Extending the assessment of outcome reporting bias to harms

Tags: Workshop
Kirkham JJ1, Saini P1, Dwan K2, Williamson P3, Loke Y4, Altman DG5, Gamble C6
1University of Liverpool, United Kingdom, 2The University of Liverpool, United Kingdom, 3University of Liverpool, Afghanistan, 4University of East Anglia, UK, 5University of Oxford, UK, 6University of Liverpool, UK


a) To discuss how reviewers decide what harm outcomes to include in their reviews. b) To provide the reviewer with a background to the problem and mechanisms that may lead to incomplete reporting of harms data. c) To discuss how a reviewer might identify outcome reporting bias (ORB) for harms in their review. d) To present techniques for assessing the robustness of the meta-analysis to such bias.


Empirical research has demonstrated that harms are poorly reported in clinical trials. Important harm outcomes maybe subject to outcome reporting bias (ORB) where trialists prefer to focus on the positive benefits of an experimental intervention. Methods for the identification of ORB in harm outcomes for an individual study will be described and illustrated using examples. We will also remind reviewers about assessing ORB for benefit outcomes. Participants will be encouraged to undertake such assessments for examples provided and to discuss issues for their reviews. Sensitivity approaches for adjusting for this form of bias will be described and methods for implementation will be provided. Considerations for the impact ORB can have on the benefit-harm ratio with and without adjustment will also be presented.