tritium dilution space of 36.7 liter. Changes in total body water which occurred over a treatment period of at least 1 week were also measured by the two methods. The change as measured by isotopic dilution could be estimated from the change as measured by BIA with a standard error of 2.25 liter. Although the results suggest that BIA should not be used as a research tool, it is concluded that in clinical practice, when nutrient requirements are being estimated, fat-free mass measured by BIA may have a useful place.