Beynon

Appraisal of: "Beynon R, Leeflang MM, McDonald S, Eisinga A, Mitchell RL, Whiting P, Glanville JM. Search strategies to identify diagnostic accuracy studies in MEDLINE and EMBASE. Cochrane Database Syst Rev. 2013 Sep 11;(9):MR000022. doi: 10.1002/14651858.MR000022.pub3."

Reviewer(s):

Caroline Higgins

Anita Fitzgerald

Full reference:

Beynon R, Leeflang MM, McDonald S, Eisinga A, Mitchell RL, Whiting P, Glanville JM. Search strategies to identify diagnostic accuracy studies in MEDLINE and EMBASE. Cochrane Database Syst Rev. 2013 Sep 11;(9):MR000022. doi: 10.1002/14651858.MR000022.pub3.

Short description:

The objective of this methodology review was to assess the value of adding methodological search filters to search strategies to identify records for inclusion in DTA reviews.. The authors included nineteen studies reporting the development and/or evaluation of 57 MEDLINE filters and 13 EMBASE filters.

The authors assessed filter performance on standard measurements (as reported in the studies) including sensitivity, specificity and precision. Authors established a priori threshold levels of sensitivity (>90%) and precision (>10%) as the minimum levels of filter performance. Authors also assessed potential bias in the development of the filters as well as the methodological quality of filters using the ISSG Search Filter Appraisal Checklist.

Based on this review’s findings, none of the identified DTA filters had sufficiently high sensitivity required for systematic reviews with a reasonable degree of precision. This study’s findings support Whiting (2010) and the current Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy recommendation that the combination of methodological filter search terms for the index test and target condition should not be used as the only approach to inform systematic reviews of DTA.

Limitations stated by the study author(s):

Authors limited search to published studies, and did not search for unpublished filters. A potential source of bias could be linked to the ephemeral nature of filters; filters can be difficult to identify, some filters’ names may have been altered during publication, filters are not always listed in the paper or appendix and some may have been used inaccurately or adapted. In these situations the authors explain that they may have ascribed some performance data to avariant filter than the one described in the original study.

Limitations stated by the reviewer(s):

None.

Tags:

Search filters; MEDLINE; EMBASE; Diagnostic Test Accuracy

Study type:

Methodology Review

Supplemental publications to the study:

No.