Prediction of Fat-Free Mass in Children

Abstract

Background: Fat-free mass (FFM) is an important covariate for predicting drug clearance. Models for predicting FFM have been developed in adults but there is currently a paucity of mechanism-based models developed to predict FFM in children.

Objective: The aim of this study was to develop and evaluate a model to predict FFM in children.

Methods: A large dataset (496 females and 515 males) was available for model building. Subjects had a relatively wide range of age (3-29 years) and body mass index values (12-44.9 kg/m(2)). Two types of models (M1 and M2) were developed to describe FFM in children. M1 was fully empirical and based on a linear model that contained all statistically significant covariates and their interactions. M2 was a simpler model that incorporated a maturation process. M1 was developed to provide the best possible description of the data (i.e. a positive control). In addition, a published adult model (M3) was applied directly as a reference description of the data. The predictive performances of the three models were assessed by visual predictive checks and by using mean error (ME) and root mean squared error (RMSE). A test dataset (90 females and 86 males) was available for external evaluation.

Results: M1 consisted of nine terms with up to second-level interactions. M2 was a sigmoid hyperbolic model based on postnatal age with an asymptote at the adult prediction (M3). For the index dataset, the ME and 95 % CI for M1, M2 and M3 were 0.09 (0.03-0.16), 0.24 (0.14-0.33) and 0.29 (0.06-0.51) kg, respectively, and RMSEs were 1.12 (1.03-1.23), 1.58 (1.46-1.72) and 3.76 (3.54-3.97) kg.

Conclusions: A maturation model that asymptoted to an established adult model was developed for prediction of FFM in children. This model was found to perform well in both male and female children; however, the adult model performed similarly to the maturation model for females. The ability to predict FFM in children from simple demographic measurements is expected to improve understanding of human body structure and function with direct application to pharmacokinetics.