Three major tenets govern therapy for atrial fibrillation: maintenance of sinus rhythm, control of ventricular rate, and prevention of thromboembolism. The relative benefit of sinus rhythm over a controlled fibrillating rate is controversial. The advantages of sinus rhythm include elimination of arrhythmic symptoms; improvement in the hemodynamic profile; possible prevention of established (cardioversion-resistant) atrial fibrillation; and a likely, although unproven, decrease in the risk for cardiac thromboemboli [1-3]. Because antiarrhythmic drugs are the method most commonly used to maintain sinus rhythm, the major associated risks are pharmacologic side effects, including potentially lethal ventricular arrhythmias [4]. Judicious selection of antiarrhythmic drugs, done on the basis of the causes of atrial fibrillation and the type of cardiac pathology, and initiation of therapy in the hospital in selected patients should decrease the incidence of and the morbidity and mortality associated with such arrhythmias [5].

Regardless of the potential benefits of sinus rhythm, many physicians think that the risks of drug therapy and the difficulty of maintaining sinus rhythm are arguments for the alternative approach to the treatment of atrial fibrillation: ventricular rate control. In the absence of congestive heart failure, verapamil, diltiazem, and ß-adrenergic blockers are more effective than digoxin for long-term control [1]. If this approach to atrial fibrillation is chosen, it is important to document good rate control (for example, with 24-hour electrocardiography) because patients who have persistently elevated ventricular rates during atrial fibrillation may develop tachycardia-induced cardiomyopathy [1]. In reality, the controversy over sinus rhythm versus rate control applies only to patients in whom symptoms markedly decrease with slowing of the heart rate. A substantial proportion of patients, especially those with fatigue related to atrial fibrillation, will have a sense of well-being only during sinus rhythm. In patients with symptoms, my own bias is to try to maintain sinus rhythm. Of course, it is difficult to make an asymptomatic patient (such as a patient in whom atrial fibrillation is documented during a routine physical examination) feel better, and management decisions in such patients will be unrelated to control of symptoms.

The third and possibly the most important goal of treatment for atrial fibrillation is to prevent cardiac thromboembolism, especially stroke. Two recent studies [6, 7] analyzed many patients with chronic and paroxysmal atrial fibrillation and have reported five independent predictors of risk for thromboembolism: history of hypertension, previous stroke or transient ischemic attack, diabetes, age older than 65 years, and recent heart failure. Patients with any of these clinical variables are at high risk (5% to 7% per year or more) for thromboembolism. Importantly, five randomized clinical trials showed a mean reduction in ischemic stroke of 68% in patients with atrial fibrillation assigned to receive warfarin for anticoagulation [1]. The efficacy of aspirin is less clear and is therefore controversial. Given these data and pending further clinical studies, high-risk patients who can safely receive anticoagulation therapy should be treated with warfarin at a dosage adjusted to achieve an international normalized ratio (INR) of 2.0 to 3.0 [1]. Importantly, warfarin therapy was not discontinued if patients maintained sinus rhythm during some periods of time. Thus, the potential benefit of sinus rhythm in preventing stroke in the absence of anticoagulation is not established and requires further investigation. This caution is supported by preliminary results from a recent study [8] that showed a rate of thromboembolic complications of 1% per year at the moment of recurrent atrial fibrillation in patients in whom anticoagulation therapy had been discontinued when sinus rhythm persisted after conversion.

Electrical [9, 10] and pharmacologic [11] conversion of atrial fibrillation to sinus rhythm is itself associated with a risk for thromboembolism, and anticoagulation therapy given before cardioversion appears to decrease this risk. Bjerkelund and Orning [12] reported embolic complications in 1.1% of patients receiving anticoagulation therapy and 6.8% of patients not receiving anticoagulation therapy after conversion to sinus rhythm. Time to emboli was 6 hours in one patient and 1 to 6 days in the remaining 12 patients. An analysis of pooled trials showed embolic events in 4 of 1221 patients (0.33%) who had received anticoagulation therapy compared with 41 of 2050 patients (2.0%) who did not receive anticoagulation therapy before cardioversion [13]. In some patients, it is the transition from sinus rhythm to atrial fibrillation that appears to produce added risk for emboli [8, 11].

The traditional explanation for embolic events that occur after the conversion of atrial fibrillation to sinus rhythm is that improved atrial contractility dislodges preexisting thrombi; this can be delayed for several weeks [14]. Atrial thrombi, typically in the left atrial appendage, are present on transesophageal echocardiography (TEE) before cardioversion in many patients (15% in one series [15]). However, new thrombi can be detected after successful cardioversion and may also result in embolic events [16]; this phenomenon may be related to a thrombogenic milieu caused by decreased left atrial mechanical function, or atrial "stunning," immediately after cardioversion [17]. In essence, an atrial myopathy results from the constant rapid electrical barrage of atrial fibrillation. Thus, at least two mechanisms for thromboembolism after cardioversion are possible.

Considerable evidence now supports the concept that atrial mechanical dysfunction after cardioversion is probably caused by tachycardia-mediated alterations in atrial function over time rather than by electric shock [2, 3, 14]. Transthoracic cardioversion for ventricular tachycardia is often induced by programmed ventricular stimulation at electrophysiologic study, yet postcardioversion thromboembolism rarely occurs. Further, decreased atrial function has been seen after spontaneous, drug-induced, and electrical conversion to sinus rhythm. An additive effect of electric shock on atrial "stunning" is controversial.

It is currently recommended that warfarin be given to maintain the INR at 2.0 to 3.0 for approximately 3 weeks before and 4 weeks after cardioversion, to decrease the risk for thromboemboli in patients in whom atrial fibrillation has been present for more than 48 hours [1, 18]. This approach appears to be well founded in light of the above data and the recent observation [19] that 89% of atrial thrombi resolve during this period. An alternative approach, in which TEE is used to facilitate early cardioversion, was proposed by Manning and coworkers [16] and shown to be safe.

In this issue, Klein and colleagues [20] present results from the ACUTE (Assessment of Cardioversion Using Transesophageal Echocardiography) pilot study. This multicenter, randomized trial was designed to compare two strategies in patients with atrial fibrillation-TEE-guided cardioversion with anticoagulation therapy and the conventional approach to cardioversion. If an atrial thrombus was detected on TEE, cardioversion was postponed, the patient received warfarin for 4 weeks, and TEE was repeated. Cardioversion was then done if no thrombus was present; otherwise, cardioversion was not done. Inpatients were initially given intravenous heparin, often with concomitant warfarin, and TEE was done within 1 to 2 days. These patients were discharged when anticoagulation after cardioversion was adequate. Outpatients were given warfarin for at least 5 days, and TEE and cardioversion were scheduled to be done when a stable INR (2.0 to 3.0) was assured. All patients received warfarin for 4 weeks after cardioversion.

If it ain't broke, why fix it? What are the perceived problems with the conventional approach to cardioversion? Klein and colleagues [20] state that the major limitations of this approach are the delay of cardioversion for 3 weeks, the increased risk for bleeding, the inconvenience of readmission for cardioversion, a low rate of stroke, and the lack of controlled studies showing the efficacy of this approach. Too few patients were enrolled in this pilot trial to allow any meaningful conclusions to be drawn about these issues. Regardless, they should be discussed. Patients with hemodynamic instability require urgent cardioversion and were excluded from the trial. For outpatients, the difference in time to cardioversion between groups should be approximately 2 weeks. It is unclear why the mean time to cardioversion in the conventional therapy group was 4.7 weeks. However, the successful cardioversion rate and sinus rhythm at follow-up were similar in both groups. A potential advantage for the conventional therapy group was a substantial rate of spontaneous conversion to sinus rhythm that precluded the need for cardioversion. Nine patients reverted to sinus rhythm while receiving antiarrhythmic drugs. Administration of antiarrhythmic drugs before 3 weeks of anticoagulation therapy should be avoided because of the well-described risk for thromboemboli associated with pharmacologic conversion.

With either approach, risk for bleeding should be minimal if anticoagulation is carefully regulated. This regulation may be much easier to achieve in the future with devices that allow for home monitoring of INRs. It is also assumed that patients will discontinue warfarin therapy 4 weeks after cardioversion. However, many patients in this study [20] had a high risk for stroke (for example, more than 25% had hypertension), and long-term anticoagulation therapy is recommended for such patients. A difference of a few weeks in the duration of warfarin therapy hardly matters in these patients.

Admission for cardioversion is generally necessary only for patients for whom the in-hospital initiation of antiarrhythmic drug therapy is contemplated. This would pertain to the outpatients in the TEE-guided group and the conventional therapy group. Cardioversion guided by TEE may offer an advantage for the initial in-hospital patient group. The raison d'etre for this study appears to be the need to establish the comparative safety of the two approaches with regard to thromboembolic risk. The one embolic event in the conventional therapy group occurred in a patient who was inadequately anticoagulated and therefore should not have been cardioverted. Results from the completed ACUTE study should provide valuable data about risk for thromboembolism.

Two potential problems are associated with the TEE-guided approach. First, it requires an echocardiographer with expertise in TEE; this limits its use. A false-negative result could lead to an embolic event. Second, some patients may never be cardioverted because of the persistence of atrial thrombi during adequate anticoagulation therapy. The advantages of sinus rhythm in some of these patients may outweigh the perceived risks of cardioversion.

In conclusion, given the limited data available to date, both the TEE-guided approach and the conventional approach to cardioversion for atrial fibrillation appear reasonable and safe. Until more information is available, the method of cardioversion should be selected for each patient individually.