Author + information
- Received January 13, 1997
- Revision received December 3, 1997
- Accepted March 5, 1998
- Published online June 1, 1998.
- Roberto F.E Pedretti, MD, FESCa,* (, )
- Giovanni B Migliori, MDb,
- Vittorio Mapelli, PhD∗,
- Gabriele Daniele, BSc,
- Philip J Podrid, MD, FACC† and
- Roberto Tramarin, MD, FESCa
- ↵*Address for correspondence: Dr. Roberto F. E. Pedretti, Fondazione Salvatore Maugeri, Care and Research Institute, Division of Cardiology, Rehabilitation Institute, Via Roncaccio, 16, 21049, Tradate (VA), Italy
Objectives. We sought to evaluate 1) the cost-effectiveness of amiodarone therapy in postinfarction patients; and 2) the influence of alternative diagnostic strategies (noninvasive only vs. noninvasive and electrophysiologic testing) on survival benefit and cost-effectiveness ratio of amiodarone therapy.
Background. The cost-effectiveness of amiodarone therapy in postinfarction patients is still unknown, and no study has determined which diagnostic strategy should be used to maximize amiodarone survival benefit while improving its cost-effectiveness ratio.
Methods. We designed a postinfarction scenario wherein heart rate variability analysis on 24-h Holter monitoring was used as a screening test for 2-year amiodarone therapy in a cohort of survivors (mean age 57 years) of a recent myocardial infarction. Three different therapeutic strategies were compared: 1) no amiodarone; 2) amiodarone in patients with depressed heart rate variability; 3) amiodarone in patients with depressed heart rate variability and a positive programmed ventricular stimulation. Total variable costs and quality-adjusted life expectancy during a 20-year period were predicted with use of a Markov simulation model. Costs and charges were calculated with reference to an Italian and American hospital.
Results. Amiodarone therapy in patients with depressed heart rate variability and a positive programmed ventricular stimulation was dominated by a blend of the two alternatives. Compared with the no-treatment strategy, the incremental cost-effectiveness ratio of amiodarone therapy in patients with depressed heart rate variability was $10,633 and $39,422 per gained quality-adjusted life-year using Italian costs and American charges, respectively.
Conclusions. Compared with a noninterventional option, amiodarone prescription in all patients with depressed heart rate variability seems to be a more appropriate approach than the alternative based on the combined use of heart rate variability and electrophysiologic study.
Prevention of malignant ventricular tachyarrhythmias remains a major problem after acute myocardial infarction. At present, only beta-adrenergic blocking agents have been shown (1)to be effective for diminishing the incidence of sudden death after myocardial infarction. Therefore, alternative and complementary therapeutic approaches are under investigation. Initial trials with empiric low dose amiodarone treatment have been promising (2–5). However, amiodarone therapy is an expensive treatment; the incidence of serious side effects may be relatively high; and a properly constructed program for long-term follow-up, including methods for monitoring drug adverse reactions on a regular basis, is necessary (6). Moreover, results from the Canadian Amiodarone Myocardial Infarction Arrhythmia (CAMIAT) (7)and European Myocardial Infarct Amiodarone (EMIAT) trials (8)did not show that the drug reduced total or cardiac mortality, although arrhythmic mortality was reduced. These findings do not support a systematic prophylactic use of amiodarone in postinfarction patients but suggest a role for the drug in patients at high risk of arrhythmias for whom amiodarone will offer a substantial benefit. Nevertheless, the cost-effectiveness of low dose amiodarone therapy in postinfarction patients is still unknown. Moreover, no study has pointed out which diagnostic strategy should be used to maximize the amiodarone survival benefit and improve the cost-effectiveness ratio. With regard to risk assessment, depressed heart rate variability was found to be associated with cardiac mortality, particularly sudden death (9). However, our group (10)and others (11)have reported that programmed ventricular stimulation in patients preselected by noninvasive techniques has an additional benefit for improving diagnostic accuracy. Thus, the combined use of programmed ventricular stimulation and heart rate variability analysis could be more cost-effective than a simple noninvasive approach. Because of its higher specificity, the combined approach could reduce false positive results and decrease the number of patients needing amiodarone treatment during follow-up. Therefore, the objectives of the present study were to evaluate: 1) the cost-effectiveness of amiodarone therapy in postinfarction patients at high risk of arrhythmic death; and 2) whether the use of alternative diagnostic strategies (heart rate variability assessment vs. heart rate variability and electrophysiologic testing) may influence amiodarone survival benefit and cost-effectiveness.
We constructed a Markov simulation model (12)to compare three alternative diagnostic and therapeutic strategies in three groups of postinfarction patients. A Markov model assumes that a hypothetical cohort of patients moves through a sequence of health states. Changes from one state to another are determined by a set of transition probabilities that are either time dependent or constant. We used DATA by TreeAge software (13)to simulate the prognosis of each cohort and “monitor” their progress at 1-year intervals for a 20-year period tracking relevant events, total survival time and costs. To strengthen the results, both Italian and American costs were introduced into the analysis.
We designed a scenario in which heart rate variability analysis on 24-h Holter monitoring was used as a screening test for amiodarone therapy after myocardial infarction. Our base-case analysis evaluated survivors of a recent myocardial infarction (men in 79% of cases, mean age 57 years) free from contraindications to amiodarone, of whom 54% and 40% were treated with thrombolysis and beta-blockers, respectively (9). As shown in Figure 1, three groups of patients were considered: 1) In the first group, no amiodarone therapy was prescribed. 2) In the second group, all patients with depressed heart rate variability were treated with amiodarone during the first 2 years after myocardial infarction, after receipt of an oral loading dose during an additional week in the hospital and undergoing baseline tests to screen for potential drug toxicity during follow-up. 3) In the third group, all patients with depressed heart rate variability underwent programmed ventricular stimulation, and only those with positive programmed stimulation, defined as the induction of sustained monomorphic ventricular tachycardia of <270 beats/min, received amiodarone therapy.
Estimation of mortality
After hospital discharge, they were entered into the Markov process, where for each 1-year period they were exposed to three time-dependent forces of mortality: 1) arrhythmic cardiac mortality; 2) nonarrhythmic cardiac mortality; 3) noncardiac mortality.
We estimated mortality rates during the first year and constant mortality rates for each subsequent year up to year 10 and for years 11 to 20 (14). First-year cardiac mortality rates and proportion of arrhythmic deaths were estimated from published reports (9–11,15,16)and are shown in Table 1. To obtain subsequent annual cardiac mortality rates for years 2 to 10, these estimates were patterned on data from other studies (14,17–19)in which patients had a first-year mortality rate similar to ours. The proportion of arrhythmic death was assumed to be constant for years 1 to 10. The yearly noncardiac mortality rate was 0.012 and maintained constant for years 1 to 10 (19).
Patients at very high risk who showed a first-year cardiac mortality rate >0.11 were assumed to have a 0.075 annual cardiac mortality rate for years 2 to 10. Patients with a first-year cardiac mortality rate of 0.10 to 0.11 were defined as being at high risk and were assumed to have a 0.065 subsequent annual cardiac mortality rate. Patients with a first-year cardiac mortality rate of 0.07 to 0.08 were defined as being at intermediate risk and were assumed to have a 0.05 annual cardiac mortality rate in the following years. Patients with negative results on noninvasive testing were defined as being at low risk, with a first-year cardiac mortality rate of 0.015, and were assumed to have a 0.015 annual cardiac mortality rate for years 2 to 10. As shown in Table 1, by the sensitivity analysis, the first-year cardiac mortality rate in the low-risk group ranged from 0.016 to 0.032, on the basis of the scenario analyzed. The annual cardiac mortality rate for subsequent years was adjusted to maintain an unchanged life expectancy for the no-amiodarone strategy and ranged from 0.016 to 0.024.
Ten years after myocardial infarction, the annual all-cause mortality rate of the survivors was assumed to be 0.037, identical to that of a cohort of the same age and gender according to national life-tables (20).
According to data on treatment analysis of the CAMIAT (7)and EMIAT (8)trials, we assumed that amiodarone reduces arrhythmic cardiac mortality by 39%, increases nonarrhythmic cardiac and noncardiac mortality by 3% and 18%, respectively, in the first 2 years after myocardial infarction. Patients receiving amiodarone therapy were seen routinely according to structured follow-up protocols summarized in Table 2; the monitoring schedule recommended by Wilson and Podrid (6)was adopted in our model. Amiodarone use might be discontinued because of poor compliance or occurrence of nonfatal “drug-related” complications. The difference in the adverse event rate between amiodarone and placebo was used in the model according to data from the CAMIAT (7)and EMIAT (8)trials.
Medical rates and probabilities used in the model were based on published reports, as shown in Table 1 (7,8,10,11,14–22).
Measurement of costs
As shown in Table 3, assumptions used in the model were derived from costs calculated at the Tradate Rehabilitation Institute (Italy) and from actual charges at Boston University Medical Center Hospital (Massachusetts). The Italian center is a small hospital size (130 beds), whereas the American center is a teaching and high cost hospital. Costs and charges are shown in Table 3, expressed in 1994 U.S. dollars (1 U.S. dollar = 1,586 Italian lire). Indirect costs, such as patient time or production losses, are not included.
A detailed micro-costing methodology was used to calculate most costs at the Tradate Rehabilitation Institute. The costs of diagnostic tests include staff time, supplies, equipment and overhead. Treatment costs include costs of therapy (including drug and testing during follow-up or monitoring and surveillance of patients) and costs of treating side effects related to long-term amiodarone toxicity. Italian costs are lower than American costs because of lower prices, namely hourly wages, and because they are production costs rather than charges to third-party payers. Costs and charges of noninvasive tests were not included in the model because they are common to all alternatives in each scenario analyzed.
Quality of life adjustments
The best and worst quality of life values were assigned to the “well” and “dead” states, respectively. Patients remaining in the well state for 1 year were credited with 1 quality-adjusted life-year (QALY). Patients who died no longer accrued QALYs. Short-term morbidity was reflected by subtracting an amount of time from total life expectancy. According to previous published estimates (23,24), 2 and 3 weeks are deducted for execution of electrophysiologic testing and drug side effects, respectively. Life expectancy unadjusted for quality of life was also reported (25).
Future costs and benefits were discounted at 3% (reference case) and 5%/year (for comparison with past analyses). Undiscounted results, which are often of interest, were also reported (25).
Calculation of cost-effectiveness of alternative diagnostic strategies
Quality-adjusted life expectancies of patients under each regimen were combined with costs and charges to give the total costs and effectiveness and incremental cost-effectiveness ratios.
Sensitivity analyses were used to assess whether variations in our estimates or assumptions significantly altered the results.
The results of base-case analysis are reported in Table 4. Using a discount rate of 3%, amiodarone therapy, when given to all patients with depressed heart rate variability, was the most expensive of the regimens but resulted in the greatest quality-adjusted life expectancy. The strategy based on amiodarone therapy in patients with low heart rate variability and positive programmed ventricular stimulation resulted in lower costs. However, compared with the no-therapy option, the gain in life expectancy was small. Thus, the strategy based on amiodarone therapy after combined use of heart rate variability and electrophysiologic testing was dominated by a blend of the two alternatives. Compared with the no-treatment strategy, the incremental cost-effectiveness ratio of amiodarone therapy according to heart rate variability assessment was $10,633 and $39,422/QALY gained using Italian costs and American charges, respectively.
Data were reanalyzed using a discount rate of 5% for comparison with past analyses, and undiscounted results were also calculated. Discount rate did not affect the stability of the results. Results unadjusted for quality of life were also reported: Data showed that the results were insensitive to the quality of life adjustment according to both Italian costs and American charges.
Results of the sensitivity analysis are reported in Figure 2. The output of the model for different combinations of noninvasive testing was tested. The strategy based on amiodarone after the combined use of noninvasive testing and electrophysiologic study was dominated by a blend of the two alternatives in five of the eight analyzed scenarios for both Italian costs and American charges. Compared with the no-therapy option, when amiodarone was given only to patients with abnormal results on noninvasive testing, the incremental cost-effectiveness ratios ranged from $16,275 to $10,167 and from $60,350 to $37,767/QALY gained for Italian costs and American charges, respectively. However, in three scenarios, all based on the use of a single noninvasive test, no strategy was clearly dominated by any other, and no strategy was eliminated by extended dominance. Compared with the no-treatment option, the strategy based on amiodarone given to patients with abnormalities on both noninvasive testing and electrophysiologic study had an incremental cost-effectiveness ratio that ranged from $12,500 to $6,967 and from $53,650 to $29,900/QALY gained for Italian costs and American charges, respectively. The strategy based on amiodarone in all patients with positive results on noninvasive testing further increased quality-adjusted life expectancy, with an incremental cost-effectiveness ratio that ranged from $35,050 to $17,400 and from $115,000 to $57,133 for Italian costs and American charges, respectively.
In the base-case analysis, mortality estimates for patients evaluated with electrophysiologic study were based on a sensitivity of programmed stimulation for cardiac arrhythmic and nonarrhythmic death of 40% and 14%, respectively, and a specificity of 77% (10,11,15,16). As shown in Figure 2, the model was sensitive to an improvement in the diagnostic accuracy of electrophysiologic testing, assuming a sensitivity of programmed stimulation for cardiac arrhythmic and nonarrhythmic death of 78% and 20%, respectively, and a specificity of 80% (10,11,16). Compared with the no-treatment option, amiodarone after electrophysiologic study increased quality-adjusted life expectancy, with an incremental cost-effectiveness ratio of $4,900 and $21,150/QALY gained for Italian costs and American charges, respectively. Amiodarone, when given to all patients with depressed heart rate variability, further increased quality-adjusted life expectancy, with an incremental cost-effectiveness ratio of $56,500 and $185,600/QALY gained for Italian costs and American charges, respectively.
Finally, we analyzed the sensitivity of the model to changes in amiodarone effectiveness and follow-up design. As shown in Figure 2, the results of the baseline analysis were confirmed in three of the four tested alternative scenarios for both Italian costs and American charges. Conversely, no strategy was clearly dominated by any other, and no strategy was eliminated by extended dominance when amiodarone effectiveness observed in the EMIAT trial (8)was assumed in the model. The incremental cost-effectiveness ratio of electrophysiologic study-guided therapy was $19,950 and $85,800 for Italian costs and American charges, respectively. Amiodarone, if given after heart rate variability assessment only, further increased quality-adjusted life expectancy, with an incremental cost-effectiveness ratio of $27,600 and $90,600 using Italian costs and American charges, respectively.
On the basis of costs and charges from two centers, one in Italy and the other in the United States, the present investigation provides an estimate of cost-effectiveness of amiodarone therapy in postinfarction patients. The influence of alternative diagnostic strategies (noninvasive only, based on heart rate variability assessment, vs. the combined use of noninvasive and electrophysiologic testing) on survival benefit and the cost-effectiveness ratio of amiodarone therapy was also evaluated. The combined use of electrophysiologic study and heart rate variability assessment may significantly reduce total costs of therapy, but the gain in quality-adjusted life expectancy due to amiodarone is small. Therefore, compared with the no-therapy option, amiodarone in patients at risk after heart rate variability analysis seems to be a more cost-effective approach. Moreover, the cost-effectiveness ratio of amiodarone therapy in postinfarction patients with depressed heart rate variability seems to be consistent with that of most currently accepted programs. These findings should be taken into account in this era of limited health care resources.
Amiodarone after acute myocardial infarction: which diagnostic tests should be used to select patients suitable for long-term treatment?
Results from the CAMIAT (7)and EMIAT trials (8)showed that amiodarone does not reduce total or cardiac mortality in postinfarction patients with frequent premature ventricular complexes or left ventricular dysfunction. However, arrhythmic mortality was significantly reduced in both trials, suggesting a potential role for the drug in patients at high risk for life-threatening ventricular arrhythmias for whom amiodarone could offer a substantial survival benefit (7,8).
It is currently extremely difficult to identify with a high predictive accuracy the postinfarction patients who are at high risk for an arrhythmic event and who might benefit from a prophylactic treatment such as amiodarone. Among noninvasive risk markers, depressed heart rate variability was an independent predictor of arrhythmic death, both in patients with either a low or a normal ejection fraction (9). Inducibility of sustained monomorphic ventricular tachycardia at electrophysiologic study has proved to be the single best predictor of spontaneous ventricular tachycardia and sudden death after myocardial infarction (26). The key advantage of stratifying patients after infarction by programmed stimulation versus less invasive methods is that it may identify a small subgroup at sufficient high risk of electrical events to justify prophylactic antiarrhythmic intervention. Moreover, as previously described in a group of patients preselected by the presence of noninvasive risk factors (10,11), programmed ventricular stimulation is useful in improving diagnostic accuracy. As the first step, noninvasive tests can be effectively used to limit the number of postinfarction patients undergoing an electrophysiologic test. As a second step, electrophysiologic study minimizes false positive results because of its high negative predictive value. This feature is of great relevance when the therapy given to patients positive results is associated with both potential toxicity and high costs, such is the case with amiodarone. Patients without inducible arrhythmias, who have a good prognosis despite the presence of different noninvasive markers, will be spared the risks of long-term treatment with antiarrhythmic agents (10,11). From an economic point of view, a reduction in the number of inappropriate treatments after hospital discharge can save on health care resource use.
Cost-effectiveness of alternative diagnostic strategies in arrhythmic risk stratification after acute myocardial infarction
Cost-effectiveness analysis allows a comparison of costs of different strategies that achieve the same effect and may facilitate choices between alternatives. Results of base-case analysis showed that amiodarone therapy, when given to all patients at risk after heart rate variability assessment, was the most expensive of the regimens studied. Specifically, amiodarone therapy according to a combined use of heart rate variability analysis and electrophysiologic testing resulted in costs that were 58% and 52% lower, using Italian costs and American charges, respectively. However, compared with the no-therapy option, the gain in quality-adjusted life expectancy was small. Therefore, the strategy based on amiodarone in patients with both low heart rate variability and a positive result on electrophysiologic testing was dominated by a blend of the two alternatives. If amiodarone is given to prevent arrhythmic death after myocardial infarction, selection of patients according to heart rate variability analysis is a sufficiently cost-effective approach. Goldman et al. (27)suggested that an incremental cost-effectiveness ratio <$20,000/additional QALY is very attractive. Incremental cost-effectiveness ratios between $20,000 and ∼$40,000/QALY gained are consistent with other currently founded programs, such as hemodialysis or treatment of mild hypertension with diuretic drugs or propranolol. Incremental cost-effectiveness ratios between $60,000 and $100,000/additional QALY are clearly higher than most currently accepted programs, whereas ratios >$100,000 are generally agreed to be unattractive. With reference to American charges, the present investigation showed that amiodarone therapy in postinfarction patients at high risk for arrhythmic death because of low heart rate variability at 24-h Holter monitoring had an incremental cost-effectiveness ratio of $39,422/QALY gained, which was consistent with that of other commonly accepted health programs. Different discount rates as well as adjustment for quality of life did not significantly change results.
We tested sensitivity to changes of those variables that may be relevant as determinants of cost-effectiveness in the model. With regard to amiodarone effectiveness, we performed two additional analyses according to data reported in the CAMIAT (7)and EMIAT (8)trials. Changes in arrhythmic, nonarrhythmic cardiac and noncardiac death ranged from −33% to −45%, from −27% to +30% and from 0% to +37%, respectively. In the CAMIAT scenario, results were not significantly different from those reported in the base-case analysis. According to the model data from the EMIAT trial, no strategy was clearly dominated by any other, and no strategy was eliminated by extended dominance. However, by American charges, the incremental cost-effectiveness ratios for both alternative therapeutic strategies were higher than most currently accepted programs, ranging from $60,000 to $100,000/additional QALY gained.
We also tested the model using costs based on monitoring in the CAMIAT (7)and EMIAT (8)trials. Validity of the model was still supported by concordance of the results with those observed in the baseline analysis, suggesting that study results were independent of the adopted monitoring program.
Different combinations of noninvasive risk markers (i.e., left ventricular ejection fraction, premature ventricular complexes, unsustained ventricular tachycardia and ventricular late potentials) were introduced into the model. The results did not support the use of an additional electrophysiologic study if the first-level screening was based on abnormalities of two noninvasive tests. Conversely, data suggest the use of programmed stimulation in those patients with only one positive marker among low ejection fraction, late potentials and unsustained ventricular tachycardia. Nevertheless, despite a similar cost-effectiveness ratio, the increase in quality-adjusted life expectancy induced by amiodarone in patients with low heart rate variability was 125% to 200% higher. This finding seems to agree with recent data from a prospective substudy of the EMIAT trial (28)that suggests a potential role for heart rate variability as the best noninvasive test for screening postinfarction patients who may benefit from amiodarone therapy.
Finally, we changed the model with regard to probability of cardiac death in patients with positive and negative results on the electrophysiologic study. After exclusion of data from Bourke et al. (15), new estimates were calculated. When a programmed ventricular stimulation sensitivity for cardiac arrhythmic and nonarrhythmic death of 78% and 20%, respectively, and a specificity of 80% (10,11,16)were assumed, the results were found to be sensitive, and no strategy was clearly dominated by any other. Using American charges, electrophysiologic-guided amiodarone strategy had a very attractive incremental cost-effectiveness ratio of $21,150/QALY gained. Nevertheless, overall examination of sensitivity analysis showed that results were robust, and the model results had only slight changes compared with baseline assumptions.
The present investigation was designed to estimate the cost-effectiveness of amiodarone therapy in postinfarction patients at high risk for arrhythmic death. The influence of alternative diagnostic strategies on amiodarone-induced survival benefit and the cost-effectiveness ratio was also evaluated. The results do not support the use of an electrophysiologic study after heart rate variability analysis for selecting candidates for long-term amiodarone therapy. Moreover, the cost-effectiveness ratio of amiodarone therapy in patients with depressed heart rate variability after myocardial infarction seems to be consistent with that of most currently accepted health care programs. These findings might be important in the allocation of limited health care resources.
☆ This study was supported in part by funds for 1995 current research from the Italian Ministry of Health, Rome, Italy.
- Canadian Amiodarone Myocardial Infarction Arrhythmia Trial
- European Myocardial Infarct Amiodarone Trial
- quality-adjusted life-year
- Received January 13, 1997.
- Revision received December 3, 1997.
- Accepted March 5, 1998.
- by the American College of Cardiology
- Burkart F,
- Pfisterer M,
- Kiowski W,
- Follath F,
- Burckhardt D,
- Jordi H
- Cairns J.A,
- Connolly S.J,
- Gent M,
- Roberts R
- Ceremuzynski L,
- Kleczar E,
- Krzeminska-Pakula M,
- et al.
- 5.Navarro-López F, Cosin J, Marrugat J, Guindo J, Bayes de Luna A, for the SSSD Investigators. Comparison of the effects of amiodarone versus metoprolol on the frequency of ventricular arrhythmias and on mortality after acute myocardial infarction. Am J Cardiol 1993;72:1243–8.
- 7.↵Cairns JA, Connolly SJ, Roberts R, Gent M, for the Canadian Amiodarone Myocardial Infarction Arrhythmia Trial Investigators. Randomised trial of outcome after myocardial infarction in patients with frequent or repetitive ventricular premature depolarisations: CAMIAT. Lancet 1997;349:675–82.
- 8.↵Julian DG, Camm AJ, Frangin G, et al., for the European Myocardial Infarct Amiodarone Trial Investigators. Randomised trial of effect of amiodarone on mortality in patients with left-ventricular dysfunction after recent myocardial infarction: EMIAT. Lancet 1997;349:667–74.
- Hartikainen J.E.K,
- Malik M,
- Staunton A,
- Poloniecki J,
- Camm A.J
- Pedretti R,
- Etro M.D,
- Laporta A,
- Sarzi Braga S,
- Carù B
- Sonnenberg F.A,
- Beck J.R
- 13.↵(1996) DATA, Version 3.0 (TreeAge Software Inc, Williamstone (MA)).
- Bourke J.P,
- Richards D.A,
- Ross D.L,
- Wallace E.M,
- McGuire M.A,
- Uther J.B
- 17.Cucherat M, Boissel JP, Leizorovicz A, for the APSI Investigators. Persistent reduction of mortality for five years after one year of acebutolol treatment initiated during acute myocardial infarction. Am J Cardiol 1997;79:587–9.
- 18.Haim M, Hod H, Reisin L, et al., for the Secondary Prevention Israeli Nifedipine Trial (SPRINT) Study Group. Comparison of short-and long-term prognosis in patients with anterior wall versus inferior or lateral wall non-Q-wave acute myocardial infarction. Am J Cardiol 1997;79:717–21.
- Bacci M
- Herre J.M,
- Sauve M.J,
- Malone P,
- et al.
- Larsen G.G,
- Manolis A.S,
- Sonnenberg F.A,
- Beshansky J.R,
- Estes N.A.M,
- Pauker S.G
- Hogenhuis W,
- Stevens S.K,
- Wang P,
- et al.
- 25.↵Siegel JE, Weinstein MC, Russell LB, Gold MR, for the panel on cost-effectiveness in health and medicine. JAMA 1996;276:1339–41.
- Richards D.A.B,
- Byth K,
- Ross D.L,
- Uther J.B
- Goldman L,
- Gordon D.J,
- Rifkind B.M,
- et al.
- 28.↵Malik M, Camm AJ, Julian DG, et al, on behalf of the EMIAT Investigators. Reduction of post-infarction mortality on amiodarone in patients with reduced heart rate variability: prospective substudy of the EMIAT trial [abstr]. Eur Heart J 1997;18 Abstr Suppl:66.