OBJECTIVE: To validate the ability of DXA to measure total body composition before and after weight loss and the composition of the lost body mass.

DESIGN: Cross sectional and follow-up study of body composition before and after a weight loss of 10.6 +/- 6.8 kg.

SUBJECTS: 31 obese subjects with a mean body weight of 105.2 +/- 15.2 kg at baseline, and a mean body weight of 94.6 +/- 16.5 kg at follow-up.

MEASUREMENTS: Body composition was measured by dual X-ray absorptiometry, total body potassium counting, and high precision scales before and after a weight loss.

RESULTS: DXA and the scales showed a strong linear relation (r = 1). At baseline, however, DXA underestimated the body weight by a maximum of 3.2 kg because the subjects were too large for the scan table. After weight loss body weight measurements were accurate. Before and after weight loss the linear relations between DXA and TBK for FFM were strong (r = 0.92 and 0.93). Bland and Altman plots showed limits of agreement of +/-9 kg before and after weight loss; DXA underestimated FFM in women and overestimated FFM in men. DXA accounted for 80% of the lost body weight. The composition of the lost body mass did not differ from that estimated by TBK (7.6% FFM and 92.4% FM by TBK; 11% FFM and 89% FM by DXA).

CONCLUSION: DXA estimates accurately the body composition and the composition of weight loss in groups of obese subjects. However, the scan table may be too small for patients weighing more than 95 kg.