Assessing the Appropriateness of Oral Anticoagulation for Atrial Fibrillation in Advanced Frailty: Use of Stroke and Bleeding Risk-Prediction Models.

Background: Atrial fibrillation (AF) is common among frail older adults. Oral anticoagulation (OAC) is particularly challenging for these due to overlapping stroke and bleeding risk factor profiles.

Objective: To compare the utility of stroke and haemorrhage risk-prediction instruments in the treatment of AF among frail older adults.

Design: Cross-sectional study.

Settings And Participants: Frail residents in four nursing homes with a Clinical Frailty Scale score ≥5 (median 7±0).

Measurements: The prevalence of AF was assessed by ECG and chart review. Stroke (CHADS2 and CHA2DS2-VASc) and bleeding (HASBLED and HEMORR2HAGES) risk-prediction scores were then applied. A validated, risk-based, colour-coded decision support tool, incorporating these instruments, was then used to create a risk matrix and assess the appropriateness of OAC.

Results: In total, 225 patients were included. The distribution of CFS scores was similar irrespective of AF status. In all, 86/225 (38%) had any history of AF. Of these, only 15/86 (17%) were prescribed OAC. All those in AF scored ≥2 on the CHA2DS2-VASc. One-third also scored high-risk of bleeding using HAS-BLED or HEMORR2HAGES. Risk-prediction scores were similar between those with 'known' (documented) and occult (only on ECG) AF. The colour-coded decision tree suggested that OAC would be recommended for the majority in AF when HAS-BLED (60/86, 70%) was used as the bleeding risk-prediction instrument. Despite this, only 12/60 (20%) were anticoagulated. When HEMORR2HAGES was incorporated instead, one patient was advised OAC, the remainder no treatment (57%) or an antiplatelet (42%).

Discussion: Stroke risk was high and bleeding risk levels comparatively low, suggesting that the balance of risk may favor OAC for AF in this cohort of patients with advanced frailty. Despite this and the high prevalence of AF, OAC prescription rates were low. The decision-support tool used showed mixed results, depending on the bleeding-risk score incorporated, suggesting that while useful, they should not replace clinical judgement.