Limitations of Ejection Fraction for Prediction of Sudden Death Risk in Patients With Coronary Artery Disease

Alfred E. Buxton, MD, FACC; Kerry L. Lee, PhD; Gail E. Hafley, MSl; Luis A. Pires, MD; John D. Fisher, MD; Michael R. Gold, MD; Mark E. Josephson, MD; Michael H. Lehmann, MD; Eric N. Prystowsky, MD; for the MUSTT Investigators

J Am Coll Cardiol.  2007;50(12) ©2007 Elsevier Science, Inc.
Posted 09/17/2007

Abstract and Introduction


Objectives: We determined the contribution of multiple variables to predict arrhythmic death and total mortality risk in patients with coronary disease and left ventricular dysfunction. We then constructed an algorithm to predict risk of mortality and sudden death.
Background: Many factors in addition to ejection fraction (EF) influence the prognosis of patients with coronary disease. However, there are few tools to use this information to guide clinical decisions.
Methods: We evaluated the relationship between 25 variables and total mortality and arrhythmic death in 674 patients enrolled in the MUSTT (Multicenter Unsustained Tachycardia Trial) study that did not receive antiarrhythmic therapy. We then constructed risk-stratification algorithms to weight the prognostic impact of each variable on arrhythmic death and total mortality risk.
Results: The variables having the greatest prognostic impact in multivariable analysis were functional class, history of heart failure, nonsustained ventricular tachycardia not related to bypass surgery, EF, age, left ventricular conduction abnormalities, inducible sustained ventricular tachycardia, enrollment as an inpatient, and atrial fibrillation. The model demonstrates that patients whose only risk factor is EF ≤30% have a predicted 2-year arrhythmic death risk <5%.
Conclusions: Multiple variables influence arrhythmic death and total mortality risk. Patients with EF ≤30% but no other risk factor have low predicted mortality risk. Patients with EF >30% and other risk factors may have higher mortality and a higher risk of sudden death than some patients with EF ≤30%. Thus, risk of sudden death in patients with coronary disease depends on multiple variables in addition to EF.


Sudden cardiac death accounts for 450,000 deaths yearly in the U.S..[1] Furthermore, the proportion of all cardiac deaths accounted for by sudden death is increasing.[1] Multiple clinical trials completed over the past decade have documented the effectiveness of the implantable cardioverter-defibrillator (ICD) to reduce the risk of sudden death and overall mortality in patients at high risk for sudden death. However, no recent study to date has examined the most effective means of deploying this technology to make it available to the greatest number of people in a cost-effective manner.

A number of variables have been demonstrated to identify patients at increased risk for sudden death. Recent trials have focused on left ventricular ejection fraction (EF), because of its demonstrated association with mortality risk in patients with recent myocardial infarction. However, EF lacks sensitivity for prediction of sudden death; less than 50% of patients with prior infarction who die suddenly have EF ≤30%.[2-4] Additionally, many factors besides EF affect the prognosis of patients with coronary artery disease, and several lines of evidence suggest that reduced EF is a risk factor only when it exists in combination with other risk factors.[5,6] Given earlier studies pointing to factors other than EF that influence prognosis after MI, the purpose of the present study is to evaluate the relative importance of multiple factors and to compare their relative contribution to risk of arrhythmic death as well as total mortality using the MUSTT (Multicenter Unsustained Tachycardia Trial) database.[7,8] We then constructed a risk stratification tool that could be used in clinical practice. We demonstrate that use of such a model may enable more precise risk stratification of patients with coronary disease considered for ICD implantation for primary prevention of sudden death.


The MUSTT study was a randomized, controlled study whose primary aim was to test the ability of electrophysiologically guided therapy to reduce the risk of arrhythmic death in patients with documented coronary artery disease, left ventricular EF ≤40%, and spontaneous nonsustained ventricular tachycardia (NSVT). All patients underwent a standardized electrophysiologic test.[9] Patients with inducible sustained ventricular tachycardia (VT) (monomorphic VT induced by 1 to 3 ventricular extrastimuli or sustained polymorphic VT induced by 1 or 2 extrastimuli) at the baseline electrophysiologic study were randomized equally into 2 groups. One-half of the patients with inducible ventricular tachyarrhythmias were randomized to electrophysiologically guided therapy, which consisted of serial antiarrhythmic drug trials. Patients whose tachycardia remained inducible at electrophysiologic testing after treatment with at least 1 antiarrhythmic drug were advised to undergo defibrillator implantation. These patients who received pharmacologic antiarrhythmic therapy or implanted defibrillators are excluded from the present analysis. The remaining one-half of patients with inducible tachycardias were randomized to the control group and received no antiarrhythmic therapy. Patients without either inducible sustained monomorphic VT or sustained polymorphic VT or fibrillation induced by 1 or 2 ventricular extrastimuli were followed prospectively in a registry. The latter patients were not given antiarrhythmic therapy. The protocol encouraged administration of beta-adrenergic blocking agents and angiotensin-converting enzyme inhibitors to all patients. In the present study, we examined the effect of multiple parameters on the outcome of patients followed prospectively in the registry and the patients with inducible tachycardia randomized to the control group without antiarrhythmic treatment.

Study Population

A total of 2,202 patients were enrolled in the study between 1990 and 1996. One thousand four-hundred thirty-five (65%) had no inducible sustained VT and were followed in the registry. At the time of hospital discharge after enrollment in the trial, 35% of the registry patients were receiving beta-blockers and 66% were receiving angiotensin-converting enzyme inhibitors. Seven hundred sixty-seven patients (35%) had inducible sustained VT. Of these, 704 (92%) agreed to be randomized. The 63 patients who refused randomization were followed on the same schedule with the registry patients. Of the 704 patients with inducible ventricular tachyarrhythmias who agreed to randomization, 353 were assigned to the control (nonantiarrhythmic therapy) arm. Although the importance of heart failure as a risk factor for both total mortality and sudden death is clear today, when the trial was initiated in 1990 this relation was not as well appreciated. We did not collect data regarding history of heart failure or New York Heart Association (NYHA) functional class for the first half of the trial. As a result, complete data for all prognostic variables, including descriptors of heart failure, were available for 674 patients that did not receive antiarrhythmic therapy (Fig. 1). These patients constitute the data set for analysis of total mortality. The cause of death could not be ascertained with certainty in 4 patients. Thus, 670 patients constituted the dataset for the evaluation of arrhythmic death/cardiac arrest. The characteristics of these patients did not differ significantly from those of the entire study population.

Figure 1. 


We evaluated 25 baseline variables for their relationship with the risk of total mortality or arrhythmic death/cardiac arrest ( Table 1 ). The variables included demographics, clinical history, variables derived from the enrolling 12-lead electrocardiogram, characteristics of spontaneous NSVT, characteristics of a prior myocardial infarction, results of the baseline electrophysiologic test, EF, and the extent of coronary disease. It should be noted that the qualifying NSVT for entry into the trial had to occur within 6 months of enrollment and be documented 4 or more days after the most recent myocardial infarction or revascularization procedure (coronary artery bypass graft surgery [CABG] or percutaneous revascularization).[7,8] We have previously demonstrated that patients whose qualifying NSVT was discovered within 10 days of CABG had significantly lower mortality than patients who had never had CABG or whose NSVT was discovered more than 10 days after CABG.[10] Another previous analysis demonstrated that patients who were enrolled while they were inpatients had significantly higher mortality than those enrolled as outpatients.[11]


Arrhythmic Death. Witnessed instantaneous death, unwitnessed death in a patient who was stable in usual state of health when last seen (most often persons found dead in bed in the morning), deaths due to incessant tachycardia, or sequelae of cardiac arrest. Deaths of patients with end-stage heart failure or cardiogenic shock were not classified as arrhythmic.

Cardiac Arrest. Sudden loss of consciousness requiring direct current countershock to restore consciousness or a stable blood pressure and rhythm.

Left Bundle Branch Block (LBBB). A QRS duration of ≥0.12 s; delayed onset of intrinsicoid deflection in lead 1, V5, and V6 ≥0.05 s; broad monophasic usually notched R waves in lead 1, V5, and V6; and rS or QS complexes in right precordial leads.

Intraventricular Conduction Delay (IVCD). A QRS duration of ≥0.11 s but morphology different from LBBB or right bundle branch block.

Statistical Methods

The Cox proportional hazards regression model was used to assess the relationship of each baseline clinical variable (both individually and jointly) with the time until the occurrence of: 1) total mortality; and 2) arrhythmic death or cardiac arrest.[12] For continuous variables, we examined the shape and strength of their relationship with each of the 2 end points through use of a flexible model-fitting approach involving cubic spline functions (cubic polynomials).[13-17] These functions were graphically and statistically examined to assess the assumption of this regression model that patient characteristics are linearly related to the log of the hazard ratio. Where relationships were nonlinear, their shape was characterized either through a transformation to achieve linearity or using spline functions.[17,18] Determining how variables should be modeled was an important step in characterizing the prognostic relationships and identifying which variables were most strongly related to the 2 end points. The ability of the prognostic models to discriminate among patients with respect to their length of survival was characterized using a generalized c-index, an extension to survival analysis of the c-index (area under the receiver-operating characteristic curve) frequently used with a dichotomous end point.[19] After identifying the significant predictor variables from the multivariable modeling process, weights for each factor were derived from the Cox model regression coefficients to develop a prognostic score for each end point.[20] The possible values of the scores were then translated into estimates of: 1) the probability of dying; or 2) the probability of having an arrhythmic event within 2 years of enrollment. The relationships between the prognostic scores and the respective outcome probabilities were then graphically presented. Confidence intervals for the prognostic estimates were generated using bootstrapping techniques. Our goal in these analyses was to develop predictive models for each end point that would be relatively simple to use in clinical practice yet provide adequate predictive accuracy in the assessment of risk.


The median duration of follow-up was 39 months in the trial. Over the course of the trial, 241 of 674 patients in the study population died; 130 experienced arrhythmic death or cardiac arrest. Of 200 patients with inducible VT, 84 died and 53 patients experienced arrhythmic death or cardiac arrest. In contrast, of 474 patients without inducible VT, 157 died and 77 experienced arrhythmic death or cardiac arrest. The Kaplan-Meier 2-year total mortality rate was 22% for the entire study population. The Kaplan-Meier 2-year rate of arrhythmic death or cardiac arrest was 14%.

The factors having statistically significant associations in multivariable analysis with the end point of total mortality included EF, the presence of LBBB or nonspecific IVCD (these 2 variables were grouped together), NYHA functional class at the time of enrollment, inducible VT, age, prior CABG, atrial fibrillation at the time of enrollment, and history of heart failure ( Table 2 ). The index of discrimination (c-index) for this model was 0.78.

The factors having statistically significant associations in multivariable analysis with the end point of arrhythmic death/cardiac arrest included inducible VT, history of heart failure, patients enrolled while inhospital, EF, NSVT not discovered within 10 days after CABG, LBBB, or IVCD ( Table 3 ). Age and atrial fibrillation were not statistically significant predictors of arrhythmic death/cardiac arrest. As expected, the c-index for the end point of arrhythmic death/cardiac arrest was lower (0.70) than for total mortality, reflecting the fact that it is more difficult to predict 1 specific mode of death (arrhythmic) than to predict overall mortality.

A risk stratification algorithm was then constructed and weights assigned to each variable for the end points of total mortality and arrhythmic death/cardiac arrest ( Table 4 and Table 5 ). An individual patient's score for mortality or arrhythmic death risk was computed by ascertaining which variables characterize that individual and summing the corresponding points in Table 4 and Table 5 . This total score was then entered into the x-axis of Figure 2, which shows the curves depicting the 2-year event rates for total mortality and for arrhythmic death/cardiac arrest for the range of possible scores for patients in whom this model was developed. The score was then transposed onto the curve for the end point of interest (total mortality or arrhythmic death/cardiac arrest) and the corresponding predicted 2-year event rate read on the y-axis of the graph.

Figure 2. 


Examples of application of the risk-stratification algorithm appear in Figures 3 to 5. In example A (Fig. 3), a 60-year-old patient whose only risk factor is an EF of 25%, the predicted 2-year total mortality and arrhythmic death risks are quite low (5% and 2%, respectively) even though the EF is <30%.

Figure 3. 


Figure 4. 


Figure 5. 


In contrast, the patient depicted in example B (Fig. 4), who has multiple risk factors in addition to EF ≤30%, has much higher risks of both total mortality and arrhythmic death. Note that if this patient had inducible VT, the arrhythmic death risk is almost as high as the total mortality risk.

In general, the risks of both arrhythmic death and total mortality are lower for patients whose EF is >30%. However, example C (Fig. 5) demonstrates that some patients whose EF is >30% may be at similar or higher risk than certain patients whose EF is <30%. For example, in this case of a hypothetical patient with an EF of 35%, the total mortality and arrhythmic death risks are virtually identical to those of the patient in the first part of example B (Fig. 4) whose EF was 25%.

The execution and implications of this approach to risk stratification are influenced by the relative prevalence of various risk factors. Patients having an EF ≤30% had a higher prevalence of associated risk factors, such as electrocardiographic conduction abnormalities and symptomatic heart failure ( Table 6 ). However, Table 6 demonstrates that a significant minority (25%) of patients whose EF was 30% or less did not have these additional risk factors.


This analysis of patients enrolled in the MUSTT study who received neither pharmacologic antiarrhythmic therapy nor an implanted defibrillator demonstrates that multiple factors influence mortality of patients with chronic coronary artery disease. In addition, this study illustrates that much prognostic information can be gained from easily determined historical factors. We have shown that although EF is an important risk predictor, in this study population several other variables carry similar prognostic significance. Furthermore, the present study demonstrates the potential danger of focusing efforts to reduce risk of sudden death only on patients with EF ≤30%. We demonstrate that depending on the presence of other risk factors, patients with EF 30% to 40% may have total mortality and sudden death risks that exceed those of some patients with EF ≤30%.

The algorithm we developed does not represent the "ultimate" risk-stratification tool. Rather, it serves as an example of the potential utility of this approach to risk stratification. The algorithm developed in this study, or a similar one, could be applied easily to risk-stratify patients having the characteristics of those enrolled in the MUSTT study (and should only be applied to such patients). Of note, except for the electrophysiologic study, the variables in these prediction models are noninvasive and easily determined in the office setting without any special equipment.

The variables that were associated with highest risk for total mortality as well as arrhythmic death are not surprising. The association between inpatient (vs. outpatient) status and higher mortality risk was demonstrated for patients experiencing cardiac arrest in the AVID (Antiarrhythmics Versus Implantable Defibrillators) trial.[21] Older age would be expected to increase total mortality risk but not necessarily risk for arrhythmic death. Conversely, the independent prognostic significance of inducible sustained VT for prediction of arrhythmic death has been demonstrated and is expected.[22]

Application of this model demonstrates that patients whose only risk factor is having an EF of 30% or less have a predicted 2-year total mortality risk of approximately 5%. This rate is considerably lower than the observed 2-year total mortality rate of 22% for control patients enrolled in the MADIT (Multicenter Automatic Defibrillator Implantation Trial)-II study.[23] This difference in risk is especially striking when one considers 2 additional differences between these 2 studies. First, patients enrolled in the MUSTT study were required to have nonsustained VT, which is associated with higher mortality after myocardial infarction,[9] whereas patients enrolled in the MADIT-II study were not required to have nonsustained VT documented. Second, only one-third of patients in the MUSTT study who did not have inducible sustained VT received beta-blocking agents versus 70% of patients enrolled in the MADIT-II study (because MADIT-II was conducted in the late 1990s, when beta-blockade was more widely accepted in this patient population). It seems likely that if the MUSTT population were treated with beta-blocking agents at rates consistent with current practice, the total mortality and arrhythmic death/cardiac arrest rates would have been even lower than those observed.

We believe the explanation for the much higher mortality observed in the control arm of the MADIT-II study compared with that of patients enrolled in the MUSTT study matching the entry requirements of the MADIT-II (chronic coronary artery disease and EF ≤30%, but having no other variables from the current model associated with increased mortality) lies in the characteristics of patients enrolled in the MADIT-II study. Sixty-one percent of patients enrolled in the MADIT-II study had symptomatic heart failure (NYHA functional class ≥2), and 44% had LBBB or nonspecific IVCD.[23] The influence of these patient characteristics on the observed 2-year mortality of 22% in the MADIT-II control patients is substantiated by comparison of this event rate with that observed in the SCD-HeFT (Sudden Cardiac Death in Heart Failure Trial).[24] In that trial, which required patients to have symptomatic heart failure as well as EF ≤35%, patients with coronary disease as the cause of heart failure randomized to placebo had a 2-year mortality of 18%, similar to the mortality observed in MADIT-II patients. Note that 41% of SCD-HeFT patients with ischemic disease had the additional risk factor of QRS duration ≥120 ms, similar to the MADIT-II study population.

Study Limitations

This model requires validation by a prospective trial. The use of this model must be restricted to patients having documented coronary artery disease, left ventricular EF of 40% or less, and asymptomatic spontaneous nonsustained VT. The application of this model is also dependent on performing an electrophysiologic study to determine if sustained VT is inducible. This algorithm should not be applied to patients with ventricular dysfunction due to noncoronary disease. It is also possible that use of other noninvasive tests, such as measurement of T-wave alternans or the signal-averaged electrocardiogram, might improve the performance of this model. This study should not be construed to be a test of defibrillator efficacy. However, the intelligent use of defibrillators demands an understanding of the risk of both sudden and nonsudden death risks tailored to individual patients. This model demonstrates both the complexity and the feasibility of such risk modeling.


In summary, it is accepted that the ICD is the best therapy currently available to prevent sudden death in high-risk patients. However, recent clinical trials, which demonstrated the spectrum of efficacy of the ICD, were not designed to evaluate optimal methods for risk stratification of myocardial infarction survivors. Given the expense and risks of ICDs, it is logical to search for methods of using this technology in the most cost-effective manner. The present analysis presents one approach to solving this problem. Our model demonstrates that multiple variables in addition to left ventricular dysfunction (reflected by EF) influence mortality of patients with coronary artery disease. We have demonstrated that consideration of multiple risk factors has the potential to provide more accurate prediction for risk of sudden death as well as total mortality. As a result, the model identifies a population of patients that meets current guidelines for prophylactic ICD implantation in patients with coronary disease but is unlikely to derive a significant improvement in 2-year survival with the ICD.

Table 1. Univariate Prognostic Relationships

Table 1: Univariate Prognostic Relationships


Table 2. Multivariable Relationships With Total Mortality

Table 2: Multivariable Relationships With Total Mortality


Table 3. Multivariable Relationships With Arrhythmic Death or Cardiac Arrest

Table 3: Multivariable Relationships With Arrhythmic Death or Cardiac Arrest


Table 4. Calculation of Total Mortality Score

Table 4: Calculation of Total Mortality Score


Table 5. Calculation of Arrhythmic Death/Cardiac Arrest Score

Table 5: Calculation of Arrhythmic Death/Cardiac Arrest Score


Table 6. Prevalence of Additional Risk Factors in Relation to Ejection Fraction

Table 6: Prevalence of Additional Risk Factors in Relation to Ejection Fraction



  1. Zheng Z-J, Croft JB, Giles WH, Mensah GA. Sudden cardiac death in the United States, 1989 to 1998 Circulation 2001;104:2158-2163.
  2. de Vreede-Swagemakers JJ, Gorgels AP, Dubois-Arbouw WI, et al. Out-of-hospital cardiac arrest in the 1990's: a population-based study in the Maastricht area on incidence, characteristics and survival J Am Coll Cardiol 1997;30:1500-1505.
  3. Gorgels APM, Gijsbers C, de Vreede-Swagemakers J, Lousberg A, Wellens HJJ. Out-of-hospital cardiac arrest—the relevance of heart failureThe Maastricht Circulatory Arrest Registry. Eur Heart J 2003;24:1204-1209.
  4. Stecker EC, Vickers C, Waltz J, et al. Population-based analysis of sudden cardiac death with and without left ventricular systolic dysfunctionTwo-year findings from the Oregon Sudden Unexpected Death Study. J Am Coll Cardiol 2006;47:1161-1166.
  5. Kuchar DL, Thorburn CW, Sammel NL. Prediction of serious arrhythmic events after myocardial infarction: signal-averaged electrocardiogram, Holter monitoring and radionuclide ventriculography J Am Coll Cardiol 1987;9:531-538.
  6. La Rovere MT, Pinna GD, Hohnloser SH, et al. Baroreflex sensitivity and heart rate variability in the identification of patients at risk for life-threatening arrhythmias: implications for clinical trials Circulation 2001;103:2072-2077.
  7. Buxton AE, Lee KL, Fisher JD, Josephson ME, Prystowsky EN, Hafley G, Multicenter Unsustained Tachycardia Trial Investigators A randomized study of the prevention of sudden death in patients with coronary artery disease N Engl J Med 1999;341:1882-1890.
  8. Buxton AE, Lee KL, DiCarlo L, et al. Multicenter Unsustained Tachycardia Trial Investigators Electrophysiologic testing to identify patients with coronary artery disease who are at risk for sudden death N Engl J Med 2000;342:1937-1945.
  9. Buxton AE, Fisher JD, Josephson ME, et al. Prevention of sudden death in patients with coronary artery disease: the Multicenter Unsustained Tachycardia Trial (MUSTT) Prog Cardiovasc Dis 1993;36:215-226.
  10. Pires L, Hafley G, Lee K, et al. Multicenter Unsustained Tachycardia Trial Investigators Significance of nonsustained ventricular tachycardia identified postoperatively after coronary bypass surgery in patients with left ventricular dysfunction J Cardiovasc Electrophysiol 2002;13:757-763.
  11. Pires LA, Lehmann MH, Buxton AE, Hafley G, Lee KL, Multicenter Unsustained Tachycardia Trial Investigators Differences in inducibility and prognosis of in-hospital versus out-of-hospital identified nonsustained ventricular tachycardia in patients with coronary artery disease: clinical and trial design implications J Am Coll Cardiol 2001;38:1156-1162.
  12. Cox DR. Regression models and life-tables (with discussion) J R Statist Soc B 1972;34:187-220.
  13. Harrell Jr. FE, Polluck BG, Lee KL. Graphical methods for the analysis of survival data Proceedings of the Twelfth Annual SAS Users Group International Conference. Cary, NC: SAS Institute; 1987. pp. 1107-1115.
  14. Smith PL. Splines as a useful and convenient statistical tool Am Stat 1979;33:57-62.
  15. Stone CJ, Koo C. Additive splines in statistics Proceedings of the Statistical Computing Section. Alexandria, VA: American Statistical Association; 1985. pp. 45-48.
  16. Devlin TF, Weeks BJ. Spline functions for logistic regression modeling Proceedings of the Eleventh Annual SAS Users Group International Conference. Cary, NC: SAS Institute; 1986. pp. 646-651.
  17. Harrell Jr. FE, Lee KL, Pollock BG. Regression models in clinical studies: determining relationships between predictors and response J Natl Cancer Inst 1988;80:1198-1202.
  18. Lee KL, Woodlief LH, Topol EJ, et al. Predictors of 30-day mortality in the era of reperfusion for acute myocardial infarction—results from an international trial of 41 021 patients Circulation 1995;91:1659-1668.
  19. Harrell FE, Califf RM, Pryor DB, Lee KL, Rosati RA. Evaluating the yield of medical tests JAMA 1982;247:2543-2546.
  20. Harrell FE. Regression Modeling Strategies With Applications to Linear Models, Logistic Regression, and Survival AnalysisNew York, NY: Springer; 2001.
  21. Epstein AE, Powell J, Yao Q, et al. AVID Investigators In-hospital versus out-of-hospital presentation of life-threatening ventricular arrhythmias predicts survivalResults from the AVID Registry. J Am Coll Cardiol 1999;34:1111-1116.
  22. Buxton A, Hafley G, Lee K, et al. MUSTT Investigators Relation of ejection fraction and inducible ventricular tachycardia to mode of death in patients with coronary artery disease Circulation 2002;106:2466-2472.
  23. Moss AJ, Zareba W, Hall WJ, et al. Multicenter Automatic Defibrillator Implantation Trial II Investigators Prophylactic implantation of a defibrillator in patients with myocardial infarction and reduced ejection fraction N Engl J Med 2002;346:877-883.
  24. Bardy GH, Lee KL, Mark DB, et al. Amiodarone or an implantable cardioverter-defibrillator for congestive heart failure N Engl J Med 2005;352:225-237.
Funding Information

Supported by grants UO1 HL45700 and UO1 HL45726 from the National Heart, Lung, and Blood Institute, National Institutes of Health, and by grants from C.R. Bard, Berlex Laboratories, Boehringer-Ingelheim Pharmaceuticals, Cardiac Pacemakers/Guidant, Knoll Pharmaceutical Co., Medtronic, Searle Pharmaceutical, Ventritex, and Wyeth-Ayerst Laboratories.

Abbreviation Notes

EF = ejection fraction; ICD = implantable cardioverter-defibrillator; IVCD = intraventricular conduction delay; LBBB = left bundle branch block; NSVT = nonsustained ventricular tachycardia; VT = ventricular tachycardia

Reprint Address

Dr. Alfred E. Buxton, Cardiology Division, Brown Medical School, 2 Dudley Street, Suite 360, Providence, Rhode Island 02905. Email:

Alfred E. Buxton, MD, FACC,* Kerry L. Lee, PhD, Gail E. Hafley, MSl, Luis A. Pires, MD,& John D. Fisher, MD,§ Michael R. Gold, MD,|| Mark E. Josephson, MD,# Michael H. Lehmann, MD,** Eric N. Prystowsky, MD,†† for the MUSTT Investigators

* Department of Medicine, Cardiology Division, Brown Medical School and Lifespan Academic Medical Center, Providence, Rhode Island
Duke University Medical Center and Duke Clinical Research Institute, Durham, North Carolina
St. John Hospital and Medical Center and Wayne State University School of Medicine, Detroit, Michigan
§ Montefiore Medical Center and Albert Einstein College of Medicine, Bronx, New York
|| Medical University of South Carolina, Charleston, South Carolina
# Beth Israel Deaconess Medical Center, Boston, Massachusetts
** Department of Internal Medicine, University of Michigan Medical Center, Ann Arbor, Michigan
†† Care Group, Indianapolis, Indiana.