Improving surveillance for pediatric Clostridium difficile infection: derivation and validation of an accurate...

The incidence of Clostridium difficile infection (CDI) is increasing. Multicenter studies of CDI have been limited by the lack of valid case-finding tools. To facilitate pediatric studies of CDI, we constructed a case-finding tool using administrative data.

A cross-sectional study was performed using the Pediatric Health Information System database and microbiologic data from 4 member hospitals. Using patients with laboratory-confirmed CDI as the standard, we determined the sensitivity, specificity, positive (PPV), and negative (NPV) predictive value of an ICD-9-CM code for identifying children with laboratory-confirmed CDI.

We identified 109 patients with laboratory-confirmed CDI and 119 patients with CDI ICD-9-CM code. The sensitivity, specificity, PPV, and NPV were 80.73%, 99.89%, 73.95%, and 99.92%, respectively, for this comparison. The addition of a billing charge for both C. difficile laboratory test and treatment medication to the ICD-9-CM code increased the specificity and PPV, but resulted in a slight decrease in the sensitivity and NPV. The use of administrative data for identifying pediatric cases of CDI was also compared with that of chart review, and was found to be a stronger surrogate for identifying cases of CDI when compared with microbiology data alone.

These results demonstrate that the use of administrative data for CDI is a reliable and accurate method for identifying pediatric patients with CDI. The use of administrative data could facilitate the completion of larger studies due to its greater accessibility and reduced costs.