Residual Risks of Thrombotic Complications in Anticoagulated Patients with Atrial Fibrillation: A Cluster Analysis Approach from the GLORIA-AF Registry

Assessment of residual thromboembolic risk in patients with atrial fibrillation (AF) prescribed oral anticoagulants (OACs) remains unexplored. We performed hierarchical cluster analysis to identify phenotypic profiles of these patients and their risks of residual thromboembolic events. We utilised data from non-valvular AF patients on OACs, as documented in phases II and III of the GLORIA-AF (Global Registry on Long-Term Oral Anti-thrombotic Treatment in Patients With Atrial Fibrillation) registry. We performed a hierarchical cluster analysis to identify distinct phenotypic profiles. We compared the incidence and risks of thromboembolic events (composite of ischaemic stroke, transient ischaemic attack, or systemic embolism) and related outcomes (major bleeding and all-cause death) across the profiles. We determined the optimal number of profiles through visual inspection of the generated dendrograms. We included 22,410 patients (mean age 70 ± 8 years; 56% male), from which five phenotypes were identified: profile 1 ("uncontrolled hypertension"), profile 2 ("young with a history of coronary artery disease"), profile 3 ("young and obese"), profile 4 ("frailty"), and profile 5 ("non-paroxysmal AF with tachycardia"). Profile 4 was associated with the highest rates of thromboembolic events (1.66/100 person-years [95% confidence interval, 1.46-1.89]), major bleeding (1.92/100 person-years [1.70-2.16]), and death (6.02/100 person-years [5.62-6.43]). Profile 3 was associated with the lowest risk across all measured outcomes (thromboembolic events, 0.64 events/100 person-years [0.48-0.82]; major bleeding, 0.83 events/100 person-years [0.65-1.04]; and death, 1.44 events/100 person-years [1.21-1.71]). Profile 1 had a moderate thromboembolic event rate (1.04/100 person-years [0.91-1.08]), while profiles 2 and 5 showed lower rates. The phenotypic profiles of patients with AF prescribed OACs identified using hierarchical cluster analysis are associated with distinct residual thromboembolic risks and related outcomes. This approach has the potential to enhance patient risk-stratification and holistic approaches to management.