Searching for diagnostic test accuracy (DTA) studies can be frustrating and time-consuming, as those of us now engaged in producing systematic reviews of diagnostic test accuracy can attest! EMBASE is the first of the major electronic medical literature databases to introduce an indexing term specifically for DTA studies. The lack of such terms has been one of the factors hampering the development of more reliable and precise search filters to identify these studies in electronic databases, thus leading to the current recommendation that search strategies for Cochrane DTA reviews should not include a methodological filter. The introduction of a check tag for DTA studies will, prospectively, enhance retrieval of currently difficult to identify reports of DTA studies in EMBASE which meet the check tag definition.

We began discussions with Elsevier, the producers of EMBASE, in late 2008 about the desirability of introducing an indexing term for diagnostic test accuracy studies. We developed an initial proposal in 2009, and in July 2010 Elsevier asked us to submit a detailed proposal for the new check tag to the EMBASE indexing team. Our proposal included a Scope Note (definition) for diagnostic test accuracy study, some more comprehensive information about characteristics of DTA studies for the indexers’ manual, suggestions regarding the location and use of existing EMTREE terms relating to diagnostic test accuracy, and training materials for the EMBASE indexers.

Developing the Scope Note was quite a challenge. It had to be succinct, yet detailed enough to alert indexers to the possibility of a relevant study. Indexers are then able, if necessary, to refer to their manual for more comprehensive information to reach a decision. We had originally intended that the check tag would apply to primary studies only; however, Elsevier wished also to apply it to relevant systematic reviews, and we could see value in this approach. The Scope Note reads as follows: Diagnostic test accuracy study - Use for an original study or systematic review which assesses how accurately a test distinguishes humans (or animals) having a condition or disease, from those who do not. Typically, the test under evaluation is called the index test and its results are compared to the results of the best available standard test (reference standard), which defines the condition or disease.

We selected a series of reports of DTA studies and DTA systematic reviews, graded in difficulty, and already indexed in EMBASE for the indexer training materials. We marked up the title and abstract of each to show the triggers that should alert an indexer to the possibility of its being a DTA study or DTA systematic review. These included mention of index test, target condition, and analytical methods reported by the authors. We also highlighted the existing indexing to show where we thought each had (or had not) been appropriately indexed with the currently available EMTREE terms.

We should like to thank Elsevier for their enthusiasm and commitment to achieving this long-awaited outcome! We hope to report on progress with this development in future issues.

We would also like to acknowledge the contribution of Carol Lefebvre, co-convenor of the Cochrane Information Retrieval Methods Group. Over the years, improvements in information retrieval of difficult to identify studies have been achieved, most notably through the introduction of specific indexing terms for the randomised controlled trial. Carol has been instrumental in establishing good working relationships with database producers and lobbying successfully for these improvements on behalf of the Collaboration. Our work in the area of diagnostic test accuracy has undoubtedly benefited from this vital groundwork.

Ruth Mitchell
Cochrane Register of Diagnostic Test Accuracy Studies
ruthm4@chw.edu.au

Anne Eisinga
UK Support Unit for Cochrane DTA Reviews
aeisinga@cochrane.ac.uk

Julie Glanville
Information Retrieval Methods Group
jmg1@york.ac.uk

Mariska Leeflang
Continental European Support Unit for Cochrane DTA Reviews
m.m.leeflang@amc.uva.nl