CHA₂DS₂-VASc (maximum 9). C — congestive heart failure or left ventricular dysfunction, 1 · H — hypertension, 1 · A₂ — age 75 or over, 2 · D — diabetes mellitus, 1 · S₂ — previous stroke, TIA or systemic thromboembolism, 2 · V — vascular disease (prior myocardial infarction, peripheral arterial disease or aortic plaque), 1 · A — age 65–74, 1 · Sc — sex category female, 1. Age points are derived from the age you enter, so 65–74 scores 1 and 75 or over scores 2 and they can never both be ticked.

Anticoagulation thresholds, sex-adjusted. Female sex on its own is a risk modifier, not a risk factor, so the thresholds are not the same number for both sexes: recommend oral anticoagulation at a score of 2 or more in men and 3 or more in women (ESC 2020 Class I; NICE NG196 says offer at 2 or more); consider it at 1 in men and 2 in women (ESC Class IIa); no anticoagulation at 0 in men or 1 in women where the single point is female sex. DOACs are preferred over warfarin in almost all of these patients.

Annual ischaemic stroke risk without anticoagulation — Friberg, Benson and Rosenqvist, BMJ 2012, 182 678 patients with atrial fibrillation: score 0 → 0.2 % · 1 → 0.6 % · 2 → 2.2 % · 3 → 3.2 % · 4 → 4.8 % · 5 → 7.2 % · 6 → 9.7 % · 7 → 11.2 % · 8 → 10.8 % · 9 → 12.2 % per year. The Lip 2010 cohort, which is smaller, gives lower figures at the top of the range; the confidence intervals at scores of 7 and above are very wide and those numbers should be treated as indicative. The clinically useful part of the table is not the decimals — it is that the risk roughly doubles between a score of 1 and a score of 3, which is why the threshold sits where it does.

HAS-BLED (maximum 9). H — hypertension, defined as uncontrolled systolic above 160 mmHg, 1 · A — abnormal renal function (dialysis, transplant, or creatinine above about 200 µmol/L), 1 · A — abnormal liver function (cirrhosis, or bilirubin over twice the upper limit with transaminases or alkaline phosphatase over three times), 1 · S — previous stroke, 1 · B — previous major bleeding or a predisposition to bleeding including anaemia, 1 · L — labile INR (time in therapeutic range under 60 %), 1 · E — elderly, age over 65, 1 · D — drugs, meaning antiplatelet agents or NSAIDs, 1 · D — alcohol, 8 or more units a week or binge drinking, 1. Renal and liver abnormality, and drugs and alcohol, each score independently, which is why the maximum is 9 and not 7.

How to read HAS-BLED. A score of 3 or more means high bleeding risk. That is not a reason to withhold anticoagulation — the guideline point of the score is to identify the modifiable contributors and fix them: control the blood pressure, stop the NSAID, address the alcohol, stabilise the INR or switch from warfarin to a DOAC, correct the anaemia, review the antiplatelet. What changes is the frequency of review, not the decision. In the original validation cohort major bleeding rose steeply above a score of 3, while the stroke prevented by anticoagulating still usually outweighs the bleeding provoked.

Worked example. A 72-year-old woman with hypertension, diabetes and a previous TIA, on aspirin, drinking about 10 units a week, with normal renal and liver function. CHA₂DS₂-VASc: age 65–74 (1) + hypertension (1) + diabetes (1) + TIA (2) + female (1) = 6, annual stroke risk about 9.7 %, anticoagulation clearly indicated. HAS-BLED: age over 65 (1) + drugs (1) + alcohol (1) = 3, high bleeding risk — which means stop the aspirin if there is no separate indication for it, address the alcohol, and review more often, not "don't anticoagulate".

What these scores do not cover. CHA₂DS₂-VASc is validated for non-valvular atrial fibrillation. A mechanical heart valve or moderate-to-severe mitral stenosis with atrial fibrillation is an indication for anticoagulation with warfarin regardless of the score, and DOACs are contraindicated in mechanical valves. Neither score is validated in children, in AF under 65 with no other risk factor for whom the absolute risk is very low anyway, or in atrial flutter treated as though it were not AF. They say nothing about the patient's ability to take a tablet, their falls risk (which is routinely over-weighted — a patient would need to fall hundreds of times a year for a subdural to outweigh the stroke benefit in most cases), their frailty, their preferences or their life expectancy. Both scores are also dynamic: age alone moves the score every birthday, so a score of 1 in a 64-year-old becomes 2 a year later.

Disclaimer. A decision-support aid for qualified clinicians — not medical advice and not a substitute for the shared decision-making conversation the guidelines ask for. Sources: Lip et al., Chest 2010 · Friberg, Benson & Rosenqvist, BMJ 2012 · Pisters et al., Chest 2010 · ESC 2020 atrial fibrillation guideline · NICE NG196. Last reviewed: 2026-09-21.