Using hospital claim data to monitor surgical site infections for inpatient procedures

BackgroundTo facilitate surveillance for surgical site infections, alternative screening models developed on the basis of postoperative antimicrobial dispenses and discharge diagnosis have demonstrated robust performance for identifying possible infections. Interest remains on using such models to extrapolate and monitor secular trend of infection rates.ObjectivesOur objective was to evaluate the feasibility of monitoring surgical site infection rates for coronary bypass graft and craniotomy procedures using information extracted from a hospital claim database.MethodsWe used retrospective cohort study that included adult patients (age ≥18 years) undergoing coronary bypass graft or craniotomy procedures between January 2003 and June 2006. The alternative screening model for case identification was developed using procedures performed in year 2005. The model used up to 5 criteria including postoperative antimicrobial dispenses, discharge diagnosis suggesting a postoperative complication, and/or postoperative follow-up inpatient visits within 60 days from the surgery. The sensitivity and positive predictive values for the screening models were calculated by comparing with the infection status determined in the traditional surveillance using the former National Nosocomial Infection Surveillance definitions. Surgical site infection rates during the study period were extrapolated and compared with those obtained from the traditional surveillance.ResultsThe alternative screening model had a sensitivity of 81.1% and 82.9% to detect surgical site infections following coronary bypass graft and craniotomy procedures, respectively. The surgical site infection rates per 100 procedures extrapolated from the model were 8.6 for coronary bypass graft procedure and 3.4 for craniotomy procedure. These rates were comparable with those determined by the traditional surveillance (8.8 for coronary bypass graft; 2.7 for craniotomy).ConclusionClaim data obtained from a hospital database have potential to facilitate surveillance for surgical site infections following inpatient coronary bypass graft and craniotomy procedures.