Generating empirical evidence to support methods for overviews of reviews

Article type
Authors
Hartling L1, Dryden D1, Vandermeer B1, Fernandes R2
1University of Alberta, Canada
2Hospital de Santa Maria, Portugal
Abstract
Background: Overviews of reviews (overviews) compile data from multiple systematic reviews (SRs) and provide a single synthesis of relevant evidence for decision-making. Current methodological guidance for overviews is driven by personal experience and ‘good practice’.

Objectives: To examine methodological considerations when conducting overviews focusing on inclusion criteria, statistical synthesis, and grading evidence.

Methods: Weselected four overviews published in Evidence-based Child Health: A Cochrane Review Journal (acute otitis media, croup, bronchiolitis, gastroenteritis). We examined issues related to including SRs published outside of the Cochrane Database of Systematic Reviews (CDSR), including how to deal with multiple SRs of the same intervention. We explored the feasibility of conducting network meta-analyses and issues related to grading the evidence based on data reported in SRs.

Results: We found a number of SRs published outside of the CDSR. For example, 6 Cochrane and 8 non-Cochrane SRs examined acute otitis media; and, 3 Cochrane and 15 non-Cochrane Reviews examined gastroenteritis. Some SRs overlapped in content while others examined different interventions or populations. Methodological questions that arose in selecting SRs for an overview were whether to include SRs on the same topic and the basis for selecting SRs (i.e., methodological quality, search dates, organization producing SR). We also identified issues about how to assess methodological quality of SRs and limitations of existing tools. Network meta-analyses were possible where there was clinical homogeneity; however, analyses were reliant on data presented in the SR and decisions made by SR authors. A challenge for grading the evidence based on SRs was inconsistency in methods used to assess risk of bias of primary studies.

Conclusions: We identified a number of methodological issues when conducting overviews. Many issues stem from reliance on methods and decisions made at the SR level. These findings contribute to an evidence base to guide overview methods.