Background: Bibliographic databases such as Medline and PsycINFO are major sources of initial information for systematic reviews. This study investigates the adequacy of Medline and PsycINFO in retrieving articles describing outcome studies of group-based psychological treatments.
Method: Ten journals publishing a high number of group psychotherapy articles were hand searched for the publication years 1993 and 1994. They were then searched using narrow and broad strategies developed for each database.
Results: The sensitivities (number of relevant articles identified by each search divided by the total number of known relevant articles identified by all searches) of narrow and broad Medline searches were 28% and 67% respectively. Precision (the number of relevant articles identified by a search divided by the number of both relevant and irrelevant articles identified by the search) was 91% and 35% for narrow and broad searches respectively. The sensitivities of narrow and broad PsycINFO searches were 8% and 47% respectively. Precision was 30% for the narrow search and 36% for the broad search. The hand search identified 94% of all known articles.
Conclusions: PsycINFO performed more poorly than Medline. Although it was possible to identify 67% of outcome studies using Medline, the low precision of the necessary search strategies would make it uneconomical in practice. Abstracts of papers not identified by either search strategy were examined for possible reasons for this failure. For Medline this was generally due to the terms used to identify group psychotherapy. For PsycINFO it appears that the indexing system does not allow the easy identification of outcome studies.