313. Using a Clinical Decision Prediction Tool to Improve Empirical Antimicrobial Therapy in Ceftriaxone-Resistant Enterobacterales Bloodstream Infections

Abstract

BackgroundThere are several clinical tools for prediction of antimicrobial resistance. However, their utility in management of bloodstream infections (BSI) due to resistant bacteria remains unclear. This quasi-experimental cohort examined the impact of utilization of the extended-spectrum beta-lactamase (ESBL) prediction score on time to initiation of appropriate antimicrobial therapy (AAT) in BSI due to ceftriaxone-resistant (CRO-R) Enterobacterales.MethodsAdults with first episodes of monomicrobial BSI due to CRO-R Enterobacterales at Prisma Health-Midlands Hospitals in South Carolina from January 2010 to December 2017 were included. Antimicrobial stewardship intervention was implemented in January 2014 consisting of real-time alerts for positive blood cultures and rapid diagnostics for identification of bloodstream isolates. The ESBL prediction score was used to stratify risk of resistance prior to conventional antimicrobial susceptibility results. Student’s t-test was used to compare mean time to AAT before and after intervention.ResultsAmong 92 patients with BSI due to CRO-R Enterobacterales, median age was 66 years, 52 (57%) were men, 68 (74%) had community-onset BSI, and 52 (57%) had urinary source of infection. Escherichia coli 52 (57%) and Klebsiella species 27 (29%) were the most common bloodstream isolates and the majority 63 (68%) produced ESBLs. There were no significant differences in demographics or clinical characteristics of patients before (n=45) and after (n=47) antimicrobial stewardship intervention. Mean time to AAT was 3.2 days prior and 1.7 days after utilization of ESBL prediction score (p=0.021). Results were consistent among ESBL-producing Enterobacterales (time to AAT 2.8 before and 1.8 days after intervention, p=0.025).ConclusionUtilization of the ESBL prediction score as part of a real-time antimicrobial stewardship intervention significantly reduced time to AAT in BSI due to CRO-R and ESBL-producing Enterobacterales. The ESBL prediction score represents a viable tool to improve antimicrobial management in clinical settings that lack access to multiplex PCRs for detection of antimicrobial resistance genes and novel methods for rapid phenotypic antimicrobial susceptibility testing.DisclosuresJulie Ann Justo, PharmD, MS, BCPS-AQ ID, bioMerieux (Speaker’s Bureau)TRC Healthcare (Speaker’s Bureau)