Rungpradubvong V.Lip G.Y.H.Krittayaphong R.Mahidol University2026-03-312026-03-312026-01-01Thrombosis and Haemostasis (2026)03406245https://repository.li.mahidol.ac.th/handle/123456789/115889Background We examined the association of oral anticoagulants (OACs) discontinuation following a rise in HAS-BLED score among patients enrolled in a nationwide AF registry. Methods We analyzed 1,030 patients from the prospective, multicenter COOL-AF registry who were receiving OACs at baseline and subsequently experienced an increase in HAS-BLED score. Patients were categorized according to OAC continuation (n ¼ 936) or discontinuation (n ¼ 94) at the time of score increase. The primary composite outcome included all-cause mortality, stroke/systemic embolism (SSE), and major bleeding. Multivariable Cox regression and propensity score matching (1:2) were performed to adjust for confounders. Results During a median follow-up of 23.7 months (IQR 13.3–29.9), patients who discontinued OACs had substantially higher event rates than those who continued therapy (20.2 vs. 7.9 per 100 person-years; p < 0.001). OAC discontinuation independently associated with an increased risk of the composite endpoint (adjusted HR 2.65, 95% CI 1.75–4.02; p < 0.001), all-cause mortality (aHR 3.66, 95% CI 2.25–5.96; p < 0.001), and major bleeding (aHR 2.60, 95% CI 1.27–5.33; p ¼ 0.009), without a significant difference in SSE (aHR 0.39, 95% CI 0.05–2.89). Findings were confirmed in the propensity-matched cohort. No significant difference in stroke/systemic embolism was observed, potentially reflecting competing mortality risk and residual confounding inherent to the observational design. Conclusion OAC discontinuation after a HAS-BLED score increase was associated with worse outcomes, likely reflecting residual confounding rather than a direct causal effect, and should prompt risk optimization rather than automatic cessation.MedicineArticleSCOPUS10.1055/a-2815-53002-s2.0-1050332015892567689X41734781