Antiarrhythmic Drugs for Maintaining Sinus Rhythm After Cardioversion of Atrial Fibrillation a Systematic Review of Randomized Controlled Trials
Total Page:16
File Type:pdf, Size:1020Kb
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 con- trols (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 car- dioversion 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 ©2006 American Medical Association. All rights reserved. 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.