Atrial fibrillation is a common complication after cardiac surgery. Postoperative atrial fibrillation is associated with increased risks of morbidity and mortality, and, therefore, preventive strategies using oral amiodarone have been developed but are often unpractical. Intravenous amiodarone administered after the induction of anesthesia and continued postoperatively for 48 h could represent an effective strategy to prevent postoperative atrial fibrillation in patients undergoing cardiac valvular surgery.
Single-center, double-blinded, double-dummy, randomized controlled trial in patients undergoing valvular surgery. Patients received either an intravenous loading dose of 300 mg of amiodarone or placebo in the operating room, followed by a perfusion of 15 mg . kg(-1) . 24 h(-1) for 2 days. The primary endpoint was the development of atrial fibrillation occurring at any time within the postoperative period.
One hundred twenty patients were randomly assigned (mean age was 65 +/- 11 yr). Overall atrial fibrillation occurred more frequently in the perioperative intravenous amiodarone group compared with the placebo group (59.3 vs. 40.0%; P = 0.035). Four preoperative factors were found to be independently associated with a higher risk of developing postoperative atrial fibrillation: older age (P = 0.0003), recent myocardial infarction (<6 months; P = 0.026), preoperative angina (P = 0.0326), and use of a calcium channel blocker preoperatively (P = 0.0078) when controlling for groups.
In patients undergoing cardiac valvular surgery, a strategy using intravenous amiodarone for 48 h is not efficacious in reducing the risk of atrial fibrillation during cardiac valvular surgery.
What We Already Know about This Topic
❖ New onset atrial fibrillation after cardiac surgery carries significant morbidity.
❖ The incidence of atrial fibrillation can be reduced with oral amiodarone begun before surgery.
What This Article Tells Us That Is New
❖ In patients with elective valve surgery, intravenous amiodarone, begun after induction of anesthesia up to 48 h failed to reduce the incidence of atrial fibrillation.
A TRIAL fibrillation (AF) is an important and frequent complication after cardiac surgery occurring in almost one third of patients undergoing coronary artery bypass grafting (CABG)1and in up to 44% of patients undergoing a valvular procedure.2Heart failure, hypotension, increased risk of stroke, need for anticoagulation, increased length of stay in the hospital, and long-term mortality are some of the various potential consequences of postoperative AF.3–5Although valvular surgery poses a greater risk of AF,2,6,7most studies in cardiac surgery have been performed in patients undergoing coronary revascularization.2,6,8–11Reduction in the incidence of postoperative AF of up to 50% have been demonstrated in some of these studies,12when amiodarone was started before entry in the operating room. In several of these protocols, the preoperative loading regimen was administered orally2,6,9–11; however, this approach is currently impractical because most patients undergoing elective cardiac surgery are admitted the night before or even on the day of the procedure. However, none of the studies have looked at the hemodynamic safety and efficacy of starting amiodarone loading and infusion after anesthesia induction, before the onset of cardiopulmonary bypass (CPB), without a previous loading dose. Our hypothesis was that perioperative use of intravenous (IV) amiodarone initiated after the induction of anesthesia would be hemodynamically safe and effective in preventing postoperative AF until hospital discharge in patients undergoing cardiac valvular surgery.
Materials and Methods
The study protocol was reviewed and approved by the Ethics and Research Committee of the Montreal Heart Institute, Montreal, Quebec, Canada. All patients gave written informed consent.
The primary objective of this single-center double-blinded, double-dummy, randomized controlled trial was to demonstrate the efficacy and safety of a 48-h IV infusion of amiodarone in reducing AF prevalence in patients undergoing valvular surgery. From November 2001 to May 2003, patients more than 18 yr old undergoing an isolated cardiac valvular surgery or a valvular surgery combined with a coronary revascularization procedure were screened to be included in the study. To be eligible, they also had to be in sinus rhythm and have a left ventricular ejection fraction more than 20% at the time of screening. Patients were excluded from the study if they met one of the following criteria: amiodarone intake in the previous 6 months, a history of anaphylactic reaction to iodine, a history of severe reaction or toxicity to amiodarone, intake of class I or III antiarrhythmic agents within the previous 48 h before surgery, significant hypotension (systolic blood pressure < 80 mmHg) necessitating sustained treatment with vasoactive agents for more than 1 h preoperatively, urgent surgery, and participation in other investigational trials.
Randomization and Study Protocol
Patients were randomly assigned (1:1 ratio) in a double-blinded fashion, from a computer-generated list implemented by the hospital pharmacists not involved in the trial, to receive either IV amiodarone (Sabex, Boucherville, Quebec, Canada) or placebo (dextrose 5% in normal saline) infused through a central venous access. Because the solvent contained in the amiodarone solution is a soap (Polysorbate 80), shaking the bottle containing the amiodarone solution causes bubbles. To maintain the blind, the bottles containing amiodarone or placebo were covered with opaque or aluminum paper. Induction of anesthesia was performed using a combination of fentanyl (5–10 μg/kg) or sufentanil (0.7–1 μg/kg), midazolam (up to 0.1 mg/kg), and pancuronium (0.1 mg/kg). Isoflurane was used to control blood pressure during maintenance of anesthesia. After anesthesia induction, an IV loading dose of 300 mg of amiodarone (or placebo) was given for more than 10 min followed by an infusion of amiodarone (15 mg · kg−1· 24 h−1, max 1,500 mg/24 h) or placebo for 48 h. Complete hemodynamic assessment using the pulmonary artery catheter (7.5F 931HF75; Baxter Healthcare, Irvine, CA), transesophageal echocardiographic examination (to verify the adequacy of the valvular procedure), and complete laboratory data including arterial and venous blood gases were obtained just before and after the bolus and after weaning from CPB, during sternal closure. The surgical procedure was performed according to established guidelines. Blood cardioplegia was used in all patients. Induction and maintenance of cardioplegia were cold to tepid (15°–29°C). The blood to crystalloid ratio was 4:1. The pump flow was adjusted to obtain an output of 2.2 l · min−1· m−2of body surface area and was reduced to 0.5 l · min−1· m−2for aortic clamping and unclamping. SIII (Stockert, Munich, Germany) roller pumps were used in all patients. The oxygenator was Sorin Monolyth (Mirandola, Italy). Valve and complex procedures were done with temperatures of 32°–34°C. Weaning from CPB was attempted after systemic temperature (central and vesical) was more than 36°C. All patients in the study had epicardial pacemaker wires (atrial or ventricular) placed at the end of surgery. The use of vasoactive drugs (noradrenaline, adrenaline, vasopressin, and milrinone) and the process of weaning from CPB were done using established protocols13(see figs. 1 and 2for the Vasoactive Protocol Management, Supplemental Digital Content 1, http://links.lww.com/ALN/A562). Temporary pacing was subsequently initiated if judged necessary by the anesthesiologist and the surgeon.
On arrival in the cardiothoracic intensive care unit (CTICU), a Holter monitor (Marquette Electronics Series 8500, Boston, MA) was installed on each patient. Three-lead continuous Holter monitoring was performed for the first 4 postoperative days. The recorded data were stored for 24 h and reviewed by an independent electrophysiologist on a daily basis. Three-lead continuous telemetric monitoring (Fukuda DF 3310 and LW 3100, Fukuda, Japan) was concomitantly performed from the time of admission to the CTICU until hospital discharge. Daily 12-lead electrocardiogram recordings were also performed on all patients. Postoperative AF was treated by the CTICU and surgical teams at their discretion, in accordance with the American College of Cardiology and American Heart Association Guidelines.14AF is defined as an uncoordinated atrial activation with consequent deterioration of atrial mechanical function.14Rescue amiodarone (IV or oral) could be used for the treatment of an AF episode if judged necessary by the treating physicians. To avoid administering potentially excessive additional doses of amiodarone to patients in the active treatment group, we developed a weight-based chart, indicating the maximal amount of additional amiodarone a patient was allowed to receive during the first 24 and 48 h after the surgery. The chart was made so as not to exceed a total IV amiodarone dose of 2,000 mg/24 h ( appendix). Hemodynamic assessment, electrocardiographic, and laboratory data were systematically recorded for the first 6 postoperative days and then were recorded if an event occurred. Assessment of clinical variables was performed daily for the entire hospitalization period until the day of discharge. One-month telephone follow-up was performed to assess AF recurrences necessitating hospitalization.
The primary endpoint was the development of AF occurring at any time within the postoperative period (defined as the interval from the time of sternal closure to hospital discharge). AF was defined as an irregular rhythm with episode lasting more than 30 min or any episode requiring urgent treatment because of associated hemodynamic compromise (heart failure, hypotension, and ischemia) or symptomatic discomfort (shortness of breath, palpitations, and chest pain). Secondary endpoints consisted in adverse hemodynamic and electrophysiologic effects of IV amiodarone, temporary pacemaker requirement, length of CTICU and hospital stay, morbidity and mortality rates, and the rate of recurrent AF necessitating hospitalization at 1 month. The relationship between IV amiodarone use and postoperative nausea, the time of AF onset, and the mean heart rate response were also studied. Finally, the relationship between postoperative AF, and multiple preoperative, intraoperative, or postoperative variables were also studied.
Clinical Variables and Endpoints
Age, gender, body mass index, and body surface area were determined for each patient along with their relevant medications. The presence of hypertension, diabetes, chronic renal failure, smoking history, recent myocardial infarction (MI, before or after 6 months), signs and symptoms of congestive heart failure, chronic obstructive pulmonary disease, previous cerebrovascular disease, thyroid disorders, and left ventricular ejection fraction measured through angiographic ventriculography or echocardiography were also documented. The different types of surgical procedures were classified as isolated valvular or valvular and CABG. The number of bypass grafts and the use of a mammary artery were noted in addition to the CPB time and aortic crossclamping time. The amounts of vasopressor and inotropic support required during and after the procedure were also noted.
On the basis of previous studies, we estimated that 50% of patients undergoing cardiac valvular surgery would develop AF. To detect the expected reduction in AF from 50% to at least 25% in the amiodarone group, 58 patients/group would be necessary to reach a power of 80% with a two-sided chi-square test at an alpha of 5%. Assuming a 3% loss, we recruited 60 patients/group. The results are presented as mean ± SD or median (minimum–maximum) according to the distribution for continuous variables and number (percentage) for categorical variables. The analyses were performed using the intent-to-treat principle. Chi-square tests were used to compare categorical variables between groups (with or without amiodarone) (tables 1 and 2). For continuous variables, the Student t test or Wilcoxon test was used to compare groups (tables 1 and 2). The means and proportions of patients with or without AF were compared using a two-way analysis of variance and the Mantel-Haenszel chi-square test adjusting for treatment group (amiodarone vs. placebo), respectively (table 3). To establish which variables predicted the risk of developing AF (overall), multiple logistic regressions were used. Daily contrasts (amiodarone vs. placebo) for AF, heart rate, and β blockers as depicted in figures 1 to 3were estimated using the generalized estimating equation framework because of the presence of correlated data introduced by days (repeated factor). To take into account the multiplicity of testing, the P values were corrected using closed multiple testing procedures, that is, the permutational minimum P value method for AF and β blockers and the Westfall-Young bootstrap minimum P value method for heart rate. To evaluate the long-term effect of AF treatment on mortality, a log-rank test was performed. Statistical analyses were performed using SAS version 8.02 (SAS Institute Inc., Cary, NC). A P value ≤ 0.05 was considered significant.
Demographic and Surgical Variables
Baseline preoperative and intraoperative characteristics of patients are shown in table 1. A total of 120 patients were randomly assigned, and one patient in the amiodarone group died intraoperatively (right ventricular failure). The final intention-to-treat analysis was performed on the remaining 119 patients (59 amiodarone and 60 placebo). The mean age was 65 ± 11 yr, and 67 (56%) patients were men. Baseline demographic characteristics were similar among groups except for a higher proportion of patients with diabetes (23 vs. 8%; P = 0.0244) and chronic obstructive pulmonary disease (23 vs. 10%; P = 0.05) in the placebo group. Two thirds of the total population (68.3%) underwent an isolated valvular surgery and one third (31.7%) underwent combined valvular and CABG surgery. Patients in the amiodarone group underwent more isolated valvular surgeries compared with the placebo group (76.7 vs. 60%; P = 0.0497). There were no significant differences in the proportion of patients undergoing mitral and/or aortic valve procedures between the two groups. The total CPB time (97 ± 32 vs. 110 ± 37 min; P = 0.0426) and aortic crossclamp time (73 ± 28 vs. 85 ± 30 min; P = 0.0271) were shorter in the amiodarone group. There were no differences between the two groups in terms of baseline hemodynamic variables, laboratory, or blood gas data for all the perioperative phases (prebolus, postbolus, and post-CPB).
Primary and Other Endpoints
Postoperative AF occurred in 59 patients (49.6%) and was more frequent in the treatment group (n = 35/59 or 59.3%) compared with the placebo group (n = 24/60 or 40.0%; P = 0.035) (table 2). The difference remained significant even after adjusting for diabetes, chronic obstructive pulmonary disease, the duration of CPB, and the use of preoperative β blockers (P = 0.0292). AF occurred at a mean of 2.29 days postoperatively in the placebo group and at 3.03 days postoperatively in the amiodarone group (P = 0.0208). The median total dose of IV amiodarone administered to patients in the treatment group was 2,444 mg (410–3,790 mg) during the first 48 h.
As depicted in figure 1, both groups had a similarly low incidence of postoperative AF at day 1 (10.7 vs. 6.8%; P = 0.8625). These incidences increased during the following days, but the frequencies were similar between groups. The proportion of patients in sinus rhythm was similar among groups throughout the study. Figure 2illustrates the mean heart rate (beats/min) for both groups. Patients in the amiodarone group had a statistically significant reduction in mean heart rate compared with the placebo group on postoperative day 1 (67 ± 12 vs. 79 ± 11; P < 0.0001), day 2 (74 ± 14 vs. 82 ± 14; P = 0.0277), and day 3 (80 ± 13 vs. 88 ± 16; P = 0.0290). At the onset of AF, mean heart rate did not differ between groups (amiodarone group [84 ± 21], placebo group [86 ± 21]; P = 0.9851). Figure 3illustrates the use of β blocker from days 1 to 6.
Other postoperative outcomes are shown in table 2. There were no differences in the length of stay in the CTICU (64 ± 81 vs. 51 ± 39 h; P = 0.4898) or total hospitalization duration (311 ± 270 [median 228]vs. 253 ± 146 [median 192] h; P = 0.1996) between groups. No differences were observed in terms of rhythm on discharge, use of rescue oral and IV amiodarone, nonsustained ventricular tachycardia, acute respiratory distress syndrome, MI, stroke, acute renal failure, and rehospitalization for AF. The overall mortality was 2.5% (3 of 120). Excluding the patient who died intraoperatively, one patient in each group died during hospitalization. One patient in the control group died the next day after acute tamponade secondary to a rupture of the atrioventricular groove. The other patient in the amiodarone group died also the next day after repair through laparotomy of an accidental rupture of an iliac artery during the insertion of an intraaortic balloon pump. At 6 yr, 107 patients were still alive (89.2% survival rate), and one amiodarone patient was lost to follow-up. Survival at 6 yr was not different among groups. There were five deaths in the amiodarone group and eight deaths in the placebo group (P = 0.4307) (fig. 4).
There were no differences among groups in the number of patients requiring defibrillation to restore sinus rhythm after aortic crossclamp release (35 [58%]vs. 32 [53%]; P = 0.58) and in the use of vasoactive agents in the immediate postoperative period. The only significant difference in postoperative hemodynamics between the amiodarone and placebo groups observed during the hospitalization was that patients in the amiodarone group developed lower cardiac indices on the first postoperative day (2.5 ± 0.52 vs. 2.9 ± 0.67 l · min−1· m−2; P = 0.0003), with associated higher pulmonary vascular resistance indices (283 ± 104 vs. 247 ± 87 dynes · s · cm−5· m2; P = 0.0441). Prescribed postoperative medications were not different among groups, except for a higher number of patients on vasoactive agents on postoperative day 2, which was higher in the placebo group (13 [22%]vs. 4 [6.8%]; P = 0.018). There were no differences in laboratory data and blood gas values between groups. Temporary pacemaker use was approximately twice as frequent in the amiodarone group compared with the placebo group, but this difference was only statistically significant for the first postoperative day (24 [41%]vs. 11 [18%]; P = 0.0075).
Outcome of Patients with AF Compared with Those without AF
Patients who developed AF presented with more complications, even adjusting for therapies received (table 3). A longer duration of postoperative mechanical ventilation (P = 0.0207), more frequent ventricular tachycardia (P = 0.004), and acute renal failure (P = 0.0384) were observed in patients with postoperative AF. Lengths of CTICU and hospital stays were similar in patients who developed AF compared with those who did not. At hospital discharge, a majority of patients were in sinus rhythm with no differences between groups.
Predictors of Postoperative AF and Outcome with AF
Four preoperative factors were found to be independently associated with a higher risk of developing postoperative AF: older age (P = 0.0003), recent MI (< 6 months; P = 0.026), preoperative angina (P = 0.0326), and use of a calcium channel blocker (CCB) preoperatively (P = 0.0078) when controlling for groups. Eighteen of 26 patients (69%) who were receiving a CCB preoperatively and 15 of 19 patients (80%) who were put on a CCB after surgery developed AF (P = 0.0078 and 0.0065, respectively). Preoperative use of β blockers was similar in the amiodarone and placebo group (19 [31.7%]vs. 15 [25%]; P = 0.42) and not correlated with the development of postoperative AF. Figure 3illustrates the use of β blockers postoperatively. No intraoperative (type of surgery, total CPB time, and aortic crossclamp time) or postoperative factors were found to be predictive of postoperative AF.
In this double-blinded, randomized, controlled trial on AF prevention in patients undergoing cardiac valvular surgery, we observed that a strategy using IV amiodarone initiated in the operating room before CPB followed by an infusion for 48 h is hemodynamically safe. There was a trend in reduction of the risk of AF while patients were receiving the medication but was no longer effective after the infusion was stopped. However, this potential benefit was lost at 4 days when the trend was reversed, and the amiodarone group had more AF than the placebo group. Therefore, this 48-h IV strategy resulted in an overall occurrence of AF that was higher in the patients who received perioperative IV amiodarone compared with placebo.
Most of the studies using amiodarone to prevent incident AF have been conducted in patients undergoing CABG with very few including patients undergoing isolated valvular surgery or combined with CABG.12Only six studies have included valvular surgical patients.2,6,8–11In addition, only four studies involved IV amiodarone only in the perioperative period8,11,15,16and only two studies included valvular patients.8,11
Our results are similar to those observed in a study by Daoud et al. 6in which 71 patients (57% incident AF) underwent valvular surgery. AF occurred in 46% of patients undergoing valvular surgery, and incidence was reduced by approximately 50% in the amiodarone-treated patients. However, Daoud et al. used pre- and postoperative oral amiodarone. Guarnieri et al. 8used IV amiodarone, which was started postoperatively for 48 h and observed a 26% reduction in AF; however, only 42 patients (14%) underwent valvular surgery in that study. The largest and most recent Prophylactic Amiodarone for Prevention of Arrhythmias that Begin Early After Revascularization (PAPABEAR) trial included 212 patients (35.3% of total population) undergoing valvular surgery.2Using a 12-day strategy of oral amiodarone (6 days before and 6 days after surgery), the investigators observed a 13.4% absolute risk reduction in the incidence of AF. In patients undergoing valvular surgery with or without CABG, AF occurred in 23.8% in the amiodarone group compared with 44.1% in the placebo group. We observed similar rates of AF in the placebo group (40.1%) in our study. The PAPABEAR trial patients were younger, the majority were men, the use of angiotensin-converting enzyme inhibitors and β blockers was almost twice more frequent, and the duration of CPB was shorter. The larger proportion of patients undergoing revascularization explains to some extent the latter difference. Despite a 50% reduction in the occurrence of postoperative AF with the use of oral amiodarone in the PAPABEAR study, the authors observed that the total burden of AF was similar in both the amiodarone and the placebo groups (25.1 vs. 23.7 h; P = not significant).
The definition of AF and the monitoring tool to diagnose the arrhythmia are also important to consider when comparing such studies. Holter monitoring offers a continuous, more reliable beat-to-beat analysis of the cardiac rhythm and may thus be best suited for arrhythmia detection and diagnosis. In our study, we defined AF as an episode lasting more than 30 min or any episode requiring urgent treatment because of associated hemodynamic compromise or symptomatic discomfort. This precise definition, which we used for AF occurrence, may explain in part why our results did not show a reduction in AF occurrence with amiodarone. Holter monitoring was performed on all patients, and the use of a more rigorous diagnostic tool could be partly responsible for better assessment of AF occurrence in both groups. Prevention of any episode of sustained AF should be the primary goal to be achieved.
The higher trend of AF at day 4 in the amiodarone group was associated with 50% less patients taking a β blocker in this group on that day compared with the placebo group. This may suggest that a strategy in which β blockers are introduced in the postoperative phase could represent a potentially effective prophylactic measure, as suggested by a meta-analysis.12As a consequence of the lower heart rate in the amiodarone group, temporary pacemaker use was approximately twice as important in the amiodarone group for the first postoperative day compared with the placebo group, which is similar to the AF Suppression Trial study.10
Older age, recent MI (< 6 months), angina, and use of a CCB were found to be associated with postoperative AF. Older age has been shown to be a strong predictive factor of postoperative AF in previous studies.4,17,18The occurrence of a recent MI has been related to cardiac remodeling and atrial dilatation, which is a risk factor for AF.9Preoperative and postoperative use of a CCB was associated with a higher risk of developing postoperative AF. This higher occurrence of AF in patients taking a CCB has previously been described in the literature.19–22
In patients undergoing cardiac valvular surgery, a strategy using IV amiodarone perioperatively did not reduce the burden of postoperative AF in patients after cardiac valvular surgery.
The authors thank the Anesthesiologists (Christian Ayoub, M.D., Sylvain Bélisle, M.D., Robert Blain, M.D., Jennifer Cogan, M.D., Alain Deschamps, M.D., Ph.D., Gisèle Hemmings, M.D., J. Sébastien Lebon, M.D., Baqir Qizilbash, M.D., Antoine Rochon, M.D., Jean Taillefer, M.D., and Karine Toledano, M.D.), the Cardiothoracic Surgeons (Denis Bouchard, M.D., Raymond Cartier, M.D., Philippe Demers, M.D., Yves Hébert, M.D., Hugues Jeanmart, M.D., Michel Pellerin, M.D., and Louis P. Perrault, M.D., Ph.D.), the Clinicians (Brigitte Ducharme, M.D., Daniel Parent, M.D., and Jacynthe Thibodeau, M.D.) of the Cardiothoracic Intensive Care Unit of the Montreal Heart Institute, Montreal, Quebec, Canada, for their support of this study. The authors also thank Denis Babin, M.Sc., Research Assistant, Micheline Roy, Research Assistant, and Luce Begin, Administrative Agent of the Research Centre, Montreal Heart Institute.