Original Scientific Paper. Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Transcription

1 Original Scientific Paper Five-year follow-up findings from a randomized controlled trial of cardiac rehabilitation for heart failure Jacky Austin a, Wynford Robert Williams b, Linda Ross b and Stephen Hutchison a a Gwent Healthcare Trust, Nevill Hall Hospital, Abergavenny, Monmouthshire and b School of Care Sciences, University of Glamorgan, Pontypridd, UK Received 4 May 2007 Accepted 10 August 2007 Background A follow-up study was carried out on the 5-year status of the surviving patients (n = 179 at 6 months) of a 24-week randomized controlled trial comparing cardiac rehabilitation (CR) with heart failure outpatient clinic care (standard care). Methods In the original randomized controlled trial, 200 patients (60 89 years, 132 men) with New York Heart Association II/III heart failure confirmed by echocardiography had been randomized ( ). At the 5-year follow-up, the initial trial measures (6-min walk test, Minnesota living with heart failure, EuroQol health-related quality of life, and routine biochemistry) were repeated if the patient was in a satisfactory condition. Data on deaths and admissions were obtained from the medical records department. Results Over half of the original participants (n = 119, 59.5%) were alive at 5 years (mean age 75.2 years), and most (94%) attended the clinic for assessment. A sustained improvement from baseline for both groups in Minnesota living with heart failure, but not in EuroQol was observed, and the majority of the other measures had deteriorated. In contrast to the CR group, the standard care group showed a significant deterioration in walking distance (5 versus 11%; P < 0.05). More patients in the CR group were taking regular exercise (71 versus 51%; P < 0.05). No significant differences between the groups in health care utilization or survival were observed. Conclusion A 24-week CR programme for patients with stable heart failure showed some long-term benefit at 5 years. Differences in the mean values of most of the functional and quality of life measures were evidently to the advantage of the CR group, which also showed a better exercise profile. Eur J Cardiovasc Prev Rehabil 15: c 2008 The European Society of Cardiology European Journal of Cardiovascular Prevention and Rehabilitation 2008, 15: Keywords: cardiac rehabilitation, elderly, functional status, health-related quality of life, heart failure Introduction Patients with chronic heart failure suffer from a complex and progressive disease with impaired physical function and health-related quality of life (HRQL), and a poor prognosis [1 4]. A coordinated multidisciplinary heart failure-management programme, bridging the gap between community and acute care, can achieve a substantial improvement in health for patients at high risk of readmission [5]. This case-management approach tends, however, not to cater to stable heart failure patients who are less symptomatic; a broader strategy is Correspondence to Jacky Austin, Gwent Healthcare Trust, Nevill Hall Hospital, Abergavenny, Monmouthshire, NP7 9SA, UK Tel: ; fax: ; jackie.austin@gwent.wales.nhs.uk required. Current research objectives focus on identifying the components or combinations of treatment interventions that achieve the maximum benefit for patients [6]. Exercise training can benefit patients with stable heart failure [New York Heart Association (NYHA) classes II III] by reducing morbidity and mortality [7]. In the UK, the National Institute for Clinical Excellence guidelines (2003) state that patients with chronic heart failure should be encouraged to adopt regular aerobic and/or resistive exercise [8]. Cardiac rehabilitation (CR) provides an individualized programme of exercise, education, and psychological and social support in the contexts of both secondary and primary care, and is integral to the majority of cardiology c 2008 The European Society of Cardiology

2 Cardiac rehabilitation for heart failure RCT Austin et al. 163 departments [9]. Poor logistics and lack of resources contribute to the limited uptake of heart failure patients into CR, but specific issues relating to patient benefit also remain unclear [10]. In clinical practice, the majority of patients are elderly, with multiple comorbidity: their under-representation in clinical trials might reduce confidence in the application of research findings to practice. We have previously reported on the findings from a randomized controlled trial (RCT) of CR in elderly patients with stable heart failure [11]. The 6-month evaluation of the programme showed significant improvements over standard care (SC) in functional performance and HRQL. The intervention group (CR) had significantly fewer admissions and spent fewer days in hospital; mortality was similar for both groups. The achievement of favourable patient outcome measures might, however, have been in the short term. Improved mortality and reduced hospital admissions have been demonstrated in an Australian study of case-managed patients (median follow-up was 4.2 years) [12]. Otherwise, there has been limited assessment of the impact of heart failure CR on the functional performance and HRQL of these patients in the longer term. Within this context, we have evaluated 5-year data relating to the surviving participants from our original trial on CR, supplementary to a nurse-led clinic. Participants and methods As described previously, we conducted a randomized study in South-East Wales (UK). Patients in the original RCT were recruited ( ) from a district general hospital cardiology outpatient clinic (60%), from the hospital medical wards (25%), and from general practice if they met the following study criteria: left-ventricular systolic dysfunction (ejection fraction < 40%) confirmed by echocardiography, NYHA class II III, and age above 60 years. Patients were excluded from the original study if they had significant comorbidity that would prevent entry into the study because of terminal disease or inability to exercise. No significant differences in baseline measures at entry between the study groups were observed; mean ages were 71.8 years (SC) and 71.9 years (CR), with the respective participations of men being 64% and 67%. The aetiology of heart failure was primarily coronary (77%), with 15.5% hypertensive and 5.5% myopathy components. Echocardiography, graded semiquantitatively, revealed that the majority of patients had moderate (45%) or severe (38.5%) left-ventricular systolic dysfunction. Atrial fibrillation was evident in 29% of the sample. The mean 6-min walking distance was 267 m. With regard to medication, 88% were on loop diuretics, and 82% were on angiotensinconverting enzyme inhibitors. The study intervention was as follows. Under the supervision of a cardiologist, patients in the SC group received an 8-weekly monitoring of clinical status, optimization of pharmacotherapy, and self-care education in a specialist nurse outpatient clinic. In addition to SC, the CR group attended a twice-weekly 8-week multidisciplinary programme, including prescriptive exercise, individualized health education and information, and psychological and social support. Transport was provided by the hospital for those unable to make their own way. Partner and carer participation was at the discretion of the patients. The exercise component, based on the standards set for CR, followed recommendations for exercise training in chronic heart failure patients [13,14]. Patients performed aerobic exercise training and low resistance training/high repetitive muscular strength work in a circuit format. To encourage exercise at the prescribed level, patients were issued with guidance, to facilitate exercise an additional three times per week at home. Patients graduated from the programme to a 16-week community-based exercise class. Run by a British Association of CR fitness instructor, the weekly 1-h sessions took place in local community halls. No adverse events occurred during the exercise component. For the 5-year follow up, eligible patients who had given written informed consent were invited to attend for an assessment interview to provide clinical and psychosocial data. Information on the number of deaths in the original cohort and hospital admissions was obtained from the records department of Nevill Hall Hospital. Ethical approval for this study was obtained from the Gwent Local Research Ethics Committee. Endpoints and statistical analysis The primary endpoints of the original trial and the 5-year follow-up relate to functional status (NYHA class I IV), functional performance (6-min walk test), and perceived exertion (Borg rating of perceived exertion). HRQL measures, in terms of the disease-specific Minnesota living with heart failure (MLHF) and cost-utility questionnaires (EuroQol and EuroQol-visual analogue scale) were also repeated. The EuroQol responses were subjected to further analyses, which were based on each component part: mobility problem, self-care problem, usual activity problem, pain/discomfort, and anxiety/ depression. The secondary endpoints included healthcare use (number and duration of cardiac-related hospital admissions), routine biochemistry, and prescribed heart failure medication. A current exercise history was taken from each patient. All-cause mortality was also recorded, and Kaplan Meier survival curves were constructed using time-dependent all-cause survival for both groups. Data were entered on SPSS (Statistical Package for the Social

3 164 European Journal of Cardiovascular Prevention and Rehabilitation 2008, Vol 15 No 2 Sciences, version 14.0, SPSS Inc., Chicago, Illinois, USA) for statistical analysis. Statistical tests for parametric (paired and independent t-test, analysis of variance) and nonparametric data (w 2 test) were used, to evaluate baseline and 5-year data for the same patients (n = 112) in the SC and CR groups. The level of significance was set at P < Results The results focus on the differences in the primary and secondary endpoints within the SC and CR groups at baseline and at 5 years. Of the 200 original participants, there were 179 patients (SC 94, CR 85) remaining in the study at 6 months, following attrition (n = 12) and death (n = 9). The distribution of patient numbers at 6 months and at 5 years is given in Fig. 1. Between 6 months and 5 years, there were an additional 60 all-cause deaths in addition to the nine within the first 6 months (SC 38; CR 31; total 34.5%). Seven patients declined to take part, leaving 112 patients in the 5-year follow-up group. Unless otherwise indicated, the follow-up results are representative of 55 (SC group) and 57 (CR group) patients. Mean ages of the sample at baseline and at 5 years were 71.8 and 75.2 years, respectively and there was no change in sex distribution (66% male). Figure 2 shows the group changes in NYHA class over the 5-year period. Most patients (CR 86%, SC 76%) remained within NYHA classes II and III at 5 years. The main NYHA category changes from baseline to 5 years were no change (CR 37%; SC 51%), deterioration by one class (CR 33%; SC 31%), and improvement by one class (CR 25%; SC 9%). The level of prescribing at baseline and at 5 years, respectively, remained the same for angiotensin-converting enzyme inhibitors (82 and 85%) and for diuretics (85 and 85%); it increased by 19% for b-blockers (36 and 55%). Fig. 1 Standard care n=100 Baseline Cardiac rehab n=100 Withdrew n=2 Withdrew n=10 Died n=4 Died n=5 6 Months n=179 Standard care n=94 Cardiac rehab n=85 Withdrew n=5 Withdrew n=2 Died n=34 Died n=26 Standard care n=55 Cardiac rehab n=57 5 Years n=112 (56%) Distribution of patient numbers. Fig. 2 % SC(0) CR(0) SC(6) CR(6) SC(60) CR(60) NYHA4 NYHA3 NYHA2 NYHA1 SC and CR group NYHA changes at 0, 6, and 60 months. CR, cardiac rehabilitation; NYHA, New York Heart Association; SC, standard care. With regard to the 6-min walk test, 16% of the SC (n =9) and 5% of the CR (n = 3) groups were unable to participate owing to physical and symptomatic limitations. The walk test distance differential between the groups increased from 4 to 10% in favour of the CR group at 5 years (Table 1). At 5 years, the CR group demonstrated better Borg pretest and posttest values. Values for blood biochemistry were within the accepted range for the sample (data not shown) with no significant group difference apart from Na + [mean (SD) values: SC, (3.7) mmol/l; CR, (5.6) mmol/l; P < 0.05]. The MLHF HRQL measure revealed a sustained, although small, improvement on baseline mean scores for both groups in the physical and emotional subsets and in total score. The improvement seemed to be at a higher level for the SC group because of higher baseline scores. With regard to the EuroQol scores, there was some deterioration for both groups, and more so for the CR group, which had a higher baseline value. No change from baseline, for either group, on the EuroQol-visual analogue scale was observed. Further analysis of the EuroQol revealed that there were no significant differences in the distributions of the CR and SC groups of patients within the mobility, self-care, usual activity, or anxiety/depression subsections, at baseline or at 5 years (Table 2). At 5 years, there were higher proportions of patients in the most severe categories, apart from those in the anxiety/ depression category. In the extreme pain/discomfort and unable-to-perform usual activity categories, over 5 years the proportions more than doubled, respectively to 14% (CR group) and 18% (SC group). Hospital admissions were 39% higher in the CR group (n = 53, range 0 6) than in the SC group (n = 38, range 0 9). The difference in values for admissions into the

4 Cardiac rehabilitation for heart failure RCT Austin et al. 165 Table 1 Within-group primary endpoint data at baseline and at 5 years Standard care (n = 55) Cardiac rehabilitation (n = 57) Mean (SD) P value Mean (SD) P value Walk test (m) Baseline (107.8) (85.9) 5 years (126.6) 0.02* (114.2) 0.26** Borg RPE Pretest Baseline/5 years 2.1/ * 2.2/ ** Posttest Baseline/5 years 1.7/ * 1.9/ ** Minnesota scores Physical baseline 23.0 (11.0) 20.4 (11.4) 5 years 19.3 (12.5) (11.2) 0.26 Emotional baseline 8.9 (6.6) 8.7 (6.9) 5 years 7.6 (7.1) (6.5) 0.26 Total baseline 41.5 (21.7) 39.7 (23.8) 5 years 37.1 (24.9) (21.7) 0.28 EuroQol scores Baseline 0.66 (0.23) 0.69 (0.23) 5 years 0.60 (0.34) (0.32) 0.02*** EuroQol-vas scores Baseline 59.0 (17.8) 63.5 (17.6) 5 years 58.5 (21.6) (18.9) 0.99 Scoring range: Borg 0 = no effort, 10 = most effort. Minnesota total (0 best, 105 worst), physical (0 40), emotional (0 25); EuroQol thermometer (1 worst, 100 best); EuroQol score (1.0 best 0.4 worst) n =46*, 54**, 56***. P values relate to within-group comparisons at baseline and at 5 years. RPE, rating of perceived exertion. Table 2 EuroQol component profiles of standard care (SC) and cardiac rehabilitation (CR) groups Baseline n (%) 5 year SC CR SC CR Mobility problem None 12 (22) 19 (34) 14 (25) 16 (29) Some 43 (78) 37 (66) 39 (71) 40 (71) Confined to bed 0 (0) 0 (0) 2 (4) 0 (0) Self-care problem None 44 (80) 46 (82) 38 (69) 43 (77) Some 11 (22) 10 (18) 15 (27) 13 (23) Unable to wash 0 (0) 0 (0) 2 (4) 0 (0) Usual activity problem None 19 (35) 19 (34) 17 (31) 18 (32) Some 33 (60) 34 (61) 28 (51) 31 (55) Unable to perform 3 (5) 3 (5) 10 (18) 7 (13) Pain/discomfort None 19 (35) 23 (41) 19 (35) 21 (38) Moderate 31 (57) 30 (54) 28 (52) 27 (48) Extreme 4 (7) 3 (5) 7 (13) 8 (14) Anxiety/depression None 29 (53) 31 (55) 35 (63) 36 (64) Moderate 24 (44) 24 (43) 18 (33) 18 (33) Extreme 2 (4) 1 (2) 2 (4) 2 (3) Group total (n) a a Missing data for one CR group patient. CR (mean 0.9, SD 1.5) and SC (mean 0.7, SD 1.7) groups did not achieve significance (P = 0.53). The number of inpatient days was 50% lower for the CR group (n = 231, range 0 38) than for the SC group (n =463, range 0 188). The difference in values for inpatient days in the CR (mean 4.1, SD 8.3) and SC (mean 8.4, SD 29.9) groups did not achieve significance (P = 0.19). Estimated median survival time, from Kaplan Meier survival curves, was 32 months for the CR group in comparison with 23 months for the SC group (P = 0.28). More patients in the CR group were taking regular exercise at 5 years (71 versus 51%; P < 0.05, n = 103) with 36% taking independent plus community classes (phase IV exercise) in comparison with 11% in the SC group. Discussion The 6-month evaluation of the original RCT showed significant gains for the intervention group in functional performance, HRQL, and reduced hospital admissions and length of stay. The findings from this study suggest that these short-term improvements were not completely lost at 5 years. In particular, MLHF scores remained better than at baseline, and the CR group showed better values for exercise and most HRQL parameters, and inpatient days were reduced. Patients in both groups had a high baseline level of comorbidity, consistent with other studies [15,16]. At entry, participants had stable heart failure, and during the initial 6 months, they received optimal pharmacotherapy and self-management education. As a consequence, the number of deaths (n = 69, 34.5%) during the 5-year period equates to the middle point of the suggested annual mortality rate of 10 50% [17]. The estimated 5-year survival rate for patients with heart failure is 50%, and this decreases to less than 20% in those aged 80 years and above [18]. The 6-min walk test offers a convenient, economical, and clinical test of functional performance that is comparable with normal activity [19]. In this study, the benefit of CR over the longer term is evident with respect to walking

5 166 European Journal of Cardiovascular Prevention and Rehabilitation 2008, Vol 15 No 2 distance and perceived exertion: it seems to have slowed down the deterioration from baseline performance. This is an encouraging result, as improvement and sustainability of functional performance is the primary aim of CR. Particularly for the elderly, enhancement of physical functioning has positive effects on their ability to perform everyday activities and maintain independence [20]. As with the initial study, there remains a good relationship between the 6-min walk test and NYHA class. At 5 years, 24% of the patients in the SC group were within NYHA class IV as opposed to 11% of those who received CR. Further evidence of a worsening symptomatic status for the SC group is provided by the greater number of patients who were unable to undertake the walk test and the deterioration in the Borg rating of perceived exertion pretest ratings. Where CR programmes are not tailored to the individual, participation is known to be poor [21]. Even in the shortterm, compliance to exercise training for elderly patients with heart failure is known to be generally inadequate [22]. The relatively high number of patients in this study exercising at 5 years, either independently or in a community class, might be indicative of the patients perceived value of regular exercise to their well-being, and explains the improved Borg posttest values of the CR group at 5 years. In agreement with other studies, patients reported a moderate to low HRQL overall [23]. Patients, particularly those with long-established disease generally value HRQL above life span [24]. As in the original study, the information gained from using the combination of a utility measure EuroQol, which permits a trade-off between HRQL and life expectancy and a diseasespecific measure should provide a broader perspective from which to judge the value of an intervention. It is an expectation that disease-specific measures such as the MLHF are more sensitive in detecting clinically important changes over time. Nonetheless, EuroQol identified a deterioration within both trial groups at 5 years. As demonstrated in the subcomponent analysis, this might be explained by an increase in the patients experiencing extreme pain/discomfort; symptoms not particularly related to heart failure, not detectable by the MLHF, and perhaps not relevant to walking performance. We previously reported that the strongest effect of the intervention occurred during the first 8 weeks, at the end of its most intense phase, and the postintervention exercise taken by the patients has certainly been less rigorous. In our previous study, the EuroQol showed a significant improvement in quality of life for the CR group only, a differential that had been lost at 5 years. In this study, the EuroQol is also measuring the symptoms and impairments of more advanced age; hence, a discrepancy in relation to the disease-specific measure is not surprising. With regard to the MLHF scores, there is no regression below the baseline mean values and total scores for both groups remain slightly positive. The extent to which regular clinical review and education alone for NYHA II/III heart failure can have impacted on an anticipated negative trajectory in HRQL is debatable, as the benefits of health education on HRQL weaken from the point of delivery [25]. The benefits of a 6-month CR programme, as demonstrated in our original study, are consistent with other short-term heart failure multidisciplinary interventions that exclude exercise [26,27]. Several confounding factors (primarily the length of the evaluation period, heterogeneity in programme management, and prioritization of the different measures), however, make betweenstudy comparisons difficult. The diverse, but clinically stable population in our study (chosen to reflect typical referral to CR) seems to have a lower risk of prolonged hospital stay. This is particularly evident when the findings are compared with a prospective long-term study of heart failure patients (n = 297; age > 55 years; leftventricular ejection fraction < 55%; and NYHA II/III and IV, 9.5%: at least one admission for acute heart failure) promoted the health and economic benefits of a homebased intervention in Australia [28]. The Australian study reported 5-year unplanned readmissions to be 480 (usual care) and 395 (home-based intervention), with a 72% overall mortality rate (84% in the usual care group). This study has several limitations. Naturally there are methodological issues, as it was not designed to be a longterm longitudinal study; as such, it is underpowered. Although the number of dropouts was comparatively small, selective dropouts by patients who felt either too ill or too well to participate might have biased the results. Some contamination between the study groups was also observed; the SC group had the subsequent option to attend CR, and this was taken up by five patients. Categorizing hospital admissions as all-cause cardiac might not give a true picture of the treatment effect in terms of lack of differentiation between those admissions and planned cardiac admissions to prevent further clinical deterioration. The study contains no economic evaluation of the CR programme. An acknowledgement of the requirement for programme-specific outcomes, that is, the necessary expertise for successfully delivering prescriptive exercise both within the outpatient and the community setting for this particularly vulnerable group of patients, is also given. In response to the call for a reduction in the everincreasing burden on healthcare resources, the focus of heart failure management has undoubtedly been on those

6 Cardiac rehabilitation for heart failure RCT Austin et al. 167 patients severely affected by the disease. The success of recent strategies cannot be denied, but with the emphasis on the gold-standard management of patients with chronic to end-stage heart failure, little has progressed for those with early to moderate forms of the disease. Nearly a decade has passed since the rationale of routine CR for patients with heart failure was first raised. CR has the potential to improve the longterm outlook for patients presenting with the early stages of heart failure. More long-term, suitably powered studies are required to evaluate this form of intervention. Acknowledgements The source of support for this study is the Chief Medical Officer s Budget (Wales). The authors gratefully acknowledge the support of professor Lawrie Moseley. Conflicts of interest: none declared. References 1 Hobbs FDR. Unmet need for diagnosis of heart failure: the view from primary care. Heart 2002; 88:9ii 11ii. 2 Cleland J. Contemporary management of heart failure in clinical practice. Heart 2002; 88 (Suppl ii) Stewart S, MacIntyre K, Hole D. More malignant than cancer? Five year survival following a first admission for heart failure in Scotland. Eur J Heart Fail 2001; 3: Holland R, Battersby J, Harvey I, Lenaghan E, Smith J, Hay L. Systematic review of multidisciplinary interventions in heart failure. Heart 2005; 91: Driscoll A, Worrall-Carter L, Stewart S. Rationale and design of the national benchmarking and evidence based national clinical guidelines for chronic heart failure management programs study. J Cardiovasc Nurs 2006; 21: McDonald K, Conlon C, Ledwidge M. Disease management programs for heart failure: not just for the sick heart failure population. Eur J Heart Fail 2007; 9: Extra Match Collaborative. Exercise training meta-analysis of trials in patients with chronic heart failure (ExTraMATCH). BMJ 2004; 328: National Institute for Clinical Excellence. Management of chronic heart failure in adults in primary and secondary care. Clinical guideline 5. London: NICE Jollife J, Rees K, Taylor R, Thompson D, Uldridge N, Ebrahim S. Exercisebased rehabilitation for coronary heart disease. Cochrane Database of Syst Rev 2001; 1:CD Clark A. Exercise and heart failure: assessment and treatment. Heart 2006; 92: Austin J, Williams R, Ross L, Moseley L, Hutchison S. Randomised controlled trial of cardiac rehabilitation in elderly patients with heart failure. Eur J Heart Fail 2005; 7: Stewart S, Horowitz J. Home based intervention in congestive heart failure: long-term implications on readmission and survival. Circ 2002; 105: Working group on cardiac rehabilitation and exercise physiology and working group on heart failure of the European Society of Cardiology. Recommendations for exercise training in chronic heart failure patients. Eur Heart J 2001; 22: Chartered Society of Physiotherapy. Standards for the exercise component of phase III cardiac rehabilitation. London: The Chartered Society of Physiotherapy; Cowie M, Wood DA, Coates AJ, Thompson SG, Poole-Wilson PA, Suresch V, et al. Incidence and aetiology of heart failure: a population based study. Eur Heart J 1999; 20: Blue L, Lang E, McMurray JJ, Davie AP, McDonagh TA, Murdoch DR, et al. Randomised controlled trial of specialist nurse intervention in heart failure. BMJ 2001; 323: Goldberg R, Ciampa J, Lessard D, Meyer TE, Spencer FA. Long-term survival after heart failure: a contemporary population based perspective. Arch Intern Med 2007; 16: Kelly J, Kelleher K. The electrocardiogram in heart failure. Age Ageing 2000; 29: Lucas C, Stevenson LN, Johnson W, Hartley H, Hamilton MA, Walden J, et al. The 6-min-walk and peak oxygen consumption in advanced heart failure: aerobic capacity and survival. Am Heart J 1999; 138: Stokes H, Thompson D, McGee H. Guidelines for cardiac rehabilitation. British Association for Cardiac Rehabilitation. London: Blackwell Science; Brodie D, Bethell H, Breen S. Cardiac rehabilitation in England: a detailed national survey. Eur J Cardiovasc Prev Rehabil 2006; 13: Corvera-Tindel T, Doering L, Gomez T, Dracup K. Predictors of noncompliance to exercise training in heart failure. J Cardiovasc Nurs 2004; 19: Bennet S, Oldridge N, Eckert G, Embree L, Browning S, Hou N. Comparison of quality of life measures in heart failure. Nurs Res 2003; 52: Bennet SJ, Huster G, Baker SL, Milgrom LB, Kirchgassner A, Birt J, Pressler ML. Characterisation of precipitants of hospitalisation for heart failure decompensation. Am J Crit Care 1998; 7: Stromberg A. The crucial role of patient education in heart failure. Eur J Heart Fail 2005; 7: Cline C, Israelsson B, Willenheimer R, Broms K, Erhardt LR. Cost effective management programme for heart failure reduces hospitalisation. Heart 1998; 80: Stewart S, Marley J, Horowitz J. Effects of a multidisciplinary home-based intervention on unplanned readmissions and survival among patients with chronic congestive heart failure: a randomised controlled trial. Lancet 1999; 354: Inglis S, Pearson S, Treen S, Gallasch T, Horowitz JD, Stewart S. Extending the horizon in chronic heart failure: effects of a multidisciplinary, home-based intervention relative to usual care. Circulation 2006; 114: