Sensitivity of a search filter designed to identify studies reporting health state utility values

Article type
Authors
Arber M1, Garcia S1, Veale T1, Glanville J1
1York Health Economics Consortium, University of York, United Kingdom
Abstract
Background: The use of QALYs (quality-adjusted life years) by healthcare decision-makers means there is a need for methods that facilitate the effective and timely identification of studies reporting health state utility values (HSUVs). Despite the development of ScHARRHUD (a database of studies reporting HSUVs) there is still a need to identify studies not included in that database. Search filters have played a significant role in the effective and efficient retrieval of relevant studies in other areas of research such as clinical- and cost-effectiveness. The authors are unaware of any published, validated search filter that has been designed to identify studies reporting HSUVs. A validated search filter designed for this purpose would be a useful asset in systematic review (SR) and economic evaluation production.
Objectives: To test the sensitivity of a search filter designed to identify studies in MEDLINE that report HSUVs.
Methods: The relative recall method was used to assess how the search filter performed in retrieving a reference standard set of relevant studies reporting HSUVs. A pragmatic search to identify well-conducted published SRs of studies reporting HSUVs was undertaken in MEDLINE. Included studies in each identified SR that reported on HSUVs were checked to find out if they were available within MEDLINE. The identified records in MEDLINE formed the reference standard and they were saved as a saved search strategy within MEDLINE. We then assessed the performance of the search filter in retrieving records from this reference standard.
Results: This research is ongoing. Total records retrieved will be recorded and sensitivity will be reported.
Conclusions: The performance of the filter will be assessed and the limitations of the research will be highlighted. Implications for best practice and further research will be discussed.