Assessing the Risk for Stroke in Patients With Atrial Fibrillation

Atrial fibrillation, which affects up to 46.3 million people globally and 2.7 to 6.1 million individuals in the US, may represent a modifiable risk factor for acute stroke.

According to recent estimates from the Centers for Disease Control and Prevention, stroke affects more than 795,000 individuals in the United States annually and represents the fifth leading cause of death in the country.1 Stroke is also associated with substantial long-term disability and with costs estimated at $34 billion each year.

Atrial fibrillation (AF), which affects up to 46.3 million people globally and 2.7 to 6.1 million individuals in the United States, may represent a modifiable risk factor for acute stroke.2 Studies suggest that AF accounts for 15% to 20% of all strokes occurring in the United States, and higher mortality rates have been observed in patients with vs without AF-related stroke.2

“Despite numerous mechanisms, stroke in AF is preventable with anticoagulation and alternatively with the use of left atrial appendage closure (LAAC) devices in patients who are not candidates for long-term anticoagulation [therapy],” noted the authors of an article published in Current Cardiology Reports.2 “Thus, it is crucial to risk stratify patients who have AF to reduce the incidence of cardioembolic stroke.”

Clinicians typically use the CHA2DS2-VASc risk score, developed based on evidence, to assess the risk for stroke and guide anticoagulation treatment in patients with AF. In US and European professional guidelines, oral anticoagulant drugs are recommended in men and women with a CHA2DS2-VASc score ≥2 and ≥3, respectively. However, an overreliance on this measure may not adequately capture stroke risk in certain patient groups.

The authors note a number of limitations of the CHA2DS2-VASc score, including its lack of validation in ethnically diverse populations. In addition, the score does not distinguish between various types of AF, despite evidence indicating higher rates of stroke in patients with persistent- vs paroxysmal-type AF. This approach also “fails to consider not only the size of the [left atrium] and shape of the [left atrial appendage], but also — despite being the major physiological mechanism for stroke in AF — ignores thrombus presence in the [left atrium] altogether,” note the study authors.2

They describe a range of novel factors that may further influence a patient’s stroke risk beyond those accounted for in the CHA2DS2-VASc score, as highlighted below.

Clinical Risk Factors

Obstructive sleep apnea (OSA). Numerous studies indicate an independent association between OSA and stroke. OSA affects up to 70% of patients with stroke vs 4% of the general population, and the  severity of OSA is linked with the incidence and severity of stroke.2,3

An independent association has also been found between OSA and AF, with a higher prevalence of AF noted in individuals with OSA. Patients with comorbid OSA and AF were found to be more likely to experience cardioembolic strokes, and anticoagulation therapy has been shown to reduce this risk. “The underlying pathology of AF could worsen in the presence of OSA and other associated comorbidities such as high blood pressure and other cardiac myopathies,” noted the authors of a review published in 2018.3

Renal dysfunction. Renal failure was found to be associated with a composite thromboembolism endpoint of stroke, transient ischemic attack, or systemic embolism in patients with nonvalvular AF (n=90,490) who were not on anticoagulation therapy.4 A greater risk for stroke/systemic embolism was found in patients with AF and creatinine clearance <30 mL/min vs ≥50 mL/min after adjusting for risk factors (hazard ratio, 1.68; 95% CI, 1.04-2.65; P =.04).2

“Given these factors, adding renal failure to current models may improve their ability to risk-stratify patients,” noted investigators.2

Laboratory Markers

Brain natriuretic peptide (BNP)/Nterminal-pro BNP (NT-proBNP). Levels of BNP (ie, <13 pg/mL vs <34 pg/mL) were found to predict stroke.5 In another study, NT-proBNP was identified as a marker for cardioembolic stroke in 576 individuals with first-time ischemic stroke, with the risk for myocardial infarction associated with levels of NT-proBNP.6  

Imaging Markers

Left atrium (LA)/left atrial appendage (LAA) thrombus. LA/LAA thrombus has been linked with a 10% to 33% increased risk for stroke and death over 1 to 3 years in patients with AF.7 “Risk factors for LAA thrombus include a larger LA as well as a higher LAA position, reduced left ventricular ejection fraction, increased left ventricular end-diastolic volume, degree of spontaneous echocardiographic contrast, increased LAA volume, and LAA morphology,” note study authors. 

Other imaging-based criteria indicating a risk for stroke in patients with AF include left atrial spontaneous echo contrast and coronary artery calcium score.

Ethnicity and Stroke Risk in AF

Although the rates of AF are lower in blacks vs whites, the incidence of stroke is higher in blacks vs whites with AF. In a study of 517,941 patients with AF aged >65 years, the risk for stroke was found to be higher in black vs white patients (median follow-up, 20.3 months; hazard ratio, 1.66; 95% CI, 1.57-1.75; P <.001).8 In a subsequent study, the addition of 1 point to the CHA2DS2-VASc score to account for African American ethnicity to the CHA2DS2-VASc score was found to improve the model-fit significantly.9

Promising new models for stroke risk stratification in patients with AF are currently under investigation. These include the TIMI-AF, ATRIA, and GARFIELD-AF scores. “Ultimately, stroke risk in [AF] extends far beyond the CHA2DS2-VASc score, with much left to learn and refine on the topic,” the authors concluded.2

To further explore these issues, Cardiology Advisor interviewed study co-author Pooja S Jagadish, MD, a third-year internal medicine resident at the University of Tennessee Health Science Center in Memphis.

Cardiology Advisor: What are the risks of overrelying on the CHA2DS2-VASc score to assess stroke risk in patients with AF?

Dr Jagadish: By relying solely on a score to risk-stratify patients, one risks missing those who may not be completely represented by the model. For example, adding an extra point for African American ethnicity created a better model-fit for this population.9 By using the CHA2DS2-VASc score, which does not account for race, black patients may not be adequately covered by anticoagulants.

Cardiology Advisor: What are examples of other parameters that may help to predict stroke in patients with AF, and how can clinicians choose the best ones for each patient?

Dr Jagadish: Looking at a patient’s risk factors is critical for assessing which parameters to choose for each patient; however, CHA2DS2-VASc remains the only guideline-driven recommendation for risk-stratifying patients.

Other parameters are:

  • Echocardiographic predictors: spontaneous echo contrast, left atrial appendage thrombus, and complex aortic plaque. A transesophageal echocardiogram may be helpful in evaluating these parameters.
  • Renal dysfunction: Renal failure (CrCl < 60 mL/min or glomerular filtration rate < 60 mL/1.73 m2) has made its way into scores and may impact the anticoagulation strategy for older patients. It has yet to become part of guideline-directed therapy.
  • Race/ethnicity: Adding an extra point for African Americans aged ≥ 65 (as CHA2DS2-VASc-R) improves the model fit for CHA2DS2-VASc.

Related Articles

Cardiology Advisor: What are additional considerations for clinicians regarding this topic?

Dr Jagadish: Consideration of these risk factors must be a shared decision between physicians and patients. Adding extra points for CrCl, glomerular filtration rate, or race is not without risk, and patients must be aware that going on anticoagulation therapy raises the risk for bleeding. This is why scores such as ATRIA, GARFIELD, and the better known HAS-BLED can help physicians determine whether risks outweigh benefits for each patient. Thus, the decision to place a patient with AF on anticoagulant medication needs to be individualized.

Cardiology Advisor: What are remaining research or education needs in this area? 

Dr Jagadish: There is limited research on the topic of renal dysfunction and race and ethnicity as they affect risk stratification in AF, and they are not yet part of guideline-directed medical therapy. Educating physicians about the potential role of these risk factors will help to raise awareness and may alleviate stroke risk in those who would otherwise have inadequate anticoagulant coverage.


1. Centers for Disease Control and Prevention. Stroke facts. Updated September 6, 2017. Accessed December 2, 2019.

2. Jagadish PS, Kabra R. Stroke risk in atrial fibrillation: beyond the CHA2DS2-VASc score. Curr Cardiol Rep. 2019;21(9):95.

3. Jehan S, Farag M, Zizi F, et al. Obstructive sleep apnea and stroke. Sleep Med Disord. 2018;2(5):120-125.

4. Friberg L, Rosenqvist M, Lip GY. Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: the Swedish Atrial Fibrillation cohort study. Eur Heart J. 2012;33(12):1500-1510.

5. Sughrue T, Swiernik MA, Huang Y, Brody JP. Laboratory tests as short-term correlates of stroke. BMC Neurol. 2016;16:112.

6. Cushman M, Judd SE, Howard VJ, et al. N-terminal pro-B-type natriuretic peptide and stroke risk: the reasons for geographic and racial differences in stroke cohort. Stroke. 2014;45(6):1646-1650.

7. Vinereanu D, Lopes RD, Mulder H, et al; ARISTOTLE Investigators. Echocardiographic risk factors for stroke and outcomes in patients with atrial fibrillation anticoagulated with apixaban or warfarin. Stroke. 2017;48(12):3266-3273.

8. Kabra R, Cram P, Girotra S, Vaughan Sarrazin M. Effect of race on outcomes (stroke and death) in patients >65 years with atrial fibrillation. Am J Cardiol. 2015;116(2):230-235.

9. Kabra R, Girotra S, Vaughan Sarrazin M. Refining stroke prediction in atrial fibrillation patients by addition of African-American ethnicity to CHA2DS2-VASc score. J Am Coll Cardiol. 2016;68(5):461-470.

This article originally appeared on The Cardiology Advisor