Atrial fibrillation (AF) is the most commonly sustained cardiac arrhythmia observed in clinical practice with a prevalence of over 2,200,000 adults in the US.1 The prevalence of AF increases with age, from <1% among persons <60 years of age, to approximately 10% among those aged ≥80 years.2 AF is associated with an increased long-term risk of stroke, heart failure, and all-cause mortality.3,4 The age-adjusted death rate from AF, although relatively low at just over 9,000 per year, has been increasing by an average of approximately 5% per year since 1980, and hospitalizations for AF have increased two- to three-fold in recent years.1,4,5 Therefore, AF is not a benign condition.6 Indeed, the personal and public health burden of AF is enormous and is expected to continue to increase over the next decade as the population ages.
For patients experiencing a first episode of AF, initial therapy may be to restore and maintain sinus rhythm. However, physicians are faced with a decision when AF becomes paroxysmal or persistent, between a strategy of rhythm-control and one to control the ventricular rate response to AF.6 Other than the importance of anticoagulation for the prevention of stroke,7 the relative benefit of these two strategies was unknown until the results of the recently completed Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) trial became available.8-10
As with any clinical trial, the implications of the results for routine clinical practice depend on careful consideration of study end-points, patient inclusion and exclusion criteria, and treatment protocols. This article discusses the implications of the results of the main trial and several subsequent substudies.
The Main Trial
The AFFIRM trial was a randomized, parallel-group study that compared two treatment strategies for AF:
- rhythm control and anticoagulation; and
- rate control and anticoagulation.
A composite secondary end-point included death, disabling stroke, disabling anoxic encephalopathy, major bleeding, and cardiac arrest.8,10 Follow-up was for a maximum of six years.
Several study design criteria should be kept in mind when considering the clinical implications of the AFFIRM trial. First, this study was a comparison of two treatment strategies, and not a simple comparison of one drug versus another. Furthermore, to minimize cross-overs between the randomized treatment arms, investigators were allowed to use any available drug consistent with randomization sequentially and/or in combination. Also, since the trial was extended over many years, investigators were allowed to use any reasonable therapy that was introduced during the course of the study. Second, the primary end-point of the AFFIRM trial was mortality from any cause. Because the therapies within each treatment were not blinded, it was essential that the end-point be definitive and unambiguous.
Patients were eligible for inclusion in the AFFIRM trial if they had AF and at least one other condition associated with a high risk for stroke and death (age ≥65, congestive heart failure, hypertension, diabetes, poor left ventricular function, a large left atrium, or prior stroke or transient ischemic attack).8 While a history of qualifying episodes of AF prior to enrollment was specified by the protocol, the presence of persistent AF at randomization was not required.8
A total of 4,060 patients were randomized and followed for up to six years (mean 3.5 years).9,10 The average age was 69.7 years, 61.7% were male, and 11.4% were members of an ethnic minority group. Of the total population, 71% had hypertension, 38% had a history of coronary artery disease (CAD), and 23% had a history of congestive heart failure. The frequency of symptomatic episodes of AF in the six months prior to randomization was quite variable, ranging from more than once per day in 11% of patients to once in six months in 24%.The duration of the qualifying episode was also quite variable.While more than two-thirds of patients had continuous AF for more than 48 hours, the duration was <6 hours in 8% and >6 months in 6%.
Before randomization, 18% of patients had failed antiarrhythmic drug (AAD) therapy - most (69%) had failed only one drug.9 Quinidine and sotalol were the two most commonly failed AADs prior to randomization. Other medications taken by these patients prior to randomization included digoxin in 53%, a diuretic in 43%,a beta-blocker in 43%, and an ACE inhibitor in 39%.
The rate control and rhythm control groups were balanced according to these baseline characteristics,9,10 which are consistent with, and representative of, the majority of patients with AF.11,12
In the rate control group, the therapeutic goal was a heart rate (80 beats per minute (bpm) at rest and not higher than 110bpm during a six-minute walk test). Acceptable drugs per protocol included beta-blockers, calcium channel-blockers (verapamil and diltiazem), digoxin, or their combinations. Beta-blockers were used as initial therapy in nearly half of patients in the rate control group. However, changes in therapy were frequent. At the five-year visit, 34.6% of patients were in sinus rhythm and more than 80% of those in AF had adequate heart rate control.10 During the study, 248 (14.1%) patients crossed over to the rhythm control strategy most commonly due to uncontrolled symptoms of AF or congestive heart failure. Subsequently, 86 of these patients crossed back to the rate control arm.
The AAD or combinations used in the rhythm control group was chosen by the treating physician from among the currently available agents. Cardioversion could be included as necessary. More than two-thirds of patients in the rhythm control group began therapy with either amiodarone or sotalol.10 The prevalence of sinus rhythm in the rhythm control arm was 82.4%, 73%, and 62.6% at one, three, and five years, respectively. During the study, 594 (29.2%) patients randomized to rhythm control crossed over to the rate control strategy, a significantly (P<0.001) higher cross-over rate as compared with the rate-control group. However, 61 of these patients subsequently crossed back to the rhythm control arm by the end of the study. Failure to maintain sinus rhythm and drug intolerance were the main reasons for discontinuation.
Continuous warfarin use was mandated in the rate control group, whereas in the rhythm control group continuous anticoagulation was encouraged, but could be stopped at the physician's discretion if sinus rhythm had been maintained for at least four, and preferably 12, consecutive weeks.8,10 During the study, the overall percentage of patients receiving warfarin was 85% to 95% in the rate control arm.10 After the first four months of the trial, there was a decline in warfarin use in the rhythm control arm, although the overall proportion of patients receiving warfarin remained at approximately 70%.A total of 62.3% of international normalized ratio (INR) values measured during the course of the study were within the recommended range (2.0 to 3.0).
A total of 356 patients in the rhythm control arm and 306 patients in the rate control arm died during the course of the study.10 Overall survival at one, three, and five years was 96%, 87%, and 76% in the rhythm control group and 96%, 89%, and 79% in the rate control group. The between-group difference in mortality was not statistically significant (P=0.08).The secondary composite end-point of death, disabling stroke, major bleeding, or cardiac arrest was also not significantly different between the groups (P=0.283). Ischemic strokes occurred in 77 and 80 patients in the rate control and rhythm control groups, respectively, for an annual of rate of approximately 1% per year in each group. Most strokes occurred in patients who were not taking warfarin or who had an INR <2.0. The incidence of other adverse events was low. However, compared with the rate-control group, pulmonary events, gastrointestinal complaints, bradycardia and prolongation of the QT interval that necessitated discontinuation of a drug were all higher in the rhythm control group. The number of patients needing hospitalization during the follow-up period was significantly greater in the rhythm control group (80.1%) than in the rate control group (73.0%, P<0.001).
Therefore, the results of the main AFFIRM trial showed that management of AF with a rhythm control strategy offers no clear survival advantage over a rate control strategy in patients who are at a high risk for stroke and death. Furthermore, a rate control strategy to manage AF may have potential advantages, such as a lower risk of adverse drug effects and hospitalizations. Finally, the importance of continuous anticoagulation in all at-risk patients regardless of treatment strategy was again demonstrated by the AFFIRM trial.
A retrospective analysis was subsequently performed to evaluate the relationships between sinus rhythm, treatment, and survival in the AFFIRM trial.13 In this study, an 'on-treatment' analysis of the relationship of survival to cardiac rhythm and treatment as a function of time was reported. Thus, patients were analyzed according to the treatment they actually received, rather than the treatment assigned through randomization (as is done in an intention-to-treat analysis). Because a patient's rhythm and treatment could change over time, as happened in the AFFIRM trial, time-dependent covariates were used to allow analysis of factors that changed during follow-up.
The time-dependent covariates that were significantly associated with survival were sinus rhythm, warfarin use, digoxin use, and antiarrhythmic drug use. The presence of sinus rhythm and the use of warfarin were associated with 47% and 50% reductions in mortality, as manifested by the hazard ratios of 0.53 and 0.50, respectively. Conversely, the use of digoxin and antiarrhythmic drugs were associated with a 42% and a 49% increase in mortality, respectively. To summarize, the presence of sinus rhythm was associated with a 47% decreased risk of death,warfarin use was also associated with a 50% decreased risk of death, and among rate-control drugs, digoxin was associated with a 42% increased risk of death.
Several hypotheses may explain these results with regard to the adverse effect of antiarrhythmic drugs. Perhaps most important is that antiarrhythmic drugs may have both beneficial and detrimental effects. Indeed, when sinus rhythm was included in the survival analysis as a separate factor, the sinus rhythm variable expressed the beneficial effect of antiarrhythmic drug therapy, while the antiarrhythmic drug variable expressed only the detrimental effects. In this analysis, antiarrhythmic drugs are associated with increased mortality. Conversely, when sinus rhythm as a separate factor was removed from the analysis, the beneficial effect of antiarrhythmic drugs (specifically the maintenance of sinus rhythm) offset the detrimental effects of toxicity, morbidity, and mortality. Thus, in this instance, antiarrhythmic drugs are not associated with increased mortality.
Despite these strong conclusions, this sub-study does have limitations. First, because of its retrospective and non-randomized nature, the analysis may suffer from biases and other unidentified confounders. Second, the analysis did not determine whether sinus rhythm is an important determinant of survival or simply a marker for other factors associated with survival. It should be noted that, because there were protocol-defined specifications for antiarrhythmic drug use and exclusions (e.g. patients with CAD could not receive class IC drugs), comparisons of outcomes for different drugs cannot be made.
This study, however, has several potentially important implications. First, the association of sinus rhythm, but not antiarrhythmic drugs, with improved survival may reflect that currently available antiarrhythmic drugs are neither highly efficacious nor completely safe. Second, the results suggest that if an effective treatment for maintaining sinus rhythm with minimal adverse affects were available, it might be expected to improve survival. A subsequent analysis sought to compare cause-specific modes of death in the rate control and rhythm control groups in the AFFIRM trial.14
In the rhythm control group, 129 (9%) patients died of a cardiac cause compared with 130 (10%) in the rate control group. The rates of both arrhythmic and non-arrhythmic cardiac deaths were similar. The number of vascular deaths were also similar (35 and 37) Non-cardiovascular deaths comprised 47.5% of all deaths in the rhythm control group and only 36.5% of all deaths in the rate control group (P=0.0008). These differences were due to pulmonary and cancer-related deaths. Thus, rhythm control may be associated with an increased risk of non-cardiovascular mortality but not cardiac or vascular mortality.
Conclusions and Implications
The AFFIRM trial clearly demonstrated that a rhythm-control strategy does not provide a survival advantage over a rate-control strategy in patients with one or more risk factors for stroke or death. The results of a smaller (522 patients) European study15 published simultaneously with the AFFIRM trial are consistent with those from the AFFIRM trial. Although the primary end-point (composite of mortality and serious events) and inclusion criteria (persistent AF after a previous electrical cardioversion) differed somewhat from those of the AFFIRM trial, the Rate Control Versus Electrical Cardioversion for Persistent Atrial Fibrillation (RACE) trial found that rate control was not inferior to rhythm control for the prevention of death and morbidity from cardiovascular causes in patients with persistent AF.
What are the clinical practice implications of these two trials? Although the population of patients enrolled in the AFFIRM trial was similar to that of previous studies, the requirement that a patient have a risk factor for stroke or death for inclusion could mean that the average AFFIRM patient had serious cardiac disease or other co-existing illnesses. Similarly, patients in the RACE trial had persistent AF refractory to electrical cardioversion. Therefore, the results of these studies cannot be generalized to younger patients with 'lone' AF, for example, and others with no risk factor for stroke. Furthermore, individuals with severe symptoms were not eligible for inclusion in the AFFIRM trial. It is just this population that a rhythm-control approach is most suited to.
Finally, since all patients in the RACE trial, and the majority of those in the AFFIRM trial had already had an episode of AF, they were at risk of a recurrence. Thus, it is unclear whether the results of these two studies can be generalized to patients with a first episode of AF.
Therefore, while reversion to, and maintenance of, sinus rhythm can no longer be considered imperative, there remains a population of patients with AF for whom rhythm control still is appropriate.