1312. Use of Electronic Data to Identify Risk Factors Associated with Clostridium difficile Infection (CDI) and to Develop CDI Risk Scores
Session: Poster Abstract Session: HAI: C. difficile Risk Assessment and Prevention
Friday, October 6, 2017
Room: Poster Hall CD
Posters
  • 1312_IDWPoster_C-diff Epi_19Sept2017_FINAL.pdf (226.5 kB)
  • Background:  Clostridium difficile is a major cause of severe diarrhea in the U.S. We described characteristics of Kaiser Permanente Northern California (KPNC) members with C. difficile infection (CDI), identified risk factors associated with CDI, and developed risk scores to predict who may develop CDI.

    Methods:  Retrospective cohort study with all KPNC members ≥18 years old from May 2011 to July 2014 comparing demographic and clinical characteristics for those with and without lab-confirmed incident CDI. We included CDI risk factors in logistic regression models to estimate the risk of developing future CDI after an Identification Recruitment Date (IRD), a time when an individual might be a good candidate for a C. difficile vaccine clinical trial. Two risk score models were created and cross validated (70% of the data used for development and 30% for testing).

    Results: During the study period, there were 9,986 CDI cases and 2,230,354 members without CDI. CDI cases tended to be ≥65 years old (59% vs. 21%), female (61% vs. 53%), and white race (70% vs. 53%), with more hospitalizations (42% vs. 3%), emergency room visits (51% vs. 14%), and skilled nursing facility stays (25% vs. 0.6%) in the year prior to CDI compared with members without CDI. At least 10 office visits within the prior year (53% vs. 16%), use of antibiotics in last 12 weeks (81% vs. 11%), proton pump inhibitors in the last year (36% vs. 7%), and multiple medical conditions within the prior year (e.g., chronic kidney disease, congestive heart failure, and pneumonia) were important risk factors for CDI. Using a hospital discharge event as the IRD, our risk score model yielded excellent performance in predicting the likelihood of developing CDI in the subsequent 31 – 365 days (C-statistic of 0.851). Using a random date as the IRD, our model also predicted CDI risk in the subsequent 1 - 30 days (C-statistic 0.658) and 31 - 365 days (C–statistic 0.722) reasonably well.

    Conclusion:  CDI can be predicted by increasing age, medications, comorbidities and healthcare exposure, particularly ≥10 office visits, hospitalizations, and skilled nursing stays in the prior year and recent antibiotics. Such risk factors can be used to identify high risk populations for C. difficile vaccine clinical studies.

    Laurie Aukes, RN1, Bruce Fireman, MA1, Edwin Lewis, MPH1, Julius Timbol, MS1, John Hansen, MPH1, Holly Yu, MSPH2, Bing Cai, PhD3, Elisa Gonzalez, MS2, Jody Lawrence, MD3 and Nicola P. Klein, MD, PhD1, (1)Kaiser Permanente Vaccine Study Center, Oakland, CA, (2)Pfizer, Inc., Collegeville, PA, (3)VRD Clinical Research & Development, Pfizer, Inc., Collegeville, PA

    Disclosures:

    L. Aukes, None

    B. Fireman, None

    E. Lewis, None

    J. Timbol, None

    J. Hansen, None

    H. Yu, Pfizer, Inc.: Employee , Salary

    B. Cai, Pfizer, Inc.: Employee , Salary

    E. Gonzalez, Pfizer, Inc.: Employee , Salary

    J. Lawrence, Pfizer, Inc.: Employee , Salary

    N. P. Klein, GSK: Investigator , Grant recipient
    Sanofi Pasteur: Investigator , Grant recipient
    Merck & Co: Investigator , Grant recipient
    MedImmune: Investigator , Grant recipient
    Protein Sciences: Investigator , Grant recipient
    Pfizer: Investigator , Grant recipient

    Findings in the abstracts are embargoed until 12:01 a.m. PDT, Wednesday Oct. 4th with the exception of research findings presented at the IDWeek press conferences.