Preoperative amiodarone as prophylaxis against atrial fibrillation after heart surgery 

Authors Daoud E, Strickberger S, Man K, Goyal R, Deeb G, Bolling S, Morady F, et al. 
Source New England Journal of Medicine. 337:1785-91. December 18, 1997. 
Institutions University of Michigan Hospital, Ann Arbor; Harper Hospital, Wayne State University, Detroit. 
Support Wyeth-Ayerst Laboratories. 


Atrial fibrillation is a common arrhythmia after heart surgery and can prolong hospitalization. Amiodarone is highly effective against atrial fibrillation and therapy can be initiated fairly safely on an outpatient basis. The purpose of this study was to examine the effect of amiodarone, started at least one week prior to cardiac surgery, on the incidence of postoperative atrial fibrillation. 



Patients were screened at the cardiac surgery clinics of the two hospitals. They had to be at least 18 years old, in normal sinus rhythm, scheduled for surgery requiring cardio-pulmonary bypass with at least one week preoperatively available for drug therapy. 

Main exclusion criteria were those making amiodarone therapy inadvisable: history of amiodarone allergy or toxicity, abnormal liver functions, severe congestive heart failure, thyroid disease, pregnancy or bradycardia. Patients were also excluded if they had recently taken amiodarone or were taking anti-arrhythmic drugs other than beta-blockers, calcium-blockers or digitalis. 


Patients were randomized to either amiodarone or placebo at the time they were screened in the clinic. Patients randomized to amiodarone received 600 mg daily (200 mg tid) for 7 days, followed by 200 mg daily until hospital discharge after surgery. 

Patients on digitalis or warfarin had their doses of these drugs halved; patients on warfarin then had a prothrombin time checked 1 week later. 

Patients were monitored by telemetry after surgery and the EKG data was reviewed daily. Episodes of atrial fibrillation were counted if they lasted at least 5 minutes; management was by the cardiac surgical team. Post-discharge, patients were evaluated by a nurse a mean of 7 days after discharge and also at the routine post-surgical visit, at which time they were queried about hospitalizations and ER visits. 


No primary endpoints are mentioned in the methods section. The authors looked at the number of patients remaining free of atrial fibrillation (using Kaplan-Meier analysis) and also at hospital length of stay and costs. The data were analyzed on an intention-to-treat basis. 



373 patients were screened, 145 were excluded, mainly because they had participated in another investigational protocol or had atrial fibrillation. Of the remaining 228 patients, 131 consented but 7 had their surgery cancelled, leaving 124 patients (64 received amiodarone, 60 placebo). 

Characteristics were well matched between the two groups. Some pertinent characteristics included (averaging the data from the two groups): 

  • Age 59; male: 68% 
  • EF: 48%; history of CHF: 1/2 
  • Previous MI: 1/3 
  • Surgery: CABG only: 42%; Valvular only: 33%; Combined: 18% 
  • Pre-op beta-blocker use: 35% 
Amiodarone therapy

The mean number of preoperative days of amiodarone therapy was 12. The mean total outpatient amiodarone dose was 4.8 g. 

There was no elevation in liver enzymes during pre-operative therapy with amiodarone; only one patient discontinued amiodarone therapy because of mild GI side-effects. 

Atrial fibrillation
  Amiodarone (64) Placebo (60)
AF occurring at least once during hospitalization 15 (23%) 25 (42%)
Maximum HR during AF 112/min 135/min
Duration of first episode of AF 13.8 hours 9.3 hours (NS)
Spontaneous conversion of AF 5/15 = 33% 9/25=36% (NS)
AF noted only after hospital discharge 1 7


In both groups, beta-blocker therapy did not significantly influence the incidence of atrial fibrillation: 
  amiodarone (64) placebo (60)
beta-blockers (44) 7/26=27% 11/18=61%
no beta-blockers (80) 8/38=21% 14/42=33%
yes vs no beta-blockers p=0.76 p=0.09

AF occurred significantly more often among patients undergoing valve surgery than among those undergoing only CABG (46% vs. 29%). 

Length of stay, costs, morbidity and mortality

In both groups, the length of stay was significantly shorter in patients without AF than in those with AF. In the amiodarone group, the mean length of stay was 6.5 days, vs. 7.9 days for the placebo group (p=0.04). 

As a result, the estimated hospital costs were $18,375 in the amiodarone group vs. $26,491 for the placebo group (p=0.03). 

There was no significant difference in morbidity and mortality between the two groups. 

Author's discussion

The authors conclude that low-dose oral amiodarone therapy begun at least one week before elective cardiac surgery significantly reduces the incidence of post-operative and symptomatic post-discharge atrial fibrillation, cutting its incidence by about 50%. The drug was well tolerated in this study, with no evidence for acute pulmonary toxicity, cardiac toxicity or pro-arrhythmia. 

They note that one study of intravenous amiodarone given just prior to heart surgery showed a significant reduction in AF, but the drug had to be discontinued in 18% of patients because of side effects. 

Beta-blockers have been shown to reduce the incidence of AF in a number of studies. Unfortunately, in some of these trials, patients assigned to placebo who had previously been on beta-blockade probably suffered from beta-blocker withdrawal, thus biasing the results against placebo. Also, the authors note that beta-blockers may not be well tolerated in many patients, including those with severe LV dysfunction and chronic lung disease. 

Whether a shorter, accelerated amiodarone loading regimen would be more effective remains to be determined. 



In this study, administering low dose oral amiodarone for two weeks prior to heart surgery significantly reduced the incidence of post-operative atrial fibrillation (from 42% to 23%). As a result, hospital length of stay and cost were reduced as well. 

It is unfortunate that no primary endpoints were specified in the methods section. This leaves the authors open to the criticism that, had the results been less impressive, the data might have been analyzed differently in an attempt to come up with something significant. 

It is worth noting that, although the authors compare beta-blocker vs. no beta-blocker in the two study groups and found no statistically significant difference, they did not give the numbers for the other "grouping": amiodarone vs. placebo in the subgroups of patients taking and not taking beta-blockers. According to the two by two table I constructed above, the difference between amiodarone and placebo is much more impressive in the subgroup of patients taking beta-blockers preoperatively (27% vs. 61%), than in the subgroup not on beta-blockers (21% vs. 33%). It is not clear to me why this should be the case. Beta-blocker withdrawal might be a contributing factor, although over 90% of patients who took beta-blockers before surgery continued to take them post-operatively. Perhaps the post-op doses or absorption were lower. Or perhaps this is just a fluke of post-hoc analysis. 

The role of amiodarone vs. beta-blockade remains to be clarified. In patients not on beta-blockers who can tolerate beta-blocker therapy, it is not clear that amiodarone is much better for preventing atrial fibrillation. In patients already on beta-blockers and at risk for beta-blocker withdrawal, on the other hand, it would seem to be more useful. In patients who cannot tolerate beta-blockers, amiodarone certainly seems like a reasonable alternative, although, as noted above, the benefits in this subgroup might be less than the 50% found by the authors in the entire study group. 

January 12, 1998 


References related to this article from the NLM's PubMed database. 

Reader Comments

Date: Wed, 21 Jan 1998 
From: "C. Sitges Serra" <> 

Surprisingly, authors make no comment about thyroid function before and/or after receiving Amiodarone. It´s well known that Amiodarone has a high content of Iodine in its molecule and, in a high percentage of patients it may provoke clinical and/or subclinical changes on respect to thyroid gland function such as hyperthyroidism or mild changes in TSH or Thyroxine. 

Amiodarone has been very questioned among cardiologists because of its benefit/risk ratio and also because there are the other therapies such as Propafenone Clorhidrate that might have similar benefits at lower risk. Lung and liver fibrosis are among other side effects that Amiodarone might cause, though probably that would need long periods of treatment. 

C. Sitges-Serra(General Practitioner). 
(Barcelona. Spain) 

    Amiodarone does have multiple, complex effects on thyroid function (see, for example, Ann Intern Med 1997;125:63-73). Some of these side-effects could appear after short-term treatment. Thyroid function studies after amiodarone therapy would have been interesting, although the number of patients studied here was probably too small to reliably exclude any alteration in thyroid function occurring in a small number of patients. 

    A study looking at amiodarone vs. beta-blockers or drugs such as propafenone (presumably initiated in the hospital) would be very useful. 

Date: Fri, 06 Feb 1998 
From: Kowshik Manohar <> 

It is an interesting study. Have there been any studies in the past regarding alterations in the thyroid functions after cardio pulmonary bypass? If there are any, giving Amiodarone intra / perioperatively may have effects, and might make things look differently? 


    Cardiopulmonary bypass can affect thyroid hormone levels (see these references). Before this therapy is widely used, a closer look at thyroid function before and after surgery (in the setting of amiodarone treatment) would need to be undertaken. -- mj 

Date: Wed, 25 Feb 1998 

I just got around to analyzing this article carefully for our internal medicine residents' journal club today, and noticed a few more problems. 

One is that results are liberally mixed in with methods (eg., fourth sentence under Methods: Study Protocol, pp. 1786): "Patients were enrolled a mean (+/- SD) of 13 +/- 7 day before surgery." And "Outpt compliance by pill count was 96%." And "A home-nurse evaluation was performed 7 +/- 2 days after the pts were discharged..."   etc. 

There is no power analysis (more important with a negative trial, but nevertheless should be performed a priori). 

In Table 1, standard deviations are given for data one would not expect to be normally distributed (eg., NYHA functional class and others). 

On page 1788 in the second full sentence, 7/64 is not 14%, as reported, but 11%. Likewise, 11/60 is not 31%, but 18%. 

No "fatal" errors, but concerning nonetheless. 

Best wishes. 

John Foxworth, PharmD 
Clinical Pharmacology Section, Dept of Medicine, Truman Medical Center-West 
Associate Professor, University of Missouri-Kansas City School of Medicine 
Clinical Pharmacology Web Page: 

    I believe the lack of a power analysis is related to the fact that the authors never specified primary endpoints for this study. If you don't specify what you're studying a priori, you can't figure out how many subjects you will need to enroll!
    The discrepancies in the percentages you point out are probably due to the fact that patients who developed atrial fibrillation were excluded from further analysis. Thus, the denominators are smaller than the original 64 and 60, and the percentages are correspondingly higher.
    I agree with all of your other comments.  -- mj

June 19, 1997

Letters to the editor about this article, from the May 7, 1998 New England Journal of Medicine.

Submit a comment

Journal Club homepage

Site Meter