Author + information
- Received August 12, 2003
- Revision received October 29, 2003
- Accepted November 20, 2003
- Published online April 7, 2004.
- Brian Olshansky, MD, FACC*,* (, )
- Lynda E. Rosenfeld, MD, FACC†,
- Alberta L. Warner, MD, FACC‡,
- Allen J. Solomon, MD, FACC§,
- Gearoid O'Neill, MD, FACC∥,
- Arjun Sharma, MD, FACC∥,
- Edward Platia, MD, FACC¶,
- Gregory K. Feld, MD, FACC#,
- Toshio Akiyama, MD, FACC**,
- Michael A. Brodsky, MD, FACC††,
- H.Leon Greene, MD, FACC‡‡,
- AFFIRM Investigators§§
- ↵*Reprint requests and correspondence:
Dr. Brian Olshansky, University of Iowa Hospitals, 200 Hawkins Drive, Iowa City, Iowa 52242, USA.
Objectives We sought to evaluate approaches used to control rate, the effectiveness of rate control, and switches from one drug class to another in the Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) study.
Background The AFFIRM study showed that atrial fibrillation (AF) can be treated effectively with rate control and anticoagulation, but drug efficacy to control rate remains uncertain.
Methods Patients (n = 2,027) randomized to rate control in the AFFIRM study were given rate-controlling drugs by their treating physicians. Standardized rate-control efficacy criteria developed a priori included resting heart rate and 6-min walk tests and/or ambulatory electrocardiographic results.
Results Average follow-up was 3.5 ± 1.3 years. Initial treatment included a beta-adrenergic blocker (beta-blocker) alone in 24%, a calcium channel blocker alone in 17%, digoxin alone in 16%, a beta-blocker and digoxin in 14%, or a calcium channel blocker and digoxin in 14% of patients. Overall rate control was achieved in 70% of patients given beta-blockers as the first drug (with or without digoxin), 54% with calcium channel blockers (with or without digoxin), and 58% with digoxin alone. Adequate overall rate control was achieved in 58% of patients with the first drug or combination. Multivariate analysis revealed an association between first drug class and several clinical variables. There were more changes to beta-blockers than to the other two-drug classes (p < 0.0001).
Conclusions Rate control in AF is possible in the majority of patients with AF. Beta-blockers were the most effective drugs. To achieve the goal of adequate rate control in all patients, frequent medication changes and drug combinations were needed.
The general approaches used to manage atrial fibrillation (AF) include: 1) ventricular rate control, and 2) rhythm control (attempts to maintain sinus rhythm [SR]), both in combination with anticoagulation. Despite the large number of patients treated with rate-controlling therapies for AF, the best pharmacologic approach remains unknown. Beta-adrenergic blockers (beta-blockers), calcium channel blockers, and digoxin have been compared in small studies (1–29), but no large trial has determined the optimal pharmacologic method to control the ventricular response to AF in ambulatory patients.
The Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) study was a large, multicenter, randomized trial sponsored by the National Heart, Lung, and Blood Institute. This trial compared rate-control therapy with rhythm-control therapy to treat AF in a population at high risk of stroke or death (30–34). The primary goal was to determine which approach was associated with a better survival outcome. The AFFIRM study showed that rate control was an acceptable, if not the preferable, option (33). A substudy of this trial offered the opportunity to evaluate and compare several drug classes for the long-term ventricular rate control of AF.
This investigation assessed the effectiveness of ventricular rate-control approaches in AF in a large, general population of predominantly elderly patients. Data on all AFFIRM study patients assigned to the rate-control arm were reviewed to determine the effectiveness of rate control by the drug classes chosen by the treating physician.
The overall AFFIRM study was approved by the AFFIRM Steering Committee, the Institutional Review Board of the University of Washington (Clinical Trial Center), and by each participating site's local Institutional Review Board. All patients gave informed, written consent to participate in the AFFIRM study.
Definition of rate control
Prespecified criteria for adequate heart rate (HR) control in AF are outlined in Table 1. Rate control was assessed only if the patient was in AF at the time of evaluation. If a patient in the rate-control arm was in SR at a specified visit, rate control was not assessed and not recorded at that visit. Adequate rate control required the initial resting HR to be ≤80 beats/min. If that goal was achieved, a standard 6-min walk test (35)or 24-h ambulatory electrocardiographic recording (Holter monitor) was required to confirm adequate rate control. The approach used was left to the discretion of the treating physician. If both Holter monitoring and a 6-min walk test were performed, rate-control criteria had to be met for both tests. Overall rate control required evidence for resting and exercise rate control, as defined. Tests were repeated until rate control was achieved or when the patient had a change in status or change in drug dosage. Further drug titration was based on the results of these studies. The drug dose was adjusted and optimized by the treating physician in an attempt to achieve the preselected rate-control guidelines. If adequate rate control could not be achieved at a low dosage of medication, the dose was titrated upward, or a different medication was used and titrated in an attempt to achieve an acceptable rate response. If adequate rate control could still not be achieved despite a maximally tolerated drug dose, the treating physician could add another drug. A medication was stopped if the patient developed adverse effects, such as worsening congestive heart failure (HF) or hypotension. If a pharmacologic approach was not effective, the treating physician could proceed with atrioventricular (AV) junctional ablation, as defined in the protocol (31).
Patients were evaluated at two months using the same methods and criteria. The resting HR was determined at each routine clinic visit (every four months). A 6-min walk test or Holter monitoring was to be performed at a minimum of once a year, even if no changes had been made in the drug regimen and no changes were noted in the patient's clinical status.
Choice of drugs
Patients in the rate-control arm of the AFFIRM study were treated with an AV nodal blocking drug chosen by their treating physician. Drug choice was not randomized and was not expected to be distributed equally across all patient types or conditions. Treatment included beta-blockers, calcium channel blockers, or digoxin, alone or in combination.
The primary end point of this study was adequate “overall” pharmacologic rate control at rest and with exercise while taking the initial drug class chosen, after dose optimization by the treating physician. The adequacy of rate control was determined sequentially over the entire period of patient follow-up, but the specific end points of this substudy were: 1) achievement of adequate rate control, as defined earlier, with the first drug class chosen for each patient; and 2) the rate of switching from the first drug class chosen to another drug class.
Only drug classes (not individual drugs) administered during the last follow-up period were reported on the data forms submitted to the Clinical Trial Center. For example, only the category of beta-blockers would have been reported, not propranolol, metoprolol, and the like. Drug doses and start/stop dates were not reported. The rate-control drug classes were digoxin, beta-blockers, and calcium channel blockers.
Patients were classified by the drug group initially administered after randomization. Rate control for the initial drug class chosen for a patient was defined as the last determination of rate control while taking that drug before discontinuing the drug or adding a new drug. Patients who crossed over to rhythm control (248 of 2,027 patients) (33)were censored from further analysis at the time of crossover.
Baseline comparisons among all four major drug treatment groups (digoxin alone, beta-blockers with or without digoxin, calcium channel blockers with or without digoxin, and beta-blockers plus calcium channel blockers with or without digoxin) were performed using two-way chi-square tests for homogeneity for dichotomous variables or analysis of variance for continuous variables. Drug discontinuation was assessed using Kaplan-Meier time-to-event analysis and log-rank statistics.
Multivariate analyses evaluated the association between baseline clinical factors and the initial choice of drugs. These analyses were performed using linear model techniques generalized to categorical dependent variables (PROC CATMOD in SAS, Cary, North Carolina). For these analyses, only main effects were included, and a stepwise backward selection method was used. The criteria for remaining in the model was p < 0.01, as well as the factors associated with the success or failure of the first drug class chosen. The factors tested included age ≥65 years, gender, ethnicity, primary cardiac diagnosis, duration of qualifying AF, history of pulmonary disease, coronary artery disease, congestive HF, hypertension, diabetes, angina pectoris, myocardial infarction (MI), previous coronary artery bypass graft surgery or revascularization procedure, hepatic or renal disease, qualifying episode being the first episode of AF, frequency of episodes of AF, previous drug failure, left ventricular ejection fraction, left atrial size, and baseline HR.
Differences in mortality and long-term symptoms among drug classes could not be compared because patients were switched to different drug classes frequently throughout the study. Comparisons of rate control across different groups could not be performed for the same reasons; thus, this article is primarily a descriptive report. Furthermore, this substudy did not have sufficient power to detect differences in mortality among the different drug treatment groups.
This evaluation included the 2,027 patients randomized to the rate-control arm of the AFFIRM study (Table 2). Average follow-up was 3.5 ± 1.3 years.
Patient demographics and clinical characteristics
Before entering the AFFIRM study, 53% of patients had received digoxin, 40% had received a calcium channel blocker, and 42% a beta-blocker.
Clinical characteristics are listed in Table 3, as categorized by the drug class chosen initially. Significant differences existed among drug classes with regard to gender, coronary artery disease, angina pectoris, MI, coronary artery bypass graft surgery, interventional coronary procedure, hypertension, cardiomyopathy, pulmonary disease, baseline HR, SR at randomization, previous drug use, and ejection fraction.
Initially, 485 patients (24%) were treated with a beta-blocker alone, 344 (17%) with a calcium channel blocker alone, and 315 (16%) with digoxin alone. A drug combination was used frequently—a beta-blocking drug and digoxin in 292 (14%) and a calcium channel blocker plus digoxin in 287 (14%). Other drugs, other combinations, and no therapy accounted for 154 (8%), and missing data accounted for 150 (7%), yielding a total of 304 patients (15%).
The multivariate analysis revealed the following items to be associated with the first drug class used for rate control: gender (p = 0.0001), history of coronary artery disease (p < 0.0001), pulmonary disease (p < 0.0001), congestive HF (p < 0.0001), hypertension (p = 0.0001), qualifying episode being the first episode of AF (p = 0.005), and baseline HR (p = 0.006). Patients were likely to be prescribed the drug being administered before randomization. Beta-blockers were more likely to be used in patients having coronary disease, angina, previous MI, and previous coronary interventions. Calcium channel blockers were more likely to be used in patients with pulmonary disease and in females. Digoxin was used more often in patients with cardiomyopathy and in non-whites (Table 3).
Initial rate control at rest
Rate control at rest was achieved with a beta-blocker (with or without digoxin) in 75% of patients, with digoxin alone in 68%, and with a calcium channel blocker (with or without digoxin) in 66% (Table 4). Similarly, rate control at rest was achieved with a beta-blocker alone in 68% and with a calcium channel blocker alone in 60%.
Initial rate control during exertion
Beta-blockers (with or without digoxin) provided rate control during exertion in 85%, as compared with calcium channel blockers (with or without digoxin) in 72% and digoxin alone in 70%. Similar findings were noted with beta-blockers alone (72%) and calcium channel blockers alone (58%), although digoxin alone was associated with good rate control with exertion (70%).
Initial overall rate control
The overall goal of rate control was to achieve a satisfactory HR both during rest and with exertion. This goal was reached in 70% of patients treated with a beta-blocker (with or without digoxin), as compared with 54% treated with a calcium channel blocker (with or without digoxin) and 58% with digoxin alone. The use of a beta-blocker alone showed overall rate control similar to that of digoxin alone (59% vs. 58%), with only 38% adequately controlled with a calcium channel blocker alone.
Rate control over time
Adequate overall rate control was achieved in 58% of patients with the first drug or combination (Table 4). A total of 635 patients (37%) required a change in or the addition of another drug over the course of five years. Changes in rate-controlling drug therapy in the first year were common: 23% of patients switched from calcium channel blockers to beta-blockers, 19% switched from beta-blockers to calcium channel blockers, and 34% began taking either beta-blockers or calcium channel blockers after having taken only digoxin. Over time, more patients taking digoxin or a calcium channel blocker were changed to another drug, as compared with patients taking beta-blockers (p < 0.0001) (Fig. 1). Overall rate control, rate control at rest, and rate control with exertion improved over time (Fig. 2).
Rate control by HF status
Among patients without HF symptoms and with a left ventricular ejection fraction >0.40, adequate rate control at rest and exertion, as defined previously, was 55% with a beta-blocker (with or without digoxin), 51% with a calcium channel blocker (with or without digoxin), and 71% with digoxin alone. For patients with a history of HF symptoms or a left ventricular ejection fraction <0.40, successful rate control was observed with a beta-blocker (with or without digoxin) in 81%, with a calcium channel blocker (with or without digoxin) in 59%, and with digoxin alone in 54% at one year.
Adverse drug effects
All drug classes used for rate control were tolerated reasonably well (Table 5). Serious adverse effects were uncommon.
Number of patients in SR
There were 1,055 of 2,027 rate-controlled patients in SR at the time of randomization. Among these 1,055 patients, 427 (40%) were initially treated with a beta-blocker (with or without digoxin), 361 (34%) were initially treated with a calcium channel blocker (with or without digoxin), and 149 (14%) were initially treated with digoxin alone. On follow-up, 2,162 (58%), 1,349 (56%), and 244 (49%) were in SR at one, three, and five years, respectively.
A total of 108 patients underwent ablation of the AV junction and insertion of a permanent pacemaker to control the ventricular rate during AF. Before ablation, these patients received a mean of 2.4 ± 0.7 drug classes for rate control. Combination drug therapy was not individually reported. Furthermore, a patient might have received many different drugs and doses in a class before ablation, but would only have been reported as having received that drug class. For example, if a patient received propranolol, metoprolol, and atenolol during the course of the study before AV junctional ablation, he or she would only have been reported as receiving beta-blocker therapy. The reasons for undertaking nonpharmacologic therapy were individual but generally related to failure of pharmacologic therapy.
An additional 147 patients had a pacemaker implanted for symptomatic bradycardia.
Ventricular rate control is an important aspect of the management of AF. It can improve symptoms, exercise capacity, and cardiac function (36–40). Although it was not designed to assess these rate-control issues specifically in a randomized method, the AFFIRM study offered the opportunity to evaluate how physicians use rate-controlling drugs.
Rate control can be difficult, and drugs had to be changed in approximately one-third of patients. Drug therapy was ultimately successful by the AFFIRM definitions in two-thirds of the patients, and rate control improved over time. More patients were switched to beta-blockers than to other drug classes. Often, combination therapy was needed. Few patients underwent AV junctional ablation.
Definition of adequate rate control
The best criteria for rate control in AF are not well defined. Effectiveness criteria used in the AFFIRM study are based on a consensus of what was considered to be reasonable. The criteria may have been either too lenient or too strict, or the criteria may be addressing the wrong issue (ventricular rate, instead of symptoms). In many practice settings, the effectiveness of rate control may not be tested at rest and with frequent 6-min walk tests or Holter monitoring. It is not known whether rate control based on the AFFIRM study guidelines has any impact on symptoms or survival.
Adequate rate control at rest and with exertion
Rate control at rest does not ensure rate control with exertion, but the AFFIRM study showed that use of a functional test was feasible. It is uncertain what rate control criteria provide the best compromise between hemodynamic and symptomatic benefit and drug-induced adverse effects. No drug class was uniformly effective. Differences in patients' clinical characteristics led to a selection bias of one therapy over another. However, patients often required multiple drug iterations to achieve adequate rate control, so initial drug selections based on disease type and illness severity are unlikely to be uniformly successful.
In the AFFIRM study, digoxin was used often as a single drug. Rate control with digoxin during exercise was similar to that with beta-blocker use. This finding is perplexing, but it is possible that digoxin alone was given primarily to patients who were chronotropically incompetent, as evidenced by a slower initial ventricular response rate to AF. Another possibility is that this group was preselected. Patients whose rate was controlled with digoxin had often been started on this drug before enrollment in AFFIRM. It is possible that these results do not portray a fair representation of the effects of digoxin in an unselected population. On the other hand, these data do indicate that digoxin can be associated with adequate rate control when the drug is used clinically and selected by clinical criteria. It was not possible to know the relationship between these issues and apparent rate control with digoxin.
Similarly, it appeared that combinations of digoxin with other drug classes were associated with improved rate control. Such results may not be achievable in an unselected population, but as prescribed clinically, rate control can be achieved with these drug combinations without undue risk.
Drug titration occurred in some patients before enrollment in AFFIRM. The success of initial drug therapy was thus partly the result of drug exposure before randomization. Because no placebo control was used in this trial, it is possible that no medication would have worked as well as digoxin did to control the rate.
Calcium channel blockers
Verapamil, alone or in combination with digoxin, can be superior to digoxin alone in decreasing the resting and exercise rate (17). In two studies (6,41), verapamil combined with digoxin improved exercise capacity. Diltiazem, alone or combined with digoxin, was similar to digoxin for control of the resting HR (18), but high-dose diltiazem can have frequent adverse effects (19). Verapamil and diltiazem, when combined with digoxin, may have similar effects in patients with chronic AF (9). Their use may be problematic after MI and with HF. In the AFFIRM study, calcium channel blockers were less effective than beta-blockers at rest and with exertion. More patients in AFFIRM were switched from calcium channel blockers to beta-blockers than vice versa.
By using nadolol plus digoxin, mean and exercise HRs can be reduced, as compared with digoxin, and the exercise time can be reduced too (23). The beta1-selective drug, celiprolol, can improve HR control to exercise at the expense of decreased exercise capacity and oxygen consumption (24). Excessive dosing may result in an impaired exercise HR response (25).
In the AFFIRM study, a beta-blocker, used alone or in combination with digoxin, or a calcium channel blocker was effective, and their use increased with time. Beta-blockers may help to maintain SR better than other rate-control options, but it was not possible to evaluate this hypothesis in the AFFIRM study.
In the AFFIRM study, changing from one rate-controlling drug class to another was common. More patients were switched to beta-blockers than to other drug classes. Drug combinations were associated with improved rate control. Allowing changes in rate-controlling drugs and combination therapy may explain the improved rate control seen in the AFFIRM study over time. The success of achieving rate control in the AFFIRM study may hinge on the flexibility of the investigators to use more than one drug class for rate control. It is possible that any drug class was not fully tested and any drug titration was not complete.
Relationship among patient characteristics, initial drug dosing, and rate control
The results from this substudy do not allow conclusions to be drawn regarding the success or failure of any specific rate-controlling drug class based on patient- or disease-specific characteristics. Initial drug selection was partly based on drug use before enrollment in AFFIRM and on the patient's clinical characteristics. These selection biases could not be controlled in our analysis.
Rate control was possible for the majority of patients, and it improved over time. Initial rate control may have been influenced by previous drug testing, but long-term rate control was achievable in this elderly population with AF.
Other methods of rate control
Radiofrequency ablation of the AV junction (with implantation of a pacemaker) can be used for rate control in AF. This approach could be reserved for patients in whom the rate cannot otherwise be controlled. In the AFFIRM study, this method was used infrequently.
The treating physician was free to choose the first drug administered. Drug use was partly dependent on historical and clinical variables. Drugs used before randomization in the AFFIRM study influenced first drug use after randomization. Neither the patient nor investigator was blinded to the study drugs. Patients in SR at any time point were not included in the percent success of rate control at that time point. Rate control was analyzed by drug class, not by individual drug, and doses and start/stop dates were not reported. It is important to emphasize that there was no placebo group in this study, and there was no mandatory baseline period during which drug therapy was prohibited. Drug selection, dose selection and titration, and drug changes were discretionary. We could not exclude the contribution of natural changes in AV nodal conduction with time. Without a placebo control group or washout period, it was not possible to assess the mechanism of improved rate control over time.
Based on rate-control criteria developed for the AFFIRM study, most patients have an adequate response to the available drug classes, although frequent medication changes may be necessary and combinations may be needed. Based on the results of this study, beta-blockers tend to be used more commonly over time, and fewer patients abandoned this drug class.
- atrial fibrillation
- Atrial Fibrillation Follow-up Investigation of Rhythm Management study
- heart failure
- heart rate
- myocardial infarction
- sinus rhythm
- Received August 12, 2003.
- Revision received October 29, 2003.
- Accepted November 20, 2003.
- American College of Cardiology Foundation
- Beasley R.,
- Smith D.A.,
- McHaffie D.J.
- The Digitalis in Acute Atrial Fibrillation (DAAF) Trial Group
- Ellenbogen K.A.,
- Dias V.C.,
- Plumb V.J.,
- Heywood J.T.,
- Mirvis D.M.
- Roth A.,
- Harrison E.,
- Mitani G.,
- Cohen J.,
- Rahimtoola S.H.,
- Elkayam U.
- Farshi R.,
- Kistner D.,
- Sarma J.S.,
- Longmate J.A.,
- Singh B.N.
- Lewis R.,
- Lakhani M.,
- Moreland T.A.,
- McDevitt D.G.
- Atwood J.E.,
- Sullivan M.,
- Forbes S.,
- et al.
- Steinberg J.S.,
- Katz R.J.,
- Bren G.B.,
- Buff L.A.,
- Varghese P.J.
- The AFFIRM First Antiarrhythmic Drug Substudy Investigators
- Lipkin D.P.,
- Scriven A.J.,
- Crake T.,
- Poole-Wilson P.A.
- Levy T.,
- Walker S.,
- Mason M.,
- et al.