Development of AF in Embolic Stroke of Undetermined Source

Quick Takes

  • Detection of AF in patients with embolic stroke of uncertain source (ESUS) is important because it typically changes management (prompts initiation of anticoagulation or pursuit of left atrial appendage occlusion).
  • This secondary analysis of the RE-SPECT ESUS randomized controlled trial sought to determine what clinical factors predict detection of AF in an ESUS population.
  • The most important predictors of detection of AF were older age, hypertension, lack of diabetes, higher BMI, and higher NT-proBNP.

Study Questions:

What clinical factors predict atrial fibrillation (AF) detection in patients with recent embolic stroke of uncertain source (ESUS)?


This is a secondary analysis of data from the RE-SPECT ESUS placebo-controlled trial (May 2019), which randomized patients with recent ESUS to either dabigatran or aspirin to prevent recurrent stroke. AF was identified through investigators’ standard of care as well as by a protocol-required 1-year electrocardiography. N-terminal pro–B-type natriuretic peptide (NT-proBNP) level was evaluated as an AF predictor for a subset of patients (n = 1,117) who had a baseline level available. Of the 5,390 enrolled patients, 403 (7.5%) had AF detected over a median 19 months of follow-up. Predictors of AF detection were sought by comparing ESUS subjects with AF to those without AF.


On multivariable analysis, independent predictors of AF were older age (odds ratio [OR] for 10-year increase, 1.99; 95% confidence interval [CI], 1.78-2.23), hypertension (OR, 1.36; 95% CI, 1.03-1.79), lack of diabetes (OR, 0.74; 95% CI, 0.56-0.96), and body mass index (BMI) (OR for 5-unit increase, 1.29; 95% CI, 1.16-1.43). On multivariable analysis of the 1,117 patients with baseline NT-proBNP data, independent predictors of AF were older age (OR for 10-year increase, 1.34; 95% CI, 1.08-1.66) and NT-proBNP (OR for 1-unit increase on log scale, 1.74; 95% CI, 1.40-2.16). The annualized rate of recurrent stroke was higher for patients who had AF detected than for those who did not (hazard ratio, 1.75; 95% CI, 1.30-2.35).


In this ESUS cohort, the most important predictors of detection of AF were older age, hypertension, lack of diabetes, higher BMI, and higher NT-proBNP.


The results of this secondary analysis of RE-SPECT ESUS suggest that ESUS patients with older age, hypertension, elevated BMI, and/or elevated NT-proBNP levels may represent a subgroup who would benefit from enhanced workup for AF detection, such as prolonged external cardiac rhythm monitoring or loop recorder implantation. However, these data should be considered hypothesis-generating and not conclusive, as they are vulnerable to detection bias. For instance, older patients in the study may have had more extensive workup for AF, thus leading to greater detection of AF.

Clinical Topics: Anticoagulation Management, Arrhythmias and Clinical EP, Cardiovascular Care Team, Geriatric Cardiology, Prevention, Anticoagulation Management and Atrial Fibrillation, Implantable Devices, SCD/Ventricular Arrhythmias, Atrial Fibrillation/Supraventricular Arrhythmias, Hypertension

Keywords: Aged, Anticoagulants, Aspirin, Arrhythmias, Cardiac, Atrial Appendage, Atrial Fibrillation, Body Mass Index, Dabigatran, Diabetes Mellitus, Electrocardiography, Embolic Stroke, Geriatrics, Hypertension, Natriuretic Peptide, Brain, Peptide Fragments, Secondary Prevention, Stroke

< Back to Listings