The prevalence of atrial fibrillation increases substantially with age. As atrial fibrillation carries a higher risk of thromboembolic events, several scores have been developed to estimate thromboembolic and bleeding risk in order to help with the prophylactic decision.

To determine thromboembolic and bleeding risk of elderly with atrial fibrillation according to CHADS₂, CHA₂DS₂-VASc and HASBLED and its repercussions on thromboembolic prophylaxis.

Retrospective, observational study including 142 consecutively hospitalized patients over 65 years old, with non-valvular atrial fibrillation/flutter. CHADS₂ and CHA₂DS₂-VASc were applied and compared and HASBLED score was used to estimate haemorrhagic risk. The adequacy of prescribed antithrombotic therapy was evaluated. Long-term follow-up of thromboembolic and haemorrhagic events was carried out.

None of the elderly patients were allocated to the low-risk category according to CHADS₂ and CHA₂DS₂-VASc risk stratification. CHADS₂ classified 32 (22.5%) patients at moderate risk, while CHA₂DS₂-VASc score classified all patients at high risk. Applying the HASBLED score, 57 (40.1%) had high haemorrhagic risk. Although by CHA₂DS₂-VASc all patients had a formal indication for anticoagulation, only 77 (54.2%) were anticoagulated. Age was found to be a common criteria for withholding oral anticoagulation. The thromboembolic event rate was 2.6% for anticoagulated patients and 11.5% for not anticoagulated ones, while major haemorrhages occurred in 6.5% anticoagulated and 1.5% not anticoagulated patients.

All elderly with atrial fibrillation had high thromboembolic risk, better predicted by CHA₂DS₂-VASc. Anticoagulation, the only factor that can alter prognosis, was underused despite the evidence of the scores. Paradoxically, age alone was frequently considered a contra-indication for anticoagulation.