Abstract Body: Introduction
Patients with atherosclerotic carotid artery disease are at high risk of mortality in the long-term follow-up after carotid endarterectomy (CEA), partly due to dysrhythmia. Atrial fibrillation (AF) is a common cardiac dysrhythmia linked to stroke and cardiovascular events. An artificial intelligence-based electrocardiogram (ECG) AF algorithm (AI-AF) can effectively identify silent AF. We aimed to assess the effectiveness of AI-AF in predicting mortality in patients with carotid artery disease undergoing CEA.
Methods
Patients with carotid artery disease undergoing CEA at Mayo Clinic from 2002-2023 were included if they had >30 days follow-up and an ECG within one year before CEA. The ECG closest to the CEA date was used to calculate AI-AF (probability of AF ranging from 0 to 1). All-cause mortality rates after CEA were obtained from clinical records. The optimal AI-AF cut-off value for predicting mortality was calculated using a receiver operating characteristic curve and the Youden index. Long-term survival after CEA was depicted with a Kaplan Meier plot and compared with the Log-rank test. Univariate and multivariate Cox Regression models were used to assess the predictive value of AI-AF for mortality, adjusting for age, sex, body mass index, diabetes mellitus, hypertension, and hyperlipidemia.
Results
A total of 636 patients with a median age of 72 [IQR: 66, 78] years and 414 (65.1%) males were included. The median AI-AF was 0.056 [IQR: <0, 0.171]. Over a median follow-up time of 7.1 [IQR: 2.7, 11.5] years, mortality occurred in 326 (51.3%) patients. AI-AF significantly predicted mortality in univariate and multivariate Cox regression analysis (HR: 3.47, 95%CI: 2.34-5.16, P<0.001; HR: 2.30, 95%CI: 1.44-3.68, P<0.001, respectively). The best cut-off value for AI-AF for mortality prediction was 0.025 (area under the curve: 0.608). High and low AI-AF were defined as AI-AF≥ 0.025 and <0.025, respectively. Mortality was significantly higher in high AI-AF compared to low AI-AF patients (P<0.001, Figure 1A,1B), and high AI-AF was a significant predictor of mortality (univariate: Figure 2; multivariate Cox regression: HR: 2.11, 95%CI: 1.56-2.84, P<0.001).
Conclusion
AI-AF can identify patients with a higher risk of mortality after CEA. This finding supports a role for AI algorithms in risk stratification of patients with carotid artery disease following CEA. Future studies are needed to elucidate the contribution of AF to all-cause mortality.