H B Dumbrigue1, J F Esquivel, J S Jones. 1. Department of Prosthodontics, P.O. Box 100435, University of Florida College of Dentistry, Gainesville, FL 32610, USA. hdumbrigue@dental.ufl.edu
Abstract
PURPOSE: The purpose of this study was to evaluate the sensitivity, specificity, and precision of MEDLINE search strategies for detecting randomized controlled trials (RCTs) in prosthodontics. MATERIALS AND METHODS: A list of RCTs was obtained by hand-searching the International Journal of Prosthodontics, Journal of Prosthetic Dentistry, and Journal of Prosthodontics published between 1988 and 1997. RCTs identified served as the "gold standard" with which different MEDLINE search strategies were compared. Subject headings, keywords, MEDLINE checktags, publication types, and combinations thereof were used, and their results were compared with the "gold standard". RESULTS: Sensitivity ranged from 9% to 67%, specificity was at least 96% for all searches, and precision ranged from 22% to 59%. The most productive searches were by publication type clinical trial and publication type randomized controlled trial. Combining search terms did not improve sensitivity for detecting RCTs published in prosthodontic journals (p > .10). CONCLUSION: These findings support the need for more detailed reporting of trial methodologies to facilitate accurate indexing of prosthodontic articles in MEDLINE.
PURPOSE: The purpose of this study was to evaluate the sensitivity, specificity, and precision of MEDLINE search strategies for detecting randomized controlled trials (RCTs) in prosthodontics. MATERIALS AND METHODS: A list of RCTs was obtained by hand-searching the International Journal of Prosthodontics, Journal of Prosthetic Dentistry, and Journal of Prosthodontics published between 1988 and 1997. RCTs identified served as the "gold standard" with which different MEDLINE search strategies were compared. Subject headings, keywords, MEDLINE checktags, publication types, and combinations thereof were used, and their results were compared with the "gold standard". RESULTS: Sensitivity ranged from 9% to 67%, specificity was at least 96% for all searches, and precision ranged from 22% to 59%. The most productive searches were by publication type clinical trial and publication type randomized controlled trial. Combining search terms did not improve sensitivity for detecting RCTs published in prosthodontic journals (p > .10). CONCLUSION: These findings support the need for more detailed reporting of trial methodologies to facilitate accurate indexing of prosthodontic articles in MEDLINE.
Authors: Ellen T Crumley; Natasha Wiebe; Kristie Cramer; Terry P Klassen; Lisa Hartling Journal: BMC Med Res Methodol Date: 2005-08-10 Impact factor: 4.615