No studies have directly compared the effectiveness and safety of direct oral anticoagulants (DOACs) and warfarin in patients with atrial fibrillation (AF), with or without a history of ischemic stroke and transient ischemic attack (TIA). This is important for two reasons: first, previous research reports important differences between DOACs and warfarin across other patient subgroups, and second, patients with previous stroke or TIA have a high risk of recurrent stroke.

Using 2012–2014 Medicare claims data, we identified patients newly diagnosed with AF in 2013–014 who started taking apixaban, dabigatran, rivaroxaban, or warfarin. We categorized the patients according to whether they had a history of stroke or TIA. We constructed Cox proportional hazard models that included indicator variables for treatment groups, a history of stroke or TIA, and the interaction between them, and controlled for demographic and clinical characteristics.

The hazard ratio (HR) for stroke with dabigatran, compared with warfarin, was 0.64 (95% confidence interval [CI]: 0.48–0.85) for patients with a history of stroke or TIA and 0.94 (95% CI: 0.75–1.16) for patients without a history of stroke or TIA (p-value for interaction = 0.034). In patients with previous stroke or TIA, the risk of stroke was lower with dabigatran (HR 0.64, 95% CI: 0.48–0.85) and rivaroxaban (HR 0.70, 95%CI: 0.56–0.87), compared with apixaban, but there was no difference for patients in the other subgroup.

DOACs were generally more effective than warfarin for preventing stroke. The superiority of dabigatran was more pronounced in patients with a history of stroke or TIA. The comparative effectiveness of DOACs differed substantially between patients with and without a history of stroke or TIA; specifically, apixaban was less effective in patients with a history of stroke or TIA. Our results reinforce the need to tailor anticoagulation to patient characteristics and to support the investigation of the underlying mechanisms associated with DOACs.