A systematic review of the effectiveness of interruptive medication prescribing alerts in hospital CPOE systems to change prescriber behavior and improve patient safety.

OBJECTIVES: To assess the evidence of the effectiveness of different categories of interruptive medication prescribing alerts to change prescriber behavior and/or improve patient outcomes in hospital computerized provider order entry (CPOE) systems.

METHODS: PubMed, Embase, CINAHL and the Cochrane Library were searched for relevant articles published between January 2000 and February 2016. Studies were included if they compared the outcomes of automatic, interruptive medication prescribing alert/s to a control/comparison group to determine alert effectiveness.

RESULTS: Twenty-three studies describing 32 alerts classified into 11 alert categories were identified. The most common alert categories studied were drug-condition interaction (n=6), drug-drug interaction alerts (n=6) and corollary order alerts (n=6). All 23 papers investigated the effect of the intervention alert on at least one outcome measure of prescriber behavior. Just over half of the studies (53%, n=17) reported a statistically significant beneficial effect from the intervention alert; 34% (n=11) reported no statistically significant effect, and 6% (n=2) reported a significant detrimental effect. Two studies also evaluated the effect of alerts on patient outcome measures; neither finding that patient outcomes significantly improved following alert implementation (6%, n=2). The greatest volume of evidence relates to three alert categories: drug-condition, drug-drug and corollary order alerts. Of these, drug-condition alerts had the greatest number of studies reporting positive effects (five out of six studies). Only two of six studies of drug-drug interaction and one of six of corollary alerts reported positive benefits.

DISCUSSION AND CONCLUSION: The current evidence-base does not show a clear indication that particular categories of alerts are more effective than others. While the majority of alert categories were shown to improve outcomes in some studies, there were also many cases where outcomes did not improve. This lack of evidence hinders decisions about the amount and type of decision support that should be integrated into CPOE systems to increase safety while reducing the risk of alert fatigue. Virtually no studies have sought to investigate the impact on changes to prescriber behavior and outcomes overall when alerts from multiple categories are incorporated within the same system.