Background: CDI guidelines recommend hospitals relying on Nucleic Acid Amplification Testing (NAAT) apply criteria for appropriate submission. CDS tools could improve test utilization.
Methods: A CDS Epic Best Practice Advisory was added to the electronic health record of a 624 bed academic center on 9/21/17. Hard stops were not available but the CDS prompted ordering NAAT with: notification of cancellation if a prior negative/positive test was done within 7/14 days; recommendation to cancel for laxatives within 48 hours; and to reconsider if tube feeds were initiated within 24 hours. Presence of diarrhea was not incorporated due to lack of standardized discrete documentation. Testing appropriateness was assessed pre and post CDS, as well as how alerts were handled Phase 1 (9/21 to 11/15/17), and after CDS modifications in Phase 2 (11/16/17 to 1/18/18) and Phase 3 (1/19 to 3/13/18). Charts of patients with healthcare onset CDI (HO CDI) LabID Events were reviewed for symptoms of true CDI defined as diarrhea (≥ 3 loose stools over 24 hours) without laxatives. LabID HO CDI standardized infection ratios were tracked.
Results: Ongoing review of response to CDS alerts led to changes in the algorithm (Table 1). Inaccurate interpretation of indeterminate tests were corrected and a notification the lab would reject repeat tests and formed stool despite overriding a cancellation was added. Evaluation of declinations for unhelpful triggers led to modification of the laxative list (e.g.; removed bulk forming agents) which decreased laxative declinations from 75-79% to 54%. Changes to the CDS did not drop the rate of alerts (3.8 to 3.6 on average per day) and providers continue to test for inappropriate indications. Review of HO CDI cases (Table 2) show patients without diarrhea continue to be tested (21% pre vs. 32% post CDS), but more of those with diarrhea have not been on laxatives (38% pre, 60% post). Pre to post CDS, the HO CDI SIR has started to drop (Figure 1).
Conclusion: CDS with provider prompting improved ordering practices for CDI, but iterative changes to the tool were needed. Additional steps, such as enforcing hard stops should be explored. Greater nurse involvement, as with standardized discrete documentation to capture diarrhea, would enhance testing algorithms.
T. T. Timbrook, None
C. Ciarkowski, None
E. Spivak, None