We performed a meta-analysis of the effect of long-term treatment with antiarrhythmic drugs for the prevention of recurrent atrial fibrillation after conversion to sinus rhythm. We found that several class IA, IC, and III drugs are effective in maintaining sinus rhythm, but virtually all of them increase adverse effects, including proarrhythmia. In addition, class IA drugs are associated with increased mortality. The final risk-benefit ratio of antiarrhythmic drugs on clinically relevant outcomes is still unclear.
The most frequent sustained arrhythmia is atrial fibrillation, which carries significant morbidity and mortality. A large number of patients recover from atrial fibrillation, either spontaneously or as a result of electrical or pharmacologic cardioversion. But after 1 year, just 20% to 30% of patients will maintain sinus rhythm if no maintenance treatment is provided.1 Many antiarrhythmic drugs have been used and tested for the prevention of atrial fibrillation and maintenance of sinus rhythm. To assess the efficacy of these drugs, as well as to ascertain their effect on proarrhythmia, adverse events, stroke, and mortality, we performed an analysis of randomized controlled trials that evaluated their long-term effects.
Patients and methods
We searched the EMBASE, MEDLINE, and Cochrane Central Register of Controlled Trials databases for randomized controlled trials comparing antiarrhythmic drugs with a control or with other antiarrhythmic drugs up through May 2005, with no language restrictions. We also examined the references cited in the studies we found, as well as the references in recent publications on atrial fibrillation. The studies included patients whose sinus rhythm had been restored, with all types of atrial fibrillation except atrial fibrillation following cardiac surgery. The treatment groups were not significantly different regarding cardiac disease and other medical treatments. The following outcomes were assessed: anticoagulant drugs taken at the end of the follow-up period, recurrence of atrial fibrillation, proarrhythmia, adverse events that resulted in treatment cessation, cardiac embolism, and all-cause mortality.
The trials accepted for our analysis were chosen by 2 authors independently, who also evaluated the quality of the methods and extracted data on an intention-to-treat basis. For each antiarrhythmic agent, the data were grouped, analyzed, and classified according to pharmacologic class. For all outcomes, the Peto odds ratio was determined. We tested for homogeneity and, if present, a random-effects model was employed. To investigate the robustness of the results, we performed sensitivity analysis by grouping the highest quality studies and those that included more than 250 patients, as well as by counting missing patients as free of events or as experiencing an event.
The trials included in the analysis were all randomized, controlled prospective, parallel-group studies. Our analysis included a total of 11,322 patients in 44 studies, which were selected from 2576 original references. In 14 of the studies, 2 or more antiarrhythmic agents were compared with each other, among a total of 2112 patients. Two antiarrhythmic agents were compared with a control in 9 studies, with a total of 3265 patients, and 1 antiarrhythmic agent was compared with a control in 21 studies, with a total of 5935 patients. The controls were digoxin (Lanoxin), beta blockers, no treatment, and placebo in 1, 1, 3, and 25 studies, respectively. The majority of studies that compared 2 antiarrhythmic agents were open label, and the majority of those that compared a control with an antiarrhythmic drug were blinded.
About 60% of subjects had persistent atrial fibrillation, which was the form of atrial fibrillation studied most often. Between 35% and 100% of patients had underlying heart disease, mostly coronary artery disease with or without hypertension, although the mean left ventricular ejection fraction was more than 50% in all except 3 studies. The follow-up period was 1 year, during which 90% to 95% of patients in 30 trials were followed up. The
shows the results regarding the various outcomes.
Class IA drugs (disopyramide [Norpace], quinidine), class IC drugs (flecainide [Tambocor], propafenone [Rythmol]), and some class III drugs (amiodarone [Cordarone, Pacerone], dofetilide [Tikosyn], and sotalol [Betapace, Sorine]) markedly decreased the recurrence of atrial fibrillation to 44% to 67% among patients treated with antiarrhythmic agents compared with a rate of 71% to 89% among control subjects. The average number needed to treat was 8 with quinidine and sotalol, 5 with dofetilide and propafenone, 4 with flecainide, and 3 with amiodarone. Combined class I drugs and sotalol did not decrease recurrences of atrial fibrillation as much as amiodarone.
In all studies except the Danish Investigations of Arrhythmia and Mortality on Dofetilide (DIAMOND) trial,2 the mortality rate was low, only 0% to 4.4%. Compared with controls, class IA drugs combined (quinidine and disopyramide) showed a significant increase in mortality (OR = 2.39; 95% confidence interval [CI], 1.03-5.59; = .04). For combined class IA drugs, the number needed to harm was 109 patients treated for 1 year for 1 excess death, with a large 95% CI of 34 to 495. After the highest quality trials were grouped, only the Suppression of Paroxysmal Atrial Tachyarrhythmias (SOPAT)3 and Prevention of Atrial Fibrillation After Cardioversion (PAFAC)4 trials, which showed no effect on death rate, remained. Quinidine was given at a dose of only 320 to 480 mg/day in these 2 studies, which was lower than in other trials, and was combined with verapamil (Calan, Covera-HS, Isoptin, Verelan).
Differences in death rate were also shown with sotalol and amiodarone. The death rate increased with sotalol, although statistical significance was not reached. Amiodarone and placebo were similar with regard to mortality, but amiodarone had a significantly lower mortality rate compared with combined class I drugs.
Except for dofetilide and aprindine (Fibocil), withdrawal because of adverse effects occurred markedly more frequently with antiarrhythmic agents compared with placebo. As a result of the SOPAT3 and PAFAC4 trials, heterogeneity among studies was shown for sotalol and quinidine. Except for amiodarone and propafenone, all antiarrhythmics significantly increased treatment-related proarrhythmic effects. Depending on the antiarrhythmic drug, pooled event rates varied from 9% to 23% for treatment withdrawals and from 1% to 7% for proarrhythmia.
Among class I agents, the largest number of treatment withdrawals occurred with quinidine. Compared with combined class I drugs, fewer proarrhythmic events and markedly fewer withdrawals occurred with amiodarone.
The use of anticoagulant medications during the follow-up period was not reported in any of the studies. Information on the occurrence of heart failure was only reported in 5 trials, and among trials that compared controls with antiarrhythmic agents, the incidence of strokes was reported in only 7 of the studies. The frequency of heart failure and strokes was low. Results of analyses of the subgroup of patients with persistent atrial fibrillation were similar to the overall results.
As far as we know, this analysis is the most comprehensive examination of the effects of antiarrhythmic drugs on important clinical outcomes, including mortality, in patients with atrial fibrillation. Previous studies and meta-analyses5,6 have focused mainly on the effectiveness of antiarrhythmic drugs in maintaining a normal electrical cardiac rhythm, assuming that maintaining normal rhythm would logically be followed by better clinical outcomes. However, sinus rhythm induced by drugs may not have the same prognostic significance as spontaneous sinus rhythm, and the potential for adverse effects of antiarrhythmic drugs could potentially outweigh their benefits.
The results of our study showed that various class IA, IC, and III drugs are beneficial in reducing the recurrence of atrial fibrillation. The high recurrence rate of 71% to 84% at 1 year in controls was reduced to 44% to 67% in treated patients. This effectiveness, however, is modest, as atrial fibrillation recurred in an important proportion of patients despite treatment.
The major outcome we assessed was the effect of antiarrhythmic drugs on mortality. Compared with controls, no variation in mortality was observed, with the exception of class IA drugs, which were associated with a mild but significant risk of increased mortality. Coplen and colleagues also found an increased risk of death with quinidine in their meta-analysis.7 In contrast, 2 recent large-scale studies3,4 included in our review that compared quinidine, sotalol, and placebo presented significant heterogeneity with other trials and did not show any mortality risk. These 2 studies included fewer patients with structural heart disease, used lower doses of quinidine, and combined quinidine with verapamil. Nevertheless, concern about an increased mortality risk with long-term use of class IA antiarrhythmic drugs persists.
Not unexpectedly, significantly more adverse events occurred with all antiarrhythmic drugs. Although amiodarone showed no significant incidence of proarrhythmia and fewer withdrawals than combined class I drugs at the 1-year follow-up, this drug clearly augmented treatment withdrawal compared with controls, and amiodarone toxicity is known to increase in frequency over time with long-term use.
The impossibility of assessing other relevant clinical end points, including stroke, heart failure, and anticoagulation use, was the chief limitation of the study. These outcomes were rarely reported in the primary trials; therefore, analyses regarding them were not feasible.
After cardioversion of atrial fibrillation, several antiarrhythmic drugs were shown to be moderately effective in preventing recurrences of atrial fibrillation over the long term. It appears that most antiarrhythmic drugs do not modify mortality, except class IA drugs, which are associated with a slight increase in the risk of death. In addition, virtually all antiarrhythmic drugs are accompanied by adverse effects.
There are few data available about the effect of antiarrhythmic drugs on important clinical outcomes. Among all the different antiarrhythmic drugs, class IA drugs should be avoided, whereas amiodarone seems to be preferable. Amiodarone is more beneficial in preventing atrial fibrillation recurrences; it also results in fewer adverse events and lower mortality than class I drugs, and it has not been shown to increase mortality compared with controls. However, long-term use of amiodarone may result in cumulative toxicity.
Overall, when the general population of patients with atrial fibrillation is considered, the final risk—benefit ratio of long-term treatment with antiarrhythmic drugs is not clear. More data regarding clinical outcomes compared with alternative strategies, including rate control,8 drugs that terminate recurrences of atrial fibrillation,9 and benefits of radiofrequency ablation,10 are needed.