We also expected to find a greater impact of CDSSs on prescribing outcomes than clinical outcomes and examined whether multi-faceted Cisplatin in vivo interventions would have a greater impact than CDSS alone. We included English-language studies, published between 1990 and March 2009, reporting RCTs and strong quasi-experiments (non-randomised studies with comparison groups or interrupted

time-series designs with or without comparison groups). The studies had to: target pharmacists; compare the performance of the CDSSs to routine care and/or paper-based decision support; provide information that could be applied to a specific patient (e.g. provide advice to prescribe a particular drug, to monitor a drug or adjust the dose or to perform laboratory tests related to safe prescribing); generate information or advice to the pharmacist in an electronic format (however, subsequent delivery of information to physicians or patients could be in electronic or paper formats); and report data on at least one outcome relating to prescribing, clinical or patient outcomes (see Table 1). Studies were excluded if: interventions were based around hypothetical scenarios rather than actual clinical

practice, studies did not undertake statistical analyses or studies reported only cost outcomes. We searched Medline (1990–March see more 2009), PreMedline (18 March 2009), Embase (1990–March 2009), CINAHL (1990–March 2009) and PsycINFO (1990–March 2009). We combined keywords and subject headings to identify computer-based

decision support (e.g. decision support systems clinical, decision making computer assisted), medicines use (e.g. prescription Vasopressin Receptor drug, drug utilization) and pharmacy or pharmacists. We also searched INSPEC (March 2009) and the Cochrane Database of Systematic Reviews (March 2009) including reviews and protocols published under the Effective Practice and Organisation of Care Group (EPOC). Finally, we hand-searched the reference lists of retrieved articles and reviews. Table 2 details the full search strategy. Two reviewers (JR, EW) evaluated independently the study titles and abstracts identified in the search. Full-text articles were retrieved if either reviewer considered a citation potentially relevant. Studies deemed eligible for review underwent data extraction by two reviewers (JR, EW). Disagreements were resolved by discussion to reach consensus. We extracted the following information from eligible studies: objectives, clinical setting (ambulatory or institutional care) and details of the decision-support intervention (e.g. system-initiated or user-initiated, multi-faceted or CDSS alone, clinical target). We classified studies as having a safety or a QUM focus. Given the lack of uniformity in relation to terminology about prompts, alerts and reminders we extracted details as they were reported in the articles.