Cochrane overviews of reviews: purpose and quality

Session: 

Oral session: Overviews of reviews and network meta-analysis (2)

Date: 

Monday 17 September 2018 - 14:30 to 14:40

Location: 

Fintry Auditorium

All authors in correct order:

Dalsbø TK1, Dahm KT1, Strømme H1
1 Norwegian Institute of Public Health, Norway
Presenting author and contact person

Presenting author:

Therese Dalsbø

Contact person:

Therese Dalsbø
Abstract text
Background: The Cochrane Handbook guides authors of Cochrane overviews of reviews. The Handbook gives advice about five relevant reasons for overviews and to distinguish between the quality of included reviews and the studies they include. The recommendations are: to use Grading of Recommendations, Assessment, Development and Evaluation (GRADE), conduct network meta-analysis if appropriate, and to interpret findings and present confidence in findings using a plain language format style. Our aim was to see if overviews published in the Cochrane Library were performed in accordance with key guidance from the Cochrane Handbook.

Methods: We searched for overviews of reviews in the Cochrane Database of Systematic Reviews (CDSR). We extracted data about their: main purpose, search date, number of included reviews, and if they planned to use a checklist for quality of included reviews such as AMSTAR (A MeaSurement Tool to Assess systematic Reviews). Then we assessed if they used GRADE, performed network meta-analysis (if appropriate), and gave a balanced summary in plain language about the benefit and harms and level of confidence in the results, and if there was patient or consumer involvement.

Main findings: There were 10,062 references in CDSR up to 12 January 2018 (issue1). We searched for overview* in title/abstract and found 99 references. We found and included 33 overviews of reviews. The main purpose of 20 overviews was to summarize evidence from more than one systematic review of different interventions for the same condition or problem. Also, five overviews had reasons other than those presented in the Handbook. The median search date was 2015, the oldest was from 2009 and the newest was 2017, one did not report a search date. One had no included reviews. Another had not included reviews, but used the studies from selected reviews. The average number of included reviews in the remaining overviews was 18.77. A total of 582 reviews, mostly Cochrane Reviews were included in the 31 overviews. Most, 93%, had used a checklist, mainly AMSTAR; 71% had used GRADE, and 9% had performed a network meta-analysis; 21% presented a balanced summary in plain language. Patient or consumer involvement was mentioned in two overviews, equivalent to 6%.

Conclusion: The quality of overviews could be improved by adhering to the Handbook expectations.

Relevance to patients and consumers: 

Patients or healthcare consumers does not appear to participate in many overviews. Especially overviews on effectiveness of multiple interventions for a condition or disease should be of high interest to the patients with the conditions in question and could be involved more often. A balanced and consistent plain language summary is helpful to both patients and healthcare professionals.