Introduction: Non-variceal upper gastrointestinal bleeding (NVUGIB) is a common presentation to the emergency department (ED) accounting for significant morbidity, mortality and health care resource usage. In Alberta, a provincial care pathway was recently developed to provide an evidence informed approach to managing patients with an UGIBs in the ED. Pantoprazole infusions are a commonly used treatment despite evidence that suggests they are generally not indicated prior to endoscopy in the ED. The goal of this project was to optimize management of patients with a NVUGIB, in particular reduce pre-endoscopy pantoprazole infusions. Methods: In July 2016, we implemented a multi-faceted intervention to optimize management of ED patients with NVUGIB including 1. de-emphasizing IV pantoprazole infusions in the ED, 2. clinical decision support (CDS) embedded (for endoscopy, disposition and transfusions) within the order set and 3. educating clinicians about the care pathway. We used a pre/post-order set design, analyzing 391 days pre and 189 days post-order set changes. Data was extracted from our fully integrated electronic health records system. The primary outcome was the % of patients receiving IV pantoprazole infusion ordered by an emergency physician (EP) among all patients with NVUGIB. Secondary outcomes included % transfused with hgb >70g/L and whether using the GIB order set impacted management of NVUGIB patients. Results: In the 391 days pre-order set changes, there were 2165 patients included and in the 189 days post-order set changes, there were 901 patients. For baseline characteristics, patients in the post-order set change group were significantly older (64.4 yrs vs 60.9 yrs p-value=0.0016) and had a lower hgb (115 vs 118, p-value=0.049) but otherwise for gender, measures of severity of illness (systolic blood pressure, heart rate, CTAS, % admitted) there were no significantly differences. For the primary outcome, in the pre-order set phase, 47.1% received a pantoprazole infusion ordered by an EP, compared to 31.5% in the post-order phase, for an absolute reduction of 15.6% (p-value= <0.001). For the secondary outcomes, transfusion rates were similar pre/post (22.08% vs 22.75%). Significant inter-site variability exists with respect to the reduction in pantoprazole infusion rates across the four sites (-23.3% to +6.12%). Conclusion: Our interventions resulted in a significant overall reduction in pantoprazole infusions in ED patients with NVUGIB. Reductions in pantoprazole infusions varied significantly across the different sites, future work in our department will explore and address this variability. Keys to the success of this project included engaging clinicians as well as leveraging the SCM order sets as well as the provincial care pathway. Although there were no changes in transfusion rates, it in unclear if this a function of the CDS not being effective or whether these transfusions were clinically indicated.