Direct Oral Anticoagulants and Dementia Risk in Nonvalvular Atrial Fibrillation: Evidence to Guide Clinical Selection

As our clinical landscape shifts toward managing increasingly multi-morbid aging populations, the intersection of cardiovascular health and neurodegenerative decline represents one of our most pressing challenges. For patients diagnosed with nonvalvular atrial fibrillation (NVAF), stroke prevention using oral anticoagulation is an absolute cornerstone of therapy. However, emerging evidence has increasingly pointed to an additional, profound therapeutic goal: the preservation of cognitive function.

Previous Studies to Recent Studies

While previous cohort studies established that direct oral anticoagulants (DOACs) as a class are associated with a reduced risk of incident dementia compared to traditional vitamin K antagonists like warfarin, a persistent knowledge gap has remained unaddressed. Do individual pharmacological and pharmacokinetic variances among specific DOACs—namely apixaban, edoxaban, and rivaroxaban—translate into differential risks or benefits regarding dementia prevention? A landmark population-based cohort study by Le Van and colleagues (2026) offers robust, head-to-head clarity on this question, providing vital insights that directly reshape how we approach shared clinical decision-making.

Why It Matters

The Interplay of Atrial Fibrillation and Cognitive Decline

To understand the clinical value of this study, we must first contextualize the underlying physiological pathways that tie atrial fibrillation directly to cognitive deterioration. Atrial fibrillation is far more than a rhythm disruption that predisposes patients to large-vessel ischemic strokes. Even in the complete absence of clinically overt, symptomatic thromboembolic events, patients with AF suffer from chronic, insidious cerebral insults.

The mechanism is multi-factorial. Atrial fibrillation increases the risk of silent cerebral infarcts by more than two-fold. These micro-thrombi accumulate silently over years within the cerebral parenchyma, progressively stripping away cognitive reserve and destroying white matter integrity. Over time, this ischemic burden drives vascular dementia and exacerbates underlying neurodegenerative conditions like Alzheimer’s disease by worsening mixed neurodegenerative and vascular pathologies. Furthermore, the erratic ventricular response in AF results in intermittent cerebral hypoperfusion, while the microthromboembolic environment encourages localized neuroinflammation and blood-brain barrier disruption.

The Therapeutic Rationale for Anticoagulation in Cognitive Health

By establishing a stable, therapeutic level of anticoagulation, oral anticoagulants protect the brain from both micro- and macro-embolic events. Historically, warfarin was the primary tool available, but its narrow therapeutic index and volatile Time in Therapeutic Range (TTR) often exposed patients to periods of under-coagulation (increasing ischemic micro-infarctions) or over-coagulation (increasing microbleeds).

DOACs revolutionized this landscape by offering more predictable pharmacokinetics, more stable anticoagulation profiles, and significantly lower rates of catastrophic intracranial hemorrhage. However, because individual DOACs feature distinct binding properties, dosing frequencies, half-lives, and metabolic clearance pathways, clinicians have long questioned whether one agent might possess a superior neuroprotective profile over another. For instance, differences in the rates of minor subclinical microbleeds or variations in real-world adherence due to once-daily versus twice-daily dosing regimens could theoretically tilt the scale in favor of a specific drug.

With dementia representing a profound economic, social, and personal burden, understanding whether a specific DOAC choice can actively reduce a patient’s long-term risk of cognitive decline is of paramount importance. Clinicians require concrete data to know if their prescribing habits are optimized not just for systemic stroke prevention, but for long-term brain preservation.

Who It Affects

Demographics and the Real-World Target Population

Atrial fibrillation is overwhelmingly a disease of aging. Its prevalence surges dramatically across successive decades of life, affecting approximately 37% of men and 19% of women aged 75 years and older. Concurrently, the incidence of dementia rises exponentially in this same demographic group. Consequently, the intersection of these two conditions represents an everyday clinical scenario for primary care providers, cardiologists, and geriatricians.

The study by Le Van et al. tapped into the UK Clinical Practice Research Datalink (CPRD), utilizing both the Gold and Aurum databases. This immense electronic medical records infrastructure encompasses over 60 million patients across more than 2,000 general practices, capturing a highly representative cross-section of the UK national population in terms of urban-rural distribution, socio-economic standing, and clinical complexity.

Cohort Characteristics and Inclusions

The investigators meticulously structured three distinct head-to-head new-user cohorts drawn from patients aged 50 years or older who received a new diagnosis of AF between 2011 and 2023. This focus on “new users” with incident AF is clinically significant because it eliminates the confounding associated with “prevalent users”—individuals who may have already tolerated a drug for years or survived early subclinical events, which skews observational analyses.

• Total Apixaban Initiators Evaluated: Up to 131,663 patients
• Total Rivaroxaban Initiators Evaluated: Up to 70,038 patients
• Total Edoxaban Initiators Evaluated: Up to 48,405 patients
• Patient Age Spectrum (Mean Across Cohorts): 74.7 to 77.9 years
• Gender Representation (Female Range): 42.2% to 46.2%

What makes this target population uniquely relevant to everyday practice is the extensive inclusion of patients with complex baseline comorbidity profiles. Unlike highly sanitized phase III randomized controlled trials (RCTs), which frequently exclude individuals with severe multi-morbidity, this real-world population included significant proportions of patients with long-standing hypertension, diabetes, chronic kidney disease (CKD), peripheral arterial disease, liver disease, and prior histories of clinical stroke, transient ischemic attack (TIA), or major bleeding.

Furthermore, the study took a major step forward by including individuals who had switched oral anticoagulants after their initial AF diagnosis but before cohort entry. In routine clinical care, patients frequently rotate through different OACs due to transient side effects, fluctuating renal function, or adherence struggles. Prior studies excluded these patients, filtering out 10% to 40% of the true clinical population. By retaining them, this study mirrors the exact, complex patient population currently sitting in our waiting rooms.

What Changes

Clinical Findings and Shifting Paradigms

The core finding of this extensive investigation is simple yet profoundly practice-changing: there are no meaningful differences in the risk of incident dementia when comparing apixaban, edoxaban, and rivaroxaban head-to-head.

To reach this conclusion, the authors employed robust statistical safeguards to overcome the traditional limitations of observational research. They utilized propensity score (PS) fine stratification weighting to achieve a meticulous balance across all baseline covariates, including demographics, body mass index, smoking, lifestyle factors, extensive cardiovascular and psychiatric comorbidities, concomitant medications, and even markers of health-seeking behavior. They also integrated Inverse Probability of Censoring Weighting (IPCW) to handle potential informative censoring resulting from treatment discontinuation, switching, or the competing risk of all-cause mortality.

Head-to-Head Comparisons Summary

The primary analysis evaluated exposure using an “as-treated” methodology, enforcing a rigorous 6-month lag period after drug initiation. This lag is a vital methodological feature: it ensures that cases of prodromal, undiagnosed dementia already present at the start of therapy were not falsely attributed to the DOAC, and it satisfies the biological plausibility that an anticoagulant requires an extended duration of action to meaningfully alter a chronic neurodegenerative process.

• Apixaban vs. Rivaroxaban: Adjusted Hazard Ratio (aHR) of 1.01 (95% CI 0.96 – 1.06)
• Edoxaban vs. Rivaroxaban: Adjusted Hazard Ratio (aHR) of 1.05 (95% CI 0.96 – 1.16)
• Edoxaban vs. Apixaban: Adjusted Hazard Ratio (aHR) of 1.06 (95% CI 0.98 – 1.15)

As demonstrated by the data, the adjusted hazard ratios across all three major comparative groups sit squarely around unity, with tight confidence intervals that cross 1.00. This confirms a robust lack of statistical or clinical difference between the agents.

Subgroup Nuances and Sensitivity Analyses

To ensure these findings were not masked by broader cohort trends, the researchers executed extensive secondary and sensitivity analyses. They stratified results by age (<75 vs. >75 years old), sex, history of diabetes, chronic kidney disease stages, prior stroke/TIA, and baseline stroke risk as measured by the CHA\_2DS\_2VASc score.

While isolated, minor deviations were observed in a few subgroups—such as a slightly higher point estimate for apixaban in patients aged < 75 years (HR 1.14; 95% CI 1.00–1.29) and a modest elevation for edoxaban compared to apixaban in patients over 75 (HR 1.11; 95% CI 1.02–1.20) —the authors noted that these isolated findings are best interpreted as exploratory, multi-testing anomalies likely driven by chance rather than a true biological mechanism.

Crucially, the study also evaluated whether DOACs might protect cognition through a pathway distinct from their primary role of preventing large, disabling strokes. To test this, they performed a sensitivity analysis where they censored any patient who experienced an overt ischemic stroke or TIA during the follow-up period, and further excluded anyone with a baseline history of stroke or TIA. Even after eliminating the influence of clinical cerebrovascular events, the hazard ratios remained neutral. This implies that whatever protective effect DOACs possess against dementia—likely driven by suppressing silent micro-infarctions—is uniformly shared across apixaban, edoxaban, and rivaroxaban.

Clinical Impact: Paradigm Shift in DOAC Selection

Historically, small or methodologically limited observational data had suggested potential advantages for one DOAC over another regarding cognitive preservation, creating unnecessary clinical dilemmas at the point of care.

The findings from Le Van et al. establish a clear clinical directive: Dementia risk should not be used as a differentiating factor when selecting among apixaban, edoxaban, and rivaroxaban.

Instead of worrying about specific cognitive neuroprotection profiles, clinicians can confidently shift their complete focus to well-established, evidence-based parameters to customize therapy, such as:

1. Renal Function & CrCl Tracking: Tailoring dosing dynamically based on calculated creatinine clearance (especially critical given different renal elimination fractions).
2. Adherence & Dosing Preference: Selecting once-daily options (rivaroxaban, edoxaban) vs. twice-daily options (apixaban) based on patient pill burden and compliance history.
3. Concomitant Drug-Drug Interactions: Evaluating shared metabolic pathways involving dual CYP3A4 and P-gp inhibitors/inducers.
4. Gastrointestinal Safety Profiles: Weighing specific mucosal bleeding risks based on individual patient gastrointestinal history.

Summary for Practice

The comprehensive data provided by this study delivers an exceptionally reassuring message for the medical community. In the management of nonvalvular atrial fibrillation, preventing long-term slide into cognitive impairment is a primary clinical imperative. We now have high-quality, large-scale real-world data confirming that our three primary modern frontline choices—apixaban, edoxaban, and rivaroxaban—are entirely comparable in their long-term neuroprotective capacity.

By eliminating dementia risk as a confusing variable in our clinical decision matrix, this study streamlines clinical practice. It allows healthcare providers to focus cognitive efforts on what truly works: initiating oral anticoagulation promptly upon diagnosis, ensuring consistent long-term adherence, tracking renal function meticulously to prevent inappropriate under-dosing or over-dosing, and managing companion cardiovascular risk factors like hypertension and diabetes to holistically preserve our patients’ neurological longevity.

Reference

1. Brophy JM, Le Van JQ, Beradid S, et al. Direct Oral Anticoagulants and Incidence of Dementia in Patients With Nonvalvular Atrial Fibrillation: A Population-Based Cohort Study. Neurology Open Access. 2026;2(2). doi:10.1212/WN9.0000000000000077