Author + information
- Received October 30, 2012
- Revision received December 18, 2012
- Accepted January 8, 2013
- Published online April 9, 2013.
- Martin H. Ruwald, MD⁎,†,⁎ (, )
- Anne-Christine H. Ruwald, MD⁎,†,
- Christian Jons, MD, PhD†,
- Jeffrey Alexis, MD⁎,
- Scott McNitt, MS⁎,
- Wojciech Zareba, MD, PhD⁎ and
- Arthur J. Moss, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Martin H. Ruwald, Heart Research Follow-up Program, Cardiology Division, University of Rochester Medical Center, 265 Crittenden Boulevard, Box 653, Rochester, New York 14642
Objectives This study sought to compare the effects of metoprolol and carvedilol in the MADIT-CRT (Multicenter Automatic Defibrillator Implantation Trial With Cardiac Resynchronization Therapy) study.
Background The impact of beta-blockers in heart failure (HF) patients with devices is uninvestigated.
Methods All patients receiving either metoprolol or carvedilol in the MADIT-CRT study were identified and compared. Time-dependent Cox proportional hazard regression analyses were performed to assess differences in hospitalization for HF or death and ventricular arrhythmias.
Results Hospitalization for HF or death occurred in 30% of the patients on metoprolol and in 23% on carvedilol. Treatment with carvedilol was associated with a significantly decreased risk of hospitalization for HF or death when compared with metoprolol (hazard ratio [HR]: 0.70, [95% confidence interval (CI): 0.57 to 0.87], p = 0.001). This reduction in risk was further attenuated in the subgroup of cardiac resynchronization therapy with implantable cardioverter-defibrillator (CRT-D) patients (HR: 0.61 [95% CI: 0.46 to 0.82], p = 0.001) and CRT-D patients with left bundle branch block (LBBB) (HR: 0.51 [95% CI: 0.35 to 0.76], p < 0.001). Ventricular arrhythmias occurred in 26% and in 22%, respectively, of the patients receiving metoprolol or carvedilol (HR: 0.80 [95% CI: 0.63 to 1.00], p = 0.050). General use of beta-blockers and adherence in this study was high, and a clear dose-dependent relationship was found in carvedilol, but not in metoprolol.
Conclusions In HF patients in New York Heart Association functional class I and II and with wide QRS complexes, carvedilol was associated with a 30% reduction in hospitalizations for HF or death when compared with metoprolol. A novel beneficial and synergistic effect of carvedilol was seen in patients with CRT-D and LBBB. Furthermore, we found a pronounced dose-dependent relationship in carvedilol, but not in metoprolol. (MADIT-CRT: Multicenter Automatic Defibrillator Implantation Trial With Cardiac Resynchronization Therapy; NCT00180271)
Cardiac resynchronization therapy (CRT) has emerged as an important device-based therapy for selected patients with systolic heart failure (HF). Landmark clinical trials have demonstrated the efficacy of CRT in patients with mild or advanced HF symptoms despite optimal pharmacological therapy (1–5). Optimal pharmacological therapy is considered a prerequisite to consideration for CRT (6), and beta-blockers (BBs) in particular have been proven to improve quality of life and reduce mortality in large populations of patients with systolic HF (7–9).
Presently, metoprolol and carvedilol are the BBs most often used in the management of patients with HF, and the choice of drug is left to the discretion of the physician because both drugs have a class IA indication (10,11).
The COMET (Carvedilol or Metoprolol European Trial) investigated the efficacy of metoprolol tartrate and carvedilol in patients with moderate-to-severe HF, showing risk reduction in cardiovascular mortality favoring carvedilol (7). Unfortunately, no comparative randomized study has been done on metoprolol succinate, the metoprolol salt proven in MERIT-HF (Multicenter Automatic Defibrillator Implantation Trial With Cardiac Resynchronization Therapy) (9) to reduce mortality in HF patients, but indications that major differences reside in these 2 widely used BBs remain. Furthermore, numerous studies have shown that HF patients have difficulties reaching optimal BB doses, and therefore, the risk reduction may not be as pronounced in real life as in the randomized controlled trials.
The current information on BB therapy in HF patients with CRT devices is sparse.
Heywood et al. (12) found that patients with implantable cardioverter-defibrillator (ICD) or cardiac resynchronization therapy with an ICD (CRT-D) device were more likely to be treated with target doses of BBs than those without a device and that absence of BBs is associated with poor outcome (13). Furthermore, a small study has retrospectively investigated the effect of carvedilol and metoprolol in CRT patients with HF and New York Heart Association (NYHA) functional class III to IV (14) and found no significant difference in all-cause mortality between the 2 BBs. To our knowledge, no study has investigated the effects of metoprolol versus carvedilol in NYHA functional class I to II HF patients with CRT-D or ICD devices.
The MADIT-CRT study evaluated the effect of CRT-D versus ICD-only in relatively asymptomatic patients (NYHA functional class I to II) with a low ejection fraction and wide QRS complex; BBs were well used in this study (93%).
The aim of this study was to evaluate the effect of metoprolol and carvedilol in HF patients receiving CRT-D or ICD therapy in the MADIT-CRT trial.
The protocol and primary article of the MADIT-CRT study have previously been published (3,15).
The study included 1,820 patients with left ventricular ejection fraction (LVEF) ≤30%, QRS duration ≥130 ms, and ischemic cardiomyopathy NYHA functional class I or II or nonischemic cardiomyopathy NYHA functional class II. Patients were enrolled from 110 centers in Europe, Canada, and the United States and randomized in a 3:2 fashion for CRT-D or ICD devices. Patients were excluded if they had atrial fibrillation at enrollment, whereas prior history of atrial fibrillation was not an exclusion criterion.
Patients had to be on optimal pharmacotherapy in accordance with HF guidelines (6). However, the choice of BBs and other HF therapy was left to the discretion of the implanting and/or treating physician.
The study was conducted in the period from December 22, 2004 through June 24, 2009, at which time the study was stopped by recommendation of the safety monitoring board. Extended follow-up was conducted until September 10, 2010.
The MADIT-CRT study included 1,820 patients with 1,089 patients in the CRT-D arm and 731 patients in the ICD-arm.
Patients who received other BBs than metoprolol or carvedilol or who did not receive any BBs (n = 275) and patients who for some reason never received an ICD or CRT-D (n = 30) were excluded from the baseline characteristics shown in Table 1. Patients who received no BB or received other types of BBs than metoprolol or carvedilol at enrollment but changed to metoprolol or carvedilol during the course of the study were included in a time-dependent fashion. All analyses were on an intention-to-treat basis, and device crossovers were not considered. Sensitivity analyses were conducted on crossovers of ICD or CRT-D devices, which did not change the results.
Medication and doses
All medication, including type of BB and the doses were recorded on patients at baseline and clinical follow-up at 1 month and then at 3-month intervals until the termination of the trial.
The individual dose was calculated as the baseline dose divided by the patient's weight, thereby yielding precise estimates on the dose distribution in quartiles. All accounted doses of metoprolol are metoprolol succinate or recalculated doses from metoprolol tartrate.
Device programming and interrogation
The institutional review board approved the protocol at each participating organization, and each patient provided written informed consent before enrollment. Commercially available devices from Boston Scientific were used in this trial; programming has been described in the pre-specified protocol (15). All devices were interrogated 1 month after enrollment and thereafter every 3 months and after the occurrence of any device therapy. All interrogation discs were sent to an independent core laboratory for final adjudication and interpretation. Device therapy was adjudicated into pre-defined categories and, on the basis of these, categorized as appropriate or inappropriate. The ventricular tachycardia (VT) zone was set at 180 beats/min. VT was defined as ventricular rates up to 250 beats/min; ventricular fibrillation (VF) was defined as ventricular rates faster than 250 beats/min with disorganized ventricular electrograms.
The primary endpoint of the present study was defined as death from any cause or nonfatal HF events, whichever came first. The diagnosis of HF, which required symptoms consistent with congestive HF and responsive to decongestive therapy, was independently adjudicated by a mortality committee unaware of study group assignments. The pre-specified criteria have been described previously (15). The mode of death was adjudicated by the mortality committee with the modified Hinkle-Thaler classification (16). The secondary endpoint was first occurrence of appropriate therapy for VF or VT as defined in the previous text.
Patients were divided into 2 groups based on their use of metoprolol or carvedilol, and their baseline characteristics were compared using the chi-square test or the Wilcoxon rank sum test, where appropriate.
Cox proportional hazard regression was used to compare models with variables for BB therapy type use at baseline with models that incorporated time-dependent covariates for metoprolol or carvedilol. Two models were used, depending on the endpoint. In the HF/death model, adjustment was made for CRT-D treatment, left bundle branch block (LBBB), CRT-D–LBBB treatment interaction, creatinine level, hospitalization in prior year to enrollment, age, ischemic etiology, LVEF, NYHA functional class, and the use of other BBs or no BBs.
In the VT/VF model, we adjusted for CRT-D treatment, LBBB, CRT-D–LBBB treatment interaction, age, LVEF, female sex, non-U.S. centers, systolic blood pressure, prior myocardial infarction, prior non-coronary artery bypass graft (CABG) revascularization, prior ventricular arrhythmias, and the use of other BBs or no BBs.
When specific subgroups were analyzed, these models were modified omitting the variable investigated.
BB therapy was assessed in the multivariate model in a time-dependent manner (i.e., by incorporating in the Cox model variables for each patient that identified the effect of each follow-up period “on” and “off” BB therapy during the trial). The value of time-dependent BB therapy covariates in predicting the risk of various endpoints for the 2 treatment arms was estimated using interaction terms.
Univariate and multivariate Cox proportional hazard regression analyses were performed for the primary endpoint of HF/death and the secondary endpoint of VT/VF. In the multivariate model, we adjusted for relevant variables that met 2 criteria. First, they needed to vary significantly between the 2 BB therapy types at baseline, and second, they were found to be significantly predictive of the endpoint using stepwise selection, with limits for entry into the model at a significance level of 0.05. These potential confounders were adjusted for in the multivariate Cox model.
Additionally, we carried out a propensity score—adjusted Cox proportional hazards regression analysis. We quantified a propensity score for the likelihood of receiving metoprolol at baseline by multivariate logistic regression analysis using the pool of baseline covariates and the stepwise procedure to include only the statistical significant predictors. The propensity score adjustment compensates for clinical decisions affecting the choice of beta-blocker prior to enrollment in the study. The logistic regression analysis found four covariates associated with a higher likelihood of receiving metoprolol at baseline; age above 65 years, ischemic heart failure etiology, prior hospitalization, and higher weight.
Hazard ratios (HRs) with their 95% confidence intervals (CIs) and 2-sided p values are reported. The cumulative probability of HF or death and VT/VF were displayed by the method of Kaplan-Meier using the log-rank test to compare cumulative events. A 2-tailed p value below 0.05 was considered statistically significant.
Analyses were performed using SAS statistical system version 9.3 (SAS Institute, Cary, North Carolina).
The clinical characteristics of the 1,515 patients who received either metoprolol or carvedilol are presented in Table 1. In short, patients receiving metoprolol were more often male, marginally older, had more ischemic cardiomyopathy, were more likely to be enrolled at a non-U.S. center, and had a higher frequency of prior myocardial infarctions, hypertension, and past ventricular arrhythmias than patients receiving carvedilol.
During a mean follow-up period of 3.4 ± 1.1 years, the primary endpoint of hospitalization for HF or death from any cause (whichever came first) occurred in 132 patients (30%) on metoprolol and in 243 patients (23%) on carvedilol. This endpoint included 48 (11%) and 104 (10%) deaths in the 2 groups, respectively. The secondary endpoint of ventricular arrhythmias and VT occurred in 115 patients (26%) and 241 patients (22%) receiving either metoprolol or carvedilol, respectively. Kaplan-Meier estimates of the primary outcome in the 2 study groups are shown in Figure 1.
Primary and secondary endpoints
In the univariate Cox analysis, there was a significant association between the use of carvedilol and a reduction in HF or death compared with metoprolol (HR: 0.72, 95% CI: 0.58 to 0.89, p = 0.002), which was confirmed in the multivariate analysis as shown in Table 2. The relative risk reduction after adjusting for relevant confounders was 30%, which was driven primarily by a reduction in hospitalizations for HF, primarily in the CRT-D–treated part of the study. Univariate analysis in the event of VT/VF revealed a trend in difference between metoprolol and carvedilol (HR: 0.83, 95% CI: 0.66 to 1.03, p = 0.090) as shown in Figure 2. In the multivariate analysis, however (Table 2), a 20% relative risk reduction in VT/VF was found significantly associated with the use of carvedilol.
In the CRT-D arm of the population, the difference in risk of HF/death between the two BBs persisted and was even more pronounced in univariate (HR: 0.68, 95% CI: 0.51 to 0.91, p = 0.010) as well as in multivariate analyses, presented in Table 2. The relationship and direct comparison with the ICD group is depicted in Figure 3. No significant difference was found in the secondary endpoint of VT and VF in the CRT arm of the study, but a trend was evident (HR: 0.74, 95% CI: 0.54 to 1.01, p = 0.054) in univariate as well as multivariate analyses (Table 2).
CRT-D and LBBB
Selecting the subgroup of patients receiving CRT-D treatment who had LBBB, there was a significant beneficial effect in patients receiving carvedilol (HR: 0.58, 95% CI: 0.39 to 0.86, p = 0.006) as compared with metoprolol. Multivariate results in Table 2 show the independent and synergistic effect of carvedilol, CRT-D treatment, and LBBB QRS morphology. Furthermore, when comparing metoprolol and carvedilol in patients in this subgroup, there was a similar significant association between the type of BB and the risk of ventricular arrhythmias, ultimately favoring carvedilol. In Table 2, the overall relationship between outcome and LBBB is shown favoring carvedilol, with a 49% relative reduction in HF/death compared with metoprolol among patients treated with a CRT-D device.
We found no significant difference in the ICD arm of the study on either hospitalization for HF, death, or VT/VF between the 2 groups.
The propensity-scored analyses yielded nearly identical results, as shown in the Online Appendix.
BB adherence and initial doses at baseline
The mean dose of metoprolol at baseline was 66 ± 48 mg, median 50 mg (interquartile range [IQR]: 25 to 100 mg), and the mean dose of carvedilol at baseline was 18 ± 13 mg, median 12.5 mg (IQR: 6.25 to 25 mg), with recalculations to mean dose per body weight of metoprolol of 0.75 mg/kg compared with 0.22 mg/kg of carvedilol (median: 0.60 mg/kg (IQR: 0.38 to 0.98 mg/kg) and 0.18 (IQR: 0.09 to 0.29 mg/kg), in the 2 groups, respectively.
From baseline to first change, the mean increase in dose on metoprolol was 16 ± 28 mg (n = 146), while the mean change in dose on carvedilol was 13 ± 11 mg (n = 947). Only small differences were seen in the BB adherence or changes in doses through the rest of the study, 19 patients changed dosages of metoprolol and 38 patients changed dose of carvedilol. The mean change in dose of metoprolol was −2 ± 39 mg and −4 ± 12 mg in carvedilol, quite evenly distributed in the CRT-D and ICD populations. Overall, 92 patients changed from one type to the other, of which 68 patients changed from metoprolol to carvedilol, and 24 patients changed from carvedilol to metoprolol. These changes are taken into account in the time-dependent analyses. In total, 12% of those on metoprolol used the metoprolol tartrate salt, and these doses have been recalculated to metoprolol succinate equivalents.
Kaplan-Meier estimates of the dose relationship and primary endpoint as selected by the values above the 3rd quartile (0.29 mg/kg and 0.98 mg/kg) in carvedilol and metoprolol are depicted in Figure 4 presenting differences in the outcome. This dose relationship of carvedilol was further substantiated in CRT-D and LBBB patients, again in a synergistic fashion.
To our knowledge this is the first study to investigate and compare metoprolol with carvedilol in a setting of relatively asymptomatic HF patients with low EF and wide QRS receiving either a CRT-D or ICD device.
The primary composite endpoint of hospitalization for HF and death showed a marked event reduction in patients treated with carvedilol, which was primarily driven by a synergistic effect of carvedilol and CRT-D treatment in the population with LBBB QRS morphology. This major difference in hospitalizations for HF has not previously been described in the setting of NYHA functional class I and II patients and the novel association and synergistic effect of CRT-D treatment and carvedilol, primarily in the setting of LBBB patients, requires further attention.
We found a borderline significant 20% relative risk reduction between carvedilol and metoprolol on our secondary endpoint of VT or VF, but further attenuated in the subgroup of LBBB patients and a marked and significant effect in the CRT-D and LBBB subset. The effect of carvedilol, particularly in LBBB and CRT-D–treated patients, thus efficiently reduces the risk of ventricular arrhythmias.
BB adherence and therapy in the MADIT-CRT study
Are the results all about dosage? Mean carvedilol doses in MADIT-CRT were 18 ± 13 mg, mean metoprolol doses were 64 ± 47 mg, which is comparable to real-life doses administered to HF patients in the clinical setting, although somewhat lower than the doses in the previous randomized clinical drug trials. In OPTIMIZE-HF (Organized Program to Initiate Lifesaving Treatment in Hospitalized Patients with Heart Failure) (17), the mean dose of carvedilol was 22 mg and the mean dose of metoprolol was 69 mg in patients who had systolic HF. The COMET trial mean dose was 42 mg and 128 mg in carvedilol and metoprolol (recalculated from metoprolol-tartrate), respectively, whereas in the COPERNICUS (Carvedilol Prospective Randomized Cumulative Survival) trial, the mean dose of carvedilol was 45 mg. Usually a factor 4 is used to convert carvedilol to metoprolol. So, comparing the equivalent mean doses of carvedilol to metoprolol, the drug comparison is reasonable and also in line with achievable doses in clinical practice. However, when converting the doses of metoprolol into doses of carvedilol equivalents, there was a somewhat small difference in mean dosage. The mean carvedilol equivalent in the metoprolol group was 16 ± 12 mg, which was a small, but significant, difference (p = 0.004) compared with carvedilol (18 ± 13 mg). We do not believe this small difference in baseline mean carvedilol equivalents account for the major differences seen.
In our subanalysis, we did not find a decrease in risk associated with a higher dose of metoprolol, whereas there was a clear dose-dependent relationship of carvedilol on HF/death.
The use of BB in MADIT-CRT was high (93% at enrollment), compared with 68% in the COMPANION (Comparison of Medical Therapy, Pacing, and Defibrillation in Chronic Heart Failure) trial, 70% in CARE-HF (Cardiac Resynchronization–Heart Failure) (18), and about 85% in IMPROVE HF (Registry to Improve the Use of Evidence-Based Heart Failure Therapies in the Outpatient Setting) (12). MADIT-CRT patients were eligible when they had been appropriately treated with a BB at a therapeutic dose for the last 3 months, and had been stable for at least 1 month before enrollment. The choice of selective or nonselective BB was left to the physician's discretion. Thus, no optimal target dose was given in the enrollment criteria. In OPTIMIZE-HF, only 17.5% and 7.9% received target doses of carvedilol and metoprolol, respectively (17). In MADIT-CRT, maximum baseline doses of carvedilol (50 mg) and metoprolol (200 mg) were given in 55 (5.1%) and 23 patients (5.3%), respectively.
There remains no definitive evidence of a dose-response relationship between BB therapy and outcomes in a randomized trial, and a recent meta-analysis of 23 BB trials failed to show an association between BB dose and survival benefit in HF (19) and that heart rate reduction in itself yields the survival benefit of BBs in HF. However, a recent study from HF-ACTION (Heart Failure: A Controlled Trial Investigating Outcomes of Exercise Training) showed a linear correlation between up to 50 mg of daily dose of carvedilol in ambulatory patients and significantly lower rates of all-cause death and all-cause hospitalization (20). Our study was not designed to address the relationship of dose benefit; however, it does clearly raise the question of the achievability of currently recommended BB dose targets.
We speculate that low baseline resting heart rates in our study indicate that the patients generally were well titrated at baseline and that the rate may have contributed to a relatively low dose in both metoprolol and carvedilol compared with the BB dosage in clinical trials. The mean heart rate in the MADIT-CRT was 68 beats/min, suggesting a reasonable amount of beta blockage, and importantly, no difference was found between the 2 groups of carvedilol and metoprolol.
So what causes patients on metoprolol to be hospitalized more often for HF than patients on carvedilol? It is not evident from shocks given for appropriate therapy; however, a borderline significant trend supported less VT/VF in the patients receiving carvedilol. In the COMET trial, the primary endpoint was driven by cardiovascular deaths, whereas our findings, in a lower-risk population than the COMET population, indicate that carvedilol also reduces hospitalizations for HF with a 30% relative risk reduction, with further benefit in CRT-D and LBBB subgroups. As noted, unfortunately, COMET was designed with the use of metoprolol tartrate instead of metoprolol succinate, whereas metoprolol used in our study consists of both salts as in real life.
BBs and CRT may have complementary salutary effects as shown by Voigt et al. (13), and our results show that carvedilol in particular is associated with reduction in the rate of hospitalizations for HF, especially in CRT patients with LBBB QRS morphology.
The conclusion in the COMET trial was that the benefit of carvedilol is driven by a reduction in mortality. In MADIT-CRT, the overall mortality rate is markedly lower than the mortality rates in the clinical trials MERIT-HF, COMET, and CIBIS (Cardiac Insufficiency Bisoprolol Study), with a yearly mortality rate of 3% compared with 7% to 10%. Extrapolating the results, it seems that carvedilol is superior to metoprolol in terms of cardiovascular death, supported by risk reductions in our subgroups of VT/VF, but most definitely in terms of HF hospitalizations in our patients. The clinical data and evidence supporting carvedilol in favor of metoprolol now exceed NYHA functional class III to IV and cardiovascular death extending it to NYHA functional class I to II and HF hospitalizations in ischemic and nonischemic cardiomyopathy with wide QRS complexes. Also it seems evident that there are synergistic effects of carvedilol and CRT-D treatment, which again is powered with increased doses of carvedilol. The use of metoprolol succinate at the target dose of 200 mg/day should, however, still be justified based on the MERIT-HF data. The definitive head-to-head comparison of carvedilol and metoprolol succinate at the doses proven to reduce mortality remains to be done, but we find this is a relevant comparison on “real-life” obtainable doses in mild symptomatic HF patients.
One limitation is that the dosages of carvedilol and metoprolol were lower than generally achieved in the clinical randomized trials. Another limitation was that although multivariate analysis showed that carvedilol was superior to metoprolol when taking many confounders into consideration, it was not a prospective randomized trial comparing these drugs, and other confounders not included in the analyses may have biased our results. However, our findings are consistent with the overall results of the COMET trial.
In HF patients in NYHA functional class I and II, with low ejection fraction, wide QRS, and either a CRT-D or ICD device therapy, carvedilol proved superior in terms of a 30% reduction in hospitalizations for HF or death compared with metoprolol. A beneficial and synergistic effect of carvedilol was seen in patients with CRT-D and LBBB QRS morphology with a 49% reduction in HF/death compared with metoprolol. We found similar reductions in risk of ventricular arrhythmias associated with the use of carvedilol. Furthermore, we found a pronounced dosage-dependent relationship between outcome and dose in carvedilol, which was not found in metoprolol-treated patients.
For a supplementary table, please see the online version of this article.
The MADIT-CRT study was supported by a research grant from Boston Scientific to the University of Rochester, with funds distributed to the coordination and data center, enrolling centers, core laboratories, committees, and boards under subcontracts from the University of Rochester. The current study was not funded by Boston Scientific Corporation. Dr. M. H. Ruwald has received unrestricted funding grants from the Danish Heart Association, the Lundbeck Foundation, Helsefonden, Arvid Nilssons Fond, and Knud Hoejgaard Fonden. Dr. Zareba has received lecture fees from Boston Scientific. Dr. Moss has received grant support from Boston Scientific and lecture fees from Boston Scientific, Medtronic, and St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary artery bypass graft
- confidence interval
- cardiac resynchronization therapy
- cardiac resynchronization therapy with implantable cardioverter-defibrillator
- heart failure
- hazard ratio
- implantable cardioverter-defibrillator
- interquartile range
- left bundle branch block
- left ventricular ejection fraction
- New York Heart Association
- ventricular fibrillation
- ventricular tachycardia
- Received October 30, 2012.
- Revision received December 18, 2012.
- Accepted January 8, 2013.
- American College of Cardiology Foundation
- Epstein A.E.,
- Dimarco J.P.,
- Ellenbogen K.A.,
- et al.
- Packer M.,
- Fowler M.B.,
- Roecker E.B.,
- et al.
- Hjalmarson A.,
- Goldstein S.,
- Fagerberg B.,
- et al.
- McMurray J.J.,
- Adamopoulos S.,
- Anker S.D.,
- et al.
- Jessup M.,
- Abraham W.T.,
- Casey D.E.,
- et al.
- Heywood J.T.,
- Fonarow G.C.,
- Yancy C.W.,
- et al.
- Voigt A.,
- Shalaby A.,
- Adelstein E.,
- Saba S.
- Shen X.,
- Nair C.K.,
- Aronow W.S.,
- Hee T.,
- Esterbrooks D.J.
- Greenberg H.,
- Case R.B.,
- Moss A.J.,
- Brown M.W.,
- Carroll E.R.,
- Andrews M.L.
- Fiuzat M.,
- Wojdyla D.,
- Kitzman D.,
- et al.