754 Non-conducted premature atrial complexes: a new predictor for atrial fibrillation in cryptogenic stroke patients

Atrial fibrillation (AF) is the main cardiac cause of stroke, but it frequently remains undetected. In these patient monitoring for AF is recommended using a Holter electrocardiogram (ECG). The aim of the present study is to study non-conducted atrial complexes (ncPAC) recorded on Holter ECG as a new predictor of AF.

Patients admitted to the Stroke Unit of our hospital for cryptogenic stroke from December 2018 to January 2020 who underwent 24-h electrocardiographic monitoring were prospectively enrolled in the study and were subsequently submitted to 3-month and 6-month follow-up to investigate the occurrence of AF. The study recruited 112 patients. At follow-up visit, AF was diagnosed in 21.4% of the population. The only statistically significant difference between the group with and without a AF diagnosis was the presence of ncPAC (83.3% vs. 16.7%; P < 0.0001). ROC analysis was performed and showed that ncPAC had the best diagnostic accuracy in the AF diagnosis [AUC: 0.798; confidence interval (CI): 0.675–0.921]. The AUC of ncPAC was significantly better than the AUC of premature atrial complexes (PACs) (P < 0.05), CHA₂DS₂-VASc, HATCH, HAVOC, and C₂HEST scores (P < 0.01). Kaplan–Meier curve survival estimate for AF onset by the presence of ncPAC revealed that there was a significant difference in the AF onset between patients with ncPAC and those without (P < 0.0001) and multivariate Cox-proportional hazard analysis revealed that ncPAC presence was an independent predictors of AF onset [hazards ratio (HR): 9.28; CI 95%: 2,66–32,40; P = 0.0001].

The presence of ncPAC represents a new predictor of AF that could further guide the investigation of AF in patients with cryptogenic stroke.