Facility-Level Predictors of Publicly-Reported Hospital-Associated Clostridium Difficile Infection Rates in California

BACKGROUND:   California hospitals are mandated to report hospital-associated Clostridium difficile infection (HA-CDI) rates.  However, risk adjustment methods are limited.  We sought to identify readily-available facility-level adjustors for HA-CDI rates in California hospitals. 

METHODS: We conducted a retrospective cohort study of all California hospitals between April 2010 and March 2011.  HA-CDI rates were derived from 2010-2011 data reported to the National Healthcare Safety Network.  We used 2009-2010 patient discharge datasets from the California Office of Statewide Health Planning and Development and 2005-2010 U.S. Census American Community Survey data to generate hospital population characteristics.  Information on hospital type, teaching status, and laboratory testing method were also included.  We used Least Absolute Shrinkage and Selection Operator regression with 10-fold cross validation for initial variable selection.  Facility-level predictors were then assessed using multivariate Poisson regression testing.          

RESULTS:   We identified hospital population characteristics for 366 hospitals reporting HA-CDI rates.  One outlier facility was excluded.  The median number of admissions was 7,933. 6% were teaching hospitals, 5% long-term acute care hospitals, and 28% used PCR. Overall, the mean HA-CDI rate was 8.3 cases per 10,000 patient-days (range, 0 to 29.8).  In  final regression models, variables associated with HA-CDI included increased number of hospital beds (incidence rate ratio (IRR) = 1.08, 95% CI: 1.05-1.11), teaching status (IRR=1.25, 95% CI: 1.04-1.51), PCR testing method (IRR=1.12, 95% CI: 1.00-1.25), and increased annual admissions involving patients ≥ 85 years (IRR=1.36, 95% CI: 1.12-1.66), increasing comorbidity index (IRR=1.37, 95% CI: 1.18-1.59), increasing admissions with infections (proxy for antibiotic use, IRR=1.44, 95% CI: 1.00-1.16), and commercially-insured patients (IRR=1.16, 95% CI: 1.08-1.25). 

CONCLUSIONS:   HA-CDI rates in California hospitals are associated with multiple facility-level characteristics including hospital volume, C. difficile testing method, insurance type, and patient case mix.  These results suggest that risk-adjustment for non-modifiable factors such as patient population age, comorbidity index, reasons for admission, and insurance type maybe important for valid inter-hospital comparisons of HA-CDI rates.