Introduction to meta-analysis 1: meta-analysis of binary and continuous outcomes

Session Type
Workshop - training
Statistical methods
Target audience
Review authors
Level of difficulty

Background: A core component of many systematic reviews is meta-analysis, which is a method for statistically combining results across studies. Meta-analysis results often underpin healthcare decision-making. Most commonly, meta-analysis of binary or continuous outcomes are undertaken. Many issues need to be considered when meta-analysing binary or continuous outcomes ranging from data extraction through to the selected meta-analysis method.
Objectives: To provide review authors with knowledge to undertake meta-analysis of binary and continuous outcomes. This workshop is part of a series of workshops delivered by the Cochrane Statistical Methods Group.
Description: We will use a combination of presentations and practicals. We will begin with a brief recap of meta-analysis models. We will then cover issues specific to meta-analysis of binary and continuous outcomes including: data extraction (e.g. for binary outcomes, extraction of event frequencies and/or effect estimates; and for continuous outcomes, extraction of standard deviations, or calculation from standard errors, confidence intervals, test statistics and P values); effect measures (e.g. risk ratio, odds ratio, standardised mean difference), with rationale for choosing between them; and different meta-analysis methods (e.g. inverse-variance, Mantel-Haenszel).

McKenzie J1
1Monash University, Australia