Complex Surgical Site Infections and the Devilish Details of Risk Adjustment: Important Implications for Public Reporting

OBJECTIVE. To validate the National Nosocomial Infection Surveillance (NNIS) risk index as a tool to account for differences in case mix when reporting rates of complex surgical site infection (SSI). DESIGN. Prospective cohort study. SETTING. Twenty-four community hospitals in the southeastern United States. METHODS. We identified surgical procedures performed between January 1, 2005, and June 30, 2007. The Goodman-Kruskal gamma or G statistic was used to determine the correlation between the NNIS risk index score and the rates of complex SSI (not including superficial incisional SSI). Procedure-specific analyses were performed for SSI after abdominal hysterectomy, cardiothoracic procedures, colon procedures, insertion of a hip prosthesis, insertion of a knee prosthesis, and vascular procedures. RESULTS. A total of 2,257 SSIs were identified during the study period (overall rate, 1.19 SSIs per 100 procedures), of which 1,093 (48.4%) were complex (0.58 complex SSIs per 100 procedures). There were 45 complex SSIs identified following 7,032 abdominal hysterectomies (rate, 0.64 SSIs per 100 procedures); 63 following 5,318 cardiothoracic procedures (1.18 SSIs per 100 procedures); 139 following 5,144 colon procedures (2.70 SSIs per 100 procedures); 63 following 6,639 hip prosthesis insertions (0.94 SSIs per 100 procedures); 73 following 9,658 knee prosthesis insertions (0.76 SSIs per 100 procedures); and 55 following 6,575 vascular procedures (0.84 SSIs per 100 procedures). All 6 procedure-specific rates of complex SSI were significantly correlated with increasing NNIS risk index score (P < .05) CONCLUSIONS. Some experts recommend reporting rates of complex SSI to overcome the widely acknowledged detection bias associated with superficial incisional infection. Furthermore, it is necessary to compensate for case-mix differences in patient populations, to ensure that intrahospital comparisons are meaningful. Our results indicate that the NNIS risk index is a reasonable method for the risk stratification of complex SSIs for several commonly performed procedures.