P05.25 LB
Purpose: Review the comparative effectiveness and efficiency of databases at retrieving articles include in a systematic review of yoga for musculoskeletal conditions.
Methods: Twenty electronic databases previously used in systematic reviews of complementary therapies and musculoskeletal conditions were searched using database-specific search strategies to identify full-text, peer reviewed, randomized controlled trials of yoga interventions for musculoskeletal conditions. Of the 774 articles returned by the 20 databases, 17 met criteria for inclusion in the systematic review. Each database was subsequently analyzed to determine how many of these 17 articles it retrieved (effectiveness), and how many of the total number of articles the database retrieved were included in the review (efficiency).
Results: Fourteen of the 20 databases retrieved at least one of the 17 articles included in the systematic review. Of these 14 databases, effectiveness ranged from 12% (PsycInfo: 2/17 articles) to 88% (PubMed: 15/17 articles); and efficiency ranged from 8% (Google Scholar: 12/143 articles) to 41% (Cochrane Central Register of Controlled Trials: 13/31 articles). There was differential retrieval of the 17 articles across databases, based on country of journal publication. Articles published in journals from India were retrieved from a maximum of five of the 20 databases (range 1–5); in contrast, articles published during a corresponding time period in Western journals were retrieved from 6–12 databases. Results also indicated some databases may take longer to index recently published articles, with only six of the 20 databases retrieving articles published in the three months prior to the review.
Conclusion: Variability in article retrieval suggests a range of databases should be used when designing systematic reviews of complementary therapies. While highly effective databases such as PubMed are currently regularly used, the addition of less efficient databases such as Google Scholar and Scopus is recommended to reduce retrieval bias based on time or place of publication.
Contact: Lesley Ward, lesley.ward@otago.ac.nz
