Assessing risk prediction models using individual participant data from multiple studies

Lisa Pennells, Stephen Kaptoge, Ian R. White, Simon G. Thompson, Angela M. Wood, Robert W. Tipping, Aaron R Folsom, Christie M. Ballantyne, Josef Coresh, S. Goya Wannamethee, Richard W. Morris, Stefan Kiechl, Johann Willeit, Peter Willeit, Georg Schett, Shah Ebrahim, Debbie A. Lawlor, John W. Yarnell, John Gallacher, Mary CushmanBruce M. Psaty, Russ Tracy, Anne Tybjærg-Hansen, Ruth Frikke-Schmidt, Marianne Benn, Børge G. Nordestgaard, Jackie F. Price, Amanda J. Lee, Stela McLachlan, Kay Tee Khaw, Nicholas J. Wareham, Hermann Brenner, Ben Schöttker, Heiko Müller, Dietrich Rothenbacher, Jan Håkan Jansson, Patrik Wennberg, Veikko Salomaa, Kennet Harald, Pekka Jousilahti, Erkki Vartiainen, Mark Woodward, Ralph B. D'Agostino, Philip A. Wolf, Ramachandran S. Vasan, Emelia J. Benjamin, Else Marie Bladbjerg, Torben Jørgensen, Richard F. Gillum, the Emerging Risk Factors Collaboration

Research output: Contribution to journalArticlepeer-review

32 Scopus citations


Individual participant time-to-event data from multiple prospective epidemiologic studies enable detailed investigation into the predictive ability of risk models. Here we address the challenges in appropriately combining such information across studies. Methods are exemplified by analyses of log C-reactive protein and conventional risk factors for coronary heart disease in the Emerging Risk Factors Collaboration, a collation of individual data from multiple prospective studies with an average follow-up duration of 9.8 years (dates varied).We derive risk prediction models using Cox proportional hazards regression analysis stratified by study and obtain estimates of risk discrimination, Harrell’s concordance index, and Royston’s discrimination measure within each study; we then combine the estimates across studies using aweighted meta-analysis. Various weighting approaches are compared and lead us to recommend using the number of events in each study. We also discuss the calculation of measures of reclassification for multiple studies. We further show that comparison of differences in predictive ability across subgroups should be based only on within-study information and that combining measures of risk discrimination from casecontrol studies and prospective studies is problematic. The concordance index and discrimination measure gave qualitatively similar results throughout. While the concordance index was very heterogeneous between studies, principally because of differing age ranges, the increments in the concordance index from adding log C-reactive protein to conventional risk factors were more homogeneous.

Original languageEnglish (US)
Pages (from-to)621-632
Number of pages12
JournalAmerican journal of epidemiology
Issue number5
StatePublished - Mar 1 2014

Bibliographical note

Funding Information:
This work was supported the United Kingdom Medical Research Council (grant G0701619 and Unit Programme U105260558). The Emerging Risk Factors Collaboration Coordinating Centre was supported by the British Heart Foundation (grant RG/08/014), the Medical Research Council, the United Kingdom National Institute of Health Research Cambridge Biomedical Research Centre, a specific grant from the Bupa Foundation, and an unrestricted educational grant from GlaxoSmithKline. Various sources have supported recruitment, follow-up, and laboratory measurements in the cohorts contributing to the Emerging Risk Factors Collaboration. Investigators in several of these studies have contributed to a list of relevant funding sources (

Publisher Copyright:
© The Author 2013.


  • C index
  • Coronary heart disease
  • D measure
  • Individual participant data
  • Inverse variance
  • Meta-analysis
  • Risk prediction
  • Weighting


Dive into the research topics of 'Assessing risk prediction models using individual participant data from multiple studies'. Together they form a unique fingerprint.

Cite this