Long-Term Healthcare and Cost Outcomes of Disease Management in a Large, Randomized, Community-Based Population With Heart Failure
Background— Because of the prevalence and expense of congestive heart failure (CHF), significant efforts have been made to develop disease management (DM) programs that will improve clinical and financial outcomes. The effectiveness of such programs in a large, heterogeneous population of CHF patients remains unknown.
Methods and Results— We randomized 1069 patients (aged 70.9±10.3 years) with systolic (ejection fraction 35±9%) or echocardiographically confirmed diastolic heart failure (HF) to assess telephonic DM over an 18-month period. Data were collected at baseline and at 6-month intervals. Survival analysis was performed by Kaplan-Meier and Cox regression methods. Healthcare utilization was defined after extensive record review, with an attempt to account for all inpatient and outpatient visits, medications, and diagnostic tests. We obtained data on 92% of the patients, from nearly 53 000 health-related encounters. Total cost per patient was defined by adding estimated costs for the observed encounters, excluding the cost of the DM. Kaplan-Meier analysis showed that DM patients had a reduced mortality rate (P=0.037), with DM patients surviving an average of 76 days longer than controls. Subgroup analysis showed that DM had beneficial outcomes in patients with systolic HF (hazard ratio 0.62; P=0.040), which was more pronounced in NYHA classes III and IV. Although improvements in NYHA class were more likely with DM (P<0.001), 6-minute walk data from 217 patients in whom data were available at each visit showed no significant benefit from DM (P=0.08). Total and CHF-related healthcare utilization, including medications, office or emergency department visits, procedures, or hospitalizations, was not decreased by DM. Repeated-measures ANOVA for cost by group showed no significant differences, even in the higher NYHA class groups.
Conclusions— Participation in DM resulted in a significant survival benefit, most notably in symptomatic systolic HF patients. Although DM was associated with improved NYHA class, 6-minute walk test results did not improve. Healthcare utilization was not reduced by DM, and it conferred no cost savings. DM in HF results in improved life expectancy but does not improve objective measures of functional capacity and does not reduce cost.
Received September 1, 2004; revision received October 12, 2004; accepted October 13, 2004.
Congestive heart failure (CHF) is one of the most prevalent and expensive chronic diseases in the United States. It is also the only cardiovascular disorder that is increasing in both incidence and prevalence,1 a likely result of the aging of the US population, reductions in death rate for myocardial infarction, and the greater longevity of CHF patients.2 Prevalence of this disease is expected to reach 10 million cases by the year 2007.3 The symptoms of CHF are responsible for >11 million physician office visits and are directly related to >3.5 million hospitalizations per year.1 In addition, readmission rates as high as 15% to 50% within 6 months of discharge have been reported.3 Symptoms of CHF remain the leading cause of hospitalization in individuals aged >65 years,1,2 with an average admission for heart failure (HF) costing between $7174 and $10 000.4,5 In the United States, annual direct expenditure of healthcare dollars for symptom management of CHF has been estimated to be $20 billion to $56 billion.1,3,5 In the Medicare population, estimates show that CHF costs approximately $8 billion per year.2
Disease management (DM) has recently been proposed as a useful approach to reduce repeated hospitalization in CHF. The goal of DM is to improve health outcomes while simultaneously reducing healthcare cost. Although a number of prior studies have demonstrated that these goals can be met,6–9 the studies have been of limited duration and small sample size, generally focusing on a patient population with homogeneous demographic characteristics. Given the fact that CHF due to systolic dysfunction is a progressive disease for which there is no curative treatment, whether DM can meet its stated goals over a longer period of time in this condition is unclear. Moreover, little is known about the effectiveness of DM in diastolic HF. The purpose of our study was to evaluate the effectiveness of DM as a clinical and cost-containment strategy in both systolic and diastolic CHF over a longer time period and with a larger and more heterogeneous sample than those of previous studies.
The South Texas Congestive Heart Failure Disease Management Project was a single-center, randomized, controlled clinical trial that was performed between 1999 and 2003 through the University of Texas Health Science Center at San Antonio, in partnership with Wilford Hall Medical Center, Brooke Army Medical Center, South Texas Veterans Health Care System, TRICARE Region 6, and University Health System. A total of 1069 male and female subjects, aged ≥18 years, with symptoms of CHF and documented systolic or diastolic dysfunction were enrolled. Potential patients were identified through lists generated from the databases of partner institutions. In addition, the study entered an academic partnership with the Centers for Medicare and Medicaid Services and obtained a list of potential participants from the Centers for Medicare and Medicaid Services. These patients were contacted by letter and offered an opportunity to participate. All lists were generated on the basis of International Classification of Diseases, Ninth Revision diagnosis of HF. Patients were identified and tracked by practice site and physician specialty.
A screening questionnaire administered by telephone contained 9 questions about previous history suggestive of CHF signs or symptoms. The questions were based in part on the National Health and Nutrition Examination Survey (NHANES) Cardiovascular Disease Questionnaire and were meant to be broadly inclusive. A positive answer to any question was considered suggestive of CHF, and these patients proceeded to screening echocardiography. Echocardiograms were performed by research staff and interpreted by a single cardiologist who was board certified in echocardiography and not associated with the study. Standard 2D echocardiographic examinations were performed. Systolic dysfunction was broadly defined as an ejection fraction of ≤49%, and criteria for diastolic dysfunction included any of the following: left ventricular hypertrophy (interventricular septal or posterior wall thickness ≥1.2 cm), E-to-A reversal, or other abnormal transmitral flow pattern. Echocardiographs showing ejection fraction of ≥50%, absence of left ventricular hypertrophy, and lack of E-to-A reversal or abnormal mitral flow pattern were considered negative for CHF.
Approval was obtained from the institutional review boards of all partner institutions. After informed consent was obtained and patients were screened by history and echocardiography, 1069 adult subjects were enrolled for a period of 18 months per subject. Participants were randomized in a 2:1 ratio between the treatment and control groups. All DM patients received bathroom scales. Within the treatment group, there was additional randomization, with half of the treatment group given in-home technology, including an electronic blood pressure monitor and finger pulse oximeter. Blood pressure and pulse oximeter readings were reported by the patient to the disease managers, but these data were not forwarded to the primary care provider (PCP). This group also wore activity monitors for 2 of each 6 weeks and had thoracic bioimpedance cardiac output measurements performed at 6-month intervals, although the results were not reported to their physicians. Because data derived from the technology were not used in clinical management, we combined results from the 2 treatment groups for the purposes of this analysis.
All subjects underwent an echocardiogram at baseline and at 18 months and were assessed every 6 months by medical history, physical examination, 6-minute walk test, and serum chemistries. Subjects in the traditional care group were managed as usual by their physicians. Subjects in the DM group were assigned a disease manager, a registered nurse with specialized cardiac training, who provided patient education and medication management in conjunction with the PCP.
Disease managers were employed by CorSolutions, Inc, an established DM firm contracted by the study. They used the proprietary MULTIFIT protocol, under which the patients’ care was directed by their physicians, with recommendations made by disease managers in accordance with the American College of Cardiology/American Heart Association guidelines for CHF.9,10 In addition, disease managers recommended antilipemic and antianginal medications when indicated. The recommendations were a part of the study protocol, but whether or not to follow them was left to the physician’s discretion. The DM program was administered telephonically. The first encounter entailed a medical history, elucidation of current medications, smoking history, dietary and lifestyle assessments, and a confidence rating. All patients were mailed educational material on CHF with emphasis on the patient’s role in management through daily weight, activity level, sodium intake, and medication compliance. Smokers were offered smoking cessation instruction including both written material and verbal reinforcement from disease managers. In the event that a patient was thought to be unstable by the disease manager, face-to-face evaluation with a home healthcare nurse could be arranged.
Initial call frequency was weekly, with a transition to monthly for the duration of the intervention. The frequency of calls could be adjusted for acuity or need. After each call, a call summary was faxed to the patient’s PCP. Information or orders from the PCP could likewise be faxed to the disease manager.
Patient education included instruction in appropriate cardiac diet (low fat, low sodium, fluid restricted), medication compliance, exercise, and appropriate reaction to signs of the onset of a CHF exacerbation. Patients were given a 24-hour-a-day, 7-day-a-week, toll-free telephone number that they could call with questions about CHF management. Reported symptoms were addressed on the basis of the DM algorithm unless the patient’s physician had given other orders; physicians were called for all management issues.
To assess economic and utilization outcomes, we thoroughly reviewed patient medical records. Records covering the period of a patient’s enrollment in the trial were requested from both PCPs and specialty physicians. Records of inpatient and outpatient encounters and emergency department visits were culled from patient self-reported data, reviews of electronic hospital records, and documents received from physician and clinic charts. Reviews were performed by study staff, consisting of physicians, nurses, and ancillary health providers.
The central hypothesis of the study was that DM would improve outcomes in patients with CHF. The primary study end point was all-cause mortality. Five secondary outcomes were also examined. Three were assessed across all patients: performance on a 6-minute walk test, improvement in functional therapeutic class, and total healthcare costs. In the subgroup of systolic HF patients, 2 additional outcomes were used: improvement in ejection fraction and adherence to guideline-based medications (defined as percentage on a combination of diuretic, ACE inhibitor or angiotensin receptor blocker or hydralazine/nitrite combination, and a β-blocker). This report addresses the clinical outcomes and the associated cost of healthcare utilization. A subsequent report will detail the results of a full cost-effectiveness analysis.
The data were analyzed by an intent-to-treat approach. Continuous variables were expressed as mean±SD, and categorical variables were expressed as percentages. Bivariate hypotheses involving continuous variables were tested with a t test for independent groups. For tests of whether the distribution of categorical variables differed across study groups, the Fisher exact test was used in cases in which the outcome was dichotomous, and a χ2 test was used in cases in which the outcome was polytomous. Overall survival analyses were conducted with the Kaplan-Meier technique; differences in the survival curves were assessed with a Wilcoxon log-rank test. Cox regression, adjusted for study group membership, was used to ascertain whether the effect of the intervention differed across key demographic and clinical subgroups. A series of repeated-measures ANOVAs was used to assess whether clinical or utilization outcomes differed between study groups and/or over time. Time trends and intergroup differences in the use of guideline-based medicines were evaluated by repeated-measures logistic regression. We set the overall α at 0.05. Half of this value was allocated to the overall analysis of all-cause mortality; the other half was split equally among the 5 secondary outcomes. Additional subgroup analyses of all-cause mortality and cardiac event–free survival were conducted but were not included in the overall α allocation because of their exploratory nature. Sample size calculations showed that a total sample size of 1058 would yield 80% power to detect clinically and economically significant differences across the primary and secondary outcomes. Analyses were conducted with SAS version 8.2.
Demographic data are shown in Table 1. There was a slightly higher average diastolic blood pressure (1.8 mm Hg) in the experimental group, which was statistically significant (P=0.032). For all other descriptors the groups were closely matched. For the entire study population, 29% of the patients were female, 70% had systolic HF, 72% had hypertension, 62% had coronary artery disease, and 28% had diabetes.
Comparison of Survival and Event-Free Survival
As can be seen in Figure 1, Kaplan-Meier analysis showed that patients in the intervention group had a statistically significant reduction in mortality (P=0.037). Over the course of the trial, patients in the intervention group lived approximately 76 days longer than patients in the control group (mean survival of 526.9 versus 450.5 days). By contrast, although there was a trend toward fewer events in the treatment group, the Kaplan-Meier analysis of cardiac event–free survival shown in Figure 1 demonstrated that the intervention did not have a statistically significant effect on cardiac event–free survival (P=0.074).
As can be seen in Figure 2, when the analysis was restricted to the subgroup of patients with systolic dysfunction, significant differences in mortality rate and in cardiac event–free survival were found (P=0.012). Additional analyses suggest that the overall positive effect of DM was driven by a strong protective effect among the most severely ill patients. Cox regression analysis of all-cause mortality showed that hazard ratios for experimental group membership were 1.32 for asymptomatic patients (NYHA class I; P=NS), 0.76 (P=NS) for patients in NYHA class II, and 0.54 (P=0.048) for patients in NYHA classes III and IV. When survival data were analyzed for only the patients with systolic HF, the benefit derived by the intervention group was lengthened to 81 days (445 days for control and 526 days for intervention group). For this subgroup, a substantially stronger effect on mortality was observed among patients in NYHA classes III and IV (hazard ratio 0.37; P=0.004). Of note, however, for patients with diastolic HF (Figure 3), no difference in mortality rate or in cardiac event–free survival was derived from the intervention.
Clinical End Points
Analysis of 6-minute walk test results, the study’s primary objective measure of exercise capacity, was performed with the use of repeated-measures ANOVA comparing the control with the intervention group (Table 2). As a result of patient dropout or refusal to perform the test, there were missing data points, and the analysis was performed on 217 patients in whom data were available at each 6-month interval from the time of entrance to 18 months of follow-up. No statistically significant difference in distance walked was observed between groups (P=0.08) at baseline or over the course of the intervention. Of note, there was a minor reduction in 6-minute walk distance in both groups at 18 months.
Analysis of the ejection fraction was performed for patients with systolic HF and was limited to those with both an initial and final ejection fraction recorded (Table 2), a total of 502 patients. Repeated-measures ANOVA revealed a small increase in average ejection fraction over time (P<0.001), which tended to be greater for the control group, although the difference in ejection fraction between the groups at baseline or at 18 months was not statistically significant (P=0.82).
We also assessed the likelihood that patients would experience an improvement in NYHA functional class, another secondary measure of physical well-being. At 18 months, 24.8% of DM patients had a better NYHA functional class score than at baseline, compared with improvement in 12.6% of the control group. NYHA score improvements were significantly more likely among patients in the DM group than in control patients (P<0.001).
Medication regimens were analyzed to assess differences in the percentage of patients on appropriate medication regimens at the end of the trial (Figure 4). Because specific medication guidelines have not been developed for diastolic HF, this analysis was limited to patients with systolic dysfunction. Appropriate medication regimen was defined as a combination of a diuretic, an ACE inhibitor (or angiotensin receptor blocker or hydralazine/nitrite combination), and a β-blocker (for patients with NYHA class II to IV HF). Repeated-measures logistic regression analysis revealed a statistically significant interaction effect between group membership and time (P=0.002). At the end of the trial, 43.3% of the control patients and 54.4% of the patients in the treatment group were on guideline-recommended medications. Thus, DM significantly increased the fraction of patients with systolic HF who were prescribed guideline-based therapy.
Healthcare utilization data were collected through extensive medical record review, in which we attempted to define all inpatient and outpatient encounters, as well as drug use and use of diagnostic tests. We obtained data on 92% of the patients, and 66% of requested records were returned for review. This entailed approximately 53 000 health-related encounters. Data audits were performed on a randomly selected 5% of all charts as an assessment of review accuracy, with an interrater reliability of 96.8%.
Analysis of total healthcare utilization as well as CHF-related healthcare utilization showed that in no area, including drug use, office visits, emergency department visits, procedures, or hospitalizations, was there a decrease with the intervention. Moreover, total healthcare cost was not statistically different between groups (Table 3). Total costs per patient were computed by summing the estimated costs for the 5 resource categories by the three 6-month study periods, and then summing these 3 totals. The cost of administering the DM intervention was not included in this analysis. Repeated-measures ANOVA for total estimated cost by study group showed no significant differences. In several categories the average costs for the experimental group were higher than those for the control group. A lack of cost benefit persisted even when the analysis was limited to more severe NYHA classes.
Our study shows for the first time that when DM is applied for an 18-month period to a large, diverse, community-based population with CHF, there is a moderate increase in life span and in NYHA functional class but no reduction in overall healthcare cost. Patients with systolic HF were most likely to benefit from the DM program, and the effect was more pronounced for patients in NYHA classes III and IV. Patients with diastolic HF did not benefit. Although patients enrolled in DM were more likely to be on appropriate guideline-based therapy at the end of the study, there was no impact on objective measures of function such as ejection fraction or performance on a 6-minute walk test.
The nature of our patient population provided us with an opportunity to evaluate DM in patients with a heterogeneous geographic and demographic distribution. Our catchment area spanned approximately 70 000 square miles in South Texas, and patients came from urban, suburban, and rural settings. Unlike most prior reports, which have been limited to small populations from a single healthcare system,8–12 our study directly addresses the question of how useful DM can be in a more general application. Our results demonstrate that in this real-world setting, DM can improve longevity and subjective symptoms, although its impact on healthcare usage is limited.
Another key distinction of our study is the inclusion of patients with both systolic and diastolic HF. Although most prior studies have focused on systolic dysfunction, it is possible that in some trials in which enrollment was based on symptoms alone, patients with diastolic HF have been included.8,11,13,14 Because all of our patients were evaluated by echocardiography, we can be certain about the classification of each patient. In this regard our results are unique. The patients with diastolic HF had substantially higher blood pressures at the time of entry into the trial, and fewer achieved the guideline-based blood pressure, whether or not they were in the DM program. Moreover, the patients in the diastolic HF group had lower rates of all-cause mortality and cardiac mortality. The implications of our findings are that the major underlying condition in patients with diastolic HF is hypertension and that it appears to be resistant to treatment. Furthermore, although DM did little to affect blood pressure control, we cannot discern whether this resulted from refractoriness to drug treatment or to the fact that physicians were willing to accept blood pressures that were above guideline recommendations in this subgroup, a possible reflection of the difficulty of reaching target blood pressures. In either case, over the 18-month study period, patients with diastolic HF had lower overall and cardiac-specific mortality than patients with left ventricular systolic dysfunction.
Many prior studies have reported DM in CHF to have more favorable results, in terms of both reduced hospitalizations and reduced costs. We believe that there are a number of explanations for our contrasting results. First, many of the trials did not use randomized control groups but relied on historical controls or past cost data on individual patients enrolled in the study.6–10,15–18 Because the management of CHF has undergone substantial evolution in recent years, use of historical control data may overestimate the impact of the intervention. Second, our study was considerably larger than most of those for which published data are available. Historically, there are many examples of small pilot studies providing optimism for larger trials that had much less favorable results.19,20 Third, our diagnosis of CHF did not rest on historical evidence or past hospital discharge codes alone. Because we confirmed the presence of CHF by echocardiography, this problem was obviated, and we can be confident that the diagnosis was correct. Fourth, our study had a longer follow-up period than all but 2 published trials of DM.21,22 It is possible that the early gains noted in previous studies would have not have persisted had the follow-up been longer.12,23
Our findings are in accord with those of 3 earlier reports. Riegel and colleagues24 monitored 240 patients who were identified as having CHF at the time of hospital admission. Half of the patients were in a control group, constructed of patients matched to the treatment group patients for preadmission functional status, age by decade, and comorbidities. The patients were monitored for a 6-month period after enrollment in a multidisciplinary DM program that included intensive predischarge counseling and postdischarge telephonic and in-home follow-up. Despite this intervention, no difference was noted in overall or CHF-specific hospitalization rate, days in hospital, or total cost. In contrast to our finding of similar expenditures between groups, in the patients monitored by Riegel et al, functional class I patients showed a marked increase in hospitalization rate and total cost compared with patients in the control group. In a second study, Laramee and colleagues25 studied 287 patients identified at the time of hospitalization for CHF as being at risk for readmission. The intervention included predischarge education and postdischarge telephonic management. Although patients in the treatment group were more likely to be on guideline-based therapy and had higher satisfaction with their care, there was no reduction in hospitalizations, emergency department visits, or cost. In the third trial, Mejhert et al21 studied elderly patients with systolic HF, showing that posthospitalization nurse case management improved appropriate drug utilization but did not reduce mortality, readmission rates, or quality of life. Our randomized study, performed in a large group of outpatients recruited from a variety of different healthcare settings, strongly suggests that widespread application of DM programs for all patients with CHF may have limited impact on healthcare cost.
Our study must be interpreted in view of several limitations. First, although our questionnaire was intentionally broad and inclusive, it was not specifically validated as a diagnostic tool to select patients with HF. The frequency with which patients with past histories that should make them positive responders gave negative answers is not known, and thus some patients eligible for the study may have been missed. Although false-positive responses may have increased numbers of patients with negative screening echocardiographs, this should not have altered the impact of the trial on patients who ultimately qualified. Second, we used self-reported data on healthcare encounters, and it is possible that this underestimated the incidence of these events. We used actual medical record review for estimation of costs, and because all groups had the same number of interactions with the research teams, there should have been no intergroup bias in regard to reporting healthcare use. Third, our definition of diastolic HF was broad and was less sophisticated than more recently proposed classifications.26 Because the patients all carried a diagnosis of CHF, we believed that these abnormal echocardiographic findings were sufficient to categorize them as having diastolic dysfunction.
In terms of healthcare policy, the critical finding of unchanged healthcare expenditure has significant implications. On the basis of previous studies, we hypothesized that there would be a reduction in overall healthcare utilization and cost in the DM group. On the contrary, our results showed that no area of healthcare utilization decreased, including acute hospitalizations, emergency department visits, office visits, procedures, or medication use. Total healthcare cost did not differ between the DM group and the control group. We base our findings on an extensive collation of data from a variety of sources, not simply on extrapolation from hospital discharge records. We collected approximately 53 000 individual charges, and the cost data were equally distributed among the groups. Of note, we found that the lack of cost difference persisted even in patients with higher NYHA class, suggesting that the improved survival in sicker patients was not associated with cost savings, an observation that has been reported previously.10,24
From a conceptual viewpoint, DM will be most useful when there is a marked disparity between the best-case management for a disease and the management that is being applied. Earlier DM trials were performed when use of guideline-based therapy was less prevalent than it is currently. Although many prior trials have noted prescription of ACE inhibitors in 30% to 50% of HF patients,11,17,27,28 77% of our patients were on an ACE inhibitor or angiotensin receptor blocking drug at the time of enrollment. As such, the margin for improvement in management is smaller, and the value added by a DM program will be more difficult to demonstrate. Although there continues to be considerable room for improvement in outpatient management of CHF, our findings suggest that the startling reductions in hospitalizations and healthcare costs reported by some earlier investigators9,17,18 may not be achievable at present. Our results suggest that DM programs will be most useful when targeted to patients with worse functional class.
We used a commercially available DM program that has been shown to be cost saving in prior trials.9,18 This approach is superimposed on the routine care of the patient and requires that alterations in management be ordered by the patient’s PCP. A second challenge facing DM is that physicians may not welcome input from disease managers. A small number of potential patients for our trial withdrew from considering participation after they were advised by their physicians they should not enroll, some of whom stated they would no longer see the patient if the patient participated in DM. Despite guideline-based recommendations, upward titration of drugs was not optimized in many patients. Whether or not DM can be more effective at improving physician adherence to guideline recommendations than other systematic efforts at education and compliance monitoring continues to remain unclear.
In conclusion, in a large, community-based CHF population, we found that DM provided a significant improvement in survival, most notably in patients with systolic dysfunction, particularly in those in NYHA classes III and IV. DM did not reduce hospitalizations or cost and did not increase exercise performance. Differences in response were not present in patients with more advanced CHF, and DM had little impact on patients with diastolic dysfunction. Our results suggest that DM improves survival, but as a cost-saving measure in CHF its widespread use may not be warranted.
This study was supported by a grant from the US Department of Defense, US Army Medical Research Acquisition Activity. Clinical space was provided by TEAM Research of Seguin, McKenna Neighborhood Clinic, and Hill Country Medical Associates. The authors also thank Dina Goytia-Leos, MD, Mario Garza, MD, Devinda Desoyza, MD, Greg Olsovsky, MD, Shayne Skarda, MD, Jody Jordan, MD, Stephen Dunder, MD, and Monica Lee, MD, for their assistance in data collection. Special thanks to Dr Marilyn Harrington and Dennis Driscoll, RN, without whose early support the project would not have been possible, as well as the staff of the University of Texas Disease Management Center, who worked tirelessly to complete this study.
Presented in part at the 77th Scientific Sessions of the American Heart Association, New Orleans, La, November 7–10, 2004.
Kornowski R, Zeeli D, Averbuch M, Finkelstein A, Schwartz D, Moshkovitz M, Weinreb B, Hershkovitz R, Eyal D, Miller M. Intensive home-care surveillance prevents hospitalization and improves morbidity rates among elderly patients with severe congestive heart failure. Am Heart J. 1995; 129: 762–766.
Jaarsma T, Halfens R, Huijer Abu-Saad H, Dracup K, Gorgels T, van Ree J, Stappers J. Effects of education and support on self-care and resource utilization in patients with heart failure. Eur Heart J. 1999; 20: 673–682.
Fonarow GC, Stevenson LW, Walden JA, Livingston NA, Steimle AE, Hamilton MA, Moriguchi J, Tillisch JH, Woo MA. Impact of a comprehensive heart failure management program on hospital readmission and functional status of patients with advanced heart failure. J Am Coll Cardiol. 1997; 30: 725–732.
Hanumanthu S, Butler J, Chomsky D, Davis S, Wilson JR. Effect of a heart failure program on hospitalization frequency and exercise tolerance. Circulation. 1997; 96: 2842–2848.
Tilney C, Whiting SB, Horrar JL, Perkins BD, Vance RP, Reeves JD. Improved clinical and financial outcomes associated with a comprehensive congestive heart failure program. Dis Management. 1998; 1: 175–183.
Cohn JN, Goldstein SO, Greenberg BH, Lorell BH, Bourge RC, Jaski BE, Gottlieb SO, McGrew F III, DeMets DL, White BG, for the Vesnarinone Trial Investigators. A dose-dependent increase in mortality with vesnarinone among patients with severe heart failure. N Engl J Med. 1998; 339: 1810–1816.
Pitt B, Poole-Wilson PA, Segal R, Martinez FA, Dickstein K, Camm AJ, Konstam MA, Riegger G, Klinger GH, Neaton J, Sharma D, Thiyagarajan B. Effect of losartan compared with captopril on mortality in patients with symptomatic heart failure: randomised trial—the Losartan Heart Failure Survival Study ELITE II. Lancet. 2000; 355: 1582–1587.
Mejhert M, Kahan T, Persson H, Edner M. Limited long term effects of a management programme for heart failure. Heart. 2004; 90: 1010–1015.
Cline CM, Israelsson BY, Willenheimer RB, Broms K, Erhardt LR. Cost effective management programme for heart failure reduces hospitalisation. Heart. 1998; 80: 442–446.