Use mainly in adults with non-valvular atrial fibrillation who are being considered for, or already receiving, long-term oral anticoagulation, particularly warfarin.

The HEMORR₂HAGES scheme was derived from the US National Registry of Atrial Fibrillation (NRAF), including 3,791 adults with atrial fibrillation, of whom 1,604 were treated with warfarin. The score combines 11 baseline bleeding risk factors and was designed to predict major bleeding during anticoagulation.

HEMORR₂HAGES score	Major bleeds per 100 patient-years on warfarin*
0	1.9
1	2.5
2	5.3
3	8.4
4	10.4
≥5	12.3

*Approximate rates from the original warfarin-treated NRAF cohort.

Subsequent studies comparing HEMORR₂HAGES with other bleeding scores (HAS-BLED, ATRIA) in anticoagulated atrial fibrillation cohorts show that all have only modest discrimination, with HAS-BLED generally performing slightly better for overall clinically relevant bleeding and intracranial haemorrhage.

Evidence is strongest in older, warfarin-treated atrial fibrillation populations; calibration in patients on direct oral anticoagulants or with other indications (for example venous thromboembolism) is less certain.
The genetic-factors item was not directly measured in the original NRAF dataset and remains conceptual; use it only when a clinically meaningful bleeding-related variant is documented.
Current atrial fibrillation guidance supports using a validated bleeding risk score to identify and address modifiable bleeding risks and to inform shared decision-making, rather than as a stand-alone reason to withhold anticoagulation when stroke risk is high.

The primary source or current guidance provides the model definition, source population and supporting evidence.

HEMORR2HAGES Score for Major Bleeding Risk in Atrial Fibrillation