ABSTRACT
Most systematic reviews and meta-analyses are based on data extracted from publications. Individual participant data (IPD) meta-analyses, by contrast, collect, validate, and re-analyze individual-level data recorded for each participant recruited in each included study. IPD meta-analyses are more time-consuming and costly to perform than standard meta-analyses but offer considerable advantages. These include: ensuring data and analyses are consistent across studies; inclusion of unpublished studies or studies not reporting the desired results; and greater range and flexibility in the meta-analyses that may be performed. This chapter discusses how to conduct an IPD systematic review and meta-analysis, and how it can differ from a conventional review. The chapter describes the statistical methods used for IPD meta-analysis, with particular focus on regression models and how these models are used to investigate whether patient characteristics modify the effect of interventions. While the chapter primarily considers the meta-analysis of randomized trials, the application of IPD methods to other fields is discussed.
