The ORBIT study: outcome reporting bias in trials – inter­views with trialists who do and do not report all outcomes AU:Rebecca Smyth

Article type
Authors
Jacoby A, Gamble C, Altman D, Williamson P
Abstract
Background: Outcome reporting bias refers to the process of selective reporting based on results of some outcomes but not others in trial publications. Such bias has been recognised as a potential threat to the validity of meta-analysis. Without complete reporting of research findings any attempt to recommend a treatment is likely to be based on biased information. Little research, however, has been undertaken to understand the process of reporting research findings from the viewpoint or situation of the researcher. To our knowledge, no qualitative study has assessed researchers’ understanding of the potential sources of bias in the analysis and reporting of research findings. Interviewing trialists as part of the ORBIT project (UK MRC G0500952) provided a unique opportunity to gain insight into researchers’ understanding of and their contributions to bias in the research process. Objectives: To further our understanding of the processes resulting in outcome reporting bias. Methods: Semi-structured interviews with trialists identified from two separate sources (Cochrane reviews and PubMed) and three groups (trials with and without suspected outcome reporting bias, and cohort of recently published trials in PubMed). The interview schedule dealt with four topic areas: specific questions relating to outcome reporting bias; general questions relating to writing and getting the paper published; general questions relating to publication bias more generally; and questions regarding understanding of the research process. Results: Fifty-nine interviews with trialists were performed. The majority (76%) did not report outcomes as they intended, although for some (58%) their reasons were legitimate. The findings from the interviews give an explanation and understanding of why outcome reporting bias happens. Conclusions: Interviews with trialists are feasible. This study provides an understanding of the processes resulting in outcome reporting bias. Interviews suggest that few trialists are aware of this potential source of bias and how to prevent it occurring.