Direct oral anticoagulants in very elderly (≥90 years) patients with atrial fibrillation: observed versus predicted thromboembolism and bleeding risk, and HAS-BLED performance
K Abd El Hay, A Abu Husseine, R Hilu, N Ivan, D Pereg, H Vaknin-Assa, A AssaliAbstract
Background
Very elderly patients with atrial fibrillation (AF) carry high predicted thromboembolic risk yet are underrepresented in randomised trials of direct oral anticoagulants (DOACs). Evidence on real-world stroke protection at extreme age and the performance of bleeding risk scores in this setting remains limited.
Purpose
To evaluate thromboembolic and major bleeding outcomes in DOAC-treated AF patients aged ≥90 years, contextualise observed 1-year event rates against CHA2DS2-VASc and HAS-BLED–based expected risks, and assess HAS-BLED discrimination for major bleeding.
Methods
Retrospective single-centre cohort study at Meir Medical Center (Israel) including consecutive AF patients aged ≥90 years treated with DOACs between 2016–2021. Follow-up started at DOAC initiation and continued until end of follow-up or death. Outcomes were stroke/transient ischaemic attack (TIA) and major bleeding. Event rates were reported as incidence per 100 person-years and cumulative 1-year rates. HAS-BLED performance was assessed by AUC for 1-year bleeding and Harrell’s C-index; association with bleeding was tested using Cox regression.
Results
The cohort included 480 patients (mean age 93.0±2.6 years; 63.5% women), mean CHA2DS2-VASc 4.7±1.2 and mean HAS-BLED 2.6±0.7. Over 1556 person-years (median follow-up 2.9 years), stroke/TIA occurred in 31 patients (2.0 per 100 person-years) with a 1-year rate ~1.3%, markedly lower than the ~5–6% annual risk expected without anticoagulation. Major bleeding occurred in 49 patients (3.15 per 100 person-years) with a 1-year rate 4.4%, consistent with the ~4–6% annual risk expected from HAS-BLED. HAS-BLED showed poor discrimination (AUC 0.49; 95% CI 0.38–0.60; C-index 0.47) and was not associated with major bleeding (HR per point 0.80; 95% CI 0.54–1.19; p=0.29). All-cause mortality was high (85.0%).
Conclusion
In AF patients aged ≥90 years, DOAC therapy was associated with substantially lower observed thromboembolic risk than predicted by CHA2DS2-VASc, while major bleeding remained clinically relevant. HAS-BLED demonstrated limited ability to discriminate individual bleeding risk at extreme age, supporting the need for geriatric-informed risk assessment to guide anticoagulation decisions.