66 A computational phenotype for carbapenem-resistant Enterobacterales and extended-spectrum beta-lactamase-producing Enterobacterales infections to support electronic health record-based surveillance efforts

Objectives/Goals: The objective of this project was to develop electronic health record-based algorithms, or computational phenotypes, for carbapenem-resistant Enterobacterales (CRE) and extended-spectrum beta-lactamase-producing Enterobacterales (ESBL) infections based on public health surveillance case definitions. Methods/Study Population: We analyzed electronic health records (EHR) from a retrospective cohort of patients within the University of Colorado Health System. We detected CRE and ESBL infections using a set of string-matching algorithms for bacterial organism, specimen source, and antibiotic susceptibility test results. Infections were limited to the first case of each organism per patient in a 30-day period. We conducted manual chart review on a subset of cases and non-cases for validation purposes. We then performed a descriptive analysis to examine demographic characteristics of patients with CRE and ESBL infections including sex, age, infection type, and organism. All analyses were conducted using R statistical software. Results/Anticipated Results: There were 634 CRE infections from 2013 to 2023, with the majority from urine (n = 448, 70.7%) or blood specimens (n = 56, 8.8%). Over half of CRE case-patients were female (n = 362, 57.1%). The mean age was 65.0 years. Most isolates were identified as Enterobacter cloacae complex species (n = 353, 55.6%). There was a total of 4,363 ESBL infections from 2019 to 2023, and most were urine (n = 3,550, 81.4%) or blood specimens (n = 457, 10.5%). Most ESBL case-patients were female (n = 3,030, 69.4%) with a mean age of 61.2 years. The majority of isolates were identified as Escherichia coli (n = 2,932, 67.2%), followed by Klebsiella pneumoniae (n = 368, 8.4%). Validation results are pending, including sensitivity and specificity of the algorithms. Discussion/Significance of Impact: Demographic characteristics of CRE and ESBL case-patients were similar to findings from traditional public health surveillance. The algorithms will supplement traditional surveillance methods, inform future epidemiological research using EHR data, as well as contribute to the development of standardized EHR-based definitions for CRE and ESBL.