Evaluation of a novel citation-based search method vs. Ottawa/RAND Search Method for Prioritising systematic reviews for update




Poster session 2 Thursday: Evidence synthesis - methods / improving conduct and reporting


Thursday 14 September 2017 - 12:30 to 14:00


All authors in correct order:

Veazie S1, Winchell K1, Relevo R1, Helfand M1
1 Scientific Resource Center, AHRQ Effective Health Care Program, USA
Presenting author and contact person

Presenting author:

Stephanie Veazie

Contact person:

Abstract text
Background: Although systematic reviews (SRs) should be updated as new research is published, most organisations cannot update every SR. The Agency for Healthcare Research and Quality (AHRQ) Scientific Resource Center (SRC) assesses the currency of AHRQ SRs to determine priority for update. The SRC uses a gap literature search method based on the work of University of Ottawa and the RAND Corporation (Ottawa/RAND method). Citation-based search methods have been proposed as an alternative way to efficiently identify articles that meet original SR criteria. The SRC developed a novel citation-based search method (SRC method) that searches for articles that either cite or use the same subject headings as the original SR’s included articles.

Objective: To evaluate if the SRC method is more efficient than the Ottawa/RAND method at identifying articles that meet original SR criteria.

Methods: We conducted gap literature searches using the SRC method and Ottawa/RAND method for two SRs. We determined the total number of reviewed articles and applied the original SRs inclusion criteria to determine the inclusion percentage (included articles/articles reviewed), randomised-controlled trial (RCT) inclusion percentage (RCTs/ included articles), and number of 'unique' included articles identified through one method but not the other.

Results: Fewer total articles were identified using the SRC method than the Ottawa/RAND method for one review, but more were identified for the other review (SR#1: 96 vs. 111 articles; SR#2: 335 vs. 253 articles). The SRC method resulted in a greater inclusion percentage for both reviews (SR#1: 24% vs. 14%; SR#2: 7% vs. 3%). The SRC method resulted in a greater RCT inclusion percentage for one review (SR#1: 87% vs. 81%), and an equal RCT inclusion percentage for the other review (SR#2:100%) since its inclusion criteria stipulated that only RCTs be included. For both reviews, the SRC method identified more 'unique' included articles (SR#1: 14 vs. 7 articles; SR#2: 20 vs. 2 articles).

Conclusion: The SRC method is more efficient than the Ottawa/RAND method at identifying articles meeting original SR inclusion criteria.