Assessing bias due to confounding in a Cochrane Systematic Review that includes non-randomised studies of an intervention (NRSI)

Article type
Authors
Shea B1, Reeves B2, Savovic J3, Higgins J3, Sterne. J3, Wells G4
1OHRI and University of Ottawa
2Non-Randomized Studies Methods Group
3Bristol University, UK
4University of Ottawa, Canada
Abstract
Objective: To train authors to assess the risk of bias due to confounding using an extended risk of bias tool, 'A Cochrane Risk Of Bias Assessment Tool: for Non-Randomized Studies of Interventions' (ACROBAT-NRSI), when including NRSI in a systematic intervention review.
Description: Cochrane recommends that review authors consider and justify whether or not to include NRSI for all research questions about the effects of interventions. Decisions to include NRSI may arise when there are inadequate or no RCTs, but where the question addressed by the review is a considered a priority. Topics about possible harmful or long-term effects of interventions, or review questions about the effects of public health and non-pharmacological interventions, may have these characteristics. This workshop aims to give review authors and others intending to include NRSI in Cochrane Reviews experience in applying ACROBAT-NRSI. Participants will work mainly in small groups to apply the ‘signalling questions’ for the confounding domain of the tool, a domain that does not apply to RCTs. Signalling questions prompt the user to assess key aspects of studies and then to judge whether the finding from a study for a particular review outcome is at high or low risk of material bias in the domain. Responses to signalling questions and domain specific bias judgements are made on 4-point scales and include a ‘no information’ option; these features will be contrasted with the existing risk of bias tool.