Pattern of atrial fibrillation and risk of outcomes: The Loire Valley Atrial Fibrillation Project

Background: Risk of stroke and thromboembolism (TE) in patients with non-valvular atrial fibrillation (NVAF) is categorised in stroke risk stratification scores. The role of pattern of NVAF in risk prediction is unclear in contemporary 'real world' cohorts. Methods and results: Patients with NVAF in a four-hospital-institution between 2000 and 2010 were included. Stroke/TE event rates were calculated according to pattern of AF, i.e. paroxysmal, persistent and permanent. Risk factors were investigated by Cox regression. Among 7156 NVAF patients, 4176 (58.4%) patients with paroxysmal, 376 (5.3%) with persistent and 2604 (36.3%) with permanent patterns of NVAF were included. In non-anticoagulated patients, overall stroke/TE event rate per 100 person-years was 1.29 (95% CI 1.13-1.47). Compared with paroxysmal NVAF, rates of stroke/TE, bleeding and all-cause mortality (p<0.001) were significantly higher in permanent NVAF patients but not in persistent NVAF patients. In multivariate analyses, previous stroke (hazard ratio, HR 2.58, 95% CI 2.08-3.21), vascular disease (HR 1.34, 1.12-1.61), heart failure (HR 1.20, 1.00-1.44), age >= 75 years (HR 2.75, 2.16-3.50) and age 65-74 years (HR 1.60, 1.22-2.09) independently increased stroke/TE risk, but not persistent (HR 1.13, 0.76-1.70) and permanent (HR 1.44, 0.96-2.16) NVAF patterns. Conclusion: In this large 'real world' NVAF cohort, rates of stroke, TE, death and bleeding differed significantly by patterns of NVAF. However, only previous stroke, age, heart failure and vascular disease (not pattern of NVAF) independently increased risk of adverse outcomes in multivariate analyses. Thus, stroke risk is similar across all patterns of NVAF and antithrombotic therapy should be based on clinical risk factors, not on arrhythmia pattern. (C) 2012 Elsevier Ireland Ltd. All rights reserved.