Guidelines have recommended oral anticoagulation as a class IIa option for atrial fibrillation at intermediate stroke risk — a CHA2DS2-VASc score of 1 in men and 2 in women — without a randomised trial behind it. SINGLE-AF is that trial. It was multicentre, open-label with masked adjudication, conducted in South Korea, and randomised 1,803 patients to a direct oral anticoagulant or to no anticoagulation. Mean age was 60.4 years and 23.7% were women. The primary endpoint at 24 months was a composite of stroke, systemic embolism, major bleeding or cardiovascular death.
DOAC therapy won. Primary events occurred in 4 patients on DOAC (cumulative incidence 0.5%) against 13 on no anticoagulation (1.5%): a difference of 1.0 percentage point (95% CI -2.0 to -0.1, p=0.03) and a hazard ratio of 0.31 (0.10 to 0.94). Stroke drove it, 3 events against 10. Systemic embolism and major bleeding looked similar between arms and there were no cardiovascular deaths at all. Serious adverse events were 8.9% against 9.3%.
What this settles is direction, and what it does not settle is enthusiasm. Seventeen primary events across 1,803 patients over two years is a very low event rate, and the entire benefit rests on those numbers. The number needed to treat for two years is about 100. That is a real benefit worth having in a patient who is happy to take a tablet, and a thin one to insist on in a patient who is not.
For Indian practice, the population matters. AF here presents younger, so more patients sit in exactly this intermediate band than in Western cohorts, and this trial's mean age of 60 is closer to the Indian norm than most anticoagulation evidence. Two cautions: valvular AF from rheumatic heart disease is a different disease with a different answer, and bleeding risk in patients with limited access to monitoring and to emergency care is not the bleeding risk of a Korean trial site. Use the result to stop hedging on the indication, not to stop assessing the individual.
- Offer a DOAC to men with CHA2DS2-VASc 1 and women with 2 rather than leaving the decision open-ended.
- Quote the size honestly: about one primary event avoided per 100 patients treated for two years.
- Bleeding was not increased in this trial, which strengthens the case in patients with no specific bleeding risk.
- Do not extend this to rheumatic or valvular AF, which was not the population studied.
- Reassess the score periodically — age and new comorbidity move patients out of this band quickly.
The statistics, in plain English
A hazard ratio of 0.31 sounds dramatic, but its confidence interval runs from 0.10 to 0.94 — wide, and only just clear of 1.0. That width is what happens when a result rests on 17 events in total. The absolute difference of 1.0 percentage point, with an interval of -2.0 to -0.1, is the more honest number to carry into a consultation. The trial was open-label, so patients and doctors knew the allocation; endpoint adjudication was masked, which protects the outcome assessment but not decisions about who gets investigated in the first place.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for cardiology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free