2006. Evaluation of Rapid Blood Culture Identification with Antimicrobial Stewardship Treatment Recommendations at a Community Health System in Patients with Gram-negative Bacteremia: Adequacy, Adherence, and Outcomes

BACKGROUND: Sepsis mortality is greatly affected by the timely receipt of appropriate antibiotics. FilmArray Blood Culture Identification (BCID) is used at Cone Health to identify organisms in blood cultures within one to 2 hours after growth detected. The Cone Health antimicrobial stewardship (AMS) team has created treatment recommendations for each organism and resistance mechanism identifiable by BCID. Results and antibiotic recommendations are communicated in real time to providers by clinical pharmacists. The purpose of this evaluation was to validate the adequacy of antibiotics recommended by the BCID treatment algorithm for Gram-negative rods (GNR); assess proper implementation of the BCID notification procedure; and evaluate its effect on AMS. METHODS: Patients with GNR BCID results in January and April 2018 were retrospectively identified. Information collected for each patient included: demographics, location, organism, admission antibiotics, pharmacist compliance with BCID procedure, recommendation acceptance rate, organism susceptibility, changes to antibiotics post-BCID and final cultures, extended-spectrum β-lactamase (ESBL) incidence, length of antibiotic therapy, and patient outcome. RESULTS: A total of 101 patients were evaluated. The BCID treatment algorithm recommendations covered 97% of identified organisms (Figures 1–4). Resistant isolates were ESBL producers. Pharmacist antibiotic recommendations matched the treatment algorithm 66% of the time. Providers accepted 90% of pharmacist recommendations. Twenty-two percent of antibiotics were not de-escalated after BCID results without identifiable reason. CONCLUSION: The BCID treatment algorithm provided adequate coverage for nearly all identified organisms, except ESBLs. However, patients with ESBL organisms all survived to hospital discharge. Pharmacists are following the BCID protocol in a majority of cases. One-third of recommendations deviated from the algorithm but only 17% did not have documented reasoning. Providers are very receptive to pharmacist input, with only 8% of recommendations rejected without documented reasoning. Finally, nearly a quarter of empiric antibiotics were not de-escalated despite organism identification, which represents opportunity for improvement. [Image: see text] [Image: see text] [Image: see text] [Image: see text] DISCLOSURES: All authors: No reported disclosures.