topeeng.gif (8416 bytes)

[ Scientific Activities - Actividades Científicas ]

Methods of identifying post-myocardial infarction
patients at high risk for subsequent arrhythmic death

Francesco Santoni-Rugiu, MD
Division of Electrophysiology and Electrocardiology
Cardiovascular Institute
Mount Sinai Medical Center

1 Gustave L. Levy Place 10029
New York


Measures of Autonomic Tone – Heart Rate Variability

Sudden Arrhythmic Death Post-MI

Measures of Autonomic Response - Baroreflex Sensitivity

Risk Factors for Sudden Arrhythmic death Post-MI

Other Methods

Left Ventricular Function

Combined Risk Factor Assessment

ECG Monitoring

A Proposed Approach

Signal Averaged ECG

Future trends

Programmed stimulation

Clinical Implications and Conclusion



Sudden cardiac death (SCD) can be defined as instantaneous death or death occurring within an hour of the onset of an abrupt change in clinical status (1). It represents a major public health problem, claiming approximately 400,000 lives per annum in the United States, roughly half of the deaths due to cardiovascular disease (2). Given this scenario, great emphasis has been placed on identifying high-risk groups who would benefit from effective preventive interventions.

As many as 75% of patients who die suddenly have evidence of a prior myocardial infarction (MI)(3). Indeed, patients who survive the acute phase of a myocardial infarction constitute as a whole a high-risk group for sudden death. The risk for sudden death in the post-MI population is particularly high in the first years after the infarction (4-6). As the index MI becomes more remote, the relative contribution to cardiac mortality of non-sudden causes, such as progression of heart failure, increases.

Post-myocardial infarction sudden cardiac death includes both non-arrhythmic and arrhythmic events.

Sudden non-arrhythmic cardiac death encompasses a variety of distinct conditions including acute myocardial failure, usually in the setting of a repeated infarction, free wall or interventricular septal rupture and acute valvular dysfunction. Sudden arrhythmic death (SAD) is caused by either ventricular tachyarrhythmias or severe bradycardic/asystolic episodes; the former mechanism has been found to be more common (6-8), while severe bradycardia or asystole seem to be associated with profound myocardial dysfunction (3).

Despite differences in definition, most follow-up studies of patients post myocardial infarction suggest that approximately 50 to 60% of deaths are due to arrhythmic events (4,5). Besides being of epidemiological interest, the dichotomy between arrhythmic and non-arrhythmic events is of paramount importance. The vast majority of primary arrhythmic death victims could indeed be saved by prompt therapy. The impact of appropriately targeted preventive and therapeutic strategies could be tremendous. The focus of this monograph is arrhythmic death post myocardial infarction and the methods of identifying patients at high risk for this event.

Sudden Arrhythmic Death Post MI


Sudden cardiac death represents a significant proportion of deaths occurring after the acute phase of a myocardial infarction (6,9,10). Although SCD should not be equated to sudden arrhythmic death, most SCD events post-MI are caused by ventricular tachyarrhythmias (7). The incidence of arrhythmic death appears to be highest in the first year post-MI (10,11). It has been estimated that lethal arrhythmias represent approximately half of the causes of death in the first year post-MI (10-12). A major hurdle to the study of this condition is represented by the frequent lack of a clearly established cause of death. An analysis of the Aspirin Myocardial Infarction Study database focused on witnessed deaths, revealed that sudden arrhythmic death was the mechanism of death in 41% of fatal events recorded over a mean follow-up of 38 months, in a group of 4,524 post-MI patients (7). A report from the Multicenter Post Infarction Research Group classified as arrhythmic 55% of 108 deaths observed in a 48 months follow-up period (13). Unexpectedly, in both studies arrhythmias caused a sizable percentage of non-sudden cardiac deaths, occurring more than one hour after the onset of symptoms. It is critical to understand that populations of MI survivors at high risk for arrhythmic and non-arrhythmic death overlap significantly. A major effort is currently oriented towards methods of identifying groups for which arrhythmias are a more likely mode of exitus than non-arrhythmic events.


Clinical studies of post-MI patients and experimental post-infarction models suggest that two distinct mechanisms are responsible for the majority of lethal arrhythmic episodes: acute ischemia and primary arrhythmias based on a "myocardial substrate"(14). Acute ischemia may trigger ventricular tachyarrhythmias, mainly polymorphic ventricular tachycardia (VT) and/or ventricular fibrillation (Vfib), both in patients with and without a history of previous MI. These arrhythmias probably arise as a result of sudden and inhomogeneous electrophysiological changes induced by ischemia. The electrophysiological properties of the ventricular muscle are significantly altered by coronary hypoperfusion via impaired transmembrane electrolyte balance and acid-base disturbances (15). As elucidated in animal models, ischemia and acute infarction create abnormalities in impulse conduction velocity, changes in refractoriness, areas of block and enhanced automaticity (16). These alterations are inhomogeneous in distinct areas of the ventricular muscle and set the scene for ventricular tachyarrhythmias. Since the etiology of these events is primarily ischemic, they should not be defined as sudden arrhythmic death. The events should be correctly viewed as acute coronary syndromes complicated by lethal ventricular arrhythmias. The occurrence of acute coronary syndromes post-MI is predicted by markers of instability, progression and severity of coronary artery disease (17); markers of electrophysiological abnormalities may not be present before the onset of ischemia. One should consider different preventive strategies than those proposed for sudden arrhythmic death (17).

The second mechanism responsible for lethal arrhythmic episodes is represented by reentrant circuits that course within and around the infarct scar (14). Presence of areas of slow conduction and surrounding areas of block are crucial for the initiation and persistence of the reentrant circuit. There frequently are areas of surviving myocytes interspersed among fibrotic tissue within the infarct zone (15). Intervening fibrotic areas and altered intercellular gap junctions have been identified as some of the possible explanations for slow conduction along these strands of myocardial tissue encased within the infarct scar (15). These pathophysiologic changes are currently referred to as "myocardial substrate". There is currently a consensus that such substrate is responsible for the majority of monomorphic VT associated with infarction. This regular monomorphic arrhythmia can degenerate into polymorphic VT or Vfib.

Insight into the mechanisms of these arrhythmic events derives from a variety of sources. Direct evidence of the electrical phenomena prior to and during the event is provided by Holter tracings of patients with a history of infarction, who die of sudden arrhythmic death during the monitoring hours (9). Analysis of Holter-ST monitoring and clinical symptoms preceding the event support the notion that up to 42% of arrhythmic deaths are not related to ischemia (13). In the absence of an ischemic trigger, arrhythmic events probably occur as a combination of conditions including the presence of a myocardial substrate, inhomogeneous repolarization, premature depolarizations and neurohormonal factors.

It is possible that an interaction between ischemia and myocardial substrate plays some role in SAD events post-MI, although methodological limitations relegate this possibility mostly to speculation. Interestingly, a study of ICD patients using non-invasive assessment of ischemic and scar areas after MI has revealed that the incidence of life-threatening arrhythmias associates with infarct size more than ischemic burden (18).

Risk Factors for Sudden Arrhythmic Death Post-MI

Many markers for sudden cardiac death post-MI have been identified. These include left ventricular function, ECG monitoring findings, signal averaged ECG and others. Some of these predict both arrhythmic and non-arrhythmic death; others appear to be more specific for arrhythmic death and arrhythmic events. Notably, exercise or pharmacological stress testing and other indexes of ischemia have been excluded from our discussion, as they primarily assess residual ischemia. These tests predict the occurrence of coronary events and consequently possible secondary, ischemia induced arrhythmic episodes. Other investigators have reported a correlation between ischemic markers and SAD. For instance, analysis of the Coronary Artery Surgery Study (CASS) registry revealed that angiographic severity of CAD and electrocardiographic evidence of ischemia are correlated with SAD (19), but do not differentiate between SCD and non-sudden cardiac death risk.

Many investigations of post-MI patients have used as endpoints both arrhythmic death and sustained ventricular arrhythmias. Indeed an argument could be made in favor of a distinction of these two events and possibly a different approach. Moreover, markers of sudden arrhythmic death may differ from markers of non-lethal arrhythmic events. For the purposes of our discussion, we have considered studies that analyzed lethal and also non-lethal ventricular arrhythmias (including both hemodynamically stable VT and resuscitated cardiac arrest).

Left Ventricular Function

A marked impairment of left ventricular function (LVF) has been shown to be one of the most accurate predictors of mortality after MI (6,20). Different methods such as two dimensional echocardiography, radionuclide or angiographic ventriculography, indexes of functional capacity or physical findings of heart failure (S3 gallop or rales) have been consistently proven to be predictive of mortality post-MI (17). LVF impairment also correlates with the incidence of SAD. Indeed, even mild degrees of impairment correlate with an increase in lethal arrhythmic events, although the sharpest increase is noted when the LVEF decreases below 40% (10). This predictive value is strongest in the first 6 months after the infarction (10).

LVF has been used as a reference prognostic index in various studies designed to assess the role of other proposed markers such as SAECG (21), HRV (22) and more recently QT dispersion (23). These studies have confirmed the value of LVF indexes in predicting SAD post-MI as an independent risk factor. Depressed left ventricular function has been used to select post-MI patients at high risk for sudden arrhythmic death in large post-MI survival studies such as CAST (24) and MADIT (25). LVF indexes though do not help identify a post-MI population at higher risk for arrhythmic versus non-arrhythmic cardiac death. In fact the progressive overall mortality increase with decreasing LVF is not accompanied by an equal increase in the incidence of SAD. For patients with the lowest LVF values the relative impact of non-sudden, non-arrhythmic causes is stronger than SAD (3).

In a population of post-MI patients who have received thrombolysis, the prognostic value of LV function measures remains significant, although the incidence of arrhythmic events is significantly lower than in patients not treated with thrombolysis (26,27).

Arrhythmias on ECG Monitoring

Frequent and complex ventricular arrhythmias

Ventricular premature depolarizations (PVDs) have been identified as a marker of higher total mortality in post-MI patients (28). Presence, frequency and complexity of PVDs on ambulatory ECG recordings also correlate with an increased risk for sudden arrhythmic mortality in the first 1 to 2 years after MI (10,11,28). The risk appears to increase significantly from baseline when the frequency of PVDs exceeds 10 per minute (6,10,29). The relative risk for SAD associated with the presence of frequent (>10/hour) ventricular premature depolarizations is around 2 to 3. (10,11,28,30). Complexity has been defined as changing morphology and/or repetitive forms of VPDs, and has been found to be a predictive marker for SAD (10). The most widely utilized methods to grade VPDs according to their potential risk are the Lown and the Moss grading systems, which take into account frequency, morphology, repetitive character and specific timing of the VPDs (31,32). The most powerful predictor has been identified by various authors as the presence of runs of nonsustained ventricular tachycardia, although it may be less sensitive than the criterion of >10 PVDs /hour (10,33). It appears that any nonsustained VT run increases significantly the risk of subsequent SAD and total cardiac mortality, irrespective of rate, duration or frequency (34). Higher beat-to-beat cycle length variability but not faster rates of nonsustained VT also appear to portend a higher risk (35).

As concerns the optimal duration of monitoring, evidence supports the concept that recordings as short as 1 hour, obtained within the first 4 weeks post-MI, have almost equal prognostic accuracy as 6 or 24-hour recordings. A longer duration of monitoring has slightly higher sensitivity but lower specificity (36-37).

The prevalence and predictive value of ventricular arrhythmias after thrombolytic therapy has been evaluated by the GISSI-2 investigators (38). Analyzing their large cohort (8676 patients), frequent and complex PVDs were observed respectively in 19.7% and 33% of patients, values comparable to previous non-thrombolysis studies. On the contrary, nonsustained VT was found in only 6.8% of patients (versus 11% found by Bigger et al [ 10] ). During a 6 months follow-up, 256 deaths were recorded; of these, interestingly, only 84 (33%) were classified as sudden. On multivariate analysis, more than 10 PVDs/hour and complex PVDs carried respectively a 2.2 and 2.1 relative risk for sudden death, while nonsustained VT was not an independent predictor of total nor sudden mortality. Therefore, nonsustained VT should probably be viewed as a less accurate risk factor for SAD than frequent or complex PVDs after thrombolysis.

It must be stressed that the ECG recording indexes described above carry a very low positive predictive value, although they appear to be independent risk factors for sudden arrhythmic death after correcting for other risk factors such as left ventricular dysfunction (10,38).

The CAST trial showed that in patients with a depressed LVEF after MI and frequent VPDs, the use of flecainide and encainide to suppress VPDs was associated with a higher total and sudden death rate (24). Analysis of this study data base revealed that disappearance of VPDs with the lowest dose of the first drug tried correlated with a low risk of sudden death. In these patients in fact the incidence of sudden death was not statistically different from control patients who had been randomized to placebo (39).

Sustained ventricular arrhythmias

Sustained VT or Vfib occurring in the first 48-72 hours of MI onset should be considered as secondary arrhythmias induced by acute ischemia or infarction. Although the prognostic significance of early ventricular tachyarrhythmias in this clinical setting remains controversial (40,41), the prevailing view considers these arrhythmias secondary phenomena and as such expression of acute and transient changes in the electrophysiological properties of the ventricular myocardium. Currently, no further investigation is proposed for patients suffering early ventricular fibrillation, while sustained VT even in the first 48 hours seems to portend a worse prognosis (106).

A few remarks need to be made about the subgroup of patients who develop spontaneous, hemodynamically tolerated episodes of sustained VT more than 3 days after the MI. The risk of SAD after stable VT varies as a function of the underlying LVF, impact of the arrhythmia on the hemodynamic status and precipitating circumstances (42,43). As an example of the wide variation of clinical significance of this entity, the GISSI-2 investigators observed a mortality of 16.7% during a 6-months follow-up of 12 patients with sustained VT after MI during Holter monitoring (38). Kleiman reported on the survival of 87 patients experiencing sustained VT 3 to 90 days after infarction. The incidence of SAD or arrhythmia recurrence was 25% over a follow-up period of 26 months, with a 41% overall mortality. In this study, a short time from MI to arrhythmic events and anterior wall MI portended a worse prognosis (44). On the other hand, J. Brugada et al found that patients with sustained VT after a MI had a 0% risk of SAD during a follow-up of 2 years when hemodynamically stable VT occurred more than 2 months after the MI. For patients developing sustained VT within 2 months, the risk was 11% if NYHA functional class was ³ 3, or 4% for NYHA class <2 (45).In conclusion, early occurrence of ventricular arrhythmias, hemodynamic compromise and reduced LVF have repeatedly been shown to be predictors of higher SAD and total mortality in post-MI patients.

The current use of implantable defibrillators in patient with a history of MI and episodes of sustained ventricular tachyarrhythmias will probably change significantly the natural history of this population. The widespread use of these devices though should not dissuade the clinician from assessing the patient's prognosis based on the criteria mentioned above.


The technique of signal averaging of the QRS (SA-QRS) with high-gain amplification and filtering is used to detect "late potentials" in the terminal QRS complex or the ST segment. It is a non-invasive test for patients post-myocardial infarction to assess their risk of sustained ventricular arrhythmias. This is an attractive non-invasive test for risk stratification, since the presence of late potentials on the signal-averaged electrocardiogram (SAECG) is considered a non-invasive marker of the arrhythmic substrate.


The QRS complex on the surface electrocardiogram results from the sum of the individual electrical action potentials of depolarizing Purkinje fibers and ventricular myocytes. However, slow and inhomogeneous propagation of conduction in viable tissue around a myocardial scar is usually not apparent on the surface ECG because the voltage generated is too low in amplitude to be detected. These concealed low-amplitude electrical waveforms are clinically important, since slow and inhomogeneous conduction propagation is an important condition for the occurrence of reentry. Indeed, most clinically significant ventricular arrhythmias in the setting of a previous myocardial infarction are reentrant in origin. In humans, late potentials occurring in the terminal QRS complex or the ST segment of the ECG have been recorded with intracardiac electrodes during cardiac catheterization or during intra-operative mapping. Their presence can also be detected noninvasively by signal processing techniques.

The purpose of signal processing is to decrease random noise in the amplified ECG in order to enhance detection of low-amplitude signals measuring only a few mV.

The primary source of noise is skeletal muscle activity, which measures 5 to 25 mV and cannot be eliminated by filtering because its frequency content is more than 25 Hz and falls into the frequency range of cardiac potentials. In ensemble averaging, sequential beats are accepted for averaging if they satisfy a predetermined morphology called template. To reduce noise effectively, several criteria need to be met. These include the following: 1) the waveform of interest must be repetitive so that multiple samples can be aligned to obtain an averaged signal. Ventricular premature beats are excluded by comparing a new beat against a template of previous beats before averaging. 2) The waveform of interest must have a fixed temporal relationship with a reference event in time represented by the QRS complex, or the averaged waveform will be smooth and high-frequency details will be lost. 3) The noise has to fall randomly in relationship to the signal of interest. Averaging 100 to 400 beats will reduce noise by a factor of 10 to 20. Averages of approximately 200 beats will usually yield levels of less than 1 mV. Signal averaging can be performed for a specific number of beats or a specific noise level. After averaging, the signal-averaged QRS complex is filtered at corner frequency of 25 to 40 Hz (high pass = frequencies lower than 25 to 40 Hz are filtered out) and 250 Hz (low pass = frequencies high than 250 Hz are filtered out). Although the best lead system in terms of sensitivity remains undefined, most laboratories used the orthogonal X, Y, Z lead system.

Signal-averaged ECG data thus obtained can be analyzed in the time-domain or in the frequency-domain mode. In the time-domain analysis, voltage changes with time are examined over a window of frequencies defined by filtering. In the frequency-domain analysis, the signal is mathematically decomposed into different frequencies by a method known as the Fourier transform.

Quantitative Analysis of SAECG in the Time-Domain

The quantitative signal-averaged parameters measured on the vector magnitude include 1) the entire duration of the SA-QRS complex (tQRS), 2) the root mean square (RMS) voltage of the terminal 40 msec (RMS-40) of the SA-QRS and 3) the duration of the low-amplitude signals of <40 mV (LAS). It should be noted that these quantitative parameters vary with the high-pass filter selected for analysis.

Analysis of SAECG in the Frequency domain

Limitations of SAECG time-domain analysis such as the need for a high pass filter or the exclusion of patients with bundle branch block have led some investigators to examine SAECG data using the frequency-domain approach. This method analyses the SAECG in terms of the relative contributions of specific frequencies that comprise the ECG signal, in particular for segments during and right after the QRS. The rational behind this technique is to identify signals from arrhythmogenic substrates by picking up changes in the frequency component of the ECG. These frequency content changes would arise as the depolarization wavefront propagates from fast, homogeneously conducting myocardium to semi-fibrotic areas with slow and inhomogeneous conduction, thus altering specific frequency components. Different frequency-domain analysis methods have been proposed, including spectral temporal mapping, spectral turbulence analysis and acceleration spectrum analysis. Velasquez et al have compared the predictive accuracy of time domain parameters with various frequency-domain analyses in a group of post-MI patients. In this study, spectral turbolence analysis had the best predictive accuracy for all patients, but time domain analysis was the most predictive for patients without a bundle branch block (46). Stimulating findings emerged from a recent study by Schmidt et al. who looked at the MPIP and EMIAT databases applying turbulence analysis of post PVDs heart rate. The authors found that absence of heart rate turbulence – that is a biphasic, transient oscillation of heart rate after isolated PVDs, was a very potent and independent risk stratifier post-MI, stronger then conventional ones such as low LVEF and PVDs (107).

Frequency-domain analysis is still not used in clinical practice. Most of the following discussion will focus on time-domain analysis of the SAECG.

Definition of Late Potentials

The definition of what constitutes late potentials and therefore an abnormal SAECG is shown on Table X. At 40 Hz high-pass filtering, an abnormal SAECG or the presence of late potentials is considered if one or more of the following three criteria are met 1.) The filtered QRS complex is greater than 114 msec 2.) The duration of LAS is >38 msec 3.) The RMS-40 is <20 mV (47). The predicted value of these variables depends on how they are used to define a normal SA-ECG. In our experience the RMS-40 has been the most sensitive parameter, but less specific than tQRS for predicting an arrhythmic event. In contrast, tQRS has shown a lower sensitivity but the best specificity. Our studies have indicated that the combination of the SA-QRS duration and RMS-40 provides the highest predictive value when compared with the three variables used alone or in combination (48). A substudy of the CAST investigation by El Sherif et al indicated that a tQRS >120 msec was the most predictive SA-ECG criterion for arrhythmic events in the first year after myocardial infarction (49). Borggrefe et al. have reviewed the predictive effect of different SAECG criteria in post-myocardial infarction patients (50). These authors concluded that time domain SAECG analysis yielded the most reproducible data and had a definite and independent role in improving the predictive accuracy of SAD risk stratification (51-53).

Electrophysiologic Basis of Late Potentials

Studies in experimental myocardial infarction have shown that fragmented electrograms outlasting the QRS complex can be recorded from infarcted areas. These electrograms correspond to surviving Purkinje fibers and myocytes interspersed with connective tissue. The increased coupling resistance between surviving myocytes causes slow and inhomogeneous impulse conduction and hence potentials that appear "late" when recorded from the body surface. Removal of these areas partially infarcted areas by endocardial resection abolishes late potentials; this has been correlated with noninducibility of ventricular tachycardia in (54). However the presence of this arrhythmic substrate should not be automatically equated with the occurrence of sustained ventricular tachycardias. Other electrophysiologic characteristics such as inhomogeneous refractoriness of myocardial areas involved in the tachycardia circuit are equally important for the occurrence of these arrhythmias. Late potentials may originate from partially fibrotic areas, which do not pose a risk for reentrant arrhythmias due to surrounding impulse conduction characteristics. This lack of specificity may account for the relatively low positive predictive value for arrhythmic events.

Prevalence of Late Potentials after Myocardial Infarction

The prevalence of late potentials following myocardial infarction according to different studies ranges around 25% to 50% (55). This variability relates to differences in patient groups, time of recording, techniques, and definition of late potentials. McGuire has reported (56) that the prevalence of late potentials increases from 32% on day 1 to 52% on day 7-10. An initially normal SA-ECG at one week after myocardial infarction rarely becomes abnormal, but approximately 16% of abnormal recordings become normal at six weeks and 30% at 1 year (56). The prevalence of late potentials also depends on the site of myocardial infarction (48). The incidence of late potentials defined as at least one quantitative variable being abnormal, was 56% in patients with inferior or inferoposterior infarcts, as compared with 27% in patients with anterior or anteroseptal infarcts. These observations are most likely explained by differences in activation timing of different segments of the left ventricle, where the inferoposterior area is activated later than the anteroseptal. A late potential arising from the inferoposterior area is therefore more likely to be apparent on the surface ECG.

Relationship between SA-ECG and Other Predictors of Arrhythmia Risk

A lack of correlation between late potentials and left ventricular ejection fraction has been shown in various studies (48,57). A positive SA-ECG has been shown to be independent of the presence of high-grade ventricular ectopy following myocardial infarction (48,57,58). A more recent study by El Sherif confirmed that SA-ECG positivity did not correlate with complex ventricular arrhythmias on a 24-hour Holter monitor and also showed that time-dependent changes in the SA-ECG were discordant with changes in prevalence of nonsustained ventricular arrhythmias (59).

Prognostic Significance of Late Potentials after Myocardial Infarction

Several prospective studies have established the prognostic significance of SA-ECG parameters following myocardial infarction as concerns the occurrence of malignant ventricular arrhythmias (58-61). Of note, there is a high variability among different studies in the definition of abnormal SA-QRS (abnormal total SA-QRS duration, RMS-40, LAS, or a combination of the above three parameters). Despite this limitation, which in part explains the different sensitivity and specificity that were reported, SA-ECG parameters are recognized as an independent prognostic marker for ventricular arrhythmias following a myocardial infarction.

In recent years, several prospective follow-up studies have evaluated the prognostic value of SA-ECG following MI. For example, in a study of 115 patients post-MI, 16 experienced an arrhythmic event during a follow-up of 14 ± 8 months (48). In this study, the sensitivity of late potentials for predicting an arrhythmic event was higher at 40-Hz high-pass filtering than at 25-Hz. Also, it was noted that sensitivity and specificity of late potentials were dependent on their definition. If abnormal SA-ECG was defined as at least one abnormal variable, the sensitivity was 81% and the specificity was 65% to predict an arrhythmic event. When it was defined as at least two abnormal variables, the sensitivity decreased to 69% and the specificity increased to 80%. If an abnormal SA-ECG was defined as a prolonged SA-QRS duration plus an abnormal RMS-40, than the specificity increased remarkably to 95%, but led to a decline in sensitivity to 56%.

Interestingly, a positive SA-ECG seems to retain its prognostic significance even over a longer follow-up period. Zimmermann et al have found that SA-ECG performed at enrollment, with cutpoints of 118 msec for tQRS duration and 45 msec for LAS, had a 99% and 95% negative predictive value (percentage of patients with a negative study remaining arrhythmia free) at 1 and 5 years respectively. At the same intervals the positive predictive value was 13% and 20% (percentage of patients with a positive study experiencing arrhythmic events) (62).

Predictive Value of SA-ECG Relative to the Site of Myocardial Infarction

The predictive value of late potentials depends on the site of myocardial infarction. In particular, in patients with anterior or anteroseptal myocardial infarction, the absence of late potentials has a weaker negative predictive value and the presence of late potentials has a stronger positive predictive value. Conversely, in inferior wall myocardial infarction, late potentials have a weaker positive predictive value and a stronger negative predictive value (personal communication). A recent study has assessed the utility of adjusting SA-QRS duration criteria for the site of the MI. The authors concluded that this adjustment produced a more uniform predictive performance (63).

Impact of Thrombolysis on Late Potentials

Reperfusion of the infarct related coronary artery (IRA) territory following thrombolytic therapy has contributed to the improvement in survival of post-MI patients. This effect is likely secondary to a decrease in infarct size, better post-MI remodeling, and improved electrical stability. In recent years, several studies (27,64-68) have reported on the effect of a patent IRA on late potentials and subsequent arrhythmic events (see table I). In a study by Gang et al, late potentials were seen in 6% of patients with an open IRA and in 32% of patients with a closed IRA (p < 0.02) (64). Zimmerman et al (66) observed late potentials I 13% of patients with a patent and 26% with a closed IRA. Vatterot et al (67) reported a 20% incidence of late potentials in patients with a patent IRA versus 71% incidence in patients with a closed IRA. This study showed that the best predictor of late potentials was a closed coronary artery, prior MI, and age, whereas left ventricular ejection fraction was not a predictor for late potentials. Again, differences in definition of late potentials may account for interstudy variability.

In conclusion, early thrombolysis results in significant reduction of late potentials and this effect appears to be related to the presence of an open IRA. Infarct size reduction does not appear to explain the decrease in late potentials observed after thrombolysis. These observations have led to the hypothesis that thrombolysis may prevent the formation of a myocardial substrate for ventricular arrhythmias not only by preserving the left ventricular function but also by altering favorably the scarring process. The exact nature of this improved electrical stability after successful thrombolysis has not been fully elucidated.

Conflicting data exist concerning the impact of primary angioplasty on the incidence of late potentials and the comparison between this reperfusion modality and thrombolysis, but it appears that angioplasty is equally (69) if not more effective (70) than thrombolytic therapy in reducing SAECG abnormalities after acute MI.

The reduced prevalence of abnormal SA-ECG after thrombolysis leads to a decreased predictive accuracy of this test in patients undergoing thrombolysis. Thus, in the era of thrombolytic therapy, risk stratification utilizing SA-ECG alone is even less appropriate. In table II we have summarized the results of studies assessing the impact of thrombolytic therapy on prevalence of late potentials and subsequent arrhythmic events.

Limitations of SA-ECG

Time-domain analysis of the SA-ECG cannot be carried out on patients with bundle branch block. Frequency-domain analysis can still be performed on bundle branch block QRS signals, although its accuracy in this population has remained controversial. Neither time- nor frequency-domain QRS signal averaging should be carried out before day 5 post-MI, as the few studies that have evaluated the prognostic value of these techniques before day 5 have revealed a high false negative rate (59). The biggest limitation of the technique probably still lies in its low positive predictive value; for this reason it is recommended that it should be used as part of a more comprehensive risk stratification approach in post-MI patients.

Newer SAECG analyses

A new use of time-domain study of the SAECG is represented by the analysis of abnormal intra-QRS potentials or AIQPs. AIQPs are low amplitude notches or slurs, which may occur anywhere in the QRS. They are possibly related to abnormal myocardial activation related to scarring. Abnormal signals related to a reentrant substrate can occur anywhere within the QRS during normal sinus rhythm, not necessarily at the end. AIQP are identified with a computerized processing method that builds a QRS "model", representing the smooth, predictable part of the QRS. Subtraction of the model QRS from the SA-QRS yields the unpredictable, residual signal, or AIQPs. These are then quantified with a RMS calculation. Lander et al (71) examined a group of 173 patients followed post-MI for a mean of 14± 7 months. Sixteen arrhythmic events were observed during this period. Mean AIQP amplitudes were greater in the arrhythmic-event group and appeared to be independent of late potentials. Compared to other non-invasive variables for the prediction of arrhythmic events, AIQP yielded the best accuracy (90%) and positive predictive value (46%).

Programmed Stimulation Technique

Electrophysiological testing is used to assess the risk of ventricular tachyarrhythmias in a variety of different conditions. Electrode-tip catheters are advanced percutaneously through the superior or inferior venae cavae tributaries and placed in contact with the endocardial surface of the right ventricle. Pacing stimuli are delivered according to specific protocols defined as programmed electrical stimulation. This includes a train of regularly spaced stimuli (S1) followed by an increasing number of prematurely placed stimuli or extrastimuli (S2,S3,S4) and burst rapid pacing. These sequences are repeated at least in two different sites, usually the right ventricular apex and the right ventricular outflow tract. The stimulation protocols are designed to induce ventricular tachyarrhythmias, typically those based on a reentrant mechanism.

Prognostic value of programmed stimulation

A number of studies evaluated the ability of programmed stimulation to predict the occurrence of arrhythmic events and sudden arrhythmic death after MI in unselected patients groups, i.e. without prior spontaneous ventricular tachyarrhythmias or risk factors for SAD (60,72). Unfortunately, no standard stimulation protocol has been used. These studies report a positive predictive value of inducible tachycardia ranging from 12.5% to 33% and a negative predictive value >90%. Given this disappointing yield this test is not recommended for a low risk population.

The role of programmed stimulation in post-MI populations identified by at least one non-invasive marker of SAD has been evaluated more extensively. Various studies have assessed the predictive significance of programmed stimulation in post-MI patients with nonsustained ventricular arrhythmias. Gomes et al. (73) induced sustained VT in 17 of 40 post-MI patients with high-grade ventricular ectopy, using a protocol with up to 2 extrastimuli. Other investigators using up to 3 extrastimuli (74-78) obtained inducibility in 40% to 50% of their patients. After a follow-up period of 14 to 30 months, the incidence of SAD was around 30% for inducible patients and 2%-6% for non-inducible. The negative predictive value is in the range of 90%, and the positive predictive value is approximately 10-20%. These studies indicate that the induction of sustained ventricular arrhythmias in patients with history of MI and spontaneous nonsustained ventricular tachycardia may identify a subgroup at increased risk of subsequent SAD and sustained ventricular arrhythmias. Induction of sustained monomorphic VT is a more specific marker of subsequent SAD and spontaneous sustained VT then induction of polymorphic VT or Vfib (78). The vast majority of patients with inducible arrhythmias enrolled in these studies were treated with antiarrhythmic drugs during follow-up. An extremely significant difference in the incidence of SAD was found between patients whose inducible arrhythmias were suppressed by anti-arrhythmic agents and patients with non-suppressible arrhythmias. The incidence of SAD in patients whose inducible arrhythmias were not suppressed by antiarrhythmic drugs can be as high as 50% at 2 years (75). Fewer data exist concerning the risk of SAD in post-MI patients with inducible ventricular tachyarrhythmias not suppressible by drugs who receive no antiarrhythmic drugs. Some evidence suggests that this group represents a high risk for SAD (42). The ongoing Multicenter Unsustained Tachycardia Trial (79) randomized post-MI patients with a LV ejection fraction <40% and asymptomatic nonsustained VT to programmed electrical stimulation or no treatment. Patients with an inducible ventricular arrhythmia underwent serial electrophysiological drug testing and are being treated with an antiarrhythmic drug. Patients whose induced arrhythmia was not suppressed nor stabilized by a drug are being followed off antiarrhythmic agents. The results of this study may clarify the prognostic significance off non-suppressible ventricular arrhythmias in the absence of antiarrhythmic therapy.

In addition, the risk carried by inducibility of ventricular arrhythmias is strongly modulated by ventricular function. In those studies (74-77) the rate of SAD was very low in patients with LVEF >40%, regardless of the result of programmed stimulation.

The MADIT study (25) enrolled post-MI patients with depressed LVF (<35%), nonsustained ventricular tachycardia, inducible VT or Vfib at programmed stimulation, which could not be suppressed by intravenous procainamide. The study randomized to ICD implant or conventional medical therapy (74% were given amiodarone). Twenty seven months follow-up of these patients demonstrated a significantly higher total mortality (38% vs. 15%, p=0.009) and arrhythmic mortality (12% vs. 3%) in the group randomized to medical therapy versus the ICD group, leading to the interruption of the randomizing phase. The high mortality observed in the group of patients receiving standard therapy emphasizes the poor prognosis implied by a positive (and non-suppressible) programmed stimulation study in a low LVF post-MI population. Pedretti et al evaluated 303 patients with an algorithm combining LVF, Holter results and late potentials. Sixty-seven patients thus classified at high risk were referred for programmed electrical stimulation, to which 47 consented. Monomorphic VT at a rate <270 beats/min was induced in 20 and resulted to be a strong predictor of arrhythmic events, with a 65% positive predictive value (61).

In conclusion, programmed stimulation appears to be particularly useful to provide further risk stratification of patients with depressed LVF post-MI, especially in the presence of positive non-invasive markers of SAD, such as nonsustained VT. It helps to identify a population at extremely high risk of SAD. Its role in combination with newer markers of SAD risk remains to be investigated. The use of programmed electrical stimulation is not justified in unselected groups of MI survivors (17).

Measures of Autonomic Tone - Heart Rate Variability

The heart rate manifests continuous changes over time. During normal sinus rhythm, the heart rate and its inverse measurement, the RR interval, vary from beat to beat in response mainly to changes in autonomic function. Heart rate variability (HRV) reflects the balance in the activity of parasympathetic and sympathetic nerve fibers innervating the sinoatrial node and of circulating catacholamines. Vagal denervation of the atria significantly decreases HRV and abolishes baroreflex sensitivity (see next chapter) in an experimental model (80). HRV can be calculated from a routine Holter or telemetry recording but requires a special software analysis. Premature beats and noise need to be taken out of the analysis in order to avoid the introduction of spurious variability data. Similarly to the SAECG, analysis of HRV can be performed in a time-domain or in a frequency-domain mode.

In the time-domain, HRV can be simply expressed as the standard deviation of all normal RR intervals (SDNN) or as more complex indexes such as the percent of adjacent normal RR intervals whose difference exceeds 50msec (NN50+) or the root-mean-square of successive RR interval differences (RMSSD). The higher the value of SNDD, the greater the HRV. On the other side, frequency-domain analysis of HRV evaluates different frequencies of periodic oscillations of the heart rate. It is obtained by applying a mathematical technique called fast Fourier transform. With this method, different frequency components are derived and usually grouped into four major bands: high, low, very low, ultralow frequency bands. The higher the frequency, the faster the oscillations in heart rate. For example, high frequency is defined as 0.15 to 0.40 Hz and corresponds to oscillations in heart rate lasting 2.5 to 6.6 seconds. High-frequency is considered as a reflection of the parasympathetic autonomic nervous system tone (81). Low-frequency is an expression of both the parasympathetic and the sympathetic system and is influenced strongly by the baroreceptor system (82). Very low frequency and ultralow frequency bands appear to be influenced by various other factors. Total power is the sum of all these bands.

Prognostic significance of HRV after MI

Wolf et al (83) measured the HRV of 176 patients admitted to the coronary care unit with acute myocardial infarction on a 60-second ECG recording and correlated the variability with hospital mortality rate. Kleiger et al retrospectively analyzed the Holter data of the MPIP study (22). They defined HRV as standard deviation of normal RR intervals in a 24-hour continuous monitoring obtained 11± 3 days post-MI. HRV represented the strongest univariate predictor of mortality over a follow-up period of 31 months, with a relative risk of 5.3 for patients with HRV <50msec compared to those with HRV >100msec. After adjusting for other variables including frequency of VPDs and left ventricular ejection fraction, HRV remained a significant predictor of mortality, despite an inverse relation to left ventricular ejection fraction. It is noteworthy that sudden death or arrhythmic events were not an endpoint of the study.

The association between all-caused mortality, cardiac death, and arrhythmic death post-MI with frequency-domain measures of HRV has been studied by Bigger et al (84). The 24-hour Holter recording of 715 patients included in the MPIP study were analyzed to obtain six frequency-domain measures (high, low, very low, ultralow frequency bands plus total power and low frequency/high frequency ratio). The follow-up period of this study was four years during which there were 112 all-cause deaths, 84 cardiac deaths, and 64 arrhythmic deaths. After adjusting for age, New York Heart Association functional class, rales in the coronary care unit, LVEF, and ventricular arrhythmias on the Holter recording, ULF and VLF remained significantly associated with mortality (p<0.001). A trend was found for VLF to be more strongly associated with arrhythmic death than all-cause or cardiac death.

The GISSI-2 investigators have studied the prognostic value of HRV in 567 post-MI patients who had received thrombolysis and were followed over a mean of 30 months period. Three time-domain HRV indexes, including SDNN and NN50+ were strongly associated with increased cardiac mortality. Analysis of sudden versus non-sudden cardiac death was not performed (85).

In summary, HRV is a predictive marker of cardiac mortality and SAD after MI and appears to be independent of other methods. It maintains prognostic significance in patients treated with thrombolytics. The ability of HRV to discriminate between sudden and non-sudden mortality risk has to be investigated further before the test can be proposed as a clinical tool in the risk stratification for SAD after infarction.

Measures of Autonomic Response - Baroreflex Sensitivity

Baroreflex sensitivity (BRS) is another measure of autonomic nervous system activity on the heart rate. It is commonly measured by assessing the heart rate response to an intravenous infusion of phenylephrine, an a -adrenergic agonist. This infusion induces an elevation in systemic arterial blood pressure and consequently a slowing of the heart rate via a vagal reflex. BRS is expressed as the prolongation of the RR interval in milliseconds over the blood pressure increase in millimeters of mercury. Initial work by Schwartz et al using an experimental dog infarction model showed that BRS was significantly lower after anterior wall MI and that decreased BRS correlated with susceptibility to ventricular fibrillation during exercise stress test and partial coronary occlusion 30-days post-MI (86).

Human studies of BRS after myocardial infarction have shown that BRS is lower in post-MI patients than in age-matched controls and that it slowly returns to pre-infarction levels within three months. The first clinical study to assess the prognostic significance of BRS followed 78 patients less than 65 years of age after their first myocardial infarction (87). In this study, BRS did not correlate with LVEF. BRS was found to be remarkably lower in the six patients who died after a mean follow-up of two years than in the survivors (7.6%). A BRS <3 msec/mm Hg carried 50% mortality rate versus 3% in patients with BRS >3 msec.

A recently published multi-center study (ATRAMI) has evaluated prospectively the prognostic value of HRV and BRS in 1,284 patients less than 80 years of age with a recent MI (less than 4 weeks) (88). Sixty three percent of these patients received thrombolytic therapy. During 21± 8 months of follow-up, 44 cardiac deaths and 5 cardiac arrests were observed. Decreased HRV (SDNN <70 msec) or depressed BRS (<3 msec/mm Hg) were both associated with a higher risk of mortality (3.2 and 2.8 relative risk, respectively). The relative risk of low BRS was still significant after adjusting for 2 pre-specified covariates (LVEF and VPDs). Moreover, the combination of low BRS and HRV values translated into a significantly increased mortality with a relative risk of 7.3. Analyzing the patient group younger than 65 years of age the multivariate relative risk for low BRS was 3.4 versus only .8 in patients older than 65 years of age, whereas, the multivariate relative risk for low SDNN was 2.4 versus 4.7 respectively in patients younger and older than 65. The combination of LVEF <35% and low BRS carried the highest relative risk in patients younger than 65 (11.5, p=0.0001). The authors postulated that low BRS and low SDNN identified patients with an altered autonomic balance. This alteration may favor life-threatening arrhythmias via electrical instability and possibly by inducing ischemia. This hypothetical mechanism could not be clarified further since the investigators did not subcategorized cardiac mortality into arrhythmic and non-arrhythmic.

Various studies support the concept that BRS and HRV are not redundant indexes (88,89), consistent with the hypothesis that HRV measures vagal tone while BRS is a measure of vagal response to an increase in sympathetic activity.

Based on the data reported above and other clinical and experimental evidence (90), it is currently thought that increased vagal tone exerts a protective antiarrhythmic effect. HRV and BRS express cardiac vagal activity, respectively ambient and reflex; their impairment is thus considered a predictor of SAD. This interpretation is probably too simplistic. In fact, as we noted above, HRV is depressed after the MI but tends to return to pre-infarction levels within a few months. If the vagal tone also returns to normal levels, how would depressed HRV in the immediate post-MI period identify a group at persistent high SAD risk? Although the pathophysiological mechanisms need to be clarified, the evidence linking HRV and BRS to SAD post-MI is undoubtedly very strong.

Other Methods

QT interval dispersion

The QT interval on a surface ECG reflects ventricular repolarization; its duration varies in different standard ECG leads. QT dispersion is an index of this variability and is defined as the difference between the longest and the shortest corrected QT (QTc) interval on a 12 lead ECG. QT dispersion has been used as an index of repolarization heterogeneity . Its potential value in risk stratification for sudden cardiac death has been investigated in the setting of congestive heart failure (91) and congenital prolonged QT syndrome (92). A large prospective study has indicated that QT dispersion is a predictor of cardiac mortality and sudden cardiac death in an asymptomatic population of subjects over 55 years of age (93).

Retrospective case-control studies of post-MI patients suggested that increased QT dispersion could be associated with occurrence of ventricular arrhythmias (94-96). A prospective study following 280 consecutive MI survivors assessed the predictive value for all-cause mortality and arrhythmic events of QT dispersion and other repolarization variables such as area under the T wave (23). After a 32± 10 months period, 30 patients (11%) reached one of the end-points. Neither QT dispersion nor any of the other indexes of repolarization were predictive of all-cause mortality or arrhythmic events. Despite the negative results of this study, data of ongoing investigations will have to be evaluated before concluding that this index of repolarization heterogeneity does not have a prognostic value for SAD post-MI.

T wave alternans

Beat to beat variation in the energy or morphology of the T wave has been recognized in a variety of different heart diseases (97). This beat to beat variation (T wave alternans) is not to be confused with the gradual variability of T wave duration over time, which depends in part on heart rate variability (98). T wave alternans is often not apparent on a regular 12-lead ECG and needs to be analyzed with the aid of filtering and amplifying techniques. It may not become manifest without the use of atrial pacing or other chronotropic stimuli (99). This beat to beat fluctuation would arise from a loss of the normal high coherence between heart rate and repolarization duration (97) and has been proposed as a marker of repolarization heterogeneity and ischemia-induced susceptibility to arrhythmias (99). Rest and exercise T wave alternans, maesured with a sophisticated spectral technique have been reported to predict inducibility of sustained ventricular tachyarrhythmias at programmed stimulation with an accuracy of 80%, in a small group of patients with different cardiac diseases including coronary artery disease (100). No prospective study has evaluated to date the prognostic significance of T wave alternans for SAD in MI survivors. Despite a growing interest in the field, this tool has to be considered still experimental.

Combined Risk Factor Assessment

Various combination algorithms have been tested prospectively for their accuracy in identifying post-MI arrhythmic events and SAD (33,61,49,101,102)(see Table III). The great majority of these studies have reported a significant advantage in this multi-step approach for risk stratification. Gomes et al proposed to utilize both LVF measurements, late potentials and nonsustained VT to recognize a subgroup of patients at high risk for ventricular arrhythmias (101). Farrel et al followed 416 post MI patients studied prior to discharge with 24-hr Holter, HRV, SAECG, LVEF and stress test (33). Their investigation was designed to determine the respective predictive value of all these variables and the relationship among each other. Over a mean follow up of two years, they noted an 11% mortality and a 6% arrhythmic event incidence. Impaired HRV was the most sensitive predictor of arrhythmic events, but alone had a low positive predictive accuracy. After analyzing all possible combinations, HRV and SAECG used together were the most sensitive predictors, attributing a risk ratio of 18.5. In this study LVEF alone was not a marker of arrhythmic versus non-arrhythmic death.

A more recent investigation examined prospectively a larger group of patients and studied the predictive value of 24-hr Holter, HRV, SAECG and LVEF for differentiating arrhythmic from non-arrhythmic death risk (CAPS definition) (103). The authors concluded that depressed HRV and runs of nonsustained VT were the most sensitive predictors of arrhythmic death in multivariate analysis. In particular, the highest arrhythmic to non-arrhythmic death ratio was seen in a group of patients without severe LVEF impairment but depressed HRV and positive SAECG (total duration of the signal-averaged QRS, see chapter 6). Their analysis clearly showed that the risk imparted by one factor was influenced by the concomitance of other factors in the same patient.

It has to be emphasized that there is convincing evidence that patients with normal non-invasive tests have a risk of SAD post-MI of approximately 1% per year.

As noted above, programmed stimulation is recommended as a second line test for further risk stratification of patients with at least one positive non-invasive marker of SAD.

A Proposed Approach

Risk stratification after MI is crucial before hospital discharge, as these patients are at highest risk for sudden death in the early months after the event. It may also be more cost-effective to obtain prognostic tests while the patients are still in a hospital setting. Which tests should be therefore obtained to assess the risk of SAD after discharge and potentially to provide timely and effective preventive therapies? The clinician should be aware that recent AHA/ACC guidelines for the management of post-MI patients do not recommend any test as class 1 or class 2 for this specific purpose (17). Yet, routine hospital care of patients admitted with MI includes tests that are useful to this effect.

The following is a proposed algorithm for the assessment of SAD risk in infarct survivors, combining the different methods discussed so far. LVF indexes, obtained either by two dimensional echocardiography, radionuclide or angiographic ventriculography would provide the first line of assessment. The second method of evaluation would be ECG monitoring, which is also routinely available and should be analyzed for presence, frequency and pattern of ventricular complexes. ECG monitor data together with LVF measurements would identify a first subgroup composed by patients with depressed LVF (LVEF<40%) and frequent PVDs (more than 10/Hr) or nonsustained VT. Patients with a normal LVEF would be eliminated from further work-up even in the presence of nonsustained ventricular arrhythmias, as they represent the lowest risk group. Patients with minimal to mild impairment of LVF would be referred for a third non-invasive study, currently represented by SAECG. Patients with moderate to severe LVF impairment should be directly studied with programmed electrical stimulation. This test has been proposed by various authors as an effective method to provide further risk stratification by recognizing the presence of an anatomical substrate of ventricular arrhythmias. Alternatively, the third step could be represented by HRV studies. HRV measurements may, in the next future, result to be an effective tool, particularly for patients who received thrombolytic therapy, in which SAECG and possibly nonsustained VT have a lower predictive accuracy. The advent of modern telemetry stations in CCUs and cardiology wards could potentially provide the clinician with a routine predischarge combined assessment of ventricular arrhythmias, SAECG and HRV. Utilizing a two or three step approach, a subgroup of infarct survivors at highest risk for SAD would be identified. Currently we advocate that this group should be referred for programmed stimulation study. Programmed stimulation would be carried out in this subgroup in order to identify a population that could benefit from further intervention before discharge, such as implant of an ICD. Programmed stimulation also may guide subsequent therapy by establishing both the characteristics of inducible arrhythmias and the response of these arrhythmias to antiarrhythmic drugs.

Future trends

A more thorough understanding of the pathophysiology and dimensions of the phenomena that we currently group under the term of SAD post-MI will be critical to approach this problem effectively. A new direction of study is represented by the analysis of electrocardiographic recordings obtained from implanted ICD after episodes of SAD. These data may help clarify the mechanisms of sudden arrhythmias in survivors of MI. Other ongoing efforts in this field pursue more specific non-invasive markers of sudden death risks or a modification of the established ones to achieve higher specificity tools. The ideal test would retain the high negative predictive value of available methods and show a higher positive predictive value, yielding consequently a better accuracy. It would complement routine tests such as LVF measurements and be specific for arrhythmic events as opposed to other etiologies of cardiac death and morbidity. It would then allow the clinician to make focused therapeutic decisions and optimize the use of a changing and often expensive armamentarium to prevent SAD. Also, a useful method needs to retain its prognostic value after thrombolytic therapy, as the latter changes the natural history of post-MI patients. Since advances in the care of cardiac patients, including thrombolytic therapy, have decreased overall mortality and SAD after MI, accuracy of available methods of risk assessment has to be assessed in the specific clinical setting. The ideal approach to this lower risk population would include simple non-invasive tests with the highest positive predictive value.

Clinical Implications and Conclusion

A search for an ideal risk stratification algorithm should progress along with an effort to establish the pathophysiology of lethal arrhythmic events. The risk markers so far analyzed correspond to distinct contributing factors in the genesis of lethal ventricular arrhythmias post-MI. It is critical to appreciate the diversity of these contributing factors. They can be schematically separated into three distinct types (104). Persistent factors such as the MI scar and the arrangement of surviving strands of myocytes are a direct consequence of the infarction and tend to be reproducible with instantaneous measurements, such as LVF studies, or SAECG. Longitudinal or dynamic factors such as the autonomic nervous system effects on the heart can be assessed by HRV or Holter findings; they are not constantly reproducible and change over time after the index MI, either progressively or randomly. Finally, transient factors such as ischemia or electrolyte changes or decompensation of heart failure, the timing or magnitude of which cannot be predicted directly. When they manifest, further electrophysiological alterations arise, creating a predisposing setting for SAD. This third type of factors is also most amenable to interventions, such as revascularization or electrolyte repletion. Not only some of these factors may correlate to others (e.g., LVF and PVDs) but their coexistence may create a potentiating effect in terms of SAD risk. Our discussion has focused on persistent and dynamic factors, given the elusive nature of the third type.

This schematic view, although simplistic, is useful to understand the need for a multi-step approach in risk stratification for SAD. Also, it stresses the need for caution in interpreting the results of the proposed tests. In fact, the impact of transient factors can be evaluated only indirectly (e.g., considering the use of diuretics as a predictor of risk for electrolyte imbalances).

Currently, a critical application of risk assessment is the selection of candidates for ICD implant. This represents at the present moment an intervention that modifies significantly both total mortality and SAD in very high-risk patients. Issues of cost and health care resources make the selection of candidates for ICD therapy a demanding task for the clinician in the absence of clear guidelines. Identifying high-risk populations may also help in the selection of premises for the placement of external defibrillators devices, the efficacy of which is currently being tested on a national scale (105). There is no doubt that SAD post-MI will continue to be a major challenge for the clinician; better tools are needed in a field where effective interventions may produce tremendous impact on public health.



1. Report of the working group on atherosclerosis of the National Heart, Lung, and blood Institute (Vol. 2) Patient oriented research-fundamental and applied, sudden cardiac death. DHEW, NIH Publication 83-2085. Washington DC, US Government Printing Office 1981:114-22.

2. Gillum RF. Sudden coronary death in the United States. Circulation 1989;79:756-65.

3. Myerburg RJ, Castellanos A. Cardiac Arrest and Sudden Cardiac Death. In: Braunwald E, editor. Heart disease: a textbook of cardiovascular medicine. 5th ed. Philadelphia; WB Saunders: 1997. p. 742-59.

4. Greene HL, Richardson DW, Barker AH, Roden DM, Capone RJ, Echt DS, et al, and the CAPS Investigators. Classification of deaths after myocardial infarction as arrhythmic or nonarrhythmic (the Cardiac Arrhythmia Pilot Study) Am J Cardiol 1989;63:1-6.

5. Davis HT, DeCamilla J, Bayer LW, Moss AJ. Survivorship patterns in the posthospital phase of myocardial infarction. Circulation 1979;60:1252-58.

6. The Multicenter Post Infarction Research Group. Risk Stratification and Survival After Myocardial Infarction. N Engl J Med 1983;309:331-36.

7. Goldstein S, Friedman L, Hutchinson R, Canner P, Romhilt D, Schlant R, Sobrino R, Verter J, Wasserman A. Timing, mechanism and clinical setting of witnessed deaths in postmyocardial infarction patients. J Am Coll Cardiol 1984;3:1111-17.

8. Meissner MD, Akhtar M, Lehmann MH. Nonischemic Sudden Tachyarrhythmic Death in Atherosclerotic heart disease. Circulation 1991;84:905-12.

9. Leclercq JF, Maisonplanche P, Cauchemez B, Coumel P, Respective role of sympathetic tone and cardiac pauses in the genesis of 62 cases of ventricular fibrillation recorded during holter monitoring. Eur Heart J 1988;9:1276-83.

10. Bigger JT, Fleiss JL, Kleiger R, Miller JP, Rolnitski LM, and the Multicenter Post Infarction Research Group. The relationship among ventricular arrhythmias, left ventricular dysfunction and mortality in the two years after myocardial infarction. Circulation 1984:69:250-58.

11. Moss AJ, DeCamilla J, Davis H. Cardiac death in the first six months after myocardial infarction: Potential for mortality reduction in the early posthospital period. Am J Cardiol 1977;39:816-20.

12. Pratt CM, Francis MJ, Luck JC, Wyndham CR, Miller RR, Quiñones MA. Analysis of ambulatory electrocardiograms in 15 patients during spontaneous ventricular fibrillation with special reference to preceding arrhythmic events. J Am Coll Cardiol 1983;2:789-97.

13. Marcus FI, Cobb LA, Edwards JE, Kuller L, Moss AJ, Bigger JT Jr, Fleiss JL, Rolnitzky L, Serokman R. Mechanism of death and prevalence of myocardial ischemic symptoms in the terminal event after acute myocardial infarction. Am J Cardiol 1988;61:8-15.

14. Mehta D, Curwin J, Gomes JA, Fuster V. Sudden death in coronary artery disease. Acute ischemia versus myocardial substrate. Circulation. 1997;96:3215-23.

15. Peters NS, Wit AL. Myocardial architecture and ventricular arrhythmogenesis. Circulation 1998;97:1746-54.

16. Denniss AR, Richards DA, Waywood JA, Yung T, Kam CA, Ross DL, Uther JB. Electrophysiological and anatomic differences between canine hearts with inducible ventricular tachycardia and fibrillation associated with chronic myocardial infarction. Circ Res 1989;64:155-66.

17. Ryan TJ, Anderson JL, Antman EM, Braniff BA, Brooks NH, Califf RM et al. ACC/AHA Guidelines for the Management of Patients with Acute Myocardial Infarction. J Am Coll Cardiol 1996;28:1328-1428.

18. Gioia G, Bagheri B, Gottlieb CD, Schwartzman DS, Callans DJ, Marchlinski FE, et al. Prediction of outcome of patients with life-threatening ventricular arrhythmias treated with automatic implantable cardioverter-defibrillators using SPECT perfusion imaging. Circulation 1997;95:390-94.

19. Holmes DR Jr, Davis K, Gersh BJ, Mock MB, Pettinger M. Risk factor profiles of patients with sudden cardiac death and death from other cardiac causes: a report from the Coronary Artery Surgery Study (CASS). J Am Coll Cardiol 1989;13:524-30.

20. Roig E, Magrina J, Garcia A, Armengol X, Muxi A, Melis G, Sanz G. Prognostic value of exercise radionucleid angiography in the low risk acute myocardial infarction survivors. Eur Heart J 1993;14:213-18.

21. Kuchar DL, Thornburn CW, Sammel NL. Prediction of serious arrhythmic events after myocardial infarction: signal-averaged electrocardiogram, Holter monitoring and radionucleid ventriculography. J Am Coll Cardiol 1987;9:531-38.

22. Kleiger RE, Miller JP, Bigger JT, Moss AJ, and the Multicenter Post-Infarction Research Group. Decreased heart rate variability and its association with increased mortality after acute myocardial infarction. Am J Cardiol 1987;59:256-62.

23. Zabel M, Klingenheben T, Franz MR, Hohnloser SH. Assessment of QT Dispersion for Prediction of Mortality or Arrhythmic Events after Myocardial Infarction. Results of a prospective, long-term follow-up study. Circulation 1998;97:2543-50.

24. Echt DS, Liebson PR, Mitchell LB, Peters RW, Obias-Manno D, Barker AH et al. and the CAST investigators. Mortality and morbidity in patients receiving encainide, flecainide, or placebo. N Engl J Med 1992;327:227-33.

25. Moss AJ, Hall WJ, Cannom DS, Daubert JP, Higgins SL, Klein H, Levine JH, Saksena S, Waldo AL, Wilber D, Brown MW, Heo M, for the Multicenter Automatic Defibrillator Implantation Trial Investigators. Improved Survival with an implanted defibrillator in patients with coronary disease at high risk for ventricular arrhythmia. N Engl J Med 1996;335:1933-40.

26. Turitto G, Risa AL, Zanchi E, Prati L. The signal-averaged electrocardiogram and ventricular arrhythmias after thrombolysis for acute myocardial infarction. J Am Coll Cardiol 1990;15:1270-76.

27. Pedretti R, Laporta A, Etro MD, Gementi A, Bonelli R, Anza C, Colombo E, Maslowsky F, Santoro F, Caru B. Influence of thrombolysis on signal-averaged electrocardiography and late arrhythmic events after acute myocardial infarction. Am J Cardiol 1992;69:866-72.

28. Ruberman W, Weinblatt E, Goldberg JD, Frank CW, Shapiro S. Ventricular premature beats and mortality after myocardial infarction. N Engl J Med 1977;297:750-57

29. Myerburg RJ, Kessler KM, Luceri RM, Zaman L, Trohman RG, Estes D, Castellanos A. Classification of ventricular arrhythmias based on parallel hierarchies of frequency and form. Am J Cardiol 1984;54: 1355-58.

30. The Cardiac Arrhythmia Pilot Study (CAPS) Investigators. Effects of encainide, flecainide, imipramine, and moricizine on ventricular arrhythmias during the year after acute myocardial infarction: The CAPS. Am J Cardiol 1988;61:501-9.

31. Lown B, Wolf M. Approaches to sudden death from coronary heart disease. Circulation 1971; 44:130-42.

32. Moss AJ, Davis HT, DeCamilla J, Bayer LW. Ventricular ectopic beats and their relation to sudden and nonsudden cardiac death after myocardial infarction. Circulation 1979;60:998-1003.

33. Farrell TG, Bashir Y, Cripps T, Malik M, Poloniecki J, Bennett ED, et al. Stratification for arrhythmic events in postinfarction patients based on heart rate variability, ambulatory electrocardiographic variables and the signal-averaged electrocardiogram. J Am Coll Cardiol 1991;18:687-97.

34. Buxton AE, Lee KL, DiCarlo L, Echt DS, Fisher JD, Greer GS, et al, for the MUSTT investigators. Nonsustained ventricular tachycardia in coronary artery disease: relation to inducible sustained ventricular tachycardia. Ann Int Med 1996;125:35-39.

35. Dabrowski A, Kramarz E, Piotrowicz R. Low variability of cycle lengths in nonsustained ventricular tachycardia as an independent predictor of mortality after myocardial infarction. Am J Cardiol 1997;80:1347-50.

36. Connolly SJ, Cairns JA, on behalf of the CAMIAT Pilot Study Group. Comparison of one-, six- and 24-hour ambulatory electrocardiographic monitoring for ventricular arrhythmia as a predictor of mortality in survivors of acute myocardial infarction. Am J Cardiol 1992;69:308-13.

37. Petretta M, Bianchi V, Pulcino A, Carpinelli A, Valva G, Themistoclakis S, et al. Continuous electrocardiographic monitoring for more than one hour does not improve the prognostic value of ventricular arrhythmias in survivors of first acute myocardial infarction. Am J Cardiol 1994;73:139-42.

38. Maggioni AP, Zuanetti G, Franzosi MG, Rovelli F, Santoro E, Staszewsky L, et al. Prevalence and prognostic significance of ventricular arrhythmias after acute myocardial infarction in the fibrinolytic era. Circulation 1993;97:312-22.

39. Goldstein S, Mori Brooks M, Ledingham R, Kennedy HL, Epstein AE, Pawatin Y, et al. Association between ease of suppression of ventricular arrhythmia and survival. Circulation 1995;91:79-83.

40. Jensen GV, Torp-Pedersen C, Kober L, Steensgaard-Hansen F, Rasmussen YH, Berning J, Skagen K, Pedersen A. Prognosis of late versus early ventricular fibrillation in acute myocardial infarction. Am J Cardiol 1990;66:10-15.

41. Behar S, Reicher-Reiss H, Shechter M, Rabinowitz B, Kaplinsky E, Abinader E, Agmon J, Friedman Y, Barzilai J, Kauli N. Frequency and prognostic significance of secondary ventricular fibrillation complicating acute myocardial infarction. SPRINT Study Group. Am J Cardiol 1993;71:152-56

42. Swerdlow CD, Winkle RA, Mason JW. Determinants of survival in patients with ventricular tachyarrhythmias. N Engl J Med 1983;308:1436-39.

43. Gomes JA; Mehta D; Ip J; Winters SL; Camunas J; Ergin A; Newhouse TT; Pe E. Predictors of long-term survival in patients with malignant ventricular arrhythmias. Am J Cardiol, 1997;79:1054-60.

44. Kleiman RB, Miller JM, Buxton AE, Josephson ME, Marchlinski FE. Prognosis following sustained ventricular tahcycardia occurring early after myocardial infarction. Am J Cardiol 1988;62:528-33.

45. Brugada P, Talajic M, Smeets J, Mulleneers R Wellens JJ. The value of the clinical history to assess prognosis of patients with ventricular tachycardia or ventricular fibrillation after myocardial infarction. European Heart Journal 1989;10:747-52.

46. Vazquez R, Caref EB, Torres F, Reina M, Huet J, Guerrero JA et al. Comparison of the new acceleration spectrum analysis with other time- and frequency-domain analyses of the signal -averaged electrocardiogram. Eur Heart J 1998;19:628-37.

47. Gomes JA, Winters SL, Stewart D, Targonski A, Barreca P. Optimal band pass filters for time domain analysis of the signal-averaged electrocardiogram. Am J Cardiol 1987;60:1290-98.

48. Gomes JA, Winter SL, Martinson M, Machac J, Stewart D, Targonsky A. The prognostic significance of quantitative of signal-averaged variables relative to clinical variables, sight of myocardial infarction, ejection fraction and ventricular premature beats: A prospective study. J Am Coll Cardiol 1988;13:377-84.

49. El-Sherif N, Denes P, Katz R, Capone R, Mitchell LB, Carolson M, Reynolds-Haertle R, For the Cardiac Arrhythmia Suppression Trial/Signal-Averaged Electrocardiogram (CAST/SAECG) Substudy Investigators. Definition of the best prediction criteria of the time domain signal-averaged electrocardiogram for serious arrhythmic events in the post infarction period. J Am Coll Cardiol 1995;25:908-14.

50. Borggrefe M, Fetsch T, Martinez-Rubio A, Makijarvi M, Breithardt G. Prediction of Arrhythmia Risk Based on Signal-Averaged ECG in Postinfarction Patients. PACE 1997;20(Pt II):2566-76.

51. Malik M, Kulakowski P, Poloniecki J, Staunton A, Odemuyiwa O, Farrell T, et al. Frequency versus time domain analysis of signal-averaged electrocardiograms. I. Reproducibility of the results. J Am Coll Cardiol 1992;20:127-34.

52. Kulakowski P, Malik M, Poloniecki J, Bashir Y, Odemuyiwa O, Farrell T, et al. Frequency versus time domain analysis of signal-averaged electrocardiograms. II. Identification of patients with ventricular tachycardia after myocardial infarction. J Am Coll Cardiol 1992 Jul;20(1):135-43

53. Odemuyiwa O, Malik M, Poloniecki J, Farrell T, Kulakowski P, Millane T, et al. Frequency versus time domain analysis of signal-averaged electrocardiograms. III. Stratification of postinfarction patients for arrhythmic events. J Am Coll Cardiol 1992;20:144-50.

54. Denniss AR, Johnson DC, Richards DA, Ross DL, Uther JB. Effect of excision of ventricular myocardium on delayed potentials detected by the signal-averaged electrocardiogram in patients with ventricular tachycardia. Am J Cardiol 1987;59:591-95.

55. Gomes JA. The Prognostic Significance of the Signal Averaged Electrocardiogram in the Infarct Survivor. In: Acute Myocardial Infarction. Gersh BJ, Rahimtoola SH, editors. Acute Myocardial Infarction. 2nd ed.; New York; Chapman & Hall.1997. p. 687-703.

56. McGuire M, Kuchar DL, Ganes J, Sammel NL, Thornburn CW. Natural history of late potentials in the first ten days after acute myocardial infarction and relation to early ventricular arrhythmias. Am J Cardiol 1998;61:1187-90.

57. Steinberg JS, Regan A, Sciacca RR, Bigger JT, Fleiss JL. Predicting arrhythmic events after acute myocardial infarction using the signal-averaged electrocardiogram. Am J Cardiol 1992;69:13-21.

58. Kuchar DL, Thornburn CW, Sammel NL. Prediction of serious arrhythmic events after myocardial infarction: signal-averaged electrocardiogram, Holter monitoring and radionucleid ventriculography. J Am Coll Cardiol 1987;9:531-38.

59. El-Sherif N, Ursell SN, Bekheit S, Fontaine J, Turitto G, Henkin R, et al. Prognostic significance of the signal-averaged electrocariogram depends on the time of recording in the post-infarction. Am Heart J 1989;118:256-64.

60. Denniss AR, Richards DA, Cody DV, Russell PA, Young AA, Cooper MJ, et al. Prognostic significance of ventricular tachycardia and fibrillation induced at programmed stimulation and delayed potentials detected on the signal-averaged electrocardiograms of survivors of acute myocardial infarction. Circulation 1986;74:731-45.

61. Pedretti R, Etro MD, Laporta A, Sarzi Braga S, Caru B. Prediction of late arrhythmic events after acute myocardial infarction from combined use of noninvasive prognostic variables and inducibility of sustained monomorphic ventricular tachycardia. Am J Cardiol 1993;71:1131-41.

62. Zimmermann M, Sentici A, Adamec R, Metzger J, Mermillod B, Rutishauser W. Long-term prognostic significance of ventricular late potentials after a first acute myocardial infarction. Am Heart J 1997;134:1019-28.

63. Savard P, Rouleau JL, Ferguson J, Poitras N, Morel P, Davies RF, Stewart DJ, Talajic M, Gardner M, Dupuis R, Lauzon C, Sussex B, Potvin L, Warnica W. Risk Stratification after myocardial infarction using signal-averaged electrocardiographic criteria adjusted for sex, age, and myocardial infarction location. Circulation 1997;96:202-13.

64. Gang ES, Lew AS, Hong M, Wang FZ, Sibst CA, Peter P. Decreased incidence of late potentials after successful thrombolytic therapy of myocardial infarction. N Engl J Med 1989;231:712-16.

65. McClements BM, Adgey AAJ. Value of signal-averaged electrocardiography, radionuclide ventriculography, Holter monitoring and clinical variables for prediction of arrhythmic events in survivors of acute myocardial infarction in the thrombolytic era. J Am Coll Cardiol 1993;21:1419-27.

66. Zimmerman M, Adamee R, Ciaroni S, Malbois F, Tieche R. Reduction in the frequency of ventricular late potentials after acute myocardial infarction by early thrombolytic therapy. Am J Cardiol 1991;67:697-703.

67. Vatterot PJ, Hammill SC, Bailey WR, Wiltgen CM, Gersh BJ. Late potentials on signal-averaged electrocardiograms and patency of the infarct-related artery in survivors of acute myocardial infarction. J Am Coll Cardiol 1991;17:330-37.

68. de Chillou C, Sadoul N, Bizeau O, Feldmann L, Gazakuré E, Ismaïl M, et al. Prognostic value of thrombolysis, coronary artery patency, signal-averaged electrocardiography, left ventricular ejection fraction, and Holter electrocardiographic monitoring for life-threatening ventricular arrhythmias after a first acute myocardial infarction. Am J Cardiol 1997;80:852-58.

69. Bruna C, Rossetti G, Vado A, Racca E, Steffenino G, Dellavalle A, Ribichini F, Ferrero V, Menardi E, Uslenghi E. Prevalence of late potentials after myocardial infarction treated with systemic thrombolysis or primary percutaneous transluminal coronary angioplasty. G Ital Cardiol 1998;28:3-11.

70. Karam G, Golmard J, Steg PG. Decreased prevalence of late potentials with mechanical versus thrombolysis-induced reperfusion in acute myocardial infarction. J Am Coll Cardiol 1996;27:1343-48.

71. Lander P, Gomis P, Goyal R, Berbari EJ, Caminal P, Lazzara R, Steinberg JS. Analysis of abnormal intra-QRS potentials: Improved predictive value for arrhythmic events with the signal-averaged electrocardiogram. Circulation 1997;95:1386-93.

72. Roy D, Marchand E, Theroux P, Waters DD, Pelletier GB, Bourassa MG, et al. Programmed ventricular stimulation in survivors of an acute myocardial infarction. Circulation 1985;72:487-94.

73. Gomes JA, Hariman RI, Kang PS, El-Sherif N, Chowdhry I, Lyons Julianne. Programmed electrical stimulation in patients with high-grade ventricular ectopy. Circulation 1984;70:43-51.

74. Buxton AE, Marchlinski FE, Flores BT, Miller JM, Doherty JU, Josephson ME. Nonsustained ventricular tachycardia in patients with coronary artery disease: role of electrophysiologic study. Circulation 1987;75:1178-85.

75. Wilber DJ, Olshansky B, Moran JF, Scanlon PJ. Electrophysiological testing and nonsustained ventricular tachycardia: Use and limitations in patients with coronary artery disease and impaired ventricular function. Circulation 1990;82:350-58.

76. Klein RC, Machell C. Use of electrophysiologic testing in patients with nonsustained ventricular tachycardia: Prognostic and therapeutic implications. J Am Coll Cardiol 1989;14:155-61.

77. Manolis AS, Estes NAM. Value of programmed stimulation in the evalution and management of patients with nonsustained ventricular tachycardia associated with coronary artery disease. Am J Cardiol 1990;65:201-5.

78. Brembilla-Perrot B, Terrier de la Chaise A, Briançon S, Suty-Selton C, Beurrier D, Martin N, et al. Programmed ventricular stimulation in survivors of acute myocardial infarction: long-term follow-up. Int J Cardiol 1995;49:55-65.

79. Buxton AE, Fisher JD, Josephson ME, Lee KL, Pryor DB, Prystowsky EN, et al. Prevention of sudden death in patients with coronary artery disease: The Multicenter Unsustained Tachycardia Trial (MUSTT). Prog Cardiovasc Dis 1993;36:215-26.

80. Chiou CW, Zipes DP. Selective vagal denervation of the atria eliminates heart rate variability and baroreflex sensitivity while preserving ventricular innervation. Circulation 1998;98:360-68.

81. Berger RD, Akselrod S, Gordon D, Cohen RJ. An efficient algorithm for spectral analysis of heart rate variability. IEEE Trans Biomed Eng 1986;33:900-4.

82. Kamath MV, Ghista DN, Fallen EL, Fitchett D, Miller D, McKelvie R. Heart rate variability power spectrogram as a potential noninvasive signature of cardiac regulatory system response, mechanisms, and disorders. Heart Vessels 1987;3:33-41.

83. Wolf MW, Varigos GA, Hunt D, Sloman JG. Sinus arrhythmia in acute myocardial infarction. Med J Australia 1978;2:52-53.

84. Bigger JT, Fleiss JL, Steinman RC, Rolnitzky LM, Kleiger RE, Rottman JN. Frequence domain measures of heart period variability and mortality after myocardial infarction. Circulation 1992;85:164-71.

85. Zuanetti G, Neilson JM, Latini R, Santoro E, Maggioni AP, Ewing DJ. Prognostic significance of heart rate variability in post-myocardial infarction patients in the fibrinolytic era. The GISSI-2 results. Gruppo Italiano per lo Studio della Sopravvivenza nell' Infarto Miocardico. Circulation 1996;94:432-36.

86. Billman GE, Schwartz PJ, Stone HL. Baroreceptor reflex control of heart rate: A Predictor of sudden cardiac death. Circulation 1982;66:874-80.

87. La Rovere MT, Specchia G, Mortara A, Schwartz PJ. Baroreflex sensitivity, clinical correlates and cardiovascular mortality among patients with a first myocardial infarction. Circulation 1988;78:816-24.

88. La Rovere MT, Bigger JT Jr, Marcus FI, Mortara A, Schwartz PJ. Baroreflex sensitivity and heart-rate variability in prediction of total cardiac mortality after myocardial infarction. ATRAMI (Autonomic Tone and Reflexes After Myocardial Infarction) Investigators. Lancet 1998;351:478-84.

89. Bigger JT Jr, La Rovere MT, Steinman RC, Fleiss JL, Rottman JN, Rolnitsky LM et al. Comparison of baroreflex sensitivity and heart period variability after myocardial infarction. J Am Coll Cardiol 1989;14:1511-18.

90. Schwartz PJ, Snebold NG, Brown AM. Effects of unilateral cardiac sympathetic denervation on the ventricular fibrillation threshold. Am J Cardiol 1976;37:1034-40.

91. Barr CS, Naas A, Freeman M, Lang CC, Struthers AD. QT dispersion and sudden unexpected death in chronic heart failure. Lancet 1994;343:327-29.

92. Statters DJ, Malik M, Ward DE, Camm AJ. QT dispersion: problems of methodology and clinical significance. J Cardiovascular Electrophysiol 1994;5:672-85.

93. de Bruyne MC, Hoes AW, Kors JA, Hofman A, van Bemmel JH, Grobee DE. QTc dispersion predicts cardiac mortality in the elderly: The Rotterdam study. Circulation 1998;97:467-72.

94. Perkiömäki JS, Koistinen MJ, Yli-Mayry S, Huikuri HV. Dispersion of the QT interval in patients with and without susceptibility to ventricular tachyarrhythmias after previous myocardial infarction. J Am Coll Cardiol. 1995;26:174-79.

95. Glancy JM, Garratt CJ, Woods KL, de Bono DP. QT dispersion and mortality after myocardial infarction. Lancet 1995;345:945-48.

96. Oikarinene L, Viitasalo M, Toivonene L. Dispersion of the QT Interval in Postmyocardial Infarction Patients Presenting with Ventricular Tachycardia or with Ventricular Fibrillation. Am J Cardiol 1998;81:694-97.

97. Berger RD, Kasper EK, Baughman KL, Marban E, Calkins H, Tomaselli GF. Beta to beat QT interval variability. Novel evidence for repolarization lability in ischemic and nonischemic dilated cardiomyopathy. Circulation 1997;96:1557-65.

98. Sarma JSM, Singh N, Schoenbaum MP, Venkataraman K, Singh BN. Circadian and power spectral changes of RR and QT intervals during treatment of patients with angina pectoris with nadolol providing evidence for differential autonomic modulation of heart rate and ventricular repolarization. Am J Cardiol 1997;74:131-36.

99. Rosenbaum DS, Jackson LE, Smith JN, Garan H, Ruskin JN, Cohen RJ. Electrical alternans and vulnerability to ventricular arrhythmias. N Engl J Med 1994;330:235-41.

100. Estes NAM, Michaud G, Zipes DP, El-Sherif N, Venditti FJ, Rosenbaum DS, Albrecht P, Wang PJ, Cohen RJ. Electrical alternans during rest and exercise as predictors of vulnerability to ventricular arrhythmias. Am J Cardiol 1997;80:1314-18.

101. Gomes JA, Winters SL, Stewart D, Horowitz S, Milner M, Barreca P. A new non-invasive index to predict sustained ventricular tachycardia and sudden death in the first year after myocardial infarction based on signal averaged electrocardiogram, radionuclide ejection fraction and Holter monitoring. J Am Coll Cardiol 1995;25:908-14.

102. Reinhardt L, Mäkijärvi M, Fetsch T, Schulte G, Sierra G, Martinez-Rubio A, et al. Noninvasive risk modeling after myocardial infarction. Am J Cardiol 1996;78:627-32.

103. Hartikainen JEK, Malik M, Staunton A, Poloniecki J, Camm AJ. Distinction between arrhythmic and non-arrhythmic death after acute myocardial infarction based on heart rate variability, signal-averaged electrocardiogram, ventricular arrhythmias and left ventricular ejection fraction. J Am Coll Cardiol 1996;28:296-04.

104. Myerburg RJ, Kessler KM, Kimura S et al. Life threatening ventricular arrhythmias: the link between epidemiology and pathophysiology. In: Zipes DP, Jalife J, editors. Cardiac Electrophysiology: from cell to bedside. 2nd ed. Philadelphia: WB Saunders; 1995. p.723-31.

105. Nichol G, Hallstrom AP, Ornato JP, Riegel B, Stiell IG, Valenzuela T, et al. Potential cost-effectiveness of public access defibrillation in the United States. Circulation 1998;97:1315-20

106. Newby KH, Thompson T, Stebbins A, Topol EJ, Califf RM, Natale A. Sustained ventricular arrhythmias in patients receiving thrombolytic therapy: incidence and outcomes. The GUSTO Investigators. Circulation 1998;98:2567-73

107. Schmidt G, Malik M, Barthel P, Schneider R, Ulm K, Rolnitsky L, Camm AJ, Bigger JT JR, Schomig A. Heart-rate turbulence after ventricular premature beats as a predictor of mortality after acute myocardial infarction. Lancet 1999;353:1390-96