Publication bias in comprehensive meta analysis

Publication bias was assessed on all homogeneous subsets (3.8% of all subsets of meta-analyses published in Psychological Bulletin) of primary studies included in meta-analyses, because publication bias methods do not have good statistical properties if the true effect size is heterogeneous. We examined evidence of publication bias in a large-scale data set of primary studies that were included in 83 meta-analyses published in Psychological Bulletin (representing meta-analyses from psychology) and 499 systematic reviews from the Cochrane Database of Systematic Reviews (CDSR representing meta-analyses from medicine).

Although there is consensus that publication bias exists, how strongly it affects different scientific literatures is currently less well-known. He is currently particularly interested in methods for visual communication of information and its application to synthesis methods.Publication bias is a substantial problem for the credibility of research in general and of meta-analyses in particular, as it yields overestimated effects and may suggest the existence of non-existing effects. Her research interests include network meta-analysis, individual participant data meta-analysis, synthesis of time-to-event and continuous outcomes and diagnostic test accuracy meta-analysis.ДЄlex Sutton has had a long-standing interest in methodology for evidence synthesis and has worked on developing methods for publication bias, diagnostic test accuracy meta-analysis and network meta-analysis. Suzanne Freeman is an NIHR Research Fellow and member of the NIHR Complex Reviews Support Unit.

Part 3: Simulation study, limitations and conclusions The webinar was delivered in July 2020В and below you will find the videos from the webinar, together with accompanying slides to download. Although there is some statistical content, concepts are explained visually where possible keeping much of the material accessible. These videos, originally part of the Cochrane Learning Live webinar series, areВ highly relevant to reviewers, editors and statisticians with interests in dealing with bias in meta-analysis. Through application to a motivating collection of meta-analyses of post-operative analgesics, and simulation studies, we will show that EggerвЂ™s test is potentially misleading for continuous outcomes and a test which regresses the residuals from a meta-regression model, including baseline risk as a study-level covariate, against inverse sample size has better statistical properties. When this is the case, funnel plots can appear highly asymmetric, even when publication bias is not present, since correlations between outcome and both effect size and its standard error exist. Comparative continuous outcomes are commonly measured on an absolute (mean) difference scale, and it is not uncommon for the magnitude of effect to be related to response in the control arm (i.e. The performance of EggerвЂ™s and related tests has been extensively studied for binary outcomes, but not for continuous ones. EggerвЂ™s test is commonly used to assess potential publication bias in a meta-analysis via funnel plot asymmetry (EggerвЂ™s test is a linear regression of the intervention effect estimates on their standard errors weighted by their inverse variance).

Publication bias in comprehensive meta analysis