Benchmarks and control charts for surgical site infections

Abstract

Abstract Background Although benchmarks and control charts are basic quality improvement tools, few surgeons use them to monitor surgical site infection (SSI). Obstacles to widespread acceptance include: (1) small denominators, (2) complexities of adjusting for patient risk and (3) scepticism about their true purpose (cost cutting, surgical privilege determination or improving outcomes). Methods The application of benchmark charts (using US national SSI rates as limits) and control charts (using facility rates as limits) was studied in 51 hospitals submitting data to the AICE National Database Initiative. SSI rates were risk adjusted by calculating a new statistic, the standardized infection ratio (SIR), based on the risk index suggested by the Centers for Disease Control National Nosocomial Infection Surveillance Study. Fourteen different types of control chart were examined and 115 suspiciously high or low monthly rates were flagged. Participating hospital epidemiologists investigated and classified each flag as ‘a real problem’ (potentially preventable) or ‘not a problem’ (beyond the control of personnel at this facility). Results None of the standard, widely recommended, control charts studied showed practical value for identifying either preventable rate increases or outbreaks (clusters due to a single organism). On the other hand, several types of risk-adjusted control chart based on the SIR correctly identified most true opportunities for improvement. Sensitivity, specificity and receiver–operator characteristic (ROC) analysis revealed that the XmR chart of monthly SIRs would be useful in hospitals with smaller surgical volumes (ROC area = 0·732, P = 0·001). For larger hospitals, the most sensitive and robust SIR chart for real-time monitoring of surgical infections was the mXmR chart (ROC area = 0·753, P = 0·0005).