Bleeding Risk Assessment in Anticoagulated Patients
Use the HAS-BLED score as your primary bleeding risk calculator for all patients with atrial fibrillation being considered for anticoagulation. 1, 2
Why HAS-BLED is the Preferred Tool
The 2024 ESC Guidelines and 2018 CHEST Guidelines both give strong recommendations for HAS-BLED as the bleeding risk assessment tool of choice. 1 Here's the evidence-based rationale:
Superior Predictive Performance
- HAS-BLED demonstrates moderate discriminative ability with C-statistics of 0.63–0.72, which is validated across both VKA-treated and NOAC-treated patients. 2, 3
- HAS-BLED significantly outperforms ATRIA (C-statistic 0.71 vs 0.59, P=0.035) and shows superior net reclassification improvement over HEMORR2HAGES, ORBIT, and GARFIELD-AF scores. 4
- It is the only bleeding score validated to predict intracranial hemorrhage specifically, making it particularly valuable for identifying the most devastating bleeding complication. 2
Unique Advantage for VKA Patients
- HAS-BLED includes labile INR/time in therapeutic range (TTR), a critical modifiable risk factor that ORBIT and ATRIA completely omit. 1, 2, 5
- For patients on warfarin with poor INR control (TTR <65%), the "L" component directly identifies this high-risk feature. 1, 2
How to Calculate HAS-BLED (1 point each, maximum 9)
| Component | Definition |
|---|---|
| H – Hypertension | Uncontrolled systolic BP >160 mmHg [2] |
| A – Abnormal renal/liver function | Dialysis, transplant, creatinine >200 µmol/L; OR cirrhosis, bilirubin >2× normal, AST/ALT/ALP >3× normal [2] |
| S – Stroke | Prior ischemic stroke or TIA [2] |
| B – Bleeding history | Prior major bleed, anemia, or thrombocytopenia [2] |
| L – Labile INRs | TTR <60% (for VKA patients only) [2] |
| E – Elderly | Age >65 years [2] |
| D – Drugs/Alcohol | Concomitant antiplatelet agents or NSAIDs OR ≥8 drinks/week [2] |
Risk Stratification and Action Thresholds
A HAS-BLED score ≥3 defines "high bleeding risk" and mandates specific management actions. 1, 2
Annual Major Bleeding Rates by Score
Critical Management Algorithm
Step 1: Calculate HAS-BLED for Every Patient
- Perform bleeding risk assessment at every patient contact, not just at baseline, because bleeding risk is highly dynamic. 1, 2, 6
- Serial "delta HAS-BLED" assessments are more predictive than a single baseline score. 2
Step 2: Address Modifiable Risk Factors (Do NOT Withhold Anticoagulation)
Even with HAS-BLED ≥3, anticoagulation should NOT be withheld—the net clinical benefit outweighs bleeding risk in almost all patients. 1, 2, 6 Instead, focus on these modifiable factors:
- Uncontrolled hypertension (H): Target systolic BP <160 mmHg 1, 2, 6
- Labile INR (L): Increase monitoring frequency, address medication adherence, or switch to NOACs 1, 2, 6
- Alcohol excess (D): Counsel reduction to ≤3 standard drinks per week 1, 6
- Concomitant NSAIDs/aspirin (D): Discontinue if not absolutely necessary for another indication 1, 2, 6
- Bleeding predisposition (B): Treat peptic ulcers, optimize renal/hepatic function 1, 6
Step 3: Intensify Monitoring for High-Risk Patients
Patients with HAS-BLED ≥3 require more frequent clinical reviews and follow-up. 1, 2, 6 This is a strong recommendation from both CHEST and ESC guidelines.
Step 4: Consider NOAC Selection Based on Bleeding Profile
For patients with HAS-BLED ≥3 or prior bleeding history:
- Prefer apixaban, edoxaban, or dabigatran 110 mg (where available), as all demonstrate significantly less major bleeding compared with warfarin. 1, 6
- For prior gastrointestinal bleeding specifically, apixaban or dabigatran 110 mg are preferable as they do not increase GI bleeding risk versus warfarin. 1
Common Pitfalls to Avoid
Pitfall #1: Using HAS-BLED to Withhold Anticoagulation
Never use HAS-BLED ≥3 as a standalone reason to avoid anticoagulation. 1, 2, 6 High bleeding risk patients often have high stroke risk, and the net benefit of anticoagulation is typically greatest in this population. 6
Pitfall #2: Static Risk Assessment
Do not rely on a single baseline HAS-BLED score—bleeding risk evolves over time with changes in blood pressure control, renal function, medication regimens, and alcohol use. 1, 2, 6 Reassess at every encounter.
Pitfall #3: Ignoring the "L" Component in VKA Patients
For patients on warfarin with TTR <65%, the labile INR component adds 1 point and should trigger interventions to improve INR control or consideration of switching to a NOAC. 1, 2
Pitfall #4: Withholding Anticoagulation Due to Fall Risk
Do not withhold oral anticoagulation solely because of fall risk—this is a strong recommendation against this practice. 2 The number of falls required to offset the stroke prevention benefit is extraordinarily high (>295 falls per year in one analysis).
When to Consider Alternative Scores
ORBIT may be considered as a secondary tool if you need a simpler score that does not require INR data, but it is inferior to HAS-BLED for VKA-treated patients because it omits labile INR/TTR. 5 ORBIT is most useful in NOAC-treated populations where the "L" component is not applicable.
Do NOT use CHADS₂ or CHA₂DS₂-VASc for bleeding risk assessment—these are stroke risk scores, and HAS-BLED demonstrates significantly higher C-statistics for bleeding prediction. 3
Special Populations
Cancer Patients with AF
- CHA₂DS₂-VASc and HAS-BLED have not been fully validated in cancer patients, so use the T-B-I-P algorithm (thromboembolic risk, bleeding risk, drug-drug interactions, patient preferences) as recommended by 2024 ESC guidelines. 1
- HAS-BLED should still be calculated to identify modifiable factors, but recognize that cancer-specific bleeding risks (e.g., intracranial tumors, GI malignancies, thrombocytopenia) require additional clinical judgment. 1