Discussion section of Cochrane Reviews: is it supported by systematic reviews?

Article type
Authors
Glujovsky D1, Bardach A1, Comande D1, Ciapponi A1
1Institute for Clinical Effectiveness and Health Policy (IECS-CONICET)
Abstract
Background:
Any Cochrane Review is usually considered by readers to be an important piece of research. Although the Methods and Results sections are based on a systematic approach, the Discussion section may not be. Some readers (patients, other researchers and healthcare workers) retrieve information directly from the Discussion as if it were strong evidence and as if the authors have carried out a comprehensive search, simply because it comes from a Cochrane Review. However, it should be noted that if the references supporting this section are not of high quality, the information provided will not be either.

Objectives:
To analyze whether the 'Agreements and disagreements' section of the Discussion is based on systematic reviews and to evaluate whether there were other available systematic reviews that were not used and this section.

Methods:
We analyzed all the intervention Cochrane Reviews published in November/December 2017. We excluded updates. We retrieved the references to other studies or reviews from the 'Agreements and disagreements' sections. We ran the same search strategy in PubMed as the authors had used in the systematic review and added a filter for systematic reviews and another for burden of disease. We analyzed whether the retrieved studies were systematic reviews that could support the information given in the section and whether these references were different from those found by the review authors.

Results:
At the time of submission, we have analyzed 12 reviews. Full information will be provided at the Colloquium. We found that in four of the reviews (33%), no systematic reviews had been used to support the information used to describe the condition. When checking the studies retrieved by our search strategy, we found systematic reviews that could have been used in three of them or used more systematically in another six (75% were not comprehensive).

Conclusions:
A large proportion of references in the Discussion section were not systematic reviews. Authors should make more efforts to retrieve the best evidence for this section. Readers should critically appraise this information as sometimes it is not based on the highest quality evidence. This study aims to alert patients and consumers to this potential issue when they are reading a Cochrane Review.

Patient or healthcare consumer involvement:
Healthcare consumers participated as authors.