CHA2DS2-VASc Calculator – Your AF Stroke Risk Score, Explained

Related tools

How it works, in three steps

Going deeper

Four people, four scores, one table

The CHA2DS2-VASc score is a point score used to estimate the yearly risk of stroke in people with atrial fibrillation, and the yearly rates on this page come from the original study by Lip and colleagues, published in 2010. This section works through four profiles that were run through the calculator above: the default profile, the same person ten years older, a person with no risk factors, and a person with every factor. The score counts points only; it does not weigh which factors produced them, and the yearly rate for each score is a group average from that study, not a personal prediction. Nothing here recommends any treatment, and whether any treatment is right depends on the full clinical picture, including bleeding risk, which this score does not measure.

The default profile is the page’s own worked example, and it is worth adding slowly, because it shows how the letters map to points. C stands for heart failure, worth 1 point, and it is not ticked here. H stands for hypertension, worth 1, and it is ticked. A stands for age, worth 1 for age 65 to 74 and 2 for age 75 or older, and at 70 it contributes 1. D stands for diabetes, worth 1, and it is ticked. S stands for prior stroke, TIA or thromboembolism, worth 2, and it is not ticked here. V stands for vascular disease, worth 1, and it is not ticked. Sc stands for female sex, worth 1, and it is ticked. Add them: 1 + 1 + 1 + 1 = 4, and the published adjusted stroke rate for a score of 4 in the original study is 4.0 percent per year. The same total can be reached by a very different route, as the page notes: a man of 76 who has had a prior stroke, with nothing else ticked, also scores 4, from age 2 plus prior stroke 2. The score only counts points, and two people with the same score can have entirely different clinical pictures.

The second row shows what a birthday can do. The same woman at 80 instead of 70 moves from the 65-to-74 age band to the 75-or-older band, so the age contribution rises from 1 to 2, the total rises from 4 to 5, and the published rate rises from 4.0 percent to 6.7 percent per year. Nothing about her health changed; only her age band did, and the score moved by a full point with it. That is also why the age bands never stack: someone aged 80 gets 2 age points in total, not 3, because the two bands are alternatives, not additions. The page states that rule explicitly, and it is the most common scoring error to correct: adding 1 for 65-to-74 and 2 for 75-or-older together would give a maximum score of 10 and a systematically inflated total for every older person.

The third and fourth rows mark the ends of the scale. A man of 40 with no risk factors scores 0, and the published rate for a score of 0 in the original study is 0 percent per year: in that study’s population, no one in the score-0 group had the outcome measured, which is a statement about that group, not a guarantee for any individual. A woman of 80 with every factor scores 9, the maximum, and the published rate is 15.2 percent per year. Between those ends, the full published table, which the page prints in full, runs: 0 at 0 percent, 1 at 1.3, 2 at 2.2, 3 at 3.2, 4 at 4.0, 5 at 6.7, 6 at 9.8, 7 at 9.6, 8 at 6.7 with small numbers, and 9 at 15.2 percent per year. The score-8 figure sits below the score-7 figure because very few patients in that study had a score of 8, so that one estimate is shaky, and the page flags it rather than smoothing it away. A table that dips where it should rise is a reminder that these rates are measured averages from one group of patients at one time, not a smooth law of nature.

Why the same score carries different rates in different studies

The percentages on this page come from the original 2010 study, a hospital-based survey of European patients with atrial fibrillation. Later studies that followed whole national populations, including large Danish and Swedish cohorts, found somewhat different rates for the same scores, often lower at the low end and higher at the top end. The score itself, the points, is stable and is what guidelines are built on; the percentages are averages measured in one group of patients at one time. That distinction matters for how the calculator should be used: the points are the portable part, and the rate is the part that must be read with its source attached. A score of 4 computed here is the same score of 4 anywhere, but “4.0 percent per year” is specifically the Lip 2010 figure, and a clinician discussing the score today will place it alongside more recent population data and the individual patient’s circumstances.

The sex point illustrates the same distinction between a stable score and a moving interpretation. In the studies behind the score, women with atrial fibrillation had a higher stroke risk than men with the same other factors, so sex is one of the scored items, worth 1 point. But sex works more like a risk modifier than a stand-alone risk factor: the page notes that a woman whose only point is for sex, giving a score of 1, is usually considered low risk. The point is counted the same way in every case, and the interpretation of a lone sex point is more cautious than the interpretation of a point from heart failure or diabetes. That nuance is exactly the kind of judgement that belongs in a clinical conversation, not in a calculator: the tool counts the point, and the clinician weighs what the point means for this patient, alongside bleeding risk, which the score does not measure at all.

Age illustrates a different nuance: it is the strongest single driver of stroke risk in atrial fibrillation, which is why the score gives it extra weight, 2 points from 75, against 1 point for most other factors and 2 for a prior stroke event. The double weight is a statement about relative importance in the original data, not a statement that age causes stroke in the way a prior stroke does. A person cannot change their age, and the score’s purpose is not to assign blame but to stratify risk so that a conversation about prevention can be prioritised. The practical consequence is that two people with identical conditions but ages of 74 and 75 can differ by a full point, and a birthday can move a score even when nothing about health has changed. That discontinuity is a property of any banded score, and it should be read as a reminder that the underlying risk rises smoothly with age while the score rises in steps.

How to use this score in a conversation, and what it cannot do

The page’s own framing for using the score is a three-step process: tick the factors that apply, add the points, and read the published rate, using it to inform a conversation with a clinician, never to decide treatment on its own. That framing is worth following literally. The calculator’s output is most useful as preparation: it lets a person arrive at a clinical conversation knowing what the number means, which factors contributed to it, and what yearly rate was published for that score in the original study, so the conversation can focus on the decisions that the score cannot make. Those decisions include whether any preventive treatment is appropriate, which depends on the full clinical picture, including bleeding risk, other conditions, medicines, and the patient’s own preferences, none of which this page measures or models.

The page’s own FAQ answers the treatment question directly, and this section repeats that answer because it is the most important sentence on the page: no, a high score does not by itself mean a person should take a blood thinner. This calculator cannot make that decision, and it does not recommend any treatment. Whether blood-thinning treatment is right depends on the full clinical picture, and the score is one input among many to that decision, not the decision itself. The same restraint applies at the low end: a score of 0 with a published rate of 0 percent per year in the original study is not a guarantee of no risk, and a score of 1 from the sex point alone is described on the page as usually low risk, not as no risk. In both directions, the score informs but does not decide, and any result from this page should be discussed with a clinician, who can place it in the context of the individual patient.

Worked checks: adding a score by hand

The default score is the hand check to memorise. Heart failure 0, hypertension 1, age points 1 for age 65 to 74, diabetes 1, prior stroke 0, vascular disease 0, female sex 1: the total is 4, and the published rate for a score of 4 is 4.0 percent per year. The age-80 variant changes only the age line: heart failure 0, hypertension 1, age points 2 for age 75 or older, diabetes 1, prior stroke 0, vascular disease 0, female sex 1, for a total of 5, and a published rate of 6.7 percent per year. The no-risk profile is all zeros for a total of 0 and a published rate of 0 percent per year. The every-factor profile is heart failure 1, hypertension 1, age points 2, diabetes 1, prior stroke 2, vascular disease 1, female sex 1, for a total of 9, and a published rate of 15.2 percent per year. Every one of those totals was run through the calculator, and every one follows from the same point values, with the age bands never stacking.

The check that catches the most scoring errors is the age check: confirm the age band first, because it is the only line worth 2 points besides prior stroke, and the only line where two values compete. A person aged 74 takes 1 age point; a person aged 75 takes 2; nobody takes 3. The second check is the prior stroke line, which is worth 2 points for a prior stroke, TIA or thromboembolism, not 1: undercounting it by one point understates the total and the rate together. The third check is the maximum: the highest possible score is 9, from 1 + 1 + 2 + 1 + 2 + 1 + 1, and any total above 9 means the age bands were stacked or a line was double-counted. A total that passes all three checks is arithmetically correct; whether it is clinically meaningful is a separate question, and it belongs with a clinician, who can read the score alongside bleeding risk and the rest of the patient’s picture.

How a score changes over time

A CHA2DS2-VASc score is not a fixed property of a person; it is a snapshot of the factors that apply at the moment of scoring, and it moves as those factors change. The most common move is the age-band move already described: the same woman with hypertension and diabetes moves from 4 to 5 on her 75th birthday, with no change in her health, and the published rate moves from 4.0 to 6.7 percent per year. A new diagnosis moves it the same way: if the default profile, a woman of 70 with hypertension and diabetes at a score of 4, were also found to have heart failure, the heart failure point would raise the total to 5, again at a published rate of 6.7 percent per year, by a completely different route to the same total. And a prior stroke event, worth 2 points, moves any profile by two at once: the no-risk man of 40 at a score of 0 would move to 2 on a first stroke event, at a published rate of 2.2 percent per year, before any other factor changed.

That movement is why re-scoring matters, and why a score computed years ago should not be quoted as current without re-checking the factors. The factors are all things that change over a lifetime: blood pressure can rise into the hypertensive range, diabetes can develop, vascular disease can be diagnosed, and age bands are crossed on birthdays. A score that was low when computed can become moderate without any dramatic event, simply through the accumulation of one point here and one point there, and the published rate moves with it, from 1.3 percent at a score of 1 to 2.2 at 2, 3.2 at 3, and 4.0 at 4. The calculator makes re-scoring cheap: tick the factors as they stand now, read the new total, and take the new breakdown to the clinical conversation. What re-scoring never does, on its own, is change any treatment: the score informs the conversation, and the clinician, with the full picture including bleeding risk, decides what, if anything, should change.

Thresholds, the sex-category point, and CHA2DS2-VA

Anticoagulation is recommended at 2 or more in men and 3 or more in women, considered at 1 in men and 2 in women, and not recommended at 0 in men or 1 in women where the point is sex alone. Female sex alone never warrants anticoagulation.

The score adds 1 point each for heart failure, hypertension, diabetes, vascular disease, and age 65 to 74, 2 points for age 75 or older, 2 points for prior stroke, TIA, or thromboembolism, and 1 point for female sex category, to a maximum of 9. Annual stroke rates without treatment rise steeply with the score — roughly 0.2% at 0, about 2.2% at 2, about 3.2% at 3, about 4% to 5% at 4, and about 7% at 5 — and anticoagulation reduces ischaemic stroke by roughly two thirds, which is why the absolute benefit is large at high scores and smaller than the bleeding risk at low ones. Scores 3 to 9 are all high risk and all carry the recommendation; the distinction between them is the magnitude of the risk and the urgency, not the decision.

The sex-category point is a risk modifier, not a standalone risk factor: women with AF have slightly higher stroke risk than men with the same other factors, but a woman with no other factor, scoring 1 from sex alone, is low risk. Recent guidelines have moved to CHA2DS2-VA, which drops the sex point entirely and treats 2 or more as the threshold for both sexes and 1 as the consideration band. Bleeding risk, assessed alongside the stroke score, identifies modifiable factors to fix — blood pressure, interacting drugs, alcohol — rather than forbidding treatment, and direct oral anticoagulants are generally preferred over warfarin. This is general information about a published tool, not medical advice; any anticoagulation decision belongs with the treating clinician. The companion guide explains every component, the score-by-score meaning, the annual rates, the bleeding-risk balance, and the limits of the score.

Frequently asked questions

More calculators

This calculator is for education only. It estimates a published risk score for people with atrial fibrillation; it does not diagnose any condition, it does not predict what will happen to any one person, and it does not recommend, start or stop any treatment, including blood-thinning medication. Decisions about stroke prevention belong to you and your clinician, who knows your full history. If you think you may be having a stroke, seek emergency help immediately.