Impact of Patient Sharing on Interpreting Clostridium Difficile Infection Rates

BACKGROUND: In 2010, all acute care hospitals in New York State (NYS) reported Clostridium difficile infections (CDI) following the National Healthcare Safety Network (NHSN) Laboratory-Identified Event (LabID) Protocol.  CDI is classified as community- or hospital-associated within each hospital based only on laboratory testing performed within that hospital.  However, patients may be exposed to CDI through stays at multiple hospitals and have laboratory testing performed at multiple hospitals. 

METHODS:  The NYS NHSN database was matched to the NYS hospital inpatient billing database using personal identifiers to obtain more detailed information on past hospital stays.  CDI rates calculated using only hospital-specific data were compared to CDI rates calculated using the enhanced database.  Hospital-onset (HO) rates are based on incident cases in which a positive stool sample was obtained on day four or later during the hospital stay.  Hospital-associated (HA) rates include both HO cases and cases in which a positive stool sample was obtained during the first three days of hospital admission from a patient who had been discharged from a hospital within the previous four weeks.  

RESULTS: Due to the reclassification of cases from incident to recurrent based on test results reported by other hospitals, the overall HO rate decreased 1.4%, from 7.90 to 7.80 cases per 10,000 patient days (range 0% to 33% decrease between hospitals) using the enhanced database. From the state perspective, the overall HA rate increased 5.9%, from 10.31 to 10.91 cases per 10,000 patient days.  Thirteen percent of CDI patients were inpatients at another hospital in the previous 4 weeks. 

CONCLUSIONS:  NHSN CDI rates must be interpreted with caution.  Varying amounts of patient sharing between different hospitals impact the ability to compare CDI rates between hospitals.