Older adults living with both atrial fibrillation and Alzheimer’s disease who took newer oral anticoagulants declined more slowly on a standard cognitive test than similar patients on warfarin or on no anticoagulant, according to a nationwide Swedish analysis published in the European Heart Journal.
The difference was small in any single year. The annual decline on the Mini-Mental State Examination was about 0.23 points lower with these drugs than with no anticoagulation, and roughly 0.21 points lower than with warfarin, on a test scored out of 30. Researchers say the gap could become clinically meaningful if it accumulates over several years.
This is observational research. It cannot show that the medication caused the slower decline, and the researchers say so directly.
The Overlap That Makes This Decision Difficult
Studies have found that about 15 to 20 percent of people with Alzheimer’s disease also have atrial fibrillation, lead author Maria Eriksdotter of Karolinska Institutet told Medical News Today. That overlap creates a genuinely hard clinical problem.
Atrial fibrillation raises stroke risk, and anticoagulation is the standard way to lower it. But people with dementia are also at elevated risk of falls, of bleeding, and of missing doses, which has historically made clinicians cautious about prescribing blood thinners in this group. Some patients end up with nothing at all.
The study drew on SveDem, Sweden’s national dementia quality register, identifying people diagnosed with Alzheimer’s or mixed Alzheimer’s dementia between May 2007 and December 2020 who already had a recorded atrial fibrillation diagnosis. After exclusions, the published paper analyzed 7,308 people: 3,341 taking no anticoagulant, 2,277 on warfarin, and 1,690 on a newer agent. Mean age at Alzheimer’s diagnosis was 82.6 years, and 52.7 percent were women.
The design has a structural advantage worth noting. Because Sweden maintains linked national registers covering dementia diagnoses, prescriptions, hospitalizations, and deaths, the researchers could follow the same people across all these outcomes without relying on patients to report anything. That does not solve the central problem: patients were not randomly assigned to their treatment.
The class in question includes apixaban, rivaroxaban, dabigatran, and edoxaban, often called NOACs or direct oral anticoagulants. Compared with older warfarin, they require less routine blood monitoring and have more predictable effects.
The Outcomes Beyond Memory Scores
Cognition was not the only measure, and the other findings arguably matter more for a treatment decision today.
According to the Karolinska Institutet announcement, people on the newer drugs had lower risks of death, stroke, blood clots, and fractures than those receiving no anticoagulant. Both drug classes were associated with lower rates of ischemic stroke, systemic embolism, and mortality relative to no treatment, with stronger associations for the newer agents. The difference showed up in harms as well: the stroke, mortality, fracture, and bleeding results showed warfarin users experiencing a higher rate of major bleeding while newer-agent users had a lower rate of fracture.
Eriksdotter offered a plausible mechanism rather than a proven one, saying there are “reasons to believe that the treatment could have a positive effect on cognition,” for example by improving blood flow and reducing small-scale damage in the brain.
The investigators were careful about what should change. While awaiting more evidence on cognitive outcomes, they wrote, anticoagulation decisions should remain primarily guided by stroke prevention rather than anticipated cognitive benefit. Their conclusion was narrower than the headline: the findings support these drugs as preferred over warfarin in this population, which is already the direction most contemporary practice has moved.
The Confounding Problem the Data Themselves Reveal
A responsible reading turns on who was prescribed what, and the paper’s own baseline table makes the concern concrete.
Compared with either treated group, baseline differences between the groups show that non-users were slightly older, had lower baseline cognitive scores, and had lower recorded prevalence of hypertension, dyslipidemia, and pacemaker implantation, as well as lower use of blood pressure drugs, beta-blockers, and statins. In other words, the people who received nothing were not simply the same patients minus a prescription. They were sicker cognitively at the outset and less engaged with cardiovascular care. Any of that could produce a difference in decline without the drug doing anything at all.
The direction of the finding is consistent with earlier work. A systematic review of prior studies covering 22 studies and more than 600,000 patients found that most reported reduced cognitive impairment or dementia among patients receiving anticoagulation, with direct oral anticoagulants associated with a lower incidence in 10 of them. But that literature is overwhelmingly observational and shares the same vulnerability. Consistency across studies with the same design flaw is not the same as confirmation.
What is missing is a randomized trial in this specific population, meaning people who already have Alzheimer’s disease. None has been done.
The Conversation Worth Having at the Next Appointment
Nobody should start, stop, or switch an anticoagulant because of this study. That instruction comes from the researchers themselves, and it is the single most important line in this coverage.
What the findings reasonably support is a different conversation at the next appointment. For a patient with both conditions who is currently on warfarin, it is fair to ask whether a newer agent would be appropriate, since the evidence on stroke, mortality, fracture, and bleeding already favored that direction before cognition entered the discussion. For a patient on no anticoagulant, it is fair to ask why and whether the fall and bleeding concerns that drove that decision still apply.
Those questions belong with a cardiologist, neurologist, or geriatrician who knows the patient’s bleeding history, kidney function, fall risk, and other medications. Kidney function, in particular, affects dosing of these drugs and is not something a family can assess at home.
Caregivers should also know that abruptly stopping an anticoagulant carries a real stroke risk, which is why unsupervised changes are dangerous rather than merely unwise. Missed doses matter too, and for a household managing dementia, a pill organizer, a reminder system, or pharmacy blister packaging is a more useful intervention than any change of drug.
Readers should also be aware of the study’s disclosures. The work was funded by the Swedish Research Council, the Swedish Brain Foundation, and other public sources, and the senior author reported consulting participation in one-off meetings with several pharmaceutical companies and paid lectures at company-sponsored symposia.
What remains unresolved is whether the cognitive difference is real or an artifact of who gets prescribed what. Answering that would require a randomized trial in this specific population. MedicalDaily will report if one is launched or if larger analyses replicate the finding.
Key Questions Answered
What did the study find? Among 7,308 Swedish patients with both atrial fibrillation and Alzheimer’s disease, newer oral anticoagulants were associated with cognitive decline about 0.2 points per year slower on the Mini-Mental State Examination than warfarin or no anticoagulant.
Does this prove the drugs protect the brain? No. This is registry-based observational research, and the paper’s own data show that people who received no anticoagulant differed from those who did in age, baseline cognition, and other treatments.
Which medications are involved? Apixaban, rivaroxaban, dabigatran, and edoxaban are known as non-vitamin K oral anticoagulants or direct oral anticoagulants.
What else did the study show? Lower risks of death, stroke, blood clots, and fractures versus no anticoagulation. Warfarin also lowered stroke and death risk relative to no treatment but was associated with more major bleeding.
Do other studies agree? Most earlier observational work points in the same direction, but it shares the same design limitation. No randomized trial has tested this question in people who already have Alzheimer’s disease.
Should anyone change medication because of this? No. The researchers say anticoagulation decisions should still be driven by stroke prevention. Stopping an anticoagulant without supervision carries a real stroke risk.
What should a family ask a doctor? Whether the current anticoagulant choice still fits the patient’s bleeding history, fall risk, kidney function, and other medications.
