The projects in this proposal are part of a series of investigations seeking ways to develop search strategies that will improve retrieval of clinically relevant and scientifically sound study reports from large bibliographic databases including MEDLINE, Embase, PsycLIT and/or CINAHL. The purpose of the search strategies is 1) to assist health care providers to do their own searches; 2) to help reviewers of published evidence concerning health care interventions to retrieve all relevant citations; 3) to provide resources for librarians to help health care providers to construct their own searches; and 4) to provide input to the database producers on their indexing processes. The proposed projects involve perfecting ways to harness these electronic databases for clinical purposes so the effect on clinical practice can be eventually tested. In our previous research using MEDLINE, we documented the retrieval performance of over 60,000 """"""""methodologic hedges"""""""", that is, search strategies designed to detect applied clinical research studies that meet basic scientific standards for study of the etiology, prognosis, diagnosis, prevention and treatment of health care problems. To extend this research the questions to be addressed in this investigation are: 1. What are the operating characteristics, in 1999 files of MEDLINE, Embase, PsycLIT, and CINAHL, of methodologic terms that define studies of most importance, in terms of both content and quality, to clinical questions about the cause, course, diagnosis, prevention or therapy, or qualitative aspects of human disorders?; 2. Can empirical search strategies (""""""""hedges"""""""") be created that combine terms to optimize the yield from the above noted electronic databases and topics? 3. How do the databases compare in terms of their retrieval performance?; 4. How well do the methodologic search strategies derived in our previous research perform in more recent MEDLINE files when using the same journal, subset, i.e., how well do the search strategies derived in the 1986 and 1991 MEDLINE databases perform when using citations published in 1999?; 5. How do the search strategies developed in this project compare with those published by other groups? To evaluate the search strategies, index terms and textwords related to research design features will be treated as diagnostic tests and a manual review of the literature will be treated as the gold standard . Articles will be defined as having relevant content and meeting basic methodologic criteria for clinical practice by the manual review. The sensitivity, specificity, accuracy and precision of the search strategies will be determined in each of the 4 electronic databases.
Showing the most recent 10 out of 11 publications
