In this target trial emulation with clone censor weighting in a cohort of long-term care residents with atrial fibrillation, anticoagulation discontinuation was not associated with higher stroke or bleeding risk, though discontinuers had higher mortality.
Key Findings
Results
Stroke risk was similar between long-term care residents with atrial fibrillation who discontinued anticoagulation and those who continued it.
Discontinuers vs continuers: sHR, 1.07; 95% CI, 0.93-1.23.
Stroke outcome included ischemic stroke, transient ischemic attack, and systemic embolism, assessed using Medicare claims.
Discontinuers had 187 strokes over a mean (SD) of 128.9 (177.9) days; continuers had 321 strokes over a mean (SD) of 279.6 (204.9) days.
Subdistribution hazard ratios accounted for competing mortality.
Results
Bleeding risk was similar between anticoagulation discontinuers and continuers.
Discontinuers vs continuers: sHR, 1.02; 95% CI, 0.92-1.13.
Discontinuers had 297 bleeding events and continuers had 493 bleeding events.
Bleeding was assessed using Medicare claims, with competing mortality accounted for in the subdistribution hazard model.
Results
Anticoagulation discontinuers had higher mortality risk than continuers.
Cause-specific HR for death, 1.27; 95% CI, 1.23-1.31.
Discontinuers had 3684 deaths and continuers had 5475 deaths over differing follow-up periods.
Mortality in the cohort was very high overall.
Methods
The study cohort consisted of 15,426 Medicare beneficiaries with AF on anticoagulation who entered long-term care in 2020.
Mean (SD) age was 82.6 (7.6) years; 62.6% were female.
Residents entered LTC between April 1 and December 31, 2020; claims data were from 2020-2021.
Residents with additional AC indications, including prosthetic heart valve or venous thromboembolism, were excluded.
Methods
The study used a target trial emulation with clone censor weighting to compare discontinuation and continuation strategies.
Eligible residents were cloned and assigned to both strategies: AC discontinuation or AC continuation.
Clones were censored when they deviated from their assigned strategy.
Inverse probability of censoring weighting addressed informative censoring from systematic differences between those who deviated and those who did not.
Conclusions
The authors identified limitations that may have constrained understanding of the consequences of anticoagulation discontinuation.
Very high mortality in the cohort.
Unmeasured confounding.
Reliance on claims-based outcome ascertainment.
What This Means
This study looked at more than 15,000 older Medicare beneficiaries with atrial fibrillation (an irregular heart rhythm that raises stroke risk) who were taking blood thinners when they entered long-term care facilities in 2020. The researchers used claims data and a statistical method that mimics a clinical trial to compare residents who stopped blood thinners with those who continued them.
They found that stopping blood thinners was not associated with a higher risk of stroke or of bleeding. However, residents who stopped had a higher risk of death, about 27% higher than those who continued. The study cannot say whether stopping caused the higher death risk.
This research suggests that the balance of stroke and bleeding risk may not differ much for long-term care residents who stop blood thinners, but the results should be interpreted cautiously. Death rates were very high in this group, some factors could not be measured, and outcomes were identified from billing claims rather than detailed medical records. Decisions about blood thinners remain individual and involve clinicians and residents.
Kapoor A, Chen Z, Patel J, Ye Z, Lin H, Barnes G, et al.. (2026). Stroke and Bleeding Risk in Long-Term Care Residents With Atrial Fibrillation Who Discontinue Anticoagulation.. JAMA network open. https://doi.org/10.1001/jamanetworkopen.2026.40964