Development and Cross-Validation of a Simple Model to Estimate Percent Body Fat in Persons with Multiple Sclerosis.

ABSTRACT

BACKGROUND: Persons with multiple sclerosis (MS) have higher body composition variability compared with the general population. Monitoring body composition requires accurate methods for estimating percent body fat (%BF). We developed and cross-validated an equation for estimating %BF from body mass index (BMI) and sex in persons with MS. METHODS: Seventy-seven adults with MS represented the sample for the equation development. A separate sample of 33 adults with MS permitted the equation cross-validation. Dual-energy x-ray absorptiometry (DXA) provided the criterion %BF. RESULTS: The model including BMI and sex (mean ± SD age women, 49.2 ± 8.8 years; men, 48.6 ± 9.8 years) had high predictive ability for estimating %BF (P < .001, R2 = 0.77, standard error of estimate = 4.06%). Age, MS type, Patient-Determined Disease Steps score, and MS duration did not improve the model. The equation was %BF = 3.168 + (0.895 × BMI) - (10.191 × sex); sex, 0 = woman; 1 = man. The equation was cross-validated in the separate sample (age women, 48.4 ± 9.4 years; men, 43.8 ± 15.4 years) based on high accuracy as indicated by strong association (r = 0.89, P < .001), nonsignificant difference (mean 0.2%, P > .05), small absolute error (mean 2.7%), root mean square error (3.5%), and small differences and no bias in Bland-Altman analysis (mean difference 0.2%, 95% CI -6.98 to 6.55, rs = -0.07, P = .702) between DXA-determined and equation-estimated %BF. CONCLUSIONS: Health care providers can use this developed and cross-validated equation for estimating adiposity in persons with MS when DXA is unavailable.