2167. Predicting Carbapenem-Resistant Enterobacteriaceae (CRE) Carriage on Admission using Updated Statewide Hospital Discharge Data
Session: Poster Abstract Session: Healthcare Epidemiology: HAI Surveillance
Saturday, October 6, 2018
Room: S Poster Hall
Posters
  • IDWeek Poster_FINAL.pdf (779.0 kB)
  • Background:

    We previously built a patient-level prediction model to assess an individual’s risk of Carbapenem-resistant Enterobacteriaceae (CRE) carriage upon hospital admission based on the following factors: past hospital visits (short- and long-term acute care (STACHs & LTACHs)), endoscopic procedures, infection-related diagnosis codes, and patient age and sex. Our model discriminated CRE cases relatively well (c-statistic = 0.86). In the hopes of operationalizing our results, we evaluated the distribution of predicted probabilities on an updated dataset using existing model parameters.

    Methods:

    We used Illinois Hospital discharge data (CYs 2015-2016) with ICD-10 diagnosis and procedure codes to establish baseline exposure history (2015) and to generate predicted probabilities (2016). We calculated the number of hospital visits and the average number of hospital days in the past year (STACH and LTACH). We identified infection-related diagnosis codes using prior knowledge, and included procedure codes for endoscopic retrograde cholangiopancreatography (ERCP). We then used the model parameters from our previous work to generate predicted probabilities corresponding to each hospital visit.

    Results:

    Our study year (2016) included 1229158 visits by 816500 unique adult patients. 62% of patients had no inpatient visits in the previous year. Among those with a prior hospitalization, the median STACH length of stay was 4 days (IQR: 2-6). 3566 patients (0.4%) had previous LTACH exposure upon admission, with a median length of stay of 25 days (IQR: 13-40). 32% of hospital visits had an infection-related diagnosis code, and 0.5 percent had an ERCP procedure code. Of the more than 1.2 million visits, our model predicted 10614 visits associated with a CRE risk of over 1%, 946 visits of over 10%, and 96 visits by 63 unique patients with over a 50% risk. On average, highest risk patients were exposed to (median) 15 (7-97) STACH, 104 LTACH (37-174) days; 83% had infection codes.

    Conclusion:

    Using a large, de-identified statewide dataset, we were able to identify a small number of extremely high-risk individuals. Selective screening of these individuals upon admission could prove to be a valuable way to identify CRE-colonized patients in order to take proper precautions.

    Michael J. Ray, MPH1, William E. Trick, MD1,2, Angela S. Tang, MPH3,4 and Michael Y. Lin, MD, MPH2, (1)Cook County Health and Hospitals System, Chicago, IL, (2)Rush University Medical Center, Chicago, IL, (3)Illinois Department of Public Health, Chicago, IL, (4)Hektoen Institute of Medicine, Chicago, IL

    Disclosures:

    M. J. Ray, None

    W. E. Trick, None

    A. S. Tang, None

    M. Y. Lin, None

    Findings in the abstracts are embargoed until 12:01 a.m. PDT, Wednesday Oct. 3rd with the exception of research findings presented at the IDWeek press conferences.