Clinical manifestations

Presentation and course

Atrial fibrillation is the most common sustained cardiac arrhythmia and the most common cause of cardioembolism. It has been estimated that atrial fibrillation accounts for a third of ischemic strokes (34). The classic sign of atrial fibrillation is an irregular pulse; however, between atrial fibrillation episodes, the pulse may be regular. Patients can have transient events of diaphoresis, lightheadedness, or palpitations. In general, cardioembolic stroke, either due to atrial fibrillation or other causes, has similar clinical findings. Patients present with severe deficits and clinical evidence of cortical or cerebellar dysfunction, eg, sudden onset of language deficits, apraxia, unilateral weakness, sensory loss, dyscoordination, ataxia, or hemianopia (03). Other stroke subtypes, however, including large-vessel thromboembolism and lacunar infarction, can present similarly. Thus, the neurologic examination has limited utility in making the diagnosis of atrial fibrillation-associated stroke. The concomitant occurrence of cerebral and systemic embolism is suggestive of cardioembolism. Brain imaging commonly shows cortical ischemia or concurrent infarction in multiple vascular territories. In the acute setting, angiographic studies typically reveal the abrupt cutoff of large vessels, such as the terminal internal carotid artery or the trunk of the middle cerebral artery, with or without evidence of atherosclerotic disease of the large cerebral arteries. Diagnosis of atrial fibrillation-related stroke is supported by the identification of abnormal cardiac rhythm on cardiac monitoring.

Prognosis and complications

Patients with atrial fibrillation have an increased risk of stroke ranging from less than 2% to more than 10% per year. The risk of stroke increases with different traditional vascular risk factors, including age, sex, hypertension, diabetes mellitus, and previous stroke or transient ischemic attack. The deficits in cardioembolism are usually more disabling than those seen in other stroke subtypes. As an example, the proportion of patients with increased disability, defined by a modified Rankin Scale score of 3 and above, at 6 months is 66% for cardioembolism, 32% for large artery occlusive disease, and 23% for lacunar infarction (62). Due to the increased disability, atrial fibrillation patients are more likely to develop severe complications, including pneumonia, sepsis, infections, or systemic embolism. The 1-year mortality rate increases from 22.6% in nonatrial fibrillation-related strokes to approximately 40% for patients with atrial fibrillation (38).

Biological basis

Genomic data in atrial fibrillation come from linkage analysis and association studies (60). Both common and rare genetic variants have been implicated in development of atrial fibrillation. Common variants have a minor allele frequency higher than 1% and increase modestly the risk of atrial fibrillation. The majority of these mutations are located within intronic and intergenic regions where they can alter the function of promoters and enhancers of the nearby genes (71; 74). Single nucleotide polymorphism (SNP) associated with atrial fibrillation were identified in the PITX2, KCNN3, PRRX1, CAV1, C9orf3, SYNPO2L, SYNE2, HCN4, and ZFHX3 genes.

In comparison, rare genetic variants have a minor allele frequency of less than 1% but are highly predictive of atrial fibrillation. These mutations are associated with gain or loss of function of genes encoding for cardiac gap junctions, signaling molecules, or sodium or potassium channels. An identified rare variant associated with loss of function of the TTN gene has been associated with earlier onset atrial fibrillation in patients younger than 30 years and leads to an earlier diagnosis (average of 5 years) than those without this mutation (19).

These genetic markers can identify individuals who have an increased risk of incidental atrial fibrillation and stroke even after adjustment for confounders (90). Data obtained in observational studies indicate that this genetic information can improve the accuracy of stroke risk stratification tools commonly used in clinical practice, such as the CHADS₂ and CHA₂DS₂-VASc scores (discussed below). However, gene sequencing data are not usually available for decision making, and its predictive value has not been well validated in mixed cohorts.

Etiology and pathogenesis

5.1 Pathogenesis of thromboembolism in atrial fibrillation. Most infarctions in patients with atrial fibrillation are believed to be caused by emboli originating in the left atrium, particularly the left atrial appendage. However, atrial fibrillation usually coexists with different traditional vascular risk factors, such as advanced age, arterial hypertension, diabetes, and hypercholesterolemia. This “systemic substrate” can contribute to the development of both atrial fibrillation and thromboembolism. In addition, atrial fibrillation can be seen in association with decreased left ventricular function, atherosclerosis of the aortic arch, and occlusive disease of the extra- and intracranial arteries, which are additional sources of embolism.

The detrimental effect of the systemic substrate is complemented by an “atrial substrate” or “atrial cardiomyopathy,” which is represented by the elements of the Virchow’s triad of thrombogenesis: blood stasis, endothelial injury, and procoagulability. Blood stasis can be explained by the hemodynamic changes that occur in the fibrillating atrium. Endothelial injury in atrial fibrillation is represented by endocardial denudation and fibroelastic infiltration. In addition, observational studies have shown that enhanced inflammation, hemostatic dysfunction, and abnormal fibrinolysis, all characteristic of procoagulable states, are common in atrial fibrillation (96).

The systemic substrate contributes to the development of atrial fibrillation, which, once formed, worsens atrial contractility and furthers the progression of the atrial cardiomyopathy by inducing structural remodeling. Thus, it is considered that the systemic substrate and the atrial substrate commonly coexist in atrial fibrillation patients and interact with each other, both directly and indirectly, leading to the formation of microemboli and stroke (68).

5.2 Atrial fibrillation and cryptogenic stroke. Cryptogenic stroke refers to an embolic stroke of unknown origin. Large observational studies have shown that at least 50% of the cryptogenic strokes are caused by atrial fibrillation. The formulation of this category, though practical from a taxonomic and investigational stand point, is hindered by its variable and rather broad definition. As an example, in the TOAST classification, cryptogenic stroke is characterized under three different scenarios: strokes with incomplete evaluation, lack of cause despite an extensive assessment, or stroke with multiple plausible causes (03).

Embolic stroke of undetermined source is a concept that was first introduced in 2014. Embolic stroke of undetermined source is a subset of cryptogenic stroke and refers to a nonlacunar ischemic stroke that takes place in the absence of other well-defined stroke mechanisms, such as atrial fibrillation, atherosclerosis causing 50% or greater luminal stenosis in arteries supplying the area of ischemia, intracardiac thrombus, arterial dissection, or angiitis (49). From a pragmatic standpoint, the minimal workup necessary for the diagnosis of embolic stroke of undetermined source includes transthoracic echocardiogram, 24 hours of continuous heart rhythm monitoring, and vascular imaging of both the extracranial and intracranial cerebral arteries. Both cryptogenic stroke and embolic stroke of undetermined source have radiological evidence suggestive of an embolic mechanism. However, embolic stroke of undetermined source constitutes a subgroup of patients with cryptogenic stroke that have had sufficient workup to exclude common stroke etiologies.

Cryptogenic stroke is responsible for 30% of strokes and embolic strokes of undetermined source for 15% of them. Paroxysmal atrial fibrillation is typically suspected in cryptogenic stroke or embolic stroke of undetermined source; however, observational studies demonstrate that these stroke subtypes have distinct characteristics. Patients with embolic stroke of undetermined source are usually younger, have milder stroke severity, and have smaller emboli than atrial fibrillation patients (48). Also, in the Oxford Vascular Study embolic stroke of undetermined source and cryptogenic stroke patients had decreased prevalence of myocardial infarction, peripheral vascular disease, and asymptomatic extracranial carotid disease than patients with cardioembolism. The rate of recurrent stroke at 1 year was 2.8% for cryptogenic stroke and 35% for cardioembolic stroke. Also, the mortality at 1 year and 5 years was 6% and 25% for cryptogenic stroke and 41% and 65% for cardioembolic stroke, respectively (62).

The rate of detection of atrial fibrillation in patients with cryptogenic stroke increases with the length of cardiac monitoring, ranging from 0% to 23% (24). The EMBRACE study investigated the 30-day incidence of atrial fibrillation of at least 30 seconds detected by continuous cardiac event monitoring in patients with cryptogenic stroke (41). In this study, atrial fibrillation was found in 16.1% of the patients. In comparison, in the CRYSTAL-AF study, which utilized a subcutaneous implantable device that allowed continuous monitoring of the cardiac rhythm for up to 3 years, the rate of atrial fibrillation in cryptogenic stroke patients was 8.6% at 6 months, 12.4% at 12 months, and 30% at 36 months (83). Additional valuable information was obtained in the ASSERT study (17). In this study, patients with a recently implanted pacemaker or defibrillator with at least one risk factor for stroke and no history of atrial fibrillation were followed for approximately 2.5 years. Surprisingly, among the individuals with stroke, 27% of the patients had no evidence of atrial fibrillation in the 30 days preceding the stroke, and 16% had atrial fibrillation detected only after the stroke. In those patients with detected subclinical atrial fibrillation during the first 3 months, there was a 2.5-fold hazard of ischemic stroke and an attributed risk of 13%, suggesting that longer periods of monitoring are required to identify episodes of atrial fibrillation (51). In these three studies, a substantial proportion of the atrial fibrillation events were asymptomatic.

Together, these results indicate that the dysrhythmia that defines atrial fibrillation, although undoubtedly associated with cerebral ischemia, is not a necessary step in the pathogenesis of cryptogenic stroke or embolic stroke of undetermined source. These observations have led to the development of the concept of “atrial cardiomyopathy” or “atrial cardiopathy,” which refers to the anatomic and physiologic changes associated with atrial dysfunction that precede atrial fibrillation but may actively contribute to left atrial thromboembolism.

Atrial cardiomyopathy is defined by the presence of at least one of the following factors: (1) left atrial structural abnormalities evidenced by echocardiography or cardiac MR, such as endothelial dysfunction, fibrosis, impaired myocyte function, or chamber dilatation; (2) electrocardiographic evidence of left atrial dysfunction, including increased P-wave terminal force velocity in lead V1 (PTFV1) greater than 5000 μV·ms; or (3) increased levels of biochemical markers of cardiac dysfunction, including increased troponins or serum N-terminal pro-brain natriuretic peptide (NT-proBNP) levels greater than 250 pg/mL (59). The European Heart Rhythm Association/Consortium (EHRAS) 2017 classification is the first one to attempt to characterize atrial cardiomyopathies based on the histopathological changes seen within the atria. EHRAS includes four categories that denote progressive atrial disease. It involves either changes seen in: (1) atrial cardiomyocytes; (2) fibrotic changes; (3) a combination of both cardiomyocyte changes and fibrotic changes; or (4) noncollagen infiltration changes (43). Multiple mechanisms have been associated with the development of atrial cardiomyopathy, including: (1) hereditary muscular dystrophies cause myocyte degeneration and fatty or fibrotic changes in the atrial fibers; (2) congestive heart failure leads to atrial fibrosis and remodeling, along with calcium channel abnormalities; (3) obstructive sleep apnea is associated with slowing of cardiac conduction; (4) in the elderly there are degenerative atrial fibrotic changes; and (5) hypertension and obesity cause left atrium enlargement and P-wave abnormalities whereas diabetes and valvular diseases cause fibrotic changes and structural remodeling (43).

It has been observed that atrial cardiopathy is associated with cerebral embolism in 10% of cases, even in the absence of atrial fibrillation (101). Also, secondary analysis of the WARSS trials shows that patients with serologic evidence of atrial cardiopathy have a lower risk of recurrent stroke or death when treated with anticoagulation rather than aspirin (67).

Another entity thought to be a precursor of atrial fibrillation and embolic stroke is atrial-heart rate event. Atrial-heart rate events are brief asymptomatic episodes of atrial tachyarrhythmia and can include brief episodes of atrial fibrillation or atrial flutter, detected only during prolonged use of cardiac implanted electronic devices. In comparison, asymptomatic paroxysmal atrial fibrillation is readily detected using surface electrocardiograms. Atrial-heart rate events are considered to be a separate entity from atrial fibrillation, although presence of atrial-heart rate event longer than 5 minutes has a 6-fold increased risk of developing atrial fibrillation and a 2- to 2.5-fold increase in the risk of stroke. The overall absolute stroke risk is lower though compared with patients diagnosed with atrial fibrillation (33). There is uncertainty on best management for atrial-heart rate events. There are ongoing clinical trials that will address the potential therapeutic options for both primary and secondary stroke prevention in patients with cardiac implanted electronic devices-detected atrial-heart rate event.

Epidemiology

Atrial fibrillation is the most common sustained cardiac arrhythmia. It has been estimated that atrial fibrillation affects more than 2.7 million individuals in the United States, resulting in a prevalence of about 3%. This increases from 0.1% in patients younger than 55 years to 10% in patients older than 80 years (100). Due to the increase in life expectancy, the prevalence of atrial fibrillation is estimated to rise to 12.1 million by 2030. The total incidence of atrial fibrillation in the United States adult population was estimated at 1.2 million cases in 2010, and this is expected to increase to 2.6 million by 2030. The incidence of atrial fibrillation in males is estimated to be 20.6 per 100,000 people per year for patients aged 15 to 44 years and 1,077 per 100,000 people per year in those aged 85 years or older. In comparison, the incidence in females ranges from 6.6 per 100,000 people per year for patients aged 15 to 44 years to 1,203 per 100,000 people per year in those aged 85 years or older (13). The incidence of atrial fibrillation is higher among Caucasians. After adjustment for risk factors, Blacks (HR, 0.84; 95% CI, 0.82 to 0.85; P< 0.001) and Hispanics (HR, 0.78; 95% CI, 0.77 to 0.79; P< 0.001) have lower rates of atrial fibrillation than Whites (13).

In general, atrial fibrillation accounts for 15% to 30% of all the ischemic strokes. Owing the association of atrial fibrillation with age, the prevalence, incidence, and proportion of strokes associated with atrial fibrillation increase sharply with age. The proportion of strokes attributable to atrial fibrillation increases from 1.5% in individuals aged 50 to 59 years to 23.5% in those aged 80 to 89 years (97).

Nonvalvular atrial fibrillation is associated with a 4- to 5-fold increase in the risk of stroke across all age groups (97). The risk of stroke is even higher in valvular-associated atrial fibrillation. As an example, patients with rheumatic heart disease-associated atrial fibrillation have a 17-fold increase in the risk of stroke (98). Similarly, mitral stenosis increases the risk of stroke 20 times over nonatrial fibrillation patients.

Sex- and racial- or ethnic-related disparities have been described in atrial fibrillation-associated stroke. Females have a 2-fold stroke risk compared with males. In addition, Blacks (HR, 1.46; 95% CI, 1.38 to 1.55; P< 0.001) and Hispanics (HR, 1.11; 95% CI, 1.03 to 1.18; P< 0.001) with atrial fibrillation have a higher risk of stroke than their Caucasian counterparts (13).

Prevention

Traditional vascular risk factors, including age, male sex, hypertension, congestive heart disease, diabetes mellitus, coronary artery disease, and sleep apnea, are associated with both atrial fibrillation and stroke. Five borderline or abnormal risk factors explain approximately 65% of the population-attributable risk for atrial fibrillation: hypertension (24%), BMI 25 or greater (18%), smoking (12%), cardiac disease (5%), and diabetes mellitus (4%) (55). However, the studies investigating the prevention of atrial fibrillation by using different strategies, including lifestyle interventions, glycemic control, treatment of sleep apnea, and use of statins or renin-angiotensin blockers have yield mixed results. The most current guidelines for the prevention of stroke recommend the correction of modifiable vascular risk factors (58). Antithrombotics are considered the mainstay for primary and secondary stroke prevention.

Differential diagnosis

Although the majority of strokes in patients with atrial fibrillation are believed to originate in the left atrium, other potential causes of embolic stroke, such as carotid artery stenosis or aortic arch plaque may coexist in these patients. About half of the patients with atrial fibrillation have hypertension, which predisposes to small-vessel occlusive disease (06). Of major clinical importance is the association of atrial fibrillation and aortic arch atheroma. The Stroke Prevention in Atrial Fibrillation investigators reported a series of 382 patients with high-risk nonvalvular atrial fibrillation, of whom 35% had complex aortic plaques. Patients with complex plaques and atrial fibrillation had a significantly higher stroke risk per year (12% to 20%) compared to patients with atrial fibrillation alone (1.2%), suggesting that aortic arch atheroma is an important additional source of emboli (09). Complex plaques are characterized by surface abnormalities (ulcerations, mobile components) and increased plaque thickness of at least 4 mm. Descending aorta seems to be preferentially involved in cerebral embolism whereas complex plaques found in the ascending and transverse portions of the aorta appear to be involved in both cerebral and peripheral embolic events. Complex aortic plaques are also associated with a higher prevalence of left atrial abnormalities or endocardial abnormalities in patients with atrial fibrillation, suggesting a possible synergistic effect (15).

Presence of carotid atherosclerosis or plaque is another important competing mechanism for stroke in patients with atrial fibrillation. The prevalence of carotid atherosclerosis or plaque in patients with atrial fibrillation increases with age, from 38% in patients under 57-years-old (12) to 65% in patients older than 75 (11); patients who are started on anticoagulation for atrial fibrillation and have carotid atherosclerosis changes of carotid plaque of at least 50% and have a 40% nonsignificant increase in risk of stroke or transient ischemic attack compared with patients without carotid plaque (11). For patients not on anticoagulation, presence of carotid changes is associated with 56% significant increased risk of stroke, even after adjustment for CHA₂DS₂-VASc score (12).

Diagnostic workup

The aim of the diagnostic workup in patients presenting with ischemic stroke and atrial fibrillation is to exclude coexistent causes of stroke that may affect management. Carotid Doppler in addition to transcranial Doppler or other non-invasive angiographic studies, such as magnetic resonance angiography or computed tomography angiography, should be performed to rule out large artery occlusive disease. Echocardiography is recommended to assess whether cardiac disease is present or absent, in particular, valvular heart disease (eg, mitral stenosis), decreased left ventricular ejection function, left ventricular hypertrophy, or left atrial dilatation. In addition, echocardiography serves in the search for other potential sources of cardiac embolism, such as cardiac tumor, interatrial septal aneurysm, and patent foramen ovale.

Transesophageal echocardiography is clearly superior to transthoracic echocardiography for the detection of left atrial thrombi (particularly in the appendage) and is considered the gold standard in assessing aortic arch atherosclerosis. In addition, transesophageal echocardiography allows depiction of spontaneous echo contrast in the atrium and reduced left atrial appendage peak flow velocities, both being important features independently associated with increased thromboembolic risk in patients with atrial fibrillation (103). Despite the diagnostic advantages of transesophageal echocardiography, the use of this relatively invasive, albeit safe, procedure may have little impact on treatment decisions except for selected cases because anticoagulation is the treatment of choice in stroke patients with atrial fibrillation.

The standard of care for identifying atrial fibrillation after an ischemic stroke involves use of prolonged cardiac monitoring. The yield of cardiac monitoring in detection of atrial fibrillation increases with the recording duration. The rate of atrial fibrillation in stroke patients is approximately 5% after 24 hours of monitoring, 6.4% after 48 hours, and 13% after 1 week of monitoring (79). The STROKE-AF trial enrolled 496 patients who had an implantable cardiac monitor inserted within 10 days of an index stroke, with a primary outcome measure of incident atrial fibrillation lasting more than 30 seconds through 12 months. Atrial fibrillation detection at 12 months was significantly higher in the implantable cardiac monitor group versus the control group (27 patients [12.1%] vs 4 patients [1.8%]; hazard ratio, 7.4 [95% CI, 2.6-21.3]; P <  .001) (14). The risk of stroke seems to increase in proportion to the atrial fibrillation burden. The rate of thromboembolic events is 1.49% per year in patients with paroxysmal atrial fibrillation, 1.95% in those with permanent atrial fibrillation, and 1.83% in persistent atrial fibrillation (63). The KP-RHYTHM study analyzed a cohort of 1965 adults with paroxysmal atrial fibrillation not on anticoagulation. Higher atrial fibrillation burden (defined as greater than 11.4% of time spent in atrial fibrillation or atrial flutter during 14 days of continuous cardiac monitoring) was associated with a 3-fold increase in the risk of thromboembolic events after adjusting for known stroke risk factors (42).

Management

14.1 Risk stratification. Guidelines for the prevention of stroke in patients with nonvalvular atrial fibrillation recommend the use of risk stratification tools to guide management. The scores more commonly used in clinical practice are the CHADS₂ and the CHA₂DS₂-VASc scores. The CHADS₂ score (Table 1a) incorporates five conditions: (1) recent congestive heart failure, (2) hypertension, (3) age of 75 years or older, (4) diabetes, and (5) prior stroke or transient ischemic attack. Each condition is given 1 point except for the composite of stroke or transient ischemic attach, which receives 2 points.

Table 1a. CHADS2 Score Quantification of Stroke Risk for Patients with Non-Valvular Atrial Fibrillation

CHADS2 score	Number of patients (n=1733)	Number of stroke (n=94)	NRAF crude stroke rate per 100 patient-years	NRAF adjusted stroke rate (95% CI)
0	120	2	1.2	1.9 (1.2 to 3.0)
1	463	17	2.8	2.8 (2.0 to 3.8)
2	523	23	3.6	4.0 (3.1 to 5.1)
3	337	25	6.4	5.9 (4.6 to 7.3)
4	220	19	8.0	8.5 (6.3 to 11.1)
5	65	6	7.7	12.5 (8.2 to 17.5)
6	5	2	44.0	18.2 (10.5 to 27.4)
NRAF indicates National Registry of Atrial Fibrillation. From (36).

One of the drawbacks of the CHADS₂ score is that almost 30% of the general population falls in the undetermined category, raising the question of whether these patients should be treated with anticoagulants or not. The CHA₂DS₂-VASc score (Table 1b), which is an extension of the CHADS₂ score, incorporates presence of vascular disease, age 65 to 74, and female sex as prognostic factors for stroke.

Table 1b. CHA2DS2-VASc Score

Risk Factor	Points
C	Congestive heart failure or LV ejection fraction 40% or less
H	Hypertension
A2	Age 75 years or older
D	Diabetes mellitus
S2	Prior Stroke, transient ischemic attack, or thromboembolism
V	Vascular disease (prior MI, peripheral artery disease, or aortic plaque)
A	Age 65 to 74 years
Sc	Sex category (female)
From (66).

The 2018 American College of Chest Physicians guidelines and 2019 American College of Cardiology/American Heart Association clinical practice guidelines recommend:

	• Men with a CHA2DS2-VASc score of 0 and women with a score of 1 are considered low stroke risk; anticoagulation therapy may be omitted in this group.
	• For patients with a CHA2DS2-VASc score of 1 (except when the patient scores 1 point due to sex only), oral anticoagulation may be considered.
	• For men with CHA2DS2-VASc score of 2 or higher and women with a CHA2DS2-VASc score of 3 or higher, oral anticoagulation is recommended in favor of single or combined antiplatelet therapy (64; 58).

The CHA₂DS₂-VASc score is superior to the CHADS₂ score in discriminating patients at low risk. The rate of patients in the general population who fall in the undetermined risk category using the CHA₂DS₂-VASc score is only 10% (102).

Although the CHA₂DS₂-VASc score is widely used in clinical practice, a systematic review of 19 validating studies found high heterogenicity for the assessment of stroke risks, especially for low to intermediate risk groups (92). This could be attributed, at least in part, to the inclusion of different populations. Also, all the risk stratification scales have well known shortcomings. For example:

	• Several risk factors in the CHA2DS2-VASc score are given the same weight; however, their contribution to stroke may be different. History of hypertension, for example, is associated with a higher risk of ischemic stroke than diabetes (81).
	• Female sex, in the absence of other vascular risk factors, does not seem to pose an increased risk of stroke. Thus, the risk of stroke for a patient with a score of 1 due to female sex is different than the risk of a patient with score of 1 due to hypertension.
	• Age is a continuous variable and establishing strict cutoffs may not accurately determine stroke risk.
	• The CHA2DS2-VASc score does not take into account the vascular risk factor severity or how well controlled this is.
	• Risk stratification tools do not differentiate between the type and burden of atrial fibrillation (paroxysmal versus persistent or permanent).

Despite these limitations, the CHA₂DS₂-VASc score remains a fast and easy to use clinical risk stratification tool that reliably identifies patients at high risk who will benefit from initiating anticoagulation.

Table 1c. CHA2DS2-VASc Score Quantification of Stroke/TE Risk for Patients with Non-Valvular Atrial Fibrillation

CHA2DS2-VASc score	Number of patients (n=1084)	Number of stroke/TE (n=25)	TE rate during 1y (95% CI)	Adjusted TE rate for aspirin prescription
0	103	0	0% (0-0)	0%
1	162	1	0.6% (0.0-3.4)	0.7%
2	184	3	1.6% (0.3-4.7)	1.9%
3	203	8	3.9% (1.7-7.6)	4.7%
4	208	4	1.9% (0.5-4.9)	2.3%
5	95	3	3.2% (0.7-9.0)	3.9%
6	57	2	3.6% (0.4-12.3)	4.5%
7	25	2	8.0% (1.0-26.0)	10.1%
8	9	1	11.1% (0.3-48.3)	14.2%
9	1	1	100% (2.5-100)	100%
TE= thromboembolic events, including ischemic stroke, peripheral embolism, or pulmonary embolism. From (66).

14.2 Vitamin K antagonists. In the late 1980s and early 1990s, six trials compared vitamin K antagonist to placebo (05; 06). Adjusted-dose oral anticoagulation with a target international normalized ratio (INR) of 2.0 to 3.0 was associated with a 60% reduction in the risk of stoke compared to placebo (47; 65). This reduction was similar for both primary and secondary prevention of stroke.

The effect of aspirin and warfarin was investigated in the SPAF Trials I and II. In these studies, atrial fibrillation patients were randomized to warfarin or aspirin (325 mg per day). Patients with systolic hypertension greater than 160 mmHg, impaired left ventricular function, history of prior thromboembolism, or women aged 75 years or older receiving aspirin had a disproportionately elevated risk of stroke that could be as high as 6% per year (89). In the SPAF II trial, the combined use of fixed-dose warfarin (mean daily dose = 2.1 mg) with aspirin (325 mg per day) was evaluated as an alternative therapy to adjusted-dose warfarin with a target INR of 2.0 to 3.0 in patients with at least one risk factor for stroke (08). This trial was stopped early due to an elevated rate of embolism in patients treated with the combination therapy (7.9% per year) as compared to those on adjusted-dose warfarin (1.9% per year). A meta-analysis of randomized trials comparing vitamin K antagonist monotherapy and vitamin K antagonist plus aspirin with the same target INR showed an increased risk of bleeding in the combined therapy arm (OR=1.43, 95% CI 1.00 to 2.02) (26).

Based on these studies, vitamin K antagonists became the standard of care for the prevention of stroke or systemic embolism in patients with atrial fibrillation. Despite its proven efficacy in secondary prevention of stroke, anticoagulation therapy is not initiated in a substantial number of patients, especially in the elderly because of contraindications or physician preconceived ideas (36). According to a national registry, approximately only half of the patients with atrial fibrillation in the United States receive evidence-based proven preventative treatments (53). Predictors for not receiving these therapies include frailty, comorbid illness, and geographic region. Two of the most common causes for discontinuation of warfarin are hospitalization due to bleeding and recent fall.

14.3 Nonvitamin K antagonists or novel oral anticoagulants. The use of vitamin K antagonists is hindered by the interaction of these agents with other drugs, dietary restrictions, and variable therapeutic effect that requires frequent blood analysis. These limitations fueled the investigation for new anticoagulant agents with more favorable pharmacokinetic and pharmacodynamics profiles. The SPORTIF III and IV studies compared the safety and efficacy of the thrombin inhibitor ximelagatran (36 mg twice daily) to dose-adjusted warfarin (72; 04). Both studies concurred that ximelagatran is as effective as warfarin in the prevention of ischemic stroke or systemic embolism with lower risk of minor bleeding. However, concerns were raised due to the potential hepatotoxic effect of ximelagatran. The AMADEUS trial compared the long-acting parenteral factor Xa inhibitor idraparinux to warfarin. In this study, the idraparinux arm had lower rates of stroke and systemic embolism than warfarin (idraparinux 0.9% vs. warfarin 1.3%, p=0.007) (16). However, a significantly higher rate of bleeding was observed in the idraparinux arm, particularly in patients with advanced age and renal insufficiency.

Novel oral anticoagulants are non-vitamin K antagonists that have been approved by the United States Food and Drug Administration as alternatives to warfarin in the prevention of stroke and systemic embolism in patients with nonvalvular atrial fibrillation. The anticoagulant effect of these agents is exerted by either direct inhibition of thrombin (dabigatran) or factor Xa (apixaban, rivaroxaban and edoxaban). In general, these agents are considered noninferior to warfarin with the advantage of having a lower rate of intracerebral hemorrhage. In addition, they are administered as a fixed dose, do not require laboratory monitoring, and have very limited drug interactions (27). The rates of embolism and cerebral hemorrhage observed with each of these agents are shown in Table 2.

Table 2. Novel Oral Anticoagulants

	RE-LY	ROCKET-AF	ARISTOTEL	ENGAGE AF-TIMI
Drug Dose	Dabigatran 110 mg low-dose 150 mg high-dose	Rivaroxaban 15 mg low-dose 20 mg high-dose	Apixaban 2.5 mg low-dose 5 mg high-dose	Edoxaban 30 mg low-dose 60 mg high-dose
Mechanism of action	Direct factor IIa inhibitor	Direct factor Xa inhibitor	Direct factor Xa inhibitor	Direct factor Xa inhibitor
Time to maximum inhibition	0.5 to 2 hours	1 to 4 hours	1 to 4 hours	1 to 2 hours
Half life	12 to 17 hours	5 to 13 hours	8 to 15 hours	6 to 11 hours
Renal excretion	85%	66%	27%	50%
Metabolic drug interactions	P-gp inhibitors: reduce dose or avoid* P-gp inducers: avoid	CYP3A4 and P-gp: avoid CYP3A4 and P-gp inducers: caution with use	CYP3A4 and P-gp: avoid CYP3A4 and P-gp inducers: caution with use	P-gp inhibitors: reduce dose P-gp inducers: avoid
Efficacy
Stroke or systemic embolism (% per year)	1.7% warfarin 1.5% low dose 1.1% high dose	2.4% warfarin 2.3% rivaroxaban	1.6% warfarin 1.3% apixaban	1.5% warfarin 1.6% low-dose 1.2% high-dose
Safety
Major bleeding (% per year)	3.6% warfarin 2.9% low-dose 3.3% high-dose	3.4% warfarin 3.6% rivaroxaban	3.1% warfarin 2.1% apixaban	3.4% warfarin 1.6% low-dose 2.8% high-dose
Intracranial hemorrhage (% per year)	0.7% warfarin 0.2% low-dose 0.3% high-dose	0.7% warfarin 0.5% rivaroxaban	0.8% warfarin 0.3% apixaban	0.5% warfarin 0.2% low-dose 0.3% high-dose
Myocardial infarction (% per year)	0.6% warfarin 0.8% low-dose 0.8% high-dose	1.1% warfarin 0.9% rivaroxaban	0.6% warfarin 0.5% apixaban	0.8% warfarin 0.9% low-dose 0.7% high-dose
* P-gp=P-glycoprotein (reduce dose with concomitant verapamil dose; avoid with dronedarone) From (21; 75; 44; 40).

The AVERROES trial compared apixaban to aspirin (81- to 324 mg/day) in atrial fibrillation patients who had failed vitamin K antagonist treatment or were deemed unsuitable to be treated with warfarin. This study was terminated early as apixaban proved its superiority to aspirin for the reduction of stroke or systemic thromboembolism (HR 0.45, 95% CI 0.32 to 0.62; P< 0.001). The rate of hemorrhagic complications was similar between both groups (0.4% per year) (20).

It should be emphasized that the studies that investigated the efficacy of novel oral anticoagulants in atrial fibrillation excluded patients with mechanical heart valves or hemodynamically significant mitral stenosis. Based on subgroup analysis of randomized trials it was suggested that novel oral anticoagulants can potentially be used in atrial fibrillation patients with mild aortic stenosis, aortic regurgitation, and mitral regurgitation (73). However, the RE-ALIGN study showed that dabigatran was associated with an excess risk of thromboembolic events and hemorrhagic complications compared with warfarin in atrial fibrillation patients with mechanical heart valves (32).

The 2019 American College of Cardiology/American Heart Association Clinical Practice Guidelines and the Heart Rhythm Society Guideline for the Management of Patients with Atrial Fibrillation recommend using novel oral anticoagulants over warfarin in eligible patients with atrial fibrillation, except for the patients with moderate to severe mitral stenosis or mechanical heart valve, where warfarin remains the drug of choice (58). These updated guidelines are in line with the updated 2021 European Heart Rhythm Association Practical guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation, which states that novel oral anticoagulants can be safely used in patients with native valvular stenosis or valvular insufficiencies as well as in patients with bioprosthetic valves. The European guidelines also recommend against using novel oral anticoagulants in patients with atrial fibrillation and mitral valve replacement due to rheumatic heart disease; for these patients, the atria usually remain enlarged, even post valve replacement, and warfarin remains the treatment of choice (87). The term “non-valvular” atrial fibrillation has been removed from clinical practice and denotes atrial fibrillation associated with all native valvular stenosis or valvular insufficiencies and bioprosthetic valves. Valvular atrial fibrillation refers to atrial fibrillation due to mechanical heart valve or moderate to severe mitral stenosis (87).

Understanding that undiagnosed atrial fibrillation may be the culprit of stroke in individuals with cryptogenic ischemia, two large randomized studies investigated the beneficial effect of novel oral anticoagulants for the prevention of recurrent ischemic strokes in patients with embolic stroke of undetermined source. NAVIGATE-ESUS (rivaroxaban for stroke prevention after embolic stroke of undetermined source) was a randomized international trial that evaluated the efficacy of rivaroxaban 15 mg daily versus 100 mg aspirin daily for preventing recurrent strokes or systemic embolism in patients with recent embolic stroke of undetermined source. The trial was terminated early due to increased risk of bleeding seen with rivaroxaban (1.8% annual rate) compared to aspiring (0.7% annual rate). The trial also failed to demonstrate benefit of using rivaroxaban for preventing recurrent ischemic strokes (4.7% annual rate of recurrent strokes for both rivaroxaban and aspirin groups) (50). RE-SPECT ESUS (design of randomized, double-blind, evaluation in secondary stroke prevention comparing the efficacy and safety of the oral thrombin inhibitor dabigatran etexilate vs. acetylsalicylic acid in patients with embolic stroke of undetermined source) investigated the efficacy of dabigatran 150 mg twice daily versus aspirin 100 mg daily in the prevention of recurrent ischemic strokes after embolic stroke of undetermined source. The annual rate of recurrent stroke was similar in patients treated with dabigatran (4.1%) and aspirin (4.8%). The rates of major bleeding were similar between both groups (1.7% annual rate for dabigatran vs. 1.4% annual rate for aspirin) (28). Based on these trials, the use of anticoagulation in unselected cases of embolic stroke of undetermined source is not recommended. The ongoing studies ARCADIA may shed light on whether anticoagulation may have a role for stroke prevention in embolic stroke of undetermined source patients with atrial cardiopathy.

14.4 Antiplatelet agents. The Japan Atrial Fibrillation Stroke Trial was a prospective study that compared the effect of aspirin at lower doses (150 to 200 mg/day) to placebo for the prevention of cardiovascular events (death, strokes, transient ischemic attacks) in patients with atrial fibrillation. The results of this study confirmed that the effect of aspirin was comparable to placebo (3.1% events per year in the aspirin group vs. 2.4% events per year in the placebo group). In addition, there was a slightly higher risk of bleeding in the aspirin group (1.6% vs. 0.4%) (84).

The ACTIVE W trial compared the efficacy of combined antiplatelet therapy (aspirin 75 to 100 mg and clopidogrel 75 mg) to warfarin in high-risk patients with atrial fibrillation. The study was stopped prematurely when patients in the antiplatelet group had more cardiovascular events than those in the vitamin K antagonist group (RR=1.44; 95% CI 1.18–1.76; P=0.0003) (02). Combined antiplatelet-anticoagulant therapy may be indicated for patients with atrial fibrillation who also have acute coronary syndromes or those who undergo percutaneous coronary intervention with or without stenting. The ACTIVE A trial randomized patients to either clopidogrel plus aspirin versus aspirin and placebo. Although the dual antiplatelet group had fewer cardioembolic events (3.3% vs. 2.4% per year), the aspirin plus clopidogrel group had an increased risk of both major extracranial and intracranial bleeding compared with the aspirin group alone with a null net beneficial effect (01).

14.5 Initiation of anticoagulation after acute stroke. Timing to initiate anticoagulation therapy in atrial fibrillation patients with acute ischemic stroke is controversial. The potential benefits of preventing early stroke recurrence may be offset by an increased risk of hemorrhagic conversion (56). Also, the risk of early stroke recurrence in atrial fibrillation patients is smaller than previously thought and estimated at approximately 5% during the 2 to 4 weeks after initial stroke.

In most cases, anticoagulation can be safely resumed 14 days post stroke (05). However, different factors can influence this decision, including size of stroke or presence of hemorrhagic conversion. For patients at high risk of recurrent cardiogenic embolism, including those with valvular heart disease, intracardiac thrombus, or congestive heart failure, early anticoagulation is generally advised. In patients with large infarcts, uncontrolled hypertension, or those at relatively low risk for early recurrence, delaying anticoagulation for several days to a week may reduce the risk of hemorrhagic transformation. Following significant hemorrhagic infarction, anticoagulation should generally be delayed.

With the widespread use of novel oral anticoagulants for stroke prevention in patients with atrial fibrillation, timing of resuming anticoagulation is controversial: the 2018 American Heart Association/American Stroke Association guidelines recommend starting anticoagulation within 4 to 14 days post ischemic stroke (78; 85). In comparison, the European Heart Rhythm Association of the European Society of Cardiology recommends using the “1-3-6-12 days” rule. This means that anticoagulation can be resumed 1 day after a transient ischemic attack, 3 days after a minor ischemic stroke (defined as NIHSS less than 8), 6 days after a mild stroke (defined as NIHSS between 8-15), and 12 days after a large stroke (NIHSS more than 15) (52).

14.6 Rhythm- and rate-control. One of the largest trials of rate versus rhythm control in atrial fibrillation, the AFFIRM trial showed that achieving a normal sinus rhythm does not significantly reduce the risk of stroke (91). The majority of strokes occurred in patients who stopped anticoagulation, suggesting that the treatment of the atrial dysrhythmia may not completely eliminate the thrombogenic substrate observed in atrial fibrillation. Also, the rhythm control group was more likely to experience episodes of undetected atrial fibrillation, reinforcing the need of continuing anticoagulation for thromboembolism prophylaxis even after achieving normal sinus rhythm (93). When congestive heart failure is associated with atrial fibrillation, rhythm control (maintenance of sinus rhythm) had similar rates of ischemic stroke as rate-control (ventricular rate control) (82).

14.7 Nonpharmacological treatments. Besides anticoagulation, other nonpharmacological strategies are emerging for the prevention of stroke in atrial fibrillation patients. The traditional surgical treatment of atrial fibrillation includes catheter ablation therapies (23). This approach may cure atrial fibrillation in 70% to 80% of patients. However, based on the experience obtained with antiarrhythmic drugs, guidelines recommend against interruption of anticoagulation in high-risk patients with atrial fibrillation undergoing catheter ablation or pharmacologic treatment with antiarrhythmic drugs.

The PROTECT atrial fibrillation trial (WATCHMAN) randomized atrial fibrillation patients with high stroke risk and contraindications to long-term therapy with warfarin versus percutaneous closure of the left atrium appendage. Patients assigned to percutaneous closure of the left atrium appendage had to be able to tolerate warfarin for 45 days after the procedure, at which time, warfarin therapy was discontinued. The primary endpoint was a composite of stroke, systemic embolism, or cardiovascular or unexplained death. The surgical procedure was noninferior to warfarin and had a 3 per 100 patient-year efficacy rate compared with 4.9 per 100 patient-year efficacy in the warfarin group (54). In the extended period of 4 years, the device was associated with a significant reduction in primary outcome compared with warfarin (8.4% vs. 13.9%) (80). However, this benefit was largely driven by a reduction in the rate of hemorrhagic stroke with no effect in the occurrence of ischemic stroke. In addition, it is unclear how this device will perform when compared against novel oral anticoagulants. Thus, uncertainties persist in relation to the beneficial effect of left atrium appendage closure and stroke prevention in atrial fibrillation.

Outcomes

The selection of antithrombotic agents in atrial fibrillation depends on the predicted risk of thromboembolism and hemorrhagic complications. When anticoagulation is warranted, novel oral anticoagulants are usually preferred over warfarin given the same proven efficacy as vitamin K antagonists and a lower risk for bleeding. Several scales exist to quantify bleeding risk: (1) HEMORR2HAGES, (2) ATRIA and (3) HAS-BLED. The 2012 AMADEUS trial compared the predictive abilities of these scales for clinically relevant intracranial bleeding and HAS-BLED demonstrated superiority over the other scales (10). Both HEMORR2HAGES and HAS-BLED scores are being used in clinical practice.

The HEMORR2HAGES score is calculated by adding 1 point for each of the following factors: hepatic or renal disease, ethanol abuse, malignancy, old age (older than 75 years), reduced platelet counts or platelet dysfunction, uncontrolled hypertension, anemia, genetic factors, elevated fall risk, or stroke; 2 points were added for rebleeding (37). HAS-BLED score has fewer components and adds 1 point for each of these factors: hypertension, abnormal renal and liver function (1 or 2 points), stroke, bleeding, labile INRs, elderly, and drugs or alcohol (1 or 2 points) (76).

Table 3a. Incidence of Major Bleeding Stratified by the HEMORR2HAGES Score

HEMORR2HAGES score	Number of patients	Number of bleeding	Bleeding per 100 patient-years warfarin (95% CI)
0	209	4	1.9 (0.6-4.4)
1	508	11	2.5 (1.3-4.3)
2	454	20	5.3 (3.4-8.1)
3	240	15	8.4 (4.9-13.6)
4	106	9	10.4 (5.1-18.9)
5 or greater	87	8	12.3 (5.8-23.1)
Any score	1604	67	4.9 (3.9-6.3)
Data from the National Registry of Atrial Fibrillation (37).

Table 3b. Incidence of Major Bleeding Stratified by HAS-BLED Score

HAS-BLED score	Number of patients	Number of bleeding	Bleeding per 100 patient-years (%)
0	798	9	1.13
1	1286	13	1.02
2	744	14	1.88
3	187	7	3.74
4	46	4	8.70
5	8	1	12.50
6	2	0	0.0
7	0	-	-
8	0	-	-
9	0	-	-
Any score	3 071	48	1.56
From (76).

Bleeding rates for fixed doses of warfarin versus fixed dose warfarin plus aspirin versus adjusted dose warfarin (target INR 2.0-3.0) were higher for the fixed dose warfarin or combination of fixed dose warfarin plus aspirin than adjusted warfarin alone (45). Older age (median age 77 years; range 67 to 89 years) did not significantly influence the risk of bleeding, indicating that elderly patients will benefit from anticoagulation.

The annual rate of major hemorrhage is estimated at 1% for aspirin and 3% to 4% for warfarin (07). Novel oral anticoagulants have a 30% to 50% reduced risk of major bleeding, including intracranial hemorrhage, compared to warfarin. The use of dual therapy (warfarin and clopidogrel) or triple therapy (warfarin, aspirin, and clopidogrel) increases the risk of fatal and nonfatal bleeding 1.7 to 3.7 times compared to warfarin, respectively (46). Also, the addition of aspirin to novel oral anticoagulants has an additive effect on the risk of bleeding (88).

In general, the combination of antithrombotics is not recommended for patients with atrial fibrillation. However, this may be indicated in patients with coronary artery disease or those undergoing cardiac stent procedures.

In case of emergency, the effect of warfarin can be reverted using intravenous vitamin K and prothrombin complex concentrate (35). In the case of dabigatran, the monoclonal antibody fragment idarucizumab reverses the anticoagulant effect in a complete and durable fashion within minutes of administration (77). Studies have shown the efficacy of andexanet alfa in reducing the effects of apixaban and rivaroxaban. A single bolus of andexanate reduced factor Xa inhibitor activity by 94% in the apixaban group and by 92% in the rivaroxaban group (86). ANNEXA-4 investigated patients with major bleeding associated with the use of a factor Xa inhibitor who were given a bolus plus infusion protocol of andexanet alfa (22). This study demonstrated a reduction in factor Xa activity of 92% in both the apixaban and rivaroxaban groups, along with excellent or good hemostasis at 12 hours in 82% of patients. Andexanet alfa is FDA approved for reversal of anticoagulation in patients treated with apixiban or rivaroxaban.

Special considerations

16.1 Perioperative management of antithrombotics in atrial fibrillation. The question of holding anticoagulation perioperatively and using other protective agents (eg, low molecular weight heparin) is still a matter of debate. Existing data suggest that the risk of stroke increases significantly in the first 2 weeks after stopping the anticoagulant (18). The BRIDGE trial concluded that holding warfarin for elective procedures without any bridging was noninferior to using low molecular weight heparin for arterial thromboembolism prevention (29). The risk of bleeding was reduced in patients without bridging compared with patients who were bridged (RR 0.41; 95% CI, 0.20 to 0.78). Findings are similar when novel oral anticoagulants are used: a subgroup analysis of the RE-LY trial looked at patients who interrupted dabigatran before surgery and used bridging anticoagulation versus no bridging and found no significant differences in the rates of thromboembolism or systemic emboli among groups (30). Bridging anticoagulation is not recommended for unselected cases but is advised for patients at high risk for thromboembolism (eg, mechanical heart valve, atrial fibrillation, venous thromboembolism) (31).

16.2 Secondary atrial fibrillation. Secondary atrial fibrillation refers to the cases of atrial fibrillation that take place during the course of active medical conditions or surgical procedures. Typical scenarios include infections, pulmonary embolism, hyperthyroidism, or cardiac surgery. A substantial number of patients with secondary atrial fibrillation will spontaneously convert to sinus rhythm on correction of the underlying process. This has led to the notion that secondary atrial fibrillation is a self-limited condition that carries a more benign prognosis than primary atrial fibrillation. However, there is growing evidence that challenges this concept. New onset atrial fibrillation during sepsis and perioperative atrial fibrillation has been associated with increased risk of stroke (95; 39). It is plausible that patients with secondary atrial fibrillation constitute a selected group of individuals with genetic or atrial factors that predispose them to develop atrial fibrillation. Guidelines do not give specific recommendations in terms of anticoagulation for patients with secondary atrial fibrillation (57). However, it is reasonable to consider active surveillance for recurrent atrial fibrillation and tailor the use of anticoagulation based on individual risk of thromboembolism.

16.3 Management of atrial fibrillation in renal patients. Approximately 15% to 20% of the patients with chronic kidney disease have atrial fibrillation. Subgroup analysis of the SPAF III trial revealed that adjusted-dose warfarin (target INR 2.0 to 3.0) was superior to aspirin or low-dose warfarin in preventing ischemic stroke and systemic embolism in patients with chronic kidney disease with no difference in the rate of major hemorrhage. Most of the trials investigating the effectiveness of novel oral anticoagulants in atrial fibrillation used CrCl less than 50 mL/min to introduce dose adjustments. The CrCl in atrial fibrillation patients may fluctuate over time. Thus, renal function in patients with chronic kidney disease treated with novel oral anticoagulants should be determined at baseline and monitored at least annually. The dose adjustments recommended for patients with chronic kidney disease are shown in Table 4. There are no randomized trials of novel oral anticoagulants in patients with advanced chronic kidney disease defined as creatinine clearance lower than 25 to 30 mL/min or on hemodialysis. In these particular cases, the use of warfarin with target INR of 2.0 to 3.0 is recommended.

Table 4. Dose Adjustment of Novel Oral Anticoagulants in Patients with Chronic Kidney Disease

	RE-LY	ROCKET-AF	ARISTOTEL	ENGAGE AF-TIMI
Drug	Dabigatran	Rivaroxaban	Apixaban	Edoxaban
Renal clearance	80%	35%	25%	50%
Exclusion criteria for chronic kidney disease	CrCl<30 mL/min	CrCl<30 mL/min	Serum creatinine>2.5 mg/dL or CrCl<25 mL/min	CrCl<30 mL/min
Renal function
Normal or mild impairment (CrCl 51-80 mL/min)	150 mg twice daily	20 mg once daily	5 mg once daily	60 mg (or 30 mg) once daily
Moderate impairment
(CrCl 30-50 mL/min)	150 mg twice daily	15 mg once daily	5 mg once daily*	30 mg (or 15 mg) once daily
Severe impairment
(CrCl 15-29 mL/min)	75 mg twice daily**	15 mg once daily	Not recommended	30 mg (or 15 mg) once daily
End stage chronic kidney disease (with or without dialysis)	Not recommended	Not recommended	Not recommended***	Not recommended
CrCl: creatinine clearance * Use 2.5 mg twice daily if two of the following: serum creatinine 1.5 or higher, age 80 years or older, or weight 60 kg or less. ** Dose determined based on modelling studies; however, this dose was not properly investigated in randomized trials. *** In patients with end-stage chronic kidney disease on stable hemodialysis, prescribing information suggests the use of apixaban 5 mg twice daily with dose reduction to 2.5 mg twice daily if the patient is 80 years of age or older or weight is 60 kg or less. From (21; 44; 75; 40; 57).

Pregnancy

In general, atrial fibrillation is uncommon during pregnancy (61). In cases where anticoagulation is indicated, low molecular weight heparin should be used throughout pregnancy. Warfarin therapy in the first trimester of pregnancy is associated with fetal anomalies and is, therefore, not recommended. Novel oral anticoagulants have not been investigated in pregnancy.