OUP user menu

How safe is the antiarrhythmic drug therapy in atrial fibrillation?

Suresh Krishnamoorthy, Gregory Y.H. Lip
DOI: http://dx.doi.org/10.1093/europace/eup170 837-839 First published online: 22 June 2009

While the predominant symptom(s) related to atrial fibrillation (AF) are due to either a poorly controlled—or irregular—ventricular rate, still there is some debate with regard to the optimal strategy (that is, rate or rhythm control) to treat individual patients. Indeed, recent large, randomized, controlled trials have shown that neither rate nor rhythm control was superior in reducing mortality in AF patients, although achieving rhythm control appears to improve functional class,1 at least in the short term. Interestingly, a post hoc analysis from the AFFIRM trial found that sinus rhythm (and warfarin use) was related to improved survival but this was offset by a greater mortality from antiarrhythmic drugs (AADs).2

Anderson et al.,3 from a large retrospective (n = 40 823) Swedish nationwide study (conducted between 1995 and 2004), reported that AAD therapy was not associated with an increased mortality in AF patients. As expected, various AADs (mainly flecainide, propafenone, sotalol, and amiodarone) were used in the treatment of AF, but in this unselected, non-randomized cohort, considerably more patients were on sotalol (16%) when compared with the rest of the AADs (flecainide 2.4%, propafenone 2.6%, or amiodarone 7.3%). However, in patients with previous myocardial infarction (MI) or ischaemic heart disease, the preferred choice of AAD was amiodarone (24.3%) compared with sotalol (15%) or class-Ic drugs (8.6%). Similarly, more patients with heart failure were on amiodarone (21.5%) when compared with other AADs (sotalol 9.5%, class-Ic drugs 6.1%). The authors quoted annual mortality rates (per year/100 patient-years) of 7.4 vs. 5.3% vs. 4.3% vs. 2.5% with amiodarone, sotalol, propafenone and flecainide, respectively. Approximately a third of these unselected patients were noted to be on digoxin, whilst a fifth were on beta-blockers or calcium blockers—on top of prescribed AAD therapy for AF. In a Cox-regression analysis, treatment with AAD was generally associated with a lower mortality (flecainide HR 0.38, 95% CI 0.32–0.44; propafenone HR 0.65, 95% CI 0.58–0.71; sotalol HR 0.65, 95% CI 0.63–0.67) but the effect of amiodarone was borderline (HR 0.94, 95% CI 0.89–1.00).

Perhaps AADs may be safe in AF after all? The quoted mortality in the paper by Anderson et l.3 with individual AADs may well be related to disease progression rather than effect of antiarrhythmic therapy per se, and cause(s) of death in these patients is not that clear. But also, the choice of AADs were carefully selected in individual patients which also might account for the observed low mortality.

These observations are contrary to the post hoc analysis of AFFIRM trial data2,4 that there was a non-significant trend towards a higher mortality in patients assigned to rhythm when compared with rate control, predominantly due to non-cardiovascular causes. The significant predictors of higher non-cardiovascular mortality on multivariate analysis were rhythm control, male gender, heart failure, age, and coronary artery disease. Nevertheless, this increased mortality seen in rhythm control group may well be related to adverse effects of the various AADs used (amiodarone, sotalol, disopyramide, flecainide, moricizine, procainamide, propafenone, and quinidine) in these AF patients without heart failure, although the precise mechanism(s) was not well elucidated.

In a subsequent analysis using time-dependent variables towards in these patients, there was a 1.5-fold increased risk of death noted with AAD therapy after adjusting to co-variables in the presence of sinus rhythm (HR 1.49, 99% CI 1.11–2.01).4 The use of digoxin for rate control also appears to increase the risk of death (HR 1.42, 99% CI 1.18–1.89) in these patients with persistent AF.2,4 Notably, neither beta-blockers nor calcium-channel blockers had any effect on improving survival of these AF patients. Broadly similar results were noted in the Stroke Prevention in Atrial Fibrillation (SPAF)5 trial, where there was a 2.5-fold increase in cardiac mortality (HR 2.5, 95%CI 1.3–4.9) and a 2.6-fold increase in arrhythmic deaths (HR 2.6, 95% CI 1.2–5.6) in AF patients on AAD therapy, even after adjusting for cardiovascular risk factors. However, AF patients with history of congestive heart failure (CHF) are probably at higher risk of cardiac death (RR 3.3, 95% CI 0.99–11.1) and arrhythmic death (RR 5.8, 95% CI 1.5–21.7) on AAD therapy when compared with those without heart failure.

Indeed, in one systematic review of randomized trials6 (n = 11,322) assessing the efficacy of AADs, there was a 67% risk reduction in the recurrence of AF noted with AAD therapy; but significantly higher withdrawals were noted with AADs due to their adverse pro-arrhythmic and non-target organ side effects. Overall mortality associated with AADs was between 0 and 4.4%. However, higher mortality in these AF patients appears to be heterogeneous with individual AADs. Both the use of quinidine (OR 2.26, 95% CI 0.93–5.45) and sotalol (OR 2.09, 95% CI 0.97–4.49) showed an increase in mortality when compared with placebo, but the difference appears to be statistically non-significant. However, the combination of class-Ia agents (quinidine and disopyramide) does clearly show a significant increase in mortality (OR 2.39, 95% CI 1.03–5.59). Contrary to the study by Anderson et al.3 amiodarone significantly reduced mortality when compared with class-Ia drugs (OR 0.39, 95% CI 0.19–0.79), but the mortality benefit appears to be lost when compared with placebo. Overall, amiodarone was the least pro-arrhythmic AAD when compared with class-I drugs (OR 0.28) or sotalol (OR 0.60).

Clearly, one size (or drug) does not fit all. Some AADs can be rather harmful particularly with those with co-existent coronary ischaemia, heart failure, and structural heart disease, with additional data from studies that included such patient populations.

For example, the Cardiac Arrhythmic Suppression Trial (CAST)-I7 found a higher mortality with class-Ic drugs (flecainide/encainide) in patients with previous MI, with a higher risk of non-fatal cardiac arrest or sudden cardiac death (RR 3.6, 95% CI 1.7–8.5) and total mortality (RR 2.5, 95% CI 1.6–4.5) noted with flecainide/encainide when compared with placebo. In the prematurely discontinued CAST- II trial,8 a similar effect was seen with moricizine with a higher short-term mortality at 14 days in patients with previous MI when compared with placebo. In the Survival With ORal D-sotalol trial (SWORD),9 there was a significantly higher mortality observed in patients with remote MI and left ventricular ejection fraction (LVEF) <40% (RR 7.9, 95% CI 2.4–26.2) with the prophylactic administration of oral sotalol when compared with placebo, therefore prompted an early discontinuation of the study. There was also an inverse relationship noted between arrhythmic death related to sotalol and LVEF, and female subjects taking sotalol had a 4.7-fold increased mortality risk when compared with placebo.

In contrast, a systematic review of RCT's with other ‘non-sotalol’ beta-blockers (metoprolol, bisoprolol, carvedilol, and nebivolol) in patients with heart failure (both ischaemic and non-ischaemic)10 shows an impressive 31% risk reduction in death with beta-blocker therapy (OR 0.69, 95% CI 0.54–0.89) and this effect appears to be on top of mortality benefit conferred from renin–angiotensin–aldosterone system blockade. The DIG study (with a mean follow-up 37 months)11 showed no significant difference noted in overall mortality with digoxin vs. placebo in patients with heart failure in sinus rhythm; but the post hoc analysis12 revealed that patients who had higher serum digoxin concentrations (SDCs) had an increased mortality (unadjusted HR 1.19, 95% CI 0.98–1.45). The reported predictors of high SDC were increasing age, female sex, raised serum creatinine, concomitant diuretic use, and high dose of digoxin treatment (>250 µg/day).

With regard to the use of amiodarone, the European Myocardial Infarct Amiodarone Trial (EMIAT)13 showed a 35% risk reduction with arrhythmic deaths with amiodarone when compared with placebo (RR 0.65, 95% CI 0.42–1.00) in survivors of MI with LVEF ≤ 40% in a 2-year follow-up study although no significant differences were seen in all-cause or cardiac mortality per se between these groups. Similar results were seen in the Canadian Amiodarone Myocardial Infarction Arrhythmia Trial (CAMIAT),14 which showed up to a 48% relative risk reduction in arrhythmic death with amiodarone in patients who had recent MI when compared with placebo (RR 0.52, 95% CI 4.5–72.2). However, the absolute risk reduction in arrhythmic death was notably higher in patients who had previous MI (ARR 4.56%) and CHF (ARR 2.67%) with amiodarone.

However, both Antiarrhythmics Versus Implantable Defibrillator (AVID)15 and Sudden Cardiac Death in Heart Failure (SCD-HeFT)16 trials clearly confirmed the superiority of implanted cardiac defibrillators (ICDs) over amiodarone in reducing arrhythmic death both in patients who previously has had haemodynamically compromising ventricular arrhythmias or heart failure. In contrary to previous results seen with amiodarone, there was a trend towards higher non-cardiac deaths noted in patients assigned to amiodarone when compared with ICDs in the AVID trial (RR 1.78, 95% CI 0.98–3.26). Similar results were also seen in SCD-HeFT trial with a 44% increased risk of death noted with patients with NYHA class III CHF assigned to amiodarone when compared with placebo (HR 1.44, 95% CI 1.05–1.97). Whether this higher mortality observed in these patients with advanced heart failure was related to disease progression rather than amiodarone per se is uncertain.

Even in the recent highly debated ANtiarrhythmic trial with DRonedarone in Moderate to severe heart failure Evaluating Morbidity DecreAse (ANDROMEDA)17 study with dronedarone, there was a much higher mortality seen with patients randomized to dronedarone when compared with placebo (HR 2.13, 95% CI 1.07–4.25) and this trial therefore discontinued prematurely at 7 months follow-up. In the regression analysis, the most powerful predictor of death in these heart failure patients was the treatment with dronedarone (HR 2.19, 95% CI 1.0614.52), which also increased rates of CV hospitalizations and worsening heart failure.

In summary, the currently available AADs are neither highly efficacious nor completely safe in the treatment of AF patients. Individual AADs can be deleterious potentiating life-threatening arrhythmia(s) and also increases mortality in a certain subgroup of AF patients. This is important given that AF is commonly associated with co-existent conditions like hypertension, ischaemic heart disease, left ventricular dysfunction, and/or structural heart disease. Newer generation novel AADs such as multichannel blockers, atrial-specific agents, and gap junction modulators are on the horizon and a few have already shown some promising results in the early Phase 2 studies. Thus, there remains the optimism that the new generation novel AADs may ultimately prove to be highly efficacious, better tolerated, and with minimal side effects offering better pharmacotherapy for AF. Until then, the choice of appropriate antiarrhythmic therapy in AF patients is left with the discretion of responsible clinician(s) after careful risk and benefit assessment of individual AAD(s) towards better clinical outcomes taking associated comorbidities into consideration.


  • The opinions expressed in this article are not necessarily those of the Editors of Europace or of the European Society of Cardiology.


View Abstract