Geriatric Conditions Predict Discontinuation of Anticoagulation in Long-Term Care Residents With Atrial Fibrillation.

BACKGROUND Anticoagulation (AC) for stroke prevention in long-term care (LTC) residents with atrial fibrillation (AF) involves a challenging risk-benefit evaluation. We measured the association of geriatric conditions with discontinuation of AC. DESIGN Retrospective cohort analysis. SETTING LTC facilities across the United States. PARTICIPANTS A total of 48 545 individuals residing in LTC facilities in 2015 with AF and sufficient information to establish their status as someone who stopped AC vs someone who continued AC. MEASUREMENTS We measured the association of six geriatric conditions-recent fall, severe activity of daily living (ADL) dependency (21-28 on a 28-point scale), mobility impairment, cognitive impairment, body mass index (BMI) less than 18.5 kg/m(2), and weight loss (>= 5% in 1 month or >= 10% in 6 months)-with discontinuation of AC. To identify cases of discontinuation, we required a pattern of being on AC over two consecutive recordings of the Minimum Data Set, the nursing home quality control data set recorded every 90 days, followed by two assessments being off AC-pattern of "on-on-off-off." By contrast, we required a pattern of "on-on-on-on" for continuers. We then constructed six logistic regression models to measure the independent association between each geriatric condition and discontinuation of AC, adjusted for CHA(2)DS(2)-VASc stroke risk score, recent bleeding hospitalization, and other confounders. RESULTS There were 4172 discontinuers and 44 373 continuers. Recent fall predicted a 1.9-fold increase in the odds of discontinuation (odds ratio = 1.91; 95% confidence interval = 1.66-2.20), whereas mobility and cognitive impairment only increased the odds by 14% to 17%. Severe ADL dependency, BMI less than 18.5 kg/m(2), and weight loss of 10% each increased odds of discontinuation by 55% to 68%. CHA(2)DS(2)-VASc score did not predict discontinuation. CONCLUSION Several geriatric conditions predicted discontinuation of AC, whereas CHA(2)DS(2)-VASc score did not. Future research should examine the association of geriatric conditions and discontinuation of warfarin discrete from newer anticoagulants and association of geriatric conditions with development of stroke and bleeding. J Am Geriatr Soc 68:717-724, 2020