Objectives: To evaluate the use of two Canadian provincial databases by a systematic review of published studies that used them as a primary data source to answer epidemiologic and health services research questions.
Study design and setting: PubMed, EMBASE, BIOSIS, and CINAHL (keywords: "Manitoba" 1970-2004 and "Saskatchewan" 1969-2004) and the web sites of the provincial data custodians were searched to address our objective. Broad screening of citations and data abstraction were performed using a predefined collection form. Information on study characteristics, therapeutic areas studied, databases used, authors' affiliation, and issues related to data validity was recorded.
Results: Three thousand nine hundred and forty-nine citations were screened, 610 studies retrieved, and 325 included. In Saskatchewan, the principal research type was assessment of exposures and health outcomes (48.2%) with 50.4% using a cohort or case-control design, whereas, in Manitoba, it was health services utilization (47.8%) and 86.6% were descriptive. Local investigators performed 83.3% of the Manitoba studies, compared with 35.5% of the Saskatchewan studies. Only 6.2% of the studies assessed the validity and reliability of the database for research purposes and few incorporated relevant information about the validity of their diagnostic data.
Conclusion: Important differences exist in the administration and use of these databases. Similar systematic evidence synthesis should be conducted on other databases.