Error identification in search strategies of new Cochrane Reviews published in 2015

Article type
Year
Authors
Franco JVA1, Garrote V1, Escobar C1, Vázquez L1, Vietto V1
1Instituto Universitario Hospital Italiano, Cochrane Center, Argentina
Abstract
Background: Search strategy is a key component for the identification of clinical study reports. Previous research estimated error rate in search strategies of Cochrane Systematic Reviews (CSRs) to be as high as 90% (Sampson 2006). Efforts have been made since then in order to reduce errors in the design and report of search strategies, such as modifications in the Cochrane Handbook for Systematic Reviews of Interventions (Cochrane Handbook) and the development of expert consensus documents for peer-reviewing like the PRESS (Peer Review Electronic Search Strategies) guideline.

Objectives: To describe errors identified in the design and report of search strategies of new CSRs published during 2015.

Methods: In this cross-sectional study, we selected a random sample of 89 CSRs from the Cochrane Database of Systematic Reviews from the 12 issues published in 2015. Updates, withdrawals, protocols, empty reviews, dentistry, prognostic associations, safety of interventions and diagnostic test accuracy reviews were excluded. We formed a peer-reviewing team composed of a trained Cochrane Information Specialist, two medical librarians and a medical researcher. We used the Cochrane Handbook recommendations to assess reporting of search strategies and 12 items assessed by Sampson 2006 together with the six elements of the PRESS 2015 guideline to identify errors in MEDLINE search strategies.

Results: After excluding 19 CSRs, 70 reviews were eligible for inclusion. Preliminary results based on the assessment of 20 included CSRs show that most reported search strategies (19/20) lack one or more of the recommended elements from the Cochrane Handbook; 4/20 lacked a full report of the detailed search strategy; and 9/20 had at least one identifiable error in the design.

Conclusions: we will complete the assessment of the remaining 50 reviews and present the final results and conclusions at the Colloquium.