Antiarrhythmic Drugs for Maintaining Sinus Rhythm After Cardioversion of Atrial Fibrillation a Systematic Review of Randomized Controlled Trials

REVIEW ARTICLE Antiarrhythmic Drugs for Maintaining Sinus Rhythm After Cardioversion of Atrial Fibrillation A Systematic Review of Randomized Controlled Trials

Carmelo Lafuente-Lafuente, MD; Ste´phane Mouly, MD, PhD; Miguel Angel Longa´s-Tejero, MD, PhD; Isabelle Mahe´, MD, PhD; Jean-Franc¸ois Bergmann, MD

Background: A variety of antiarrhythmic drugs have phate, quinidine sulfate), class IC (flecainide acetate, been used to prevent recurrence of atrial fibrillation af- propafenone hydrochloride), and class III (amiodarone, ter conversion to sinus rhythm. We performed a system- dofetilide, sotalol hydrochloride) drugs significantly re- atic review to determine the effect of long-term treat- duced recurrence of atrial fibrillation (number needed ment with those drugs on death, embolisms, adverse to treat, 2-9), but all increased withdrawals due to ad- effects, and atrial fibrillation recurrence. verse effects (number needed to harm [NNH], 9-27) and all but amiodarone and propafenone increased proar- Methods: We searched MEDLINE, EMBASE, the Coch- rhythmia (NNH, 17-119). Class IA drugs, pooled, were rane Library (all up to May 2005), and the reference lists associated with increased mortality compared with con- of retrieved articles. We included randomized controls (Peto odds ratio, 2.39; 95% confidence interval, 1.03- trolled trials that compared any antiarrhythmic against 5.59; P=.04; NNH, 109). No other antiarrhythmic showed control (placebo or no treatment) or another antiarrhyth- a significant effect on mortality compared with con- mic, for more than 6 months. Postoperative atrial fibril- trols. We could not analyze other outcomes because data lation was excluded. Two evaluators independently re- were lacking. viewed the retrieved studies and extracted all data. Disagreements were resolved by discussion. All results Conclusion: Class IA, IC, and III drugs are effective in were calculated at 1 year of follow-up. maintaining sinus rhythm but increase adverse effects, and class IA drugs may increase mortality. Results: Forty-four trials were included, with a total of 11 322 patients. Several class IA (disopyramide phos- Arch Intern Med. 2006;166:719-728

TRIAL FIBRILLATION (AF) IS taining sinus rhythm are not well known. the most common sus- Serious adverse events are possible, as tained arrhythmia and is some of these drugs, such as quinidine sul- associated with impor- fate7,8 or flecainide acetate,9 have the po- tant morbidity and mor- tential to induce life-threatening arrhyth- tality related to stroke, other embolic mias. Overall, a rhythm-control strategy, A 1,2 complications, and heart failure. In using AAs to maintain sinus rhythm, has developed countries, AF has grown pro- not shown clear differences when com- gressively as a contributing cause of hos- pared with a rate-control strategy in out- pitalization and death in recent decades.3 comes such as mortality or stroke.10 In ad- Many patients, as many as 70% in some dition, the relative effectiveness and safety studies,4 recover sinus rhythm spontane- of the different AAs used for this indica- ously after an episode of recent-onset AF. tion are not well defined. If not, electrical and pharmacologic cardioversion are very effective in restoring CME course available at sinus rhythm. However, the problem lies www.archinternmed.com Author Affiliations: Service in the fact that the recurrence rate of AF de Me´decine Interne A, Hoˆpital is high: without treatment, only 20% to Attempts to summarize the available Lariboisière, Paris, France 30% of patients who converted remain in multitude of studies on AAs in this setting (Drs Lafuente-Lafuente, Mouly, 5,6 Mahe´, and Bergmann); and sinus rhythm at 1 year. have been incomplete. Existing meta- Servicio de Cardiologı´a, Therefore, a variety of antiarrhythmic analysis and reviews have focused on indi- 7,11,12 Hospital Prı´ncipe de Asturias, drugs (AAs) have been widely used to pre- vidual specific drugs, have pooled stud- Alcala´ de Henares, Madrid, vent recurrence of AF. However, their ef- ies using AAs for acute cardioversion Spain (Dr Longa´s-Tejero). fects on outcomes other than merely main- together with long-term treatment,13 or did

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 719

Downloaded From: https://jamanetwork.com/ on 09/26/2021 not evaluate outcomes other than si- agement of anticoagulation, heart fail- ber needed to harm, to prevent or pro- nus rhythm maintenance.14 Conse- ure, and hypertension. Finally, studies duce, respectively, one adverse out- quently, we aimed to conduct a com- had to evaluate at least 1 of the follow- come. A funnel plot was constructed, prehensive systematic review of ing outcomes: all-cause mortality, em- based on the data for mortality. randomized controlled trials study- bolic complications (stroke, peripheral Sensitivity analyses to test the ro- embolisms), adverse events leading to bustness of the results were performed ing long-term use of AAs, in patients withdrawal of treatment, proarrhyth- by (1) calculating both extremes of in- converted to sinus rhythm after hav- mia, recurrence of AF, and anticoagu- tention-to-treat analysis possibilities, ie, ing AF, with the objective of deter- lation use at the end of follow-up. We the “best case” counting all missing pa- mining the effect of the different AAs considered the following as proarrhyth- tients as being free of events, and the not only on the recurrence of AF but mia: sudden death, any new symptom- “worst case,” counting all missing pa- also on other important clinical out- atic arrhythmia (including symptom- tients as having events; and (2) selec- comes: death, stroke and other em- atic bradycardia), worsened preexisting tively pooling best-quality studies and bolisms, drug adverse effects, and pro- arrhythmias (ie, rapid AF), and newly studies with more than 250 patients. arrhythmia. appeared QRS or QT widening when Subgroup analyses were planned as they forced treatment to stop.16 Cross- follows: (1) recent-onset or persistent over studies and studies on AF after car- AF; (2) structurally normal heart or heart METHODS diac surgery were excluded. failure; and (3) studies where warfarin sodium treatment was mandatory. DATA SOURCES SELECTION PROCESS AND DATA EXTRACTION RESULTS We searched the Cochrane Central Reg- ister of Controlled Trials, MEDLINE Two of us (C.L.-L. and S.M., M.A.L.- (PubMed), and EMBASE (Ovid), all up T., or J.-F.B.) independently read the full From a total of 2576 references to May 2005, by using the following text of the studies retrieved and se- found, we assessed 151 articles in terms: (Atrial Fibrillation OR [(atrial OR lected the trials that met the inclusion more detail. Forty-four studies ful- atrium OR auricular) AND fibrillat*]) criteria, then assessed methodologic filled inclusion criteria and had us- AND (Anti-Arrhythmia Agents OR anti- quality and extracted data on an inten- able data.18-61 They represented a arrhythmi* OR anti-arrhythm* OR pro- tion-to-treat basis. Any difference be- total of 11 322 patients. Figure 1 cainamide OR disopyramide OR quini- tween reviewers was decided by discus- illustrates the selection process. dine OR mexiletine OR flecainide OR sion and consensus. Records of the study Agreement between reviewers was propafenone OR bisoprolol OR esmolol OR selection process were kept and a Qual- excellent. The funnel plot was asym- amiodarone OR dofetilide OR sotalol OR ity of Reporting of Meta-analyses state- metrical, indicating that publica- ibutilide OR azimilide OR dronedarone ment was prepared.17 Quality was rated Figure 2 OR moricizine OR cibenzoline). according to the adequacy of allocation tion bias is possible ( ). Search terms were combined with the concealment (concealing assignment un- The Table details the character- strategy to identify randomized con- til treatment had been allocated), ranked istics of included studies. All were trolled trials developed by the Coch- as A (explained and adequate) or B (un- prospective, randomized, parallel- rane Collaboration.15 In addition, we clear or not well explained). Studies group, controlled trials. Twenty- checked the reference lists of retrieved where allocation was not concealed were one trials (5935 patients) com- studies, recent guidelines, meta- not considered truly randomized and pared an AA with a control, 9 trials analyses, and general reviews on AF. Any were not included. When necessary, the (3265 patients) compared 2 AAs article that seemed to possibly meet the authors of primary studies were con- with a control, and 14 trials (2122 criteria listed in the next section was re- tacted for additional information. trieved. No limitation by language was patients) compared 2 or more AAs with each other. The control was pla- applied. During the process, publica- ␤ tions in English, German, Italian, STATISTICAL ANALYSIS cebo in 25 trials, -blockers in 1, di- French, Spanish, and Swedish were re- goxin in 1, and no treatment in 3. trieved, translated, and evaluated. Data from AAs were pooled and ana- Most trials comparing AAs vs con- lyzed individually (each specific drug) trol were single or double blind; in and grouped by pharmacologic class. contrast, most trials comparing 2 dif- CRITERIA FOR Peto odds ratios (ORs) with 95% con- SELECTING STUDIES fidence intervals (CIs) were calculated ferent AAs were open label. for all outcomes by means of a fixed- The type of AF most frequently We included only randomized con- effects model. Homogeneity between studied was persistent AF (60% of trolled trials that included patients older included studies was tested by the patients in the pooled population). than 16 years who had AF of any type Mantel-Haenszel ␹2 test. If significant Only 6 studies included exclu- and duration and in whom sinus rhythm heterogeneity between studies was ob- sively paroxysmal or recent-onset had been restored (spontaneously or by served, differences in clinical character- AF. The proportion of patients hav- any therapeutic means), and compared istics of the studies were searched and ing underlying heart disease, as de- long-term treatment (at least 6 months) either a random-effects model was used fined by each study, varied widely, with any available AA against a control or studies were not combined if clini- from 33% to 100%; only 1 study se- (placebo, no treatment, or drugs for rate cally dissimilar. We used the Cochrane lectively included patients without control) or against other AAs. In each Collaboration software RevMan (ver- 30 study, all treatment groups had to be sion 4.2.2; available at: http://www any structural heart abnormality. similar with regard to (1) cardiac dis- .cc-ims.net/RevMan) for all statistics. Sig- Nonetheless, the mean left ventricu- ease (frequency, type, and severity); (2) nificant results were also expressed as lar ejection fraction was greater than type and duration of AF; and (3) man- the number needed to treat or the num- 50% in all except 3 trials.28,35,45 The

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 720

Downloaded From: https://jamanetwork.com/ on 09/26/2021 most frequent abnormalities were,

unsurprisingly, coronary disease (5% 2576 Publications Identified and Screened to 50% of patients), hypertension, and valvular heart disease, the last 2425 Excluded by Title and/or Abstract more frequent in older studies. The percentage of patients lost 151 Retrieved for More Detailed Evaluation during follow-up was reported in 30 83 Excluded trials and was small: 5% to 10%. Im- 42 Inappropriate Design puting those missing patients as 24 Noncontrolled Trials events—the worst-case intention-to- 6 Controlled but Nonrandomized 9 Crossover treat scenario—seldom modified re- 3 Only Acute Treatment of Atrial Fibrillation sults, so the best-case intention-to- 37 Double Publication (Data Extracted When Needed) treat analysis—missing patients 4 Ongoing Studies counted as being free of events—is 68 Potentially Appropriate Randomized given by default, and when differ- Controlled Trials to Be Included ences existed, they are stated. All results are calculated at 1 year 23 Excluded 17 Not Fulfilling Inclusion Criteria of follow-up. 9 Follow-up <6 mo 7 Persistent Atrial Fibrillation Not Reverted MORTALITY in Controls 1 Randomized Trial, but Not for Allocation to Antiarrhythmics Figure 3 shows results for overall 6 Inadequate Comparison (Combination of Drugs, mortality. Mortality rate was low: 0% Comparing Drugs Not Tested vs Placebo)

to 4.4%. No death at all was re- 45 Included in Meta-analysis ported in trials of flecainide. The only exception to the general low 1 Withdrawn: Separate Data for Each mortality was the Danish Investiga- Tested Drug Not Available tions of Arrhythmia and Mortality on 28 44 Studies With Usable Information Dofetilide study, which specifi- 21 Comparing Antiarrhythmic vs Control cally recruited patients with ad- 17 vs Placebo vanced heart failure and had a mor- 3 vs No Treatment 1 vs β-Blocker tality of 31% at 1 year. 14 Comparing ≥2 Antiarrhythmics Compared with controls, a non- 9 Comparing ≥2 Antiarrhythmics and Control significant trend to increased mortality appeared with quinidine (OR, Figure 1. Selection of studies for inclusion. 2.26; 95% CI, 0.93-5.45; P=.07). This increase in mortality was significant (OR, 2.39; 95% CI, 1.03- 0 5.59; P=.04) when drugs of class IA—quinidine and disopyramide 1 phosphate—were combined. The corresponding number needed to 2 harm for combined class IA drugs was 109 patients treated for 1 year to have 1 excess death; the 95% CI 3 was very large (34-4895 patients). SE (log Peto OR) In sensitivity analyses, counting 4 missing patients as deaths con- firmed these results, showing a sig- 5 nificant increase for quinidine alone. 0.01 0.1 1 10 100 However, selectively pooling trials Peto OR with adequate allocation concealment or those including more than Figure 2. Funnel plot based on results for mortality. OR indicates odds ratio; SE, standard error. 250 patients left only 2 trials— Prevention of Atrial Fibrillation Af- to 1800 mg/d), and combined it with was less pronounced if only high- ter Cardioversion (PAFAC)44 and verapamil hydrochloride. quality trials or trials having more Suppression of Paroxysmal Atrial A nonsignificant trend to in- than 250 patients were analyzed. Tachyarrhythmias (SOPAT)56—in creased mortality appeared also with Amiodarone, when compared with which no effect on mortality was ap- sotalol hydrochloride (OR, 2.09; combined class I drugs, showed a parent. These 2 trials used a lower 95% CI, 0.97-4.49; P=.06). This significant reduction in mortality dosage of quinidine sulfate (320- trend became significant if missing (OR, 0.39; 95% CI, 0.19-0.79; 480 mg/d) than other studies (800 patients were counted as deaths but P=.009; number needed to treat,

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 721

Downloaded From: https://jamanetwork.com/ on 09/26/2021 Table. Characteristics of Included Studies

Structural Mean Heart Left Persistent Electrical Disease, Atrial Mean AF, Cardioversion, Allocation Age, %of Diameter, LVEF, %of %of Treatment Follow-up, Study Concealment* Blinding N y† Patients mm % Patients Patients Groups mo Studies Comparing Antiarrhythmic vs Control (Placebo or No Treatment) AFIB,19 1997 B No 1227 63 67 NA NA NA 30 Bidisomide, 400-1200 mg/d; 6 placebo ASAP,21 2003 B Double-blind 1380 63 73 NA NA NA 35 Azimilide dihydrochloride, 6 35-125 mg/d; placebo Benditt et al,23 A Double-blind 253 62 (24-86) 57 NA NA 23 NA Sotalol hydrochloride, 80-160 12 1999 mg/d; placebo Byrne-Quinn B Double-blind 74 54 (30-70) 80 NA NA 100 100 Quinidine sulfate, 1200 mg/d; 12 and Wing,24 placebo 1970 Channer et al,26 A Double-blind 99 67 NA 44 58 100 80 Amiodarone, 200 mg/d; placebo 12 2004 DAFNE,27 2003 B Double-blind 199 63 NA 45 55 100 85 Dronedarone, 800-1200 mg/d; 6 placebo DIAMOND,28 A Double-blind 506 72 (36-92) 100 NA Ͻ35 100 15 Dofetilide, 0.5 mg/d; placebo 12 2001 Dogan et al,29 B Single-blind 110 61 79 44 64 29 27 Propafenone hydrochloride, 450 15 2004 mg/d; placebo GEFACA,31 2001 B Double-blind 50 62 94 48 55 100 68 Amiodarone, 200 mg/d; placebo 16 Hillestad et al,32 B No 100 54 (22-77) 92 NA NA 100 100 Quinidine, 800-1200 mg/d; no 12 1971 treatment Karlson et al,36 B Double-blind 92 60 (31-72) 60 NA NA 100 100 Disopyramide phosphate, 500 12 1988 mg/d; placebo Kuhlkamp et A Double-blind 394 60 (24-86) 36 42 64 100 82 Metoprolol tartrate, 100 mg/d; 6 al,40 2000 placebo Okishige et al,43 B Single-blind 62 51 61 41 61 100 79 Pilsicainide, 150 mg/d; placebo 12 2000 Plewan et al,45 B No 128 59 72 48 41 100 100 Sotalol hydrochloride, 160 mg/d; 8 2001 bisoprolol fumarate, 5 mg/d RAFT,47 2003 B Double-blind 523 63 (22-89) 48 NA NA NA 21 Propafenone hydrochloride, 9 450-850 mg/d; placebo SAFIRE-D,51 B Double-blind 250 67 (30-88) 67 NA NA 100 85 Dofetilide, 0.25-1 mg/d; placebo 12 2000 Singh et al,52 B Double-blind 34 60 NA 44 NA 100 83 Sotalol hydrochloride, 80-320 6 1991 mg/d; placebo SMART,53 2002 A Double-blind 94 60 NA NA NA 41 50 Aprindine hydrochloride, 40 6 mg/d; placebo Sodemark B No 185 58 (24-78) 94 NA NA 100 51 Quinidine sulfate, 1200-1800 12 et al,55 1975 mg/d; no treatment Stroobandt A Double-blind 102 62 (27-84) 71 39 NA 54 66 Propafenone hydrochloride, 450 6 et al,58 1997 mg/d; placebo Van Gelder A No 73 60 82 44 NA 100 100 Flecainide acetate, 200-300 12 et al,59 1989 mg/d; no treatment Studies Comparing 2 or More Antiarrhythmics and Control Bellandi et al,22 B Double-blind 194 52 (20-75) 72 42 55 53 11 Propafenone hydrochloride, 900 12 2001 mg/d; sotalol hydrochloride, 240 mg/d; placebo Carunchio B No 66 48 (30-69) 65 36 NA NA 33 Flecainide acetate, 200 mg/d; 12 et al,25 1995 sotalol hydrochloride, 240 mg/d; placebo Kochiadakis B Single-blind 146 63 38 43 53 37 NA Amiodarone, 200 mg/d; 24 et al,37 2004 propafenone hydrochloride, 450 mg/d; placebo Kochiadakis B Single-blind 186 63 36 43 54 35 NA Amiodarone, 200 mg/d; sotalol 24 et al,38 2000 hydrochloride, 320 mg/d; placebo Lloyd et al,41 B Double-blind 82 46 (15-79) 94 NA NA 100 100 Disopyramide phosphate, 450 6 1984 mg/d; quinidine sulfate, 1400 mg/d; placebo PAFAC,44 2004 A Double-blind 848 63 NA 45 60 100 NA Quinidine sulfate, 480 mg/d; 12 sotalol hydrochloride, 320 mg/d; placebo SAFE-T,50 2005 B Double-blind 665 67 33 48 51 100 80 Amiodarone, 300 mg/d; sotalol 12 hydrochloride, 320 mg/d; placebo SOPAT,56 2004 A Double-blind 1033 60 NA 39 61 0 NA Quinidine sulfate, 320-480 mg/d; 12 sotalol hydrochloride, 320 mg/d; placebo Steinbeck B No 45 59 73 NA NA 0 0 Flecainide acetate, 200-300 12 et al,57 1988 mg/d; quinidine sulfate, 1000 mg/d; digoxin

(continued)

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 722

Downloaded From: https://jamanetwork.com/ on 09/26/2021 Table. Characteristics of Included Studies (cont)

Structural Mean Heart Left Persistent Electrical Disease, Atrial Mean AF, Cardioversion, Allocation Age, %of Diameter, LVEF, %of %of Treatment Follow-up, Study Concealment* Blinding N y† Patients mm % Patients Patients Groups mo Studies Comparing 2 or More Antiarrhythmics AFFIRM A No 410 69 85 NA 55 71 NA Amiodarone, 200 mg/d; class I 44 substudy,18 drugs (various); sotalol 2003 hydrochloride, 240 mg/d Aliot et al,20 B No 97 63 45 NA NA 0 0 Flecainide acetate, 100-200 12 1996 mg/d; propafenone hydrochloride, 600 mg/d FAPIS,30 1996 A No 200 57 0 35 61 0 0 Flecainide acetate, 200 mg/d; 12 propafenone hydrochloride, 520 mg/d Hohnloser B No 50 62 86 50 51 100 60 Quinidine sulfate, 1000 mg/d; 6 et al,33 1995 sotalol hydrochloride, 240-320 mg/d Juul-Moller B No 183 59 NA 42 NA 100 100 Quinidine sulfate, 1200 mg/d; 6 et al,34 1990 sotalol hydrochloride, 160-320 mg/d Kalusche B No 82 61 68 45 30 68 53 Quinidine sulfate, 1000 mg/d; 12 et al,35 1994 sotalol hydrochloride, 240-400 mg/d Kochiadakis B Single-blind 254 63 41 44 53 41 NA Propafenone hydrochloride, 450 24 et al,39 2004 mg/d; sotalol hydrochloride, 320 mg/d Naccarelli B No 239 58 83 NA NA 26 0 Flecainide acetate, 200-300 12 et al,42 1996 mg/d; quinidine sulfate, 1000-1500 mg/d PRODIS,46 B Double-blind 56 60 65 46 NA 100 100 Disopyramide phosphate, 750 8 1996 mg/d; propafenone hydrochloride, 900 mg/d Reimold et al,48 A No 100 61 81 46 59 53 NA Propafenone hydrochloride, 675 12 1993 mg/d; sotalol hydrochloride, 320 mg/d Richiardi et al,49 B No 200 57 (29-75) 48 45 NA 0 12 Propafenone hydrochloride, 12 1992 900 mg/d; quinidine sulfate, 1000 mg/d SOCESP,54 B No 121 54 54 39 68 39 NA Quinidine sulfate, 700 mg/d; 6 1999 sotalol hydrochloride, 240 mg/d Villani et al,60 B No 76 65 (37-85) 86 38 NA 0 26 Amiodarone, 200 mg/d; 14 1992 disopyramide phosphate, 500 mg/d Vitolo et al,61 B No 54 53 100 NA NA 100 100 Amiodarone, 400 mg/d; 6 1981 quinidine sulfate, 1200 mg/d

Abbreviations: AF, atrial fibrillation; AFFIRM, Atrial Fibrillation Follow-up Investigation of Rhythm Management; AFIB, Atrial Fibrillation Investigation with Bidisomide; ASAP, Azimilide Supraventricular Arrhythmia Program; DAFNE, Dronedarone Atrial Fibrillation Study After Electrical Cardioversion; DIAMOND, Danish Investigations of Arrhythmia and Mortality on Dofetilide; FAPIS, Flecainide and Propafenone Italian Study; GEFACA, Grupo de Estudio de Fibrilacion Auricular Con Amiodarona; LVEF, left ventricular ejection fraction; NA, not available; PAFAC, Prevention of Atrial Fibrillation After Cardioversion; PRODIS, Propafenone vs Disopyramide Study; SAFE-T, Sotalol Amiodarone Atrial Fibrillation Efficacy Trial; SAFIRE-D, Symptomatic Atrial Fibrillation Investigative Research on Dofetilide; SOCESP, Sociedade de Cardiologia do Estado de Sa˜o Paulo Investigators; SOPAT, Suppression of Paroxysmal Atrial Tachyarrhythmias. *Adequacy of the process of concealing assignment during randomization, where A represents adequate, and B, unclear. †Age is presented as mean or mean (range). 17; 95% CI, 13-52). This effect was eFigure 2; available at: http://www Almost all AAs showed signifi- mainly due to the weight of 1 .archinternmed.com). Withdraw- cantly increased proarrhythmic study18 and persisted in sensitivity als due to adverse effects were sig- effects, the only exceptions being analysis. Amiodarone showed no nificantly more frequent with all AAs amiodarone and propafenone difference when compared with compared with controls, with few hydrochloride. Pooled event rates placebo. exceptions: aprindine hydrochlo- varied depending on the drug No other significant difference in ride and dofetilide, both having re- used, from 9% to 23% for with- mortality appeared in the remain- sults from only 1 study. Heteroge- drawals due to adverse effects ing comparisons. neity between studies was detected and from 1% to 7% for proar- for quinidine (PϽ.001) and sotalol rhythmia. Number needed to WITHDRAWALS BECAUSE (P=.03). In both cases that finding harm ranged, for withdrawals, OF ADVERSE EFFECTS was due to 2 trials, again PAFAC44 from 9 with quinidine to 27 with AND PROARRHYTHMIA and SOPAT,56 in which neither amiodarone, propafenone, or quinidine nor sotalol showed more sotalol; and for proarrhythmia, Figure 4 shows results for these withdrawals than placebo, con- between 17 with flecainide and outcomes (see also eFigure 1 and trary to other studies. 119 with dofetilide.

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 723

Downloaded From: https://jamanetwork.com/ on 09/26/2021 No. of Events/Total Peto OR (95% CI) No. of Drugs Studied Studies Antiarrhythmic Control P Value

Antiarrhythmic vs Control 0.10 1 10 Class IA Disopyramide phosphate 2 2/75 0/717.56 (0.47-1.22) .16 Quinidine sulfate 7 21/1128 4/5482.26 (0.93-5.45) .07 All class IA 8 23/1203 4/594 2.39 (1.03-5.59) .04 Class IB All: aprindine hydrochloride, bidisomide 2 9/781 3/5401.89 (0.59-6.03) .28 Class IC Flecainide acetate 3 0/71 0/78Not Estimable NA Propafenone hydrochloride 5 0/720 2/3780.05 (0.00-1.02) .05 All class IC 9 1/843 2/4660.14 (0.00-1.88) .14 Class II All: metoprolol tartrate 1 3/197 0/1977.47 (0.77-72.20) .08 Class III Amiodarone 4 13/428 3/2451.96 (0.68-5.67) .21 Dofetilide 2 83/431 83/3250.97 (0.67-1.40) .88 Sotalol hydrochloride 9 30/1391 5/8152.09 (0.97-4.49) .06 Azimilide dihydrochloride + dronedarone 2 10/1042 4/5371.31 (0.43-3.97) .63 All class III 16 136/3292 95/19221.19 (0.88-1.61) .27

Comparing 2 Antiarrhythmics Drug A Drug B

Disopyramide vs other class I drugs 2 1/60 2/530.46 (0.05-4.52) .51 Quinidine vs Flecainide 2 0/132 0/137Not Estimable NA Other class I drugs 4 2/258 2/2681.04 (0.14-7.46) .97 Sotalol 6 13/1109 17/8690.71 (0.34-1.46) .35 Flecainide vs propafenone 2 0/145 1/1520.14 (0.00-6.96) .32 Amiodarone vs Class I drugs 4 10/241 26/257 0.39 (0.19-0.79) .009 Sotalol 3 28/463 39/4470.66 (0.40-1.10) .11 Sotalol vs class I except quinidine 4 15/243 17/2510.94 (0.44-1.99) .87

Figure 3. Overall mortality. Some studies compared more than 2 drugs, so the total numbers of studies and patients in the figure are greater than the absolute numbers of studies and patients included. CI indicates confidence interval; NA, not applicable; and OR, odds ratio. Boldface values are significant.

When the different AAs were ing average numbers needed to treat a control reported strokes occur- compared, quinidine caused more were 3 with amiodarone, 4 with fle- ring during the trial,23,32,36,41,50,55,56 withdrawals than the other class I cainide, 5 with dofetilide and but it is not certain that reporting was drugs (OR, 2.25; 95% CI, 1.45- propafenone, and 8 with quinidine exhaustive. They reported a total of 3.51; PϽ.001) but not more proar- and sotalol; the 95% CIs varied be- 20 strokes in 1755 patients treated rhythmia. Amiodarone was associ- tween 2 and 14. with AAs and 6 strokes in 650 ated with significantly fewer In comparisons between AAs, patients in the corresponding con- withdrawals (OR, 0.52; 95% CI, amiodarone reduced recurrences of trol groups. Five trials reported data 0.34-0.81; P=.004) and fewer pro- AF significantly more than com- on the incidence of heart fail- arrhythmic events (OR, 0.28; 95% bined class I drugs (OR, 0.31; 95% ure,30,33,40,46,48 which was low and CI, 0.13-0.59; PϽ.001) than com- CI, 0.21-0.45; PϽ.001) and more without obvious differences be- bined class I drugs; results were not than sotalol (OR, 0.43; 95% CI, 0.33- tween groups. modified in the sensitivity analysis. 0.56; PϽ.001). No other differ- Subgroup analysis of patients ences between AAs were detected. with persistent AF replicated these ATRIAL FIBRILLATION results. The rest of the planned sub- RECURRENCE OTHER OUTCOMES group analyses were not possible.

Results for atrial fibrillation recur- Long-term anticoagulation with war- COMMENT rence are presented in Figure 5. All farin was mandatory for every pa- studied class IA and IC drugs and all tient during all the follow-up in only class III drugs except dronedarone 3 studies.26,32,59 In the rest, the deci- Our study confirms that several proved to significantly reduce re- sion was left to the judgment of the AAs belonging to different phar- currences of AF. Pooled recurrence attending physician. No trial re- macologic classes (all class IA and rates at 1 year were high: 71% to 84% ported the frequency of anticoagu- IC drugs, and all class III drugs in controls and reduced to 44% to lation during the follow-up. Only 7 except dronedarone) proved to be 67% in treated patients. Correspond- of the studies comparing AAs with effective in maintaining sinus

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 724

Downloaded From: https://jamanetwork.com/ on 09/26/2021 Withdrawals Proarrhythmia No. of Drugs Studied Studies† Peto OR (95% CI) Peto OR (95% CI) Antiarrhythmic vs Control 0.10 1 10 0.10 1 10 Class IA Disopyramide hydrochloride 2 3.85 (1.13-13.20) No Events Quinidine sulfate: higher dose 5 3.58 (2.01-6.40) 4.56 (1.20-17.30) Quinidine: lower dose∗ 2 0.81 (0.59-1.10) 1.53 (0.64-3.60) Quinidine: all studies 7 1.90 (0.90-4.02)† 2.10 (1.02-4.33) All class IA 8 2.02 (1.00-4.10)† 2.06 (1.00-4.26) Class IB All: aprindine hydrochloride 1 0.66 (0.11-3.95) No Events Class IC Flecainide acetate 3 9.14 (1.94-42.90) 5.97 (1.67-21.30) Propafenone hydrochloride 5 1.69 (1.09-2.62) 1.52 (0.33-7.02) All class IC 9 1.93 (1.27-2.93) 3.41 (1.28-9.09) Class II All: metoprolol tartrate 1 3.16 (1.43-6.99) 7.96 (2.84-22.30) Class III Amiodarone 3, 4 5.55 (2.24-13.70) 2.65 (0.88-8.00) Dofetilide 1, 2 1.61 (0.41-6.23) 3.77 (1.31-10.80) Sotalol hydrochloride: PAFAC, SOPAT‡ 2 0.95 (0.68-1.33) 1.42 (0.56-3.60) Sotalol: rest of studies 6, 7 3.02 (1.65-5.53) 2.67 (1.44-4.98) Sotalol: all studies 8, 9 1.47 (0.84-2.60)† 2.20 (1.31-3.69) Azimilide dihydrochloride + dronedarone 2 2.46 (1.51-4.01) 3.30 (1.01-10.80) All class III 14, 16 1.63 (1.29-2.07) 2.20 (1.31-3.69)

Comparing 2 Antiarrhythmics

Disopyramide vs other class I drugs 2 0.37 (0.14-1.03) 0.43 (0.00-4.25) Quinidine vs Flecainide 2 2.04 (1.14-3.64) 1.60 (0.64-3.96) Other class I drugs 4 2.25 (1.45-3.51) 1.59 (0.71-3.56) Sotalol 6 1.12 (0.89-1.40) 1.00 (0.60-1.68) Flecainide vs propafenone 2 0.68 (0.32-1.43) 0.43 (0.10-1.93) Amiodarone vs Class I drugs 4, 3 0.52 (0.34-0.81) 0.28 (0.13-0.59) Sotalol 2, 3 1.18 (0.67-2.07) 0.60 (0.30-1.20) Sotalol vs class I except quinidine 4 0.45 (0.28-0.72) 0.77 (0.44-1.34)

Figure 4. Withdrawals due to adverse effects and proarrhythmia. Some studies compared more than 2 drugs, so the total numbers of studies and patients in the table are greater than the absolute numbers of studies and patients included. When the numbers of studies pooled were different for the 2 outcomes, the number combined to evaluate withdrawals is given first, followed by the number combined to evaluate proarrhythmia. CI indicates confidence interval; OR, odds ratio; PAFAC, Prevention of Atrial Fibrillation After Cardioversion; SOPAT, Suppression of Paroxysmal Atrial Tachyarrhythmias; asterisk, PAFAC and SOPAT trials in both cases, which showed heterogeneity compared with other studies on quinidine or on sotalol; and dagger, the OR was calculated by the random-effects model, because the test for heterogeneity between pooled studies was significant.

rhythm after conversion of AF. disopyramide together) demon- the PAFAC and SOPAT studies Depending on the drug, recur- strated a significant effect on than in older trials. rence of AF was reduced by 30% mortality, increasing it. These When AAs were compared with to 50% with respect to controls. results were not reproduced when each other, amiodarone produced Amiodarone seemed to be the only the PAFAC and SOPAT stud- less mortality than class I drugs almost effective. However, the suc- ies were analyzed. These 2 trials together. However, it is important to cess of AAs was limited: AF still are recent, high-quality, large emphasize that amiodarone proved recurred in 42% to 67% of treated (848 and 1033 patients, respec- no advantage in mortality com- patients. Most trials focused on tively) studies that compared pared with placebo. To date, 2 pre- effects on electrical rhythm as quinidine, sotalol, and placebo vious meta-analyses have assessed the main outcome, as restoring and showed no increase in mortal- mortality in this setting: Coplen et and keeping normal auricular ity. A probable explanation is that al7 focused on quinidine and also function is believed to be the most both used a lower dose of quini- found an increased mortality with advantageous to the patient. We dine than other studies and that this drug, while Nichol et al13 found wanted to know whether, in addi- quinidine was combined with no difference with any AA regard- tion to maintaining sinus rhythm, verapamil, which has been shown ing mortality, but most of the trials long-term treatment with these to reduce some of the proarrhyth- they pooled had very short fol- drugs carried other measurable mic effects of quinidine, such low-up periods. clinical benefits. as accelerated atrioventricular We also found evidence of in- Concerning mortality, which conduction. In addition, the pro- creased adverse effects. Virtually all was low, our results show that portion of patients having struc- of the AAs caused more withdraw- only class IA drugs (quinidine and tural heart disease was lower in als due to adverse effects than con-

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 725

Downloaded From: https://jamanetwork.com/ on 09/26/2021 No. of Events/Total Peto OR (95% CI) No. of Drugs Studied Studies Antiarrhythmic Control P Value

Antiarrhythmic vs Control 0.10 1 10 Class IA Disopyramide hydrochloride 2 40/75 49/710.52 (0.27-1.01) .05 Quinidine sulfate 7 741/1106 417/518 0.51 (0.40-0.65) <.001 All class IA 8 781/1118 449/564 0.51 (0.40-0.64) <.001 Class IB All: aprindine hydrochloride, bidisomide 2 639/781 453/5400.84 (0.63-1.13) .26 Class IC Flecainide acetate 3 31/71 56/78 0.31 (0.16-0.60) <.001 Propafenone hydrochloride 5 376/720 276/378 0.37 (0.28-0.48) <.001 All class IC 9 443/843 342/466 0.36 (0.28-0.45) <.001 Class II All: metoprolol tartrate 1 127/197 140/1970.74 (0.49-1.13) .16 Class III Amiodarone 4 200/428 209/245 0.19 (0.14-0.27) <.001 Dofetilide 2 252/431 274/325 0.28 (0.20-0.38) <.001 Sotalol hydrochloride 9 916/1391 622/815 0.53 (0.44-0.65) <.001 Dronedarone 1 116/151 43/480.45 (0.20-1.02) .06 All class III 15 1484/2401 1148/1433 0.37 (0.32-0.43) <.001

Comparing 2 Antiarrhythmics Drug A Drug B

Disopyramide vs other class I drugs 2 26/60 27/530.76 (0.36-1.60) .47 Quinidine vs Flecainide 2 103/132 99/1371.38 (0.79-2.41) .26 Other class I drugs 4 176/258 168/2681.30 (0.90-1.87) .17 Sotalol 6 715/1109 556/8690.92 (0.76-1.11) .38 Flecainide vs propafenone 2 49/145 56/1520.87 (0.54-1.40) .56 Amiodarone vs Class I drugs 4 100/241 175/257 0.31 (0.21-0.45) <.001 Sotalol 3 218/463 303/447 0.43 (0.33-0.56) <.001 Sotalol vs class I except quinidine 4 150/243 157/2510.98 (0.67-1.45) .93

Figure 5. Atrial fibrillation recurrence. Some studies compared more than 2 drugs, so the total numbers of studies and patients in the figure are greater than the absolute numbers of studies and patients included. CI indicates confidence interval; OR, odds ratio.

trols did. For proarrhythmia, only To the best of our knowledge, this years, and most of them had a nor- amiodarone and propafenone showed is the most exhaustive systematic re- mal left ventricular ejection frac- no significant difference compared view performed to date in this set- tion. We do not know whether our with controls. Amiodarone was as- ting, assessing not only the effect of results can be extrapolated to other sociated with significantly fewer ad- AAs on maintaining sinus rhythm patient populations, such as older verse effects and proarrhythmic but also other relevant outcomes. We people or patients with impaired events than class I drugs combined. have synthesized the accumulated left ventricular function. However, these are results at 1 year experience of 44 good-quality In conclusion, after cardiover- of follow-up, and adverse effects of randomized controlled trials sion of AF, various AAs appear to be amiodarone are well known to in- and focused on diverse AAs, repre- moderately effective in maintaining crease in frequency over time. senting approximately 30 years of sinus rhythm in the long term, but Unfortunately, we could not evalu- research, some recently pub- all show evidence of adverse effects, ate other important clinical out- lished,37,44,50,56 and involving more and data on various important clini- comes, such as stroke and heart fail- than 11 000 patients in total. A va- cal outcomes, such as stroke, embo- ure, because they were rarely riety of sensitivity analyses pro- lisms, and heart failure, are sparse reported. Moreover, the frequency of duced the same results. among available trials. Therefore, the use of long-term anticoagulation was The main limitation of this final risk-benefit ratio of long-term not available, complicating this evalu- study is the lack of data on some treatment with those drugs remains ation. While the frequency of stroke outcomes. Other limitations are as unclear. With respect to the relative and heart failure, in the few studies follows: (1) In many studies effectiveness and risks of each drug, that reported it, was very low and did patients were followed up until AF class IA drugs should be used most not show apparent differences be- recurred and not thereafter; hence, carefully for this indication because tween treatment groups, the actual events between that point and the of the risk of increasing mortality. On effect of long-term AAs on these complete 1 year of follow-up might the other hand, on the basis of re- important end points remains have been missed. (2) Mean age of sults at 1 year, amiodarone could be unknown. included patients was about 60 preferred because it seems to be the

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 726

Downloaded From: https://jamanetwork.com/ on 09/26/2021 most effective in preventing recur- on the risk of death: the Framingham Heart Study. 18. AFFIRM First Antiarrhythmic Drug Substudy In- rences of AF, produces fewer ad- Circulation. 1998;98:946-952. vestigators. Maintenance of sinus rhythm in pa- 2. Stewart S, Hart CL, Hole DJ, McMurray JJ. A popu- tients with atrial fibrillation: an AFFIRM substudy verse events, and is associated with lation-based study of the long-term risks associ- of the first antiarrhythmic drug. J Am Coll Cardiol. less mortality than class I drugs while ated with atrial fibrillation: 20-year follow-up of 2003;42:20-29. not increasing mortality compared the Renfrew/Paisley study. Am J Med. 2002; 19. Atrial Fibrillation Investigation With Bidisomide with controls. However, we do not 113:359-364. (AFIB) Investigators. Treatment of atrial fibrilla- know whether those advantages are 3. WattigneyWA,MensahGA,CroftJB.Increasingtrends tion and paroxysmal supraventricular tachycar- in hospitalization for atrial fibrillation in the United dia with bidisomide. Circulation. 1997;96: maintained with longer treatment. A States, 1985 through 1999: implications for primary 2625-2632. more precise estimation of the ef- prevention. Circulation. 2003;108:711-716. 20. Aliot E, Denjoy I; Flecainide AF French Study Group. fects of long-term treatment with AAs 4. Geleris P, Stavrati A, Afthonidis D, Kirpizidis H, Comparison of the safety and efficacy of on appropriate clinical outcomes, as Boudoulas H. Spontaneous conversion to sinus flecainide versus propafenone in hospital out- rhythm of recent (within 24 hours) atrial fibrillation. patients with symptomatic paroxysmal atrial compared with placebo or alterna- J Cardiol. 2001;37:103-107. fibrillation/flutter. Am J Cardiol. 1996;77: tive strategies to manage AF—ie, rate 5. Dahlin J, Svendsen P, Gadsboll N. Poor mainte- 66A-71A. control,10 drugs only for terminat- nance of sinus rhythm after electrical cardiover- 21. Page RL, Tilsch TW, Connolly SJ, et al; Azimilide ing AF recurrences,62 and radiofre- sion of patients with atrial fibrillation or flutter: a Supraventricular Arrhythmia Program (ASAP) In- quency ablation63—would be needed. 5-year follow-up of 268 consecutive patients. vestigators. Asymptomatic or “silent” atrial fibril- Scand Cardiovasc J. 2003;37:324-328. lation: frequency in untreated patients and pa- 6. Golzari H, Cebul RD, Bahler RC. Atrial fibrillation: tients receiving azimilide. Circulation. 2003; Accepted for Publication: Septem- restoration and maintenance of sinus rhythm and 107:1141-1145. ber 12, 2005. indications for anticoagulation therapy. Ann Intern 22. Bellandi F, Simonetti I, Leoncini M, et al. Long- Correspondence: Carmelo Lafuente- Med. 1996;125:311-323. term efficacy and safety of propafenone and so- Lafuente, MD, Service de Me´decine 7. Coplen SE, Antman EM, Berlin JA, Hewitt P, talol for the maintenance of sinus rhythm after con- Chalmers TC. Efficacy and safety of quinidine version of recurrent symptomatic atrial fibrillation. Interne A, Hoˆpital Lariboisière, 2 rue therapy for maintenance of sinus rhythm after car- Am J Cardiol. 2001;88:640-645. Ambroise Pare´, 75010 Paris, France dioversion: a meta-analysis of randomized con- 23. Benditt DG, Williams JH, Jin J, et al; d,l-Sotalol ([email protected]). trol trials. Circulation. 1990;82:1106-1116. Atrial Fibrillation/Flutter Study Group. Mainte- Author Contributions: Dr Lafuente- 8. Flaker GC, Blackshear JL, McBride R, Kronmal RA, nance of sinus rhythm with oral d,l-sotalol therapy Lafuente had full access to all of the Halperin JL, Hart RG; Stroke Prevention in Atrial in patients with symptomatic atrial fibrillation Fibrillation Investigators. Antiarrhythmic drug and/or atrial flutter. Am J Cardiol. 1999;84: data in the study and takes respon- therapy and cardiac mortality in atrial fibrillation. 270-277. sibility for the integrity of the data and J Am Coll Cardiol. 1992;20:527-532. 24. Byrne-Quinn E, Wing AJ. Maintenance of sinus the accuracy of the data analysis. 9. Echt DS, Liebson PR, Mitchell L, et al. Mortality rhythm after DC reversion of atrial fibrilllation: a Financial Disclosure: None. and morbidity in patients receiving encainide, double-blind controlled trial of long-acting quini- flecainide, or placebo: the Cardiac Arrhythmia dine bisulphate. Br Heart J. 1970;32:370-376. Funding/Support: This work was Suppression Trial. N Engl J Med. 1991;324: 25. Carunchio A, Fera MS, Mazza A, et al. A compari- supported by the Unite´ de Recher- 781-788. son between flecainide and sotalol in the preven- che The´rapeutique, Hoˆpital Lari- 10. de Denus S, Sanoski CA, Carlsson J, Opolski G, tion of recurrences of paroxysmal atrial fibrilla- boisière, Paris. Spinler SA. Rate vs rhythm control in patients with tion [in Italian]. G Ital Cardiol. 1995;25:51-68. Role of the Sponsor: The sponsor atrial fibrillation: a meta-analysis. Arch Intern Med. 26. Channer KS, Birchall A, Steeds RP, et al. A 2005;165:258-262. randomized placebo-controlled trial of pre- had no role in the conduct of the 11. Southworth MR, Zarembski D, Viana M, Bauman treatment and short- or long-term maintenance study or in the collection, analysis, J. Comparison of sotalol versus quinidine for main- therapy with amiodarone supporting DC cardio- and interpretation of the data. It had tenance of normal sinus rhythm in patients with version for persistent atrial fibrillation. Eur Heart no role regarding the preparation, re- chronic atrial fibrillation. Am J Cardiol. 1999; J. 2004;25:144-150. view, and approval of the manu- 83:1629-1632. 27. Touboul P, Brugada J, Capucci A, Crijns HJ, Ed- 12. Zarembski DG, Nolan PE Jr, Slack MK, Caruso AC. vardsson N, Hohnloser SH. Dronedarone for pre- script, or the decision to publish. Treatment of resistant atrial fibrillation: a meta- vention of atrial fibrillation: a dose-ranging study. Additional Resources: The online- analysis comparing amiodarone and flecainide. Eur Heart J. 2003;24:1481-1487. only eFigures are available at http: Arch Intern Med. 1995;155:1885-1891. 28. Pedersen OD, Bagger H, Keller N, Marchant B, //www.archinternmed.com. 13. Nichol G, McAlister F, Pham B, et al. Meta- Kober L, Torp-Pedersen C. Efficacy of dofetilide Acknowledgment: The protocol for analysis of randomised controlled trials of the ef- in the treatment of atrial fibrillation-flutter in pa- fectiveness of antiarrhythmic agents at promot- tients with reduced left ventricular function: a Dan- this review, with the detailed search ing sinus rhythm in patients with atrial fibrillation. ish Investigations of Arrhythmia and Mortality on strategy and methods, has been pub- Heart. 2002;87:535-543. Dofetilide (DIAMOND) substudy. Circulation. 2001; lished in the Cochrane Library. We 14. Miller MR, McNamara RL, Segal JB, et al. Effi- 104:292-296. thank the Cochrane Heart Group for cacy of agents for pharmacologic conversion of 29. Dogan A, Ergene O, Nazli C, et al. Efficacy of their review process and help in atrial fibrillation and subsequent maintenance of propafenone for maintaining sinus rhythm in sinus rhythm: a meta-analysis of clinical trials. patients with recent onset or persistent atrial searching and obtaining articles; all J Fam Pract. 2000;49:1033-1046. fibrillation after conversion: a randomized, of the authors who answered our re- 15. Alderson P, Green S, Higgins JPT, eds. MEDLINE placebo-controlled study. Acta Cardiol. 2004; quest for additional data; Charles highly sensitive search strategy. Cochrane Re- 59:255-261. Caulin, MD, PhD, for his most valu- viewers’ Handbook 4.2.2 (updated December 30. Chimienti M, Cullen MT Jr, Casadei G. Safety of 2003), Appendix 5B. Available at: http://www long-term flecainide and propafenone in the able suggestions; and Barbara .cochrane.dk/cochrane/handbook/hbook.htm. Ac- management of patients with symptomatic par- Stadler, MD, for translating articles cessed March 10, 2005. oxysmal atrial fibrillation: report from the from German. 16. Friedman PL, Stevenson WG. Proarrhythmia. Am Flecainide and Propafenone Italian Study (FAPIS) J Cardiol. 1998;82:50N-58N. Investigators. Am J Cardiol. 1996;77: 17. Moher D, Cook DJ, Eastwood S, Olkin I, Rennie 60A-75A. REFERENCES D, Stroup DF. Improving the quality of reports of 31. Galperin J, Elizari MV, Chiale PA, et al; GEFCA In- meta-analyses of randomised controlled trials: the vestigators–GEMA Group. Efficacy of amioda- 1. Benjamin EJ, Wolf PA, D’Agostino RB, Silbershatz QUOROM statement, Quality of Reporting of rone for the termination of chronic atrial fibrilla- H, Kannel WB, Levy D. Impact of atrial fibrillation Meta-analyses. Lancet. 1999;354:1896-1900. tion and maintenance of normal sinus rhythm: a

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 727

Downloaded From: https://jamanetwork.com/ on 09/26/2021 prospective multicenter, randomized, con- 42. Naccarelli GV, Dorian P, Hohnloser SH, Coumel 53. Atarashi H, Inoue H, Fukunami M, Sugi K, Ha- trolled, double blind trial. J Cardiovasc Pharmacol P; Flecainide Multicenter Atrial Fibrillation Study mada C, Origasa H; Sinus Rhythm Maintenance Ther. 2001;6:341-350. Group. Prospective comparison of flecainide in Atrial Fibrillation Randomized Trial (SMART) In- 32. Hillestad L, Bjerkelund C, Dale J, Maltau J, Stor- versus quinidine for the treatment of paroxysmal vestigators. Double-blind placebo-controlled trial stein O. Quinidine in maintenance of sinus rhythm atrial fibrillation/flutter. Am J Cardiol. 1996;77: of aprindine and digoxin for the prevention of after electroconversion of chronic atrial fibrilla- 53A-59A. symptomatic atrial fibrillation. Circ J. 2002;66: tion: a controlled clinical study. Br Heart J. 1971; 43. Okishige K, Nishizaki M, Azegami K, Igawa M, Ya- 553-556. 33:518-521. mawaki N, Aonuma K. Pilsicainide for conver- 54. de Paola AA, Veloso HH. Efficacy and safety of so- 33. Hohnloser SH, van de Loo A, Baedeker F. Effi- sion and maintenance of sinus rhythm in chronic talol versus quinidine for the maintenance of si- cacy and proarrhythmic hazards of pharmaco- atrial fibrillation: a placebo-controlled, multi- nus rhythm after conversion of atrial fibrillation: logic cardioversion of atrial fibrillation: prospec- center study. Am Heart J. 2000;140:437-444. SOCESP Investigators, the Cardiology Society of tive comparison of sotalol versus quinidine. JAm 44. Fetsch T, Bauer P, Engberding R, et al; Preven- Sao Paulo. Am J Cardiol. 1999;84:1033-1037. Coll Cardiol. 1995;26:852-858. tion of Atrial Fibrillation after Cardioversion In- 55. Sodermark T, Jonsson B, Olsson A, et al. Effect 34. Juul-Moller S, Edvardsson N, Rehnqvist- vestigators. Prevention of atrial fibrillation after car- of quinidine on maintaining sinus rhythm after con- Ahlberg N. Sotalol versus quinidine for the main- dioversion: results of the PAFAC trial. Eur Heart tenance of sinus rhythm after direct current con- J. 2004;25:1385-1394. version of atrial fibrillation or flutter: a multicen- version of atrial fibrillation. Circulation. 1990; 45. Plewan A, Lehmann G, Ndrepepa G, et al. Main- tre study from Stockholm. Br Heart J. 1975; 82:1932-1939. tenance of sinus rhythm after electrical cardio- 37:486-492. 35. Kalusche D, Stockinger J, Betz P, Roskamm H. version of persistent atrial fibrillation: sotalol vs 56. Patten M, Maas R, Bauer P, et al. Suppression of Sotalol and quinidine/verapamil (Cordichin) in bisoprolol. Eur Heart J. 2001;22:1504-1510. paroxysmal atrial tachyarrhythmias: results of the chronic atrial fibrillation: conversion and 12- 46. Crijns HJ, Gosselink AT, Lie KI; PRODIS Study SOPAT trial. Eur Heart J. 2004;25:1395-1404. month follow-up, a randomized comparison [in Group. Propafenone versus disopyramide for 57. Steinbeck G, Doliwa R, Bach P. Therapy of par- German]. Z Kardiol. 1994;83:109-116. maintenance of sinus rhythm after electrical car- oxysmal atrial fibrillation: cardiac glycosides alone 36. Karlson BW, Torstensson I, Abjorn C, Jansson SO, dioversion of chronic atrial fibrillation: a random- or combined with anti-arrhythmia agents [in Peterson LE. Disopyramide in the maintenance of ized, double-blind study. Cardiovasc Drugs Ther. German]? Dtsch Med Wochenschr. 1988;113: sinus rhythm after electroconversion of atrial fi- 1996;10:145-152. 1867-1871. brillation: a placebo-controlled one-year fol- 47. Pritchett EL, Page RL, Carlson M, Undesser K, Fava 58. Stroobandt R, Stiels B, Hoebrechts R; Propafe- low-up study. Eur Heart J. 1988;9:284-290. G; Rythmol Atrial Fibrillation Trial (RAFT) Inves- none Atrial Fibrillation Trial Investigators. 37. Kochiadakis GE, Igoumenidis NE, Hamilos MI, et al. tigators. Efficacy and safety of sustained-release Propafenone for conversion and prophylaxis Long-term maintenance of normal sinus rhythm propafenone (propafenone SR) for patients of atrial fibrillation. Am J Cardiol. 1997;79: in patients with current symptomatic atrial fibril- with atrial fibrillation. Am J Cardiol. 2003;92: 418-423. lation: amiodarone vs propafenone, both in low 941-946. 59. Van Gelder IC, Crijns HJ, Van Gilst WH, Van Wijk doses. Chest. 2004;125:377-383. 48. Reimold SC, Cantillon CO, Friedman PL, Antman LM, Hamer HP, Lie KI. Efficacy and safety of fle- 38. Kochiadakis GE, Igoumenidis NE, Marketou ME, EM. Propafenone versus sotalol for suppression cainide acetate in the maintenance of sinus rhythm Kaleboubas MD, Simantirakis EN, Vardas PE. of recurrent symptomatic atrial fibrillation. Am J after electrical cardioversion of chronic atrial fi- Low dose amiodarone and sotalol in the treat- Cardiol. 1993;71:558-563. brillation or atrial flutter. Am J Cardiol. 1989; ment of recurrent, symptomatic atrial fibrillation: 49. Richiardi E, Gaita F, Greco C, et al. Propafenone 64:1317-1321. a comparative, placebo controlled study. Heart. versus hydroquinidine in long-term pharmaco- 60. Villani R, Zoletti F, Veniani M, Locati F, Nava S. 2000;84:251-257. logical prophylaxis of atrial fibrillation [in Italian]. A comparison between amiodarone and disopyr- 39. Kochiadakis GE, Igoumenidis NE, Hamilos ME, et al. Cardiologia. 1992;37:123-127. amide in a delayed-release formulation in the Sotalol versus propafenone for long-term main- 50. Singh BN, Singh SN, Reda DJ, et al. Amiodarone prevention of recurrences of symptomatic atrial fi- tenance of normal sinus rhythm in patients with versus sotalol for atrial fibrillation: the Sotalol brillation [in Italian]. Clin Ter. 1992;140:35-39. recurrent symptomatic atrial fibrillation. Am J Amiodarone Atrial Fibrillation Efficacy Trial 61. Vitolo E, Tronci M, Larovere MT, Rumolo R, Mora- Cardiol. 2004;94:1563-1566. (SAFE-T). N Engl J Med. 2005;352:1861-1872. 40. Kuhlkamp V, Schirdewan A, Stangl K, Homberg 51. Singh S, Zoble RG, Yellen L, et al. Efficacy and bito A. Amiodarone versus quinidine in the pro- M, Ploch M, Beck OA. Use of metoprolol CR/XL safety of oral dofetilide in converting to and phylaxis of atrial fibrillation. Acta Cardiol. 1981; to maintain sinus rhythm after conversion from maintaining sinus rhythm in patients with chronic 36:431-444. persistent atrial fibrillation: a randomized, double- atrial fibrillation or atrial flutter: the Symptomatic 62. Alboni P, Botto GL, Baldi N, et al. Outpatient treat- blind, placebo-controlled study. J Am Coll Cardiol. Atrial Fibrillation Investigative Research on Dofeti- ment of recent-onset atrial fibrillation with the “pill- 2000;36:139-146. lide (SAFIRE-D) study. Circulation. 2000;102: in-the-pocket” approach. N Engl J Med. 2004; 41. Lloyd EA, Gersh BJ, Forman R. The efficacy of 2385-2390. 351:2384-2391. quinidine and disopyramide in the maintenance 52. Singh S, Saini RK, DiMarco J, Kluger J, Gold R, 63. Wazni OM, Marrouche NF, Martin DO, et al. Ra- of sinus rhythm after electroconversion from atrial Chen YW. Efficacy and safety of sotalol in digi- diofrequency ablation vs antiarrhythmic drugs fibrillation: a double-blind study comparing quini- talized patients with chronic atrial fibrillation: the as first-line treatment of symptomatic atrial dine, disopyramide and placebo. S Afr Med J. 1984; Sotalol Study Group. Am J Cardiol. 1991;68: fibrillation: a randomized trial. JAMA. 2005; 65:367-369. 1227-1230. 293:2634-2640.

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM 728

Downloaded From: https://jamanetwork.com/ on 09/26/2021 WEB-ONLY CONTENT

Total No. of Events Peto OR (95% CI)

Drugs Studied No. of Studies Antiarrhythmic Control P Value Antiarrhythmic vs Control 0.10 1 10 Class IA Disopyramide phosphate2 9/75 2/71 3.85 (1.13-13.2) .03

Quinidine sulfate: higher dose5 44/233 12/209 3.58 (2.01-6.40) <.001 ∗ Quinidine: lower dose 2 181/895 77/339 0.81 (0.59-1.10) .18

Quinidine: all studies7 225/1128 89/548 1.90 (0.90-4.02)† .39

All class IA8 234/1203 91/594 2.02 (1.00-4.10)† .05

Class IB All: aprindine hydrochloride1 2/47 3/47 0.66 (0.11-3.95) .65

Class IC Flecainide acetate3 7/71 0/78 9.14 (1.94-42.9) .005

Propafenone hydrochloride5 94/720 23/378 1.69 (1.09-2.62) .02

All class IC9 103/843 23/466 1.93 (1.27-2.93) .002

Class II All: metoprolol tartrate1 20/197 6/197 3.16 (1.43-6.99) .005

Class III Amiodarone3 20/161 1/113 5.55 (2.24-13.7) <.001

Dofetilide1 9/182 2/68 1.61 (0.41-6.23) .49 ∗ Sotalol hydrochloride: PAFAC, SOPAT 2 149/647 77/339 0.95 (0.68-1.33) .77

Sotalol: rest of studies6 44/483 7/344 3.02 (1.65-5.53) <.001

Sotalol: all studies8 193/1130 84/683 1.47 (0.84-2.06)† .14

Azimilide dihydrochloride + dronedarone2 68/1042 11/537 2.46 (1.51-4.01) <.001

All class III14 290/2515 98/1401 1.63 (1.29-2.07) <.001

Comparing 2 Antiarrhythmics

Drug A Drug B 0.10 1 10 Disopyramide vs other class I drugs2 6/60 12/53 0.37 (0.14-1.03) .06

Quinidine vs flecainide2 37/132 22/137 2.04 (1.14-3.64) .02

Other class I drugs4 64/258 34/268 2.25 (1.45-3.51) <.001

Sotalol6 230/1109 171/869 1.12 (0.89-1.40) .34

Flecainide vs propafenone2 12/145 18/152 0.68 (0.32-1.43) .31

Amiodarone vs class I drugs4 41/298 60/262 0.52 (0.34-0.81) .004

Sotalol2 31/219 24/196 1.18 (0.67-2.07) .58

Sotalol vs class I except quinidine 4 32/290 56/277 0.45 (0.28-0.72) .002

eFigure 1. Withdrawals due to adverse effects. Some studies compared more than 2 drugs, so the total numbers of studies and patients shown are greater than the absolute numbers of studies and patients included. CI indicates confidence interval; OR, odds ratio; PAFAC, Prevention of Atrial Fibrillation After Cardioversion study; and SOPAT, Suppression of Paroxysmal Atrial Tachyarrhythmias. Comparisons marked with an asterisk were from the PAFAC and SOPAT trials in both cases, which showed heterogeneity compared with other studies of quinidine or sotalol. Dagger indicates that OR was calculated by random effects model because the test for heterogeneity between pooled studies was significant for those comparisons. Boldface P values are significant.

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM E1

Downloaded From: https://jamanetwork.com/ on 09/26/2021 Total No. of Events Peto OR (95% CI)

Drugs Studied No. of Studies Antiarrhythmic Control P Value Antiarrhythmic vs Control 0.10 1 10 Class IA Disopyramide phosphate 2 0/75 0/71 Not Estimable NA

Quinidine sulfate 7 33/1128 6/548 2.10 (1.02-4.33) .04

All class IA 8 33/1203 6/594 2.06 (1.00-4.26) .05

Class IB All: aprindine hydrochloride 1 0/47 0/47 Not Estimable NA

Class IC Flecainide acetate 3 9/71 1/78 5.97 (1.67-21.3) .006

Propafenone hydrochloride 5 6/720 2/378 1.52 (0.33-7.02) .59

All class IC 9 15/843 3/466 3.41 (1.28-9.09) .01

Class II All: metoprolol tartrate 1 15/197 0/197 7.96 (2.84-22.3) <.001

Class III Amiodarone 4 12/428 2/245 2.65 (0.88-8.00) .08

Dofetilide 2 16/431 1/325 3.77 (1.31-10.8) .01

Sotalol hydrochloride 9 60/1391 13/815 2.20 (1.31-3.69) .003

Azimilide dihydrochloride + dronedarone 2 11/1042 1/537 3.30 (1.01-10.8) .05

All class III 16 99/3292 17/1922 2.20 (1.31-3.69) <.001

Comparing 2 Antiarrhythmics

Drug A Drug B 0.10 1 10 Disopyramide vs other class I drugs 2 1/60 2/53 0.43 (0.04-4.25) .47

Quinidine vs flecainide 2 12/132 8/137 1.60 (0.64-3.96) .31

Other class I drugs 4 15/258 10/268 1.59 (0.71-3.56) .26

Sotalol 6 32/1109 29/869 1.00 (0.60-1.68) >.99

Flecainide vs propafenone 2 2/145 5/152 0.43 (0.10-1.93) .27

Amiodarone vs class I drugs 3 8/264 23/221 0.28 (0.13-0.59) <.001

Sotalol 3 13/486 20/457 0.60 (0.30-1.20) .15

Sotalol vs class I except quinidine 4 26/290 31/277 0.77 (0.44-1.34) .35

eFigure 2. Proarrhythmia (either bradycardia or tachycardia). Some studies compared more than 2 drugs, so the total numbers of studies and patients shown are greater than the absolute numbers of studies and patients included. CI indicates confidence interval; NA, not applicable; OR, odds ratio. Boldface P values are significant.

(REPRINTED) ARCH INTERN MED/ VOL 166, APR 10, 2006 WWW.ARCHINTERNMED.COM E2

Downloaded From: https://jamanetwork.com/ on 09/26/2021