Skip to main content

Languages

Assessment of risk of bias due to blinding for objective and subjective outcomes: An exploratory study of Cochrane Reviews

Date and Location

Session: 

P2.019

Date

Saturday 21 September 2013 - 10:30 - 12:00

Location

Presenting author and contact person

Presenting author

Charles Okwundu

Contact person

Charles Okwundu
Abstract text
Background: In Cochrane reviews, the risk of bias (RoB) in included studies is assessed to determine the likelihood that the intervention effects have been affected by bias. The Collaboration’s tool for assessing RoB has different domains related to particular types of bias - sequence generation, allocation concealment, blinding, selective outcome reporting and incomplete outcome data. For blinding, RoB assessment should be done separately per outcome reported, especially for studies with a mix of subjective and objective outcomes. Objectives: To evaluate whether Cochrane review authors assess risk bias due to blinding separately for subjective and objective outcomes. Methods: We included all new or updated reviews published in The Cochrane library in January and February 2013. Two assessors independently classified the reported outcomes in each review as objective or subjective, and whether authors of the reviews assessed the risk of performance and detection bias separately per outcome. Discrepancies were resolved by discussion and with input from a third assessor Results: We identified a total of 158 new and updated reviews (66 new reviews and 92 updated reviews). Twenty of the reviews (12.7%) did not have included studies, seven (4.4%) included non-randomised controlled trials (RCTs), while 131 (82.9%) included RCTs. Of these, 82 (63%) reported a mix of subjective and objective outcomes. The risk of performance and detection bias was assessed per outcome in seven (5.3%) and 14 (10.7%) of these reviews respectively. Conclusions: Few Cochrane authors separate risk of performance or detection bias for objective and subjective outcomes. Objective outcomes are less likely to be influenced by lack of blinding, compared to subjective outcomes. Authors should thus be more careful in evaluating the risk of bias from lack of blinding per outcome, especially in reviews reporting a mix of objective and subjective outcomes.