Development and external validation of a prognostic model for new-onset atrial fibrillation complicating acute myocardial infarction: insights from the NOAFCAMI-China registry

Abstract Aims There is no specifically developed model to predict the risk of major adverse cardiac events (MACEs) in patients with new-onset atrial fibrillation (NOAF) complicating acute myocardial infarction (AMI). We aimed to develop and validate a prediction model for 5-year risk of MACE in patients with post-MI NOAF. Methods and results The derivation cohort comprised 457 patients, and the external validation cohort consisted of 206 patients between January 2014 and January 2022. Stepwise multivariable Cox regression analysis was used to identify candidate predictors and to establish the model for 5-year MACE prediction. Model performance was assessed using time-dependent area under the receiver-operating characteristic curve (AUC), C-index, and calibration curves. According to the stepwise multivariable Cox regression analysis, 7 variables were included in the prediction model (NOAFCAMI score): age, prior HF, Killip class, undergoing percutaneous coronary intervention, peak level of NT-pro BNP, AF burden, and symptomatic AF. The 5-year AUC was 0.83 [95% confidence interval (CI): 0.77 to 0.88]. Internal validation by optimism bootstrap-corrected C-index was 0.72 (95% CI: 0.68 to 0.76). External validation showed a 5-year AUC of 0.79 (95% CI: 0.69 to 0.89). The calibration of the NOAFCAMI score for 5-year MACE prediction was acceptable in the derivation [Brier score: 0.17 (95% CI: 0.15 to 0.19)] and the external validation [Brier score: 0.19 (95% CI: 0.16 to 0.22)] cohorts, respectively. Conclusion The NOAFCAMI score is the first externally validated prediction model to personalize MACE risk assessment in patients with post-MI NOAF, offering actionable insights for tailored management.