Critical Appraisal of a Systematic Review and Meta-Analysis

Performing Meta-Analysis and Addressing Heterogeneity

Abstract

A meta-analysis encompasses a set of statistical methods used to combine and synthesize a set of quantitative estimates or measures coming from multiple scientific primary studies. Within the context of
a systematic review, it is adequate to perform meta-analysis when we aim to synthesize quantitatively the primary studies included, as long as they are sufficiently homogeneous from both a methodological and a clinical point of view. A meta-analysis may be based on any quantitative measure summarizing the results of the primary studies included; therefore, one of the most important first topics to be appraised in a meta-analysis is the adequate choice of the effect measures to be analyzed. Next, a particular attention should be
placed on the choice and adequacy of meta-analytical model (fixed or random effects model) and the weighting methods used, taking into account the specific characteristics of the primary studies included.
Beyond the pooled meta-analytical measure, it is of paramount relevance in a meta-analysis to assess the existence and extent of heterogeneity (between studies variability beyond what would be expected). Generally, a meta-analytical pooled measure, as long as its evidence base is sufficiently comprehensive and unbiased, may very well be the best available answer to a given research question, since it encompasses and synthesizes the best available scientific evidence.
However, when substantial heterogeneity is found, pooled metaanalytical measures should be interpreted with great caution and the primary effort should be targeted to the exploration of possible causes and moderators of heterogeneity. In such situations, the main objective of the meta-analysis becomes the analysis and adequate assessment of heterogeneity, and not only the mere search for metaanalytical pooled measures that may not be per se the best available answer to the research question.