Background: Antibiotics predispose to C. difficile infection (CDI) by at least 2 mechanisms. In addition to its disruptive effects on the host fecal flora, we previously demonstrated that clindamycin use increased the risk of CDI in the setting of hospital outbreaks due to an epidemic strain (REA group J) that was highly resistant to clindamycin. Since 2001, a new epidemic strain (REA group BI) has caused CDI outbreaks in hospitals throughout North America and Europe. This strain was recovered from patients prior to 2000, but was not associated with outbreaks and was distinguished from the recent epidemic BI strains by in vitro susceptibility to the newer fluoroquinolones, gatifloxacin and moxifloxacin.
Objective: To assess specific antibiotic risks for patients with CDI due to the epidemic BI stain.
Methods: Between February 2005 and April 2006, stool specimens from patients with CDI at our hospital were cultured for CD and subjected to restriction endonuclease analysis (REA) typing. A case-control study was conducted whereby antibiotic exposures up to 6 weeks prior to the first episode of CDI were recorded by review of our hospital's electronic databases. All CDI cases during this time period in which a stool culture was positive were included. Cases and controls were defined as first episode CDI in which BI and non-BI strains were recovered, respectively.
Results: There were 97 first episode CDI cases during the study period of which 67 (69%) were due to the epidemic BI strain. Antibiotic exposure prior to development of CDI by strain type is summarized in the table. Fluoroquinolone and macrolide use was more frequent in patients who developed CDI from the epidemic strain (TABLE, column A). Use of both classes of antibiotics was associated with the BI strain as a cause of CDI (TABLE, column B). Gatifloxacin was the most frequently used fluoroquinolone in both groups (45% in those with the BI strain, 37% with non-BI strain, p=0.51). In contrast, clindamycin exposure was significantly more frequent with CDI from non-BI strains and was not associated with developing CDI due to BI. Use of other classes of antibiotics was similar between groups. The frequency of patients receiving multiple classes of antibiotics prior to onset of CDI in both groups was similar (BI 50/67 [75%] vs. non-BI 19/30 [63%], p=0.33).
Conclusions: Fluoroquinolone and macrolide use was significantly associated with the epidemic BI/NAP1 strain in patients who developed CDI and clindamycin use was associated with CDI due to non-BI strains. These observations are consistent with the previously shown fluoroquinolone resistance of BI strains and support the link between specific antimicrobial exposure and resistance of the infecting C. difficile strain to that agent. These data also provide potential targets for antimicrobial stewardship interventions.