Objective Current stroke risk schemes need improvement of predictive value in patients with atrial fibrillation. Transoesophageal echocardiography (TEE) may facilitate stroke risk assessment in such patients and guide antithrombotic treatment.

Methods We randomised 238 patients with non-valvular atrial fibrillation and a moderate stroke risk to aspirin or adjusted vitamin K antagonist therapy after TEE had ruled out thrombogenic features in the atria and aorta. The primary outcome was a composite of stroke, major bleeding, peripheral embolism and all-cause mortality.

Results Mean CHA₂DS₂-VASc score was 2.1±1.1. The incidences of the composite primary outcome at a mean follow-up of 1.6 years were 3.2% (2.02% per year) in the aspirin group compared to 6.1% (3.84% per year) in the vitamin K antagonists group with an absolute advantage of 2.9 percentage points. Aspirin was non-inferior to vitamin K antagonists (p<0.0001) because the upper limit of the 90% CI did not exceed the 7% absolute difference in event rate between the two treatment arms.

Conclusions This hypothesis-generating pilot trial has found that TEE may be used for refinement of stroke risk in paroxysmal atrial fibrillation patients. A larger trial is needed to confirm these data. (ClinicalTrials.gov number NTC00224757).