Author + information
- Received December 20, 2002
- Revision received February 14, 2003
- Accepted March 20, 2003
- Published online August 20, 2003.
- Heikki V Huikuri, MD, FACC*,* (, )
- Jari M Tapanainen, MD*,
- Kai Lindgren, MD*,
- Pekka Raatikainen, MD*,
- Timo H Mäkikallio, MD*,
- K.E Juhani Airaksinen, MD* and
- Robert J Myerburg, MD, FACC†
- ↵*Reprint requests and correspondence:
Dr. Heikki V. Huikuri, Professor of Medicine. Division of Cardiology, Department of Internal Medicine, University of Oulu, P.O. Box 5000, FIN-90014, Oulu, Finland.
Objectives This study assessed the predictive power of arrhythmia risk markers after an acute myocardial infarction (AMI).
Background Several risk variables have been suggested to predict the occurrence of sudden cardiac death (SCD), but the utility of these variables has not been well established among patients using medical therapy according to contemporary guidelines.
Methods A consecutive series of 700 patients with AMI was studied. The end points were total mortality, SCD, and nonsudden cardiac death (non-SCD). Nonsustained ventricular tachycardia (nsVT), ejection fraction (EF), heart rate variability, baroreflex sensitivity, signal-averaged electrocardiogram (SAECG), QT dispersion, and QRS duration were analyzed (n = 675). Beta-blocking therapy was used by 97% of the patients at discharge and by 95% at one and two years after AMI.
Results During a mean (±SD) follow-up of 43 ± 15 months, 37 non-SCDs (5.5%) and 22 SCDs (3.2%) occurred. All arrhythmia risk variables differed between the survivors and those with non-SCD (e.g., the standard deviation of N-N intervals was 98 ± 32 vs. 74 ± 21 ms [p < 0.001] and the QRS duration was 103 ± 22 vs.89 ± 16 ms [p < 0.001]). Sudden cardiac death was weakly predicted only by reduced EF (<0.40; p < 0.05), nsVT (p < 0.05), and abnormal SAECG (p < 0.05), but not by autonomic markers or standard ECG variables. The positive predictive accuracy of EF, nsVT, and abnormal SAECG as predictors of SCD was relatively low (8%, 12%, and 13%, respectively).
Conclusions The common arrhythmia risk variables, particularly the autonomic and standard ECG markers, have limited predictive power in identifying patients at risk of SCD after AMI in the beta-blocking era.
Data from various sources have shown that sudden cardiac death (SCD) accounts for ∼50% of all subsequent cardiac deaths among patients who have survived an acute myocardial infarction (AMI), and that there is a specific time dependence of SCD after AMI (1–6). Numerous studies have also suggested that certain arrhythmia risk variables can specifically predict the occurrence of SCD after AMI (7–15). Some of these risk markers have been suggested to be useful for widespread screening of patients for the candidacy of anti-arrhythmic interventions, such as placement of an implantable cardioverter-defibrillator (ICD) (7,15–18).
Most of the previous studies assessing the proportion and time dependence of SCD risk, as well as estimating the predictive power of various risk variables after AMI, have been performed in the absence of optimal compliance to beta-blocking therapy (7–17), which may have a major influence on the prediction and epidemiologic pattern of SCD (19). This single-center, prospective study was designed to assess the power of the risk predictors of SCD and arrhythmia events in a consecutive series of patients surviving an AMI and with an optimal compliance to beta-blocker therapy.
A single-center, prospective study—the Multiple Risk Factor Analysis Trial (MRFAT)—was started in 1996 in the Division of Cardiology, University of Oulu, Finland. The aim was to estimate the predictive power of several noninvasive arrhythmia and clinical risk markers for SCD and arrhythmia events among patients who had survived an AMI. A consecutive series of patients with AMI was included in the study. The patients were recruited to participate during the first seven days after the diagnosis of AMI, which was confirmed by using the contemporary guidelines at the beginning of the study (20). The exclusion criteria were unstable angina at recruitment, dementia, alcoholism, drug abuse, or any other condition that could impair the capacity for informed consent. The qualifying diagnosis of the patients have previously been described in detail in a smaller pilot study (20). A total of 700 consecutive patients (182 females and 518 males; mean age 62 ± 10 years) with AMI fulfilled the inclusion criteria.
Beta-blocking treatment was prescribed to all patients at the time of discharge from the hospital. The dose of beta-blocker therapy was adjusted to achieve a resting heart rate between 50 and 60 beats/min, and a special emphasis was paid to long-term compliance with the beta-blocking medication. In the total study population, beta-blocking drugs were used by 654 (97%) of 675 patients at discharge, 620 (95%) of 652 patients at one year, and 577 (95%) of 609 patients at two years after AMI. Other cardiac medications were prescribed by the primary physicians of the patients, and revascularization therapy was used according to contemporary guidelines (21). All patients were required to give written, informed consent, and the study was approved by the Ethical Committee of the institution.
Risk factor analyses
All risk factor analytical techniques have been previously described in detail (14,20). Left ventricular systolic function was measured with two-dimensional echocardiography between three and seven days after AMI. Ejection fraction (EF) <0.40 was a predefined cut-off point in risk stratification (20).
Baroreflex sensitivity (BRS) was measured between days 5 and 14 with a phenylephrine method (20). A value <3.0 ms/mm Hg was defined as abnormal (7,20). A 24-h electrocardiographic (ECG) recording was obtained with an Oxford Medilog system (Oxford Medilog 4500, Oxford Medical Ltd., London, England) between days 5 and 14 after AMI. The number of episodes of nonsustained ventricular tachycardia (nsVT) and the number of ventricular premature beats were counted. The standard deviation of all N-N intervals measured from the 24-h recording was chosen as an index of heart rate variability (HRV), and a value <70 ms was defined as abnormal (20,22).
QT intervals and QRS durations were measured between days 5 and 14 from two cycles on a standard 12-lead surface ECG recorded at a speed of 50 mm/ms (14). A QT dispersion value ≥90 ms and a QRS duration ≥120 ms were considered abnormal (20). The signal-averaged ECG (SAECG) was recorded between days 5 and 14 using the LP Plus system (Fidelity Medical Ltd., Haifa, Israel) (20). The presence of late potentials was defined as described previously (20,23).
Follow-up and end points
The patients or their families were contacted via telephone at 6, 24, 36, 48, and 60 months after AMI, and the patients had a clinical visit at 12 months after AMI. In cases of death, the reasons for death were verified from the hospital and autopsy records and from either the primary physician or those who had witnessed the death. Two independent end point committees—one at the University of Oulu and the other at the University of Miami—defined the mode of deaths. Cardiac deaths were defined as sudden or nonsudden. Cardiac death was defined as sudden if it was: 1) a witnessed death occurring within 60 min from the onset of new symptoms, unless a cause other than cardiac was obvious; 2) unwitnessed death (<24 h) in the absence of preexisting, progressive circulatory failure or other causes of death; or 3) death during attempted resuscitation (19). In addition to the clinical definition of SCD, probable ventricular tachyarrhythmia events were separately defined as events with a high probability of prevention by the ICD. These events were defined as: 1) successful resuscitation from sustained ventricular or fibrillation; or 2) probable arrhythmic death. Probable arrhythmic death was defined as SCD with verified ECG recordings of ventricular tachycardia or fibrillation at the time of resuscitation, or a witnessed instantaneous death without evidence of a nonarrhythmic cause of death at autopsy.
The data were analyzed using the SPSS software (version 9.0, SPSS Inc., Chicago, Illinois). Comparisons of the baseline characteristics between the groups were performed with analysis of variance with Bonferroni post-hoc analysis and with the chi-square test for categorical variables. Analysis of variance was performed by including survivors in the non-SCD and SCD groups and survivors in the non-SCD and arrhythmia event groups separately. Hazard ratios and 95% confidence intervals were calculated for each categorical variable by using predefined cut-off values in the Cox regression model. To estimate the independent predictive power of the variables, each arrhythmia risk variable was included in the Cox proportional hazards regression analyses after stratification with the clinical factors. Kaplan-Meier estimates of the distribution of times from baseline to total mortality, cardiac death, SCD, and non-SCD were computed. A p value <0.05 was considered statistically significant.
Mortality and time distribution of events during follow-up
Twenty-five of the enrolled patients died during their hospital stay after AMI, and 675 who were discharged alive were included in the long-term follow-up sample. During a mean follow-up of 43 ± 15 months (range 30 to 60), total mortality was 15.0%; among these, 59 (8.7%) were cardiac deaths and 42 (6.2%) were noncardiac deaths. Among the cardiac deaths, 37 (5.5%) were nonsudden and 22 (3.3%; 37% of cardiac deaths) were sudden. Arrhythmia events occurred in 17 patients (2.5%).
Figure 1demonstrates the cumulative event rates for total mortality, cardiac deaths, non-SCDs, and SCDs during follow-up. The temporal distribution of events did not differ from that expected from previous studies for all-cause mortality. In addition, the majority of non-SCDs (68%) occurred during the first 18 months after AMI. However, the incidence and temporal distribution of SCDs (14%) was significantly lower during the early time period (p < 0.001). Similarly, only 3 of the 17 arrhythmia events occurred during the first 18 months after AMI.
Differences in baseline clinical data
The clinical data of survivors and those who experienced non-SCD or SCD or arrhythmia events are shown in Table 1. There were several differences in the baseline clinical characteristics between those who were alive and those who had died due to a cardiac cause, either sudden or nonsudden. For example, a high proportion of the patients who experienced either SCD or non-SCD had diabetes. The only significant difference between the patients who died nonsuddenly or suddenly due to a cardiac cause was in the New York Heart Association (NYHA) functional class—class I being more common among those who died suddenly (55%) than among those who had a non-SCD (24%; p < 0.05). Patients who experienced a probable tachyarrhythmia event were also younger and more commonly in NYHA class I (76%; p < 0.001 vs. non-SCD).
Arrhythmia risk variables
The mean values for arrhythmia risk variables are shown in Table 2and Figure 2. All arrhythmia risk variables differed between the survivors and those who had a non-SCD. The patients who died suddenly did not differ from those who remained alive in terms of any of these variables, except that nsVT was more commonly observed on Holter recordings and the EF was lower among those who experienced SCD (p = 0.048) or an arrhythmia event (p = 0.023).
The hazard ratios of each categorized arrhythmia risk variable with predefined cut-off values are shown in Table 3. On univariate analysis, SCD and arrhythmia events were predicted by reduced EF, nsVT, and abnormal SAECG. After including clinical variables as co-variates, such as age, functional class, diabetes, revascularization, and use of beta-blocking medication, the same variables still predicted SCD. The numbers of patients with true-positive, false-negative, true-negative, and false-positive test results are shown in Table 4. The positive predictive accuracy values of low EF, nsVT, and abnormal SACEG in predicting SCD (8%, 12%, and 13%, respectively) were lower than those of non-SCD (15%, 15%, and 18%, respectively).
Two main differences were observed in this study among the post-AMI patients with a good adherence to beta-blocking therapy, compared with earlier studies. First, the common arrhythmia risk markers provided only limited predictive power on the risk of future SCD or arrhythmia events. Many risk variables that were considered to provide information on the risk of SCD appeared to better predict the occurrence of non-SCD in this population, possibly because of a specific benefit of beta-blocking therapy on SCD, leaving non-SCD risk relatively unopposed early after AMI. Secondly, the epidemiologic pattern of SCD was different from that reported in previous studies (2–15). Arrhythmia events or SCDs did not concentrate early after the index event, but most of them occurred more than 18 months after AMI.
Prediction of SCD after AMI
Although several clinical and arrhythmia risk markers were associated with an increased risk of cardiac mortality in this study, they did not identify risk for a specific type of cardiac death, either sudden or nonsudden. Similarities in the risk profile between the patients who died suddenly or nonsuddenly imply that there are common factors predisposing to both SCD and non-SCD, with severe coronary artery disease and left ventricular dysfunction probably being the most important.
The only arrhythmia risk variables that were able to separate the patients who subsequently experienced SCD or an arrhythmia event from survivors were reduced EF, nsVT, and abnormal SAECG. However, the positive predictive accuracy of even these variables was lower in the prediction of SCD versus non-SCD. Patients who experienced an arrhythmia event also had less severe functional impairment, assessed by NYHA class, compared with those who had a non-SCD, despite similar impairment of left ventricular systolic function. This is in agreement with a recent trial showing that a higher proportion of patients with less functional impairment will die suddenly, whereas the majority of patients with advanced functional impairment will experience a non-SCD (19).
In contrast to the present observations, previous studies have suggested that the markers of autonomic nervous function (7,12)and measurements from the standard 12-lead ECG (4,13–15)specifically predict the risk of SCD or arrhythmia events, alone or in combination with other variables (7,12). The only obvious difference between the previous studies and the present one is in the usage of medication. For example, in the Autonomic Tone and Reflexes After Myocardial Infarction (ATRAMI) trial, showing that autonomic markers are powerful predictors of SCD, only 20% of post-AMI patients were taking beta-blocking medication (7). Beta-blocking drugs have significant influences on both HRV and BRS (24,25)and also on the incidence of SCD (19), perhaps explaining the lack of predictive power of autonomic markers among patients who are treated with continuous beta-blocking medication.
The proportion of patients receiving beta-blocking medication has varied between 10% and 70% in previous observational studies and randomized trials (4,7–18). Data from several studies have convincingly shown that beta-blocking medication specifically reduces the incidence of SCDs in high-risk post-AMI populations (19,26–29). An obvious explanation for the lack of predictive power of many arrhythmia risk variables was the altered temporal distribution of SCD after AMI, because beta-blockers seem to provide an increased benefit for the early occurrence of SCD. Similarly, a recent randomized trial including patients with a depressed EF and remote AMI (18)showed that in the presence of effective therapy of heart failure, including beta-blockers, the survival benefit of ICD therapy began relatively late as compared with previous trials without optimized medical therapy (16,17).
A relatively small number of sudden deaths (n = 22) may prevent the generalization of the results to other post-AMI populations. However, this is one of the largest prospective, observational studies assessing the prediction of sudden death in a consecutive post-AMI population, including several predefined arrhythmia risk markers in the study design. In the largest prospective, observational study assessing the predictive power autonomic markers, 30 patients reached the combined end point of sudden death and nonfatal ventricular tachycardia (7). In most of the previous observational studies, SCD and nonfatal ventricular tachyarrhythmia have been used as a combined end point (7,9,10,12,13,30). This end point is partly biased, because many other conditions than arrhythmias that evolve rapidly can also lead to sudden death. In fact, recent studies on patients with an ICD indicate that many of the deaths defined as sudden were not due to cardiac arrhythmia (31,32). Therefore, we separately analyzed the cases of SCD, defined by clinical criteria, and arrhythmia events, defined by documentation of life-threatening arrhythmia and/or autopsy findings. Despite the relatively low number of SCDs and arrhythmia events in the current study, the arrhythmia risk markers would have been expected to give a better predictive accuracy if they are considered to be useful for clinical purposes (e.g., for evaluation of the candidacy of an ICD).
The observational nature of the study design may not allow definite conclusions about the clinical utility of the arrhythmia risk variables. Therefore, the predictive power of the arrhythmia risk markers should ideally be confirmed in larger randomized trials, which should pay special attention to adherence to evidence-based medical therapy, including beta-blockers.
Common arrhythmia risk markers, particularly the autonomic and standard ECG markers, seem to lose some of their predictive power among patients who are receiving beta-blocking therapy after AMI. Despite the beta-blocking medication, still >3% of post-AMI patients experienced SCD. These patients tended to have a better functional capacity, despite left ventricular dysfunction, as compared with those who had a non-SCD. The prediction and prevention of fatal arrhythmia events in these patients still remain a challenge for future attempts to reduce the unexpected deaths among patients surviving an AMI.
The authors wish to express their sincere acknowledgements to Pirkko Huikuri, RN, Päivi Karjalainen, RN, and Ms. Anne Lehtinen.
☆ This study was supported by the Medical Council of the Academy of Science, Helsinki, Finland, and the Foundation for Cardiovascular Research, Helsinki, Finland. Dr. Myerburg is supported, in part, by the American Heart Association Chair in Cardiovascular Research at the University of Miami and by the Louis Lemberg Chair in Cardiology.
- acute myocardial infarction
- baroreflex sensitivity
- ejection fraction
- heat rate variability
- implantable cardioverter-defibrillator
- nonsustained ventricular tachycardia
- New York Heart Association
- signal-averaged electrocardiogram
- sudden cardiac death
- Received December 20, 2002.
- Revision received February 14, 2003.
- Accepted March 20, 2003.
- American College of Cardiology Foundation
- Myerburg R.J.,
- Interian A. Jr..,
- Mitrani R.M.,
- Kessler K.M.,
- Castellanos A.
- Zipes D.P.,
- Wellens H.J.J.
- La Rovere M.T.,
- Pinna G.D.,
- Hohnloser S.H.,
- et al.
- Maggioni A.P.,
- Zuanetti G.,
- Franzosi M.G.,
- et al.
- McClements B.M.,
- Adgey A.A.J.
- Cardiac Arrhythmia Suppression Trial/Signal-Averaged Electrocardiogram (CAST/SAECG) Substudy Investigators,
- el-Sherif N.,
- Denes P.,
- Katz R.,
- et al.
- Hartikainen J.E.K.,
- Malik M.,
- Staunton A.,
- Poloniecki J.,
- Camm A.J.
- Perkiömäki J.S.,
- Koistinen M.J.,
- Yli-Mäyry S.,
- Huikuri H.V.
- Bailey J.J.,
- Berson A.S.,
- Handelsman H.,
- Hodges M.
- Ryan T.J.,
- Anderson J.L.,
- Braniff B.A.,
- et al.
- Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology
- Breithardt G.,
- Cain M.E.,
- El-Sherif N.,
- et al.
- Niemelä M.J.,
- Airaksinen K.E.J.,
- Huikuri H.V.
- Airaksinen K.E.J.,
- Niemelä M.J.,
- Huikuri H.V.
- Hohnloser S.H.,
- Klingenheben T.,
- Zabel M.,
- Schöpperl M.,
- Mauss O.
- Pratt C.M.,
- Greenway P.S.,
- Schoenfeld M.H.,
- et al.
- Grubman E.M.,
- Pavri B.B.,
- Shipman T.,
- et al.