CHAMPION-AF Trial Compares Left Atrial Appendage Closure to Blood Thinners for AFib

On March 28, 2026, Boston Scientific announced that its CHAMPION-AF trial had met all primary and secondary endpoints, declaring the Watchman FLX left atrial appendage closure (LAAC) device a viable first-line alternative to blood-thinning medications for patients with atrial fibrillation [1]. The results, published simultaneously in the New England Journal of Medicine and presented at the American College of Cardiology's annual meeting, were hailed by the company as "practice-transforming" [2].

But behind the press releases, the trial data reveal a more ambiguous picture — one in which ischemic stroke rates were actually higher in the device arm, the noninferiority margin has drawn criticism from independent cardiologists, and a competing European trial found the opposite result entirely. The question is no longer just whether plugging a small pouch in the heart can replace lifelong medication. It is whether the evidence supports expanding a $25,000–$45,000 procedure to millions of patients who are already well-served by pills costing a few dollars a day.

What CHAMPION-AF Actually Found

CHAMPION-AF enrolled 3,000 patients across multiple countries, randomizing them to receive either the Watchman FLX device or a direct oral anticoagulant (DOAC) — the current standard of care for stroke prevention in atrial fibrillation [1]. The study population had a mean age of 71.7 years, was 31.9% women, and carried a mean CHA₂DS₂-VASc stroke risk score of 3.5 [3]. Notably, 47.8% had undergone prior catheter ablation for AFib, and 68.9% had paroxysmal (intermittent) atrial fibrillation — a profile skewing toward moderate rather than high stroke risk [3].

At three years, the primary efficacy endpoint — a composite of cardiovascular death, stroke, or systemic embolism — occurred in 5.7% of the LAAC arm versus 4.8% of the DOAC arm [3]. The trial declared noninferiority with a P-value below 0.001 [1].

Source: NEJM / ACC 2026

Data as of Mar 28, 2026CSV

The headline safety finding was that non-procedure-related bleeding was significantly lower in the device group: 10.9% versus 19.0% (P < 0.001 for superiority) [1][3]. This 45% relative reduction in non-procedural bleeding became the trial's primary marketing point.

But the stroke data told a different story. Ischemic stroke or systemic embolism occurred in 3.2% of the LAAC arm versus 2.2% of the DOAC arm — a 46% relative increase in the device group [3]. The annualized absolute difference was 0.33 percentage points per year [3]. In a trial designed to show that the device is at least as good as medication, the device arm had more strokes.

The Noninferiority Margin Problem

CHAMPION-AF was a noninferiority trial, meaning it was not designed to show the device was better than DOACs, but rather that it was not unacceptably worse. The trial used a noninferiority margin of 4.8 percentage points for the primary efficacy endpoint [3]. That margin matters enormously: it defines how much worse the device can perform and still be declared equivalent.

Cardiologist John Mandrola, writing in Medscape, argued this margin was far too generous. In his analysis, "Six Reasons Why CHAMPION-AF Should Not Change Practice," he noted that including cardiovascular death in the composite endpoint — when neither treatment meaningfully affects death rates — inflates the denominator of events and makes noninferiority easier to achieve [4]. He also pointed out that the primary safety endpoint excluded procedural bleeding, effectively guaranteeing a positive result for the device arm by not counting complications from the procedure itself [4].

When procedural bleeding was folded back in, the gap narrowed substantially: combined ISTH major bleeding was 5.9% in the LAAC arm versus 6.4% in the DOAC arm [3]. The bleeding advantage of the device largely evaporated once the procedure's own complications were included.

How Earlier Trials Set the Stage — and Fell Short

CHAMPION-AF is not the first randomized trial of left atrial appendage closure. It follows PROTECT AF (2009) and PREVAIL (2014), both of which tested the original Watchman device against warfarin — an older blood thinner with a worse safety profile than modern DOACs [5].

PROTECT AF demonstrated noninferiority to warfarin but had an 8.7% major procedural complication rate [5]. PREVAIL failed to meet its first coprimary endpoint for noninferiority (posterior probability of only 88.4%) — meaning the trial could not statistically confirm that the device was not worse than warfarin for the composite of stroke, systemic embolism, and cardiovascular death [5]. Long-term pooled analyses of both trials eventually showed comparable stroke prevention and reduced hemorrhagic stroke, but only after years of follow-up and combined datasets [6].

Source: PROTECT AF, PREVAIL, NCDR Registry, CHAMPION-AF

Data as of Mar 28, 2026CSV

CHAMPION-AF's contribution is that it tested the newer Watchman FLX device against DOACs rather than warfarin — a more relevant comparator. Procedural complication rates have dropped substantially: from 8.7% in PROTECT AF to 2.3% in CHAMPION-AF [3]. But the shift to a better comparator also means the bar for the device is higher, and the trial's own stroke data suggest the device did not fully clear it.

CLOSURE-AF: The Contradictory Evidence

Complicating the picture further, the CLOSURE-AF trial — a German multicenter study of 912 patients — reported results at AHA 2025 and was published in the NEJM in March 2026 [7]. CLOSURE-AF enrolled a higher-risk population (mean CHA₂DS₂-VASc score of 5.2, mean age 79.1) and found that LAAC failed to demonstrate noninferiority to physician-directed medical therapy [7].

The primary composite endpoint occurred at 16.83 per 100 person-years in the LAAC arm versus 13.27 in the medical therapy arm, with an adjusted hazard ratio of 1.28 favoring standard care (P = 0.44 for noninferiority) [7]. Lead investigator Ulf Landmesser acknowledged: "We expected that catheter-based LAAC would be comparable to physician-directed standard medical care... However, this was not the case" [7].

The divergence between CHAMPION-AF and CLOSURE-AF may partly reflect their different patient populations: CHAMPION-AF enrolled moderate-risk patients (mean CHA₂DS₂-VASc 3.5, mean HAS-BLED 1.3), while CLOSURE-AF enrolled high-risk patients with both elevated stroke and bleeding risk scores [3][7]. This raises a paradox: the patients who arguably need an alternative to blood thinners the most — those at highest bleeding risk — may be the ones who benefit least from the procedure.

Follow the Money

CHAMPION-AF was funded entirely by Boston Scientific, the manufacturer of the Watchman FLX device [2]. The Watchman franchise generated $535 million in Q4 2025 sales alone, growing 29% year over year, and is one of Boston Scientific's fastest-growing product lines within the company's $20 billion annual revenue [8].

The trial's principal investigators disclosed significant financial relationships with the sponsor. Co-PI Dr. Kar serves as a consultant to Boston Scientific and is co-PI of multiple Boston Scientific-sponsored studies [2]. Co-PI Dr. Doshi is also a consultant to Boston Scientific [2]. Other key investigators, including Drs. Alkhouli, Coylewright, and Reddy, disclosed advisory board roles, consulting fees, or research grants from the company [2].

These relationships do not invalidate the trial results, but they are relevant context when evaluating a study whose design choices — a generous noninferiority margin, exclusion of procedural bleeding from the primary safety endpoint, and a patient population with moderate rather than high stroke risk — each tended to favor the device arm.

Who Is Actually Eligible — and Who Is Getting the Procedure

The Centers for Medicare and Medicaid Services (CMS) currently covers percutaneous LAAC under a Coverage with Evidence Development (CED) framework, requiring that patients be treated by a multidisciplinary team at a hospital with an established structural heart program [9]. The original indication was for patients with nonvalvular AFib who had a contraindication to long-term anticoagulation.

CHAMPION-AF's significance is that it tested the device in patients who could take DOACs — a much larger population. Approximately 10.5 million U.S. adults have atrial fibrillation, according to 2024 estimates from the NIH [10]. The subset with true contraindications to anticoagulation — active bleeding disorders, prior life-threatening hemorrhage, or intolerance to all available agents — is considerably smaller, though precise estimates vary.

If CHAMPION-AF's results lead to expanded indications, the addressable market for the Watchman device could grow by an order of magnitude. Boston Scientific's own press materials describe the trial as supporting "first-line therapy" for stroke risk reduction — language that implies use in the general AFib population, not just those who cannot take pills [2].

The Cost Equation

The total cost of a Watchman implantation procedure in the United States ranges from $25,000 to $45,000, with a median hospitalization cost of $24,500 according to data from over 30,000 procedures [11]. For Medicare beneficiaries, out-of-pocket costs are capped at roughly $1,500 to $3,286, with the federal program absorbing the remainder [12]. CMS has proposed reducing physician reimbursement for the procedure by approximately 16% in 2026 [13].

Source: Published cost-effectiveness analyses

Data as of Jan 1, 2025CSV

Published cost-effectiveness analyses have found that LAAC becomes cost-saving compared to DOAC therapy at approximately 8 years, driven by the cumulative year-over-year cost of medications and their associated monitoring [14]. A generic version of apixaban (Eliquis) — the most commonly prescribed DOAC — became available in early 2026, which could substantially shift this break-even calculation by lowering the annual drug cost from approximately $6,000–$7,000 to a fraction of that amount.

Higher-volume centers have been shown to achieve lower procedure costs and fewer complications [11]. But those centers are disproportionately large urban academic hospitals. Rural areas, where AFib prevalence is actually higher (5.29% vs. 3.66% in urban areas) [10], have less access to structural heart programs. If LAAC becomes standard of care, patients in rural and underserved communities could face geographic barriers to a procedure that is most cost-effective when performed at experienced, high-volume centers.

Procedural Risks in the Real World

Within the controlled setting of CHAMPION-AF, procedure-related serious adverse events occurred in 2.3% of patients [3]. Pericardial effusion — fluid accumulation around the heart — occurred in 0.7% of cases [3]. Device-related thrombus (blood clot on the device) was detected in 4.8%, though only 1.8% was deemed clinically relevant [3]. The device was successfully implanted in 92.5% of patients [3].

The NCDR LAAO Registry, which tracks real-world outcomes across hundreds of U.S. hospitals, provides a broader picture. Among 65,355 patients who received the Watchman between 2016 and 2019, pericardial effusion requiring intervention occurred in 1.35%, major bleeding in 1.25%, stroke in 0.17%, and death in 0.19% [15]. Device embolization — where the plug migrates from its intended position — occurred in 0.07% of cases [16].

Operator volume matters. The median annual case volume per physician in the NCDR was 12 procedures (interquartile range: 8–20) [15]. Higher-volume operators and centers have consistently lower complication rates. As the indication for LAAC potentially broadens, the gap between trial-center performance and community-hospital outcomes could widen.

The Strokes That a Plug Cannot Prevent

Left atrial appendage closure operates on a specific anatomical premise: that in AFib, the left atrial appendage (LAA) is where most blood clots form before traveling to the brain. Studies of surgical and autopsy specimens have found the LAA to be the thrombus source in approximately 90% of patients with nonvalvular AFib [17].

But that leaves roughly 10% of AFib-related thrombi originating elsewhere — from the left atrial body, aortic arch plaques, or other cardiac chambers [17]. DOACs, by contrast, reduce clotting systemically and protect against thrombus formation regardless of anatomic origin.

CHAMPION-AF's design does not account for non-LAA stroke mechanisms. Patients who undergo LAAC and subsequently stop anticoagulation — as is the protocol's intent — lose systemic anticoagulant protection. Residual ischemic stroke risk of 1.5% to 2.5% annually persists even in anticoagulated AFib patients, driven partly by non-thromboembolic mechanisms including small-vessel disease and atherosclerosis [17]. Closing the appendage addresses one source of clot but does not eliminate the broader stroke risk that AFib confers.

Gregory Marcus of UCSF raised additional questions about long-term consequences of occluding the LAA, including potential effects on hemodynamic and endocrine function — the appendage produces atrial natriuretic peptide, a hormone involved in blood pressure regulation — as well as the clinical significance of residual leaks around the device [3].

The Case for Staying on Pills

Electrophysiologist Jose Joglar noted that CHAMPION-AF's favorable results for the device may partly reflect the lower-risk profile of the enrolled patients [3]. With a mean CHA₂DS₂-VASc score of 3.5 and a mean HAS-BLED score of only 1.3, many participants may not have been at sufficient risk to benefit substantially from either intervention [3]. Without a control group receiving no treatment, it is impossible to determine whether the Watchman prevented strokes that would have occurred, or whether stroke rates would have been similarly low regardless.

Sanjay Kaul of Cedars-Sinai summarized the tension: the trial is "a win on the scoreboard but a loss... delivering meaningful therapeutic value" [3]. Gregory Katz of NYU called the increased stroke rate in the device arm "a pretty concerning endpoint" [3].

The strongest case for continued DOAC therapy rests on several pillars: DOACs reduce ischemic stroke more effectively than LAAC in head-to-head data [3]; they protect against non-LAA thrombus sources [17]; they carry no procedural risk; generic formulations are driving costs down; and their benefits are immediately reversible if complications arise. Proponents of medical therapy argue that a procedure carrying a 2–3% serious complication rate should not be positioned as equivalent to a well-tolerated daily pill in patients who have no contraindication to that pill.

Source: OpenAlex

Data as of Jan 1, 2026CSV

What Comes Next

The academic interest in LAAC has exploded over the past decade, with over 10,000 papers published since 2011 and a peak of 1,354 in 2023 alone [18]. CHAMPION-AF will fuel further debate, but several unresolved questions will shape whether and how practice changes.

Extended follow-up data — the trial plans to report at 5 years — will be critical. If the ischemic stroke signal in the device arm grows over time, the noninferiority finding could reverse. Conversely, if bleeding reductions persist and translate into mortality benefits, the case for LAAC strengthens.

Regulatory and payer decisions will also matter. If the FDA and CMS expand the approved indication for Watchman to include DOAC-eligible patients, procedure volumes could surge. Boston Scientific's 29% annual growth in Watchman sales suggests the commercial infrastructure is already in place [8].

For now, the evidence supports LAAC as a reasonable option for patients who genuinely cannot tolerate anticoagulation. Whether it should be offered as a first-line alternative to pills — in a population that tolerates those pills well — remains an open question that CHAMPION-AF, despite its sponsors' framing, has not definitively answered.