Combining studies 12

In the previous module we discussed choosing an effect measure for presenting the results of studies reporting dichotomous data within your review. If any of the outcomes you are planning to analyse are continuous, you should complete the additional module A1, either before or immediately after completing this module.

This module is about meta-analysis. We will look at the principles for the sensible combination of results from separate studies, and overview the methods that are most commonly used in Cochrane reviews.

Learning objectives

  • Understand what a weighted average is, and how it differs from a straightforward average
  • Understand the concept of standard error relating to individual trial results
  • Be aware that there are alternative methods for meta-analysis (Peto, Mantel-Haenszel) which differ in the way that study weights are calculated
  • Be aware that there are a group of methods known as random effects methods which take a different approach to fixed effect meta-analysis

Relevant sections of the Reviewers’ Handbook

  • Section 8

Where does this go in a Cochrane review?

The information in this module will be relevant to many parts of your review

  • In the data analysis part of the Methods section of a protocol or review, where you will describe what statistical techniques you are planning to use
  • When (or if) you actually perform meta-analyses using the analysis part of RevMan or other software
  • In the presentation of results in the Results section of the review
  • In the interpretation of results, in the Discussion of your review