Author + information
- Received April 27, 2017
- Accepted April 28, 2017
- Published online June 19, 2017.
- Stephen J. Greene, MDa,b,∗ (, )
- Adrian F. Hernandez, MD, MHSa,b,
- Allison Dunning, MSa,
- Andrew P. Ambrosy, MDa,b,
- Paul W. Armstrong, MDc,
- Javed Butler, MD, MPH, MBAd,
- Lukasz P. Cerbin, MDe,
- Adrian Coles, PhDa,
- Justin A. Ezekowitz, MBBCh, MScc,
- Marco Metra, MDf,
- Randall C. Starling, MD, MPHg,
- John R. Teerlink, MDh,
- Adriaan A. Voors, MD, PhDi,
- Christopher M. O’Connor, MDj and
- Robert J. Mentz, MDa,b
- aDuke Clinical Research Institute, Durham, North Carolina
- bDivision of Cardiology, Duke University Medical Center, Durham, North Carolina
- cCanadian VIGOUR Centre, University of Alberta, Edmonton, Alberta, Canada
- dDivision of Cardiology, Stony Brook University, Stony Brook, New York
- eDepartment of Medicine, Duke University Medical Center, Durham, North Carolina
- fCardiology, University of Brescia, Brescia, Italy
- gHeart and Vascular Institute, Cleveland Clinic Foundation, Cleveland, Ohio
- hSection of Cardiology, San Francisco Veterans Affairs Medical Center, and School of Medicine, University of California-San Francisco, San Francisco, California
- iUniversity of Groningen, Groningen, the Netherlands
- jInova Heart and Vascular Institute, Falls Church, Virginia
- ↵∗Address for correspondence:
Dr. Stephen J. Greene, Duke Clinical Research Institute and Division of Cardiology, Duke University Medical Center, 2301 Erwin Road, Suite 7400, Durham, North Carolina 27705.
Background It is unclear how patients hospitalized for acute heart failure (HF) who are long-term chronic HF survivors differ from those with more recent HF diagnoses.
Objectives The goal of this study was to evaluate the influence of HF chronicity on acute HF patient profiles and outcomes.
Methods The ASCEND-HF (Acute Study of Clinical Effectiveness of Nesiritide in Decompensated Heart Failure) trial randomized 7,141 hospitalized patients with acute HF with reduced or preserved ejection fraction (EF) to receive nesiritide or placebo in addition to standard care. The present analysis compared patients according to duration of HF diagnosis before index hospitalization by using pre-specified cutoffs (0 to 1 month [i.e., “recently diagnosed”], >1 to 12 months, >12 to 60 months, and >60 months).
Results Overall, 5,741 (80.4%) patients had documentation of duration of HF diagnosis (recently diagnosed, n = 1,536; >1 to 12 months, n = 1,020; >12 to 60 months, n = 1,653; and >60 months, n = 1,532). Across HF duration groups, mean age ranged from 64 to 66 years, and mean ejection fraction ranged from 29% to 32%. Compared with patients with longer HF duration, recently diagnosed patients were more likely to be women with nonischemic HF etiology, higher baseline blood pressure, better baseline renal function, and fewer comorbidities. After adjustment, compared with recently diagnosed patients, patients with longer HF duration were associated with more persistent dyspnea at 24 h (>1 to 12 months, odds ratio [OR]: 1.20; 95% confidence interval [CI]: 0.97 to 1.48; >12 to 60 months, OR: 1.34; 95% CI: 1.11 to 1.62; and >60 months, OR: 1.31; 95% CI: 1.08 to 1.60) and increased 180-day mortality (>1 to 12 months, hazard ratio [HR]: 1.89; 95% CI: 1.35 to 2.65; >12 to 60 months, HR: 1.82; 95% CI: 1.33 to 2.48; and >60 months, HR: 2.02; 95% CI: 1.47 to 2.77). The influence of HF duration on mortality was potentially more pronounced among female patients (interaction p = 0.05), but did not differ according to age, race, prior ischemic heart disease, or ejection fraction (all interactions, p ≥ 0.23).
Conclusions In this acute HF trial, patient profile differed according to duration of the HF diagnosis. A diagnosis of HF for ≤1 month before hospitalization was independently associated with greater early dyspnea relief and improved post-discharge survival compared to patients with chronic HF diagnoses. The distinction between de novo or recently diagnosed HF and worsening chronic HF should be considered in the design of future acute HF trials. (A Study Testing the Effectiveness of Nesiritide in Patients With Acute Decompensated Heart Failure; NCT00475852)
On a population level, heart failure (HF) carries a prognosis comparable to many cancers, with 5-year mortality rates as high as 50% (1,2). For many of these patients, the clinical course is punctuated by at least 1 hospitalization for acute heart failure (AHF), representing an inflection point in the natural history whereby subsequent mortality risk is 3-fold higher than in patients never hospitalized (3). Indeed, during the “vulnerable phase” within 60 to 90 days’ post-discharge, rates of mortality and readmission may be as high as 15% and 30%, respectively (4,5). However, although overall event rates are high, HF is a heterogeneous clinical syndrome coinciding with various cardiac and noncardiac disease processes, and individual patient survival after hospitalization may vary widely (4,6). It is unclear how hospitalized patients who are previous long-term chronic HF survivors differ from those with more recent or de novo HF diagnoses.
Determining the characteristics of long-term HF survivors and whether they continue to experience favorable post-discharge prognoses may offer incremental insights into risk stratification, therapeutic management, and clinical trial design. If previous long-term HF survivors continue to experience favorable outcomes, such data could be instrumental in defining a true low-risk phenotype and could suggest differing HF biology. Similarly, there are limited data describing the subset of patients with de novo HF and their respective short- and long-term clinical risk as they transition toward either chronic HF or functional recovery. In this context, the ASCEND-HF (Acute Study of Clinical Effectiveness of Nesiritide in Decompensated Heart Failure) database affords the opportunity to characterize the association between HF chronicity, clinical characteristics, and post-discharge outcomes within a large, international cohort of hospitalized AHF patients.
The design and primary results of the ASCEND-HF trial have been reported previously (7,8). Briefly, ASCEND-HF was a prospective, multinational, randomized, placebo-controlled trial studying the effects of nesiritide, an intravenous recombinant B-type natriuretic peptide (BNP), on dyspnea relief and clinical outcomes among hospitalized patients with AHF and either reduced or preserved ejection fraction. Eligible patients were enrolled within 24 h of first intravenous HF therapy and had dyspnea at rest or with minimal exertion, ≥1 clinical sign of HF, and ≥1 objective measure of HF. All patients provided written informed consent. The trial was conducted in accordance with the Declaration of Helsinki and with institutional review board/ethics committee approval at all sites.
Duration of HF diagnosis
Date of initial HF diagnosis was specified in the trial case report form and prospectively recorded by local site investigators. Date of diagnosis information could be obtained by using any available method (e.g., medical record, patient-reported history) at the discretion of the local investigator. Duration of HF diagnosis in reference to the index AHF hospital admission was then calculated. For descriptive purposes, patients were assigned to 1 of 4 HF duration groups: 0 to 1 month (i.e., “recently diagnosed”), >1 and ≤12 months, >12 months and ≤60 months, and >60 months. These cutoffs were chosen a priori based on the distribution of HF duration within the study population (i.e., roughly even distribution of patients in each group) and for ease of communication (i.e., cutoffs at 1 month, 1 year, and 5 years). Duration was also assessed as a continuous variable.
The pre-specified endpoints for the present study were as follows: 1) 180-day all-cause death; 2) 30-day all-cause death or HF hospitalization; 3) 30-day all-cause death or all-cause rehospitalization; and 4) persistent dyspnea (i.e., no dyspnea improvement) at 6 and 24 h. Causes of death and hospitalization within 30 days were adjudicated by an independent blinded clinical events committee (University of Glasgow, Glasgow, United Kingdom). Persistent dyspnea was defined by using a self-reported 7-point categorical Likert scale as markedly worse, moderately worse, minimally worse, no change, or minimally better.
Baseline characteristics were compared between HF duration groups. Continuous variables were reported as median (25th percentile, 75th percentile) and compared by using Kruskal-Wallis tests. Categorical variables were presented as frequencies and percentages and were compared by using the Pearson chi-square test or the Fisher exact test. Raw clinical event rates were reported across groups.
HF duration was pre-specified to be examined as both a categorical variable according to HF duration group and a continuous variable per 12-month increase. Unadjusted and adjusted logistic regression models were constructed to evaluate associations between HF duration and 30-day endpoints and persistent dyspnea. Unadjusted and adjusted Cox models were used to evaluate associations with 180-day all-cause death. To account for potential clustering of patients within countries, hierarchical modeling was used. Specifically, random intercept multilevel logistic regressions (generalized linear mixed models) and shared frailty Cox regression, each using random effects for country of patient enrollment, were used. Adjusted models included 17 pre-specified covariates either used previously in ASCEND-HF mortality and dyspnea models or added a priori per clinical judgment (9,10).
Linearity and proportional hazards assumptions were tested for all models, and no violations were found. In addition, the influence of HF duration on 180-day mortality was tested across multiple pre-specified subgroups of interest, including ischemic/nonischemic heart disease, age ≥50 and <50 years, white/non-white race, sex, and ejection fraction (EF) <40% and ≥40%. Multiple imputation (under the missing at random assumption) was used for missing baseline covariate data (<10% for all variables), and reported results were summarized across 25 imputation datasets. All statistical analyses were performed using SAS version 9.4 or later (SAS Institute, Inc., Cary, North Carolina). Two-tailed p values <0.05 were considered statistically significant.
Of the 7,141 patients enrolled in the ASCEND-HF trial, 5,741 (80.4%) had documentation of the duration of the HF diagnosis. Characteristics of patients according to study inclusion status are presented in Online Table 1. Compared with excluded patients (i.e., those who lacked documentation of date of initial HF diagnosis), included patients tended to be younger with less ischemic heart disease, lower systolic blood pressure, and more signs and symptoms of congestion.
Baseline patient characteristics (at time of trial randomization) according to HF duration group are displayed in Table 1. Mean age within each group ranged from 64 to 66 years. Compared to patients with worsening chronic HF (i.e., >1 month duration), patients with recently diagnosed HF (i.e., 0 to 1 month duration) were more likely to be female, Asian, and have a nonischemic HF etiology with preserved EF. Recently diagnosed patients tended to have higher blood pressure, better renal function, and less severe New York Heart Association functional class. Rates of comorbidities and guideline-directed therapies for HF with reduced EF were lowest among patients recently diagnosed with HF, with progressively higher rates with increasing HF duration. In contrast, although natriuretic peptide levels were lowest among recently diagnosed patients, there was no clear relationship between level and increasing HF chronicity.
Measures of congestion
Table 2 presents data on baseline and in-hospital change in measures of congestion. With the exception of an increased prevalence of pulmonary rales, baseline signs and symptoms of congestion were generally less common among patients with recently diagnosed HF. There were no significant relationships between HF duration and measures of in-hospital decongestion, including urine volume and change in weight.
Event rates according to HF duration are displayed in Table 3 and the Central Illustration. Patients with recently diagnosed HF consistently had the lowest rates of 30-day all-cause death or HF rehospitalization, 30-day all-cause death or hospitalization, and 180-day all-cause death. In contrast, these event rates were highest among patients with a duration of HF >60 months. There was no graded relationship between event rates and increasing HF chronicity, with the >12- to 60-month group having lower rates than the >1- to 12-month group for all 3 endpoints.
Unadjusted and adjusted analyses for study endpoints are presented in Table 4. Compared with recently diagnosed patients, all other HF duration groups carried a >2-fold greater risk of 180-day all-cause death. These associations remained statistically significant after accounting for patient characteristics (all p < 0.001). Point estimates between HF duration groups did not show a graded correlation with increasing HF duration. After adjustment, analysis of continuous HF duration exhibited a trend toward increased 180-day all-cause death per 12-month increase in HF chronicity (p = 0.099). In unadjusted analyses of the 30-day all-cause death or HF hospitalization and 30-day all-cause death or hospitalization composite endpoints, robust and statistically significant associations were seen among all HF duration groups, compared with recently diagnosed patients. The majority of these associations remained statistically significant after adjustment. However, after accounting for patient characteristics, continuous HF duration did not predict 30-day endpoints (all p ≥ 0.372).
In both unadjusted and adjusted analyses, HF duration group was predictive of persistent dyspnea at 24 h. Specifically, after adjustment, patients with >12-month HF chronicity carried a 31% to 34% greater risk of persistent dyspnea at 24 h compared with patients recently diagnosed with HF. There was a trend toward greater risk of 24-h persistent dyspnea per 12-month increase in HF duration (p = 0.096). After adjustment, there were no significant associations between HF duration and persistent dyspnea at 6 h.
HF duration, nesiritide efficacy, and subgroups of interest
There were no significant differences in nesiritide efficacy across HF duration groups for 180-day all-cause death or 30-day all-cause death or HF rehospitalization (p for interaction ≥0.585) (Online Table 2). In contrast, there was a significant interaction for 24-h persistent dyspnea, whereby nesiritide was associated with decreased likelihood of persistent dyspnea among patients with >1- to 12-month and >60-month HF duration (p for interaction = 0.040).
Figure 1 displays interaction analyses for HF duration groups and 180-day all-cause death across pre-specified subgroups of interest. The association between HF duration and death was consistent, irrespective of age, race, ejection fraction <40% or ≥40%, and presence of ischemic heart disease (all p for interaction ≥0.227). Interaction analysis according to sex yielded borderline statistically significant results (p for interaction = 0.050), with a potentially stronger relationship between HF duration and mortality among female patients.
In this large international cohort of patients hospitalized for AHF, patient characteristics varied with duration of the existing HF diagnosis, with particularly marked differences between those with recently diagnosed or de novo HF and worsening chronic HF. Notably, recently diagnosed patients were more likely to be female and tended to have a nonischemic HF etiology, higher baseline blood pressure, better baseline renal function and functional class, and fewer comorbidities. Compared with patients with worsening chronic HF, recently diagnosed HF status was independently associated with lower rates of post-discharge death and rehospitalization and improved rates of early dyspnea relief. Among patients with HF duration >1 month, there was no strong evidence of a graded relationship between increasing HF chronicity and outcomes, with the overall influence of continuous HF duration in this study driven by differences between recently diagnosed and worsening chronic HF status. With potential exception of a borderline significant interaction by sex, the predictive value of HF duration on mortality was consistent, irrespective of age, race, EF, and presence of ischemic heart disease. The ability of nesiritide to provide dyspnea relief at 24 h varied according to duration of the HF diagnosis, but there was no differential influence on post-discharge outcomes.
Previously reported proportions of patients with de novo versus worsening chronic HF have varied widely, with de novo rates ranging from 13% in North America to 49% in Finland (11–14). However, to our knowledge, only 2 previous, smaller studies have specifically explored the influence of HF duration on AHF patient profiles and outcomes (12,14). Similar to these existing analyses, we found that the group of recently diagnosed and de novo patients had fewer comorbidities, had higher baseline blood pressure and better baseline renal function, and experienced superior post-discharge outcomes compared to patients with established chronic HF. However, the current data from ASCEND-HF offer multiple novel features that shed incremental and more granular insight into the role of HF chronicity on clinical course during and after AHF hospitalization. First, in contrast to previous studies, ASCEND-HF contains precise dates of original HF diagnoses and thus allowed for assessment of a graded effect of HF duration on trial endpoints via both categorical and continuous variable analyses. Second, although previous studies solely report associations with all-cause mortality, the present analysis is the first from a clinical trial cohort and the first to document relationships with dyspnea-related and adjudicated mortality and hospitalization endpoints. Third, we explored, for the first time, the interaction between HF chronicity and multiple AHF subgroups of interest and found no differential influence regardless of age, race, etiology of HF, or EF and a borderline interaction according to sex. Fourth, we provide data on baseline signs and symptoms of congestion and document associations between HF chronicity, urine output, and change in weight during AHF hospitalization. Lastly, the large sample size from a global AHF population allowed for rigorous statistical adjustment to determine the independent prognostic impact of HF duration on outcomes.
In the present study, clinical characteristics of the recently diagnosed HF population were generally concordant with the superior post-discharge outcomes observed, as previous investigations have established features such as higher systolic blood pressure, better baseline renal function, and nonischemic HF etiology as strong markers of more favorable prognosis (4,15,16). However, the robust association with improved outcomes seen here persisted despite adjustment for these factors and others in a multivariate model, suggesting that other characteristics inherent to this population drove the differences in event rates. One could speculate that the improved outcomes among recently diagnosed patients were, in part, due to increased rates of downstream myocardial recovery in this group, compared with patients with longstanding persistent HF despite therapy. Unfortunately, longitudinal data on resolution of HF are not available in the ASCEND-HF database to confirm this hypothesis. An additional notable finding is the lack of an obvious graded relationship between HF duration and outcomes among patients with established chronic HF. Before these data, one may have postulated that given the overall poor long-term prognosis of chronic HF, individual patients proven to be long-term survivors may represent a true low-risk phenotype with relative impermeability to death or hospitalization. However, the current findings do not support this theory. The lack of a graded relationship between outcomes and increasing chronic HF duration suggests no protective effect from previous long-term survivorship, with comparable risk of post-discharge mortality and hospitalization for patients with HF diagnoses of several years versus several months.
Clinical trial implications
Over the last 2 decades, a multitude of large international trials has tested various therapies in the AHF population, but there remain no approved therapies definitively proven to improve clinical outcomes (17). Although explanations for the persistent lack of trial success are debated and likely multifactorial, increasing attention has centered on the heterogeneity of the AHF population (17,18). Whether secondary to differences in patient-level factors (i.e., HF biology, comorbidities), trial site and study execution factors (i.e., interpretation of selection criteria, enrollment rate, geographic region), or regional health care practices (i.e., physician and patient thresholds for hospital admission and discharge decisions), patients enrolled in AHF trials vary widely despite strict and uniform selection criteria (19). The current data from ASCEND-HF suggest that specification of de novo or recently diagnosed versus worsening chronic HF status represents a valuable additional dimension of AHF trial heterogeneity that correlates with clinical profile and independently predicts patient outcomes. Previous studies have not consistently differentiated between enrollment of de novo versus worsening chronic HF and, in the case of ASCEND-HF and other trials, patients are included regardless of the duration of their HF diagnosis (7,20). The present findings challenge this convention, as the dramatically different event rates among patients with recently diagnosed and chronic HF strongly suggest different populations. Not only may inclusion of both groups make accurate study power calculations difficult, but it is entirely plausible that the efficacy of an investigational therapy may vary according to duration of the HF diagnosis, with the effect of nesiritide on 24-h dyspnea as a potential illustrative example. Moreover, previous AHF trials have generally not pre-specified subgroup analyses for efficacy and safety according to de novo versus worsening chronic HF status.
To date, all Phase III AHF trials have required an index hospitalization for HF with the intuitive belief that it represents a reliable surrogate for worsening HF. Although this approach provides a practical means of easily identifying decompensated patients at high risk, patients with worsening chronic HF who are not hospitalized may carry a similarly poor prognosis (21,22). Likewise, although hospitalization is undoubtedly a strong marker of risk, a recent analysis from another large AHF trial cohort suggests that individual patient characteristics are the dominant predictors of subsequent mortality, regardless of recentness of previous hospitalization (3,23). The current findings from ASCEND-HF build on this theme by showing that although all included patients experienced an AHF hospitalization, exclusion of de novo and recently diagnosed patients enriches the population toward a more homogeneous cohort with worsening chronic HF with a high event rate. Thus, when taken in aggregate, the collective available data suggest that specification of a recent worsening chronic HF event, irrespective of the location of care, may prove a superior strategy for identifying a homogeneous population for future AHF trials, compared with the traditional mandate for an index hospitalization. The recent Phase II SOCRATES-REDUCED (Soluble Guanylate Cyclase Stimulator in Heart Failure with Reduced Ejection Fraction) trial used such a novel strategy, whereby eligible patients had a diagnosis of chronic HF of ≥30 days’ duration and a recent episode of worsening symptoms that required either hospitalization or outpatient administration of intravenous diuretic (24).
First, despite rigorous multivariable modeling, the retrospective observational design of this study prohibits definitive determination of cause–effect relationships. Second, notation of date of original HF diagnosis was done by study investigators, and there was no requirement for additional documentation confirming specific dates. Nonetheless, this approach may be most similar and generalizable to routine clinical practice. Third, nearly 20% of the ASCEND-HF population lacked data on date of HF diagnosis and were excluded from this analysis. Although there were few clinically relevant differences between patients included and excluded in this study, we cannot exclude the potential influence of selection bias on the results seen here (Online Table 1).
In this large international AHF trial, patient profile differed according to the duration of the HF diagnosis at the time of hospitalization, with marked differences between patients with de novo or recently diagnosed HF and worsening chronic HF. A diagnosis of HF for ≤1 month was independently associated with greater early dyspnea relief and improved post-discharge outcomes compared to patients presenting with worsening chronic HF. The distinction between de novo or recently diagnosed HF and worsening chronic HF is clinically relevant and should be considered in the design of future AHF trials.
COMPETENCY IN MEDICAL KNOWLEDGE: Compared to patients hospitalized for worsening chronic HF, patients hospitalized for de novo or recently diagnosed AHF tend to have nonischemic HF etiology, higher baseline blood pressure, better baseline renal function and functional class, and fewer comorbidities. Patients hospitalized for de novo or recently diagnosed HF experience greater early dyspnea relief and have improved post-discharge outcomes compared to patients hospitalized with worsening chronic HF. Among patients hospitalized for worsening chronic HF, there is no protective effect from prior long-term survivorship, with comparable post-discharge outcomes for patients with HF diagnoses of several years versus several months.
TRANSLATIONAL OUTLOOK: AHF clinical trials do not consistently specify de novo versus worsening chronic HF status in trial selection criteria. This distinction is clinically relevant and should be accounted for in the design of future AHF trials.
For supplemental tables, please see the online version of this article.
Scios Inc. provided financial and material support for the ASCEND-HF trial. Database management and statistical analysis were performed by the Duke Clinical Research Institute. Dr. Hernandez has received consulting fees from Sanofi, Johnson & Johnson, AstraZeneca, and Corthera; and research support from Amylin and Scios/Johnson & Johnson. Dr. Butler has received research support from the National Institutes of Health, the European Union, and the Patient Centered Outcomes Research Institute; and has served a consultant for Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, CardioCell, Janssen, Novartis, Relypsa, ZS Pharma, Medtronic, Merck, and CVRx. Dr. Ezekowitz has received consulting fees from Pfizer, Abbott Laboratories, and Servier; and research support from Amgen and Johnson & Johnson. Dr. Metra has received consulting incomes from Bayer, Novartis, and Servier. Dr. Starling has received consulting fees from Novartis, BioControl, and Medtronic; has ownership/partnership/principal in CardioMEMS; has received research support from the National Institutes of Health, Medtronic, Biotronik, Novartis, and Thoratec; and is in receipt of benefits from the American Board of Internal Medicine. Dr. Teerlink has received research/consulting fees from Amgen, Madeleine, Mast Therapeutics, Novartis, Relypsa, and Trevena. Dr. Voors has received consultancy fees and/or research grants from Alere, Amgen, Anexon, Bayer, Boehringer Ingelheim, Cardio3BioSciences, Celladon, Merck, Novartis, Servier, Torrent, and Vifor Pharma. Dr. O’Connor has received consulting fees from Novella and Amgen; has ownership/partnership/principal in Biscardia, LLC; and has received research support from Otsuka, Roche Diagnostics, BG Medicine, Critical Diagnostics, Astellas, Gilead, GE Healthcare, and ResMed. Dr. Mentz has received research support from Amgen, AstraZeneca, Bristol-Myers Squibb, GlaxoSmithKline, Gilead, Novartis, Otsuka, and ResMed; and has received honoraria from Thoratec. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute heart failure
- B-type natriuretic peptide
- ejection fraction
- heart failure
- Received April 27, 2017.
- Accepted April 28, 2017.
- 2017 American College of Cardiology Foundation
- Solomon S.D.,
- Dobson J.,
- Pocock S.,
- et al.
- Greene S.J.,
- Fonarow G.C.,
- Vaduganathan M.,
- Khan S.S.,
- Butler J.,
- Gheorghiade M.
- Gheorghiade M.,
- Pang P.S.,
- Ambrosy A.P.,
- et al.
- Khazanie P.,
- Heizer G.M.,
- Hasselblad V.,
- et al.
- Abraham W.T.,
- Fonarow G.C.,
- Albert N.M.,
- et al.
- Okumura N.,
- Jhund P.S.,
- Gong J.,
- et al.
- Cook T.D.,
- Greene S.J.,
- Kalogeropoulos A.P.,
- et al.
- Gheorghiade M.,
- Greene S.J.,
- Butler J.,
- et al.