How is my stroke risk score used to decide whether to stop?

The short version: your stroke risk score is a rough guide but the real question is whether your atrial fibrillation is genuinely gone.

You are asking how the CHA2DS2-VASc score is used to decide whether to stop a NOAC after successful cardioversion. The answer is that the score is used as a threshold, but the evidence behind that threshold is weaker than most patients are told.

The CHA2DS2-VASc score was built from observational data. It assigns points for age, sex, and conditions like hypertension, diabetes, and prior stroke [2, 14]. Higher scores mean higher estimated annual stroke risk. For a score of 0, the estimated rate is about 1 stroke per 100 people per year. For a score of 4, it is about 5.8 per 100 per year [2]. The score's ability to predict who will actually have a stroke is modest. The C-statistic, a measure of predictive accuracy, is only about 0.67, meaning it is barely better than a coin flip [14].

The clinical logic goes: if your score is low (0-1 in men, 1-2 in women), the risk of stroke off anticoagulation may be low enough to stop, especially if the atrial fibrillation has been eliminated. But the threshold was set by expert consensus, not by a trial that compared stopping versus continuing in people with the same score. The 2010 ablation study found that even patients with CHADS2 scores of 2 or higher who stopped anticoagulation had no strokes during follow-up [12]. That suggests the score alone may not tell you who is safe to stop.

What the score does not capture is whether your atrial fibrillation is actually gone. A 2020 study found that patients who stopped anticoagulation but later had recurrence of atrial fibrillation had a thromboembolism rate of 1.62 per 100 person-years, compared to 0.33 for those who stayed in sinus rhythm [11]. The adjusted hazard ratio for recurrence was 4.49 [11]. That means the single most important predictor is not your risk score but whether the arrhythmia stays suppressed.

The score also does not measure blood stasis directly. A 2014 study using 4D flow MRI found that higher CHA2DS2-VASc scores correlated with greater left atrial blood stasis, but the correlation was modest (r = -0.41) [1]. The score is a proxy for what matters most: how well blood moves through your atrium.

My call: your stroke risk score is a starting point, but the decision to stop a NOAC should rest more on objective evidence that the atrial fibrillation is truly gone (monitoring, repeated ECGs, or an implantable loop recorder) than on the score alone. The evidence for using the score as a stopping threshold is weak and observational. Confidence: moderate.