Poster presentation: Hand-searching for studies other than randomised trials

Bero L., Grimshaw J. and Hyde C. Hyde C., UK Cochrane Centre, Summertown Pavilion, Middle Way, Oxford 0X2 7LG, UK.

Introduction: CCEPP (the Cochrane Collaboration on Effective Professional Practice) is reviewing the effects of educational, organizational and financial interventions on professional behaviour. The effects of these interventions cannot always be examined using randomized controlled trials. Therefore, the specialised register being developed by CCEPP will contain a range of study designs.
Objective: To develop a hand-search strategy identifying such a range of studies.
Methods: Three journals were hand-searched independently by the authors, Medical Care (1973-4,1991 & 1993-4), Journal of Health Politics, Policy and Law (1976-1995) and American Journal of Hospital Pharmacy (1990, 1994-5). Problems identifying studies were catalogued and discussed.
Results: The greatest difficulty encountered was distinguishing between some categories of study design. Poor reporting was the main reason for this, rather than the absence of an adequate theoretical construct. However, there was a clear watershed between those studies which had a comparative element, either in time or person, and those which did not. The CCEPP hand-search criteria were hence established as studies where: (A) an educational, financial or organizational intervention has been applied with the purpose of changing the behaviour of health professionals, AND (B) there is a control group OR pre-intervention period of observation, AND (C) there is an objective measure of professional performance AND/OR patient outcome. The numbers of studies identified on the basis of these criteria are shown below:

Journals (Years) Randomised or quasirandomised controlled trials Controlled trials Time series analyses & other pre-post designs

J.Health Politics, Policy & Law (76-95) 2 1 9

Amer. J.Hospital Pharmacy (90, 94-5) 1 0 2

Medical care (73-4, 91, 93-4) 8 12 18

Total 11 3 29

Discussion: These results suggest an approximate doubling of study yield in comparison with that using existing Cochrane Collaboration hand-search criteria. Further work is in progress to formally compare the yield of studies identified by the Cochrane Collaboration criteria (as reflected by the results of an existing hand-search co-ordinated by the Baltimore Cochrane Centre) with the yield resulting from application of the CCEPP hand-search criteria to all years of Medical Care, and to measure the additional resources consumed in using the CCEPP hand-search criteria. The results of this work will be available to amplify the presentation at the Colloquium.

Oslo 1995 P32