ISSG structured abstract for:
Bachmann LM, Estermann P, Kronenberg C, ter Riet G. Identifying diagnostic accuracy studies in EMBASE. Journal of the Medical Library Association 2003;91(3):341-6.
Structured abstract prepared by: Julie Glanville
Objective:
To develop and test search strategies to identify diagnostic test accuracy records in EMBASE.
Methods:
Four medical journals were hand searched for diagnostic test studies for the years 1999. Potential strategies were developed on and tested on the same gold standard. Candidate search terms were identified by word frequency analysis. Sensitivity and precision were calculated for each term and terms with the highest sensitivity x precision product were tested in a series of search strategies.
Results:
A gold standard of 61 records was identified. 8 strategies were developed and tested. 3 filters were recommended. A sensitive strategy achieving 100% sensitivity in the test gold standard, and a Number Needed to Read of 27. A sensitivity and precision compromise strategy is offered with a sensitivity of 73.8% and a NNR of 5.7.
Discussion:
The search strategies offer reasonable performance, but the authors acknowledged that the generalisability of the strategies, given a small gold standard representing records in one year, was untested. The strategies were not validated against gold standards other than the one from which they were derived.
ISSG commentary:
The strategies were developed and tested on a single small data set for one year, and it would be desirable to see performance data for other validation sets of records.