1068. Real-time Computerized Flagging of Potential Clostridium difficile Infection (CDI) Agrees with National Health and Safety Network (NHSN) LabID Case Identification
Session: Poster Abstract Session: Surveillance of HAIs: Evaluating National Strategy
Friday, October 4, 2013
Room: The Moscone Center: Poster Hall C
  • IDSA poster.pdf (178.6 kB)
  • Background:

    CDI is associated with excess morbidity, mortality, and cost and is now used in public quality reporting.  The NHSN LabID surveillance CDI definition may allow more standardized and efficient CDI case identification but is still labor intensive.  We undertook a retrospective cohort study to test if a real-time computerized CDI LabID flag (RTCflag) agreed with the NHSN LabID case identification.


    An RTCflag was developed to mirror the NHSN LabID CDI case definition.  A 483-bed community teaching hospital was chosen as a validation cohort of the RTCflag.  Cases identified by the flag for one year after it went live as well as cases identified by a primary discharge diagnosis of CDI from the same time period were matched 1:1 to controls by age and year of admission.  Each medical record was independently abstracted by two researchers blinded to the case-control status of the patients.  The two cohorts and the operating characteristics of the flag were described.


    147 RTCflag or primary discharge diagnosis CDI cases were matched to 147 controls. 120 patients had LabID CDI on chart abstraction and 174 patients did not.  The mean age (65.8 vs. 65.5 yr, p = 0.9) and sex (41.7 vs. 38.5% male, p = 0.6) were similar between the LabID CDI positive and negative groups, respectively.  The LabID CDI positive group was not more significantly likely to have been in the ICU (p = 0.2) or to be readmitted within 30 days (p = 0.1), but was more likely to die in the hospital (9.2 vs. 3.2%, p = 0.04) and had a longer hospital length of stay (median 6 vs. 4 days, p < 0.01).  Compared with chart review LabID as the reference, the RTCflag had a sensitivity of 100%, specificity of 92%, positive predictive value of 86.7%, and a negative predictive value of 100%.  The interobserver Kappa was excellent (0.88, 95% CI 0.77-1.00, p < 0.01).


    An RTCflag is able to accurately identify NHSN LabID CDI events.  Computerized flags have the potential to optimize Infection Prevention and disease surveillance efforts across health systems, geographic areas or networks.

    Kristina Bajema, MD1, Adam Alter, MD1, Chris Dale, MD, MPH2 and James Leggett, MD1, (1)Providence Portland Medical Center, Portland, OR, (2)Providence Health & Services, Seattle, WA


    K. Bajema, None

    A. Alter, None

    C. Dale, None

    J. Leggett, None

    Findings in the abstracts are embargoed until 12:01 a.m. PST, Oct. 2nd with the exception of research findings presented at the IDWeek press conferences.