Figure 2.

Average marginal effect of individual geriatric syndromes on anticoagulant use adjusted for stroke risk. A negative marginal effect indicates lower use of anticoagulants. Models adjusted for stroke risk using CHA₂DS₂-VASc score. Red point estimates and confidence intervals denote values that are not, statistically, different from the reference group. Blue point estimate and confidence intervals denote values that are statistically different from the reference group. All levels within a syndrome are hierarchical and mutually exclusive. We present tabular results in Supplementary Table S5. Falls: If a participant reported multiple falls and any fall that resulted in an injury requiring medical attention, the participant was categorized as having had an injurious fall. Analysis on 773 participants; excludes 3 missing falls data and 3 missing anticoagulation data. Activity of daily living (ADL): Difficulty defined as participant reported difficulty completing one or more ADLs and not requiring help with any ADL. ADLs include bathing, getting out of bed, dressing, eating, toileting, and walking. Analysis on 773 participants; excludes 3 missing ADL data and 3 missing anticoagulation data. Instrumental ADL (IADL): Difficulty defined as participant reported difficulty completing one or more IADLs and not requiring help with any IADL. IADLs include shopping for groceries, preparing hot meals, taking medications, making telephone calls, and managing money. Analysis on 776 participants; excludes 3 missing anticoagulation data. Cognitive status: defined using the Langa-Wier score.³⁸ Analysis on 776 participants; excludes 3 missing anticoagulation data. Incontinence analysis on 774 participants; excludes 2 missing incontinence data, 2 missing anticoagulation data, and 1 missing both.