Comparison of dual-energy x-ray absorptiometry to four other methods to determine body composition in underweight patients with chronic gastrointestinal disease
Kv. Haderslev et M. Staun, Comparison of dual-energy x-ray absorptiometry to four other methods to determine body composition in underweight patients with chronic gastrointestinal disease, METABOLISM, 49(3), 2000, pp. 360-366
Assessment of body composition may provide important information about the
nutritional status. The applicability of two safe and convenient methods fo
r body composition analysis, bioelectrical impedance analysis (BIA) and dua
l-energy x-ray absorptiometry (DXA), in underweight patients with chronic g
astrointestinal disease has been sparsely elucidated. Our objective was to
compare measurements by DXA with four other methods. Furthermore, we compar
ed total body water (TBW) by BIA using three different BIA equations with m
easurement of TBW by tritium dilution (TBW-(H2O)-H-3). Nineteen clinically
stable underweight patients with chronic gastrointestinal disease were incl
uded in the study (body mass index [BMI], 19.3 +/- 1.2 kg/m(2)). Body compo
sition was assessed using total body potassium (TBK), isotope dilution of t
ritium ((H2O)-H-3), anthropometry (skinfold thickness [SF]), BIA, and DXA.
Fat-free mass (FFM) by DXA was in reasonable agreement with body compositio
n measurements by TBK (mean difference(TBK.DXA) = -1.61 kg, r = .88, standa
rd error of the estimate [SEE] = 4.66 kg) and (H2O)-H-3 (mean difference(3H
2O.DXA) = 0.98 kg, r = .93, SEE = 3.34 kg). Although mean values for FFM by
DXA differed significantly versus BIA and SF, we found highly significant
correlations between the measurements (r = .97 and r = .97, respectively).
The mean TBW by BIA was overestimated by 1.9 and 3.1 L compared with TBW-(H
2O)-H-3 when prediction equations for normal-weight subjects were used. We
conclude that the DXA method is a valuable addition to the list of methods
available for body composition studies in clinically stable underweight pat
ients. Our data show that BIA equations for normal-weight subjects overesti
mated TBW in the patients studied. Copyright (C) 2000 by W.B. Saunders Comp
any.