New York Risk Model and Simplified Risk Score for In-Hospital/30-Day Mortality for Percutaneous Coronary Intervention

Abstract

Risk models and risk scores derived from those models require periodic updating to account for changes in procedural performance, patient mix, and new risk factors added to existing systems. No risk model or risk score exists for predicting in-hospital/30-day mortality for percutaneous coronary interventions (PCIs) using contemporary data. This study develops an updated risk model and simplified risk score for in-hospital/30-day mortality following PCI. To accomplish this, New York's Percutaneous Coronary Intervention Reporting System was used to develop a logistic regression model and a simplified risk score model for predicting in-hospital/30-day mortality and to validate both models based on New York data from the previous year. A total of 54,770 PCI patients from 2019 were used to develop the models. Twelve different risk factors and 27 risk factor categories were used in the models. Both models displayed excellent discrimination for the development and validation samples (range from 0.894 to 0.896) and acceptable calibration, but the full logistic model had superior calibration, particularly among higher-risk patients. In conclusion, both the PCI risk model and its simplified risk score model provide excellent discrimination and although the full risk model requires the use of a hand-held device for estimating individual patient risk, it provides somewhat better calibration, especially among higher-risk patients.

Keywords: PCI risk score; percutaneous coronary intervention; short-term mortality.