Predictors of Occult Atrial Fibrillation After Cryptogenic Stroke | Journal Scan

Study Questions:

What predicts atrial fibrillation (AF) during outpatient cardiac monitoring of patients with cryptogenic stroke?


This was a retrospective study of patients at the University of Pennsylvania with cryptogenic stroke or transient ischemic attack (TIA) who had 28 days of ambulatory cardiac monitoring between April 2010 and November 2012. Abstractors, blinded to the results of the cardiac monitoring, reviewed clinical and demographic data, including echocardiography results and imaging studies. Patients completed 28 days of ambulatory cardiac monitoring with the CardioNet Mobile Cardiac Outpatient Telemetry device. AF was detected by automated analysis with verification by cardiologist review. Patients could also manually activate the device. AF of any duration was classified as AF. Multivariable analysis with logistic regression was used to look for predictors of AF, adjusting for age (>60 vs. ≤60 years), prior cortical or cerebellar stroke, sex, and race.


There were 227 patients included in the analysis (179 with stroke and 48 with TIA). Cardiac monitoring was started a median of 64 days after the index stroke. AF was found in 31 patients (14%) and the median duration of AF was 4.5 minutes. Ten percent of patients with AF reported palpitations that were associated with the arrhythmia. The time between the stroke and initiation of ambulatory cardiac monitoring was not associated with detecting AF (p = 0.61). In univariate analysis, older age (p = 0.005) and prior cortical/cerebellar infarction (p = 0.021) predicted AF. These prior infarctions were not always symptomatic strokes (i.e., some were asymptomatic infarcts). Median CHA2DS2-VASc score (p = 0.28), left atrial size (p = 0.39), and multiple acute infarctions in multiple vascular territories (p = 0.36) did not predict AF. In the multivariate analysis, age >60 years (odds ratio [OR], 3.7; 95% confidence interval [CI], 1.3-11) and prior cortical or cerebellar infarction (OR, 3.1; 95% CI, 1.2-7.6) were associated with detecting AF during monitoring. The probability of detecting AF was 33% in patients with these two risk factors. A sensitivity analysis only looking at patients with AF >30 seconds yielded similar results.


Older age and prior cortical or cerebellar infarction (clinical or asymptomatic) predicted the development of AF during 28 days of ambulatory cardiac monitoring after cryptogenic stroke or TIA.


More than 30% of ischemic strokes are considered cryptogenic. AF is an important stroke risk factor and paroxysmal AF can be difficult to detect. Ambulatory cardiac monitoring, using various devices, is commonly used after cryptogenic stroke and TIA to detect AF. This study further validates the role of ambulatory cardiac monitoring in detecting AF and identifies predictors that can help identify which patients will have AF during ambulatory cardiac monitoring. It is somewhat surprising that echocardiographic findings and acute infarctions in multiple vascular territories were not associated with detecting AF. These findings will need to be replicated in large multicenter studies to determine if these associations are truly absent.

Clinical Topics: Arrhythmias and Clinical EP, Noninvasive Imaging, Implantable Devices, SCD/Ventricular Arrhythmias, Atrial Fibrillation/Supraventricular Arrhythmias, Echocardiography/Ultrasound

Keywords: Arrhythmias, Cardiac, Atrial Fibrillation, Demography, Echocardiography, Heart Atria, Heart Conduction System, Ischemic Attack, Transient, Monitoring, Ambulatory, Outpatients, Risk Factors, Stroke, Telemetry, Logistic Models, Multivariate Analysis, Retrospective Studies

< Back to Listings