This Article Free upon publication Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Prystowsky, E. N. Articles by Wyse, D. G. Search for Related Content PubMed PubMed Citation Articles by Prystowsky, E. N. Articles by Wyse, D. G.

(Circulation. 1996;93:1262-1277.)
© 1996 American Heart Association, Inc.

Articles

Management of Patients With Atrial Fibrillation

A Statement for Healthcare Professionals From the Subcommittee on Electrocardiography and Electrophysiology, American Heart Association

Eric N. Prystowsky, MD, Chair; D. Woodrow Benson, Jr, MD, PhD; Valentin Fuster, MD, PhD; Robert G. Hart, MD; G. Neal Kay, MD; Robert J. Myerburg, MD; Gerald V. Naccarelli, MD; D. George Wyse, MD, PhD

	Executive Summary

Top Executive Summary Epidemiology Pathophysiology Clinical Presentations Approach to Treatment Restoration of Sinus Rhythm... Control of Ventricular Rate... Preventing Thromboembolism in... References

 
Atrial fibrillation (AF) is the most common sustained arrhythmia encountered in clinical practice. Its incidence increases with age and the presence of structural heart disease. It is a major cause of stroke, especially in the elderly. Although the causes are diverse, hypertension is common. Most patients experience palpitations, but fatigue, dyspnea, and dizziness are not uncommon. Patients with an uncontrolled ventricular response during AF may occasionally develop a tachycardia-induced cardiomyopathy. There are three therapeutic goals to consider for patients with AF: rate control, maintenance of sinus rhythm, and prevention of thromboembolism. The risks and benefits of each treatment must be considered for each patient.

Restoration and Maintenance of Sinus Rhythm
Several drugs effectively restore and maintain sinus rhythm in patients with AF. To date few data are available to confirm superiority of any particular drug over another for this purpose. Agents such as digitalis, verapamil, diltiazem, and ß-adrenergic blockers may be useful during AF to decrease the ventricular response that occurs over the atrioventricular (AV) node, but they rarely terminate AF. Intravenous procainamide is the treatment of choice for patients with Wolff-Parkinson-White syndrome who have a preexcited ventricular response during AF, provided they are hemodynamically stable. Patients who are unstable (eg, those with hypotension or significant heart failure) may require immediate cardioversion. Drugs selected for long-term oral therapy should be given initially in low to moderate doses and titrated upward, depending on effectiveness and side effects. Drug interactions with warfarin and digoxin should be monitored. Proarrhythmia is the most important risk associated with antiarrhythmic drug therapy. Bradyarrhythmias, especially sinus bradycardia, and ventricular tachyarrhythmias, especially torsade de pointes, can occur. Proarrhythmia often occurs during initiation of antiarrhythmic drug treatment. In patients without heart disease who have a normal baseline QT interval, ventricular proarrhythmia is relatively rare, and outpatient initiation of treatment is reasonable. However, patients with structural heart disease, especially those with a history of congestive heart failure, are at highest risk for proarrhythmia. Inpatient initiation of antiarrhythmic drug therapy is recommended for these patients. Nonpharmacological approaches to prevention of AF include surgery, atrial pacing, and endocardial catheter ablation. Too few data are available to make any specific recommendations.

Control of Ventricular Rate
In patients without ventricular preexcitation, acute rate control is most effective with intravenous verapamil, diltiazem, or ß-blockers. ß-Adrenergic blockers are especially effective in the presence of thyrotoxicosis and increased sympathetic tone. Verapamil, diltiazem, and ß-blockers are more effective than digoxin when given orally for long-term rate control and should be the initial drugs of choice. Digoxin should be considered as first-line treatment in patients with congestive heart failure secondary to impaired systolic ventricular function. Some patients may require combinations of digoxin, calcium channel blockers, and ß-adrenergic blockers to control ventricular response during AF. Intravenous procainamide is the treatment of choice if conduction is over an accessory pathway. Nonpharmacological methods to control ventricular rate include endocardial catheter ablation or modification of the AV junction and surgically induced AV block. Surgery, however, is rarely indicated. Catheter ablation is effective and is recommended in patients who have not responded to or are intolerant of drugs used for rate control.

Prevention of Thromboembolism
Patients at "low risk" may be given aspirin 325 mg/d to prevent stroke. "High-risk" patients who can safely receive anticoagulation should be treated with warfarin. For high-risk AF patients aged 75 years or younger, an International Normalized Ratio (INR) range of 2.0 to 3.0 is safe and effective; for those older than 75, close surveillance of INR levels is recommended because of the apparently greater likelihood of bleeding complications. Patients with AF who cannot safely receive anticoagulation should be given aspirin. The long-term recurrence rate for stroke is high, and warfarin anticoagulation is recommended; aspirin is an alternative therapy for those who cannot take anticoagulants. Regarding cardioversion, in patients who have AF of unknown duration or for more than 48 hours, anticoagulation should be given for 3 weeks before electrical or pharmacological cardioversion and continued for 4 weeks after cardioversion. An alternative approach involves use of intravenous heparin and subsequent transesophageal echocardiography (TEE). Patients without atrial thrombi may undergo cardioversion and be given warfarin for 4 weeks. Minimal data are available about embolic risk and the need for anticoagulation in patients with AF of 48 hours' duration or less.

	Epidemiology

Top Executive Summary Epidemiology Pathophysiology Clinical Presentations Approach to Treatment Restoration of Sinus Rhythm... Control of Ventricular Rate... Preventing Thromboembolism in... References

 
Atrial fibrillation is the most common sustained arrhythmia encountered in clinical practice. Recent data suggest that hospital stays for AF are markedly greater than for any other arrhythmia.¹ Nevertheless, information about its incidence and prevalence in a general population is rather sparse. There are fewer data on atrial flutter. Data from clinical populations are subject to the influence of a number of factors that tend to introduce bias. The single best sources of data are reports from the Framingham Study.^{2 3 4 5} It should be noted that the Framingham population may not be representative of the ethnic and racial diversity found in other parts of the country.

Atrial fibrillation commonly occurs with rheumatic heart disease, particularly mitral stenosis. It also occurs with many other cardiac disorders, including coronary heart disease, congestive or hypertrophic cardiomyopathy, mitral valve prolapse, and mitral valve annular calcification. In the setting of acute myocardial infarction or following cardiac surgery, AF is a common but usually self-limited problem. A number of potentially reversible, noncardiac factors are also associated with transient AF. The latter include hyperthyroidism, acute alcohol intoxication, cholinergic drugs, noncardiac surgery or diagnostic procedures, and pulmonary conditions leading to hypoxemia. Continuing problems with AF are most commonly associated with rheumatic heart disease; hypertension, especially when left ventricular hypertrophy is present; and chronic coronary heart disease.² Of course, AF can also occur in the absence of other preexisting conditions; in this situation it is called lone or primary AF.^{3 6 7}

The reported frequency of lone AF, usually about 10% or less,^{3 6 7} may be lower than that currently observed in an office-based setting. At least two factors could be responsible: a referral bias caused by patients seen at tertiary medical centers, from which most data are reported, who are more sick; a change in the epidemiology of AF; or both. Other data (E.N. Prystowsky, unpublished data, 1996) demonstrate that approximately 28% of 172 consecutive outpatients referred for evaluation of AF had lone AF, defined as the absence of any known etiologic factors plus normal ventricular function by echocardiography. The majority of patients were younger than 65 years, although age was not used to define lone AF. Future investigations involving patients of various ages and from various types of practices should clarify this issue.

Prevalence of AF increases with age and is slightly more common in men than in women.² The prevalence of AF is 0.5% for the group aged 50 to 59 years and rises to 8.8% in the group aged 80 to 89 years.⁵ In 1982 the cumulative incidence of development of AF over 22 years in the Framingham Study was 2.2% in men and 1.7% in women.² Excluding persons with rheumatic heart disease, the 2-year incidence of development of AF in 1987 in Framingham was 0.04% for men and 0% for women aged 30 to 39 years; the corresponding figures in the 80-to-89 age group were 4.6% and 3.6%.⁴ In the more recently initiated Cardiovascular Health Study, a cross-sectional population study of Americans older than 65 years, the prevalence of AF on a 24-hour electrocardiographic (ECG) recording was approximately 5%.^{7 8} Because women have a greater life expectancy than men, the actual number of cases in elderly women (older than 75) is greater than it is in elderly men. Based on biennial examination in the Framingham Study, the age-adjusted prevalence of AF has increased substantially, particularly in men, over a 30-year period beginning in the mid 1950s (unpublished data, P.A. Wolf, 1995). The importance of this observation, if it can be generalized to the nation, is that the impact of AF with respect to stroke and other consequences may actually be much greater than estimated.

The cardiac precursors of AF were slightly different in men and women in the 1982 report from Framingham.² In men the significant associations were stroke, cardiac failure, rheumatic heart disease, and hypertensive cardiovascular disease. The strongest associations of these were cardiac failure and rheumatic heart disease. In women the only two significant precursors were cardiac failure and rheumatic heart disease, and the latter was much stronger than in men. Echocardiographic correlates of these clinical features included left atrial enlargement, increased left ventricular wall thickness, and reduced left ventricular shortening.⁹ Cardiovascular risk factors significantly associated with AF in Framingham were diabetes and left ventricular hypertrophy.²

The importance of AF in a general population is that its appearance heralds a mortality rate double that of control subjects. Much of the morbidity and some of the mortality from AF are due to stroke. The risk of stroke is not due solely to AF, and it substantially increases in the presence of other cardiovascular disease. The attributable risk of stroke from AF is estimated to be 1.5% for the 50-to-59-year age group and approaches 30% for those aged 80 to 89.^{4 5}

There is conflict in the literature concerning prognosis for lone or primary AF. In a 1985 report from Framingham, lone AF accounted for 11.4% of all cases of AF. Importantly, in these persons the relative risk for development of stroke was 4.1% compared with control subjects, whereas incidences for coronary heart disease events and congestive heart failure were the same in both groups.³ In Olmsted County, Minnesota, 2.7% of all patients aged 60 years or younger with AF were identified as having lone AF, and among these patients there was no increased risk of stroke.⁶ The major difference between the two studies is that the latter⁶ restricted observations to younger patients free of diabetes or hypertension, whereas in the former³ the majority of the cases of AF occurred in those older than 60 years, 31.5% of whom had hypertension, an important risk factor for stroke in patients with AF.³ Thus, it appears that there is an increased risk of stroke with lone AF in those older than 60 but not in younger patients. In the Cardiovascular Health Study, which included only patients over 65, 6% of cases of AF in women were categorized as lone AF; 9% of cases of AF in men were lone AF.⁷ The investigators have questioned the clinical usefulness of this diagnosis in the elderly.

Atrial fibrillation is relatively rare in the first 2 decades of life. It has been reported in the fetus, neonate, child, and adolescent, but except for the study by Radford and Izukawa,¹⁰ the reports have usually been single case. The rarity of cases has resulted in a limited clinical experience, which has hindered understanding of the etiology of AF, development of management strategies, and characterization of prognosis in the young patient.

Despite the rarity of cases, a few generalizations can be made. In the fetus and neonate, AF is almost always associated with an accessory AV connection.^{11 12 13} In this setting it is postulated that rapid AV reentry degenerates into AF¹⁴ ; the natural history favors spontaneous resolution of tachycardia during the first year of life.¹⁵ A similar mechanism has been proposed for development of AF during paroxysmal supraventricular tachycardia in adolescents.¹⁶ Atrial fibrillation has been reported in association with dilated and hypertrophic cardiomyopathy. The precise reason for this association has not been established, but it may be related to other aspects of the heart disease rather than as a consequence of the hemodynamic derangement. For example, Spirito et al¹⁷ noted that patients with hypertrophic cardiomyopathy and AF usually have the unobstructed type with mild left ventricular hypertrophy.

Long-standing, severe AV valve disease is a risk factor for development of AF. Such a problem is unusual in young persons in the industrialized world, but in areas in which rheumatic heart disease is prevalent, AF is common in young patients. Of note, AF is uncommon in young patients with tachycardia-bradycardia or sick sinus syndrome, which is relatively common following atrial surgery for congenital heart defects.¹⁸ Atrial fibrillation may occur in an adult with previous surgery to correct congenital heart disease. In this setting the tachycardia is usually atrial with some features of atrial flutter.¹⁹ Furthermore, when such patients undergo pacing conversion of atrial tachycardia, AF develops in about 25% of them, but this is transient and supervened by sinus rhythm within minutes.²⁰ Finally, in ostensibly normal adolescents AF is rare, but in such cases hyperthyroidism has been noted.²¹ Atrial fibrillation has infrequently been reported in association with intracardiac tumor and muscular dystrophy.

	Pathophysiology

Top Executive Summary Epidemiology Pathophysiology Clinical Presentations Approach to Treatment Restoration of Sinus Rhythm... Control of Ventricular Rate... Preventing Thromboembolism in... References

 
Atrial tachyarrhythmias in general and AF in particular occur in three distinct clinical circumstances: (1) as a primary arrhythmia in the absence of identifiable structural heart disease; (2) as a secondary arrhythmia in the absence of structural heart disease but the presence of a systemic abnormality that predisposes the individual to the arrhythmia; and (3) as a secondary arrhythmia associated with cardiac disease that affects the atria. At this time it is unclear whether the pathophysiological factors responsible for maintaining AF after it has been induced are common to all three, but it is likely that the initiating factors differ.

There are distinct differences between pathological findings in AF secondary to cardiac disorders and other forms of AF. Atrial dilatation and patchy fibrosis ranging from scattered foci to diffuse involvement, including evidence of destruction of the sinoatrial node, are commonly present when AF is associated with structural heart disease.^{22 23} In contrast, AF secondary to systemic disorders, such as thyrotoxicosis or electrolyte disturbances, is usually unaccompanied by pathological abnormalities or at most by nonspecific scattered fibrosis. From the limited information available, paroxysmal AF in otherwise healthy persons has no pathological correlate and has its basis in either abnormal function of atrial myocyte ion channels, a functional disorder of atrial myocardium, or is associated with unidentified nonpathological structural abnormalities. In some patients AF may be the earliest manifestation of sick sinus syndrome, often a panatrial disease.

There are defined clinical associations between patterns of AF and pathophysiology based on underlying cause. In the presence of heart disease, AF may be paroxysmal at onset and appears to begin when the disease has progressed to clinically significant levels. Over time it has a tendency to become chronic and fixed, rather than remaining paroxysmal. In contrast, systemic conditions that predispose a person to AF are usually associated with persistence of arrhythmia during the period of time that the abnormality persists, which is then followed by either spontaneous reversion after treatment of the predisposing condition or ability to maintain sinus rhythm after cardioversion. In the absence of cardiac or noncardiac disorders, AF is most commonly paroxysmal and recurrent and only occasionally more persistent.²⁴

Electrophysiology of Atrial Fibrillation
The most widely accepted theory of the mechanism of AF is the multiple wavelet hypothesis of Moe.²⁵ This hypothesis envisions multiple reentrant impulses of various sizes wandering through the atria, creating continuous electrical activity. The wavelets are of the leading circle type, with a functionally determined area of conduction block at the center of the circle preventing its collapse and extinction.²⁶ The critical number of wavelets required for the perpetuation of AF is approximately six²⁶ ; the number of wavelets appears to be smaller with longer wavelengths in coarse AF and greater with smaller wavelengths in fine AF.

The wavelength, or the product of conduction velocity and refractory period, is a critical determinant for maintaining AF. Influences that increase the wavelength tend to prevent or terminate AF, whereas those that tend to shorten the wavelength of the atrial impulse tend to favor the onset and perpetuation of AF. Wavelength can be prolonged by antiarrhythmic drugs and shortened by increased parasympathetic tone, rapid atrial pacing, or intra-atrial conduction abnormalities.

Recent data suggest that AF may cause electrical changes in the atria, which may lead to persistence or recurrence of AF.²⁷ In a chronic instrumented goat model, Wijffels et al²⁷ noted that (1) artificial maintenance of AF resulted in a prolonged duration of subsequent episodes of AF; (2) atrial refractoriness was shortened markedly during the first 24 hours of AF; and (3) an inverse rate adaptation of atrial refractoriness was manifested by shortening of the effective refractory period at slower paced rates. Anatomic and genetic correlates of these findings remain to be elucidated. However, these exciting new observations may explain, in part, the clinical observation that maintenance of sinus rhythm is more difficult in patients who have had persistent AF for many months.

Functional Mechanisms of Atrial Fibrillation
It has long been recognized that increased parasympathetic tone predisposes otherwise normal hearts to the onset of AF.²⁸ This may occur through several mechanisms, but it is clear that increased parasympathetic activity, as well as the muscarinic agonist acetylcholine, can abruptly shorten the time course of repolarization through activation of a muscarinic potassium channel in atrial muscle.²⁹ This action shortens the refractory period of atrial tissue and therefore shortens the wavelength of the atrial impulse. Increased sympathetic tone may also lead to AF. In addition, a "maladaptation" of atrial refractory periods to variations in heart rate has been associated with a propensity to AF.³⁰ The absence of physiological shortening of the refractory period in response to an increased heart rate has been observed and found highly predictive for atrial tachyarrhythmias. This correlation appears to be at variance with the mechanism related to the response to parasympathetic stimulation, since maladaptation produces the reverse effect of the shortening induced by acetylcholine. It is possible that two fundamentally different mechanisms can both be responsible for potentiation of atrial tachyarrhythmias under different conditions, or that maladaptation to pacing may correlate with increased response to parasympathetic surges, a possibility that has not yet been studied. Furthermore, it is possible that changes in refractoriness may differ in various parts of the atria.

Regardless, present knowledge suggests that one feature common to all patterns is the requirement for multiple wavelets of activation to sustain AF in normal and abnormal hearts. The possibility that multiple areas of focal automaticity could produce the same pattern should not be dismissed, particularly in the presence of underlying structural heart disease and the systemic abnormalities associated with AF. In this regard, a recent observation in a small group of patients suggests that a unifocal atrial mechanism may be the initiating factor in some patients with apparent lone AF. Such a mechanism could also produce multiple wavelets of activation, although by different pathophysiological mechanisms.

Atrioventricular Node Conduction
The AV node exhibits several properties that tend to limit ventricular rate during AF. First, the excitability of cells within the AV node is significantly less than the adjacent atrial myocardium.³¹ The inexcitable period of AV nodal cells is delayed beyond the repolarization phase of the action potential. Thus, the refractory period of the AV node tends to be relatively prolonged. Second, the AV node demonstrates decremental conduction properties; that is, the amplitude and rate of rise of cardiac action potentials decrease progressively from cell to cell. Therefore, as action potentials are conducted along the course of the AV node, there is a progressive decrease in their ability to induce new action potentials in the cells that lie ahead.³² Because of this property of decremental conduction, impulses may traverse a portion of the AV node before encountering conduction block. Among the clinical manifestations of this property is the phenomenon of concealed conduction, in which an atrial impulse that itself does not conduct to the ventricles may impair conduction of subsequent impulses.³³ Thus, a premature electrical impulse may slow conduction of another impulse through the AV node, blocking an impulse that otherwise would have conducted.³⁴ The conduction interval through the AV node is inversely related to the coupling interval of the preceding atrial impulses, with short atrial cycles resulting in longer AV nodal conduction intervals than longer atrial cycles. Moe and colleagues³⁵ observed that when the atrial rate during AF was relatively slow, there was a tendency for the ventricular rate to increase; conversely, when the atrial rate increased, the ventricular rate slowed. The combined properties of concealed conduction, delayed refractoriness, and the rapid, variable cycle length of the wavelets of atrial activation tend to slow conduction of impulses through the AV node in AF. The result is a significantly slower ventricular rate than atrial rate. Accessory AV pathways typically do not share these electrophysiological properties of the normal AV node, and patients with Wolff-Parkinson-White syndrome require special therapeutic considerations.

The electrophysiological properties of the AV node are profoundly affected by autonomic influences. Withdrawal of vagal inhibition or an increase in sympathetic stimulation facilitates AV nodal conduction. Exercise is associated with both of these changes in autonomic tone, and the ventricular response during AF may substantially increase along with the metabolic demands of the individual. Changes in AV nodal conduction will reflect the adequacy of heart rate control in patients with AF during varying levels of exertion, emotion, and other metabolic stresses. The marked variation in AV nodal conduction properties as a consequence of varying autonomic tone often presents a therapeutic dilemma for physicians treating patients with AF in whom the ventricular rate may be excessively rapid during exercise, yet inappropriately slow during rest.

Hemodynamic Effects Related to Loss of Atrioventricular Synchrony and Irregular RR Intervals
In addition to an inappropriately rapid heart rate, patients with AF experience the loss of normal AV synchrony and an irregular ventricular rhythm. The loss of effective atrial contraction may result in a marked decrease in cardiac output, especially for persons with impaired diastolic filling of the ventricles. Patients with mitral stenosis, restrictive or hypertrophic cardiomyopathy, pericardial diseases, or ventricular hypertrophy are especially likely to experience hemodynamic deterioration with development of AF. In contrast, patients with impaired systolic function with elevated filling pressures and dilated, compliant ventricles may experience only minor hemodynamic deterioration as a consequence of losing AV synchrony.

The random variation in RR intervals during AF results in a constantly changing diastolic filling interval. This fluctuation in diastolic filling interval results in a widely varying stroke volume. Naito and colleagues³⁶ studied the effects of an irregular ventricular rhythm on cardiac output in dogs with complete AV block during ventricular pacing. These investigators found that an irregular ventricular rhythm was associated with a 15% decline in cardiac output compared with a regular rhythm at the same average pacing rate. Mitral regurgitation was observed in these animals during an irregular paced rhythm but not with constant-rate pacing. Thus, it appears that both loss of AV synchrony and irregularity of the ventricular rhythm have an adverse impact on cardiac output. It is quite likely that many symptoms related to AF are due to the variation in left ventricular stroke volume as a result of irregular RR intervals. Chronically elevated ventricular rates during AF or any supraventricular tachycardia may result in a reversible form of ventricular dysfunction characterized by global hypokinesis and dilatation. Persons with continuously elevated ventricular rates (usually greater than 130 beats/min) for a period of several months are at risk of developing a tachycardia-induced cardiomyopathy.³⁷ This form of cardiomyopathy is often reversible following effective control of the ventricular rate. In fact, several investigators have found this to be true in patients with AF when pharmacological or nonpharmacological methods are used to control ventricular rate.^{38 39 40}

	Clinical Presentations

Top Executive Summary Epidemiology Pathophysiology Clinical Presentations Approach to Treatment Restoration of Sinus Rhythm... Control of Ventricular Rate... Preventing Thromboembolism in... References

 
Symptoms associated with AF vary and depend on several factors, including ventricular rate, cardiac function, concomitant medical problems, and individual patient perceptions. Most patients experience palpitations, but presyncope, dizziness, fatigue, and dyspnea are not uncommon. Furthermore, a minority of patients are asymptomatic and AF is discovered by chance. Although asymptomatic patients usually have a relatively controlled ventricular rate, even in the absence of drugs, in some instances the ventricular rate is greater than 100 beats/min. Some patients may have left ventricular dysfunction presumably secondary to a persistently fast ventricular rate during AF.³⁸ Three specific, although relatively uncommon clinical presentations, are noted below.

Tachycardia-Induced Tachycardia
Tachycardia-induced tachycardia is the phenomenon in which one tachycardia degenerates into another.¹⁴ For example, atrial flutter and certain atrial tachycardias often degenerate into AF. More interesting is the induction of AF as a result of a nonatrial tachycardia; for example, AV or AV node reentry.^{14 41 42 43 44} Even ventricular tachycardia, with or without ventriculoatrial conduction, can initiate AF.¹⁴ The mechanism of tachycardia-induced tachycardia in nonatrial arrhythmias is unclear and probably multifactorial. Rate of tachycardia, accessory pathway electrophysiological properties, intrinsic atrial vulnerability, and contraction-excitation feedback⁴⁵ may be causative. In patients who underwent surgery for Wolff-Parkinson-White syndrome, Chen et al⁴² reported that cycle length of AV reentry was shorter in patients with a history of AF. Spach et al⁴⁶ showed that micro-reentry based on anisotropy can occur in very small muscle bundles in the atrial appendage, and in patients with AV reentry, spontaneous degeneration into AF during electrophysiological study commonly occurs first on the high right atrial recording.^{14 47} Sudden dilatation of the atria, which can be observed at surgery after onset of AV reentry, can affect cardiac membrane potential⁴⁵ and lead to development of AF. Whatever the exact mechanism for tachycardia-induced tachycardia in patients with otherwise normal atria, it is important to be aware of its occurrence. In patients with a history of regular palpitations preceding AF, AV or AV node reentry may be a cause of AF. In these persons, elimination of the primary arrhythmia almost always prevents further episodes of AF.¹⁴

Atrial Fibrillation in Wolff-Parkinson-White Syndrome
Atrioventricular reentry can initiate AF, which can lead to disastrous consequences if the patient is capable of sustaining a very rapid preexcited ventricular response with conduction over the accessory pathway. The rapid heart rate can produce syncope or, more important, AF may cause ventricular fibrillation and sudden cardiac death.^{48 49 50} Patients who have been resuscitated from ventricular fibrillation have a rapid preexcited ventricular response as demonstrated during induction of AF in electrophysiological study.^{48 49} These patients require aggressive therapy, most commonly radiofrequency catheter ablation of the accessory pathway. Sudden cardiac death in patients with ventricular preexcitation who have symptomatic arrhythmias should be preventable, and these persons should undergo electrophysiological evaluation. If they are capable of sustaining a rapid preexcited ventricular response during AF (eg, the shortest preexcited RR interval is less than 260 ms), catheter ablation of the accessory pathway or aggressive antiarrhythmic drug therapy to prevent conduction over the accessory pathway is necessary.

Neurogenic Atrial Fibrillation
Coumel⁵¹ described a vagal and adrenergic form of AF. Vagal origin of AF is characterized by (1) predominance in men rather than in women (approximately 4:1); (2) age at onset approximately 40 to 50 years; (3) lone AF with minimal tendency to permanent AF; (4) occurrence at night, during rest, after eating, or with absorption of alcohol; and (5) AF usually preceded by progressive bradycardia. Importantly, both ß-adrenergic blocking drugs and digitalis may increase frequency of AF.

According to Coumel,⁵¹ adrenergic AF has the following features: (1) occurs less frequently than vagal AF; (2) onset is exclusively during daytime; (3) often preceded by exercise and emotional stress; (4) polyuria is common; (5) onset typically occurs with a specific sinus rate, often near 90 beats/min. In contrast to vagally induced AF, ß-adrenergic blockers are usually the treatment of choice.

Patients with "pure" vagally dependent or adrenergic AF are very uncommon. However, history taking may reveal a pattern of onset of AF that has elements of one of these syndromes. This is important because it allows the clinician to select agents that are more likely to prevent episodes of AF in these patients.

	Approach to Treatment

Top Executive Summary Epidemiology Pathophysiology Clinical Presentations Approach to Treatment Restoration of Sinus Rhythm... Control of Ventricular Rate... Preventing Thromboembolism in... References

 
Patient symptoms often direct the physician's therapeutic decisions. For example, a patient with bothersome palpitations in the presence of a rapid ventricular response requires treatment directed at AV nodal conduction to decrease the rate, and if this cannot be achieved with drugs, then nonpharmacological methods may be needed. Restoration and maintenance of sinus rhythm may also be appropriate. In contrast, a patient with a controlled ventricular response during AF and fatigue or shortness of breath often requires maintenance of sinus rhythm as the primary therapeutic goal. In this instance, nonpharmacological methods to control ventricular rate often will not ameliorate symptoms. Asymptomatic persons, especially those with good rate control and depending on the circumstances, may not need further therapy or may be considered candidates for anticoagulation.

In summary, three therapeutic goals should be considered for each patient: rate control, maintenance of sinus rhythm, and prevention of thromboembolism. Heart rate control is a requisite for all patients. The risks and benefits of other therapy must be considered for each patient. A recently initiated NIH trial (AFFIRM) will evaluate the merits and problems of rate control versus maintenance of sinus rhythm.

	Restoration of Sinus Rhythm and Primary Prevention of Atrial Fibrillation

Top Executive Summary Epidemiology Pathophysiology Clinical Presentations Approach to Treatment Restoration of Sinus Rhythm... Control of Ventricular Rate... Preventing Thromboembolism in... References

 
Pharmacological Therapy
Restoration of sinus rhythm may improve symptoms and hemodynamics and has even been associated with an increase in cerebral blood flow.⁵² Several antiarrhythmic drugs with disparate electrophysiological effects on atrial tissue often restore sinus rhythm. About 50% of patients with new-onset AF will convert spontaneously to sinus rhythm within 24 to 48 hours of presentation.⁵³

Digitalis, verapamil, propranolol, and esmolol rarely terminate AF.^{53 54 55} Falk et al⁵³ noted no difference in acute reversion of AF in patients with and without digitalis. Agents such as quinidine or procainamide will frequently terminate paroxysmal AF.^{56 57} The ability to acutely terminate and maintain sinus rhythm over the long term may be enhanced when pharmacological therapy is instituted before electrical cardioversion and then continued after sinus rhythm is restored.

Several antiarrhythmic drugs that affect atrial electrophysiology can terminate or prevent AF.^{56 57 58 59 60 61 62 63 64 65 66 67 68 69} These include quinidine, procainamide, disopyramide, propafenone, sotalol, flecainide, and amiodarone. Few comparative data exist to assess the efficacy of these agents in maintaining sinus rhythm. In one study sotalol was equally as effective as quinidine.⁶⁵ Sinus rhythm will be maintained in more than 50% of patients treated with propafenone, flecainide, or sotalol, often with fewer side effects than quinidine.^{60 61 64} Amiodarone is considered by some the most effective agent for drug-refractory, symptomatic, recurrent AF, although minimal prospective comparative drug data are available. Nearly two thirds of patients treated remained in sinus rhythm for up to 1-year follow-up.^{66 67 68 69 70} Frequent use of amiodarone for AF is limited by its potentially severe and life-threatening side effects, which may be minimized with lower daily doses.⁶⁹

Recurrence of AF is common, although antiarrhythmic drugs often effectively decrease the frequency of AF as measured by time to first recurrence and the arrhythmia-free interval.^{60 63} Thus, successful drug therapy should be evaluated by the decrease in number and duration of AF episodes and not its mere recurrence. Patients with long-standing AF, large left atrial size, or those with multiple previous drug failures will have the highest recurrence rates. Concomitant ventricular rate control therapy with oral digoxin, verapamil, diltiazem, or a ß-blocker should be considered.

Recommendation
There are minimal data from randomized clinical trials that confirm superiority of efficacy of any particular drug over the others. Therefore, selection of an antiarrhythmic agent should be individualized and will depend in part on renal and hepatic function, concomitant illnesses and drugs, and cardiovascular function. Initially, drug dosages should be in the low to moderate range and titrated upward if efficacy is not achieved and side effects are not limiting. Certain agents interact with warfarin and digoxin, and close observation is warranted in these circumstances.

Wolff-Parkinson-White Syndrome
Patients with anterograde conduction over an accessory pathway during AF may be hemodynamically stable and may not require emergency electrical cardioversion. However, these patients represent the one subgroup with supraventricular tachyarrhythmias that may be at risk for sudden cardiac death if the preexcited ventricular response is rapid (see above). Drugs such as digitalis, calcium channel blockers, and ß-adrenergic blockers typically will be ineffective in blocking conduction over the accessory pathway and frequently enhance conduction resulting in hypotension or precipitation of cardiac arrest.^{55 71 72} These drugs should not be given in such a situation. Lidocaine typically will not terminate AF and may rarely enhance conduction over the accessory pathway. Intravenous procainamide is the treatment of choice. Patients who are hemodynamically unstable require direct-current cardioversion.

Recommendation
Intravenous administration of procainamide may decrease conduction over the accessory pathway and terminate AF. In patients who are hemodynamically unstable, direct-current cardioversion is the treatment of choice. As noted previously, these patients require electrophysiological evaluation and aggressive antiarrhythmic treatment, preferably catheter ablation of the accessory pathway.

Risks Associated With Antiarrhythmic Drug Therapy for Atrial Fibrillation
It is assumed that antiarrhythmic drug treatment will reduce recurrence of AF and thus morbidity. This hypothesis has not been well tested. Asymptomatic short-term recurrence of AF episodes appears to be common.⁷³ It is not known if these asymptomatic recurrences are associated with an increased risk for stroke. In addition, data suggest that use of antiarrhythmic drugs may be associated with increased risk.^{56 74 75 76 77 78 79} In a meta-analysis study, Coplen et al⁵⁶ reported that use of quinidine was associated with a 2.9% mortality, compared with only 0.9% in patients not treated with quinidine (P<.05). However, many deaths were due to noncardiac causes. In the Stroke Prevention in Atrial Fibrillation trial, Flaker et al⁷⁷ noted enhanced mortality in AF patients with associated heart failure who were treated with drugs such as quinidine and procainamide. The results of the Cardiac Arrhythmia Suppression Trial (CAST) indicate that flecainide should be avoided in the post–myocardial infarction setting.⁸⁰ Pritchett et al⁸¹ reported that in patients with supraventricular tachycardia, flecainide and encainide were not associated with enhanced mortality.

Because of its vagolytic effect and resultant rapid ventricular response, quinidine should not be given without prior administration of agents that slow AV nodal conduction. Many antiarrhythmic agents, especially flecainide and propafenone, may slow the atrial rate, allowing more atrial impulses to be conducted through the AV node, which results in a faster ventricular rate.⁷⁴ In fact, many episodes of wide QRS tachycardia in patients with AF treated with flecainide or propafenone appear to be secondary to atrial flutter with 1:1 AV node conduction associated with QRS widening.^{82 83} The resultant arrhythmia is often misdiagnosed as ventricular tachycardia. This situation can be prevented or treated by the addition of agents that slow AV node conduction, for example, digitalis, ß-adrenergic blockers, or calcium channel blockers. Occasionally, treatment of patients with tachycardia-bradycardia syndrome worsens sinus node function, resulting in symptomatic bradyarrhythmias that require pacemaker support.

The most common proarrhythmic event reported with antiarrhythmic drug therapy for AF is torsade de pointes, a rapid polymorphic ventricular tachycardia^{74 78} that occurs with agents that prolong ventricular repolarization (QT interval) (eg, quinidine). Patients most at risk are those with ventricular dysfunction, hypokalemia, or baseline prolongation of the QT interval. Torsade de pointes usually occurs after sinus rhythm has been restored.^{74 75} Thus, for patients at risk, it is recommended that therapy be initiated in the hospital and the patient observed for 24 to 48 hours in sinus rhythm. Recent data suggest that patients with ventricular hypertrophy may be at increased risk of developing torsade de pointes,⁷⁴ and drugs that prolong the QT interval should be used with caution in these patients.

Drugs such as propafenone and flecainide do not prolong ventricular repolarization but may cause other forms of life-threatening ventricular tachyarrhythmias. This usually occurs with rapid dose escalations, coexisting potentially lethal ventricular arrhythmias, or left ventricular dysfunction.⁸⁴ These patients should also be monitored in hospital during drug titration for unexpected proarrhythmic responses. Finally, development of incessant atrial arrhythmias or atrial arrhythmias not previously documented is a rare but well-described atrial proarrhythmic response.⁷⁸ These arrhythmias are frequently resistant to pharmacological and/or electrical cardioversion, with spontaneous conversion occurring after the drug is metabolized. Because it is difficult to predict the hemodynamic and ventricular responses to an incessant atrial proarrhythmia after the drug is discontinued, treatment is most safe when done in the hospital with telemetry monitoring. Digitalis may cause more frequent episodes of AF in certain patients, possibly because of its vagomimetic effect.

Recommendation
Proarrhythmia is the most important risk associated with antiarrhythmic drug therapy in patients with AF. Both bradyarrhythmias, especially sinus bradycardia, and ventricular tachyarrhythmias, especially torsade de pointes, can occur. Proarrhythmia is relatively rare in patients without heart disease, and outpatient initiation of antiarrhythmic treatment is reasonable. Patients with heart disease are most at risk for proarrhythmia, especially those with a history of congestive heart failure. Inpatient initiation of antiarrhythmic therapy is recommended for these patients.

Nonpharmacological Therapy
Several surgical approaches have been used to treat patients with AF.^{85 86 87} The left atrial isolation technique and "corridor" operation^{85 86} have met with less enthusiasm compared with the maze procedure.⁸⁷ In the maze operation, multiple atrial incisions are made to channel sinus impulses through a path, or "maze," to reach the AV node. This prevents a critical mass of contiguous atrial tissue from sustaining AF while maintaining atrial contractility. Early results have been optimistic, but long-term follow-up of substantial numbers of patients with a variety of causes of AF is lacking. A substantial number of patients may require implantation of a permanent pacemaker after the maze procedure.

Results from surgical approaches suggest that large areas of the atria must be isolated electrically from each other to prevent AF. Endocardial radiofrequency catheter ablation can be used to create long linear lesions for this purpose and is under extensive study by many investigators. A recent case report documenting short-term success of catheter ablation of the right atrium only to prevent AF is noteworthy.⁸⁸ It is possible that different ablation approaches may be required, depending on the etiology of AF.

Permanent pacing has also been championed as a method to prevent AF. In this regard, dual-chamber AV systems have been considered superior to single-chamber ventricular demand pacemakers.^{89 90} However, few prospective data are available that document the efficacy of pacing in preventing AF. Even fewer data are available on implantable atrial defibrillators. Newer pacing methods such as biatrial pacing require further investigation.

Recommendation
Nonpharmacological approaches to prevention of AF include surgery, atrial pacing, and endocardial catheter ablation. Too few data are available to recommend pacing or ablation. Preliminary data with the maze operation are encouraging, but longer follow-up in a more diverse patient population is required before any specific recommendations can be made. Recommendations on the use of implantable atrial defibrillators await results from clinical trials.

Prevention of Postoperative Atrial Arrhythmias
Postoperative atrial arrhythmias are most commonly seen after thoracic surgery, especially pneumonectomy and open-heart surgery.^{91 92} Up to 30% of patients may develop an episode that is most often AF. Acute treatment to restore sinus rhythm is often ineffective or associated with early recurrence. However, most patients who develop postoperative AF will have sinus rhythm restored either spontaneously or with pharmacological therapy within 1 week of surgery and will rarely require long-term treatment. Hemodynamic and embolic complications from postoperative paroxysmal AF are rare. Multiple studies have shown that perioperative prophylaxis with low-dose ß-blockers frequently prevents acute episodes.⁹²

	Control of Ventricular Rate in Atrial Fibrillation

Top Executive Summary Epidemiology Pathophysiology Clinical Presentations Approach to Treatment Restoration of Sinus Rhythm... Control of Ventricular Rate... Preventing Thromboembolism in... References

 
Pharmacological Control of Atrioventricular Conduction During Atrial Fibrillation
Pharmacological agents that depress conduction and prolong refractoriness in the AV node are frequently required for control of symptoms and improvement of hemodynamics during AF. Commonly used drugs that prolong refractoriness and decrease conduction velocity in the AV node include digoxin, ß-adrenergic antagonists, and calcium channel blockers. The electrophysiological actions of digoxin on the AV node are indirect and depend on cardiac innervation, with minimal direct effects noted.⁹³ The physician attempting to slow the ventricular rate during AF must consider two phases of treatment: an acute phase that involves rapid control of ventricular rate and a long-term phase that involves drugs given orally.

Patients with an episode of AF often develop rapid ventricular rates and may be highly symptomatic. In the presence of important clinical symptoms, such as chest pain or exacerbation of congestive heart failure, that are related to a rapid ventricular response, intravenous drug therapy to slow the heart rate relatively quickly is often required. Although intravenous digoxin may effectively slow the ventricular rate at rest, there is a delay in onset of the therapeutic effect of at least 60 minutes in most patients, with the full effect delayed for up to 6 hours. In addition, digoxin is no more effective than placebo for converting AF to sinus rhythm⁵³ and may prolong duration of paroxysmal AF.⁹⁴ Thus, except in heart failure, it is not first-line therapy.

For patients with severe symptoms related to a rapid ventricular rate, intravenous diltiazem, verapamil, esmolol, propranolol, or metoprolol provide rapid control of heart rate. Each of these drugs has a potential role in acute management of symptomatic individuals with AF and rapid ventricular rates. For example, an intravenous bolus of diltiazem (20 to 25 mg) resulted in a slowing of ventricular rate during AF or atrial flutter in 94% of patients in one multicenter study.⁹⁵ The median time to therapeutic response was 4 minutes, and a continuous infusion of diltiazem (10 to 15 mg/h) maintained control of ventricular rate in 74%. Intravenous verapamil is also effective for slowing ventricular rate during AF.^{55 96 97 98} An initial bolus of 5 to 15 mg reduces the ventricular rate within 5 minutes and can be followed by a maintenance infusion of 0.05 to 0.2 mg/min.^{55 96 97 98} The major adverse effects of intravenous diltiazem and verapamil include hypotension (in up to 7.5% of patients receiving diltiazem).⁹⁹ Intravenous esmolol, an ultrashort-acting ß-adrenergic antagonist, may effectively control ventricular response during AF within 15 minutes.¹⁰⁰ Intravenous metoprolol (5 to 15 mg intravenously over 5 to 15 minutes) and propranolol (1 to 12 mg intravenously over 5 to 12 minutes) can also be administered for control of rapid heart rates in AF. Adenosine is a naturally occurring substance with a half life of approximately 10 seconds that produces marked inhibition of AV nodal conduction. Although adenosine is very effective for terminating reentrant arrhythmias using the AV node,¹⁰¹ this agent has no role in the management of AF because of its transient duration of action.

The control of rapid preexcited ventricular rates during AF in patients with Wolff-Parkinson-White syndrome requires reemphasis. As stated earlier, drugs such as digoxin, adenosine, calcium channel blockers, and ß-adrenergic antagonists should be avoided. Prompt electrical cardioversion is indicated for patients with severe symptoms. Others may be treated with intravenous procainamide.

In the absence of significant symptoms related to a rapid ventricular response to AF, intravenous medications are not indicated because of the potential for adverse effects such as hypotension or precipitation of congestive heart failure. In these less symptomatic persons, the ventricular rate is usually controlled with orally administered medications. Calcium channel blockers and ß-adrenergic antagonists are preferred over digoxin in patients without heart failure. In the presence of thyrotoxicosis or factors associated with increased sympathetic tone, such as exercise or metabolic stress, rapid ventricular rates during AF are most effectively controlled with ß-adrenergic blockers. When there are contraindications to the use of ß-adrenergic antagonists, such as bronchospasm, calcium channel blockers should be considered.

The choice of a drug for long-term ventricular rate control during AF requires consideration of several clinical factors. Some patients with depressed AV nodal function may have adequate rate control and do not require drug therapy. Since AV node conduction is markedly influenced by changes in autonomic tone, the ventricular rate may be well controlled during rest, yet excessive during physical exertion. Thus, although digoxin is useful for slowing the ventricular rate during AF at rest, it provides little control during exercise.¹⁰² For patients without heart failure, ß-adrenergic or calcium channel blockers are preferred for rate control.^{55 103 104 105} Perhaps the most clearly defined role for digoxin in the management of AF is for concomitant congestive heart failure related to impaired systolic ventricular function or in combination with other agents. In the latter circumstance, digoxin provides additive effects with calcium channel or ß-adrenergic blockers for ventricular rate control.^{102 103 104} The balance of rate control during varying levels of physical activity can be especially challenging for patients who have episodes of paroxysmal AF associated with rapid ventricular rates that alternate with periods of sinus rhythm with sinus node dysfunction (tachycardia-bradycardia syndrome).²² Permanent pacing is often required for these patients to allow use of pharmacological agents that slow AV nodal conduction. In some patients with tachycardia-bradycardia syndrome, pindolol, a ß-adrenergic blocker with intrinsic sympathomimetic activity, may provide rate control without the need for permanent pacing. Last, if sotalol or amiodarone is chosen to prevent AF, in some patients no other agent may be needed because these drugs also depress AV node conduction.

Recommendation
It is important to control ventricular response during AF to decrease the patient's symptoms as well as prevent a tachycardia-induced cardiomyopathy. Acute rate control is most effective with intravenous verapamil, diltiazem, or ß-blockers. Intravenous procainamide is the treatment of choice if conduction is over an accessory pathway. Some patients may require immediate cardioversion. ß-Adrenergic blockers are especially effective in the presence of thyrotoxicosis and increased sympathetic tone. For long-term rate control, verapamil, diltiazem, and ß-blockers are more effective than digoxin and should be the initial drugs of choice. Digoxin should be considered as first-line treatment only in patients with congestive heart failure secondary to impaired systolic ventricular function. In some patients, combinations of digoxin, calcium channel blockers, and ß-adrenergic blockers may be needed to control ventricular response.

Nonpharmacological Control of Ventricular Rate During Atrial Fibrillation
Many persons with either paroxysmal or chronic AF will remain significantly symptomatic despite the use of drugs that slow AV nodal conduction or attempt to maintain sinus rhythm. In some patients with severe symptoms that cannot be improved with adequate trials of antiarrhythmic medications, AV nodal ablation and permanent pacemaker implantation may provide symptomatic relief.^{106 107} This procedure has been demonstrated to improve exercise tolerance and quality of life for selected individuals with either paroxysmal or chronic AF.¹⁰⁸ Several methods for ablating AV nodal conduction have been described, including surgical cryoablation, catheter ablation with direct-current shocks,¹⁰⁹ intracoronary ethanol infusion, or radiofrequency energy.¹¹⁰ Radiofrequency catheter ablation has emerged as the most commonly used technique for ablating the AV node. There are several important clinical considerations regarding use of AV nodal ablation for treatment of AF. First, there is no evidence that this technique influences survival in patients with AF. In fact, there has been some concern about a very small possible risk of sudden death after AV junctional ablation in these patients, but it is not clear whether this is not simply the natural history of the underlying cardiac disease. Second, the results of this procedure cannot be reversed. Third, patients with fatigue who have good rate control may receive minimal benefit from this procedure. Fourth, many patients are pacemaker dependent after AV nodal ablation. For this reason, the long-term reliability of pacing leads and pulse generators must be considered. Finally, AV nodal ablation does not change the risk of systemic emboli or the need for anticoagulation. Recently, ablation techniques that modify AV conduction without inducing complete AV block have been described.^{111 112}

Recommendation
Nonpharmacological methods to control ventricular rate include endocardial catheter ablation or modification of the AV junction and surgically induced AV block. Catheter ablation is recommended in patients who have not responded to or are intolerant of drugs used to control ventricular response. Surgery is rarely indicated for rate control.

	Preventing Thromboembolism in Atrial Fibrillation

Top Executive Summary Epidemiology Pathophysiology Clinical Presentations Approach to Treatment Restoration of Sinus Rhythm... Control of Ventricular Rate... Preventing Thromboembolism in... References

 
Atrial fibrillation is a common cardiac arrhythmia of the elderly, and stroke is its most devastating complication. The high risks of thromboembolism in AF associated with mitral stenosis and prosthetic mitral valves have long been appreciated. However, AF even in the absence of these valvular disorders carries a substantially increased risk of ischemic stroke.⁵ The rate of ischemic stroke among elderly people with AF averages 5% per year, about six times that of people without AF.^{4 5 113} Considering transient ischemic attacks (which often cause radiographic evidence of brain infarction) and clinically occult stroke detected radiographically, the rate of brain ischemia accompanying nonvalvular AF exceeds 7% per year, an impressive threat to the brain.^{113 114 115 116 117} However, the absolute rate of stroke varies importantly with patient age and coexistent cardiovascular disease.^{6 113 118 119} Stratification of AF patients into those at high and low risk for thromboembolism is a crucial determinant of optimal antithrombotic prophylaxis, as discussed in detail below.

The prevalence of AF increases with advancing age, and the incidence of stroke in AF patients is similarly age related. Thus, AF is most frequent and more threatening in the very elderly.^{4 5 120} About half of AF-associated strokes occur in patients older than 75, and AF may be the most frequent cause of disabling stroke in elderly women.^{5 121 122} In short, AF-associated stroke is particularly a problem for the very elderly (older than 75 years). Special consideration of stroke prophylaxis in this age group is critical to prevention.

Most ischemic strokes associated with AF are probably due to embolism of stasis-induced thrombi forming in the left atrium and particularly its appendage. Transesophageal echocardiography shows left atrial thrombi to be more frequent in AF patients with ischemic stroke compared with AF patients without stroke. However, perhaps 25% of AF-associated stroke is due to associated intrinsic cerebrovascular diseases, other cardiac sources of embolism, or aortic arch atheroma.^{123 124} About half of elderly AF patients have chronic hypertension, a major risk factor for primary cerebrovascular disease.¹¹³ About 12% of elderly AF patients harbor cervical carotid artery stenosis, but the frequency of carotid artery stenosis is not substantially greater in AF patients with stroke, suggesting that carotid artery stenosis is a minor contributor to AF-associated stroke.¹²⁵ Routine screening of AF patients without symptoms of brain ischemia for cervical carotid artery stenosis does not appear warranted.

Atrial Fibrillation Patients With High and Low Rates of Thromboembolism
The absolute rate of ischemic stroke in patients with AF is critically influenced by coexistent cardiovascular disease. Identification of subpopulations of AF patients who have relatively high or low absolute rates of stroke determine which patients gain the greatest benefit from anticoagulant therapy, offsetting its disutility. Because clinical classification of etiologic subtypes of ischemic stroke is imperfect and not adequately validated,¹²³ risk stratification schemes are presently based on combined analysis of all ischemic strokes. Such stratification schemes for AF patients have been extensively pursued, using clinical and echocardiographic parameters.^{6 113 118 119 122 126 127 128 129 130}

Two prospective studies included sufficient numbers of patients and strokes, analyzed by multivariable techniques, and provide the most reliable stratification schemes available (Table).^{113 118} The differences in the two schemes (presence of age versus heart failure) are not contradictory or even substantially conflicting, as clinical variables overlap (age is related to heart failure, hypertension, and diabetes). Interestingly, intermittent (ie, paroxysmal) AF was not an independent predictor of thromboembolic risk in either study. In summary, the five clinical variables listed in the Table are independently predictive of thromboembolic risk and are clinically useful for characterizing AF patients with high and low risks for stroke. Other studies support the notion that AF-associated stroke is related to patient age, hypertension, coexistent heart disease, and perhaps female gender.^{122 126} Echocardiographic predictors of increased thromboembolic risk in AF are enlarged left atrial size (mitigated by mitral regurgitation) and, independently, impaired left ventricular function.^{119 131} Impaired left ventricular function may further contribute to stasis within the left atrium in AF patients.^{132 133} Precordial echocardiographic findings can be combined with clinical risk stratifiers to identify AF patients with very low inherent rates of thromboembolism.¹¹⁹

View this table: [in this window] [in a new window]   Table 1. Risk Stratification in Atrial Fibrillation1 : Independent Predictors of Thromboembolic Risk

Transesophageal echocardiography offers better visualization of the left atrium and its appendage than conventional transthoracic echocardiography. When TEE is used, left atrial appendage thrombi and spontaneous echogenic densities ("smoke") possibly indicative of stasis are more often found in AF patients with thromboembolism than those without thromboembolism.¹³⁴ Stoddard et al¹³⁵ used TEE to evaluate the frequency of left atrial thrombus in 143 patients with AF of less than 3 days' duration. Twenty patients (14%) had left atrial thrombus and 56 (39%) had spontaneous echocardiographic contrast. A recent systemic embolus occurred in 24 patients, 5 of whom had documented left atrial thrombus. Fatkin et al¹³⁶ showed that thromboembolic complications can occur after cardioversion in patients without left atrial thrombus identified with TEE before cardioversion, an observation confirmed in a multicenter study.¹³⁷ However, the predictive value of TEE findings for subsequent stroke has yet to be validated by adequate clinical studies. Thus, data are insufficient to recommend routine TEE to stratify thromboembolic potential in AF patients.

Antithrombotic Therapies to Prevent Stroke
Anticoagulation with oral vitamin K antagonists such as warfarin is highly effective for reducing ischemic stroke in AF patients. Five recent randomized clinical trials using INR ranges of approximately 1.8 to 4.2 showed a mean reduction in ischemic stroke of nearly 70% in patients assigned to receive anticoagulation (Fig 1); on-therapy analysis indicated an even greater benefit.¹¹³ The incremental risk of serious bleeding was less than 1% per year among patients on anticoagulation who were selected to participate in these clinical trials and followed carefully on protocols. In a more generalized outpatient population, risk of bleeding may be greater.¹³⁸ Low-intensity anticoagulation (INR 2.0 to 3.0) clearly confers benefit.^{113 139} Warfarin is very effective in subgroups of AF patients with a high inherent risk of thromboembolism (Table).^{113 140} Safe anticoagulation requires monitoring with the INR that corrects for varying thromboplastin sensitivities.

View larger version (22K): [in this window] [in a new window]   Figure 1. Risk-reduction plot of results of five randomized clinical trials comparing warfarin with control for prevention of ischemic stroke in atrial fibrillation patients. Horizontal lines indicate 95% confidence intervals (CIs) around point estimates (vertical lines) for each trial. Combined risk reduction was 68% (95% CI, 50% to 79%; P<.991). See Atrial Fibrillation Investigators' pooled analysis113 for specific data. AFASAK indicates Atrial Fibrillation, Aspirin, and Anticoagulant Therapy study; BAATAF, Boston Area Anticoagulation Trial for Atrial Fibrillation; CAFA, Canadian Atrial Fibrillation Anticoagulation; SPAF, Stroke Prevention in Atrial Fibrillation study; SPINAF, Stroke Prevention in Nonrheumatic Atrial Fibrillation Study. Adapted from reference 113.

The safety and tolerability of long-term anticoagulation titrated to conventional levels has been less clear in the very elderly (older than 75), the age group encompassing perhaps half of AF-associated stroke patients. All but one of the placebo-controlled clinical trials testing anticoagulation enrolled AF patients with a mean age in the late 60s.¹¹³ The single placebo-controlled trial involving AF patients with a mean age of 75 reported a 38% withdrawal rate from anticoagulation after 1 year, although withdrawal was not related merely to age.¹⁴¹ A recent clinical trial comparing anticoagulation in AF patients younger and older than 75 years found that risk of major hemorrhage during anticoagulation (INR range 2.0 to 4.5; mean INR, 2.7) was substantially increased in AF patients older than 75 compared with younger AF patients who received anticoagulation of similar intensities.¹⁴⁰ In contrast, the pooled data from five AF trials demonstrated only one intracranial hemorrhage in 223 patients older than 75 receiving warfarin.¹¹³ Further, in the European Atrial Fibrillation Trial,¹⁴² a secondary prevention study, 63% of patients were more than 70 years old, and no intracranial hemorrhages occurred during warfarin therapy (mean INR 2.9). Importantly, the yearly incidence of major bleeding complications was significantly higher during warfarin treatment versus control (2.8% versus 0.7%).¹⁴² Thus, particular caution should be taken and INR levels closely monitored when warfarin is used in the very elderly.

The efficacy of aspirin, an antiplatelet agent, for stroke prevention in AF patients is less clear and remains controversial.¹¹³ The effect of aspirin in doses between 75 and 325 mg/d has been assessed in three randomized, placebo-controlled clinical trials with a statistically significant risk reduction of about 25% (range 14 to 44%) in aspirin-treated patients, based on pooled data.^{114 142 143} However, aspirin was significantly less effective than anticoagulation in two of these clinical trials^{142 143} and also by secondary on-therapy analysis of the third trial¹⁴⁴ (Fig 2). There is no compelling evidence that the specific dose of aspirin between 75 and 325 mg/d confers more or less benefit. In short, aspirin has some degree of efficacy for preventing AF-associated stroke, but it is clearly less than that of anticoagulation.

View larger version (21K): [in this window] [in a new window]   Figure 2. Risk-reduction plot of results of three randomized clinical trials comparing warfarin with aspirin for prevention of ischemic stroke in atrial fibrillation patients. Horizontal lines indicate 95% confidence intervals (CIs) around point estimates (vertical lines) for each trial. Combined estimated risk reduction from published results (intent to treat) was 47% (95% CI, 28% to 61%) by warfarin relative to aspirin. SPAF indicates Stroke Prevention in Atrial Fibrillation study; EAFT, European Atrial Fibrillation Trial; AFASAK, Atrial Fibrillation, Aspirin, and Anticoagulant Therapy study. Adapted from reference 113.

Selecting Antithrombotic Therapy to Prevent Stroke
There is persuasive evidence that antiplatelet agents and anticoagulants differentially affect cardioembolic and noncardioembolic stroke in AF patients.^{123 145} Aspirin has a greater effect on noncardioembolic stroke than on those presumed to be cardioembolic.¹²³ In cardioembolic strokes, aspirin exerted a lesser and nonsignificant effect.¹²³ Aspirin may be particularly effective in AF patients under 75 years old with a history of diastolic hypertension, who are at special risk for noncardioembolic stroke.¹⁴⁵ Antiplatelet agents do appear to be effective in reducing stasis-induced thrombi in the lower extremities,¹⁴⁴ and aspirin may offer some degree of protection against cardioembolic stroke in AF patients.¹²³ Most AF-related stroke, particularly in women, is due to cardiogenic embolism, and these strokes are more effectively prevented by anticoagulation.^{145 146}

Recommendation
Pending further clinical studies now ongoing, low-risk AF patients may be given 325 mg aspirin daily to prevent stroke and may be carefully followed unless high-risk criteria (including systolic blood pressure 160 mm Hg) develop (the Table). This can be strongly recommended for low-risk AF patients younger than 65. High-risk patients who can safely receive anticoagulation should be treated with warfarin. For high-risk AF patients 75 or younger, an INR range of 2.0 to 3.0 is safe and effective; for those over 75, close surveillance of INR levels is recommended because of the apparently greater likelihood of bleeding complications. Patients with AF who cannot safely receive anticoagulation should be given aspirin. The value of other antiplatelet agents has not been assessed in patients with AF.

Secondary Prevention of Stroke in Atrial Fibrillation Patients
The conventional wisdom has been that most ischemic strokes in AF patients are large and disabling, but it is now clear that minor stroke and transient ischemic attacks (TIAs) frequently accompany AF.¹¹³ AF-related stroke carries a particularly high 30-day mortality rate related to advanced patient age and associated heart disease.^{124 147} Stroke in AF patients is usually due to cardiogenic embolism, but an important minority is caused by other mechanisms.^{123 124} In patients with acute stroke who have previously unrecognized AF, atrial fibrillation can be the consequence of brain infarction mediated by other mechanisms.¹⁴⁸

Recent studies indicate that risk of early recurrent stroke (within 2 weeks) is low in patients with AF.^{124 149} Initiating oral anticoagulation within a few days in submassive infarcts seems reasonable; a delay of 1 week or more in starting warfarin in AF patients with large infarcts may be prudent to avoid accentuating secondary brain hemorrhage. Prophylaxis for deep venous thrombosis with low-dose, subcutaneous heparin and/or aspirin in those with lower limb paresis is safe.¹²⁴

The sequence and extent of diagnostic evaluation is aimed at identifying patients who should receive long-term anticoagulation for secondary prevention (most patients) or carotid endarterectomy (few). Carotid sonography shows ipsilateral cervical carotid