Article type
Year
Abstract
Background: It is not untypical for (co-)authors of Cochrane Overviews to have also (co-)authored one or several of the reviews included in the overview. This type of dual (co-)authorship raises questions regarding independence of judgments and intellectual conflicts of interest (CoI). Potential for bias may arise from quality assessment, interpretation of data, drawing conclusions, and (non-)inclusion of or dealing with competing reviews.
Objectives: To examine how Cochrane Overviews deal with potential CoI regarding dual (co-)authorship and whether the issue is raised in protocols.
Methods: The CDSR was searched for overviews and protocols for overviews in January 2015. Data were extracted on: total number of included reviews; reviews with dual authorship; independence of quality assessment; mention of the issue under limitations or declaration of interest (DoI); and description of safeguards. For protocols, data were extracted on whether the issue was brought up and how it would be dealt with.
Results: Eighteen overviews and 24 protocols were identified. Overviews included a median of nine reviews (interquartile range (IQR): 6 to 18). In 16/18 overviews (89%) at least one of the included reviews was affected by dual (co-)authorship. A median of five (IQR: 2 to 7) reviews per overview were affected by dual (co-)authorship. In 6/16 (38%) overviews with dual (co-)authorship, quality assessment was conducted independently, in 9/16 (56%) overviews it was not conducted independently and in 1/16 overviews this was unclear. Potential biases arising from dual (co-)authorship at other stages of the overview, such as the selection process or interpretation of data were not addressed. In 7/24 (29%) protocols dual (co-)authorship was mentioned, typically under DoI. Only one protocol described a safeguard (independent quality assessment of reviews with dual (co-authorship).
Conclusions: Potential biases arising from dual (co-)authorship are often neglected in Cochrane Overviews. We argue that their authors and Review Groups should pay more attention to the issue, in order to avoid bias and preserve the good reputation that Cochrane Overviews will typically deserve.
Objectives: To examine how Cochrane Overviews deal with potential CoI regarding dual (co-)authorship and whether the issue is raised in protocols.
Methods: The CDSR was searched for overviews and protocols for overviews in January 2015. Data were extracted on: total number of included reviews; reviews with dual authorship; independence of quality assessment; mention of the issue under limitations or declaration of interest (DoI); and description of safeguards. For protocols, data were extracted on whether the issue was brought up and how it would be dealt with.
Results: Eighteen overviews and 24 protocols were identified. Overviews included a median of nine reviews (interquartile range (IQR): 6 to 18). In 16/18 overviews (89%) at least one of the included reviews was affected by dual (co-)authorship. A median of five (IQR: 2 to 7) reviews per overview were affected by dual (co-)authorship. In 6/16 (38%) overviews with dual (co-)authorship, quality assessment was conducted independently, in 9/16 (56%) overviews it was not conducted independently and in 1/16 overviews this was unclear. Potential biases arising from dual (co-)authorship at other stages of the overview, such as the selection process or interpretation of data were not addressed. In 7/24 (29%) protocols dual (co-)authorship was mentioned, typically under DoI. Only one protocol described a safeguard (independent quality assessment of reviews with dual (co-authorship).
Conclusions: Potential biases arising from dual (co-)authorship are often neglected in Cochrane Overviews. We argue that their authors and Review Groups should pay more attention to the issue, in order to avoid bias and preserve the good reputation that Cochrane Overviews will typically deserve.