IPD Meta‐Analysis for Clinical Prediction Model Research

Richard D. Riley, Kym I.e. Snell, Laure Wynants, Valentijn M T de Jong, Karel G M Moons, Thomas P A Debray

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

This chapter describes the opportunities and challenges involved in prediction model research using individual participant data (IPD) meta-analysis. It begins by outlining the various types of prediction model research, and then describes the importance and conduct of IPD meta-analysis projects for each type. The chapter emphasizes the importance of evaluating prediction model performance in terms of calibration, discrimination and clinical utility, and the need to examine heterogeneity in performance across studies, settings and subgroups of interest. by meta-analysing standardised estimates of model performance, any remaining heterogeneity in performance only reflects the use of invalid model coefficients, thereby highlighting whether local updating of model coefficients is necessary for that target population. External validation of an existing prediction model may incorporate updating or tailoring of the prediction model equation, which is often needed to improve the performance in the setting or population at hand.
Original languageEnglish
Title of host publicationIndividual Participant Data Meta‐Analysis
Subtitle of host publicationA Handbook for Healthcare Research
EditorsRichard D. Riley, Jayne F. Tierney, Lesley A. Stewart
PublisherWiley
Chapter17
Pages447-498
Number of pages52
ISBN (Electronic)9781119333784, 9781119333753
ISBN (Print)9781119333722
DOIs
Publication statusPublished - 22 Apr 2021

Publication series

NameStatistics in Practice
PublisherWiley

Keywords

  • clinical prediction model research
  • individual participant data meta-analysis
  • local model updating
  • tailoring strategies

Cite this