Quality of clinical heterogeneity investigations in systematic reviews: an assessment of Cochrane and non-Cochrane reviews

Tags: Poster
Gagnier J1, Morgenstern H1, Moher D1
1University of Michigan, USA

Background: Systematic reviews frequently have heterogeneity between the included studies. In particular, clinical characteristics (e.g., patient and treatment characteristic) frequently differ between the included trials. Recently a set of consensus-based recommendations for investigating clinical heterogeneity were developed.

Objective: To apply the recommendations for investigating clinical heterogeneity in systematic reviews to a sample of systematic reviews to highlight the methods being used for such investigations and to make recommendations on how such investigations can be improved.

Methods: Wecollected the 100 most recent Cochrane reviews and 100 non-Cochrane reviews published within 24 months prior to February 2012. We then assessed each review on each of the recommendations for investigating clinical heterogeneity. Each review was assessed as having performed each method or not (y, n) or if insufficient information (dk) was reported to determine fulfillment. We then calculated mean fulfillment of each item and across all items. We compared the methods of investigation in Cochrane verses non-Cochrane reviews. We also applied regression analyses to determine the role of predictors (e.g., whether a statistician was involved, funding source, journal impact factor) on investigations of clinical heterogeneity.

Results: We are currently conducting this research and will present the results at the upcoming meeting.

Conclusions: We expect that these findings will reinforce the need for systematic reviewers to conduct valid and reliable investigations of clinical heterogeneity.