Skip to main content

Languages

Reporting methodological search filter performance comparisons: a review

Date and Location

Session: 

P3.087

Date

Sunday 22 September 2013 - 10:30 - 12:00

Location

Presenting author and contact person

Presenting author

Jenny Harbour

Contact person

Jenny Harbour
Jenny Harbour
Abstract text
Background: Effective retrieval of literature is essential for evidence-based healthcare. Methodological search filters are tools for retrieving studies using specific research method. As the number of filters proliferates, choice of filter is likely to be based on filter performance data. Aim: To review measures reported and presentation methods in comparing methodological search filter performance. Methods: Studies were identified from the InterTasc Information Specialists Subgroup (ISSG) Search Filter Resource website, a search by the Cochrane UK Centre and references from a concurrent review. Included studies compared two or more methodological search filters for randomised controlled trials (RCTs), diagnostic test accuracy studies (DTA), systematic reviews (SR) or economic evaluations. Results: Eighteen papers met the review inclusion criteria comparing DTA (8), RCT (5), SR (3), economic evaluations (1) and both RCT and SR filters (1). The number of filters compared in a single study ranged from 2 to 38. Seventeen studies assessed performance against a gold standard set of studies derived from hand-searching journals (10), included studies from systematic reviews (5) or from database records (2). The most commonly reported measures were sensitivity/recall (16) and precision (13). Specificity was calculated for seven studies. All studies displayed results in tables, generally listing filters in descending order by the measure of interest. Results were given as percentages or proportions in all studies. Two studies supplemented results tables with graphical displays of data: bar graph of the proportion of retrieved and missed gold standard references per filter; forest plot of the overall sensitivity and specificity of each filter. Conclusions: The most commonly reported measures in filter performance comparisons were sensitivity and precision. Methods of presenting results of filter performance comparisons were limited to tables and a small number of graphs.