Managing overlap in overviews of reviews: a cross-sectional survey of the published literature between 2015 and 2017

Article type
Authors
Bajpai R1, Posadzki P1, Soljak M1, Car J1
1Lee Kong Chian School of Medicine
Abstract
Background: Overviews of reviews (or simply 'overviews') have become an increasingly established approach to synthesising research. Overviews are becoming prevalent because they have potential advantages over systematic reviews. However, they impose unique methodological challenges due to undeveloped synthesis methods. While significant empirical work has been undertaken to inform and improve overview methods, limited research and guidance is available for handling overlap in overviews.

Objectives: To examine the methodology used to handle overlap issues in Cochrane and non-Cochrane overviews.

Methods: We searched MEDLINE, Embase, and the Cochrane Database of Systematic Reviews (CDSR) via Ovid for overviews published between January 2015 and December 2017. We used the following search terms: systematic review of systematic reviews[title] OR umbrella review[title] OR overview of reviews[title] OR meta-epidemiologic study[title] OR overview of systematic reviews[tittle] OR overview of Cochrane reviews[tittle]. Studies were analysed descriptively. We aimed to elucidate how overlap issues are handled in Cochrane and non-Cochrane overviews.

Results: A total of 105 non-Cochrane and 21 Cochrane overviews met the eligibility criteria for this study. Among Cochrane overviews, six overviews (28.6%) explained overlap in their study, with deletion used as the preferred method to handle overlap. Among non-Cochrane overviews, 41 overviews (39.1%) did mention overlap issue in any form (with or without handling this issue). Out of these, 22 overviews (53.7%) described the extent of overlap; four overviews (9.8%) used Pieper’s approach to quantify the amount of overlap; three overviews reported no overlap (7.3%); and the remaining 12 overviews (29.3%) just mentioned overlap without using any method to handle it.

Conclusions: Overviews offer an exciting and challenging new way of synthesising evidence to inform practice and policy. However, there is no uniform methodology or consensus among overview authors about how to handle the issue of overlap in overviews. Therefore, to ensure the validity and utility of overviews, care must be taken in interpreting the results.