Predictors of newly diagnosed atrial fibrillation in cryptogenic stroke: a cohort study


  • See editorial by Seet on page 1331.

Correspondence: J.-M. Bugnicourt, Service de Neurologie, CHU Amiens, Place Victor Pauchet, F-80054 Amiens Cedex 1, France (tel.: +33 322 668240; fax: +33 322 668244; e-mail:


Background and purpose

A significant proportion of cryptogenic ischaemic strokes are due to paroxysmal atrial fibrillation (AF). As paroxysmal AF appears to inexorably progress to persistent or permanent AF, this study with long-term follow-up was designed to establish the profile of patients who developed AF after hospital discharge.


All patients with cryptogenic ischaemic stroke over a 1-year period were included (n = 164). Patients were prospectively followed up at the outpatient clinic. Information on long-term outcome included the presence of newly diagnosed AF (NDAF). A specific NDAF assessment was performed at least 2 years after the index stroke using a structured telephone interview. Baseline clinical, laboratory, and echocardiographic data of these patients were retrospectively recorded. Independent predictive factors were then used to produce a predictive grading score for NDAF, derived by logistic regression analysis.


With a median follow-up of 854 days, 22 cases of NDAF (13%) were observed. On multivariate analysis, factors associated with NDAF were age ≥72 years (two points), history of coronary artery disease (one point) or stroke (one point), and left atrial area ≥16 cm2 (two points) (total score ranging from 0 to 6). Patients with a score ≤1 point did not have NDAF during follow-up.


In cryptogenic ischaemic stroke, the NDAF score can be used to target patients at high risk of developing AF after hospital discharge, as a score of 0–1 was highly predictive of the absence of NDAF during follow-up. These results need to be confirmed in prospective studies.