Objectives: To construct a calculator to assess the risk of 30-day mortality following PCI.
Background: Predictors of 30-day mortality are commonly used to aid management decisions for cardiac surgical patients. There is a need for an equivalent risk-score for 30-day mortality for percutaneous coronary intervention (PCI) as many patients are suitable for both procedures.
Methods: The British Columbia Cardiac Registry (BCCR) is a population-based registry that collects information on all PCI procedures performed in British Columbia (BC). We used data from the BCCR to identify risk factors for mortality in PCI patients and construct a calculator that predicts 30-day mortality.
Results: Patients (total n = 32,899) were divided into a training set (n = 26,350, PCI between 2000 and 2004) and validation set (n = 6,549, PCI in 2005). Univariate predictors of mortality were identified. Multivariable logistic regression analysis was performed on the training set to develop a statistical model for prediction of 30-day mortality. This model was tested in the validation set. Variables that were objective and available before PCI were included in the final risk score calculator. The 30-day mortality for the overall population was 1.5% (n = 500). Area under the ROC curve was 90.2% for the training set and 91.1% for the validation set indicating that the model also performed well in this group.
Conclusions: We describe a large, contemporary cohort of patients undergoing PCI with complete follow-up for 30-day mortality. A robust, validated model of 30-day mortality after PCI was used to construct a risk calculator, the BC-PCI risk score, which can be accessed at www.bcpci.org.
2009 Wiley-Liss, Inc.