Timing of anticoagulation restart after serious bleeding in atrial fibrillation.

BACKGROUND

Restarting direct oral anticoagulants (DOACs) after a serious bleeding event in patients with atrial fibrillation (AF) presents a clinical dilemma, with limited evidence on the balance between stroke prevention and recurrent bleeding risk.

METHODS

Using nationwide Danish registries (2012-2021), we identified AF patients (Congestive heart failure, Hypertension, Age ≥75 (doubled), Diabetes, Stroke (doubled), Vascular disease, Age 65-74, and Sex category (female) (CHA₂DS₂-VASc score ≥2)) who experienced a first serious bleeding event while on DOAC therapy. Patients were grouped by timing of DOAC restart: within 60 days ('early restarters') vs after 60 days ('late restarters'). HRs for stroke, recurrent bleeding and a composite endpoint (stroke or serious bleeding) were estimated using multivariable Cox models. A secondary analysis examined outcomes across six time-varying antithrombotic treatment regimens.

RESULTS

Among 10 291 patients who survived 60 days postbleeding, 5970 restarted DOAC early and 4321 later. The early restart group had a lower rate of stroke (HR 0.89; 95% CI 0.74 to 1.08), but the interval includes both moderate benefit and no clear difference, indicating uncertainty in the stroke reduction. Recurrent bleeding was more frequent in early restarters (HR 1.21; 95% CI 1.07 to 1.36). In the time-varying analysis, DOAC monotherapy was associated with reduced stroke risk compared with no treatment (HR 0.78; 95% CI 0.68 to 0.89). However, bleeding risk was also higher during DOAC monotherapy (HR 1.26; 95% CI 1.15 to 1.38).

CONCLUSIONS

Restarting DOACs early after a serious bleeding event in AF patients may reduce stroke risk but is associated with an increased risk of recurrent bleeding. DOAC monotherapy appears to offer the best stroke protection, though with elevated bleeding risk. These findings highlight the need for individualised decision-making and further trials to define optimal timing for DOAC resumption.