New CathPCI Registry Mortality Risk Model With Higher-Risk Variables Accurately Predicts Risk

A new in-hospital mortality risk model that incorporates contemporary variables that reflect clinical severity may be accurate in predicting mortality risk following PCI, according to a study published May 3 in the Journal of the American College of Cardiology, and being presented at ACC.21.

Yulanka S. Castro-Dominguez, MD, et al., used data from ACC's CathPCI Registry to develop a model to predict in-hospital mortality risk that incorporated five new registry variables – frailty, cardiovascular instability type, level of consciousness after cardiac arrest, and decision for PCI with surgical consult – and accounted for case mix and hospital volume. The researchers also evaluated the model's performance in different risk cohorts. The models were tested using logistic regression.

The researchers developed three models – a full model including all candidate variables; a precatherization model that excluded angiographic data; and a simplified bedside risk score. Hospital-specific risk-standardized mortality rates were calculated based on the hierarchical model. All three models were assessed with the validation sample.

The final cohort included 706,263 total PCI cases from 1,608 sites. The sample was randomly split, with 495,005 (70%) in a development cohort and 211,258 (30%) in a validation cohort. Clinical, demographic and angiographic features were similar in both cohorts. Patients had an average age of 66 years and 30.8% were female, 85% were white, 40.8% had a history of diabetes and 41% had prior PCI. Elective PCIs represented 39.2% of all procedures. Overall, in-hospital mortality was 1.9% following PCI and increased with worsening clinical instability. Mortality rates were similar in the development and validation cohorts.

The researchers found that the full model performed well, with excellent discrimination (c-index: 0.943) in the validation cohort and good calibration across different clinical and procedural risk groups. Variables that were predictive of in-hospital mortality included procedural urgency, cardiovascular instability and level of consciousness after cardiac arrest. In addition, the full model performed well in cohorts undergoing PCI without cardiac arrest or shock (c-index: 0.883), all PCI without STEMI (c-index: 0.926), and patients with STEMI without cardiogenic shock or cardiac arrest (c-index: 0.859). The precatheterization and bedside risk adjustment models also performed well with excellent discrimination in the validation sample (c-indexes: 0.940 for precatheterization model; 0.925 for bedside risk model). Median hospital risk-standardized mortality was 1.9%, with a range from 1.1% to 3.3% (interquartile range: 1.7%-2.1%).

According to the researchers, the new in-hospital mortality model incorporates new variables that reflect clinical severity and accurately predict mortality risk following PCI. They conclude that the use of the model, "both in public reporting and in quality improvement efforts, will help standardize the assessment of risk associated with PCI both for hospitals and patients."

Clinical Topics: Arrhythmias and Clinical EP, Heart Failure and Cardiomyopathies, Invasive Cardiovascular Angiography and Intervention, Stable Ischemic Heart Disease, Vascular Medicine, Implantable Devices, SCD/Ventricular Arrhythmias, Acute Heart Failure, Interventions and Vascular Medicine, Chronic Angina

Keywords: ACC Annual Scientific Session, ACC21, National Cardiovascular Data Registries, CathPCI Registry, Shock, Cardiogenic, Hospital Mortality, Logistic Models, Risk Adjustment, ST Elevation Myocardial Infarction, Percutaneous Coronary Intervention, Quality Improvement, Calibration, Registries, Heart Arrest, Cohort Studies, Diabetes Mellitus, Diagnosis-Related Groups, Risk Factors

< Back to Listings