Elsevier

The Lancet

Volume 350, Issue 9089, 15 November 1997, Pages 1417-1424
The Lancet

Articles
Effect of prophylactic amiodarone on mortality after acute myocardial infarction and in congestive heart failure: meta-analysis of individual data from 6500 patients in randomised trials

https://doi.org/10.1016/S0140-6736(97)05281-1Get rights and content

Summary

Background

There have been 13 randomised controlled trials of prophylactic amiodarone in patients with recent myocardial infarction (MI) or congestive heart failure (CHF). None of these was powered to detect a mortality reduction of about 20%. We undertook a meta-analysis, based on data from individual patients, to provide a more sensitive and accurate assessment of the benefits and risks of prophylactic amiodarone.

Methods

Individual data from the studies were abstracted according to a predefined protocol. The summary odds ratios were calculated according to standard methods.

Findings

There were eight post-MI and five CHF trials; nine trials were double-blind and placebo-controlled, and four compared amiodarone with usual care. 6553 patients were randomly assigned treatment, of which 78% were in post-MI trials and 22% in CHF trials. 89% had had previous MI. The mean left-ventricular ejection fraction was 31%, and median frequency of ventricular premature depolarisation 18 per h. Total mortality was reduced by 13% (odds ratio 0·87 [95% CI 0·78–0·99], p=0·030) based on classic fixed-effects meta-analysis and by 15% (0·85 [0·71–1·02], p=0·081) with the more conservative random-effects approach. Arrhythmic/sudden death was reduced by 29% (0·71 [0·59–0·85], p=0·0003). There was no effect on non-arrhythmic deaths (1·02 [0·87–1·19], p=0·84). There was no difference in treatment effect between post-MI and CHF studies. The risk of arrhythmic/sudden death in control-group patients was higher in CHF than in post-MI studies (10·7 vs 4·1%), and the best single predictor of risk of arrhythmic/sudden death among all patients was symptomatic CHF. The excess (amiodarone minus control) risk of pulmonary toxicity was 1% per year.

Interpretation

Prophylactic amiodarone reduces the rate of arrhythmic/sudden death in high-risk patients with recent MI or CHF and this effect results in an overall reduction of 13% in total mortality.

Introduction

Significant progress has been made during the past decade in reducing the mortality of patients with acute myocardial infarction (MI) by means of thrombolysis and antiplatelet agents. Nonetheless, 1-year mortality in survivors of MI remains unacceptably high, with population-based studies continuing to report rates of more than 10%.1 A substantial proportion of deaths after hospital discharge are sudden and caused by ventricular fibrillation. Patients with congestive heart failure (CHF) also have a substantial risk of death from arrhythmia.

During the past decade randomised clinical trials have investigated the ability of several antiarrhythmic drugs to reduce premature death in patients at high risk of arrhythmia.2, 3, 4, 5 Apart from β-blockers, no other agent has been conclusively shown to reduce mortality. Indeed, there have been clear increases in mortality with some class 1 and class 3 agents. Amiodarone, however, has several antiarrhythmic actions and unusual pharmacokinetics.6 The role of prophylactic amiodarone in patients at risk of death from cardiac arrhythmia has been addressed by 13 randomised controlled clinical trials.7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 These trials have investigated amiodarone in two overlapping high-risk populations—survivors of MI and patients with CHF. These populations overlap because many patients with previous MI develop left-ventricular dysfunction. Conversely, the most common cause of heart failure is ischaemic heart disease and MI. Only three of the randomised trials of amiodarone12, 16, 17 showed a significant reduction in all-cause mortality. However, only three of the studies had a sample size of more than 1000 patients. None was sufficiently large to detect reliably a moderate reduction in mortality of 10–20%. The possibility remains, therefore, that amiodarone does reduce total mortality but that this effect was not detected because of insufficient statistical power.

The two largest amiodarone trials have raised much controversy.20 The Canadian Amiodarone Myocardial Infarction Arrhythmia Trial (CAMIAT)7 and the European Myocardial Infarction Amiodarone Trial (EMIAT)8 both reported substantial and statistically significant reduction in the risk of arrhythmic death or resuscitated ventricular fibrillation. Neither trial showed a significant reduction in overall mortality. CAMIAT, with 1202 participants, was designed primarily to detect a 50% reduction in the combined outcome of arrhythmic death or resuscitated ventricular fibrillation; total mortality was a secondary outcome. EMIAT, with 1486 participants, was designed to detect a reduction in total mortality of 33%. Because neither trial had sufficient power to detect modest but important reductions in total mortality, it remains unclear whether the beneficial effect of amiodarone on arrhythmic death and resuscitated ventricular fibrillation (observed in both trials) translates into a beneficial effect on total mortality or whether detrimental effects on non-arrhythmic death offset the reductions in arrhythmic death.

One practical way to address this issue is to carry out a systematic meta-analysis of all the relevant randomised trials to obtain a more precise estimate of the effect of amiodarone on arrhythmic/sudden death, total mortality, and non-arrhythmic death. All current trials of amiodarone in survivors of MI or in CHF patients have been completed, and no major amiodarone trials are known to be in progress. An overview of all available trials would also provide the most accurate assessment of the beneficial effect of amiodarone in subgroups of patients, as well as its side-effects, some of which occur infrequently. With these aims, the principal investigators of the amiodarone randomised trials have collaborated to carry out such a meta-analysis. To improve the precision of the primary analysis and to allow issues of the relation between baseline characteristics and the effect of amiodarone to be addressed, data from individual patients in the studies were merged into a master database upon which subsequent analyses were based.

Section snippets

Methods

The criteria for studies to be included were that treatment allocation was randomised and that amiodarone was compared with placebo or with usual care. Potentially eligible studies were identified by literature review, computerised literature search, and discussion with colleagues. The principal investigators of eligible studies were invited to an organisational meeting. A protocol specifying the baseline, follow-up, and outcome data to be collected, and the methods of analysis, was developed

Study characteristics

Eight of the 13 trials (table 1) had studied patients who had recently had MI (5101 participants, 78%) and five had studied patients with CHF (1452 participants, 22%). Of the post-MI trials, six required patients to be enrolled within 30 days of MI, and the other two specified periods of 45 days or less and less than 60 days. Two studies had no selection criteria for high-risk patients among the total MI population, three used an abnormal Holter electrocardiogram, two abnormal left-ventricular

Discussion

This analysis shows that amiodarone reduces the likelihood of the outcome of arrhythmic/sudden death in high-risk patients with recent MI or CHF. On the other hand, amiodarone has little or no effect on the combined outcome of all non-arrhythmic/non-sudden deaths. These two effects combine in the outcome of total mortality to yield a relative risk reduction of 13%, which is conventionally statistically significant (p=0·030) with classic fixed-effects meta-analysis, and of borderline

References (27)

Cited by (0)

Writing committee and investigators listed at end of paper

View full text