Performance monitoring in interventional cardiology: application of statistical process control to a single-site database

Abstract

Graphical Statistical Process Control (SPC) tools have been shown to promptly identify significant variations in clinical outcomes in a range of health care settings, but as yet have not been widely applied to performance monitoring in percutaneous coronary intervention (PCI). We explored the application of these techniques to a prospective PCI registry at a single site. Baseline clinical and procedural data along with one and twelve month major adverse cardiac event (MACE) details were prospectively collected in relation to 2,697 consecutive PCI procedures (2,417 patients) performed between the 1st January 2003 and the 31st December 2007. We investigated outcome measures which were both clinically relevant and occurred at a sufficient frequency (>1%) to allow valid application of SPC techniques, and found procedural and lesion failure, major postprocedural complications, and one and 12 month MACE to be suitable endpoints. Cumulative Sum (CUSUM) charts, Variable Life-Adjusted Display (VLAD) charts and Funnel Plots were employed in combination to evaluate both group and individual performance on a near "real time" basis. We found that the use of these charts provided complimentary prospective audit of clinical performance to identify variations in group and individual operator performance and to clarify these as either systemic or individual operator-related. We propose a system of integrating SPC tools as a component of the audit function of a PCI unit. SPC tools have the potential to provide near "real-time" performance monitoring and may allow early detection and intervention in altered performance for both the group and the individual operator. A clinically-integrated system of SPC tools may thus complement and enhance effectiveness of the traditional case-based morbidity and mortality audit.