AimsThe ARISTOTLE (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation) trial randomised patients with atrial fibrillation at risk of stroke to apixaban or warfarin. We sought to describe patients from ARISTOTLE who prematurely permanently discontinued study drug.Methods/ResultsWe performed a posthoc analysis of patients from ARISTOTLE who prematurely permanently discontinued study drug during the study or follow-up period. Discontinuation rates and reasons for discontinuation were described. Death, thromboembolism (stroke, transient ischaemic attack, systemic embolism), myocardial infarction and major bleeding rates were stratified by ≤30 days or >30 days after discontinuation. A total of 4063/18 140 (22.4%) patients discontinued study drug at a median of 7.3 (2.2, 15.2) months after randomisation. Patients with discontinuation were more likely to be female and had a higher prevalence of cardiovascular disease, diabetes, renal impairment and anaemia. Premature permanent discontinuation was more common in those randomised to warfarin than apixaban (23.4% vs 21.4%; p=0.002). The most common reasons for discontinuation were patient request (46.1%) and adverse event (34.9%), with no significant difference between treatment groups. The cumulative incidence of clinical events ≤30 days after premature permanent discontinuation for all-cause death, thromboembolism, myocardial infarction, and major bleeding was 5.8%, 2.6%, 0.9%, and 3.0%, respectively. No significant difference was seen between treatment groups with respect to clinical outcomes after discontinuation.ConclusionPremature permanent discontinuation of study drug in ARISTOTLE was common, less frequent in patients receiving apixaban than warfarin and was followed by high 30-day rates of death, thromboembolism and major bleeding. Initiatives are needed to reduce discontinuation of oral anticoagulation.