Related Experiment Video
Updated: Jun 8, 2026
An R-Based Landscape Validation of a Competing Risk Model
Published on: September 16, 2022
Development and validation of risk adjustment models for long-term mortality and myocardial infarction following
Mandeep Singh1, David R Holmes, Ryan J Lennon
1Division of Cardiovascular Diseases, Mayo Clinic, Rochester, MN 55905, USA. singh.mandeep@mayo.edu
Insights
New risk models predict long-term mortality and myocardial infarction (MI) after percutaneous coronary intervention (PCI) using simple patient data. This tool aids in providing personalized, evidence-based risk estimates for patients post-PCI.
Area of Science:
- Cardiology
- Clinical Risk Prediction
- Interventional Cardiology
Background:
- Current percutaneous coronary intervention (PCI) outcome models inadequately assess long-term prognosis.
- There is a need for reliable tools to predict mortality and major adverse cardiovascular events after PCI.
Purpose of the Study:
- To develop 1- and 5-year risk models for mortality and mortality/myocardial infarction (MI) after PCI.
- To utilize simple, readily available clinical and laboratory variables for risk prediction.
Main Methods:
- Analysis of a large registry of 9165 patients undergoing PCI.
- Application of Cox proportional hazards regression to identify predictors of long-term outcomes.
- Development of separate risk models for mortality and mortality/MI.
Main Results:
- Identified key predictors including older age, comorbidities, low ejection fraction, acute MI, smoking, heart failure, hyperlipidemia, 3-vessel disease, procedural failure, ventricular arrhythmia, and low medication score.
- Developed simple integer scores that effectively stratified patients into distinct risk categories.
- Models demonstrated good discrimination with areas under the ROC curve of 0.786 for mortality and 0.728 for mortality/MI.
Conclusions:
- A convenient risk scoring system using easily obtainable variables can accurately predict long-term mortality and MI after PCI.
- This model provides a valuable tool for individualized, evidence-based risk assessment for patients post-PCI.
Background:
Existing models for outcome after percutaneous coronary interventions (PCIs) lack assessment of long-term prognosis. Our goal was to derive 1- and 5-year mortality and mortality/myocardial infarction (MI) risk models for PCI outcomes from simple, easily obtainable clinical and laboratory variables.
Methods And Results:
Using the Mayo Clinic registry, we analyzed long-term mortality and mortality/MI following PCIs on 9165 unique patients from January 1, 2001, through December 31, 2007. Cox proportional hazards regression was used to model the calculated risk score and major procedural complications. A total of 1243 patients died, and 696 had MI. Separate risk models derived from clinical, procedural, and laboratory characteristics were made for mortality and mortality/MI. Older age, comorbid conditions, low ejection fraction, acute MI, history of smoking, heart failure, hyperlipidemia, 3-vessel disease, procedural failure, ventricular arrhythmia during PCI, and low medication score were predictors of long-term mortality and mortality/MI. Simple integer scores stratified patients into low, moderate, high, and very high risk for subsequent events. Models had adequate goodness of fit, and areas under the receiver operating characteristic curve were 0.786 and 0.728 for mortality and mortality/MI, respectively, indicating good overall discrimination. Bootstrap analysis indicated that the model was not overfit to the available data set.
Conclusions:
Easily obtainable variables can be combined into a convenient risk scoring system at the time of patient dismissal following PCI to accurately predict long-term mortality and mortality/MI. This model may be useful for providing patients with individualized, evidence-based estimates of long-term risk.
