Focus EMTREE terms vs. Non-focused EMTREE terms to search for clinical drug trials in EMBASE

Tags: Poster
Hausner E, Simon M, Kaiser T

Background: Systematic reviews require comprehensive literature searches in different bibliographic databases. Beside MEDLINE, EMBASE gets routinely selected for searching randomized controlled trials (RCT) to conduct systematic reviews of drugs. However, in comparison to MEDLINE, EMBASE searches for active agents generate substantial higher number of hits without yielding more relevant documents. A more focussed search may therefore be reasonable to reduce the workload. Objectives: This paper has two aims. First, to compare sensitivity of Focus EMTREE terms vs. Non-focused EMTREE terms search strategies, and second, to verify whether Focus EMTREE terms in a comprehensive search for active agents are reasonable. Methods: Different search strategies of Focus EMTREE terms or Non-focused EMTREE terms were used to search for several active agents. Based on a search in the Cochrane Central Register of Controlled Trials (CENTRAL) and in systematic reviews, a test set was built to generate the ‘‘gold standard’’ (clinical trials with these drugs). Based on this test set sensitivity was calculated. Additionally, each approach was combined with an established RCT filter to assess the differences in a ‘real world’ setting. Results: Focus EMTREE terms alone reached sensitivity of about 90%. After adding text words, sensitivity rose to about 99% to 100%. In the ‘real world’ setting the number of retrieved records was substantially reduced by up to 50%. Conclusions: The use of Focus EMTREE terms plus textwords in the search for active agents in EMBASE is justifiable. The reduced number of hits will lead to considerable time savings with almost no loss in sensitivity.