Treatment with antipsychotic agents may result in changes in body composition. Highly accurate measurement techniques are expensive and may be associated with safety concerns, e.g. radiation exposure. Cheaper alternatives are available, but their accuracy in the patient population of interest has been investigated little.

To compare two methods for estimating body fat % (%BF) in patients treated with atypical antipsychotics.

Data on %BF measured at baseline and 1 month in six patients participating in a randomized trial of olanzapine and sertindole were collected. Eight-electrode bio-electrical impedance (BIA8) and dual energy x-ray absorptiometry (DEXA) equipments were used to measure %BF.

At baseline, the mean %BFBIA8 was 26% compared to 35% for DEXA, indicating a large underestimation by BIA8. After one month, the means were unchanged, although individual patients changed between +1.7 to -1.5%BFBIA, and +1.5 to -1.3%BFDEXA. The assessed median change at one month was similar for the two methods, with an increase of 0.3%BFBIA compared to 0.2%BFDEXA. The median between methods difference was 0%BF (range -3.0 to +1.3).

Large discrepancies in absolute levels of %BF were seen between the two measurement methods. The discrepancies were, however, constant over time. Therefore the change estimates were almost identical. Judging from this small sample, it appears as assessment of change in body composition may be estimated using the cheaper and faster BIA method, although the absolute values may be large underestimates. Caution in the interpretation must be exercised due to the small sample and small magnitude of change.