Appendage closure matched anticoagulation in the industry-funded CHAMPION-AF trial

For most people with atrial fibrillation, the treatment that prevents a stroke is a daily anticoagulant, and the treatment's main cost is bleeding. Sealing off the left atrial appendage — the pouch where most clots form — with an implanted device has been an option for patients who cannot tolerate long-term anticoagulation. CHAMPION-AF asked a larger question: whether closure could replace the drugs in patients who are perfectly able to take them [s1].

The answer, in the trial's own terms, is yes. It is worth reading closely anyway, because the trial was funded by the company that makes the device [s1].

What the trial did

CHAMPION-AF was an ongoing, prospective, international, randomised trial involving patients with atrial fibrillation who were suitable candidates for anticoagulation [s1]. Patients were assigned in a 1:1 ratio to device-based left atrial appendage closure or to a non-vitamin K antagonist oral anticoagulant (a NOAC) [s1].

The primary efficacy endpoint — a composite of death from cardiovascular causes, stroke or systemic embolism — was tested for noninferiority, with a noninferiority margin of 4.8 percentage points, after three years of follow-up [s1]. The primary safety endpoint, non-procedure-related bleeding, was tested for superiority [s1].

That safety definition deserves a flag. It counts bleeding not caused by the implant procedure. Any bleeding tied to placing the device itself sits outside the endpoint by construction, so the safety comparison is between the drug's ongoing bleeding risk and the device's ongoing risk once it is in — not between the two strategies from day zero.

Of the 3,000 patients randomised, 1,499 were assigned to the device group and 1,501 to the anticoagulation group [s1]. Their mean age was 71.7 years, 31.9% were women, and the mean CHA2DS2-VASc score — a standard stroke-risk index — was 3.5 [s1]. A score of 3.5 describes a genuinely elevated stroke risk, the kind of patient for whom anticoagulation is normally recommended rather than optional, which is what makes the comparison a real test rather than a study of low-risk patients who might have done well on either strategy.

What happened

At three years, a primary efficacy endpoint event had occurred in 81 patients in the device group (Kaplan-Meier estimate, 5.7%) and in 65 patients in the anticoagulation group (Kaplan-Meier estimate, 4.8%) [s1]. The difference was 0.9 percentage points (95% confidence interval, -0.8 to 2.6), with P<0.001 for noninferiority [s1].

The point estimate favours the drugs by nearly a percentage point, and the upper bound of the confidence interval reaches 2.6 points. Noninferiority was met because that upper bound sits below the prespecified 4.8-point margin — a margin the reader has to accept for the conclusion to hold. This is not a finding that closure prevented more strokes; it is a finding that it did not prevent meaningfully fewer, within a tolerance the trial set in advance.

On safety the result is cleaner. Non-procedure-related bleeding occurred in 154 patients in the device group (Kaplan-Meier estimate, 10.9%) and in 260 patients in the anticoagulation group (Kaplan-Meier estimate, 19.0%), a hazard ratio of 0.55 (95% confidence interval, 0.45 to 0.67), with P<0.001 for superiority [s1]. Patients who stopped taking a daily blood thinner bled less over the following years, which is the mechanism the whole strategy rests on.

The funding, stated plainly

CHAMPION-AF was funded by Boston Scientific, the manufacturer of the appendage-closure device, and was registered as NCT04394546 [s1][s2]. That does not make the result wrong — it was randomised, sizeable and prespecified — but it sets the reading. A manufacturer-funded noninferiority trial is designed to clear a bar, and the bar here, a 4.8-percentage-point margin on a composite that includes cardiovascular death, is wide. The efficacy result cleared it while pointing numerically the other way; the bleeding result is where the device's case is strongest.

Limits

The efficacy comparison is a noninferiority claim, so it establishes that closure is not much worse, not that it is better [s1]. The safety endpoint excludes procedure-related bleeding by definition, which moves the up-front risk of implantation out of the headline comparison [s1]. Follow-up is three years; an implanted device and a lifetime of avoided anticoagulation both have longer horizons than that, and the trial is described as ongoing [s1].

What to watch

Whether guideline committees treat CHAMPION-AF as licence to offer closure to anticoagulation-eligible patients, or hold it to the standard its own numbers suggest — a bleeding advantage bought without a stroke-prevention advantage, in a trial the device maker paid for [s1][s2]. The distinction will decide how many people are offered an implant in place of a pill.

This article describes trial results. It is not medical advice, and nothing here should be used to start, stop or change any treatment.

Sources

• Left Atrial Appendage Closure or Anticoagulation for Atrial Fibrillation, New England Journal of Medicine, 28 March 2026
• CHAMPION-AF Clinical Trial (NCT04394546), ClinicalTrials.gov