Predicting the growth response of children with idiopathic growth hormone deficiency to one year of recombinant growth hormone treatment: derivation and validation of a useful method.

Patients with idiopathic growth hormone (GH) deficiency (iGHD) show a wide interindividual variability in their response to recombinant GH (rGH) therapy. Accurate growth prediction would provide clinicians a helpful tool for planning and optimizing growth promoting therapy. We develop and validate a model to predict the growth response, using accessible parameters and simple calculations. Data of 132 prepubertal patients with iGHD, receiving rGH treatment for at least one year, were collected. The sample was split into two statistically similar groups: a derivation set (100 children) and a validation set (32 children). Linear multiple regression, stepwise analysis, was used to derive the model. Validation was performed by calculating the difference in predictive power when applying it to the validation group ('shrinkage'), and analyzing the differences between the predicted and observed responses ('studentized residual'). The obtained prediction model predicts 97.8% of the variability of height achieved after one year of therapy (error SD = 1.767 cm). All data integrated in our model are accessible to the clinician when evaluating the patient and the calculations are fast and simple. It is an accurate method that fulfills all the requirements for becoming a useful tool to predict growth response in prepubertal children with iGHD and should be helpful optimizing rGH therapy.