Physician Prescribing Behavior in Suspected Clostridium difficile Infection

Abstract

Background Empirical antibiotic therapy for Clostridium difficile infection (CDI) is associated with adverse effects. We sought to describe physician prescribing behavior with regard to empirical CDI therapy and hypothesized that delays in stool testing results may lead to increased use of empirical therapy. Methods A sample of 100 patients was selected from the population of patients with stool specimens submitted for C difficile enzyme immunoassay testing over a 6-month period. The time between order placement and result posting was compared between patients who received empirical CDI therapy and patients who did not receive empirical therapy. A chart review was conducted to assess for other factors driving physician prescribing behavior. Results The mean time between order submission and result posting was 16.7 hours in the group receiving empirical therapy and 17.6 hours in the group receiving test-guided therapy, with no statistically significant difference between the 2 groups. In univariate analysis, significant factors associated with empirical therapy included leukocytosis (16.5 vs. 10.7), bandemia (7.1% vs. 1.9%), higher Hines Veterans Administration score (1.8 vs. 1.1), diarrhea present on admission (relative risk, 2.3), and ordering of abdominal imaging (relative risk, 2.9). On multivariate regression, leukocytosis and ordering of abdominal imaging remained significant. Conclusions Use of empirical therapy for CDI was not associated with time to stool test results. The decision to use empirical therapy seems to be most related to the presence of leukocytosis and the physician's decision to order abdominal imaging. The pattern of findings suggests a potential component of diagnostic uncertainty driving prescribing behavior.