Selecting antithrombotic therapy for patients with atrial fibrillation

Author and Disclosure Information

ABSTRACTWhen considering anticoagulant therapy for patients with atrial fibrillation, one must balance the reduction in risk of thromboembolism that this therapy offers against the risk of bleeding that it poses. The American Heart Association, American College of Cardiology, and Heart Rhythm Society updated their atrial fibrillation guidelines in 2014. This review outlines a rationale for clinical decision-making based on the new guidelines and summarizes the currently approved drugs.


  • Valvular atrial fibrillation poses a high risk of systemic embolization, particularly stroke, and nearly all patients who have valvular atrial fibrillation need anticoagulation therapy with warfarin.
  • Nonvalvular atrial fibrillation poses a somewhat lower risk. The new guidelines propose a new risk-classification scheme, called CHA2DS2-VASc; patients at very low risk of stroke may be able to forgo anticoagulation.
  • The new guidelines downplay the role of aspirin, although it is still an option in some situations.
  • Several novel oral anticoagulants have been approved in the past few years for thromboprophylaxis in patients with nonvalvular atrial fibrillation.



Antithrombotic therapy reduces the risk of systemic embolism in patients with atrial fibrillation, but one approach does not suit all patients. The decision whether to start this therapy—and which agent to use—must take into account the patient’s risk of thromboembolism as well as bleeding.

Antithrombotic therapy encompasses antiplatelet drugs such as aspirin and clopidogrel and anticoagulants such as warfarin and the target-specific oral anticoagulants (TSOACs). Oral anticoagulation is more effective than antiplatelet therapy and is preferred in all but those at lowest risk, in whom either antiplatelet therapy or no therapy is deemed adequate.

Patients with valvular atrial fibrillation, specifically those who have rheumatic mitral stenosis or a prosthetic heart valve, are at significantly higher risk of systemic embolization. Their overall risk-benefit profile is nearly always in favor of anticoagulation. But the same is not necessarily true for patients with nonvalvular atrial fibrillation.

The following discussion sets forth our rationale for clinical decision-making, based on recommendations in the 2014 guidelines from the American Heart Association, American College of Cardiology, and Heart Rhythm Society.1 The second half of this review outlines the oral anticoagulants currently available.


Atrial fibrillation is common, with an incidence that increases with age. It affects more than 10% of people over age 80 and is often associated with cardiovascular disease.2 Based on Framingham Heart Study data, a person’s lifetime risk of developing it is about 25%.3


The most serious complication of atrial fibrillation is arterial thromboembolism, of which ischemic stroke is the most common and most feared manifestation. The risk of stroke is five times higher than normal in patients with atrial fibrillation.3 More than 15% of strokes may be attributable to atrial fibrillation, and the proportion increases with age.4

The risk of thromboembolism appears to be similar in patients with clinically manifest atrial fibrillation irrespective of the type (paroxysmal, persistent, or permanent). The Stroke Prevention in Atrial Fibrillation (SPAF) study5 and the Atrial Fibrillation Clopidogrel Trial With Irbesartan for Prevention of Vascular Events (ACTIVE W)6 showed that patients who had paroxysmal atrial fibrillation and at least one risk factor for thromboembolism had stroke rates comparable to those of their counterparts who had persistent and permanent atrial fibrillation.

Subclinical atrial fibrillation may be an important cause of stroke. Clinically silent episodes can be detected by implantable electronic devices, which record episodes of atrial tachyarrhythmia (atrial high-rate events). Subclinical episodes have been detected in 10% to 28% of monitored patients who did not have a history of atrial fibrillation.7,8 Patients who have atrial high-rate events detected by implantable devices have a higher risk of future clinically manifest atrial fibrillation, thromboembolic events, or both.7–9 Yet characteristics of atrial high-rate episodes that predict risk are not well defined and warrant further investigation.


To date, thousands of patients with nonvalvular atrial fibrillation have participated in randomized clinical trials of stroke prevention. The placebo groups from these trials provide a sizable database for retrospectively identifying clinical characteristics associated with thromboembolism. The Atrial Fibrillation Investigators10 pooled data from five large trials and found that risk factors consistently associated with stroke in multivariate analysis included diabetes mellitus, hypertension, prior systemic embolism, and advanced age.

Though the risk of stroke increases with age with no lower limit, most studies identify age 65 as a threshold, with further escalating risk after age 75. Moreover, women were observed to be at higher risk in some but not all studies. These risk factors have become components of commonly used risk-stratification schemes.

Hypertrophic cardiomyopathy. Maron et al11 reported that atrial fibrillation in patients with hypertrophic cardiomyopathy was independently associated with thromboembolism. In 900 patients with hypertrophic cardiomyopathy, the prevalence of systemic embolism was 6%. Patients with hypertrophic cardiomyopathy and a thromboembolic complication were seven times more likely to have atrial fibrillation than matched counterparts free of thromboembolism. A notable subset of patients experienced a stroke or embolic event before age 50, and the authors advised that the risk of thromboembolism should be considered in patients of any age with hypertrophic cardiomyopathy and atrial fibrillation.

Olivotto et al12 similarly found patients with hypertrophic cardiomyopathy and atrial fibrillation to be at significantly greater risk of stroke (odds ratio [OR] 17.7, 95% confidence interval [CI] 4.1–75.9, P < .001).

Chronic kidney disease is also associated with a higher risk of thromboembolism in patients with atrial fibrillation. A glomerular filtration rate of 60 mL/min or less is independently and inversely predictive of risk.13,14

While patients with end-stage renal disease have been largely excluded from stroke prevention trials, Vázquez et al15 prospectively followed 190 dialysis patients for 12 months. In multivariate analysis, compared with matched controls without documented atrial fibrillation, patients receiving renal replacement therapy and having any form of atrial fibrillation were eight times more likely to have systemic embolization.


In addition to clinical factors, several imaging-based factors have been associated with stroke risk in patients with atrial fibrillation.

Complex aortic atheroma or markers of blood stasis within the left atrium, such as reduced left atrial appendage emptying flow (< 20 cm/second), dense spontaneous echo contrast, or left atrial appendage thrombus, seen on transesophageal echocardiography, were independently associated with increased systemic embolic risk in the third SPAF substudy.16 Moreover, multivariate analysis of SPAF data found both left ventricular dysfunction of any severity and increased left atrial size (diameter corrected for body surface area by M-mode > 2.5 cm/m2) to be independent predictors of thromboembolism.17

Although enlargement of the left atrium has not been incorporated into traditional risk stratification schemes, data from Osranek et al18 further implicate it as a marker of risk. The cohort was small (N = 46), but consisted of patients with lone atrial fibrillation followed for nearly 30 years. Patients with normal left atrial size enjoyed a benign course, while those with left atrial enlargement (> 32 mL/m2) at diagnosis or later during follow-up had significantly worse event-free survival (hazard ratio [HR] 4.46, 95% CI 1.56–12.74, P < .01). All embolic strokes occurred in the group with left atrial enlargement.


Several models for predicting systemic embolism risk in patients with nonvalvular atrial fibrillation have been proposed and validated.

The CHADS2 score has been the most widely applied, being simple to use.19,20 It assigns 1 point each for Congestive heart failure, Hypertension, Age 75 or older, and Diabetes, and 2 points for prior Stroke or systemic thromboembolism.

In patients with chronic nonvalvular atrial fibrillation, Gage et al19 reported that the stroke rate was lowest in those with a score of 0, with an annual adjusted stroke rate of 1.9% per year, and highest in those with the maximal possible score (ie, 6), with a rate of 18.2%. The rate increased by a factor of 1.5 with each point in the CHADS2 score.

CHA2DS2-VASc. Endorsed for use in both the American and European guidelines,1,21 CHA2DS2-VASc is an extension of CHADS2. Points are assigned as follows:

  • Congestive heart failure or left ventricular dysfunction (moderate to severe left ventricular dysfunction or recent heart failure exacerbation requiring hospitalization irrespective of ejection fraction): 1 point
  • Hypertension: 1 point
  • Age ≥ 75: 2 points; age 65–74: 1 point
  • Diabetes mellitus: 1 point
  • Stroke, transient ischemic attack, or thromboembolism: 2 points
  • Vascular disease (prior myocardial infarction, peripheral arterial disease, or aortic plaque): 1 point
  • Sex, female: 1 point
  • Maximum score: 9 points.

Low risk is defined as a score of 0 for a man or, for a woman with no other risk factors, 1. A score of 1 for a man indicates moderate risk, and a score of 2 or more is high risk. Lip et al22 found that, in untreated patients with nonvalvular atrial fibrillation, rates of stroke ranged from 0 with a score of 0 to 15.2% per year with a score of 9 points.

In a large cohort with over 11,000 patient-years of follow-up, 98% of the thromboembolic events occurred in people with a CHA2DS2-VASc score of 2 or more. Moreover, more than 99% of patients with a score of less than 2 were free of stroke and thromboembolism.23

Compared with CHADS2, CHA2DS2-VASc has superior negative predictive power

Compared with the CHADS2 score, CHA2DS2-VASc has superior negative predictive power. Of 1,084 patients from the European Heart Survey for Atrial Fibrillation, the newer scheme classified significantly fewer patients as being at either low risk (score of 0; 9% vs 20%) or intermediate risk (score of 1; 15% vs 35%).23 Though the overall rate of stroke was low, those categorized as being at low or intermediate risk by CHA2DS2-VASc had significantly fewer thromboembolic events than their counterparts according to CHADS2 (0.6% vs 3.3%).

Olesen et al24 similarly showed that in patients with a CHADS2 score of 0, reclassification by CHA2DS2-VASc yielded a range of annual stroke rates from 0.84% with a score of 0 up to 3.2% with a score of 3.


Next Article:

A 41-year-old man with abdominal pain

Related Articles