Background: In an effort to reduce inappropriate testing of hospital-onset Clostridioides difficile infection (HO-CDI), we sequentially implemented 2 strategies: an electronic health record-based clinical decision support tool that alerted ordering physicians about potentially inappropriate testing without a hard stop (intervention period 1), replaced by mandatory infectious diseases attending physician approval for any HO-CDI test order (intervention period 2). We analyzed appropriate HO-CDI testing rates of both intervention periods. Methods: We performed a retrospective study of patients 18 years or older who had an HO-CDI test (performed after hospital day 3) during 3 different periods: baseline (no intervention, September 2014–February 2015), intervention 1 (clinical decision support tool only, April 2015–September 2015), and intervention 2 (ID approval only, December 2017–September 2018). From each of the 3 periods, we randomly selected 150 patients who received HO-CDI testing (450 patients total). We restricted the study to the general medicine, bone marrow transplant, medical intensive care, and neurosurgical intensive care units. We assessed each HO-CDI test for appropriateness (see Table 1 for criteria), and we compared rates of appropriateness using the χ2 test or Kruskall-Wallis test, where appropriate. Results: In our cohort of 450 patients, the median age was 61 years, and the median hospital length of stay was 20 days. The median hospital day that HO-CDI testing was performed differed among the 3 groups: 12 days at baseline, 10 days during intervention 1, and 8.5 days during intervention 2 (P < .001). Appropriateness of HO-CDI testing increased from the baseline with both interventions, but mandatory ID approval was associated with the highest rate of testing appropriateness (Fig. 1). Reasons for inappropriate ordering did not differ among the periods, with <3 documented stools being the most common criterion for inappropriateness. During intervention 2, among the 33 inappropriate tests, 8 (24%) occurred where no approval from an ID attending was recorded. HO-CDI test positivity rates during the 3 time periods were 12%, 11%, and 21%, respectively (P = .03). Conclusions: We found that both the clinical decision support tool and mandatory ID attending physician approval interventions improved appropriateness of HO-CDI testing. Mandatory ID attending physician approval leading to the highest appropriateness rate. Even with mandatory ID attending physician approval, some tests continued to be ordered inappropriately per retrospective chart review; we suspect that this is partly explained by underdocumentation of criteria such as stool frequency. In healthcare settings where appropriateness of HO-CDI testing is not optimal, mandatory ID attending physician approval may provide an option beyond clinical decision-support tools.

Funding: None

Disclosures: None