Detection of Paroxysmal Atrial Fibrillation in Acute Stroke Patients

Abstract

Background: Atrial fibrillation (AF) is a frequent cause of stroke, but detecting paroxysmal AF (pAF) poses a challenge. We investigated whether continuous bedside ECG monitoring in a stroke unit detects pAF more sensitively than 24-hour Holter ECG, and tested whether examining RR interval dynamics on short-term ECG recordings using an automated screening algorithm (ASA) for pAF detection is a useful tool to predict the risk of pAF outside periods of manifest AF. Methods: Patients >60 years with acute ischemic stroke or transient ischemic attacks (TIA) were prospectively enrolled unless initial ECG revealed AF or they had a history of paroxysmal or persistent AF. ASA was performed on 1- to 2-hour ECG recordings in the emergency room and patients were classified into 5 risk categories for pAF. All patients underwent continuous bedside ECG monitoring for >48 h. Additionally, 24-hour Holter ECG was performed. Results: 136 patients were enrolled (median age: 72 years, male: 58.8%). In 29 (21.3%), pAF was newly diagnosed by continuous bedside ECG monitoring. pAF increased with age (p = 0.031). Median time to first pAF detection on continuous bedside ECG monitoring was 36 h. In 16 patients, pAF was detected by continuous bedside ECG monitoring prior to the performance of 24-hour Holter ECG. Thirteen of the remaining patients were pAF positive on continuous bedside ECG monitoring, but 24-hour Holter detected only 3 patients. Accordingly, the sensitivity of 24-hour Holter was 0.23. Sensitivity of higher-risk categories of ASA compared to continuous bedside ECG monitoring was 0.72, and specificity 0.63. Conclusion: Continuous bedside ECG monitoring is more sensitive than 24-hour Holter ECG for pAF detection in acute stroke/TIA patients. Screening patients for pAF outside AF episodes using ASA requires further development.