Abstract Body: Introduction:
Atrial Fibrillation (AF) is detected in nearly 30% of patients undergoing cardiac monitoring after ischemic stroke. Studies investigating predictors of AF showed mixed results. In this study, we aim to identify predictors of AF on insertable cardiac monitors (ICMs) and compare rates between cryptogenic stroke patients and controls.

Methods:
The ANTicoagulation And StRoke ReCurrence in ATrial FIbrillation DeteCted After Stroke (ANTARCTICA) study is an individual patient data meta-analysis of prospective observational studies of cryptogenic ischemic stroke and control patients (non-cryptogenic ischemic stroke and non-ischemic stroke) who underwent an ICM implantation. The search included prospective observational studies and randomized controlled trials of patients with non-cardioembolic ischemic stroke or transient ischemic attack or non-ischemic stroke controls who underwent prolonged cardiac monitoring with an ICM after the index event. We performed multiple imputations to derive missing covariates such as left atrial volume index. We used multivariable multi-level logistic regression models to identify clinical, imaging, and echocardiographic factors associated with AF detection. We compared AF rates and charecterisctis between cryptogenic stroke and controls.

Results:
We identified 14 studies (2 RCTs and 12 observational) that included 2036 patients (1562 cryptogenic stroke and 474 non-cryptogenic stroke and non stroke controls); AF was detected in 30.7% of cryptogenic stroke patients and 29.1% of non-cryptogenic stroke patients. In multivariable logistic regression analyses, factors associated with AF were age (OR per year increase 1.05 95% CI 1.04-1.06), left atrial volume index (OR per unit increase 1.03 95% CI 1.02-1.05), and cryptogenic stroke (adjusted OR 1.89, 95% CI 1.20-2.98, p = 0.006). When compared to controls, the time to AF detection was significantly shorter in cryptogenic stroke (median 65 days vs. 169 days, p < 0.001) and AF duration was non-significantly longer (median 90 minutes vs. 120 minutes, p = 0.144). Results remained unchanged when the control group was limited to patients with non-cryptogenic ischemic stroke.

Conclusions:
In this large, individual patient data meta-analysis of patients undergoing ICM, there is increased detection and burden of AF after cryptogenic stroke compared to controls, suggesting a likely pathogenicity of device-detected AF in cryptogenic stroke.