A facilitated home-based cardiac rehabilitation intervention for people with heart failure and their caregivers: a research programme including the REACH-HF RCT

Intervention development aims (work packages 1A and 1B combined)

The aims of the intervention development were to:

• establish the support needs of people with HF and their caregivers

• use a systematic intervention development framework (intervention mapping) to develop a home-based CR intervention that is theory based and evidence informed, involving people with HF and their caregivers

• develop a training course to support intervention delivery.

Feasibility study aims

The aims of the feasibility study were to:

• assess the feasibility and acceptability of adding the REACH-HF intervention to usual care for patients with HFrEF, their caregivers and intervention facilitators

• assess the fidelity (quality) of delivery of the REACH-HF intervention

• assess the feasibility of outcome data collection processes and measure completion/attrition rates

• identify any changes needed in the REACH-HF intervention or facilitator training.

Framework for intervention development

Following the UK Medical Research Council guidance for developing complex health-care interventions,49 we used a systematic, evidence-informed approach to develop the REACH-HF intervention. Our approach was based on intervention mapping – a six-step systematic framework for intervention development:50

1. Step 1 – ‘needs assessment’ to identify targets for change.

2. Step 2 – building matrices to ‘map’ change targets against determinants of the desired changes.

3. Step 3 – selection of appropriate behaviour change techniques and strategies to address each determinant identified in step 2.

4. Step 4 – production of detailed intervention and training materials.

5. Step 5 – anticipating adoption and implementation of the intervention.

6. Step 6 – plans for evaluation of processes and effects.

Intervention mapping seeks to ground the intervention in the context and the population to be targeted, as well as the existing evidence base. A key element of the intervention development process was the inclusion of the REACH-HF PPI group, which consisted of six people from Cornwall with a range of experiences with HF and three caregivers of people with HF. An overview of the intervention development process is provided in Figure 2, and the following sections provide a summary of the first five steps of the intervention mapping process. The process evaluation plans (step 6) are described elsewhere (see Report Supplementary Material 3 and Work package 3: process evaluation). 39

FIGURE 2.

Development of the REACH-HF intervention. CG, caregivers; Cm, commissioners; HP, health-care providers (e.g. general practitioners, cardio specialists, nurses); P, patients; qual, qualitative; quant, quantitative.

Step 1: needs assessment/identifying targets for change

Identification of needs

The process began by assessing the needs of patients with HF, their caregivers and the service providers. This included gathering information on the problem, its causes and the target population and then developing a ‘causal model’ outlining the main modifiable factors that might contribute to an improvement in HRQoL for people with HF (see Figure 2).

Reviews of qualitative and quantitative literature provided a starting point for assessing the self-care support needs of patients and caregivers. An ongoing literature search (updated every 2–3 months) identified published reviews of self-care and rehabilitation interventions for people with HF from 1994 onwards. In addition, two de novo systematic reviews were undertaken by the project team: a meta-ethnographic synthesis of qualitative literature on the attitudes, beliefs and expectations of people with HF receiving CR51 and a systematic review and meta-analysis of the efficacy and safety of CR in people with HFpEF. 52 We also reviewed national and international clinical guidelines for HF recommended by our project management group, including the ESC21 and NICE practice guidelines. 27 The key recommendations on behaviour change, information needs or other changes needed to improve the HRQoL of patients or caregivers were extracted, along with potential self-care strategies and potential determinants of such changes.

A number of systematic reviews and guidelines highlighted the importance of exercise-based rehabilitation as a central element in driving positive outcomes in HF. 14,21,27,52 As a result, a specialist working subgroup of project team members (PD, SS, KJ, JA, CG and RT) met several times (along with extensive e-mail interaction) to develop and refine the exercise and physical activity components of the intervention.

We conducted focus group interviews with two community-based HF support groups and attendees at a hospital-based rehabilitation class. Each group included 12–20 patients and 6–10 caregivers. The main topic areas were ‘coming to terms with heart failure’; benefits of and barriers to exercise/physical activity; problems and solutions associated with taking medications; information and support needs; advice for family members or caregivers; and how the REACH-HF intervention should be delivered.

To further elicit the views of key stakeholders, a needs assessment questionnaire was circulated to 10 people with HF and 24 other experts in the field, including two behavioural scientists, 14 specialist nurses (HF, CR and primary care cardiac nurses), two cardiologists, two general practitioners (GPs), two exercise physiologists with CR experience and two pharmacists. This was an opportunity sample based on contacts known to the REACH-HF project management group and people in the focus groups who had volunteered to complete the questionnaire. The questionnaire (which is published alongside the intervention development paper39) included questions about what outcomes are important for people with HF; self-care behaviours that should be targeted; information and support needs; suggested content and delivery formats; and who might deliver the intervention. Respondents were also asked how the manual could be adapted for a range of users, including those with HFpEF.

The REACH-HF PPI group helped to design the topic guide for the focus group interviews, completed and commented on the needs assessment survey and commented on summaries of information from the focus groups. The group met every 2 months throughout the 12-month needs assessment stage, with additional e-mail and postal correspondence between meetings.

A qualitative research study involving face-to-face, semistructured interviews with a purposive sample of 26 caregivers of people with HF with a range of sex, age and socioeconomic status was also conducted to specifically identify caregivers’ needs. 40

Understanding the context or community in which an intervention is delivered is another important aspect of needs assessment. 50 A member of the research team (Wendy Armitage from the Heart Manual Department) conducted site visits to HF treatment centres and a range of staff at four sites (Truro, York, Birmingham and Abergavenny) and administered a questionnaire on current service provision. 53 This identified existing strengths, limitations, competencies and capacities of potential providers.

Two team members (Carolyn Deighan and Michelle Clark from the Heart Manual Department) reviewed further literature to identify evidence on the effectiveness of relaxation and mindfulness interventions for people with HF (and other chronic illnesses) to inform the stress management component of the Heart Failure Manual. Finally (just prior to implementation in the feasibility study), a training needs questionnaire was sent to HCPs who had been selected to deliver the intervention to assess their current state of knowledge/expertise with regard to key elements of the intervention. This was used to tailor the training course in step 4.

Analysis and integration of needs assessment data

A key challenge was to summarise and integrate the data and ideas from many diverse sources. We did this using a framework for mixed-mode evidence synthesis called triangulation protocol. 54 First, a thematic synthesis of the needs assessment documents and recordings was used to generate a ‘needs assessment’ table, which listed the key recommendations from each evidence component. We then considered where the recommendations from each source agreed (convergence), offered complementary information on the same issue (complementarity) or seemed to be contradictory (dissonance). Where there was dissonance, we resolved this through further discussion with the members of the project management and PPI groups. The focus of the data synthesis was on identifying (1) targets for change and (2) modifiable determinants of the changes suggested. This analytical process was conducted separately for patients and caregivers.

The themes identified by the above synthesis were organised into a logic model55 for the intervention (Figure 3). This was developed by grouping the targets for change into broad themes (behavioural, environmental, social and psychological) and mapping them onto a generic causal modelling framework for intervention development (the PRECEDE model50,56). It was acknowledged that environmental and contextual factors (e.g. home environment, social support networks) might affect HRQoL directly or indirectly (via interaction with behavioural or psychological factors).

FIGURE 3.

Logic model/behavioural specification for the REACH-HF intervention. ADL, activities of daily living.

Prioritisation and intervention focus

The targets for change (Box 1) were then prioritised through a process that included noting the level of agreement between stakeholders from the needs assessment table, consultation with the project PPI group and further discussion within the project team. We took into account the strength of the evidence base and the potential for improving HRQoL. The highest priority targets for change (shown in dark blue in Box 1) were then grouped into the following five categories:

1. engaging in exercise training to build (and maintain) cardiovascular fitness

2. managing stress, breathlessness and anxiety

3. HF symptom monitoring (and associated help-seeking), particularly in terms of managing fluid status

4. taking prescribed medications

5. understanding HF.

BOX 1 - Developing the REACH-HF HF Manual: targets for change

Targets for change

Engage in exercise training (to improve cardiovascular fitness).

Longer-term physical activity (to maintain cardiovascular fitness).

Manage stress/anxiety.

Manage fluid status (over and under hydration).

Monitor signs and symptoms: seek help appropriately.

Take medications.

Manage breathlessness.

Understand HF.

Manage low mood.

Manage fatigue.

Healthy eating.

Manage/live with uncertainty (about possible deterioration/decompensation events/end of life).

Sleep well.

Maintain social activities/social roles.

Weight management.

Manage severe depression.

Stop smoking.

Manage alcohol intake.

Manage/organise the home and work environments.

Manage financial burden/organising benefits.

Manage comorbidities (other illnesses) that might affect the ability to manage HF.

Manage and respond appropriately to devices (e.g. implantable cardiac defibrillator, cardiac resynchronisation therapy), including managing anxiety around devices.

Understand medications/treatments.

Get vaccinations.

Engage in sexual activity if desired.

Dark blue font indicates full coverage (core topic and important for all).

Light blue font indicates needs-based intervention (topic important for some but not all patients).

Orange font indicates case management approach (topic important for some but needing external input).

Light orange font indicates information only (topic peripheral or of relatively minor importance in most cases).

Item 5 (understanding HF) was included following step 3 (see below), as it was identified as a core determinant underpinning engagement with the first four targets for change.

The project management group and PPI group agreed that the above core priorities should receive strong focused support from the intervention facilitator and that the intervention manual should contain interactive elements to support change in these areas (e.g. for exercise training, we included a choice of a walking programme or chair-based exercise programme, as well as interactive tools for goal-setting and self-monitoring).

A second set of targets (the light blue text in Box 1) were identified as important for some but not all patients (e.g. smoking cessation and healthy eating). It was agreed that these aspects should be assessed and (briefer) intervention from the facilitator provided if needed.

A third set of targets, although important for some patients, were deemed to be outside the remit of the provider (e.g. management of severe depression) or possible to address through existing services (e.g. smoking cessation). It was agreed that these topics would be dealt with using a case-management approach.

A fourth set of targets were categorised as peripheral or minor topics, such as vaccination. For these topics, the patient was assessed, given some information and, if needed, signposted to further agencies or information (e.g. websites).

Caregiver targets for change and prioritisation

The core priorities for the caregiver resource were to:

1. facilitate improvement in HRQoL for the person with HF by helping them to achieve the core priorities for change for patients (see Box 1)

2. improve the quality of life for caregivers by acting to maintain their own health and well-being.

The target of ‘understanding heart failure’ was also felt to be of core importance in underpinning engagement with the above targets. The targets for change for caregivers that emerged from the needs assessment process are shown in Box 2. A focus group with four caregivers was conducted to prioritise the targets in the same way as described for the patient manual (see Analysis and integration of needs assessment data).

BOX 2 - Developing the REACH-HF Caregiver Resource: targets for change

Caregiver targets for change

Support monitoring and management of signs/symptoms of HFa (e.g. when to ‘step in’).

Understand HF. a

Understand and manage medicines. a

Exercise/physical activity promotiona (facilitate the HF Manual exercise programme as well as maintenance).

Provide emotional support to help manage patient’s mood, anxiety, stress.

Manage caregiver stress and emotional consequences.

Manage own physical health^a (including physical activity/lifestyle, health problems, safe lifting).

Become a caregiver (psychological and practical adaptation, role negotiation).

Sleep well.

Deal with emergencies (including resuscitation issues).

Communication with health professionals.

Manage/organise the home and work environments – daily hassles.

Maintain own social activities/social roles.

Monitor signs/symptoms of depression of cared-for person. a

Understand and manage comorbidities. a

Provide and engage social support (e.g. offer information, encouragement and practical support; engage friends, relatives, carer groups).

Live with uncertaintya (e.g. possible deterioration/decompensation events, end-of-life issues).

Attend to physical needs of the cared-for person (practical tips).

Manage and respond appropriately to devices (e.g. implantable cardioverter defibrillator, cardiac resynchronisation therapy), including managing anxiety around devices. a

Engage social services (e.g. carer needs assessment).

Manage financial burden/organising benefits for caregiver.

These self-care issues are also dealt with in the HF Manual, and relevant sections are referenced from the caregiver resource.

Dark blue font indicates full coverage (core topic and important for all).

Light blue font indicates brief, needs-based intervention (topic important for some but not all patients).

Orange font indicates case management approach (topic important for some but needing external input).

Feasibility study results

A summary of the findings is available in Appendix 1. The feasibility study recruited 23 patients and 12 caregivers, and seven intervention facilitators provided the intervention.

• Patients and caregivers were highly satisfied with the REACH-HF intervention and attendance at sessions was high (all patients attended at least three face-to-face sessions and typically received four telephone contacts).

• A number of potential modifications to the content of the manual and facilitator training were identified.

• Intervention fidelity scoring of all of the delivered consultations for 18 cases indicated adequate delivery for most aspects of the intervention by all facilitators [mean scores ranged from 2.6 (item 11, caregiver health and well-being) to 5.0 (item 1, person-centred delivery)] and are presented in in the feasibility study report (see Report Supplementary Material 1, Table A).

• Two items [addressing emotional consequences of being a caregiver (item 10) and caregiver health and well-being (item 11)] had suboptimal delivery–fidelity scores.

• We also computed fidelity scores collated per facilitator, and these are presented in Table 5. These represent delivery for only three patients per facilitator and so should be treated with caution. Despite this, we observed reasonable consistency between facilitators on most items.

• Levels of outcome completion were generally excellent, and patients/caregivers perceived relatively low measurement burden.

• A number of patient and caregiver outcomes following the REACH-HF intervention showed evidence of improvement (with all of the caveats of a small population of selected participants and uncontrolled comparisons).

• No patient or caregiver safety concerns were identified.

• Physical fitness [measured through the Incremental Shuttle Walk Test (ISWT)] was successfully carried out by the majority of patients. As per previous studies, older patients with multiple comorbidities showed some improvement, but this did not achieve statistical significance over time.

A number of ideas for improving the text of the HF Manual and the training were identified, such as changing the name of the original ‘Caregiver Resource’ to ‘Family and Friends Resource’ to promote engagement with the intervention, as many cohabitees did not identify themselves as a ‘caregiver’. Following processing of the above data, the HF Manual (including the Family and Friends Resource and Progress Tracker) and training course were substantially revised. This final version was then evaluated in Work package 3: multicentre randomised controlled trial of the REACH-HF intervention in patients with heart failure with reduced ejection fraction.

The REACH-HF intervention

The REACH-HF intervention is a comprehensive self-care support programme comprising the ‘HF Manual’, including a choice of two exercise programmes for patients, a Family and Friends Resource for caregivers, a Progress Tracker booklet and a facilitator training course. For patients, the main self-care targets are engaging in exercise training, monitoring for symptom deterioration, managing stress and anxiety, managing medications and understanding HF (see Box 1 for an expanded list). Secondary targets include managing low mood and smoking cessation (where relevant). For caregivers, the main intervention targets are supporting self-care for the patient and looking after their own physical and mental health (see Box 2). The intervention is facilitated by trained HCPs with specialist cardiac experience (e.g. a HF nurse or physiotherapist) over 12 weeks via home and telephone contacts. The four main REACH-HF intervention elements as described in Figure 4.

FIGURE 4.

The REACH-HF intervention materials.

Training course for facilitators

A training manual/syllabus for a 3-day training course for the REACH-HF intervention facilitators was developed. Facilitators were defined as professionals with experience in CR or cardiac nursing. The facilitation role is crucial to the success of the REACH-HF programme. As well as being the main delivery process, it enables tailoring of the REACH-HF intervention resources to the individual needs of patients and their caregivers. The course includes the theory and process of facilitation (building rapport using patient-centred counselling techniques,66 empowerment and support of self-management, building understanding of the condition61); using behaviour change techniques; techniques for managing stress and anxiety; contents of the manual; supporting exercise and physical activity using the intervention materials; and facilitation of the Family and Friends Resource and medical/nursing issues. The training was linked by three case studies of patients with HF and opportunities to practise facilitation techniques and to problem-solve potentially difficult situations.

A set of quotations or ‘patient voices’ selected by patients and caregivers in the PPI group from our prior qualitative work was incorporated throughout the written resources in the form of speech bubbles to help illustrate key points (Figure 5).

FIGURE 5.

Extract from the HF Manual showing a ‘patient voice’ callout.

Summary

Intervention mapping gave a clear structure and process for developing the REACH-HF intervention and the associated training programme for intervention facilitators. The process took into account the needs of a range of stakeholders, including patients with HF, their caregivers, HCPs, potential facilitators and health-care commissioners.

The construction of a causal model as part of the intervention mapping method (see Figure 2) was useful as a framework to integrate the identified needs and define the intervention’s ‘targets for change’. However, as reported by other studies that have used intervention mapping to develop complex health-care interventions,73–75 the overall process was time-consuming and resource intensive.

Successes

To our knowledge, this is the only publication that provides a detailed description of the theoretical and evidentiary basis, intervention techniques and strategies for an intervention to promote quality of life in people with HF. The intervention was co-developed with patients, caregivers, clinicians and academics to optimise it for use with patients with HF in real-world CR settings, and this was a particular strength of the research. In particular, PPI ensured that the intervention was tailored to individual needs based on a diverse range of patient backgrounds, knowledge levels and severity of HF.

Limitations

The complexity of the intervention development process may affect replicability, and it is unlikely that a different team of collaborators using the same methods would have produced exactly the same intervention.

Conclusions

We developed a comprehensive, evidence-informed, theoretically driven, self-care and rehabilitation intervention that is grounded in the needs of patients and caregivers. The intervention was well received by patients and caregivers and was feasible for and acceptable to the nurses and physiotherapists who delivered it. The intervention was delivered to an acceptable standard, although some revisions to the intervention and its associated training course were suggested. After these revisions, the REACH-HF intervention was ready for evaluation in a full-scale trial.

Introduction

About 50% of patients with clinical features of HF are estimated to have HFpEF. 76 Moreover, the prevalence of HFpEF is increasing, and both HFpEF and HFrEF pose a substantial burden on patients in terms of reduced exercise intolerance, poor HRQoL and mortality. 77 For the NHS, this is evident through increased hospital admissions and higher health-care costs.

Recent systematic reviews and meta-analyses have shown promising evidence for the benefit of exercise-based CR in HFpEF. The CR programmes undertaken in these trials were predominantly group-based, supervised and delivered in centre-based settings. There is increasing recognition of the possibility of alternative delivery models of CR, such as home-based programmes, to overcome the suboptimal rates of CR uptake seen with HF.

A single-centre pilot RCT (ISRCTN78539530) was conducted in Dundee to assess the feasibility and acceptability of the novel REACH-HF intervention for patients with HFpEF and their caregivers. Detailed reports of the rationale and protocol of this pilot RCT and its findings have been published (see Appendix 1) and a report of the process evaluation has been pre-print published43 (see Appendix 1).

Research aims

The main aim of this pilot study was to assess the feasibility of undertaking a definitive RCT to assess the clinical effectiveness and cost-effectiveness of the REACH-HF intervention in patients with HFpEF and their caregivers. Our objectives were to:

• assess the acceptability of the study design and procedures to participants (patients and caregivers)

• assess the feasibility and experience of the delivery of intervention for participants and HCP facilitators

• identify barriers to participation in the intervention and study procedures

• inform a definitive study sample size

• assess methods for the collection of data, including resource use and costs

• assess the fidelity of delivery of the REACH-HF intervention by cardiac nurses.

Methods

A total of 225 patients were approached, of whom 50 were randomised (1 : 1) to the REACH-HF intervention plus usual care (intervention group) or to usual care alone (control group) (see Report Supplementary Material 1, Figure B). Participants were randomly allocated in a 1 : 1 ratio to either intervention or control group arms without stratification or minimisation. Randomisation numbers were computer-generated and assigned in strict sequence. At the point of randomisation, participants were assigned the next randomisation number in the sequence. To maintain concealment and minimise selection bias, randomisation was performed after the baseline visit by a member of Peninsula Clinical Trials Unit (CTU), independent from investigator teams, using a secure, web-based randomisation system.

As described earlier, the REACH-HF intervention is a home-based comprehensive self-care CR programme that comprises patient and caregiver manuals with supplementary tools and is delivered by trained health-care facilitators over a 12-week period. The REACH-HF intervention was originally designed for patients with HFrEF in terms of coverage of medication and explanations of the condition, as there was limited evidence to guide the development of the REACH-HF intervention for patients with HFpEF. Following a review of clinical guidance and consultation with the expert cardiology, primary care and physiology/CR specialist members of our project management group, the intervention was adapted for this feasibility study to allow evaluation in patients with HFpEF. Modifications were minor and were as follows: (1) we excluded information on beta-blockers, angiotensin-converting enzyme (ACE) inhibitors and angiotensin receptor antagonists, as these drugs have not been shown to be of prognostic benefit in patients with HFpEF, and (2) we changed text relating to the description of the condition to include a more comprehensive explanation of what HFpEF is, what causes it and what treatments are commonly offered (and why). The self-care advice in all other sections of the REACH-HF intervention was considered to be relevant to all patients with HF and corresponds to national HF guidelines. 5,21 Patient outcomes were collected by blinded assessors at baseline and at 3 and 6 months after randomisation, and included HRQoL (primary) and psychological well-being, exercise capacity, physical activity and HF-related hospitalisation (secondary). Outcomes were also collected for caregivers. The parallel process evaluation includes interviews with patients and caregivers (see Work package 2: process evaluation).

Analysis

Our sample size was adequate to enable an estimate of attrition, estimates of the standard deviation (SD) of the primary and secondary outcomes to inform power for a future definitive trial and sufficient numbers of participants for qualitative interviews. We report the mean and SD (or relevant summary statistics) for both groups for all patient and caregiver outcomes at each follow-up point and the mean (and 95% CIs) for between-group differences in outcomes at 6-month follow-up using linear regression models adjusting for baseline outcome.

Key findings

The study recruited 50 symptomatic patients with HFpEF (LVEF ≥ 45%) (mean age 73.9 years, 54% female) and 21 caregivers. Study retention and intervention uptake were excellent (90% and 92%, respectively) (see Report Supplementary Material 1, Figure B).

At 6 months, data from 45 patients showed a mean between-group difference in favour of the intervention group for the primary outcome, MLHFQ score, although the CI was wide and included 0 (between-group mean difference −11.5, 95% CI −22.8 to 0.3). A total of 11 patients (four in the intervention group and seven in the control group) were admitted to hospital over the 6 months of follow-up, with four of these admissions (all in the control group) being related to HF. In the context of this study being a single-centre pilot and not fully powered to demonstrate between-group differences in patients or caregivers, these findings should be considered indicative.

These findings support the feasibility of and rationale for delivering the facilitated, home-based REACH-HF intervention for patients with HFpEF and their caregivers and progression to a full multicentre RCT to test its clinical effectiveness and cost-effectiveness (see Report Supplementary Material 4).

Successes

We recruited to target (Figure 6); the intervention was well received by patients, caregivers and health-care facilitators, and intervention adherence was good. At follow-up, the primary outcome (MLHFQ) showed a significant improvement, and a number of other patient outcomes showed a direction of effect in favour of the intervention group over control, although this pilot trial lacked power to detect a statistically significant between-group difference.

FIGURE 6.

Work package 2: target vs. actual number of participants randomised for REACH-HFpEF pilot trial.

Limitations

Given that this was a pilot RCT, there were several limitations:

• Originally, this study was going to be run in Birmingham, but the local NHS trust was facing challenges in agreeing to recruit research staff during a staff recruitment freeze, so the research team enrolled a centre in Dundee that holds the Scottish database for patients with HF, including patients with HFpEF. An excellent rate of initial recruitment faltered as the number of patients from the database who were potential participants diminished, and we had to seek patients from outpatient medical clinics, which enabled us to recruit to target after extending the recruitment period by 3 months (see Figure 6).

• As the trial was conducted in a single centre in a population lacking ethnic diversity, the generalisability of our findings is limited.

• Patient and clinician blinding was not possible in this study because of the nature of the intervention, although we did show that it was possible to blind outcome assessors to group allocation.

• An issue raised in patient interviews was the method of exercise testing in the trial, as several patients found that the ISWT exacerbated symptoms such as fatigue and shortness of breath – 27% were unable to undertake the ISWT at final follow-up.

• The open-label design of the study may have resulted in improvements in patient-reported outcomes in intervention participants because of placebo effects.

Recommendations for future research

This pilot study produced compelling evidence for a full multicentre RCT to assess the clinical effectiveness and cost-effectiveness of the REACH-HF intervention in patients with HFpEF and their caregivers.

Trial registration

This trial is registered as ISRCTN78539530.

Introduction

Process evaluation is becoming an important part of research when assessing complex interventions such as the REACH-HF intervention (see Work package 1: intervention development and feasibility study). Currently, evidence is lacking about the experience and needs of patients living with HFpEF and their caregivers, the factors that influence their self-care decision-making and the provision of support by caregivers. 78 The complexities of living with HF in the real world and the challenges of self-care have not been adequately addressed in the literature. 79

Research aims

The aims of this process evaluation were to:

• assess the intervention fidelity delivered by the facilitators to the patients with HFpEF

• explore the experiences of patients and caregivers who participated in the REACH-HF, single-centre pilot RCT (see Work package 1: intervention development and feasibility study).

Methods

A mixed-methods process evaluation was completed alongside the single-centre pilot REACH-HFpEF RCT (see Work package 1: intervention development and feasibility study). The process evaluation consisted of a quantitative assessment of the fidelity of intervention delivery and a qualitative exploration of HFpEF patients’ and their caregivers’ experiences of the REACH-HF intervention involving semistructured interviews.

Data collection

Data were collected in parallel with the REACH-HFpEF pilot RCT (see Work package 1: intervention development and feasibility study).

Analysis

Key findings

Successes

Overall, this process evaluation suggested that the REACH-HF intervention was generally well delivered and was well received by patients with HFpEF. Qualitative data also captured some excellent examples of good practice in education, engagement and support of these patients and their caregivers by the facilitators. The findings highlight a genuine need for support in a population currently excluded from many HF and CR services.

Limitations

The small number of patients included in the fidelity analysis (n = 6), the characteristics of the participants (who were predominantly of white ethnic origin) and the single-centre setting limit the potential generalisability of the findings. Translating complex interpersonal interactions into numerical scores within the fidelity analysis does not fully illustrate some of the excellent examples of good practice. Facilitators often demonstrated high levels of skill and competence in providing tailored educational and psychological support, enabling patients to reframe negative thoughts, engage in appropriate exercise and participate in self-care. The study also highlighted how involving caregivers was challenging at times, so more could be done to enhance facilitator training on supporting caregivers.

Recommendations for future research

Given the small numbers in this study, we recommend repeating the process evaluation as part of a large multicentre RCT of the REACH-HF intervention in patients with HFpEF.

Trial registration

This trial is registered as ISRCTN78539530.

Introduction

Chronic HF is a burgeoning global health challenge that affects 1–2% of adults in the Western world. 3,6 Globally, the prevalence of HF is increasing,3 with the annual economic burden predicted to grow to more than US$108B (£82B, conversion as at 27 June 2018: 1 USD = 0.760133 GBP), as the population ages. Hospital admissions are a key driver of the rising costs,3 with the emphasis of treatment over the past two decades on pharmacological therapies and devices.

A Cochrane systematic review and meta-analysis of exercise-based CR in patients with HF showed improvements in HRQoL and a reduction in rehospitalisations. In 2010, national guidance in the UK recommended centre-based CR (supervised group exercise) for patients with HFrEF,27 with similar recommendations in Europe and the USA. 11,81 However, < 20% of people with HF in Europe and < 10% in the USA participate in CR, prompting a call to explore newer strategies to improve participation in CR and explore the clinical effectiveness of alternatives to centre-based CR.

Home-based CR programmes can widen access and have been shown to be as clinically effective as centre-based CR after MI and coronary revascularisation and with similar costs. 82 The high cost of treating people with HF is well documented,3 but little evidence is available on the cost and cost-effectiveness of home-based CR in HF. 83

We therefore developed a home-based CR programme – ‘the REACH-HF intervention’ – for people with HF and their caregivers.

Research aims

The primary aim of this multicentre RCT was to test the hypothesis that addition of the REACH-HF intervention to usual care would improve disease-specific HRQoL for patients with HFrEF at 12 months’ follow-up compared with usual care alone.

Methods

We targeted men and women aged ≥ 18 years with a confirmed diagnosis of HFrEF on echocardiography or angiography (LVEF < 45%). Patients were recruited from primary and secondary care settings in four centres in the UK (Birmingham, Cornwall, Gwent and York), excluding those who had undertaken CR within 12 months prior to the study and those with a contraindication to exercise testing or exercise training. Participants were randomly allocated in a 1 : 1 ratio, stratified by investigator site and baseline plasma NT-proBNP levels (≤ 2000 pg/ml vs. > 2000 pg/ml), using minimisation to facilitate balance between the groups. Randomisation numbers were computer-generated and assigned in strict sequence at the point of randomisation. To maintain concealment, the Peninsula CTU used a password-protected, web-based randomisation system to allocate participants after consent was obtained and baseline assessment data were entered.

Between January 2015 and February 2016, we approached 1161 patients, 216 of whom were randomised to the intervention (REACH-HF intervention plus usual care) or the control (usual care alone) (see Report Supplementary Material 1, Figures D and E). Participants were randomly allocated (see Report Supplementary Material 1, Figure D) and stratified by investigator site and baseline plasma NT-proBNP levels.

The REACH-HF intervention was uniquely co-developed with key stakeholders: patients, caregivers and clinicians (see Work package 1: intervention development and feasibility study). Briefly, the components of this complex intervention (see Work package 1: intervention development and feasibility study for more detail) included the HF Manual for patients, a Progress Tracker for the patient, a Family and Friends Resource for caregivers and facilitation by cardiac nurses or physiotherapists. Development of the intervention took just over 12 months before it was fully evaluated in our multicentre RCT.

In the trial, participants in both the intervention group and the control group continued with usual medical management and care of HF according to local and national guidelines. In most places in the UK, including the four centres participating in our RCT, usual care involved follow-up by the GP with input from a HF specialist nurse and, in some cases, a cardiologist.

Key findings

The primary analysis included 92 (86%) patients in the REACH-HF group and 93 (85%) in the control group. Table 4 summarises primary and secondary patient-reported outcomes and secondary objective outcomes at baseline and follow-up. Compared with control, the REACH-HF intervention was associated with a reduction in total MLHFQ score (i.e. improved HF-related HRQoL) at 12 months: between-group difference –5.7 (95% CI –10.6 to –0.7; p = 0.025). At 12 months, there had been eight deaths (four in each group, 4% of the total). The number of participants with at least one hospital admission was 19 in the REACH-HF group, compared with 24 participants in the control group (odds ratio 0.72, 95% CI 0.35 to 1.51; p = 0.386). Overall, there were 33 hospital admissions in the REACH-HF group (four related to HF) and 35 (10 related to HF) in the control group. The maintenance score on the SCHFI, a measure of self-care, was in favour of the REACH-HF intervention group at 12 months (p < 0.001). Within-group improvements from baseline were seen in the REACH-HF group for HADS anxiety and depression, ISWT and SCHFI (management and confidence) but did not reach statistical significance compared with the control group at 12 months. No differences were seen in the other secondary outcomes [i.e. EuroQol-5 Dimensions (EQ-5D), HeartQoL and physical activity]. The mean cost of the REACH-HF intervention was £418.39 per participant.

Recommendations for future research

With the growing availability of evidence-based CR interventions delivered at home, systematic collection of real-world data utilising vehicles, such as the NACR database in the UK, are needed to allow the tracking of potential future changes in uptake of and adherence with CR. In addition, there is a need for further well-conducted RCTs of home-based CR interventions to collect and report long-term outcomes, including hospital admissions and health-care costs.

Trial registration

This trial is registered as ISRCTN86234930.

Introduction

Caregivers have an important role in supporting chronic disease management, although this often comes at considerable personal cost and many caregivers are uncertain about how to best provide this support. Most caregivers have health and well-being problems of their own – either pre-existing or arising from the burden of their caregiving activities. 85 However, there is paucity of evidence on interventions that support caregivers in their role. 78 The needs of caregivers in the management of HF and their responses have been depicted in two recent qualitative studies. 40,86 Caregivers may also assist patients in appraising and responding to symptoms of HF. 87

Research aims

The aims of this evaluation were to:

• compare psychological well-being, quality of life, self-care activities and burden between caregivers in the intervention and control groups of the REACH-HF multicentre RCT

• explore the views of caregivers on their experience of using the REACH-HF caregiver manual (Family and Friends Resource) and its influence on their confidence in caring for patients with HF and themselves.

Methods

Patients aged ≥ 18 years with a confirmed diagnosis of HFrEF (LVEF < 45%) were recruited from four centres in the UK (Birmingham, Cornwall, Gwent and York) and invited to participate. At study entry, patients were asked whether or not they had a caregiver, that is a partner, other relative or friend. If the patient did have a caregiver who was aged ≥ 18 years, the caregiver was invited to participate in the trial alongside the patient. Participating caregivers provided unpaid support to patients who could otherwise not manage without such support. Caregivers were allocated to the REACH-HF group or control group in accordance with the random allocation of their accompanying patient.

Key findings

The REACH-HF group included 107 patients and the control group included 109 patients. A total of 97 patients declared having a caregiver, with 53 caregivers allocated to the REACH-HF group and 44 to the control group. At 12 months, data were available for 45 (85%) caregivers in the REACH-HF group and 37 (84%) in the control group.

Improvements from baseline at 4, 6 and 12 months were seen for a number of outcomes in the REACH-HF group, including HADS anxiety; CBQ-HF physical, social life and lifestyle; and all SCHFI dimension scores. However, only SCHFI confidence at 12 months was statistically significant (p < 0.05) compared with the control. Based on interviews of 17 caregivers, the REACH-HF intervention was acceptable and was associated with an increase in confidence in their caregiving role.

Recommendations for future research

The quantitative and qualitative findings for caregivers (see Work package 3: process evaluation) suggest the need to better engage caregivers in CR and self-care interventions for patients with HF. Our results suggest that caregivers require skilled and personalised facilitation by HCPs. Facilitators may need a caregiver assessment tool to help plan care. 40 Further high-quality evidence is required to confirm the benefits of actively involving caregivers in the development and delivery of CR and self-care interventions for HF and other chronic diseases.

Trial registration

This trial is registered as ISRCTN86234930.

Introduction

As mentioned earlier (see Work package 2: process evaluation), process evaluation is becoming an important part of research when assessing complex interventions, such as the REACH-HF intervention (see Work package 1: intervention development and feasibility study). However, RCTs of CR for HF have not included process evaluation, so evidence is lacking about the characteristics of CR programmes that consistently affect the success of self-care interventions across different outcomes, how different elements of the elements of CR programme interact, and the complexity of living with HF and the burden of both self-care and adherence to intervention regimens.

Research aims

The aims of this process evaluation were to:

• assess the intervention fidelity and patients’ and caregivers’ experiences (see Work package 3: caregiver outcomes) of participation in the REACH-HF multicentre RCT (see Work package 3: patient outcomes44)

• investigate the change processes that may be responsible for changes in the primary outcome measure of the RCT – MLHFQ.

Methods

A mixed-methods process evaluation used linked data to build individual case studies with links between different available data, paying particular attention to fidelity (the consistency of what is implemented with the planned intervention) and adaptations (alterations made to an intervention in order to achieve contextual fit) (see Report Supplementary Material 1, Figure G).

Data collection

Data were collected in parallel with the REACH-HF multicentre trial (see Work package 3: patient outcomes44).

Analysis

Key findings

The process evaluation sample was broadly representative of study participants. On average, 7.05 (SD 2.6) facilitated sessions were provided per process evaluation participant, of which 4 (SD 1.8) were face to face and 2.94 (SD 2.6) were by telephone.

Successes

Intervention attendance was high, with > 90% of participants receiving the minimum intended dose; this compares well with six group-based CR programmes, which typically report attendance levels of about 80% of intended contacts. How well the intervention was perceived by the participant depended not on the ‘dose’ (i.e. the contact time with the facilitator) but on other factors related to delivery of the REACH-HF intervention (Table 6). The appropriate ‘dose’ of the REACH-HF intervention was contingent on fidelity and skilled facilitation rather than the duration or frequency of contact alone. Importantly, in the case studies that we analysed, the fidelity score at the initial ‘assessment’ session was indicative of the overall fidelity for an individual participant over the course of the intervention. Intervention effectiveness was dependent on the facilitator’s style of engagement and the extent to which the trial materials were matched or adapted to the existing beliefs and goals of participants.

Limitations

Several factors were identified as possible reasons why participants may not have benefited from the intervention. Behaviour change seemed sustained when the facilitator adapted the intervention to the needs of the participant. The facilitators’ competence was key to participants’ acquisition of skills and changes in behaviour. Most participants had multiple comorbidities. Some participants valued the simplicity of the REACH-HF intervention, specifically the accessibility of the information provided. Socioeconomic factors influenced the level of engagement with the intervention – sustaining change seemed easier for participants with favourable financial or social circumstances. The main REACH-HF RCT (see Work package 3: patient outcomes) and this process evaluation included very few (< 3%) ethnic minority participants.

The fact that data were not collected during the facilitator training and that interviews with facilitators were not recorded meant that we had limited insight into the interaction between a practitioner’s existing professional repertoire and their engagement with and delivery of the REACH-HF intervention.

Recommendations for future research

Future research could explore ways of optimising the ‘workability’ of the REACH-HF intervention and integration of its components into the facilitators’ clinical practice to ensure that fidelity and adaptation are maximised in the dynamic and complex context of managing patients with HF. Given the low uptake of CR among people from minority ethnic backgrounds,88 future research should also attempt to understand the specific support needs of people from more diverse ethnic backgrounds and the extent to which cultural beliefs may moderate or mediate the effectiveness of home-based CR interventions, such as the REACH-HF intervention.

Trial registration

This trial is registered as ISRCTN86234930.

Introduction

In addition to information on the effectiveness of interventions, it is important to assess the cost-effectiveness of interventions to inform health policy decision-making within the context of the NHS in the UK. This section reports a cost-effectiveness analysis of the REACH-HF intervention versus usual care, which was carried out alongside the main trial (see Work package 3: patient outcomes).

Research aims

The aim of the trial-based economic analysis was to assess whether or not addition of the REACH-HF intervention to usual care alone (REACH-HF group) was cost-effective compared with usual care alone (control group) based on resource use, costs and HRQoL outcomes recorded for participants with HFrEF over the 12 months of the trial.

Methods

The trial-based economic analysis compared the costs associated with health and social care resource use and HRQoL of the REACH-HF intervention and control groups. Statistical analyses were used to estimate between-group differences in estimated costs and QALYs. Bootstrapping was used to provide estimates of uncertainty. Primary analysis was based on complete cases, with sensitivity analyses using imputed data, subgroups, caregiver QALYs, and informal care costs and varying the intervention cost. Full methods for the trial-based analysis are reported in Appendix 3.

Data collection

Data on resource use related to health and social care were collected from REACH-HF trial participants with HFrEF over the course of the 12 months of the trial and were costed using published estimates and national data sets. 89,90 Data on staff time, training and consumables used in providing the intervention, collected within the trial, were used to estimate the cost of the REACH-HF intervention. Data on HRQoL were collected by self-report from trial participants using the EQ-5D questionnaire.

Cost-effectiveness analyses

Cost-effectiveness was reported in terms of the incremental cost-effectiveness ratio [incremental cost-effectiveness ratio (ICER) cost per QALY] estimated over the 12 months of trial follow-up.

Key findings

The mean estimated total cost for delivery of the REACH-HF intervention was £418.39 per patient with HFrEF. We found an increase of £345 per patient for health and social care services for the REACH-HF group compared with the control group over the 12 months of the trial, although the difference was not statistically significant. No statistically significant difference was seen between the groups in EQ-5D or QALYs. With an expected additional cost associated with the addition of the REACH-HF intervention to usual care and no reported health gains, usual care alone dominates. Full results of the trial-based cost-effectiveness analysis are reported in Appendix 3.

Successes

Retention in the trial was good, with 86% of participants recruited to the trial completing and providing data for the full 12 months.

Limitations

The REACH-HF trial did not show a significant short-term improvement in generic HRQoL (EQ-5D) and it was not found to be cost-effective over the 12 months of the trial, as costs were higher in the REACH-HF group. Although meta-analyses of randomised trials studies have shown that CR can reduce all-cause and HF-specific hospital admissions, the main REACH-HF trial was not powered to detect such a change. This short-term analysis, therefore, was not able to assess whether or not there would be longer-term benefits to people with HFrEF from receiving the REACH-HF intervention.

Recommendations for future research

Robust, long-term, real-world estimates of the effectiveness of home-based CR for HFrEF in terms of hospitalisations and deaths are required to further inform the cost-effectiveness of home-based CR in patients with HF.

Optimising self-care support for people with heart failure and their caregivers: development of the REACH-HF intervention using intervention mapping

Needs of caregivers in heart failure management: a qualitative study

REACH-HF: a facilitated self-care rehabilitation intervention in patients with heart failure with preserved ejection fraction and their caregivers – rationale and protocol for a single-centre pilot randomised controlled trial

A randomised controlled trial of a facilitated home-based rehabilitation intervention in patients with heart failure with preserved ejection fraction and their caregivers: the REACH-HFpEF pilot study

Patient and caregiver experiences of the REACH-HF in patients with heart failure with preserved ejection fraction (REACH-HFpEF) study: a process evaluation

Clinical effectiveness and cost-effectiveness of the REACH-HF facilitated self-care rehabilitation intervention in heart failure patients and caregivers: rationale and protocol for a multicentre randomised controlled trial

Home-based rehabilitation for heart failure with reduced ejection fraction: the REACH-HF multicentre randomised controlled trial

Facilitated self-care and rehabilitation for patients with reduced ejection fraction heart failure (REACH-HF trial): impact on caregiver outcomes

Process evaluation of the REACH HF multicentre randomised controlled trial

Exercise-based rehabilitation for adults with heart failure (review)

Exercise-based rehabilitation for heart failure: Cochrane systematic review, and meta-analysis and trial sequential analysis

The cost-effectiveness of REACH-HF and home-based cardiac rehabilitation in the treatment of heart failure with reduced ejection fraction: a decision model-based analysis

Setting and perspective

The setting is the English health-care system and the perspective is that of the English NHS and Personal Social Services.

Target population and subgroups

Adult patients with confirmed HFrEF were enrolled (n = 216) and those who completed all elements of data collection for resource use and HRQoL at all follow-up points, including the last at 12 months, formed the participant group for the base-case analysis. The trial protocol specified subgroup-level analyses by trial centre, severity of HF (defined by NT-proBNP levels < 2000 vs. ≥ 2000 pg/ml), time since diagnosis of HF (< 1 year, 1–2 years or ≥ 2 years) and whether or not the patient had a participating carer.

Intervention cost

The REACH-HF intervention is a nurse-facilitated, home-based CR programme designed to enhance quality of life and self-care of people with HF and their caregivers. The intervention is described in full in Work package 1: intervention development and feasibility study. The REACH-HF intervention was delivered over 12 weeks. The trial methods, including economic analysis, were set out in the published trial protocol. 38 Resource use associated with delivery of the REACH-HF intervention was informed by early development research, with the primary item of resource use being the time input of the HF facilitator who delivered the intervention. Other resource use comprises consumables and some allowance for training of the HF facilitators. Data on each of these areas of resource use were collected within trial, with HF facilitator time input collected using a ‘facilitator contact sheet’ (see Report Supplementary Material 5) capturing contact time with participants and also non-contact time (e.g. planning and preparation) related to delivery of the intervention.

Unit costs for facilitation staff were based on the grade equivalent to ‘community nurse’ (including district nursing sister and district nurse) and nurse specialist (community) from Unit Costs of Health and Social Care for 2016. 90 These are based on Agenda for Change band 6, including salary, salary on-costs, overheads (management costs and non-staff costs including travel/transport) capital overheads and excluding costs for qualifications.

Training costs per facilitator, specific to delivery of the REACH-HF intervention, were estimated based on resource use data collected within trial by the trial co-ordinator.

Consumables costs (manuals and DVDs) were based on resource use data collected within trial by the trial co-ordinator.

Health, social care and other resource use and cost

Data on resource use related to health and social care were collected using a standardised resource use questionnaire (RUQ) (see Report Supplementary Material 6) at baseline (referring to the previous 6 months), at 4 months (referring to the 4 months since the trial started) and at 12 months (referring to the 8 months since the last assessment). The RUQ was developed and tested in early development research on REACH-HF. Data from assessments at 4 and 12 months were combined to calculate the total resource use during the 12 months of follow-up. Although resource use data for hospital admissions were collected in the self-reported RUQ, data on hospital admissions were reconciled against SAE reporting (with SAE reporting being the primary data source). Data on SAEs were used as the primary source of data on hospital admissions. In addition, the RUQ captured wider societal costs, including time off work for participants and caregivers, and unpaid caring activities carried out by the participant’s main caregiver and by other people.

Costs are reported in 2016 GBP. Unit costs per item of resource use (Table 7) were obtained from published estimates and, where necessary, inflated to 2016 prices using the Healthcare and Community Health Services index (see Report Supplementary Material 1). 90 These unit costs were then applied to the resource use reported at participant level to estimate the delivery costs associated with the REACH-HF intervention and the total costs associated with health and social care at baseline and over the 12 months of follow-up.

Note that the unit cost of a hospital admission represents the expected cost (published cost) of a typical HF or non-HF hospital admission in patients with HF (see Table 7) rather than the specific cost of the hospital admission events experienced by patients in the trial, as reported in SAEs. However, where mean length of stay in the trial was longer than that reported in the NHS reference cost item for hospital admission (HF related and non-HF), extra costs for additional bed-days were applied. For non-HF hospital admissions, we used data reported on type of non-HF hospital admission for people with HF in a large HF RCT and cost data from NHS reference costs103 to estimate a weighted cost for non-HF hospital admissions (see Table 7 and Report Supplementary Material 1).

Medication costs were estimated using data from the RUQ on self-report use (yes/no response) of key medication types (e.g. beta-blockers, ACE inhibitors and calcium channel blockers) and recommended dosage and medication costs reported in the British National Formulary (2016–17). We used recommend dosage or we applied half of the maximum dose where dose uptitration was relevant. Unit costs were those for generic products, where available; when multiple products were available in a single category (e.g. ACE inhibitors), we used a weighted cost based on prescribing information available from OpenPrescribing (2017),109 assuming that prescribing patterns reported were relevant for HF.

Where we have put a monetary value on informal care in sensitivity analyses, we used a replacement cost approach and assumed that the value of an hour of informal care was the same as the cost of an hour of care provided by social care/social services (£24 per hour; see Table 7).

Health-related quality of life

Data on HRQoL were collected by self-report for both the participants and their caregivers using the EQ-5D-5L health questionnaire, and health state values were derived using the mapping/crosswalk values for the UK to provide EQ-5D-3L values from EQ-5D-5L states. This is the method currently recommended by NICE for assessment of EQ-5D/QALYs in economic evaluation. 31 Estimates of QALYs for the 12 months of the trial were calculated using EQ-5D-5L data at baseline and at 4, 6 and 12 months, using the area under the curve method.

The primary economic end point was the QALY (derived from EQ-5D-5L) estimated over the 12 months of follow-up. The primary perspective taken here for analyses is that of the NHS and social care sector (third-party payer), although data collection on resource use provides an opportunity to consider wider participant level and societal impacts.

Cost-effectiveness

The trial-based analysis assessed the cost-effectiveness of the REACH-HF intervention plus usual care compared with usual care alone over the 12 months of the trial. No discounting of future costs or outcomes was required. Primary analyses were based on intention-to-treat complete-case analyses, consistent with the trial main statistical analysis plan (see the statistical analysis plan, which is available here: www.journalslibrary.nihr.ac.uk/programmes/pgfar/rp-pg-1210-12004/#/documentation; accessed December 2020).

We used statistical analyses to estimate between-group differences in estimated costs and QALYs. Generalised linear modelling regression models (with appropriate family and link functions) were used with adjustment for baseline values and prespecified covariates (see the statistical analysis plan, which is available here: www.journalslibrary.nihr.ac.uk/programmes/pgfar/rp-pg-1210-12004/#/documentation; accessed December 2020). Given the expected nature of cost data, non-parametric bootstrap methods were used to estimate CIs. Bootstrapping was used to resample (with replacement) 1000 times (i.e. simulating 1000 trials) from the empirical trial data, providing estimates of uncertainty around the modelled estimates of incremental cost, incremental effect and ICER. We produced a cost-effectiveness acceptability curve, plotting expectation of the REACH-HF intervention being cost-effective compared with usual care at a range of estimates of WTP per QALY gained. In the UK, using NICE guidance, if the cost of an intervention is below £20,000 to £30,000 per QALY gained it is considered value for money for the NHS.

In addition to the primary (complete-case) analysis, the following sensitivity analyses were carried out:

• per protocol [see the statistical analysis plan, URL: www.journalslibrary.nihr.ac.uk/programmes/pgfar/rp-pg-1210-12004/#/documentation (accessed January 2021)]

• using multiple imputation of missing values for aggregate costs

• by subgroups:

  • time since HF diagnosis (< 1 year, 1–2 years or ≥ 2 years)

  • severity of HF (BNP levels < 2000 vs. ≥ 2000 pg/ml)

  • trial centre

  • whether or not the patient had a participating carer

• including caregiver QALYs

• considering a wider societal perspective (including informal care and time off work costs)

• including intervention cost variation, based on estimated range of intervention cost.

A multiple imputation by chained equations approach was used, with predictive mean matching owing to the constrained (non-zero) nature of the cost data.

In addition to the above sensitivity analyses, we present a ‘post hoc’ sensitivity analysis, in which the EQ-5D-5L health state values and related QALY estimates are based on use of a ‘mapping’ algorithm to predict EQ-5D health state values from participant data on the MLHFQ. In this sensitivity analysis, estimates for the difference in EQ-5D/QALY were not derived from the directly collected/measured EQ-5D-5L scores (health status) but from the MLHFQ scores in the RCT using an unpublished mapping algorithm previously developed by Edlin et al. obtained via personal communication (Edlin R, Tsuchiya A, Brazier J. Mapping the Minnesota Living with Heart Failure Questionnaire to the EQ-5D index; 2002. Reproduced with permission from Brazier J, personal communication, 2018). This post hoc analysis was requested by the REACH-HF Trial Steering Committee.

Intervention cost

Facilitator contact sheets were completed and available for 613 contacts in 94 intervention group participants who received the REACH-HF intervention. Table 8 reports the mean number of contacts by type (face to face or telephone) and mean duration of contacts and related non-contact activity. The mean number of contacts was 6.52, and total contact time and non-contact time inputs were 5.3 hours (319 minutes) and 2.9 hours (176 minutes), respectively, with overall time input at 8.25 hours (495 minutes). Using the unit cost of £44 per hour (based on a salary of £32,114 per annum90), the estimated mean cost of facilitator input was £363. Training cost per facilitator, specific to delivery of the REACH-HF intervention, was estimated at £1547 [3 days’ training at £352 per day; cost for trainers per trainee at £366 (assuming eight trainees per 3-day course) and cost for REACH-HF facilitator manuals at £50 each plus an estimate of consumables for training sessions]. These were distributed across the first 100 participants who received the intervention to give mean costs attributed to each participant of £15.47. The patient’s HF Manual was costed at £40, based on one hard copy manual at £25 and two DVDs at £7.50 each. This gave a mean estimated total cost for delivery of the REACH-HF intervention of £418.39 (95% CI £203.64 to £729.78) per participant.

Resource use

Resource use for primary and secondary care and social services is reported in Table 9. Use of prescribed medications is reported in Table 10. Numbers of all-cause and HF hospital admissions and length of stay are reported in Table 11. Informal care is summarised in Table 12.

Costs

NHS and Personal Social Services costs are reported by allocation in Table 13. Informal care costs are reported in Table 14.

Health state values (EuroQol-5 Dimensions, five-level version) and quality-adjusted life-years

Table 15 summarises the estimated mean health state values and QALY estimates, derived from participant and caregiver reports for EQ-5D-5L health states (based on EQ-5D-3L values per tariff). We report data for assessments at baseline and at 4 and 12 months.

Cost-effectiveness: primary analyses

Table 16 summarises the primary statistical analyses undertaken to estimate the mean differences in costs (by cost category and for total cost) plus differences in participant QALYs between the REACH-HF intervention and usual care groups. The table reports unadjusted and adjusted analyses, including adjustment for baseline values and other covariates (trial centre, NT-proBNP group, previous MI and previous AF).

Excluding the intervention cost for the REACH-HF intervention, we report a difference of £345 in total costs for health and social care service use, with participants who used the REACH-HF intervention having a higher cost than usual care, although this was not statistically significant. When including the REACH-HF intervention cost in the analyses, we see an unadjusted difference in cost between the REACH-HF intervention and usual care of £516, with this estimated cost difference increasing to £763, higher cost for REACH-HF intervention participants, after adjustment.

Usual care had a better QALY profile than the REACH-HF intervention (i.e. a mean reduction of –0.024 QALYs in the REACH-HF intervention participants), but this difference was not statistically significant.

The REACH-HF intervention, with a higher expected mean cost and an absence of QALY gains, was therefore dominated by usual care and did not seem to be a value-for-money investment for the NHS in the UK in the short term.

When considering uncertainty using the bootstrapped data, Figure 7 illustrates the cost-effectiveness plane, showing that the incremental cost and QALY data (pairs) from the bootstrapped simulations are predominantly in the north-west quadrant, indicating additional costs and reduced QALYs. Figure 8 reports the cost-effectiveness acceptability curve, indicating that the REACH-HF intervention is expected to be cost-effective versus usual care only in a small number of trials/simulations (< 10%) regardless of the WTP threshold (per QALY gained) applied.

FIGURE 7.

Cost-effectiveness plane illustrating the results of 1000 bootstrap resamples of incremental cost and effectiveness with the REACH-HF intervention compared with usual care. Blue circles reflect bootstrap resample estimates and the orange diamond represents central value (mean QALY difference and mean cost difference across all samples).

FIGURE 8.

Cost-effectiveness acceptability curve showing probability that the REACH-HF intervention is cost-effective (probability the net benefit statistic > 0) compared with usual care at each WTP threshold.

Cost-effectiveness: sensitivity analyses

Table 17 presents summary statistics for the sensitivity/scenario analyses undertaken to assess the sensitivity of the primary base-case result to alternative assumptions/scenarios.

In all scenarios, except the MLHFQ–EQ-5D mapping scenario, the REACH-HF intervention remains dominated by usual care, with no reported QALY gains associated with the REACH-HF intervention.

Where we present analysis with EQ-5D values and QALYs derived from the indirect, participant-level MLHFQ mapping algorithm, we report a statistically significant difference in mean estimated QALYs of 0.026 (95% CI 0.005 to 0.047), favouring the intervention. When this approach is used alongside the analysis of costs, with a mean incremental cost of £763, we have an incremental cost per QALY of £29,346 for the REACH-HF intervention versus usual care. Figures 9 and 10 report the cost-effectiveness plane and cost-effectiveness acceptability curve for this analysis, where, when using the net benefit statistic approach, we see the probability that the REACH-HF intervention is cost-effective at WTP threshold values of £20,000 per QALY and £30,000 per QALY, respectively, at 31.3% and 50.3%.

FIGURE 9.

Cost-effectiveness plane showing 1000 bootstrapped resampled estimates of incremental cost and incremental QALY change, in the scenario where QALYs were derived from mapping between MLHFQ and EQ-5D scores using the algorithm developed by Edlin et al. (personal communication, 2018).

FIGURE 10.

Cost-effectiveness acceptability curve showing the probability that REACH-HF is cost-effective over a range of WTP thresholds, in the scenario where QALYs were derived from mapping between MLHFQ and EQ-5D scores using the algorithm developed by Edlin et al. (personal communication, 2018).

Review of literature

Through the searches, 6393 studies were identified. A total of 890 were duplicates and were removed, and 2010 were removed as they had been published prior to the end date of the previous systematic review. This left 3493 studies, of which 3350 were removed at the abstract screening stage because they did not include a health-care evaluation model (n = 2434), were not carried out on those with HF (n = 815), reported only short-term cost-effectiveness alongside a clinical trial (n = 63), were animal studies (n = 21) or were about HF prevention and not treatment (n = 17). Sixty-seven studies were then excluded because they were not available in full text in the English language (e.g. conference abstracts).

The remaining 76 full-text papers were reviewed and 24 were excluded because they did not meet inclusion criteria (no model, 14 papers) or were systematic reviews of modelling studies (10 papers), which left 52 papers. Of these, 50 described a unique economic model or assessment associated (two were additional reports from existing studies).

The results of the literature search are illustrated in Figure 11.

FIGURE 11.

Results of the literature search.

Summary of the review of heart failure modelling literature

The review identified a range of studies with varying decision contexts, modelling approaches, model structures, settings and parameters. In terms of model types, the reviewed models most commonly used Markov cohort models. Models for pharmaceutical or service-level interventions (as opposed to those for surgical devices, which tended to include a broader range of states) tended to use one of three main approaches: using a simple model with two basic disease states, alive (with HF) and dead, an alive/dead model plus additional states for one or more hospital admission, or modelling disease progression via NYHA class. Thokala et al. 112 included hospital admission (for HF and other causes) and additionally adjusted mortality for the time from hospital discharge to reflect the increased risk of death soon after discharge. Some models allowed both for progression in the underlying condition, based on NYHA grades, and for hospital admission, but this was less common. For example, Kansal et al. 113 reported a Markov cohort model with NYHA progression, as well as additional hospital admission states for both HF and other-cause hospital admissions for up to three events, allowing utility to decline with multiple hospital admissions. Two studies used an adaptation of a Markov cohort approach but modelled specific patient subgroups separately using relevant risk equations allowing for differences in treatment effect, disease progression, cost or mortality within different subgroups. Alternative approaches were used by some authors – for example, seven studies reported a discrete event simulation approach following individual simulated patients based on time-to-event data.