The impact of blinding on trial results: a systematic review and meta-analysis

Article type
Authors
Zeraatkar D1, Pitre T1, Kirsh S1, Jassal T1, Anderson M1, Padoan A1, Xiang A1, Mah J2
1McMaster University
2Dalhousie University
Abstract
Background: Blinding, the concealment of the arm to which participants have been randomized, is an important consideration for assessing risk of bias of randomized trials. A growing body of evidence has, however, yielded inconsistent results on whether trials without blinding produce biased findings.

Objective: To conduct a systematic review and meta-analysis of the evidence addressing whether trials with and without blinding produce different results.

Methods: We searched MEDLINE, EMBASE, Cochrane Reviews, JBI EBP, and Web of Science, from inception to May 2022, for studies comparing the results of trials with and without blinding. Pairs of reviewers worked independently and in duplicate to screen search results for eligible studies, extract data, and assess the certainty (quality) of evidence using the GRADE approach. We pooled the results of studies comparing trials with and without the blinding of patients, healthcare providers/investigators, and outcome assessors/adjudicators using frequentist random-effects meta-analyses.

Results: We identified 47 studies. We found low certainty evidence that trials without the blinding of patients and healthcare providers, outcome assessors/adjudicators, and/or patients may slightly overestimate treatment effects.

Limitations: Nearly all studies were rated at high risk of bias, primarily because of potential for confounding. Our results are further limited by the potential inconsistency in the direction of the effect of blinding. For example, although we anticipate that trials without blinding overestimate the beneficial effects of treatments, it is possible that the direction of bias caused by lack of blinding may not be consistent. That is, in certain circumstances, open-label trials may overestimate effects and in other circumstances, they may underestimate effects.

Conclusions: Our systematic review and meta-analysis suggests that blinding may influence trial results in select situations—albeit the findings are of low certainty, and the magnitude of effect is modest. Our findings have implications for clinical trialists and for evidence users interpreting the results of unblinded trials. In the absence of high certainty evidence suggesting that trials with and without blinding produce similar results, investigators should be cautious about interpreting the results of trials without blinding.

Patient, public, and/or healthcare consumer involvement: We did not involve consumers in this investigation.