Article type
Year
Abstract
Background
Review databases are promising sources for rapid reviews of reviews to provide a breadth of studies with a relatively low yield of results. A novel option is Microsoft Academic Graph (MAG) to identify related publications and citation searching of relevant studies. A rapid map of systematic reviews of interventions and views on vaccine uptake provided an opportunity to analyse the utility of sources. Searches were undertaken in Epistemonikos, specialist registers, PubMed, CINAHL, Social Sciences Citation Index, Microsoft Academic Graph and other sources.
Objectives
To explore the utility of using Epistemonikos combined with Microsoft Academic Graph for identifying reviews.
Methods
Retrospective analysis of sources used in the map of reviews. All sources apart from MAG were searched using Boolean searching techniques and 1,438 citations were screened on title and abstract after duplicate removal. MAG was searched in a beta interface within EPPI-Reviewer, using 178 relevant reviews identified from Epistemonikos as seed studies to identify related publications and bi-directional citation links. The results from MAG were filtered by searching within the title and abstracts, followed by screening on title. All reviews included in the map, that were additional to those identified in Epistemonikos, were checked to determine if present in the MAG results. The analysis of reviews identified from the other sources were based on the initial searches used to populate the map.
Results
197 reviews were included in the map, of which 154 (78%) were identified in Epistemonikos, after a date threshold was applied. Of the remaining 43 reviews: 4 were uniquely identified by MAG through the related publication feature, of which one was also found from citation links. A further 18 identified from MAG were also identified in other resources, though 2 were missed from the initial filtering process. The remaining 21 reviews were only identified from the other sources, including 10 from the SYSVAC register and 7 each from other review registers and bibliographic databases (each category provided some unique citations).
Conclusions
While Epistemonikos provided the majority of reviews, there was benefit in searching other sources as time allowed. MAG offers a promising approach to identify new studies. However, it required a two-step approach as it uses relevant seed studies from screening the initial searches, plus further deduplication steps. Filtering approaches are needed to make the MAG results manageable to screen. Efficient ways to expand the breadth of relevant search results improves the research available to stakeholders including healthcare consumers.
Review databases are promising sources for rapid reviews of reviews to provide a breadth of studies with a relatively low yield of results. A novel option is Microsoft Academic Graph (MAG) to identify related publications and citation searching of relevant studies. A rapid map of systematic reviews of interventions and views on vaccine uptake provided an opportunity to analyse the utility of sources. Searches were undertaken in Epistemonikos, specialist registers, PubMed, CINAHL, Social Sciences Citation Index, Microsoft Academic Graph and other sources.
Objectives
To explore the utility of using Epistemonikos combined with Microsoft Academic Graph for identifying reviews.
Methods
Retrospective analysis of sources used in the map of reviews. All sources apart from MAG were searched using Boolean searching techniques and 1,438 citations were screened on title and abstract after duplicate removal. MAG was searched in a beta interface within EPPI-Reviewer, using 178 relevant reviews identified from Epistemonikos as seed studies to identify related publications and bi-directional citation links. The results from MAG were filtered by searching within the title and abstracts, followed by screening on title. All reviews included in the map, that were additional to those identified in Epistemonikos, were checked to determine if present in the MAG results. The analysis of reviews identified from the other sources were based on the initial searches used to populate the map.
Results
197 reviews were included in the map, of which 154 (78%) were identified in Epistemonikos, after a date threshold was applied. Of the remaining 43 reviews: 4 were uniquely identified by MAG through the related publication feature, of which one was also found from citation links. A further 18 identified from MAG were also identified in other resources, though 2 were missed from the initial filtering process. The remaining 21 reviews were only identified from the other sources, including 10 from the SYSVAC register and 7 each from other review registers and bibliographic databases (each category provided some unique citations).
Conclusions
While Epistemonikos provided the majority of reviews, there was benefit in searching other sources as time allowed. MAG offers a promising approach to identify new studies. However, it required a two-step approach as it uses relevant seed studies from screening the initial searches, plus further deduplication steps. Filtering approaches are needed to make the MAG results manageable to screen. Efficient ways to expand the breadth of relevant search results improves the research available to stakeholders including healthcare consumers.