Clinical and radiological predictive factors of silent AF in ESUS stroke

A quarter of cerebral infarctions remain without etiology (ESUS: embolic stroke of unknown source) with a risk of recurrence under anti-platelet therapy greater than 4% per year. In this population, the challenge is the detection of subclinical atrial fibrillation (AF), the incidence of which could reach 25%. The efficacy of the probabilistic anticoagulation strategy has not been demonstrated in this population making it necessary to identify ESUS patients at high risk of subclinical AF. The aim of this study is to determine clinical and radiological predictive factors of silent AF in ESUS stroke. Monocentric descriptive and analytical historical cohort study conducted at the Neurology Department between January 2016 and December 2020. Three hundred Esus patients were included. In our study 42/300 (14%) patients had at least one episode of AF (> 30 s). In univariate analysis, the variables significantly associated with slower AF in ESUS-type stroke were age greater than or equal to 60 years (P = 0.004, OR = 3.76), female gender (P = 0.045, OR = 1.94), coronary artery disease (P = 0.008, OR = 3.43), previous stroke (P = 0.007, OR = 2.67), stroke of different territory or age (P = 0.031, OR = 2.65), stroke on awakening (P < 10−3, OR = 5.95), higher NIHSS score on admission (P = 0.023, OR = 1.08), lower NIHSS score on discharge (P = 0.031, OR = 0.89), older cerebellar stroke (P < 10−3, OR = 14.16). In multivariate analysis, higher NIHSS score at admission (P = 0.023, OR = 1.08), and lower NIHSS score (P = 0.031, OR = 0.89) at discharge appeared as independent predictors of the occurrence of a subclinical AF episode. In this cohort of ESUS patients, several factors appeared to be predictive of subclinical AF. The subgroup of ESUS patients with these factors appeared to benefit from probabilistic anticoagulation in preventing recurrent cerebral infarction. A therapeutic trial should confirm this hypothesis.