Does it matter which version of Medline you search?

Article type
Authors
Irvin E
Abstract
Background: When conducting a systematic review, reviewers typically search a variety of medical and non-medical electronic databases. Regardless of which others they choose, it is safe to assume that most will search Medline. However, depending on academic or clinical affiliation, people have access to different platforms for accessing MEDLINE. Some may use a vendor-based version such as OVID, while others may access the free version PubMed, made available through the US National Library of Medicine.

Objective: To describe the difference between OVID and PubMed in several systematic reviews conducted in the Back, Musculoskeletal, and Pain, Palliative Care and Supportive Care (PaPaS) Cochrane Review Groups.

Methods: This study is a secondary analysis of our previous work on search strategies.[1,2] Ten recent reviews and updates with comprehensive search strategies from the Back, the Musculoskeletal, and the Pain, Palliative Care and Supportive Care (PaPaS) Cochrane review groups were selected. The trials included in these reviews were searched in Medline: both Pubmed and OVID, (i) to determine if there was a difference in the yield of the databases, and if there was, (ii) to identify which database yielded the highest percentage of included trials.

Results: Having conducted the review we found that MEDLINE (both OVID and PubMed) contained 90% of the included studies; 6% of the studies searched were not located in OVID or PubMed, because they were not published in journals routinely indexed by either of these two electronic platforms. The reviews were classified into three topic areas: drugs, physical modalities and psychosocial interventions. In the five physical intervention reviews, of the 76 studies in these reviews PubMed found 43 (57%), while OVID found 37 (49%). The results for drug and psychosocial interventions will be presented at the Colloquium. Conclusions: When searching for trials in some physical interventions PubMed retrieves more citations than OVID. This is probably due in part to the fact that PubMed indexes citations from journals that precede the date those same journals were selected for OVID indexing. It is important that reviewers be aware of this fact and pay attention to what version of MEDLINE they are searching when conducting a systematic review. However, a more detailed study would be necessary to conclude that reviewers abandon OVID for complex searches.

Acknowledgements: Doreen Day, Andrea Furlan and Claire Bombardier co-authors of the previous study which spawned the idea for this abstract.

References: 1. Day D, Furlan A, Irvin E, Bombardier C. Simple searching how many lines are really needed? Institute for Work & Health Working Paper # 259, 2004. 2. Irvin E. Cochrane Colloquium: Barcelona; October 26th - 31st, 2003. Oral presentation: Simple searching - How many lines are really needed?