Related Experiment Videos

Validated multivariate models predicting the growth response to GH treatment in individual short children with a

K A Wikland1, B Kriström, S Rosberg

  • 1International Pediatric Growth Research Center, Department of Pediatrics, University of Göteborg, SE-41685 Göteborg, Sweden.

Pediatric Research
|September 27, 2000
PubMed

Insights

New models predict individual growth responses to growth hormone (GH) therapy in children. These models offer more accurate predictions than standard tests, aiding treatment decisions for short stature.

Area of Science:

  • Pediatric Endocrinology
  • Biostatistics
  • Growth Hormone Therapy

Background:

  • Accurate prediction of growth hormone (GH) therapy response is crucial for optimizing treatment in children with growth disorders.
  • Current methods, such as provocation tests, have limitations in predicting individual patient outcomes.
  • Developing robust predictive models can enhance evidence-based decision-making in pediatric growth management.

Purpose of the Study:

  • To develop and validate predictive models for individual children's growth responses to GH therapy.
  • To compare the accuracy of different models incorporating various clinical and biochemical parameters.
  • To establish a tool that assists in selecting appropriate candidates for GH treatment.

Main Methods:

  • Constructed predictive models using nonlinear multivariate data fitting on a cohort of 269 prepubertal children with growth deficiency.
  • Utilized clinical data including auxological measurements, parental heights, early-life growth, IGF-I levels, and GH secretion profiles (provocation tests and 24-hour profiles).
  • Validated model performance by comparing predicted versus observed growth responses in an independent validation cohort of 149 children.

Main Results:

  • The "Basic model" (pre-treatment auxology, parental height) achieved an SD(res) of 0.28 SDscores for the first year's growth.
  • Incorporating 24-hour GH profiles and early-life growth data into the Basic model yielded the lowest SD(res) of 0.19 SDscores, indicating the most precise predictions.
  • The developed models demonstrated superior accuracy in predicting GH responsiveness compared to maximal GH response during provocation tests.

Conclusions:

  • Validated models accurately predict individual growth responses to GH therapy in children with isolated GH deficiency or idiopathic short stature.
  • These predictive models offer enhanced accuracy over traditional methods, facilitating more informed clinical decisions.
  • The computer-based prediction tool can guide evidence-based selection of children for GH treatment, improving therapeutic outcomes.

Related Concept Videos