Combining results from several studies: systematic reviews and meta-analyses

Abstract This chapter explains systematic reviews, the PRISMA format, and meta-analysis. It discusses publication bias, outcome reporting bias, funnel plots, the issue of false positive results in small studies, along with search strategies, electronic databases, PubMed, and the Cochrane collaboration. It discusses the assessment of quality, risks of bias, limitations of meta-analysis, heterogeneity testing, effect modification, and meta-regression methods. In part two, it explains statistical methods for meta-analyses are presented, including the Mantel-Haenszel and Peto methods for individual patient data, the inverse variance weighted method using final results, and random effects methods. Forest plots and tests of heterogeneity are explained.