PIH67 Successful Development of Annual and Long Term Prediction Models to Estimate Height Outcome following Growth Hormone (GH) Therapy in Children Using Data from KIGS – A Large Pharmacoepidemiological Survey

Abstract

Annual growth prediction models following growth hormone (GH) treatment have been developed to facilitate treatment guidance. However, accurately predicting height over the long term, during pre-pubertal treatment years has not been assessed and is a prerequisite for modelling of cost effective optimum height outcomes. Annual prediction models utilised data from large cohorts sourced from the KIGS database (Pfizer International Growth Database, comprising 75,000 children with growth disorders) and describe the likely annual height gain based on patients' auxological and biochemical characteristics (e.g. GH dose, age, mid parental height standard deviation [SDS] and weight SDS score) at treatment start. The most likely long-term height development was simulated prospectively up to 4 years by sequential application of existing yearly prediction algorithms for height velocity (HV) and newly developed algorithms for weight gain in pre-pubertal children with idiopathic GH Deficiency (GHD) and Turner Syndrome (TS). The long-term prediction of height was validated in new cohorts of pre-pubertal children with GHD (n= 664) or TS (n=607) from KIGS. When height was simulated from GH start in GHD, the predicted mean (SD) gain after 4 years was 30.4 (3.4) cm; the observed height gain was 30.0 (5.0) cm. In TS the corresponding predicted and observed mean gains were 27.2 (2.2) and 26.5 (3.8) cm. Sequential application of annual KIGS growth prediction models permits accurate simulation of height development during the first four years of GH treatment in GHD and TS and is applicable for patient groups from GH start. Long-term growth simulation helps managing patient's expectations and facilitates an individualised, cost effective growth hormone (GH) therapy in children.