Reporting of Risk of Bias 2 (RoB2) in Cochrane protocols and reviews

Article type
Authors
Dwan K1, Moore TH2, Minozzi S3, Flemyng E4
1Centre for Reviews and Dissemination, University of York
2The National Institute for Health and Care Research Applied Research Collaboration West (NIHR ARC West) at University Hospitals Bristol and Weston NHS Foundation Trust, UK
3Department of Epidemiology, Lazio Regional Health Service, Rome
4Cochrane Central Executive Team, London
Abstract
Background: Risk of Bias 2 (RoB 2) was introduced in 2019 to improve the usability and address some limitations of the original version. It is used to assess bias in randomised controlled trials (RCTs) included in systematic reviews. Minozzi et al (2022) looked at any systematic review (n=208 reviews 5.3% Cochrane) published Jan 2019 to Jan 2020 and found that many applied the RoB 2 tool incorrectly. For example, for reviews with more than one outcome, only 28.2 % applied the RoB 2 tool to specific outcomes with 71.8% erroneously applying the tool to the whole study. Guidance for authors to use the tool appropriately was developed by Cochrane during a pilot phase of implementation (Flemyng 2022). Has this guidance helped Cochrane authors to use RoB2 appropriately?

Objectives: To determine whether Cochrane protocols and reviews used RoB2 appropriately and followed the available Cochrane guidance.

Methods: We searched for Cochrane protocols and reviews of interventions, published between January 2019 and December 2022 which specified use of RoB2. Each review and protocol will be assessed against the Cochrane guidance for appropriate use of RoB 2.

Results: Ninety eight Cochrane reviews, updates and 236 protocols for reviews had cited use of RoB 2, which represents 5.5% of reviews and 26% of protocols. We will summarise how many of the protocols and reviews follow both the conduct and reporting aspects of the guidance. This will include: Stating the effect of interest, listing the outcomes assessed, specifying time points and means of measurement of outcomes, stating if variants of the tool will be used for cross-over and cluster RCTs, check if the RoB 2 tool has been inappropriately adapted, state how the data collected for RoB 2 will be stored and presented.

Conclusions: We will state if the Cochrane guidance has helped authors to apply and report the use of the RoB 2 tool in the assessment of bias in RCTs appropriately. The findings will be used to inform future implementation of new methods in Cochrane with the aim of reducing misapplication in the early stages of implementation.

Patient, public and/or healthcare consumer involvement: None.