Abstract 3139: Validation of Mayo Clinic Risk Adjustment Model for In-Hospital Mortality following Percutaneous Coronary Interventions using the American College of Cardiology-National Cardiovascular Data Registry (ACC-NCDR)
Objectives The new Mayo Clinic risk score (MCRS) has 7 clinical and non-invasive variables (age, cardiogenic shock, myocardial infarction within 24 hours, congestive heart failure, peripheral vascular disease, serum creatinine, and ejection fraction) for prediction of in-hospital mortality following PCI. We sought to validate and refine this model in the ACC-NCDR data set.
Methods and Results Of 309,351 patients in the ACC-NCDR who underwent PCI 1/104 –3/ 30 – 06, 3790 (1.23%) died during their index PCI. Predicted probabilities of death were assigned to each patient, and model discrimination and calibration were assessed. Ninety seven percent of patients had a MCRS of less than 10, indicating low- to intermediate-risk. The area under the receiver operator curve was 0.884 indicating excellent overall discrimination. The model discriminated well amongst all pre-specified subgroups, including gender, diabetes mellitus, renal failure, low ejection fraction, different age groups, and multi-vessel disease. Overall and within each integer risk score, the predicted mortality rate and observed mortality rate with 95% CI were calculated and compared. The model generally under predicted mortality; overall observed in-hospital mortality was 1.23% (n=3,790) while predicted mortality was 1.10%. Therefore, the model was recalibrated in the ACC population using logistic regression to obtain a new quadratic prediction equation.
Conclusions Although the MCRS for in-hospital mortality following PCI discriminates well in the ACC-NCDR population, the recalibrated version may be useful for providing patients with individualized estimates of procedural risk as part of the informed consent process before PCI.