Efficient strategies to find diagnostic test accuracy studies in kidney journals. (August 2015)

• Record Type: Journal Article
• Authors: Rogerson, Thomas E; Ladhani, Maleeka; Mitchell, Ruth; Craig, Jonathan C; Webster, Angela C
• Is Part Of: Nephrology. Volume 20:Number 8(2015)
• Page Start: 513
• Page End: 518
• Publication Date: 2015-08
• Subjects: diagnosis -- information storage and retrieval -- MEDLINE -- sensitivity and specificity -- literature searching; Nephrology -- Periodicals; Kidneys -- Diseases -- Periodicals; Nephrologists -- Periodicals; 616.61; 616.61
• DOI: 10.1111/nep.12445
• Languages: English
• ISSNs: 1320-5358

Abstract

Abstract: Aim: Nephrologists looking for quick answers to diagnostic clinical questions in MEDLINE can use a range of published search strategies or Clinical Query limits to improve the precision of their searches. We aimed to evaluate existing search strategies for finding diagnostic test accuracy studies in nephrology journals. Methods: We assessed the accuracy of 14 search strategies for retrieving diagnostic test accuracy studies from three nephrology journals indexed in MEDLINE. Two investigators hand searched the same journals to create a reference set of diagnostic test accuracy studies to compare search strategy results against. Results: We identified 103 diagnostic test accuracy studies, accounting for 2.1% of all studies published. The most specific search strategy was the Narrow Clinical Queries limit (sensitivity: 0.20, 95% CI 0.13–0.29; specificity: 0.99, 95% CI 0.99–0.99). Using the Narrow Clinical Queries limit, a searcher would need to screen three (95% CI 2–6) articles to find one diagnostic study. The most sensitive search strategy was van der Weijden 1999 Extended (sensitivity: 0.95; 95% CI 0.89–0.98; specificity 0.55, 95% CI 0.53–0.56) but required a searcher to screen 24 (95% CI 23–26) articles to find one diagnostic study. Bachmann 2002 was the best balanced search strategy, which was sensitive (0.88, 95% CI 0.81–0.94), but also specific (0.74, 95% CI 0.73–0.75), with a number needed to screen of 15 (95% CI 14–17). Conclusion: Diagnostic studies are infrequently published in nephrology journals. The addition of a strategy for diagnostic studies to a subject search strategy in MEDLINE may reduce the records needed to screen while preserving adequate search sensitivity for routine clinical use. Summary at a Glance: This study investigated the accuracy of 14 different search strategies to retrieve studies relating to diagnostic accuracy tests in three nephrology journals. The paper makes a strong argument for adding a strategy for diagnostic studies to a subject search strategy in MEDLINE to allow efficient access to these important studies.