BACKGROUND: The New York State Department of Health collects and reports outcome data on the hospitals and cardiologists who perform percutaneous coronary intervention (PCI) to allow them to examine their quality of care. Results are provided in tabular form. However funnel plots are the display method of choice for comparison of institutions and operators, using the principles of statistical process control (SPC). We aimed to demonstrate that funnel plots, which aid a meaningful interpretation of the results, can be derived from the New York PCI dataset. METHODS: The risk-adjusted mortality rates for 48 hospitals and cardiologists performing PCI were used for this analysis. Funnel plots (with control limits at 3 and 2 sigma) of all hospitals and operators performing PCI procedures were generated. Separate plots for emergency and nonemergency PCI procedures were derived. RESULTS: 149,888 patients underwent PCI procedures between January 1, 2002 and December 31, 2004. The 3-year risk-adjusted mortality rates for all PCI patients ranged from 0.00 to 1.37%. The funnel plots show risk-adjusted mortality rates against the denominator for that percentage (number of cases), displayed as a scatter plot and compared with the binomial funnel plot calculated around the mean for all cases reported. The risk-adjusted mortality rates of all hospitals were within 3 sigma (99.8%) upper control limits. The risk-adjusted mortality rates for three hospitals were above or on the upper warning limit (2 sigma control limit, equating to the 95% confidence interval) and three hospitals had risk-adjusted mortality rates below the 2 and 3 sigma control limits. CONCLUSION: The SPC funnel plot is an easy-to-interpret, risk-adjusted means of identifying units whose performance, in terms of mortality, diverges significantly from the population mean. Funnel plots may be applied to a complex dataset and allow a visual comparison of data derived from multiple healthcare units. Variation is readily identified permitting hospitals and cardiologists to appraise their practices so that effective quality improvement may take place. Copyright 2009 Wiley-Liss, Inc.
BACKGROUND: The New York State Department of Health collects and reports outcome data on the hospitals and cardiologists who perform percutaneous coronary intervention (PCI) to allow them to examine their quality of care. Results are provided in tabular form. However funnel plots are the display method of choice for comparison of institutions and operators, using the principles of statistical process control (SPC). We aimed to demonstrate that funnel plots, which aid a meaningful interpretation of the results, can be derived from the New York PCI dataset. METHODS: The risk-adjusted mortality rates for 48 hospitals and cardiologists performing PCI were used for this analysis. Funnel plots (with control limits at 3 and 2 sigma) of all hospitals and operators performing PCI procedures were generated. Separate plots for emergency and nonemergency PCI procedures were derived. RESULTS: 149,888 patients underwent PCI procedures between January 1, 2002 and December 31, 2004. The 3-year risk-adjusted mortality rates for all PCI patients ranged from 0.00 to 1.37%. The funnel plots show risk-adjusted mortality rates against the denominator for that percentage (number of cases), displayed as a scatter plot and compared with the binomial funnel plot calculated around the mean for all cases reported. The risk-adjusted mortality rates of all hospitals were within 3 sigma (99.8%) upper control limits. The risk-adjusted mortality rates for three hospitals were above or on the upper warning limit (2 sigma control limit, equating to the 95% confidence interval) and three hospitals had risk-adjusted mortality rates below the 2 and 3 sigma control limits. CONCLUSION: The SPC funnel plot is an easy-to-interpret, risk-adjusted means of identifying units whose performance, in terms of mortality, diverges significantly from the population mean. Funnel plots may be applied to a complex dataset and allow a visual comparison of data derived from multiple healthcare units. Variation is readily identified permitting hospitals and cardiologists to appraise their practices so that effective quality improvement may take place. Copyright 2009 Wiley-Liss, Inc.