Assessing the validity of abbreviated literature searches in rapid reviews

ID: 

18113

Session: 

Long oral session 18: Efficiency in searching

Date: 

Friday 15 September 2017 - 11:00 to 12:30

Location: 

Meeting Room 1.43

All authors in correct order:

Nussbaumer-Streit B1, Klerings I1, Wagner G1, Heise TL2, Dobrescu AJ3, Armijo Olivo S4, Stratil J5, Teufer B1, Lhachimi SK2, van Noord M6, Mittermayer T7, Zeeb H2, Hemkens L8, Gartlehner G1
1 Cochrane Austria, Danube University Krems, Austria
2 Leibniz Institute for Prevention Research and Epidemiology - BIPS, Germany
3 Genetics Department, Victor Babes University of Medicine and Pharmacy Timisoara, Romania
4 Faculty of Rehabilitation Medicine, University of Alberta & Institute of Health Economics, Edmonton, Alberta, Canada
5 Institute for Medical Informatics, Biometry and Epidemiology, Ludwig-Maximilians-University, Munich, Germany
6 Medical Center Library & Archives, School of Medicine, Duke University, USA
7 Ludwig Boltzmann Institution for Health Technology Assessment, Austria
8 Basel Institute for Clinical Epidemiology and Biostatistics, Switzerland
Presenting author and contact person

Presenting author:

Barbara Nussbaumer-Streit

Contact person:

Barbara Nussbaumer-Streit
Abstract text
Background: Systematic reviews offer the most reliable and valid support for health-policy decision making, patient information and guideline development. However, they often do not meet the needs of those who have to make decisions quickly. Rapid reviews have therefore become a pragmatic alternative to systematic reviews. They are knowledge syntheses that abbreviate certain methodological aspects of systematic reviews to produce information more quickly. Methodological shortcuts often take place in literature identification. Potential disadvantages are less-reliable results. To date, the impact of abbreviated searches on estimates of treatment effects and subsequent conclusions has not been analysed systematically across multiple bodies of evidence.

Objectives: We aim to assess whether bodies of evidence that are based on abbreviated literature searches lead to different conclusions about benefits and harms of interventions compared with bodies of evidence that are based on comprehensive, systematic literature searches.

Methods: We used a non-inferiority design with the primary outcome: proportion of discordant conclusions based on different search approaches. We randomly chose 60 Cochrane reviews, and reproduced their MEDLINE, Embase and CENTRAL searches employing abbreviated search strategies. If abbreviated searches could not detect all studies included in the original review, we recalculated effect sizes, revised the original summary-of-findings table and asked review authors whether the missed evidence would change the conclusions of their report. We determined the proportion of discordant conclusions for each abbreviated search approach, and considered it as non-inferior if the lower limit of the 95% confidence interval of the proportion of discordant conclusions was below the non-inferiority margin, which was determined based on results of a survey for clinical and public-health scenarios.

Results: Will be available at the Summit.

Conclusions: This will be the first study to assess whether the reduced sensitivity of abbreviated searches has an impact on conclusions across multiple bodies of evidence, not only on effect estimates.