Percutaneous coronary intervention outcomes in a low-volume center: survival, stent thrombosis, and repeat...

American College of Cardiology (ACC) guidelines state that percutaneous coronary interventions (PCI) be performed at centers and by operators with high-volume (>400 yearly/center) whose historical and current risk-adjusted outcomes statistics are comparable to those reported in large registries. Tripler Army Medical Center is a low-volume treatment facility but has a geographic need and special mission requirement for providing this service.

We computed 30-day incidence of stent thrombosis, need for repeat revascularization, and all-cause mortality for all PCIs performed at Tripler from January 2002 through June 2008. The New York State Registry regression model was selected among 3 risk-adjustment models that we assessed in our patients. This model was used to compute expected mortality rate based on patient risk factors. The 30-day incidence of stent thrombosis and repeat revascularization was also determined, and the long-term incidence of these events was estimated with the Kaplan-Meier method as was survival. For all 546 PCI procedures, 30-day mortality was 1.47%, the incidence of stent thrombosis 2.1%, the incidence of any repeat revascularization 5.1%, and the combined event rate 5.9%. Based on risk factors used in the New York State Registry, our expected mortality was 1.93% and not significantly different from the observed rate. Although survival at 1 and 3 years appeared comparable with benchmarks at 94.6% and 89.3%, as did repeat revascularization rates at 13.0% and 21.4%, the incidence of stent thrombosis was regarded as high whether the definition included possible cases (3.2% and 3.9%) or only those regarded as definite or probable (2.7% and 3.1%). We did not identify any remediable risk factors for stent thrombosis, nor were we able to identify significant differences by year or by operator. However, visual inspection of a plot of deciles of New York State risk of death demonstrated 2 outlier cases among the 8 who died, who could have been considered candidates for thorough peer review.

We recommend other low-volume interventional programs enter all patients undergoing PCI into a database, their own local registry even if not a national one such as the American College of Cardiology National Cardiovascular Data Registry, obtain information about survival and cardiac events during follow-up, compute and risk-adjust in-hospital or 30-day mortality, and use the objective assessment of risk in individual patients to identify outliers when outcome is adverse, and possibly as a means of triaging patients to appropriate therapy before choosing PCI.