External Validation of the AS5F Score and the Role of Left Atrial Dilatation in Post-Stroke/TIA Atrial Fibrillation Detection.

Background

Prolonged cardiac monitoring increases atrial fibrillation (AF) detection after ischemic stroke or transient ischemic attack (TIA). The AS5F score was developed to identify patients at higher risk of post-stroke AF, but its performance in real-world populations remains incompletely characterized. We aimed to externally validate the AS5F score and to evaluate whether left atrial dilatation (LAD) improves risk prediction.

Methods

We conducted a retrospective single-center study including 410 patients with ischemic stroke or TIA who underwent prolonged Holter monitoring between 2021 and 2025. Logistic regression and receiver operating characteristic (ROC) curve analyses were used to assess discrimination. Model calibration was evaluated using the Hosmer-Lemeshow test.

Results

AF was detected in 33 patients (8.0%). The AS5F score was significantly associated with AF detection (OR 1.07 per point; 95% CI 1.03-1.11; p < 0.001), showing modest discrimination (AUC 0.69). Age alone demonstrated similar performance (AUC 0.69). LAD was strongly associated with AF (OR 4.00; 95% CI 1.79-8.93; p = 0.001) but had lower discriminatory ability (AUC 0.61). In patients with available echocardiographic data (n = 369), a combined age + LAD model achieved an AUC of 0.73 with adequate calibration. The improvement compared with AS5F was not statistically significant.

Conclusions

In this external real-world cohort, AS5F demonstrated moderate discrimination for post-stroke AF detection. A simplified model combining age and left atrial dilatation showed numerically higher performance and may represent a pragmatic strategy for risk stratification in clinical practice.