Can bioelectrical impedance improve the prediction of body fat in patients with eating disorders?

Prediction equations for percentage body fat have been derived for a group of 93 female patients being treated for eating disorders, using fat measured by dual-energy X-ray absorptiometry (DXA) as the reference method. Separate bioelectrical impedance analysis (BIA) prediction equations were derived for patients with a body mass index above and below 16 kg/m2. This provided an improvement compared with the equations supplied by the manufacturer. The standard error of the estimate (SEE) for the entire group was 3.2% (r = 0.922). However, an alternative prediction equation based on anthropometry alone gave an SEE of 4.2% (r = 0.861). The ability of BIA and anthropometry alone to measure changes in percentage fat was assessed in a separate group of 24 females during treatment for eating disorders. The SEEs using the BIA and anthropometry prediction equations were 2.6% (r = 0.931) and 2.7% (r = 0.901) respectively. We conclude that in patients with eating disorders BIA does not provide any significant improvement over simple anthropometry prediction equations for the assessment of percentage fat or its changes. The apparently reasonable agreement between fat predicted by BIA and that measured by DXA is largely due to the inclusion of body habitus parameters in the BIA prediction equations.