Statistical methods for meta-analysis: an overview

Article type
Authors
Beyene J
Abstract
Abstract: An understanding of the basic principles of the meta-analytic statistical methods commonly used in systematic reviews is critical for proper interpretation of synthesized research evidence. In this workshop, we will discuss various meta-analytical statistical methods including: the rationale for using meta-analysis to combine study results; the choice of measure of treatment effect (e.g., odds ratio, relative risk, risk difference, mean difference, standardized mean difference); modeling strategies and basic methods for combining measures of treatment effect from studies; testing and quantifying heterogeneity; and choice of fixed versus random effect models. The workshop will also discuss approaches for identifying sources of heterogeneity and how to deal with heterogeneity. Meta-regression will be briefly introduced and other methodological issues such as publication bias will be discussed.