The authors concluded that electronic prescribing decision support systems can be implemented and have potential to change health care provider behaviour. Aspects of this review were well-conducted, but due to the unknown quality of the included studies and the lack of reporting of study results, the reliability of the results could not be assessed.

To assess the effectiveness of prescribing decision support systems (RxCDSS) and to determine what features of RxCDSS predict successful implementation, change in provider behaviour and change in patient outcomes.

MEDLINE, EMBASE, CINAHL and Inspec were searched from inception to June 2008 for English-language publications. Search terms were provided.

Randomised controlled trials (RCTs) that evaluated a RxCDSS (considered to be use of a computer to analyse patient-specific information to advise a prescriber when writing a prescription or a pharmacist when filling a prescription) were eligible for inclusion. Support had to be generated by computer, but could be delivered in any way (such as computer terminal, fax, mail, patient record insert). Only systems that intervened before a drug therapy had been chosen by a physician or that had the ability to suggest alternate therapies other than that initially prescribed by the physician were eligible for inclusion. Systems that offered "fine-tuning" advice on a predefined therapy were excluded. Systems that focused primarily on diagnosis, vaccination or nutrition were excluded. Study outcomes focused on three domains: implementation; change in health care provider behaviour; and change in patient outcomes.

Included studies evaluated RxCDSS in a number of settings; most were in an out-patient general practice or internal medicine setting. The systems addressed a variety of problems; most were in cardiovascular care and next most was general/internal medicine. A large proportion of systems were integrated with drug order entry. Control group was usual care in most trials.

Two reviewers independently selected the studies for inclusion. Any disagreements were resolved by consensus or a third reviewer.

Validity of each study was assessed using the following criteria: unit of allocation; baseline differences between groups; objectiveness of the outcome; and completeness of follow-up. For each criteria, a score of 0 to 2 was given (maximum score was 8).

Two reviewers independently assessed validity of included studies. Any disagreements were resolved by consensus.

In each study, outcomes were assessed for success: implementation was deemed successful if an RxCDSS was implemented and utilised by clinical staff; a successful change in health care provider behaviour required a change such as a decrease in inappropriate prescribing or a change to a more cost-effective therapy; successful outcomes in participants' health included decreases in morbidity and mortality.

Two reviewers independently extracted study data.

The studies were summarised in a narrative synthesis.

Forty-one RCTs with a total of 612,556 participants (sample sizes ranged from 169 to 407,460) and 2,963 providers (sample sizes ranged from 17 to 334) were included in the review.

Mean validity assessment score was 5.9 (range 2 to 8). The authors considered this to be good quality.

Thirty-seven trials reported successful implementation. A change in health care provider behaviour was reported in 25 trials. Patient outcomes were found to be improved in five out of 23 trials that reported this outcome.

Associations between specific RxCDSS features and successful outcomes could not be statistically evaluated.

Electronic prescribing decision support systems had potential to change health care provider behaviour, but very few high-quality studies demonstrated improvement in patient outcomes. Features of RxCDSS success (or failure) were poorly presented in the literature.

This review addressed a clear research question and was supported by appropriate inclusion/exclusion criteria. A number of databases were searched. The search was limited to English-language studies, which potentially introduced language bias. Unpublished data were not sought, which may have introduced publication bias. Steps were taken to minimise reviewer bias at each stage of the review process. Validity of included studies was assessed, although criteria used were not comprehensive for a RCT and so the quality of reporting of included studies may not have been fully evaluated. The studies were appropriately summarised in a narrative synthesis. However, no detailed results of the included studies were presented in the paper or in an online supplement. Due to the unknown quality of the included studies and the lack of reporting of study results, the reliability of the results could not be assessed.

Practice: The authors did not state any implications for practice.

Research: The authors stated that RxCDSS trials that dealt with drug dosing suggestions should be systematically reviewed.

Funded by a grant from the Ontario Ministry of Health's Primary Healthcare Transition Fund.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.