Management of New Onset Atrial Fibrillation

U.S. Department of Health and Human Services
2101 East Jefferson Street
Rockville, MD 20852
http://www.ahrq.gov

Prepared by:
Johns Hopkins University, Baltimore, MD
Robert L. McNamara, M.D., M.H.S.
Principal Investigator
Eric B. Bass, M.D., M.P.H.
Co-Principal Investigator

Marlene R. Miller, M.D.
Jodi B. Segal, M.D., M.P.H.
Steven N. Goodman, M.D., Ph.D.
Nina L. Kim, M.A.
Karen A. Robinson, M.Sc.
Neil R. Powe, M.D., M.P.H., M.B.A.
Investigators

On December 6, 1999, under Public Law 106-129, the Agency for Health Care Policy and Research (AHCPR) was reauthorized and renamed the Agency for Healthcare Research and Quality (AHRQ). The law authorizes AHRQ to continue its research on the cost, quality, and outcomes of health care, and expands its role to improve patient safety and address medical errors.

This report may be used, in whole or in part, as the basis for development of clinical practice guidelines and other quality enhancement tools, or a basis for reimbursement and coverage policies. AHRQ or U.S. Department of Health and Human Services endorsement of such derivative products may not be stated or implied.

The Agency for Healthcare Research and Quality (AHRQ), formerly the Agency for Health Care Policy and Research (AHCPR), through its Evidence-Based Practice Centers (EPCs), sponsors the development of evidence reports and technology assessments to assist public- and private-sector organizations in their efforts to improve the quality of health care in the United States. The reports and assessments provide organizations with comprehensive, science-based information on common, costly medical conditions and new health care technologies. The EPCs systematically review the relevant scientific literature on topics assigned to them by AHRQ and conduct additional analyses when appropriate prior to developing their reports and assessments.

To bring the broadest range of experts into the development of evidence reports and health technology assessments, AHRQ encourages the EPCs to form partnerships and enter into collaborations with other medical and research organizations. The EPCs work with these partner organizations to ensure that the evidence reports and technology assessments they produce will become building blocks for health care quality improvement projects throughout the Nation. The reports undergo peer review prior to their release.

AHRQ expects that the EPC evidence reports and technology assessments will inform individual health plans, providers, and purchasers as well as the health care system as a whole by providing important information to help improve health care quality.

We welcome written comments on this evidence report. They may be sent to: Director, Center for Practice and Technology Assessment, Agency for Healthcare Research and Quality, 6010 Executive Blvd., Suite 300, Rockville, MD 20852.



John M. Eisenberg, M.D.	Douglas B. Kamerow, M.D.
Director Agency for Healthcare Research and Quality	Director, Center for Practice and Technology Assessment Agency for Healthcare Research and Quality

The authors of this report are responsible for its content. Statements in the report should not be construed as endorsement by the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services of a particular drug, device, test, treatment, or other clinical service.

Objectives. Atrial fibrillation (AF) is the most common arrhythmia physicians face in clinical practice, with an incidence of up to 2.3 percent per year and a prevalence of almost 9 percent in those age 80--89 years. In addition to having various symptoms, patients with AF have increased risk for stroke and death. This report synthesizes the available evidence on strategies for handling each of the following key aspects in the management of patients with new onset AF: electrical cardioversion and pharmacological conversion, maintenance of sinus rhythm, ventricular rate control, stroke prevention, outpatient management, and echocardiography.

Search Strategy. The primary sources for the literature search were the Cochrane Collaboration's database of controlled clinical trials (1948 to 1998) and MEDLINE (1966 to 1998). Specific search terms included atrial fibrillation and atrial flutter, both as medical subject headings and as text words.

Selection Criteria. Articles were included in the evidence synthesis if they reported on original data from a randomized controlled trial addressing noninvasive management of non-post-operative atrial fibrillation in adults. Many of these trials included patients with atrial flutter.

Data Collection and Analysis. A review of 521 potentially relevant citations yielded 179 articles that met eligibility criteria. Study quality was assessed and data abstracted by pairs of reviewers. Results are presented in evidence tables and, where deemed appropriate, synthesized by meta-analysis and decision analysis.

Main Results. Questions regarding pharmacological conversion of AF, maintenance of sinus rhythm, pharmacological rate control, and antithrombotic therapy were addressed directly in the clinical trial literature.

For conversion of AF to sinus rhythm, 46 trials were identified that evaluated 18 different medications. Digoxin, verapamil, and diltiazem were found to be no better than placebo and were combined with placebo to form a control group. The bulk of the evidence for any individual agent existed in comparison with this control group, limiting direct agent-to-agent comparisons. The odds ratio (OR) and 95 percent confidence intervals (CI) for successful conversion of AF revealed the strongest evidence of efficacy, and largest treatment effect sizes, for flecainide (OR 24.7, CI 9.0 to 68) and ibutilide/dofetilide (OR 29.1, CI 9.8 to 86) when compared with control. Strong evidence of efficacy with a fairly large treatment effect size also existed for propafenone (OR 4.6, CI 2.6 to 8.2). Quinidine had moderate evidence of efficacy and a modest treatment effect size compared with control treatment (OR 2.9, CI 1.2 to 7.0). Disopyramide (OR 7.0, CI 0.3 to 152) and amiodarone (OR 5.7, CI 1.0 to 33.4) had suggestive evidence of efficacy, while sotalol (OR 0.4, CI 0.0 to 3.0) had suggestive evidence of negative efficacy for conversion. For maintenance of sinus rhythm, 29 trials were identified that evaluated 10 different medications. Strong evidence of efficacy with fairly large treatment effect sizes for maintenance of sinus rhythm, when compared with control treatment, existed for quinidine (OR 4.1, CI 2.5 to 6.7), disopyramide (OR 3.4, CI 1.6 to 7.1), flecainide (OR 3.1, CI 1.5 to 6.2), propafenone (OR 3.7, CI 2.4 to 5.7), and sotalol (OR 7.1, CI 3.8 to 13.4). There were no randomized clinical trials of amiodarone versus control treatment for maintenance of sinus rhythm. One trial provided moderate evidence of efficacy for amiodarone compared with disopyramide for maintenance of sinus rhythm.

For rate control, 45 trials were identified that evaluated 17 medications. The design and outcome measures of these trials were too disparate for meta-analysis. In general, the evidence suggested that calcium-channel blockers and some beta-blockers were effective for controlling heart rate during exercise.

For antithrombotic therapy, 11 trials were identified that evaluated four medications. The aggregate odds of having a stroke for a subject on warfarin compared with a subject on placebo were 0.30 (CI 0.19 to 0.48), indicating strong evidence of the efficacy of warfarin. However, the odds of having a major bleed were 1.90 (CI 0.89 to 4.04), indicating suggestive evidence for a higher bleeding rate for subjects taking warfarin. The evidence on the efficacy of aspirin compared with placebo was moderately strong for subjects without a prior history of stroke (OR 0.65, CI 0.43 to 0.99) but was inconclusive in subjects with a prior history of stroke. The odds of having a major bleed for subjects on aspirin were 0.81 (CI 0.37 to 1.77), indicating inconclusive evidence. The evidence from trials directly comparing warfarin and aspirin was inadequate to permit strong conclusions. The evidence was inconclusive for the use of low molecular weight heparin, induprofen, or a combination of low-dose warfarin and aspirin for stroke prevention.

In decision analysis, one attempt at electrical cardioversion with pharmacological maintenance therapy was estimated to be cost-effective compared with antithrombotic therapy alone for patients 55 years old and older, regardless of risk factors. For stroke prevention in patients at low risk of stroke (~1 percent/year), aspirin was the most cost-effective therapy. In patients with a high risk of stroke (~10 percent/year), warfarin was the most cost-effective strategy for stroke prevention. In those with an intermediate risk of stroke (~3-6 percent/year), the most cost-effective treatment depended on assumptions about the quality-of-life effects of taking warfarin. In patients without risk factors for thromboembolism, transesophageal echo-cardiography was the most cost-effective test for guiding decisions about antithrombotic therapy.

High priorities for additional research are 1) improving the understanding of the effects of treatment of AF on quality of life, 2) assessing the efficacy of amiodarone in randomized controlled clinical trials, 3) evaluating the risks and benefits of each antiarrhythmic medication in subjects with coronary artery disease, and especially the risk of ventricular arrhythmia, 4) directly comparing antithrombotic agents for prevention of stroke, particularly in low- and moderate-risk populations, and 5) determining the safety of outpatient initiation of antiarrhythmic therapy for AF. Many of these priorities and specifically the second one are beginning to be addressed in trials, such as the AFFIRM Study and the PIAF trial.

Conclusions. Several medications were efficacious in conversion of AF and subsequent maintenance of sinus rhythm. Calcium-channel blockers and beta-blockers were more efficacious than digoxin in controlling ventricular rate during exercise in subjects with AF. Although subjects on warfarin had higher rates of bleeds, warfarin generally was more efficacious than aspirin in preventing stroke in subjects with AF. One attempt of electrical cardioversion with subsequent pharmacological maintenance is cost-effective compared with antithrombotic therapy alone.

This document is in the public domain and may be used and reprinted without permission except those copyrighted materials noted for which further reproduction is prohibited without the specific permission of the copyright holders.

Suggested citation:
McNamara RL, Bass, EB, Miller MR, et al. Management of New Onset Atrial Fibrillation. Evidence Report/Technology Assessment No. 12 (prepared by the Johns Hopkins University Evidence-based Practice Center in Baltimore, MD, under Contract No. 290-97-0006). AHRQ Publication Number 01-E026. Rockville, MD: Agency for Healthcare Research and Quality. January 2001.

Atrial fibrillation (AF) is the most common arrhythmia physicians face in clinical practice, accounting for about one-third of hospitalizations for arrhythmia. The prev¡alence of AF is 0.5 percent in those 50 to 59 years old and 8.8 percent in those 80 to 89 years old, and the incidence ranges from 0.2 percent per year in men 30 to 39 years old to 2.3 percent per year in men 80 to 89 years old.

The chronic cardiac conditions most commonly associated with the development of AF are rheumatic mitral valve disease, coronary artery disease, congestive heart failure, and hypertension. Noncardiac etiologies, often reversible, include hyperthyroidism, hypoxic pulmonary conditions, surgery, and alcohol intoxication. In less than 10 percent of cases no predisposing condition exists; this type of AF is called "lone AF."

One classification of AF includes acute AF (onset within 48 hours), paroxysmal AF (terminated spontaneously on at least one occasion), persistent AF (duration greater than 48 hours and has not terminated spontaneously), and permanent AF (resistant to pharmacological or electrical cardioversion). This evidence report addresses the patient who presents to a clinician for the first time with AF, whether it be persistent or paroxysmal.

The manifestations of AF can be divided into two categories: hemodynamic compromise and thromboembolic complications. Symptoms of hemodynamic compromise range from the classic complaint of irregular palpitations to the more insidious feeling of malaise. The risk of stroke is increased two- to five-fold in those with nonrheumatic AF.

In addition to treating underlying conditions, the management of AF can be divided into three areas: (1) ventricular rate control, (2) cardioversion of AF and subsequent maintenance of sinus rhythm, and (3) prevention of thromboembolism. For rate control, clinicians generally choose among digoxin, beta-adrenergic antagonists, and/or calcium-channel-blockers, without a clear consensus about which agent is the most effective.

Direct current cardioversion and numerous antiarrhythmic agents have been proposed for the cardioversion of AF. In addition, a number of pharmacological agents have been used to maintain sinus rhythm after successful cardioversion. The agents used in AF are classified by their electropharmacological actions: Class Ia (quinidine, procainamide, and disopyramide) and Ic (e.g., flecainide, propafenone) agents; Class II agents (e.g., propanolol, esmolol); Class III agents (e.g., amiodarone, sotalol, ibutilide, dofetilide); and Class IV agents (e.g., verapamil, diltiazem). Some agents have properties from different classes; particularly notable are the beta-adrenergic antagonist properties of propafenone and sotalol. The Class I and III agents are the most frequently used to convert to sinus rhythm or maintain sinus rhythm. Their use must be weighed against the risk of ventricular proarrhythmia. Again, no consensus exists regarding which strategy, electrical or pharmacological, or which agent is best.

The prevention of thromboembolic complications predominantly consists of antiplatelet therapy (e.g., aspirin, indobufen); anticoagulation (e.g., heparin, coumadin); or a combination (e.g., aspirin plus low-dose coumadin). Most clinical experts are convinced that warfarin is the most effective agent in the majority of patients, but many clinicians remain reluctant to use warfarin because of the increased risk of bleeding, particularly in the elderly. Also, it is thought that in lower-risk patients the benefits of anticoagulation may not outweigh the risks and costs, but no absolute risk threshold has been established.

Because of safety concerns, antiarrhythmic therapy generally is started in the hospital. For economic and patient convenience reasons, some clinicians are starting certain antiarrhythmic agents in an outpatient setting for some patients. No consensus exists regarding the safety of this practice, and information is needed to determine which agents and which patients are appropriate for outpatient initiation of antiarrhythmic therapy.

Echocardiography has been proposed to aid in the management of AF in three ways. First, echocardiography has been proposed to stratify patients as to risk of stroke. Second, echocardiography has been proposed to identify patients with dilated left atria, who may be less likely to respond to cardioversion attempts. Third, transesophageal echocardiography has been proposed to identify patients without left-atrial thrombus, who safely could undergo acute cardioversion without the conventional 3 weeks of precardioversion anticoagulation. Considerable uncertainty remains about the most appropriate role for echocardiography in the management of AF.

Thus, the management of patients with AF presents many options for the clinician. The initial choice of aggressive rhythm control versus appropriate rate control with treatment to prevent thromboembolism is being tested in a randomized controlled trial, but-even after the results of the trial are known-questions regarding the best strategy for a given patient likely will remain.

This report presents the results of our assessment of the evidence on key issues in the management of AF. The target study population is those with new onset non-postoperative AF, best approximating a typical outpatient presentation. The assessment is limited to first-line strategies, excluding invasive or emerging therapies that are applied more frequently to AF refractory to first-line therapies. Relevant key questions were identified (see Identifying the Questions, below), a systematic review of the evidence pertinent to these questions was performed, and evidence tables of the available information were constructed. Meta-analyses and decision analyses were performed where relevant. The overall objective of this report is to synthesize the evidence that should be used to guide clinicians in their management of patients with new onset AF.

We identified a core group of five clinically and/or methodologically oriented technical experts who provided extensive input throughout the project. This group included representatives from the American Academy of Family Physicians (AAFP) and a clinical expert identified through the American College of Cardiology (ACC) as well as cardiologists from Johns Hopkins Medical Institutions. Other technical experts aided in the identification of the relevant questions and served as peer reviewers of the evidence report. These technical experts included physicians; nurses; and representatives of professional organizations, Government agencies, health plans, and industry.

The target population consisted of ambulatory adult patients with new onset AF, defined as those who present for the first time with persistent or paroxysmal AF, regard¡less of whether the duration of the arrhythmia is known at the time of presentation. Many of the trials on management of AF included patients with atrial flutter, but data were inadequate to support separate conclusions about management of atrial flutter.

By giving a questionnaire to the above-mentioned technical experts, we identified the following key questions:

1. Which patients with new onset AF should receive attempts at cardioversion and which should receive only conservative treatment with rate control and thromboembolism prophylaxis?

2. What is the efficacy of electrical cardioversion alone compared with antiarrhythmic therapy alone compared with both together for patients with new onset AF?

3. What are the risks and benefits of each of the antiarrhythmic agents used for conversion of AF and/or the maintenance of sinus rhythm after successful cardioversion?

4. What types of therapy for AF can safely be given in an outpatient setting rather than in an inpatient setting?

5. What is the diagnostic value of tests, such as transesophageal echocardiography and transthoracic echocardiography, that can be used in the evaluation of patients with new onset AF?

In addition, two important supplementary questions, relevant to the key questions, were identified:

1. How does anticoagulation compare with aspirin in preventing thromboembolism in patients with AF?

2. How does each of the pharmacological agents used for rate control compare in efficacy?

The primary literature source was the CENTRAL database produced by the Cochrane Collaboration's international efforts to identify controlled clinical trials. The specific search terms included "atrial fibrillation" and "atrial flutter," both as subject headings and as text words. MEDLINE also was searched to identify recent publications.

We decided to limit retrieval to randomized controlled trials in order to focus on the studies that used the strongest study design. For the questions concerning outpatient strategies (key question 4) and echocardiography (key question 5), the search strategy retrieved very few citations. It was decided for these questions to conduct additional searches, expanding retrieval to studies other than randomized controlled trials.

Overall, 521 citations were identified, and 179 articles were deemed eligible for detailed review. The following criteria were used to exclude articles from detailed review: (1) article does not address management of AF or atrial flutter, (2) article does not include human data, (3) study included postoperative AF that could not be separated from data on non-postoperative AF, (4) adults are not part of the study population, (5) study contained no original data, (6) study lacked randomization, and (7) study does not enable the reader to separate subjects with AF or atrial flutter from those with other arrhythmias.

Each article meeting eligibility criteria was assessed for study quality, including representativeness of the study population; bias and confounding; description of therapy; outcomes and followup; and statistical quality and interpretation. Quantitative data were extracted on subject eligibility criteria, baseline subject characteristics that could influence outcomes (age, duration of atrial fibrillation or flutter, comorbidities, etc.), therapeutic protocols, the goal time of followup, outcomes, and adverse events. We extracted data separately for atrial fibrillation and for atrial flutter to address as pure a population of non-postoperative atrial fibrillation as possible.

Based on a preliminary review of the literature, the main questions identified as potentially appropriate for meta-analysis were acute pharmacological conversion, pharmacological maintenance of sinus rhythm, pharmacological heart rate control, and reduction in the rates of stroke associated with antithrombotic therapy. For the other questions that were not amenable to meta-analysis, we used decision analysis to synthesize the evidence.

Evidence tables were constructed for the presentation of data on pharmacological conversion, maintenance of sinus rhythm, heart rate control, and the outcomes from antithrombotic therapy. The questions with results appropriate for pooling, defined after the literature was reviewed, involved articles that met the following criteria: (1) had evidence from randomized clinical trials, (2) had an adequate number of trials with qualitatively similar subjects, and (3) had an adequate number of trials that evaluated similar outcomes. For each of the outcome measures, the treatment effect was expressed as an odds ratio; for example, the odds of converting to sinus rhythm on one drug compared with the odds of converting on another drug. In addition, we converted the pooled odds ratios into the number needed to treat (NNT) data for our summary sections to facilitate interpretation of the results.

A qualitative assessment of combinability was performed before a quantitative assessment of heterogeneity could be done. After review of the designs and results of the trials, we decided that the data from the trials using calcium-channel-blockers and digoxin as the comparison agents for antiarrhythmic drugs could be combined with trials that used placebo controls. Ibutilide and dofetilide were the only other pharmacological agents deemed similar enough to allow combination for pooling.

Estimates of the relative rates of the outcomes of interest were pooled using standard methods for combining odds ratios for the outcomes of conversion to sinus rhythm, maintenance of sinus rhythm, stroke, peripheral embolism, major bleeding, minor bleeding, and mortality. Studies were weighted on the basis of the precision of the estimate within each study. A fixed-effects model was used to summarize the evidence, as indicated by a screen for quantitative heterogeneity. In two instances with respect to the conversion data-propafenone versus control treatment and amiodarone versus control treatment-a random-effects model was used because of some quantitative heterogeneity of the data.

The impact of the baseline characteristics of the subjects within each study on the aggregate study outcomes was evaluated. Although only large subgroup effects could be detected because of the small numbers of studies, we thought it was important to explore certain clinically relevant subgroups.

Realizing that it is difficult to categorize the strength of evidence, we felt it was important to facilitate interpretation of the odds ratio (OR). The concept of "strength of evidence" depends on the estimated magnitude of effect, the precision of that estimate, and our confidence that the true effect is different from zero. Quantitatively, all of these dimensions are captured by the point estimate and confidence interval (CI), but it may be difficult to interpret results presented only in that fashion. Therefore, we provided descriptors of the strength of evidence. For the evidence on conversion and maintenance of sinus rhythm, an OR greater than 1.0 represents a higher odds of conversion or maintenance of sinus rhythm compared with the comparison group. For the evidence on anticoagulation and antiplatelet agents, an OR less than 1.0 represents a lower odds of adverse events such as stroke and bleeding compared with the comparison group. We chose the following categorization of strength of evidence by noting the placement of the point estimate and the width of the CI surrounding it:

When the point estimate was less than 1.0, we termed this negative efficacy and used the same categorization of strong, moderate, and suggestive evidence based on the CI.

When the point estimate was greater than 1.0, we termed this negative efficacy and used the same categorization of strong, moderate, and suggestive evidence based on the CI.

In some situations, the above categorizations did not completely capture some element of the results that was important for interpretation, particularly in cases with large point estimates and/or very wide confidence intervals. In those instances, we added additional descriptive text.

A Monte Carlo multistate transition model was constructed to evaluate the cost-effectiveness questions not directly or fully addressed in the clinical trial literature.

First, to address the question of what combination of electrical or pharmacological intervention is most cost-effective for cardioversion and subsequent maintenance of sinus rhythm (key questions 2 and 3), 17 strategies were evaluated.

(B-G) Pharmacological conversion using one of quinidine, flecainide, propafenone, amiodarone, sotalol, or ibutilide without subsequent pharmacological therapy.

(H-L) Pharmacological conversion with continued therapy using one of quinidine, flecainide, propafenone, amiodarone, or sotalol (ibutilide cannot be used for continued therapy).

(M-Q) Electrical cardioversion with subsequent maintenance of sinus rhythm using one of the five agents listed above.

Second, to address the question of what antithrombotic therapy is most cost-effective for the prevention of stroke (supplementary question 2), strategies employing aspirin and warfarin were evaluated.

Third, by comparing the most cost-effective strategies from each of the above two analyses, the overall question of whether to attempt cardioversion or to treat conser¡vatively with antithrombotic therapy (key question 1) was addressed.

Fourth, the implications of the costs of inpatient versus outpatient initiation of antiarrhythmic therapy on the cost-effectiveness of the strategies for cardioversion and subsequent maintenance of sinus rhythm (key question 4) were evaluated.

Finally, the question regarding the use of echocardiography to guide antithrombotic therapy (key question 5) was addressed. Strategies employing transesophageal echo-cardiography or transthoracic echocardiography to guide decisions about antithrombotic therapy were compared with the strategy employing either aspirin in all patients or warfarin in all patients.

Questions regarding pharmacological conversion of AF and maintenance of sinus rhythm, pharmacological rate control, and antithrombotic therapy were addressed directly in the clinical trial literature. Other areas were addressed only indirectly in this body of literature and needed supplementation from observational studies. Two large randomized clinical trials are under way that will address overall rate versus rhythm control and value of transesophageal echocardiography in guiding timing for acute cardioversion.

This review of the literature identified 46 randomized clinical trials on acute cardioversion of AF that included 18 antiarrhythmic agents, listed according to the Vaughan-Williams classification system.

This review of the literature identified 29 randomized clinical trials on maintenance of sinus rhythm that included 10 antiarrhythmic agents:

The overall quality scores ranged from a low of 36 percent to a high of 95 percent. Of note, however, only nine studies had an overall quality score less than 50 percent. The studies had the largest variability in terms of representativeness of the population, with a range of scores from 0 to 100 percent. All of the other four domains of quality also had considerable variation in scores with the following ranges: bias and confounding, 17 to 100 percent; therapy description, 25 to 100 percent; outcomes assessment, 25 to 90 percent; and statistical analysis, 0 to 100 percent.

Our review of the literature supported the combination of placebo, verapamil, diltiazem, and digoxin into a control treatment group. The bulk of the evidence reported on comparisons of a given antiarrhythmic agent with one of these control groups.

The evidence regarding adverse events, including ventricular arrhythmias, was reported inconsistently and sporadically, limiting its usefulness in this review in both quantitative and qualitative terms.

The evidence on clinical predictors of success in pharmacological conversion was too sparse and disparate to summarize meaningfully.

The antiarrhythmic agents with the strongest evidence of efficacy and largest treatment effect sizes for conversion of AF, compared with control treatment, were flecainide (OR 24.7, CI 9.0 to 68) and ibutilide/dofetilide (OR 29.1, CI 9.8 to 86). Assuming a control treatment cardioversion rate of 30 percent and using the upper and lower 95 percent confidence limits on these odds ratios, the estimated number of subjects needed to be treated (NNT) in order to see a benefit relative to control treatment ranged from 1.5 to 2.0 for either flecainide or ibutilide/dofetilide.

Strong evidence of efficacy compared with control treatment also existed for propafenone (OR 4.6, CI 2.6 to 8.2), with a fairly large treatment effect size. Again assuming a control treatment conversion rate of 30 percent and using the upper and lower 95 percent confidence limits on this odds ratio, the estimated NNT in order to see a benefit relative to control treatment ranged from 2.0 to 4.5 for propafenone.

We should note here that the NNT is sensitive to the baseline assumption of the control treatment cardioversion rate. For example, if we varied the control treatment cardioversion rate from 5 percent to 75 percent for propafenone, the NNT ranges became 4.0 to 14.3 for a 5 percent conversion rate and 4.7 to 7.5 for a 75 percent conversion rate.

There was moderate evidence of efficacy with a modest treatment effect size for quinidine compared with control treatment for cardioversion of AF (OR 2.9, CI 1.2 to 7.0). Assuming a control treatment cardioversion rate of 30 percent and using the 95 percent confidence limits, the range of number of subjects NNT with quinidine in order to see a benefit relative to control treatment is 2.0 to 25.0.

Disopyramide (OR 7.0, CI 0.3 to 152) and amiodarone (OR 5.7, CI 1.0 to 33.4) had suggestive evidence of benefit for acute conversion of AF relative to control treatment. The comparison of amiodarone with control treatment, however, was problematic because of the substantial qualitative and quantitative heterogeneity between the combined studies.

The evidence was suggestive of negative efficacy of sotalol (OR 0.4, CI 0.0 to 3.0) compared with control treatment for acute cardioversion of AF.

Minimal evidence existed for agents in Classes II, IV, and V, and in the miscellaneous category.

All of the following antiarrhythmic agents had strong evidence of efficacy and fairly large treatment effect sizes for maintenance of sinus rhythm in AF: quinidine (OR 4.1, CI 2.5 to 6.7), disopyramide (OR 3.4, CI 1.6 to 7.1), flecainide (OR 3.1, CI 1.5 to 6.2), propafenone (OR 3.7, CI 2.4 to 5.7), and sotalol (OR 7.1, CI 3.8 to 13.4). Our review did not support any definitive ranking of these five agents for their efficacy in maintaining sinus rhythm. Using an assumption of 30 percent recurrence of AF by 6 months in the control treatment group and the upper and lower 95 percent confidence limits of these odds ratios, the NNT range for each agent in order to see a benefit relative to control treatment is as follows: quinidine 2.3 to 4.6; disopyramide 2.2 to 9.4; flecainide 2.3 to 10.9; propafenone 2.4 to 4.8; and sotalol 1.8 to 3.1.

Evidence was scarce on use of amiodarone for maintenance of sinus rhythm after cardioversion of AF. One trial, presenting only interim results, had moderate evidence of amiodarone efficacy relative to disopyramide. Because disopyramide had strong evidence of efficacy relative to control treatment, we chose to classify the evidence on amiodarone as "potentially strong evidence" relative to control treatment. Notably, however, we identified no trials of amiodarone compared with control treatment for maintenance of sinus rhythm.

Clinical trials now in progress will help address the paucity of data on amiodarone for maintenance of sinus rhythm.

Minimal evidence existed for N-acetylprocainamide and agents in class IV and the miscellaneous category.

One attempt at electrical cardioversion with subsequent pharmacological maintenance therapy is cost-effective compared with conservative antithrombotic therapy alone for all patients 55 years old or older, regardless of risk factors.

Overall, 45 trials were identified that evaluated 17 agents. The design and outcomes of these trials were too disparate for meta-analysis. Generally, the studies were of high quality in their description of the outcomes and for having objective measures of the outcomes. The studies were weaker in their description of the study groups; it was not always possible to determine if the groups were similar.

Compared with placebo or digoxin, the calcium-channel-blockers diltiazem and verapamil were effective in reducing the heart rate at rest and during exercise in patients with AF.

Compared with placebo or digoxin, beta-blockers were effective in reducing the heart rate during exercise in patients with AF. However, exercise tolerance was decreased on beta-blockers in a number of the studies. The effect of beta-blockers on resting heart rate was inconsistent, with only about half of the studies demonstrating better control with beta-blockers than placebo.

The evidence was not convincing for the use of digoxin, particularly during exercise.

For antithrombotic therapy, 11 trials were identified that evaluated four medications.

Strong evidence for the prevention of stroke existed for use of warfarin (OR 0.30, CI 0.19 to 0.48) compared with placebo. However, the evidence was suggestive for a higher bleeding rate on warfarin (OR 1.90, CI 0.89 to 4.04) than placebo. For every 1,000 patients with AF who are treated with warfarin for 1 year, 30 strokes are prevented at the expense of six major bleeds.

The evidence for efficacy in preventing stroke was moderately strong for aspirin (OR 0.65; CI 0.43 to 0.99) compared with placebo. The evidence for an increased odds of major bleeding associated with aspirin (OR 0.81, CI 0.37 to 1.77) compared with placebo was inconclusive. For every 1,000 patients with AF who are treated with aspirin for 1 year, 12.5 strokes are prevented.

The evidence from trials directly comparing warfarin and aspirin did not permit strong conclusions.

The evidence was inconclusive about the use of a combination of low-dose warfarin with aspirin and about the use of low molecular weight heparin or induprofen for stroke prevention.

Our decision analysis estimated that for patients at low risk of stroke (~1 percent/year), aspirin is the most cost-effective therapy. For patients with a high risk of stroke (~10 percent/year), warfarin is projected to be the most cost-effective strategy. For those with an intermediate risk of ischemic stroke (~3-6 percent/year), aspirin therapy is estimated to be the most cost-effective if quality of life is assumed to be decreased by taking warfarin. If no decrease in quality of life is assumed, warfarin is estimated to be the most cost-effective strategy for the intermediate-risk groups.

Our comprehensive literature search did not identify any randomized controlled clinical trials that addressed directly the question of what types of therapy for AF can safely be given in an outpatient setting (key question 4).

Based on the decision analysis, if antiarrhythmic therapy can be started safely as an outpatient, the cost-effectiveness of an attempt of electrical cardioversion with pharma¡cological maintenance therapy is improved considerably. However, more research is needed in this area to determine the safety of this practice.

From our comprehensive literature search of clinical trials, no completed trials were identified that directly addressed the clinical utility of echocardiography in the manage¡ment of AF (key question 5).

Of the 46 studies of acute cardioversion identified in our comprehensive search of clinical trials, only 9 gave information relating left-atrial diameter from echocardiography to success of cardioversion. Although the results were not consistent, they suggested an inverse association between left-atrial diameter and successful cardioversion.

Based on the decision analysis, transthoracic echocardiography is projected to be a cost-effective test for guiding decisions about the choice of antithrombotic treatment in most patients without risk factors for thromboembolism. Transesophageal echocardi¡ography is projected to be a cost-effective test for guiding decisions about the choice of antithrombotic treatment in patients having risk factors for thromboembolism.

Two ongoing clinical trials are addressing two of the key questions. One is addressing the decision about aggressive rhythm control versus conservative rate control with anti¡thrombotic therapy. The other is evaluating the use of transesophageal echocardiography in guiding the timing of acute cardioversion of AF.

Future research should include randomized controlled trials and other types of studies:

Many of these recommended research priorities-and specifically the second one-are beginning to be addressed in trials now being conducted.