RESUMO
OBJECTIVE: To investigate prevalence and predictors of oral anticoagulant therapy (OAT) deprescribing in older inpatients with atrial fibrillation (AF), and its association with 1-year incidence of major clinical outcomes. DESIGN: Multicenter retrospective cohort study. SETTING AND PARTICIPANTS: Inpatients aged ≥75 years with known AF on OAT at admission discharged from 3 Italian acute geriatric wards between January 2014 and July 2018. METHODS: Data from a routine Comprehensive Geriatric Assessment (CGA), along with OAT status at discharge were recorded. One-year incidence of all-cause death, stroke or systemic embolism (SSE), and major and clinically relevant nonmajor bleeding (MB/CRNMB) were retrieved from administrative databases. Associations were explored through multilevel analysis. RESULTS: Among 1578 patients (median age 86 years, 56.3% female), OAT deprescription (341 patients, 21.6%) was associated with bleeding risk, functional dependence and cognitive impairment, and inversely, with previous SSE and chronic AF. Incidences of death, SSE, and MB/CRNMB were 56.6%, 1.5%, and 4.1%, respectively, in OAT-deprescribed patients, and 37.6%, 2.9%, and 4.9%, respectively, in OAT-continued patients, without significant differences between groups. OAT deprescription was associated with all-cause mortality [adjusted odds ratio (aOR) 1.41, 95% CI 1.68-1.85], along with older age, comorbidity burden, cognitive impairment, and functional dependence, but with neither SSE nor MB/CRNMB incidence, as opposed to being alive and free from SSE and MB/CNRMB, respectively (aOR 0.68, 95% CI 0.25-1.82, and aOR 0.95 95% CI 0.49-1.85, respectively). Conversely, OAT deprescription was associated with higher odds of being dead than alive both in patients free from SSE and in those free from MB/CRNMB. CONCLUSIONS AND IMPLICATIONS: CGA-based OAT deprescribing is common in acute geriatric wards and is not associated with increased SSE. The net clinical benefit of OAT in geriatric patients is strongly related with the competing risk of death, suggesting that functional and cognitive status, as well as residual life expectancy, should be considered in clinical decision making in this population.