Pruning Emtree: Does Focusing Embase Subject Headings Impact Search Strategy Precision and Sensitivity?

Although Embase is a recommended key database to be searched when undertaking health technology assessments (HTAs), it has several features that hinder efficient searching. One feature is the large number of Emtree index terms that are added to most Embase records; an average of 3 to 4 major terms and up to 50 minor terms. In comparison, MEDLINE records may contain an average of 10 to 20 index terms (both major and minor). The volume of index terms can lead to poor precision in Embase searches (large proportions of irrelevant records retrieved) if the terms that are only of marginal relevance to a specific record are added by the indexers. When this occurs, it can add to the record processing burden within the HTA process. This experience has led to informal, pragmatic recommendations that search results can be reduced by carrying out searches of subject headings combined with subheadings (qualifiers) and/or searches with subject headings limited to those with a major focus (major headings). This is achieved in Embase’s OvidSP interface by using the “Restrict to Focus” option when selecting Emtree subject headings. It is clear that research evidence to support such pragmatic decisions is needed. This project has been developed to explore whether it is safe in the context of the imperative when conducting HTAs, and will not miss relevant studies reporting relevant effects data, to carry out searches of Embase using some or all major subject headings in the search, rather than all (i.e., non-major) subject headings.