189. Optimizing Empiric Antibiotic Therapy: a Probabilistic Approach

BACKGROUND: How to start optimal antibiotic therapy before the results of cultures and antimicrobial susceptibility tests are available? Here, we use the law of total probability to present a probabilistic approach based on antibiograms of bacterial isolates from healthcare and community-acquired infections to optimizing empiric antibiotic therapy. METHODS: Data on the microbiology of healthcare and community-acquired infections were analyzed from hospitals in Belo Horizonte, a three million inhabitants city from Brazil. Healthcare infections were defined by the National Healthcare Safety Network (NHSN)/CDC protocols. Only data obtained from infections with positive culture, both hospital and community, were considered. The success rate of an antibiotic (ATB) regimen, considering just one drug individually (monotherapy), was calculated by Law of Total Probability (Fig 1). In this sense, if a microorganism has not been tested for a specific antimicrobial, then, by definition, it was considered an antibiotic failure. For a regimen with more than one antibiotic, if the microorganism is sensitive to one of them, then it was considered a success of the scheme. For calculating the success probability of two or three antimicrobials A, B, and C, simultaneously (Fig 2), i.e., P(A and B) or P(A and B and C), the sensitivity to an antimicrobial was considered independent of sensitivity to any other. Then, P(A and B) = P(A) * P(B), and P(A and B and C) = P(A)*P(B) *P(C). [Figure: see text] [Figure: see text] RESULTS: Microbiologic data from hospital acquired infections (HAI) and community-acquired infections (CAI) are analyzed once a year. Empiric antibiotic therapy to HAI were proposed for urinary tract infections (UTI), bloodstream infections (BSI), and pneumonia (Figures 2 and 3). Empiric antibiotic therapy to community-acquired infections were developed for UTI, pneumonia, gastrointestinal system infection, bone and joint infection, and skin and soft tissue infection. [Figure: see text] [Figure: see text] [Figure: see text] CONCLUSION: We presented here a probabilistic approach to empiric antibiotic therapy. The next step is to validate all proposed regimens, that can be used to improve the success likelihood of empiric antibiotic decision making. DISCLOSURES: All Authors: No reported disclosures.