Estimation of body fat in children with intellectual disability: development and cross-validation of a simple...

Population-level monitoring of body composition requires accurate, biologically-relevant, yet feasible methods for estimating percent body fat (%BF). The aim of this study was to develop and cross-validate an equation for %BF from Body Mass Index (BMI), age, and sex among children with intellectual disability (ID). This study further aimed to examine the performance of an existing BMI-based equation (Deurenberg equation) for %BF in children with ID.

Participants were 107 children (63 boys; aged 6-15 years) with ID randomly allocated to development (n = 81) and cross-validation (n = 26) samples. Dual-Energy X-Ray Absorptiometry provided the criterion %BF.

The model including BMI, age, and sex (0 = male; 1 = female) had a significant goodness-of-fit in determining %BF (p < 0.001; R² = 0.69; SEE =5.68%). The equation was: %BF = - 15.416 + (1.394 × BMI) + (4.538 × age) - (0.262 × age²) + (5.489 × sex). The equation was cross-validated in the separate sample based on (i) strong correlation (r = 0.82; p < 0.001) and non-significant differences between actual and predicted %BF (28.6 ± 9.6% and 30.1 ± 7.1%, respectively); (ii) mean absolute error (MAE) = 4.4%; and (iii) reasonable %BF estimations in Bland-Altman plot (mean: 1.48%; 95% CI: 12.5, -9.6). The Deurenberg equation had a large %BF underestimation (mean: -7.1%; 95% CI: 5.3, -19.5), significant difference between actual and estimated %BF (28.6 ± 9.7% and 21.5 ± 7.0%, respectively; p < 0.001), and MAE = 8.1%.

The developed equation with BMI, sex, and age provides valid %BF estimates for facilitating population-level body fat screening among children with ID.