Background: over the past few years, evidence synthesis has become essential to investigate and improve the generalizability of medical research findings. This strategy often involves a meta-analysis to formally summarize quantities of interest, such as relative treatment effect estimates. The use of meta-analysis methods is, however, less straightforward in prognosis research because substantial variation exists in research objectives, analysis methods and in the level of reported evidence.

Objectives: we present a gentle overview of statistical methods that can be used to summarize data of prognostic factor and prognostic model studies. We discuss how aggregate data, individual participant data, or a combination thereof can be combined through meta-analysis methods.

Methods: we present statistical models for one-stage and two-stage meta-analysis, multivariate meta-analysis and network meta-analysis in prognosis research. We also discuss the adaptation method, hierarchical-related regression, model averaging, stacked regressions and internal-external cross-validation.

Results: we illustrate various methods using recent examples, including a meta-analysis with individual participant data from 14 specialized centres for patients with amyotrophic lateral sclerosis.

Conclusions: advanced meta-analysis methods are often needed to provide (meaningful) summary estimates of primary prognosis studies, and to understand sources of between-study heterogeneity. Regardless, researchers should not be daunted by their complexity, as many of these methods have been implemented in traditional software packages and lead to an improved understanding of prognosis-related research questions.

Patient or healthcare consumer involvement: not applicable at this stage