Estimating age- and height-specific percentile curves for children using GAMLSS in the IDEFICS study

Abstract

In medical diagnostics age-specific reference values are needed for assessing the health status of children. However, for many clinical parameters such as blood cholesterol or insulin reference curves are still missing for children. To fill this gap, the IDEFICS study provides an excellent data base with 18,745 children aged 2.0–10.9 years. The generalised additive model for location, scale and shape (GAMLSS) was used to derive such reference curves while controlling for the influence of various covariates on the parameters of interest. GAMLSS, an extension of the LMS method, is able to model the influence of more than one covariate. It is also able to model the kurtosis using different distributions. The Bayesian information criterion (BIC), Q-Q plots and wormplots were applied to assess the goodness of fit of alternative models. GAMLSS has proven to be a useful tool to model the influence of more than one covariate when deriving ageand sex-specific percentile curves for clinical parameters in children. This will be demonstrated exemplarily for the bone stiffness index (SI) where age- and heightspecific percentile curves were calculated for boys and girls based on the model which showed the best goodness of fit.