Clostridium difficile is the most important cause of antibiotic-associated colitis, but its epidemiology remains unknown. Using a selective medium for the isolation of C. difficile, cultures were obtained from the environment and contacts of hospitalized patients carrying C. difficile in their stools. In areas where carriers had diarrhea, 85 (9.3%) of 910 cultures of floors and other surfaces, especially those subject to fecal contamination, were positive. In areas where there were no known carriers, only 13 (2.6%) of 497 cultures of similar sites were positive (P less than 0.005). C difficile was isolated from hands and stools of asymptomatic hospital personnel, from sewage and soil, and from the home of a patient. Environmental isolates were toxigenic. C. difficile inoculated onto a floor persisted there for five months. Further studies are needed to document how often floor persisted there for five months. Further studies are needed to document how often C. difficile shed by patients with antibiotic-associated colitis is acquired by other persons and whether isolation precautions are capable of limiting the organism's spread.

Ten cases of antibiotic-associated colitis (AAC) were identified at a hospital in Washington, D.C., from March 17 to May 9, 1979. No geographic clustering of cases was found, nor was an association with increased use of antibiotics demonstrated. Exposure to aminoglycosides, cephalosporins, and clindamycin was associated with AAC, as was a history of enemas in the seven days before the onset of illness (P=0.045). This association was strengthened when gastrointestinal procedures-defined as (1) three or more enemas per week, (2) the insertion of a nasogastric tube for two or more days, or (3) gastrointestinal surgery-were performed within seven days of the onset of illness (P=0.007). Clostridium difficile was not isolated from the hospital environments, nursing personnel, or family members of the patients. C. difficile was isolated from stool specimens of five (36%) of 14 patients who served as controls.

No abstract