Antibiotics and Clostridium difficile diarrhea in the ambulatory care setting.

OBJECTIVE: The goal of this study was to determine the prevalence of Clostridium difficile diarrhea (CDD) and the risk for CDD associated with different oral antibiotics commonly used in the ambulatory care setting.

METHODS: The prevalence of CDD was determined for enrollees in 4 UnitedHealth Group-affiliated health plans between January 1, 1992, and December 31, 1994. Cases were identified based on the presence of an inpatient or outpatient claim with a primary diagnosis of diarrhea, a pharmacy claim for a prescription drug used to treat CDD, or a physician or facility claim for the C. difficile toxin test, and were confirmed using full-text medical records. Within a retrospective cohort design, periods of risk for CDD were defined on the basis of duration of antibiotic therapy. To control for potential selection bias created by heterogeneous rates of C. difficile testing and to limit confounding due to multiple antibiotic exposures, we used a nested case-control design, restricting eligibility to subjects who underwent screening for C. difficile and who had been exposed to only 1 antibiotic risk period with a single antibiotic.

RESULTS: The global prevalence of CDD in 358,389 ambulatory care enrollees was 12 per 100,000 person-years. In the nested case-control study, after controlling for other risk factors, 2 antibiotics demonstrated an increased association with CDD: cephalexin (odds ratio [OR] = 7.5, 95% CI = 1.8 to 34.7) and cefixime (OR = 6.4, 95% CI = 1.2 to 39.0).

CONCLUSIONS: Although CDD is thought to occur primarily in hospitalized patients, it was found to be present in an ambulatory care population, but at a low frequency. In this population, it appeared to be associated with 2 cephalosporins but not with other types of antibiotics usually linked with nosocomial CDD. Because the frequency of C. difficile testing was shown to be more common with high-risk antibiotics, CDD may be underdiagnosed in the ambulatory care setting.