Blood thinners cut events in lower-risk atrial fibrillation, but the numbers are tiny

People with atrial fibrillation at the borderline — where guidelines shrug and say a blood thinner is optional — had fewer strokes and other serious events over two years when they took one, in a South Korean trial called SINGLE-AF [s1]. The catch, and it is a large one, is that the events were rare in everybody: the drug cut the combined rate from 1.5% to 0.5%, a one-percentage-point difference built on a handful of events [s1]. The result nudges a genuinely uncertain guideline recommendation toward treatment, but its small absolute numbers are the whole story.

Atrial fibrillation is a common irregular heartbeat that lets blood pool and clot in the heart, raising the risk of stroke. Doctors gauge that risk with the CHA₂DS₂-VASc score, which adds points for age, sex, and conditions like high blood pressure and diabetes [s1]. At the high end, anticoagulation is clearly worth it; at zero, it is not. The awkward middle — a score of 1 in men or 2 in women — is where guidelines from the US and Europe list anticoagulation only as a "class IIa" option, meaning reasonable but not firmly backed by trial evidence [s1]. SINGLE-AF set out to supply the missing trial.

What the trial did

This was a multicentre, open-label superiority trial in South Korea, with an important safeguard: although patients and doctors knew who was taking the drug, the committee judging whether events had occurred was kept blinded, which guards against wishful scoring of ambiguous outcomes [s1]. Patients with atrial fibrillation and exactly that intermediate score were randomly assigned in equal numbers to take a direct oral anticoagulant (DOAC) — the modern class of blood thinner that has largely replaced warfarin — or no anticoagulation at all [s1].

The primary endpoint bundled the good and the bad of anticoagulation into one composite: stroke, systemic embolism, major bleeding, or death from cardiovascular causes at 24 months [s1]. That design matters, because a blood thinner is a trade-off — it prevents clots but can cause bleeds — so a fair test has to count both on the same ledger. The trial was funded by South Korea's Ministry of Health and Welfare, a public body rather than a drug manufacturer [s1].

What it found

Of 1,803 patients randomised, 902 were assigned to a DOAC and 901 to no anticoagulant; their mean age was 60.4 years and 23.7% were women [s1]. At 24 months, a primary-endpoint event had occurred in 4 patients — a cumulative incidence of 0.5% — in the DOAC group, versus 13 patients, or 1.5%, in the no-anticoagulant group [s1]. That works out to a difference of 1.0 percentage point (95% confidence interval −2.0 to −0.1; P=0.03) and a hazard ratio of 0.31 (95% CI 0.10 to 0.94) [s1].

The composite was driven by stroke: 3 patients (0.3%) had one in the DOAC group versus 10 (1.1%) without the drug [s1]. Reassuringly on the safety side, systemic embolism and major bleeding appeared similar between the groups, and no cardiovascular deaths occurred in either [s1]. Serious adverse events of any kind were near-identical: 80 patients (8.9%) on the DOAC and 84 (9.3%) without [s1].

How to read it

Statistically, the DOAC won — the confidence interval clears 1.0 and the P value clears 0.05 [s1]. But two features demand caution before this reshapes practice. First, the entire result rests on 17 events across more than 1,800 people, so the hazard ratio, impressive as 0.31 looks, sits on a thin foundation and its confidence interval stretches nearly to 1 [s1]. Second, the trial was open-label, and although the outcome adjudication was blinded, unblinded trials can still bias softer parts of care [s1].

For a reader in that intermediate zone, the honest framing is one of small absolute stakes. Without the drug, roughly 15 in 1,000 such patients hit the composite over two years; with it, about 5 — meaning most people in this risk band will neither have an event nor be spared one, whichever they choose [s1]. That is precisely the situation where a frank conversation about bleeding risk, convenience, and preference matters more than any single trial. Related coverage has looked at how alcohol can trigger atrial fibrillation, when wearable ECGs help investigate unexplained fainting, and what counts as high blood pressure.

What to watch

SINGLE-AF was conducted in one country with a relatively young population, so whether the same modest edge holds in older, more diverse patients is unsettled [s1]. Guideline committees weighing a class IIa recommendation will have to decide how much weight a single trial of 17 events should carry, and whether the small benefit is worth committing lower-risk patients to lifelong blood thinners.

This article describes research and is not medical advice. Decisions about anticoagulation are for treating clinicians.

Sources

• Anticoagulation for Atrial Fibrillation with Intermediate Stroke Risk — New England Journal of Medicine, 28 August 2026