Construction and Validation of a Predictive Model for Long-Term Major Adverse Cardiovascular Events in Patients with Acute Myocardial Infarction.

Abstract

Purpose: Current scoring systems used to predict major adverse cardiovascular events (MACE) in patients with acute myocardial infarction (AMI) lack some key components and their predictive ability needs improvement. This study aimed to develop a more effective scoring system for predicting 3-year MACE in patients with AMI.

Patients and methods: Our statistical analyses included data for 461 patients with AMI. Eighty percent of patients (n=369) were randomly assigned to the training set and the remaining patients (n=92) to the validation set. Independent risk factors for MACE were identified in univariate and multifactorial logistic regression analyses. A nomogram was used to create the scoring system, the predictive ability of which was assessed using calibration curve, decision curve analysis, receiver-operating characteristic curve, and survival analysis.

Results: The nomogram model included the following seven variables: age, diabetes, prior myocardial infarction, Killip class, chronic kidney disease, lipoprotein(a), and percutaneous coronary intervention during hospitalization. The predicted and observed values for the nomogram model were in good agreement based on the calibration curves. Decision curve analysis showed that the clinical nomogram model had good predictive ability. The area under the curve (AUC) for the scoring system was 0.775 (95% confidence interval [CI] 0.728-0.823) in the training set and 0.789 (95% CI 0.693-0.886) in the validation set. Risk stratification based on the scoring system found that the risk of MACE was 4.51-fold higher (95% CI 3.24-6.28) in the high-risk group than in the low-risk group. Notably, this scoring system demonstrated better predictive ability than the GRACE risk score (AUC 0.776 vs 0.731; P=0.007).

Conclusion: The scoring system developed from the nomogram in this study showed favorable performance in prediction of MACE and risk stratification of patients with AMI.