Background: Transesophageal echocardiography visualizes the left atrium and its appendage, thrombi, and spontaneous echocardiographic contrast.
Objective: To assess the association of transesophageal echocardiographic characteristics with stroke or embolism in atrial fibrillation.
Design: Multicenter observational follow-up study.
Setting: Hospitals in Austria and Slovakia.
Patients: 409 outpatients with nonrheumatic atrial fibrillation and without recent stroke.
Intervention: Patients with thrombi received anticoagulation, and patients without thrombi received aspirin.
Measurements: Primary events were stroke or embolism. Secondary events were death not caused by stroke or embolism and need for anticoagulation.
Results: In the left atrium or left atrial appendage, 10 patients (2.5%) had thrombi and 47 (12%) had spontaneous echocardiographic contrast. The appendage had a mean (+/- SD) length of 44+/-10 mm, a mean width of 23+/-6 mm, and a mean area of 5.8+/-2.5 cm2. Follow-up ranged from 1 to 74 months (mean, 58 months). Fifty patients had stroke or embolism, 53 died of a cause other than stroke or embolism, and 38 required anticoagulation. On univariate analysis, thrombi (risk ratio, 3.9 [95% CI, 1.4 to 10.1]; P = 0.009), length of the left atrial appendage (risk ratio, 1.6 [CI, 1.05 to 2.5]; P = 0.03), and width of the left atrial appendage (risk ratio, 2.4 [CI, 1.2 to 4.81; P = 0.01) were associated with stroke or embolism. Multivariate analysis identified hypertension (risk ratio, 3.6 [CI, 1.8 to 8.4]; P = 0.001), previous stroke (risk ratio, 3.7 [CI, 1.5 to 7.5]; P = 0.002), and age (risk ratio, 1.1 [CI, 1.0 to 1.11; P < 0.001) as risk factors for stroke or embolism and provided evidence of an association between thrombi and stroke or embolism (risk ratio, 2.4 [CI, 0.9 to 6.9]; P = 0.09).
Conclusions: In outpatients with atrial fibrillation and without recent stroke, thrombi of the left atrium or left atrial appendage and length and width of the left atrial appendage were associated with stroke or embolism in univariate analysis. In a multivariate analysis, age, hypertension, and previous stroke were risk factors for stroke or embolism, and thrombi of the left atrium or left atrial appendage were possible risk factors. In these patients, history may be more useful than transesophageal echocardiography for the assessment of embolic risk.