2043. The “Resistance Calculator”: Refining Empiric Practices of Antimicrobials Prescription in the Era of Widespread Resistance

BACKGROUND: In the era of widespread resistance, there are two events in the course of a hospitalized septic patient where the majority of empiric prescription errors occur: (1) infections upon admission (UA) due to multi-drug-resistant organisms (MDRO) and (2) nosocomial infections due to extensively drug-resistant organisms (XDRO). These errors seriously impact patient outcomes and the ecological burden of resistance. Our objective was to develop a tool, to calculate the probability of MDRO UA, and nosocomial XDRO infections, in order to reduce delays in initiating appropriate therapy to the “right population,” i.e., with “resistant pathogens,” while avoiding overuse of broader (frequently more toxicת less efficacious) therapeutics to the “wrong population,” i.e., with “susceptible pathogens.” METHODS: Retrospective case–control analyses were conducted for septic adults at Shamir Medical Center, Israel (2016). Logistic regression was used to develop models of risk factors. All parameters incorporated into the models were readily accessible at the point of care. The performances of the development cohorts, and on 8 other validation cohorts, were assessed by the area under the receiver operating characteristic curve (ROC AUC). A web calculator (mobile modifiable) was generated. RESULTS: A total of 4,199 patients were enrolled: 2,472 with sepsis UA, and 1,727 with nosocomial sepsis. The “MDR UA score” included 10 parameters and with a cutoff of ≥22 points, had a ROC AUC of 0.85 (sensitivity 86%, NPV 98%). The “Nosocomial XDR score” included 7 parameters and with a cutoff of ≥36 points, had a ROC AUC of 0.88 (sensitivity 90%, NPV 96%). The median ROC AUC was 0.75 among the validation cohorts of the “MDR UA score,” and 0.66 among the “Nosocomial XDR score.” A free web tool was generated: https://assafharofe.azurewebsites.net/. CONCLUSION: A simple electronic calculator was generated to aid in bedside empiric prescription practices. The tool is composed of two scores to assist in common scenarios where the majority of errors occur. Prospective interventional investigations, should trial the performances of this tool in improving patient outcomes and the ecological burden in the facility. DISCLOSURES: All authors: No reported disclosures.