Stroke and Bleeding Risk in Long-Term Care Residents With Atrial Fibrillation Who Discontinue Anticoagulation
Alok Kapoor, Zhiyong Chen, Jay Patel, Zehao Ye, Honghuang Lin, Geoffrey Barnes, Sybil Crawford, Jerry H. GurwitzImportance
Atrial fibrillation (AF) leads to more than 100 000 cases of stroke yearly and is particularly prevalent in long-term care (LTC) residents. Anticoagulation (AC) is the mainstay of stroke prevention, but given its risks, clinicians and residents may consider discontinuation at the time of LTC entry.
Objective
To estimate the association between AC discontinuation and stroke and bleeding risk among LTC residents with AF.
Design, Setting, and Participants
This comparative effectiveness research study with target trial emulation used claims data (2020-2021) from LTC facilities in the US serving Medicare beneficiaries. Medicare beneficiaries with AF receiving AC who entered LTC between April 1 and December 31, 2020, were included; those with additional indications for AC, including prosthetic heart valve or venous thromboembolism, were excluded. Analysis was performed in August 2026.
Exposure
Discontinuer vs continuer assignment.
Main Outcomes and Measures
The main outcomes were stroke (including ischemic stroke, transient ischemic attack, and systemic embolism) and bleeding, assessed using Medicare claims. A target trial emulation was conducted with clone censor weighting (CCW) analysis. Specifically, eligible LTC residents were assigned (cloned) to each of 2 strategies: AC discontinuation or AC continuation. Clones were then censored when they deviated from their assigned strategy, and inverse probability of censoring weighting was applied to address informative censoring: residents who deviated differing systematically from those who did not. Subdistribution hazard ratios (sHRs) for stroke and bleeding were then calculated to account for competing mortality and a cause-specific hazard ratio (HR) for death.
Results
This study included 15 426 LTC residents (mean [SD] age, 82.6 [7.6] years). Most were female (62.6%). AC discontinuers had 187 strokes, 297 bleeding events, and 3684 deaths over a mean (SD) of 128.9 (177.9) days. AC continuers had 321 strokes, 493 bleeding events, and 5475 deaths over a mean (SD) of 279.6 (204.9) days. Discontinuers had similar stroke risk (sHR, 1.07; 95% CI, 0.93-1.23) and bleeding risk (sHR, 1.02; 95% CI, 0.92-1.13) compared with continuers, but discontinuers had higher mortality risk (HR, 1.27; 95% CI, 1.23-1.31).
Conclusions and Relevance
In this target trial emulation with CCW in a cohort of LTC residents with AF, AC discontinuation was not associated with higher stroke or bleeding risk. Very high mortality, unmeasured confounding, and reliance on claims-based outcome ascertainment may have limited the ability to better understand the consequences of AC discontinuation on stroke and bleeding.