Introduction: EPOC conducts systematic reviews of interventions to improve professional practice and the delivery of effective health services. Our scope includes the following kinds of interventions: professional (various forms of continuing education, quality assurance and informatics), organisational (changes in the structure or delivery of health care), financial (changes in health professional reimbursement, incentives and penalties), and regulatory (changes in medical liability, peer review and licensure). Although well-conducted randomised trials (RCTs) provide the most reliable evidence on the effectiveness of interventions, these are not always feasible for interventions within EPOC's scope, therefore we also allow reviewers to include controlled before and after (CBA) and interrupted time series (ITS) study designs.

Objectives: To describe the use of non-randomised evidence in EPOC reviews.

Methods: Description of the methods developed to support inclusion of non-randomised evidence. Description of the use of non-randomised evidence in current EPOC reviews.

Results: We have developed: a gold standard search strategy to identify studies in our scope including CBA and ITS designs, which is 90% sensitive and 18% specific; and standard criteria to assess the methodological quality of CBA and ITS designs. Our specialised register includes over 1000 studies, approximately 25% are either CBA or ITS designs. Currently there are 12 systematic reviews and 19 protocols published in the EPOC module of The Cochrane Library (Issue 1 1999). Five of the reviews include CBA or ITS studies and one review of the effects of mass media interventions only included studies with ITS designs. Of the 19 protocols, 13 plan to include CBA or ITS study designs.

Discussion: EPOC has developed considerable experience of using non-randomised evidence within our reviews. High quality non-randomised studies make an important contribution to EPOC systematic reviews.