To examine the usefulness of temporal and spatial analysis in identifying nosocomial transmission of Clostridium difficile among pediatric patients hospitalized on four wards at The Children's Hospital of Central California from September 8,1998, to January 16,1999.
Stool specimens obtained from the clinical microbiology laboratory during the study period were tested by culture and latex agglutination for C. difficile. Polymerase chain reaction was used to identify toxin genes. Isolates obtained were mapped to a grid for each ward and were analyzed using the Knox test. Results were compared with DNA fingerprints generated by arbitrarily primed polymerase chain reaction.
Total occupancy of these 4 wards was 438 during the study period. Stool specimens were available for 256 (58%) of these patients, yielding 67 C. difficile isolates and generating 2,211 case pairs for analysis by the Knox test. After stratification by toxin status, 5 clustered pairs of toxigenic isolates were identified on 1 of the wards by this method. Fingerprint analysis identified 4 clusters with indistinguishable banding patterns on 2 of the 4 wards. Two of the identified clusters were toxigenic and 2 were nontoxigenic. None of these clusters corresponded to clusters identified by the Knox test.
The Knox test is an ineffective method for identifying cases resulting from nosocomial transmission of C. difficile in a pediatric setting due to the persistence of C. difficile spores and the unique environment of a pediatric hospital. Molecular analysis remains the most effective method.