Background: The scope of EPOC includes professional, organizational, financial and regulatory interventions to improve the quality of health care. Since randomized controlled trials (RCT) and controlled clinical trials (CCT) are not always feasible, interrupted time series (ITS) and controlled before and after (CBA) study designs are included. The latter two study types are particularly difficult to search with good precision as no MeSH or publication types are available. A gold standard EPOC search filter was developed in 1997 by Cynthia Fraser and the EPOC editorial team.1 An updated 'gold standard' of studies known to be within the scope of EPOC was used to assist in the development of the search strategy. This included citations identified by handsearching and by screen searching full text records.
Objectives: This abstract will describe the steps taken to revise the current EPOC methodological filter used to retrieve citations from MEDLINE.
Methods: A three month interval (June - August 2006) was used for testing during regular updates of MEDLINE using the OVID interface. The search results were imported into Idealist software in separate files for pre-screening and coded to determine how many were suitable for inclusion in the EPOC specialised register. All citations accepted for inclusion in the register during the test period were then exported in SPSS software for frequency analysis of the MeSH, Label (study design) and Publication Type fields with a particular emphasis on studies labelled as CBA and ITS. Based on the frequency analysis, we will design a revised search filter and evaluate its sensitivity and sensitivity on a second dataset. A review of current search literature and of the included citations in the EPOC register will also be used to update the existing filter.
Results: 457 short filter and 4,177 long filter citations were originally sent from MEDLINE to Idealist. Of these, 233 short filter and 387 long filter citations were sent to the pending folder of the EPOC specialised register for further screening. A random number generator will be used to take a sample of the included citations to ensure the updated study design filter incorporates the results of the filter test. Current EPOC study design filter (MEDLINE): 1. randomized controlled trial.pt. 2. controlled clinical trial.pt. 3. intervention studies/ 4. experiment$.tw. 5. (time adj series).tw. 6. (pre test or pretest or (posttest or post test)).tw. 7. random allocation/ 8. impact.tw. 9. intervention?.tw. 10. chang$.tw. 11. evaluation studies/ 12. evaluat$.tw. 13. effect?.tw. 14. comparative study.pt. 15. or/1-14
Conclusions: The information provided by this analysis will provide the EPOC editorial base with an enhanced search filter which will reduce the time required to screen studies and improve the results sent to the EPOC specialised register each month. It is anticipated that the results of the testing will provide the basis for revisions of the EPOC methodology search filter in other databases.

Reference
1. Fraser C, Thomson MA on behalf of the Cochrane Effective Practice and Organisation of Care Review Group. Identifying non-randomised studies in Medline. 6th Cochrane Colloquium; 1998 Oct 22-26; Baltimore, Maryland. Available from: http://www.cochrane.org/colloquia/abstracts/baltimore/MarylandB18.htm