Randomized Trial Comparing Masked/Unmasked Meta-Analyses. Abstract, ExecutiveSummary and Final Report

摘要

Masking (blinding) of readers to identifying information about the originalpapers in meta-analyses have been recommended, but the effects of blinding on the results of actual meta-analyses have not been assessed. We conducted a randomized trial in which five, previously published meta-analyses were performed under blinded and unblinded conditions. The summary odds ratio was the main outcome measure calculated for each meta-analysis. The summary odds ratios were nearly identifical for the blinded and unblinded meta-analyses. Blinding was extremely time-consuming. Therefore, because blinding of readers to identifying information has little to no effect on the results and is time-consuming, we recommended that blinding not be used when met-analyses are performed.