186. Effective Antimicrobial StewaRdship StrategIES (ARIES): Cluster-Randomized Trial of a Clinical Decision Support System to Supplement Antibiotic Prospective Review and Feedback

Abstract

BackgroundProspective review and feedback (PRF) of antibiotic prescriptions is a tenet of antimicrobial stewardship (ASP), but labour intensive. Clinical Decision Support Systems (CDSS) have the potential to automate some of this work. We hypothesised that increasing prescriber engagement with the CDSS would reduce the requirement for PRF by the ASP team and improve prescribing behaviour without causing harmMethodsA parallel-group, 1:1 block-cluster randomized, cross-over study was conducted in 32 medical and surgical wards from March 2017 to August 2017. Participants in Arm A were allocated to voluntary use of CDSS by the clinician at first prescription of piperacillin–tazobactam or a carbapenem, while in Arm B, CDSS use was compulsory. PRF continued for both arms.ResultsSix hundred fourty-one and 616 participants were included in Arms A and B, respectively. At baseline, Charlson’s co-morbidity and APACHE II scores were comparable. Initial antibiotic prescriptions were similar, and the majority were for respiratory (67.0% vs. 68.2%) or urinary (17% vs. 19.6%) infections.CDSS recommendations were provided to 20.6% of participants in Arm A and 99.4% in Arm B (P < 0.01). Arm B adopted a higher number of CDSS antibiotic de-escalation (1.1% vs. 2.6%), dose optimization (9.7% vs. 30.7%), antibiotic optimization (8.9% vs. 31.3%), and duration setting recommendations (10.9% vs. 50%). The proportion of participants receiving PRF recommendations were not, however, significantly different between arms (8% vs. 11.5%, P = 0.13). The types of PRF recommendations and prescriber acceptance rates were also similar. The duration of antibiotic use was significantly shorter when prescribers were compelled to use the CDSS (daily defined doses ≤3: 71.8% in Arm B, 64.9% in Arm A, P < 0.01). There was no evidence of harm from the CDSS, with similar 30-day mortality (HR 0.87, 95% CI 0.67–1.12), 30-day re-infection (20.6% vs. 23.1%, P = 0.29) and 30-day re-admission rates (14.4% vs. 14.1%, P = 0.91). The median length of hospital admission was also similar (15 IQR 5–64 vs. 15, IQR 4–70 days).ConclusionCompulsory use of a CDSS at antibiotic prescription did not reduce the requirement for PRF, but limited the duration of antibiotic courses, without compromising clinical outcomesDisclosures All authors: No reported disclosures.