Core outcome sets and Cochrane Reviews – Cochrane reviewers can and should do more

Tags: Oral
Saldanha I1, Hughes K2, Kirkham J2, Li T3, Oliveira RDA4, Williamson P2
1Center for Evidence Synthesis in Health, Brown University School of Public Health, 2University of Liverpool, 3Johns Hopkins Bloomberg School of Public Health, 4Universidade Federal de Uberlândia

Background: outcomes are fundamental to Cochrane Reviews because they are yardsticks with which systematic review authors and decision makers determine intervention effectiveness and safety. Core outcome sets (COS) can help primary researchers as well as systematic review authors consistently measure outcomes that are most important and relevant to decision making. More than 300 published COS are available in an online, publicly searchable database maintained by the Core Outcome Measures for Effectiveness Trials (COMET) Initiative. The utilization and impact of COS on Cochrane Reviews has not been systematically examined.

Objectives: to examine the extent to which current Cochrane Review authors:

1) use COS in choosing outcomes for the review;

2) identify problems with outcome inconsistency in included studies; and

3) note the need for development of COS.

Methods: we examined all new or updated Cochrane Reviews of interventions addressing any topic and published by any Cochrane Review Group in 2019 (as of March 8). We extracted the relevant information noted above that may have been reported in any section of the completed version of the review (i.e. we did not examine protocols).

Results: we included all 100 eligible Cochrane Reviews that addressed a range of diseases, the most frequent of which were of the respiratory system (11/100 reviews), genitourinary system (11/100 reviews), and neoplasms (9/100 reviews). Forty-one of the 53 current Cochrane Review Groups (77%) were represented. Only nine of 100 reviews mentioned (or cited) a COS in relation to choosing outcomes for the review, but 40/100 reviews identified problems of inconsistency in outcomes in included studies. Among the 40 reviews that identified outcome inconsistency, only eight reviews (20%) noted the need for development of a COS.

Conclusions: this examination of recent Cochrane Reviews of interventions suggests that approximately 10% of reviews use COS to help inform their choice of outcomes, even though outcome inconsistency across included studies continues to be a widespread problem (40% of reviews). COS can help ensure that the most relevant outcomes are evaluated and that more included studies are incorporated into meta-analyses. We recommend that, when an appropriate COS exists, Cochrane Review authors, and especially Cochrane Review Groups, should use it to inform their process of choosing outcomes for their reviews. The COMET Database is online, free, searchable, and covers a range of health fields. In addition, authors of Cochrane Reviews, especially those that identify outcome inconsistency in included studies, should take the opportunity to push their respective fields forward by explicitly noting the need for COS, recommending COS development, and participating in the COS development process.

Patient or consumer involvement: we did not involve them in the conduct of this study, but many COS are developed or disseminated with active involvement of these (and other) stakeholders.