Carbapenem-resistant Enterobacteriaceae carriage risk for parameterization of a regional healthcare network agent-based model

• Published in: Infection control and hospital epidemiology 2020-10, Vol.41 (S1), p.s148-s149
• Main Authors: Rhea, Sarah, Li, Lei, Iyer, Pooja, DiBiase, Lauren, Jones, Kasey, Hilscher, Rainer, Sickbert-Bennett, Emily, Bobashev, Georgiy, Rineer, James, Weber, David J.
• Format: Article
• Language: English
• Subjects: Antibiotics; Comorbidity; CRE bacteria; Disease control; Disease transmission; E coli; Health care; Hospitals; Intensive care; Length of stay; Patients
• ISSN: 0899-823X; 1559-6834
• DOI: 10.1017/ice.2020.667

Background: Carbapenem-resistant Enterobacteriaceae (CRE) are increasingly common in the United States and have the potential to spread widely across healthcare networks. Only a fraction of patients with CRE carriage (ie, infection or colonization) are identified by clinical cultures. Interventions to reduce CRE transmission can be explored with agent-based models (ABMs) comprised of unique agents (eg, patients) represented by a synthetic population or model-generated representation of the population. We used electronic health record data to determine CRE carriage risk, and we discuss how these results can inform CRE transmission parameters for hospitalized agents in a regional healthcare network ABM. Methods: We reviewed the laboratory data of patients admitted during July 1, 2016−June 30, 2017, to any of 7 short-term acute-care hospitals of a regional healthcare network in North Carolina (N = 118,022 admissions) to find clinically detected cases of CRE carriage. A case was defined as the first occurrence of Enterobacter spp, Escherichia coli , or Klebsiella spp resistant to any carbapenem isolated from a clinical specimen in an admitted patient. We used Poisson regression to estimate clinically detected CRE carriage risk according to variables common to data from both the electronic health records and the ABM synthetic population, including patient demographics, systemic antibiotic administration, intensive care unit stay, comorbidities, length of stay, and admitting hospital size. Results: We identified 58 (0.05%) cases of CRE carriage among all admissions. Among these cases, 30 (52%) were ≥65 years of age and 37 (64%) were female. During their admission, 47 cases (81%) were administered systemic antibiotics and 18 cases (31%) had an intensive care unit stay. Patients administered systemic antibiotics and those with an intensive care unit stay had CRE carriage risk 6.5 times (95% CI, 3.4–12.5) and 4.9 times (95% CI, 2.8–8.5) higher, respectively, than patients without these exposures (Fig. 1). Patients ≥50 years of age and those with a higher Elixhauser comorbidity index score and with longer length of stay also had increased CRE carriage risk. Conclusions: Among admissions in our dataset, CRE carriage risk was associated with systemic antibiotic exposure, intensive care unit stay, higher Elixhauser comorbidity index score, and longer length of stay. We will use these risk estimates in the ABM to inform agents’ CRE carriage status upon hospital admissio