Understanding meta‐analysis

The statistical aspects of a systematic review of test accuracy are more challenging than those for reviews of interventions. It is therefore recommended that review teams include an individual with the statistical expertise needed to understand and implement the hierarchical models required for meta-analysis. This chapter provides guidance to the key researchers in the review team on the purpose, possibilities and interpretation of methods of meta-analysis and also provides the technical detail to assist a statistical expert in applying the methods recommended for Cochrane Reviews. It outlines the conceptual approach to meta-analysis, graphical presentations and the key meta-analysis methods that are recommended. Meta-analysis of test accuracy studies provides summaries of the results of relevant included studies, providing estimates of the average diagnostic accuracy of a test or tests, the uncertainty of these estimates and the variability between studies around the estimates.