Improving care transfers for homeless patients after hospital discharge: a realist evaluation

What is intermediate care?

In the literature on hospital discharge, it has long been recognised that there is tension between two competing notions of ‘good practice’. On the one hand, a perspective that is narrowly concerned with the most cost-effective use of hospital resources (and hence rapid throughput). On the other hand, a perspective that emphasises the importance of needs-led assessments and choice for individuals. 13 Increasingly, intermediate care and discharge to assess (D2A) are being promoted as ways of addressing the need to ensure both ‘patient flow’ and safe discharge practice.

Intermediate or ‘step-up’ and ‘step-down’ care is an umbrella term for a wide range of admission avoidance and out-of-hospital care services, including Rapid Response, Crises Response, D2A, Home First, Reablement and Safely Home. It usually comprises:

Networks of local health, housing and social care services, which deliver targeted, short-term support to individual patients, in order to: prevent inappropriate admission to NHS acute inpatient or continuing care, or long-term residential care; facilitate timely discharge from hospital; and, most importantly, maximise people’s ability to live independently.

Cowpe18

Bolton19 summarises the thinking behind out-of-hospital care services as follows. First, there should be less focus on assessment for longer-term care and support at the point of discharge and more emphasis on recovery. Second, there is a specifically commissioned set of services to help people recover post hospital. He suggests that intermediate care will not work at its best if services are solely commissioned from existing services where they were not established for that purpose (e.g. using standard home care agencies that are not geared up to take a regular flow of new people).

Guidance is clear that ‘step-down’ care should not be used to warehouse people. That is, creating an additional transfer in a person’s care pathway to free up a hospital bed without adding value to their experience of care or meeting good outcomes for the person. 20

The National Institute for Health and Care Excellence (NICE)21 guideline for intermediate care recommends that commissioners should consider making home-based intermediate care, reablement, bed-based intermediate care and crises response all available locally. These services need to be delivered in an integrated way so that people can move seamlessly between them, depending on their changing needs.

Eligibility for intermediate care

Intermediate care entered the mainstream of health and care services delivery in England through the National Service Framework for Older People. 22 Here, it was recognised that although intermediate care services are likely to be of particular importance for older people, service planning and investment need to take account of the needs of all potential services users. Updated guidance23 defined intermediate care as support for anyone with a health-related need through periods of transition and made the point that no-one should be excluded on the basis of age or ethnic or cultural group. Specific reference was made in this guidance to the eligibility of homeless people and prisoners. 23 However, later guidance has tended to move away from this broader, more inclusive definition. One reason for this may be that investment in intermediate care services (particularly reablement) has not kept pace with rising demand. 24

Housing-led intermediate care

The £10M investment that underpinned the HHDF was restricted to voluntary and community sector providers. This targeting led to the development of a new tier of specialist ‘housing-led’ intermediate care services (i.e. services that are designed and delivered mainly by housing associations and homeless charities/voluntary sector organisations). Until this point, intermediate care had been delivered mainly as a health and social care service. 25 In total, 52 schemes were funded through the HHDF. Two principal models emerged:

1. a housing-led [Home First (floating support)] ‘step-down’ model (n = 37)

2. a housing-led (residential/bed-based) ‘step-down’ model (n = 7).

In the Home First ‘step-down’ model, HHDSs provide access to a ‘link worker’. Link workers aim to support the patient to find accommodation (i.e. somewhere safe to stay). They will also assist with other aspects of discharge planning using their knowledge of local homeless services and resources. Once the patient has been discharged from hospital, the link worker will stay in touch, providing (floating) ‘step-down’ support in the community for a time-limited period until longer-term care and support are in place.

In addition to Home First (floating support) schemes, HHDF capital funding was used to develop a number of residential (bed-based) intermediate care facilities. Two of the largest HHDF capital grants (totalling in excess of £1M) were allocated to charities to establish medical respite centres with 24-hour clinical staffing. The first HHDS failed to become operational because of difficulties finding a building. The second HHDS failed to secure revenue funding for clinical staffing once the refurbishments were complete and hence did not became operational. Five smaller residential ‘step-down’ (without 24-hour clinical staffing) schemes did become operational, but only one of these (a housing association-led scheme) is currently operational.

Clinically led teams

In addition to the ‘housing-led’ schemes, we consider, as part of this evaluation, a number of ‘clinically led’ schemes (some of which predate/or were not funded by the HHDF). These are often called Pathway Homeless Teams by virtue of their affiliation with the Pathway Charity [URL: www.pathway.org.uk (accessed 3 June 2021)]. The Pathway model is a multidisciplinary team (MDT) approach that employs general practitioners (GPs), nurses, housing workers, social workers, occupational therapists and peers (i.e. people with lived experience of homelessness). The Pathway homeless teams are usually led by the GP or senior nurse and focus on individual care co-ordination (including discharge planning). As the primary aim of the clinically led team is to maximise the window of opportunity afforded through a hospital stay, these schemes usually close cases at the point at which the patient leaves hospital (i.e. they do not routinely integrate ‘step-down’ care as part of the service model).

Step 1: literature synthesis

The first step in this realist evaluation (literature synthesis) sought to identify the many different ‘intervention resources’ [including ‘simple rules’ and key practice principles (KPPs)] that might be deployed in a HHDS to facilitate the outcome of interest. The literature was then used to develop programme theories by exploring the existing evidence for each of these ‘intervention resources’.

Step 2: mapping typological configurations

Once the research team was familiar with the full range of (possible) intervention resources, the next step was to explore HHDS documentation (project leaflets, evaluation reports, etc.) and to undertake a series of preliminary site visits and discussions with key HHDS stakeholders. The objective here was to map the ‘intervention resources’ that are actually being deployed across the 52 HHDSs and, importantly, to tease out the underpinning logic for their use (i.e. why the scheme has opted to utilise these resources and practices). This preliminary mapping also informed site selection, enabling subsequent testing of the initial tentative programme theories.

Step 3: testing and refining programme theories through empirical data collection

The third step was to test and refine programme theories through empirical data collection. This study employed a mixed-methods approach comprising three WPs, allowing for both qualitative and quantitative interrogation. We will return to discuss the design of each WP later in the report once we have specified the theories being tested. In the remainder of this chapter, we focus on how our initial tentative programme theories were generated through a realist synthesis of the intermediate care literature.

Change 1: early discharge planning

Change 2: systems to monitor patient flow

Change 3: discharge co-ordination underpinned by multidisciplinary team working, including the voluntary and community sector

Change 4: Home First (intermediate care) and discharge to assess

Change 5: flexible working patterns

Change 6: trusted assessment

Change 7: focus on choice

Change 8: improved discharge to care homes

Initial hunch

Our initial hunch was that these eight high-impact changes [or mechanism intervention resources (MIRs)] might prove to be equally important in delivering consistently safe, timely transfers of care patients who are homeless. More specifically, we hypothesised that specialist HHDSs encompassing all of most of these changes would deliver better outcomes for service users than those schemes employing fewer or none.

Testing this programme theory empirically through new research was important because much of the evidence underpinning the high-impact change model (HICM) (version 2015) was based on research with older people. At the outset of the research, very little was known about if and how the HICM might work for patients who were homeless.

How we anticipated the research might be used

In terms of the potential to lever change and improve outcomes for patients who are homeless, the HICM is an important tool. Developing pathways out of hospital is a key objective of the government’s programme for ‘integration and better care funding’. 34 The Better Care Fund (BCF) is the main source of funding for intermediate care in England and places clear expectations on health and well-being boards to oversee health and social care, including (1) pooling budgets, (2) integrating services to ensure more people can leave hospital when they are ready and (3) following guidelines laid down by the HICM. Local progress to implement the HICM is actively monitored by NHS England and NHS Improvement (NHSE&I), meaning that any research that is embedded as part of this is potentially more likely to be implemented than other research that is not linked to government priorities.

In terms of how we envisaged the research would be used, we anticipated that the evaluation findings would either confirm the relevance of the HICM in its current format or make recommendations as to how it could be sensitised to better meet the needs of homeless patients.

A conceptual platform for intermediate care

In 2013, a realist synthesis of the (generic) intermediate care literature was published. 25,27–36 This identified over 10,314 sources for potential inclusion and generated an extensive list of ‘programme theories’ for testing and refinement. An iterative Delphi-style technique was then used to arrive at a conceptual framework for intermediate care. This identified three programme theories as having the most ‘explanatory power’ when seeking understanding of how intermediate care works to improve outcomes for service users in a wide range of contexts. According to this conceptual framework, improved service user outcomes are achieved when:

• the place of care and timing of transition to it are decided in consultation with service users based on the pre-arranged objectives of care and the location that is most likely to enable the service user to reach these objectives (programme theory 1)

• health and social care professionals foster the self-care skills of service users and shape the environment so as to re-enable service users (programme theory 2)

• health and social care professionals work in an integrated fashion with each other and carers (programme theory 3). 25

However, a limitation of this review was that it did not include literature on (specialist) intermediate care services for patients who are homeless.

To fill this knowledge gap and further develop our own initial programme theory, we applied the same methods and search strategy as outlined by Pearson et al. ,25 but extended the scope of the search to include ‘homelessness’. Although Pearson et al. ’s25 search strategy used, what they believed to be, a comprehensive list of phrases relating to intermediate care, we extended the scope of this search to encompass [‘medical respite’] and [‘homelessness AND ‘hospital discharge’ (schemes)] and [‘homelessness AND ‘delayed discharge’.] This is because the term intermediate care is not widely or consistently used in the homelessness sector. The search terms used are shown in Box 1.

In terms of analysis, the overall aim of this exercise was to map how this additional evidence on specialist (homeless) intermediate care ‘speaks’ to the conceptual ‘road map’ already proposed for generic intermediate care by Pearson et al. 25 The literature synthesis is reported in accordance with the RAMESES II publication standards for realist reviews. 3

Methods used in literature synthesis

Data extraction, analysis and synthesis process

The literature was synthesised to discern how it addressed Pearson’s et al. ’s25 conceptual framework. When reviewing the literature, we sought to identify programme theories that were both explicitly argued and those that were tacit or implied, making it clear which was the case. A data extraction pro forma was designed to allow the evidence to be carefully mapped against each of the three programme theories. This included space for identification of any new programme theories. The final stage of the synthesis was to take the evidence as a whole and to reflect on the overall utility of Pearson et al. ’s25 conceptual framework, highlighting where any changes or refinements could be made.

Limitations

The limitations of a synthesis approach are that, although we have outlined our search strategy, we made judgments about the interpretations of the findings. Identifying programme theories and mechanisms from sources that are not explicitly theory driven, or do not provide adequate descriptions of the services, is also problematic and risks bias arising from the perspectives of the reviewers. Using multiple reviewers for each literature source helped address any uncertainty in the utility of the findings, challenging any bias.

Results of literature synthesis

The searches yielded 87 references, of which 48 met the inclusion criteria. Additional hand-searching revealed eight further articles. Internet-searching and direct contacts with intermediate care projects yielded 13 reports. These were mostly project reports and/or small-scale external evaluations. In total, 69 reports and articles were included in the synthesis.

Document characteristics

Appendix 1 summarises the articles and reports (n = 69) that were included in the review, the methods they used, their ‘richness rating’ and to which programme theories they aligned. Most of the literature fell into the ‘thick’ category, with few papers including a theoretical perspective. Most of the empirical evidence was from the USA and focused on medical respite. The UK evidence comprised mainly local grey literature reports and was focused mainly on the hospital discharge schemes that had been set-up with the HHDF funding.

Engagement (patient in-reach) as a distinct mechanism

Compassionate kindness, dignity and respect are values that are seen to sit at the heart of psychological- or trauma-informed care and are aspired to by many specialist HHDSs. Halligan and Hewett55 observe how a visit from an empathetic team that is dedicated to the care of homeless patients in the hospital can improve patient experience and also prevent problems such as premature discharge.

Indeed, what also emerges from this literature is the importance of professionals first ‘engaging’ or building rapport with service users and how this may act as a distinct mechanism for underpinning more formal consultative or collaborative care planning processes. As Halligan and Hewett55 observe, it is only once a relationship is established that the hard work of planning community support and negotiating with housing, social care and health-care providers and the voluntary sector can begin.

Discussing a ‘nurse-led’ residential intermediate care scheme, Dorney-Smith56 describes how up to 1 month of engagement work may be needed to counter the suspicion and distrust patients may have of professional support. In the hospital setting, this work is often conceptualised as primary care or ‘patient in-reach’. This brings highly specialist clinical knowledge and understanding of homelessness onto the hospital ward, whereby specialist GPs and nurses are employed to work side by side with hospital consultants and ward staff to raise awareness about homelessness and tackle stigma and problems, such as the prevention of early self-discharge.

One of the key mechanisms for achieving patient in-reach is the ‘homeless ward round’, in which clinicians from the homeless team will identify and support homeless patients located across the hospital site. Identification of homeless patients at an early stage of admission is key to early discharge planning. Teams will use ‘concerned curiosity’ to understand a patient’s housing circumstances, mindful of the stigma of homelessness and that some patients will not want to reveal that they are homeless. They will then work with patients in psychologically informed ways to build relationships and support them to remain in hospital, have their voices heard and complete treatment.

As noted above, when working with patients who are homeless, the main challenge is not delayed discharge but more usually preventing ‘early self-discharge’. Often this can be a result of substance misuse and the patient wanting to discharge themselves against medical advice because of the onset of unpleasant withdrawal symptoms. Withdrawal can also be at the root of much challenging behaviour on the wards, for example where the patient feels that his or her withdrawal is not being appropriately managed and leaves the ward to use street drugs. Ward staff will then often address such absenteeism from the hospital bed, issuing behavioural contracts that can lead to conflict. In such circumstances, homeless team nurses and GPs can intervene to de-escalate the situation through more appropriate skilled clinical management (e.g. correct titration of opiate substitution medication such as methadone). The homeless team will also raise awareness among ward staff about many other forms of ‘silent stigma’ that lead people who are homeless to feel uncomfortable in hospital, such as being mindful that they may have no visitors to take home their washing and provide access to clean clothes and toiletries. Ultimately, the main goal of GP- or primary care-led ‘patient in-reach’ is to maximise the benefits of a hospital admission. However, Canham et al. 47 make the point that, in the absence of adequate aftercare resources, the benefits of trauma-informed care and patient in-reach can quickly evaporate.

Tomita and Herman57 undertook a randomised controlled trial (RCT) that evidenced reduced hospital readmission rates and other positive outcomes for 150 homeless psychiatric (mental health) patients receiving a care co-ordination intervention (vs. usual care). This suggested that the relationship with the social services’ worker may be as equally an important mechanism in delivering these positive results as securing housing tenure and stability.

Place of care

The strongest correlate of hospital readmission among homeless people is discharge location. 58 Kertesz et al. 59 showed that discharge to a medical respite facility was associated with significantly lower odds of readmission than discharge to ‘own care’ (including homeless shelters). Discharge to supportive housing has similar benefits. 60

Pearson et al. 25 reported that the characteristics of the local health and social care system could significantly limit care options for service users. This was evidenced in the homelessness literature, with people who are homeless having little opportunity to influence decisions about their place of care because few options might be available. 61

Doran et al. 58,62 reported that many US respite schemes will not admit homeless people who continue to drink and whose behaviour is identified as ‘challenging’. It might be argued that the lack of system capacity in the community, geared to the realities of persons’ needs, equates to acute care being a fall-back option. According to the Housing Learning and Information Network,29 having a specialist homeless hospital discharge worker does not eradicate delayed discharge if appropriate community care is not available. However, it can work to highlight where the gaps exist. The Housing Learning and Information Network caution that finding placements for people who continue to drink and have disabilities is often challenging. 29

The notion that intermediate care might itself fill some of these gaps raises questions about scope and remit and how far this should extend into the territory of longer-term care. According to van Laere et al. ,63 the high mortality rate among users of a Dutch medical respite scheme might be explained by the fact that the homeless population in Amsterdam most commonly comprises people with mental health challenges, long-term opiate users and people misusing alcohol who are not able to live independently and depend on fragmented services. In addition, many intermediate care schemes for people experiencing homelessness currently provide palliative care to compensate for the lack of provision elsewhere. 63–65

Generic or specialist?

Mainstream intermediate care facilities may not currently meet the needs of people who are homeless. 66 The argument for ‘specialist’ provision stems, in large, from the challenges of co-housing people with different challenges and vulnerabilities. 67 For example, Lane40 charts the advantages and disadvantages of admitting people who use substances and are experiencing homelessness to mainstream intermediate care. On the one hand, it is considered that when someone is in recovery they should not be exposed to hostel environments in which drug and alcohol use is commonplace. On the other hand, it is recognised that people who use substances may have ways of being that are problematised by health and care providers and other users of intermediate care services.

Rather than being understood in the context of patient choice and the need for person-centred care planning, debates around ‘place of care’ may be conflated with potentially discriminatory assumptions about the characteristics of different user groups. For example, Lane40 reports that some GPs and hostel managers he interviewed expressed concerns that ‘homeless people’ would not mix well with other users of intermediate care because they may be older and fragile both physically and emotionally. They were, therefore, overlooking the potential for challenging behaviours among older people related to conditions such as dementia.

Safe spaces for women

A study of early exit from medical respite reported that ‘respite structure’ (i.e. rules and regulations) could make some service users feel unsafe and may account for why up to one-third of people leave medical respite earlier than planned. 41 Women in this study were significantly more likely to leave respite before discharge completion than men. According to Bauer et al.,41 gender-specific treatment models or women-only spaces could enhance safety and consequently retention outcomes.

Reablement and physical rehabilitation needs

A feasibility study reviewed the caseload of a specialist homeless primary health-care team in Ireland to assess the need for a specialist homeless intermediate care centre. 68 It found that 15% of homeless people on the caseload had mobility and disability challenges attributable to health conditions, such as stroke, hip replacement, fracture or amputation.

In the literature on homeless intermediate care, ‘re-enablement’ or ‘reablement’ were not mentioned. Many studies reported difficulties collaborating with local authority adult social care, which may indicate that local authority reablement services are not easily accessible to people who are homeless. 42,45,65,69,70

Reports also suggest that the physical rehabilitation needs of people experiencing homelessness are not well catered for. 61–71 Whiteford and Simpson65 observed that many homeless people struggled to access rehabilitation not because they were homeless, but because of their age, with many services excluding people under the age of 55 years. Indeed, there is a recognised need for improved disability access in many UK hostels for homeless people. 45–73

‘Reablement’ environments

Mainstream residential intermediate care facilities in care homes or in hospitals often provide access to specially adapted environments, such as a ‘training kitchen’, in which people can practice the activities of daily living. A complaint arising from service users in one (specialist homeless) hostel-based intermediate care facility was boredom due to the lack of any kind of structured daily activity. 64 More recently, Pathway teams in London have employed occupational therapists to address this risk by promoting meaningful activity. 50

Recovery

There is emerging consensus in the intermediate care literature specific to people experiencing homelessness that to stop the ‘revolving door’ of hospital readmissions, support needs to extend beyond the discharge process itself and into the community, either by means of a residential ‘step-down’ facility or ‘floating support’ arrangement. 45,48,54 However, what is less clear is the ideal time frame for such arrangements, which may be termed intermediate care.

In the literature on specialist homelessness intermediate care, ‘recovery’ from drug and alcohol misuse and/or mental health problems emerges as the primary rehabilitative focus. O’Carroll et al. ’s68 feasibility study, for example, found that 48% of the caseload were experiencing problematic substance misuse, 33% had mental health challenges and 17% were dually diagnosed. However, the setting of goals around ‘recovery’ raises further questions about the accepted time frames for intermediate care. Dorney-Smith,56 for example, charts how service users’ health deteriorated when they were discharged from a nurse-led intermediate scheme that provided between 6 and 8 weeks of support. Dorney-Smith56 observes that recovery is a ‘long game’ for many homeless clients and time frames may be upwards of 24 months.

Resettlement

The 6- to 8-week time frame for intermediate care is further brought into question by the multiple overlapping natures of the transitions facing service users who are homeless (i.e. from ‘hospital to home’ and from ‘homelessness to housed’). Managing the transition from ‘homelessness to housed’ encompasses both the practical aspects of securing accommodation, as well as meeting what are termed ‘resettlement needs.’74 Indeed, there are many parallels to be drawn between ‘reablement’ and ‘resettlement’ work, with the latter being ‘housing’ rather than ‘social care’ led. Both share the aim of ‘doing things with rather than for people’ and have the overall aim of promoting independence. It might even be suggested that ‘resettlement’ work has a broader, more personalised focus than ‘reablement’, as it is often encompassing of both ‘citizenship goals’, such as securing employment, education and volunteering opportunities, as well as those linked to reablement and the promotion of ‘self-care’. Describing a specialist residential intermediate care facility for homeless people in northern England providing up to 3 months of resettlement support, Lowson and Hex75 describe the goal of the service as giving people the opportunity to make real life-changing decisions and to have a ‘real go’ at their lives, improving their life chances and quality of life, as well as improving independence with daily living tasks.

Pearson et al. ’s25 review noted that one drawback with (mainstream) intermediate care is that it has tended to prioritise a desire for service users to attain certain functional goals within a specified time period over service users’ self-knowledge and desire to reach a wider set of goals over longer, less clearly defined time periods.

Multidisciplinary team skill mix

Many hospitals in England now employ discharge co-ordinators to manage the most complex discharges. As recommended in the HICM,33 these roles are usually embedded in MDTs comprising senior nurses and social workers. The main focus of their work is usually on managing transfers of care for older people requiring continuing health care and/or moves into care homes. 76 Where there are large numbers of homeless patients (≥ 200 per year), it is recommended that hospital trusts develop specialist clinically led multidisciplinary ‘homeless teams’ to work alongside existing discharge teams. 77 As noted earlier, these are often called Pathway Homeless Teams by virtue of their affiliation with the Pathway charity, and these employ GPs, nurses, housing workers and a range of other staff. A RCT of the costs and benefits of two ‘clinically led’ services (vs. ‘standard care’) indicated that this clinically led approach does not alter length of stay, but improves the quality of life of homeless people (with fewer service users returning to the streets) at a cost deemed cost-effective under current guidelines. 78

An early evaluation of the HHDF similarly concluded that those schemes taking a specialist MDT approach were more effective in delivering improved health and housing outcomes than other models. 70 Without the benefit of a clinician, some of the (uniprofessional) housing link worker projects described difficulties in engaging with what they described as the ‘medical model’. 79 Wood et al. 80 suggest that the combination of nursing and social care is particularly effective for people who have complex problems or who have experienced long-term homelessness, given that the early stages of being housed can be immensely challenging, with poor physical and mental health adding to the concomitant stress of adjusting to a very different way of life. Case study evidence also highlights the important role of the GP in the MDT, particularly in advocating for homeless patients’ needs in a holistic way where there may be pressure from senior clinicians to discharge prematurely. 39–81

In the wider literature, discussions around ‘skill mix’ could reflect different ‘occupational lenses’ or types and levels of comprehensiveness around how homelessness might be addressed. Although the focus of the housing link worker schemes was often on housing and benefits advice, with ‘referrals on’ to primary care and other agencies, Hendry64 describes how staff in one medical respite scheme provided a full assessment under one roof, including a full screen blood test, screening for sexually transmitted diseases, medication compliance work, pre-detox work, smoking cessation, mental health, social services, occupational therapy referrals, benefits advice and chiropody. This comprehensiveness is a key feature of many residential step-down or medical respite schemes. 66

Involving carers and family members

In none of the material reviewed was there explicit reference to family member, carers and friends being involved in discharge and intermediate care support planning. However, there was an account of one discharge scheme that focused specifically on linking people who were experiencing homelessness back to their country of origin or their home town where they may have a ‘local connection’ and, therefore, a better chance of securing housing and social care support. 69

Mechanisms for integrating services

For many of the HHDF projects, integration into the hospital setting was described as challenging. 70 Formal protocols (e.g. developing a homeless hospital discharge protocol) were identified as important in most of the grey literature accounts, especially with regard to making sure ward staff asked about homeless and housing on admission. However, the main problem was sustaining them. 29 Successful ways of doing this and raising awareness about the schemes more generally included having the scheme championed by senior hospital staff and actively promoting the scheme through posters, leaflets and contact cards. 44 Co-location and being ‘a face’ on the ward were thought to help ensure the flow of referrals and ease of communication. 54–61,72 Participating in ward rounds, attendance at weekly hospital staff meetings to discuss patient discharge planning and running reflective practice and training sessions for hospital staff on the subject of homelessness were also considered helpful. Once the referral pathways were established in HHDF schemes, hospital staff seemed to appreciate being able to ‘hand over’ responsibility for the homeless people on their wards. 54 ‘Passing the baton’ could, however, potentially reduce opportunities for collaborative working.

Advocacy as an additional key mechanism

Although integration and co-ordination are foregrounded as key mechanisms for the successful delivery of intermediate care,25 the homeless-specific literature suggests that advocacy (‘arguing the case’) may be equally important. 44 Many grey literature accounts of the HHDF schemes alluded to the impact of austerity and depleted budgets, which meant reduced availability of housing and longer-term care and support. It was noted that housing authorities will often defend their budgets by rigidly restricting access to a defined ‘local’ population. This renders care co-ordination particularly challenging for homeless people who often have weak or no ties to any locality and lack documentary proof of any entitlements. 45

Integrating housing as the ‘third pillar’ of intermediate care

UK intermediate care has been delivered primarily as a health and social care service. 25 Canham et al. 47 recognise housing assessment as a key ‘health support’ for persons transitioning from hospital. The HHDF highlights the role of housing services in delivering improved health and well-being outcomes and, consequently, the importance of housing professionals working alongside health and social care professionals. Several grey literature accounts report that hospital staff appreciated this resource, especially in terms of its potential to free up their time. Charles et al. 54 observed that ward staff lacked the knowledge to find accommodation for homeless patients.

One report of a ‘housing link worker’ scheme describes extending its remit beyond ‘homeless people’ so that support could additionally be provided to older people who were being delayed in hospital because of ‘housing issues’. 82 This reflects the importance of integrating housing as the third pillar of health and well-being for all patient groups.

Discussion: framework utility

The additional evidence presented above broadly supports the validity or usefulness of Pearson et al. ’s25 conceptual framework for understanding ‘what works best’ in intermediate care. This is with regard to three key programme theories: (1) the importance of consulting with service users, (2) working in ways that are enabling and (3) ensuring integrated professional working. However, it might be suggested that these three ‘programme theories’ are likely to be implicated in the successful delivery of many other health and social care services. Herein lies a potential limitation of the current framework in that it may not answer some of the more complex or nuanced questions relating specifically to the development of intermediate care services.

The first challenging question to emerge from this review is how to maintain the integrity of intermediate care as a ‘time-limited’ intervention. This question arises where there is a need to encompass multiple and overlapping rehabilitative and resettlement goals, which may require housing solutions underpinned by much longer-term or continued health and social care support. Indeed, decisions around ‘time frame’ and scope are relevant to commissioners of intermediate care for both older people and people who are homeless (some of whom overlap). It is acknowledged that the rehabilitation of older people has sometimes fallen short because it has often prioritised short-term reablement goals linked to ‘physical functioning’ over and above those for inclusion and citizenship. Meanwhile, intermediate care for people who are homeless has reversed the ‘occupational lens’, prioritising longer-term resettlement and recovery outcomes over and above those for reablement. How to encompass these different needs and vulnerabilities under a single service banner is a significant additional challenge, with the danger that ‘specialist’ provision starts to confirm cultural distance (e.g. ‘elderly people’ are quiet and frail and ‘homeless people’ are challenging and disruptive).

These complexities are compounded in times of austerity, when the integrity of intermediate care is further compromised by the need not just to ‘fill the gaps’ in local provision but, on occasions to substitute for the widespread loss of longer-term support services. As Backer et al. 46 suggest, discharge planning and intermediate care will have little impact unless housing and other services are available. Interventions that were shown to work well in areas with well-resourced and efficient community support services were seen to have little or no impact where services are inadequate or lacking. 83

It is recognised that this poses perhaps the most serious threat to the viability of intermediate care as a service organisation and delivery construct. If the boundaries with longer-term care services start to blur, then intermediate care risks quickly becoming ‘blocked’. 66–71,73–84

In terms of a refined ‘conceptual framework’ that might address some of these issues, a US study57–71,73–85 is particularly insightful. It reports the findings of a RCT of a clinically led case management intervention called ‘critical time intervention’ (CTI). 57–71,73–85 CTI was designed to provide emotional and practical support over a 9-month period, with the primary objective of preventing homelessness among people being discharged from a psychiatric (mental health) hospital. In CTI, intermediate care is conceptualised as comprising the following three distinct phases:

1. Transition to the community focuses on engagement and relationship building, providing intensive support and assessing the resources that exist for the transition from in-patient care to community providers.

2. Try out is devoted to testing and adjusting the systems of support and assessing whether or not they are working as planned. By now, community providers are assumed to have adopted primary responsibility for delivering support.

3. Transfer of care focuses on completing the transfer of responsibility to community resources that deliver long-term support. 85

The findings of the RCT, which compared the outcomes of those receiving the CTI intervention with those receiving standard care, suggested that this brief, clearly focused intervention led to a reduction in the risk of homelessness that was evident 9 months after the intervention ended. In accounting for ‘what works and why’, consultation, enabling and ensuring integrated professional working are all implicated in CTI, but the cornerstone of the approach is a potential fourth programme theory. Namely, this is maintaining continuity of care during critical transition periods while responsibility gradually passes to existing community supports that will remain in place after the intervention ends. 85

In CTI, ‘scope’ is clearly defined as being about the management of transitions rather than specific kinds of ‘needs’ or ‘gaps’ in existing provision. It is, therefore, generic in that it can be applied to all client groups and can potentially be operationalised in any given local context, as the aim is to ‘weave together’ the resources and infrastructure that are already in existence. The ‘time frame’ for the intermediate care intervention is also determined not by any rigid ‘service-led’ criteria, but by the adaptive capacity of the local context to meet the person’s needs. It might be added that where CTI becomes ‘blocked’ (i.e. there are no appropriate services to take over responsibility), then this should ring alarm bells for commissioners that there are ‘cracks’ in local provision.

Indeed, CTI also seems to encapsulate the ‘how to’ of what Parker-Radford37 terms a ‘transition of care approach’. This has the additional advantage of shifting the focus of the ‘organisational lens’ from the acute sector to the management of a much wider range of transitions (e.g. ‘prison to community’ and ‘armed forces to civilian’). It is, therefore, potentially key to continuity and seamless care, as seen from the perspective of people who use or reject services.

Documentary analysis

Having completed the literature synthesis to identify the full range of (possible) interventions/changes (i.e. MIRs), the next step was to explore how these were being implemented in the HHDS sites. To achieve this, we collected and studied HHDS documentation (e.g. project leaflets, evaluation reports), monitoring data held by NHSE&I (for DToCs) and undertook a series of preliminary discussions and site visits with key HHDS stakeholders.

In carrying out the mapping exercise, we were able to make use of a database held by Homeless Link (an umbrella organisation for the homeless sector). This held typological details for all 52 HHDF schemes gathered as part of an early implementation evaluation. 70 All projects on the database were contacted for information and informed about the study. In addition to the information provided by Homeless Link, we were also able to access a wealth of information from the Pathway charity, which acts as an umbrella organisation for the ‘clinically led’ schemes. Preliminary interviews were carried out with staff at Pathway and Homeless Link. We also held three regional focus groups to which HHDS practitioners and managers were invited. These groups invited participants to discuss early progress and gauge interest in taking part in the research. Information was collected for 62 HHDSs, including the 52 HHDF-funded schemes and additional information for nine ‘clinically led’ schemes and one intermediate care network that was not funded by HHDF.

The analysis underpinning the mapping exercise was framed by means of our realist hypothesis (outlined earlier in Box 3). Documentary information was analysed to explore the interventions/changes (i.e. MIRs) that were being implemented in each of the HHDSs (e.g. protocols, patient in-reach, discharge co-ordination and step-down intermediate care) and more general information on funding history, staffing, skill mix and any evidence of effectiveness, as reported in local evaluation reports. At this initial mapping stage, we did not document KPPs, as these were not easily discernible from documentary information. The information was then compiled as a database. This was regularly updated throughout the life of the study, enabling us to chart the history of the projects and whether or not they were sustained in the longer term.

Site selection: specialist sites

The five sites with HHDSs were purposefully selected to represent different geographical contexts and typological (MIR) configurations. Selection was guided by the findings of the initial mapping exercise. The working hypothesis was that sites with more of the jigsaw pieces (i.e. more MIRs) would deliver better outcomes for patients and so we purposefully selected sites to enable interrogation of this, including sites with no HHDS (i.e. no specialist MIRs). We briefly summarise the findings of the mapping exercise below to show why the sites shown in Table 1 were recruited. The mapping also enabled us to further refine our understanding of MIRs for analysis and coding purposes.

The mapping exercise confirmed the important typological distinction between HHDSs that were ‘clinically led’, employing a (specialist) MDT to deliver patient in-reach (MIR2) and multidisciplinary discharge co-ordination (MIR3), and those that were ‘housing led’ (uniprofessional). Housing-led schemes also delivered patient in-reach (i.e. supporting and building relationships with patients prior to discharge) (MIR2) and undertook discharge co-ordination (MIR3). However, this was not necessarily multidisciplinary if, for example, the housing link worker from a uniprofessional scheme worked in isolation from other professionals and agencies. Therefore, for analysis (coding) purposes, we distinguish between MIR2 (patient in-reach) and MIR3 (discharge coordination) that is embedded in a MDT working (MIR2/3mdt) and that which is uniprofessional (MIR2/3up).

Another key typological feature to emerge from the mapping was that intermediate care (MIR4) was nearly always delivered as an integrated component of the ‘housing-led’ HHDSs (as time-limited floating support post discharge), whereas most of the ‘clinically led’ services closed cases when the patient exited the acute sector. For analysis (coding) purposes, we distinguish between intermediate care that is Home First (community-based floating support) (MIR4hf) and that which is residential/bed based (MIR4res) (i.e. where people move straight from hospital into a dedicated residential step-down unit).

The mapping exercise also demonstrated many other ways in which HHDSs were configured locally using different MIRs. For example, some localities had used HHDF funding to set up or refurbish an intermediate care facility (MIR4res), but did not then have resource to ‘in-reach’ into the hospital (MIR2-missing) and to assist ward staff with discharge co-ordination (MIR3-missing). Furthermore, in only one site could we find evidence of what might be termed sustained and purposeful engineering of a ‘complex adaptive system’. In this site, there were two HHDSs in operation. These were integrated by means of protocols (MIR1) for flexible working and ‘trusted assessment’ (MIR5) in which staff from the clinically led homeless team had direct referral rights into a housing-led residential intermediate care facility. This created an out-of-hospital care pathway that encompassed most of the jigsaw pieces:

On the basis of the mapping exercise, we hypothesised that this configuration (i.e. site 2) would deliver the best outcomes for homeless patients.

Standard care site selection

Two standard care sites were selected. The first did not have a HHDS. This site was selected opportunistically on the basis that the researcher knew one of the hospital managers and of her concerns around the need to improve hospital discharge for patients who were homeless. There was, therefore, interest on the part of the hospital trust in being part of the research. The site was also considered a good fit in terms of the wider portfolio of sites, as it was a mental health hospital (whereas trusts in other sites were acute) and also a hospital site serving a large rural area (whereas the other sites were mainly inner city). It was also recognised that this site could reflect the challenges of setting up specialist services in areas where there are few homeless patients. Another unusual feature of the site, in terms of poverty of MIRs (including non-specialist), was that it did not have a ‘generic’ complex discharge team. Here, ward staff were expected to take responsibility for all aspects of discharge planning.

The second standard care site was selected on the basis that it was a hospital site that had once had a specialist HHDS, but that scheme had recently been decommissioned. It was thought that this could shed light on the issue of sustainability and the ‘evaporation effect’. 86 Here, we were interested to explore what Dalkin et al. 32 term the ‘dimmer switch’ effect. In Pawson and Tilly’s27 seminal work, realist causation is explained by means of a gunpowder analogy in which a spark is introduced to the chemicals in gunpowder (mechanism), resulting in an explosion (outcome). However, if the conditions (context) are not right, for example if the gunpowder is damp, then there is no explosion. Here, mechanisms are seen as ‘triggering’ or ‘firing’ in context to achieve an outcome. However, Dalkin et al. 32 argue that where human volition is intertwined with an intervention, reasoning is rarely activated via a firing akin to an ‘on/off’ switch. Instead, they see activation as operating along a continuum similar to a light created by a dimmer switch where intensity varies in line with ever evolving context. Given the current austerity context, we recognised that this could be a particularly helpful heuristic for exploring the overall impact of the HHDF. We wanted to know if the different schemes were able to achieve a sustained ‘linear’ effect toward ‘good practice’ (safe discharge) or if ‘poor practice’ (unsafe discharge) crept back in once the MIRs were reduced or cut. Our hunch was that the latter may be the case, given that nationally delayed discharge figures could be seen to rise again once policy focus and resourcing shifted to other priorities.

Fieldwork

The fieldwork comprised non-participant observation in the everyday life of the HHDS (and/or hospital setting) and semistructured interviews with a range of stakeholders. Approximately 3 months was spent with each site, with the researcher spending 2 or 3 days on site each week. Non-participant observation encompassed shadowing workers and observations of assessments, case conferences, team meetings and other activities. Extensive fieldnotes were taken to record observational data.

Semistructured interviews with practitioners, managers and commissioners

In total, 77 semistructured interviews were carried out with a wide range of (non-service user) stakeholders (a minimum of 10 interviews per site) (see Appendix 2). Sampling was mainly purposeful (guided by the need to collect data to interrogate and refine programme theories), but ensured coverage of key staff groups in each site (e.g. HHDS managers and practitioners, discharge co-ordinators, ward staff, staff working on other local services and agencies, and service commissioners). Informed consent was ascertained for all participants. A topic guide (see Appendix 3) was used to gather information on a range of matters, including the history and current conceptualisation of the ‘problem’ of homeless people’s discharge from hospital in that particular site, understanding what changes/mechanisms (i.e. MIRs and KPPs) are being implemented, and perceptions about how they are working (i.e. leading to any changes in reasoning) and the associated outcomes. Key stakeholders (e.g. HHDS staff and managers) took part in a formal interview, but were then re-interviewed informally on many occasions as our programme theories evolved. Interviewing/recruitment stopped when it was felt that enough data were ascertained to have a comprehensive (rich) understanding of the programme theory under scrutiny. Interviews were digitally recorded and transcribed.

Interviews with homeless patients

Interviews were carried out with homeless patients (n = 70) at two time points, shortly after discharge and then again after 3 months. The timing of the second interview at 3 months gave sufficient time to reveal outcomes linked to continuity (e.g. what happened after step down ended), but not so far into the future as the connections with the initial hospital and discharge planning episode would be lost to memory.

Inclusion criteria (homeless patients)

In each site, we purposively recruited (≥ 10) homeless patients referred to either a HHDS or, in standard-care sites, to discharge co-ordinators/ward managers. Homeless patients (including HHDS service users) were recruited on the basis that they were ‘key informants’ with regard to different experiences of MEH. People experience MEH if they are ‘homeless’ (broadly defined to include temporary/unsuitable accommodation, as well as sleeping rough) and have also experienced one or more of the following domains of deep social exclusion: ‘institutional care’ (e.g. prison, local authority care, stays in mental health hospitals or wards), ‘substance misuse’ (e.g. drug problems, alcohol problems, abuse of solvents, glue or gas) or participation in street culture activities (e.g. begging, street drinking, ‘survival’ shop lifting or sex work). 5 Recruitment also aimed to monitor inclusion of the following groups/demographics and, whenever possible, we approached people in the order to which they referred to the service to avoid any selection bias:

• young homeless people (aged 18–24 years)

• people in the middle age range (aged 25–49 years)

• older homeless people (aged ≥ 50 years).

Recruitment/ascertaining informed consent/incentivisation

Sensitive strategies were employed to successfully recruit respondents from this ‘hard-to-reach’ group. Prior to the researcher making contact, we asked the HHDS worker/discharge co-ordinator to give the prospective participant an invite letter and study information sheet and to seek their permission to tell the interviewers if they were interested in taking part. When the interviewers subsequently visited the hospital, patients who had expressed an interest were contacted. The researcher explained that (1) participation was voluntary and they could withdraw at any time or refuse to answer any question, (2) that they would be seen twice by a researcher over the following 3 months, (3) that their consent was needed to collect information from HHDS staff and from their medical records and (4) if they said something that indicated that they or someone else was at risk of harm, the interviewer would be obliged to take appropriate action. Written consent was obtained prior to the first interview, with verbal consent obtained prior to subsequent interview. Only participants who gave informed and written consent were recruited. Participants’ capacity to give informed consent was determined before each subsequent interview. It was planned that should a participant lose capacity during the following 3 months then they would be kept in the study and, in accordance with the Mental Capacity Act 2005,87 a nominated consultee would be involved. To build trust and rapport before the first interview, the researchers visited patients on the ward on several occasions in the run up to their discharge. Participants also received a £10 voucher (per interview) for sharing their expertise and giving-up their time.

In total, 70 patients were recruited at time point 1 (T1) and 35 patients at time point 2 (T2). Finding participants at T2 was time-consuming and challenging for many reasons (e.g. people moving between areas with no fixed abode and regularly changing mobile telephone numbers):

• young homeless people (age 18–24 years) – T1 n = 5 male/n = 2 female; T2 n = 1 male

• people in the middle age range (aged 25–49 years) – T1 n = 27 male/n = 2 female; T2 n = 13 male/ n = 7 female

• older homeless people (aged ≥ 50 years) – T1 n = 20 male/n = 5 female; T2 n = 11 male/n = 3 female.

Topic guide

Interviews employed a semistructured topic guide (see Appendix 3) to unearth participants’ experiences of being admitted to hospital (i.e. the events leading up to the admission) and discharged from hospital and whether or not the HHDS (and the MIRs and KPPs therein) had the intended impact. The main focus of the second follow-up interview was on the following:

1. Patterns of service use in the intervening 3-month period, including any readmission to hospital. What support was received? Did the support materialise as planned? Did the support meet self-perceived needs?

2. What was liked and disliked about the support?

3. How was the support perceived to have helped? What outcomes were achieved? What outcomes were still aspired to and what additional support might have helped?

Interviews were digitally recorded and transcribed. We also gained participants’ written consent to access support plans and personal information held by a range of providers.

Patient and public involvement and engagement

Two ‘peer researchers’ (researchers with live experience of homelessness) accompanied us on some site visits and undertook a small number of interviews alongside the university researcher (e.g. one of the peer researchers was an ex-army veteran and so accompanied us on interviews where the patient was an ex-army veteran). Having peer researchers alongside helped establish trust and rapport with research participants and led to more targeted probing because of greater understanding (lived shared experiences) of the topics covered.

Some of the original study participants also became ‘peer researchers’ themselves (after the fieldwork had concluded), joining a special advisory group and contributing mainly to the dissemination stages (e.g. producing cases studies and attending high-level meetings and conferences to share first-hand experiences about issues such as discharge to the street). This was acknowledged by policy stakeholders as being very effective in bringing the academic findings to life and supporting a greater commitment to change.

Overview

The qualitative findings support our initial tentative programme theory about the importance of having all the ‘jigsaw pieces’ in place to deliver consistently safe, timely transfers of care. Our observations (based on 10 patient journeys per site) confirmed that consistently safer, timelier transfers of care were more likely to occur in site 2 compared with the other sites. Site 2 was purposefully selected on the basis that it came closest to NHSE&I’s description of a ‘complex adaptive system’. 20 This site had a fully integrated out-of-hospital care pathway that encompassed protocols (including those for early discharge planning) (MIR1), multidisciplinary patient in-reach (MIR2mdt), discharge co-ordination (MIR3mdt) and a residential step-down intermediate care facility (MIR4res). Patients (accepted for the ‘step-down’ service) were collected by taxi and taken straight to the facility. Here, hospital discharge was a ‘non-event’ and there were no reports of poor or unsafe discharge. Furthermore, the out-of-hospital care pathway itself was embedded as part of a wider whole system of care. This matched the description for the highest level of integration as defined by the DHSC’s Office of the Chief Analyst:

Fully coordinated primary and secondary care that provides an integrated service, including specialist primary care, outreach services, intermediate care beds, and in-reach service to acute beds.

p. 18. 88

Contains public sector information licensed under the Open Government Licence v1.0

In the remainder of this section, we describe in more detail how MIRs and KPPs are integrated into the specialist out-of-hospital care pathway in site 2 and how this works to deliver good outcomes for (some) patients. We then draw on observational and interview data from the other fieldwork sites to pinpoint how ‘gaps’ (i.e. missing jigsaw pieces) lead to poorer outcomes. We also consider, across all the sites, the factors that dampen effects, leading MIRs and KPPs to fail to deliver their intended outcomes. In Chapters 4 and 5 we then interrogate these qualitative findings by means of a quantitative lens, mindful of the limitations inherent in a case study approach.

Specialist care: ‘all the jigsaw pieces’

Site 2 contained two specialist HHDSs that work in partnership to create an out-of-hospital care pathway for patients who are homeless and who have a physical health need. A multidisciplinary (clinically led) homeless team was based at the hospital with a 14-bed (housing-led) residential step-down respite and intermediate care facility in the community.

Clinically led multidisciplinary hospital-based homeless team

In site 2, the homeless team was clinically led, comprising a manager (a band 7 nurse), a nurse, a housing worker and a care navigator, with sessional support provided by a GP and a mental health nurse. The homeless team manager was employed by the acute trust and the other clinical staff by a specialist primary care practice (holding overall management responsibility for the team). The housing workers were seconded to the team from a housing association that also managed and ran the step-down facility. The homeless team was originally co-located in the same offices as the (generic) complex discharge team, but at the time of data collection occupied its own office space. The two discharge teams complemented each other, with the homeless team taking on full responsibility for the discharge of patients who are homeless and other patients with housing problems. The manager of the generic discharge team had performance oversight of the homeless team with regard to monitoring and addressing any DToCs. The homeless team operated extended working hours on some evenings, but there was pressure for more flexible working and for the service to operate 24/7 should funding allow.

In site 2, the homeless team acted as the named point of contact for patients who were homeless, usually taking on responsibility for all aspects of their discharge planning (MIR3). Just as social workers provided detailed specialist knowledge of the Care Act 201489 and other legislation and the local care market, the homeless team provided expertise on housing legislation and local homeless service provision. The housing workers in the homeless health team also took responsibility for the considerable administrative burden that can be associated with making an approach to housing or maximising income (e.g. sorting out access to benefits or helping people ascertain identification proof or to establish a local connection). As the homeless team was co-located at the hospital (avoiding the need for ward staff to make external referrals and wait for a response), they were highly appreciated by ward staff and credited by hospital managers with playing a vital role in maintaining patient flow and reducing delays. This point was mirrored across nearly all the research sites that had HHDS:

Before we had a homeless team, homeless patients were referred to the social work team, but the social work team would say its housing not care and signpost them to outside . . . Now that we have a [homeless worker based at the hospital] it’s a massive bonus for the trust, someone we can turn to, specialising in homelessness and who knows the processes, policies, laws, benefits and everything that comes with that than an everyday nurse wouldn’t be able to deal with . . . It’s a massive bonus.

Ward manager, S3b

In addition to discharge co-ordination, clinically led homeless teams also undertook patient in-reach (MIR2mdt), with the overall aim of building relationships with patients and maximising the benefits of a hospital stay. This included preventing early self-discharge and increasing access to planned health care:

We see ourselves more as a homeless health team rather than just a discharge team . . . We work with homeless patients at the level they’re at – optimising their stay – making sure that they’ve had their jabs, that they been seen by the substance misuse services even though they’ve come in for a broken arm; that they have a sexual health screening if the need that; that they get the things done that the homeless nurses in the community have been trying to get them have done for ages, for example, a particular set of bloods. This might not be indicated in terms of why they’ve come in with, but it’s about maximising the stay and following this up with outpatient appointments.

Homeless health care team nurse, S1

Residential step-down intermediate care (MIR4res)

In site 2, HHDF capital funding had been secured to refurbish a former student hall of residence into a 14-bed dedicated residential intermediate care step-down facility. The Respite and Intermediate Care Support Service (RICSS) was ‘housing led’ and managed by a housing association. It was staffed 24/7, with housing/resettlement workers during the daytime and security staff at night. Revenue funding was mainly sourced in the same way as for other hostels in the district (e.g. from housing benefit), but the facility is only accessible to people who are homeless and assessed as needing intermediate care on the basis of physical health need. Additional funding for RICSS was also provided through the BCF and used to pay the support staff and manager who are on duty every day of the week. Without this additional funding, the service was not thought to be viable.

Seen from the perspective of the patient, the journey into RICSS starts in the hospital and is facilitated by the homeless team. The homeless team start to assess and work with patients while they are in hospital and will continue to work with them (‘boundary span’) when they move into RICSS. Importantly, the homeless team has direct referral rights into RICSS and the team’s nurse and GP are ‘trusted assessors’ (MIR5), meaning that RICSS staff will accept their assessments and judgements about a patient’s suitability for the service. Therefore, when medically optimised, people move seamlessly from hospital into RICSS, such that hospital discharge is an ‘uneventful’ event (a short taxi ride). The homeless team’s GP and nurse undertake a weekly ‘ward round’ at RICSS (reviewing the care of each resident) and work flexibly to be on hand to provide housing staff with advice and back-up on health matters. For patients, trusted assessment (MIR5) facilitates integration (KPP3), leading to a high degree of continuity (KPP4), in which ‘I only have to tell my story once’.

Once settled into RICSS, a fully integrated assessment commences of the patient’s health, housing and social care needs. Importantly, admission to RICSS enables clinical staff from the homeless team and RICSS housing staff to work together to observe needs over time and ascertain a fuller picture, rather than relying on a verbal snapshot at the point of discharge. These ‘snapshots’ can be unreliable given that people are ‘never themselves’ when in hospital and in crises or unwell. This opportunity for comprehensiveness is at the heart of intermediate care based on a D2A model. Below we describe how KPPs are embedded in intermediate care in site 2.

What binds a complex adaptive system?

Does this out-of-hospital care pathway work for everyone?

As noted above, RICSS staff worked hard to create a peaceful convalescent atmosphere. However, one consequence of this is that people with active addictions or loud, disruptive and challenging behaviour were perceived to be less suitable for the service. Rather than actively exclude this group, the aim was to have a balance of residents with different needs. Even so, having a ‘physical health need’ (as a determinant of eligibility) seemed to align more closely with physical disabilities and non-addiction-related conditions. For example, a man whose primary need was considered to be related to his heroin addiction was considered unsuitable for RICSS, despite being homeless and having a physical health need arising from a groin abscess. He returned to rough sleeping on discharge. A limitation of our case study approach is that in this site we focused on the RICSS pathway and not on hospital discharge more generally.

What dampens the effect of the mechanism intervention resources embedded in this out-of-hospital care pathway?

In site 2, although there had been budget cuts, the perception of the homeless team and RICSS staff was that there was still a good range of accommodation and support services. As compared with the other fieldwork sites, we observed that there appeared to be much more scope for specialist discharge co-ordination (MIR3mdt) and D2A to function as intended. Patients also seemed to have much more scope to exercise choice and control (KPP1). Indeed, for RICSS staff, a particular challenge was how to manage choice (KPP1) where people may have reached their goals as regards their physical health, but where they are waiting solely for the right property or post code (locality) to become available. It is anticipated that the usual length of stay at RICSS should be around 3 months. However, it was often the case that people stayed much longer (sometimes for up to a year), leading to ‘bed blocking’. It was recognised by RICSS senior managers that it was important to strengthen protocols for patient flow (MIR1) by placing reasonable limits on how choice was exercised.

In addition to ‘patient exercising choice’, the other main cause of delayed discharge from RICSS was the shortage of suitable long-term care and support services, especially for people aged < 55 years who needed adult social care or sheltered housing. Again, this could cause long delays and there was awareness of the danger that that intermediate care could start to substitute for long-term care. Indeed, in sites without intermediate care, commissioners often expressed the view that the main reason why they had not commissioned this service was because of fears that it would become quickly ‘silted up’.

There was also a recognised unmet need for some RICSS residents to have a further period of Home First ‘step-down’ care in the form of floating support in the community (MIR4hf) once they had moved into a new home of their own. To compensate for this, RICSS workers kept hold of some patients for perhaps longer than necessary and then held ‘secret caseloads’ of former residents whom they kept an eye on until they knew longer-term community services and (relationships) were in place and working well (KPP4).

Learning from site 2 for programme theory refinement

In the literature review, we found less clear evidence for MIRs linked to trusted assessment (MIR5). However, in site 2, there was strong evidence that this mechanism is crucial to the delivery of good outcomes and patient experience. The importance of MIR5 becomes even more apparent when evidence is considered from site 1 where it was missing. Site 1 had a ‘clinically led’ homeless team but no direct access to step-down intermediate care (MIR4-missing). Although a small number of residential intermediate care beds were available locally, these were earmarked for ‘step-up’ care (preventing admission to hospital and rarely available for step-down care). Furthermore, there is no protocol (MIR1) in place for trusted assessment (MIR5). As a result, patients referred to the intermediate care service were discharged from the acute sector and reassessed by the intermediate care team on presentation. In site 1, patients had to tell their story more than once. In addition, if the intermediate care bed was not available on the day of discharge as planned then this could leave them in limbo, having been discharged from hospital without an agreed discharge/support plan in place (Box 5).

Standard care: ‘no jigsaw pieces’

Without access to a HHDS, discharge co-ordinators and ward staff in standard care sites stated that they found it challenging to deliver consistently safe, timely transfers of care for their homeless patients. The most commonly reported difficulties were navigating complex rules and systems and understanding what resources and services were available for homeless people locally. They felt that there was too little time to complete the administrative work associated with supporting a housing application (e.g. helping someone to ascertain proof of identity or local connection) and also too much pressure on beds, which meant that people could not wait in hospital while housing was arranged.

In site 6 (standard care), it was not uncommon for patients to be given very short notice to leave the ward, hours in some cases, before being ‘signposted’ to the local council homeless office. This was highly distressing for patients and often led to angry and open confrontations on the ward. In one case, the patient had not disclosed domestic violence and was too afraid to go into the city centre. In this site, the relationship between the hospital and the local council housing/homeless department was underdeveloped. A consultant from the hospital was a member of the local homeless strategy group, but had been too busy to attend meetings. Meanwhile, the local housing officers were frustrated that patients were presenting as homeless with no prior warning, but did not know whom to approach inside the hospital to address this. A homeless hospital discharge protocol (MIR1) had been developed by the local council, but had been stuck somewhere in the hospital governance system for over 2 years.

Following a research interview with the homeless officer in site 6 (standard care), the details of the ward manager were passed on and a meeting was set-up. Once this meeting occurred, discharges were quickly transformed. The housing officer agreed to spend time on the ward each week to assist with discharge planning, along with a worker from a local voluntary agency who helped patients fill out forms to access support services. In this city, serving a large rural area, there was good availability of hostel places and temporary accommodation (for those not wanting to go into a hostel) plus intensive support for 6 weeks to organise benefits and move on. Therefore, this standard care site was able to start delivering safe and timely transfers of care without a specialist HHDS. However, having an informal arrangement [without a dedicated resource or protocol (MIR1-missing)] led to a lack of consistency. Should the housing officer be off sick or go on holiday, for example, the outcome was often a DToC or unsafe discharge. Therefore, the perception of ward staff in this site (as was the case in all case study sites) was that having a dedicated specialist resource to in-reach into the hospital and co-ordinate housing and homelessness arrangements (MIR2 + MIR3) was essential to more efficient and consistent patient flow.

This qualitative finding is supported by evidence gathered as part of the documentary analysis. NHSE&I statistics show that the two standard-care sites that did not have access to a HHDS had a higher rate of DToC linked to housing (category I) than the other fieldwork sites (Figure 2).

FIGURE 2.

Number of patients with a DToC due to housing (category I). The data are taken from NHS England monthly situation reports90 (2012 to March 2017). It reports the number of patients delayed on the last Thursday of the month. These data were no longer collected after April 2017. From April 2017, the reported data were the total delayed days during the month for all patients delayed throughout the month. Data are shown at provider organisation level from NHS trusts, NHS foundation trusts and primary care trusts. The six trusts with HHDS include an additional pilot site not included in the main fieldwork. The way in which DToCs are counted and recorded in England was changed again in 2020,91 with the introduction of a new category ‘Homeless/no right of recourse to public funds/no place to discharge to.’ Contains information licensed under the Open Government Licence v3.0.

Contains information licensed under the Open Government Licence v3.0.

Evaporation effect

In site 5, a standard-care site that had lost its specialist HHDS because of funding cuts, it was reported that the relationships that had been built up between voluntary workers and hospital staff sustained benefits for a time, but gradually suffered the dimmer switch effect, as voluntary workers were pulled onto new contracts. In this site, it is interesting that, once the specialist HHDS had ceased to operate, the management of discharges for patients who were homeless increasingly fell to the alcohol liaison team. This can be seen as a reflection of not only the level of specialist expertise within the alcohol liaison team around housing and homelessness, but also the extent to which the alcohol liaison service was able to link more directly (and easily) into community services and resources. Concerns were expressed, however, that the ‘specialist’ alcohol service might not be able to continue to bridge this gap without undermining its own effectiveness and long-term sustainability.

In several sites, the underpinning programme logic was that HHDSs would be time-limited interventions designed to educate and up-skill hospital staff about working with homelessness patients. However, the consensus among stakeholders was that because of high staff turnover in hospitals, the impact of the HHDS is rarely sustained once a scheme has ended.

Missing jigsaw pieces

In this section, we provide more illustrative examples to demonstrate how missing ‘jigsaw pieces’ limit the effectiveness of HHDS to deliver intended outcomes.

Example 1: out-of-hospital care pathway with Home First (MIR4hf) but no residential step-down acre (MIR4res-missing)

In England, Home First (MIR4hf) is the default pathway for all patients leaving hospital and is based on the belief that people will recover best in their own home (KPP1). However, if patients are homeless and they do not have a home or somewhere safe to stay, then the Home First model is compromised. This was evident from our observations in site 3a. This site had a ‘housing-led’ HHDS (with a single member of staff) offering patient in-reach and discharge co-ordination (MIR2/3up) with time-limited floating support post discharge (MIR4hf). This scheme had no direct access to a specialist residential step-down unit (MIR4-res missing) or any other units of specially earmarked accommodation. As illustrated in Box 6, where patients are being discharged to bed and breakfasts or other unsuitable accommodation, this makes for poorer patient experience and almost certainly has a negative impact on the health outcomes that are achieved (e.g. speed of recovery after surgery). Nevertheless, a caveat here is that having a HHDS link worker to help advocate and negotiate the hardship of unsafe discharge is hugely appreciated by the patient and may go some way to negating some of the impacts of this.

In site 3b, which had a similar housing-led scheme to site 3a, the Clinical Commissioning Group commissioned a number of ‘health beds’ in a local homeless hostel that were then earmarked for use by people leaving hospital. However, the ‘health beds’ did not come with any additional support other than that usually provided in the hostel, which meant that this could place an extra burden on staff in terms of having to manage the more complex health and care needs of residents who had recently left hospital. Furthermore, the ‘health beds’ were not always occupied by people recently discharged from hospital, which resulted in knock-on effects for patient flow out of the hospital.

Example 2: uniprofessional homeless hospital discharge scheme with no clinically led patient in-reach (MIR2mdt-missing)

Link workers in ‘housing-led’ HHDSs play a vital role in supporting homeless patients while they are in hospital. This support is concerned with engagement and relationship building and also ensuring that patients have toiletries and other essentials, especially clean clothes, and, therefore, addresses factors that can lead to stigmatisation. However, in site 3b, a housing-led scheme with no primary care in-reach (MIR2mdt), early self-discharge due to poor methadone prescribing was recognised as a significant problem at the hospital. Hostel staff in this site also spoke of their difficulty of getting residents to go into hospital (site 3b) and to stay there. This was because they could not guarantee that the hospital would prescribe methadone. Hostel staff described the consequences of this as being left with very ‘poorly residents’ who were at risk of worsening health conditions, which was distressing for staff.

Example 3: multidisciplinary homeless hospital discharge scheme with no access to step-down care – Home First (floating support) (MIR4hf-missing and KPP4-missing)

It is common practice for (generic) hospital discharge teams and discharge co-ordinators to close cases at the point at which the patient leaves the acute sector. Most of the clinically led HHDSs also managed their caseload in this way, with some limited outpatient follow-up (i.e. where a housing application has been started). However, for many homeless patients this unsupported transition out of hospital can be problematic (Box 7). Indeed, although charities such as the British Red Cross (London, UK) provide a vital service for older patients (e.g. turning their heating on and putting milk in the fridge), a much enhanced logistical and intensive support service may be required for patients who are homeless. For example, in addition to the logistical challenges that go with trying to ‘move house’ from a hospital bed (often by means of patient transport), patients may also need to organise a methadone prescription on the day of discharge and may be given a strict time slot when they can present for this. If this element of the discharge plan goes wrong, then the huge health and well-being benefits of a hospital stay (i.e. rest, care, kindness, treatment, emotional support, good food, detox for alcohol and getting off heroin and onto methadone) can evaporate in an instant. Similarly, sending someone to a cold, empty, depressing flat, alone, late in the evening and with no belongings (because of patient transport rules or lack of money) is a recipe for disaster that can quickly lead back to rough sleeping. These problems were much less likely to occur where the patient had a link worker to accompany them on the day of discharge and to help with all aspects of resettlement, including making sure community services were in place and working well (KPP4).

Example 4: residential intermediate care facility with no resource for patient in-reach and multidisciplinary discharge co-ordination (MIR2/3-missing)

Site 4 contained a three-bed respite/intermediate care facility housed within a self-contained wing of a hostel. The scheme did not include any resource for the worker to ‘in-reach’ (MIR2/3-missing) into the hospital other than when a referral was received to undertake an assessment. After a few months, the scheme closed because of lack of referrals, reflecting the importance of HHDSs maintaining networks and a high-profile ‘face-to-face’ presence in the hospital to ensure that ward staff know what services are available ‘out’ in the community. In summary, although residential intermediate care (MIR4res) is key to safe, timely transfers for patients who have nowhere safe to stay on discharge, it will not work effectively or be sustainable if it is not linked to other MIRs.

What dampens the effects of specialist care?

There are three key contextual factors that can dampen the effect of the MIRs and KPPs in out-of-hospital care: (1) a lack of adequate funding for the HHDS itself; (2) situations where housing and wider community support services are poorly resourced, inadequate or lacking; and (3) circumstances where stigma and cultural distance persist. Where these factors have an impact on a HHDS, the outcome can be ‘poor practice’ (e.g. discharge to the street).

Interplay between mechanisms and context

As outlined in Chapter 1, the HHDF introduced additional ‘resources’ into contexts that were heavily affected by austerity. Indeed, there is a strong case to be made that this additional resource per se has improved outcomes (without necessarily firing any change in reasoning as is anticipated in realist theory). For example, HHDSs bring a significant additional staffing resource into the hospital, which allows ward staff to free-up their time by handing over responsibility for complex homeless patients. Without this extra resource, ward staff acknowledge that they would not have the time to undertake the work needed to secure safe discharge beyond ‘signposting’ to relevant agencies. In sites that had lost their HHDSs or where schemes were reduced in scale (i.e. where resourcing was not sustained), then the benefits of the HHDF were seen to evaporate.

Our observations also chart the dampening effects of a wider lack of resources across the whole system. We observed, for example, that discharge to the street was commonplace in sites where HHDS staff felt that the patient had no options left open to them (i.e. where there was a lack of housing, care and support). These findings mirror those of the earlier national evaluations of intermediate care for older people that we reported earlier. Here, it was concluded that the effectiveness of interventions to improve the speed and quality of discharge will depend, to a large extent, on the broader service context in which they take place. Interventions that are shown to work well in areas with well-resourced and efficient community support services may have little or no impact where services are inadequate or lacking. 83

Changes in reasoning

Although HHDSs ‘work’ effectively (in certain contexts) to deliver outcomes linked to patient flow (e.g. reducing DToCs), there is little evidence to suggest that they have ‘fired’ a change in reasoning with regard to discharge to the street. Eradicating the ‘poor practice’ that led to discharge to the street was a key objective of the HHDF and other government guidance at the time:

Too often discharge is seen as freeing-up a hospital bed rather than acting in the patient’s best interest to move them . . . To a safer environment that will encourage supported self-management, speed . . . recovery and have them feel better. We must make every effort to shift understanding to this reality.

p. 2092

Contains public sector information licensed under the Open Government Licence v3.0

Prior to the introduction of new COVID-19 hospital discharge guidance in 2020,91 situation reports for DToCs were indicative of large numbers of older adults being permitted to remain in an acute hospital bed once they were clinically optimised. This was for a variety of reasons, including ‘awaiting a care package’ or ‘patient and family exercising choice’ (i.e. waiting for a specific care home place to become available). For the research team undertaking the fieldwork, this raised the question as to why patients who were homeless were not accorded the same leeway to remain in a hospital bed while they waited for the housing, care and support of their choice to be arranged. Our observations suggested that this difference lay in the perpetuation of stigma and cultural distance that positioned ‘homeless patients’ as somehow less vulnerable and/or deserving than other groups of patients.

We observed that patients who were homeless and using substances were particularly affected by unsafe discharge, especially when their behaviour was perceived to be challenging. NHS incident reports suggest that the demographic most likely to be involved in aggressive incidents in the acute sector is men aged between 75 and 95 years (with their female peers not far behind) and that dementia and delirium lie behind much of this distress. 93 However, it would be unusual for this group of older patients to be discharged unsafely on ‘behavioural grounds’, alerting us as to how different conditions and vulnerabilities garner more or less sympathy.

Viewed through the lenses of professional ethics and adult safeguarding, it is difficult to see how discharging a patient to the street can be justified within the parameters of ‘safe to transfer’. 94 Local safeguarding adult reviews have increasingly drawn attention to instances where poor discharge arrangements have contributed to the deaths of people who are homeless. In particular, they have highlighted the poor practice associated with seeing each hospital admission in isolation, the failure to provide appropriate multidisciplinary responses and the failure to initiate safeguarding where unsafe discharge occurs. In one case, a man was signposted to a homeless resource centre, shortly before his death, with known risks that he was unlikely to receive any support there because of his incontinence. 95 There is a specific category of abuse defined in the Care Act 201489 under ‘Neglect – Acts of Omission’ to describe circumstances where there is a ‘failure to provide access to appropriate health, care and support; a neglect of medical, emotional and physical needs; and withholding necessities of life, such as adequate nutrition and heating’ (contains information licensed under the Open Government Licence v3.0). 89 This suggests that MIRs for adult safeguarding (MIR7) are currently the missing piece of the jigsaw and may be a necessary driver for changes in reason.

Indeed, there is a case to be made that this change in reason is necessary to secure adequate provision of specialist intermediate care. When DToCs reach crisis point, the government will usually intervene to create extra capacity in the system. In the winter of 2016/17, DToCs hit an all-time high when, on average, 6660 people were delayed every day in NHS beds. In March 2017, the government introduced a new target to reduce the level of bed-days ‘lost’ to 3.5% and announced an extra £2B of funding for social care services. It might be argued that because people who are homeless are more likely to be discharged prematurely, out-of-hospital care for this group remains underdeveloped and under-resourced, further exacerbating the drivers to discharge people unsafely.

Comparator groups

Three groups of individuals admitted to hospital were included in the analysis. The first group comprised homeless individuals admitted to hospital at any one of 17 sites with a HHDS between 1 November 2013 and 30 November 2016. An initial audit of HHDSs established the date on which they were implemented.

The second group comprised individuals seen by a community homeless service in London, UK [Find & Treat (F&T)] and subsequently admitted to a hospital that did not have a HHDS. This group was, therefore, selected to provide a comparison group of people experiencing homelessness, admitted to hospital, but not seen by a HHDS. F&T is a specialist outreach team that works alongside over 200 NHS and third-sector front-line services to prevent or treat tuberculosis among homeless people. It is primarily based in London. We looked for hospital admissions during the study period for all individuals seen by F&T between 1 November 2013 and 30 November 2016.

The third group was a random sample of individuals equal in size to the F&T group and living in lower-layer super output areas in England in the most deprived quintile, as measured by the Index of Multiple Deprivation (IMD), who were recorded as having a fixed address. These individuals came from hospitals offering HHDSs with an admission during the study period but without being seen by the HHDS. This group was used to confirm or refute our hypothesis that the homeless people included in the study have worse outcomes than non-homeless people living in deprived areas. This group is subsequently referred to as IMD5 (Index of Multiple Deprivation quintile 5).

Data linkage

Data were linked in respect of each group from HES, which includes dates, causes and length of admission, and the Office for National Statistics (ONS) mortality database, which includes dates and causes of death.

Data collection, processing and linkage

Data flows used in this analysis are described in Figure 3. We obtained demographic-identifying variables under our legal and ethics approvals. Where available, NHS numbers were collected from the 17 HHDS sites. NHS Digital (formally the Health and Social Care Information Centre) used the personal demographics service to identify and add NHS numbers to as many individual records as possible.

FIGURE 3.

Study data flows. PDS, personal demographics service; UCL, University College London.

NHS Digital undertook all data linkage to HES data using patient identifiers obtained (in combination with personal demographics service tracing) and we securely uploaded a de-identified copy of the data to the University College London (London, UK) Institute of Health Informatics’ Data Safe Haven, which is a robust infrastructure certified for processing and analysing identifiable data in accordance with international and national information security standards (ISO/IEC 27001:2013 and NHS Information Governance Toolkit). The deterministic linkage process was undertaken in three steps to ensure that the three groups were mutually exclusive:

1. HES linkage to data collected at the HHDS sites

2. HES linkage to data collected from F&T services, with admissions stratified into those sites with and without HHDSs

3. identification of people in the most deprived IMD quintile from HES at the HHDS sites who were admitted after the implementation date at a given HHDS site.

In the main analysis, individuals identified in more than one of the steps described above were classified using the hierarchy 1 > 2 > 3, such that data on homeless patients who were admitted multiple times to a combination of hospitals with, and without, HHDSs were analysed according to admission to the HHDS site only. Patients identified as both homeless (i.e. HHDS or F&T) as well as those in the non-homeless (IMD5) group were assigned to the homeless HHDS category. In addition to collecting data on admissions after being seen by an HHDS, we obtained data on all individuals from HES in respect of admissions to any hospital in England from 1 January 2008 to 31 October 2013 to create a profile of their existing health conditions.

Sample size

Our sample size calculation was based on historical data from the two types of HHDS, suggesting an average of 92 patients per month across the 17 sites. We estimated that each type would have data from approximately 2208 patients for the duration of the study.

From previous health service evaluations, we expected 0.7 hospital episodes per person-year for homeless individuals. A clinically important reduction in readmission rates would be 10%. To undertake a sample size calculation to determine the study size required to detect such a reduction in readmission rates, the following variables were defined:

• µ_0 – mean readmission rate in the baseline (F&T) group = 0.7 episodes per person-year.

• µ_1 – mean readmission rate in the HHDS group = 0.6 episodes per person-year.

• v = 1.96 – percentage point of the normal distribution corresponding to a 5% two-sided significance level.

• u = 0.84 – one-sided percentage point of the normal distribution corresponding to {100% – type II error [false negative/(true positive + false negative)]} at 80% power.

Using the following equation, we estimated minimum sample size (per group) required for comparison of two rates (readmissions per person-year):

Adjustment for confounders (i.e. age, sex, ethnicity, calendar quarter of admission and existing comorbidities), as far as is possible within the analysis, resulted in a required doubling of sample size (additional 10% per confounder) required. Therefore, 2038 person-years were required per group. Assuming that the HHDS sites see, on average, 90 patients each month during the study period (which, for the purposes of the initial calculation we assumed to be November 2013 to November 2015, as many schemes may have been shut down before this time period and in some data may have been collected at an earlier point), we estimated that this translated to a total of 2160 individuals at each of the two types of scheme (e.g. housing led and clinically led). Given that the average follow-up period was likely to exceed 1 year, we anticipated that the study would be powered to detect a 10% difference in readmission rates between HHDS sites and non-HHDS sites for both the clinically- and the housing-led HHDSs.

Outcomes

A series of primary and secondary outcomes were included to ensure that our analysis was both consistent with outcomes used in previous published analyses and included outcomes that previous studies were not powered to collect. We chose several secondary outcomes that collectively reflect the priorities of health policy-makers and individuals attending patient engagement workshops. Full definitions of primary and secondary outcomes are provided in Tables 2 and 3.

Throughout the report we refer to three terms, avoidable,96 preventable and amenable, for which we use the following definitions and their associated ICD-10 (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision) codes:

• Avoidable mortality – avoidable deaths are all those defined as preventable, amenable or both, where each death is counted only once. Where a cause of death falls within both the preventable and amenable definition, all deaths from that cause are counted in both categories when they are presented separately.

• Amenable mortality – a death is amenable if, in the light of medical knowledge and technology at the time of death, all or most deaths from that cause (subject to age limits, if appropriate) could be avoided through good quality health care.

• Preventable mortality – a death is preventable if, in the light of understanding of the determinants of health at the time of death, all or most deaths from that cause (subject to age limits, if appropriate) could be avoided by public health interventions in the broadest sense.

Analysis plan

We undertook the analysis in two phases. In the first phase, we analysed baseline characteristics (Table 4) of all participants to describe the characteristics of each of the study groups at or before the index admission (Figure 4). With the exception of ethnicity, all baseline characteristics were anticipated to be fully observed (chronic disease is presumed to be absent unless recorded). Missing values of ethnicity were analysed grouped as ‘not recorded’.

FIGURE 4.

Schematic outlining hypothetical patients. Data on the characteristics of patients before their index admission were collated from admissions occurring before the implementation of HHDSs at a given site (or for comparator groups a randomly selected date within the range of implementation dates for HHDS sites).

We then summarised each of the primary and secondary outcomes by comparison group and explored the geographical spread of our 17 sites to explore their representativeness. This work enabled us to confirm the suitability of the proposed statistical methods and analysis protocol proposed. Where applicable, we compared our baseline mortality data with those in the recent analysis of homeless deaths by the ONS (see Blackburn et al. ,96 table S1, supplementary file, for definitions and full ICD-10 code lists used).

In the second phase, we identified evidence of differences in the baseline characteristics, including age, sex, chronic disease and reason for hospital admission at the time of their index admission. We estimated the crude association between each of the primary and secondary outcomes and the study population groups. The baseline comparator group was IMD5. We then re-estimated the association between each of the outcomes and the study population group after adjusting for the following characteristics at the time of admission: age, sex, chronic disease and reason for hospital admission. We undertook supplementary subgroup analyses to evaluate evidence of a difference in the outcomes of people (1) admitted to clinically- compared with housing-led schemes and (2) schemes with step-down care compared with schemes without access to step-down care.

Crude models were fitted prior to adjustment for ‘baseline’ measurements at or before the index admission. We then wrote-up the analysis in accordance with the RECORD (REporting of studies Conducted using Observational Routinely collected Data) statement.

Protocol deviations

The data linkage study was significantly delayed at commencement because of the launch of the Health Research Authority and changes to the procedures for securing ethics approvals. It was further delayed by NHS Digital in relation to approving and returning the linked data set and then, finally, because of NHS Digital initially returning a data set without an IMD5 comparison group or linked death data. As a result, we did not have time to complete the analyses for the following secondary outcomes:

• time from discharge to next elective admission

• all-cause mortality expressed as a SMR

• ICD-10 chapter-specific SMR.

Having completed the main literature synthesis (some 6 months after the study began), we developed our original realist hypothesis to capture new evidence about the importance of HHDSs having direct access to ‘step-down’ intermediate care. To enable us to interrogate this revised theory, we introduced a further comparative element across the evaluation to establish the effectiveness and cost-effectiveness of HHDSs with access to ‘step-down’ care (MIR4) compared with those with no direct access to ‘step-down’ care.

Causes of death

In addition to understanding the geographical spread of the data, the demographic differences in each group after linkage and the levels of comorbidities, we also examined causes of death before undertaking a full analysis of the primary and secondary outcomes. Table 7 shows the demographic characteristics of decedents in each group. Males made up 77.7% (466/600) of deaths in the HHDS group and 56.4% (1416/2512) of deaths in the IMD5 group. The median age of death was 51.6 [interquartile range (IQR) 42.7–60.2] years for the HHDS group and 71.5 (IQR 60.67–79.0) years for the IMD5 group. Ethnicity was broadly similar across comparison groups, but 12.9% (403/3,112) of ethnicities were either not known or not stated. The top three underlying causes of death by ICD-10 chapter in the HHDS group were external causes of death (21.7%, 130/600), cancer (19.0%, 114/600) and digestive disease (19.0%, 114/600) in the unweighted analysis.

In an age- and sex-weighted analysis (Table 8 and Figure 6), the top three underlying causes of death in the HHDS group were cardiovascular disease (30.1%, 180/600), cancer (21.8%, 131/600) and respiratory disease (16.7%, 100/600). In this weighted analysis, the underlying causes of death for each cohort by ICD-10 chapter showed a greater proportion of deaths with an underlying cause in ‘External Causes of Morbidity and Mortality’ chapter in the homeless group (7.4%, 44/600) than in the IMD5 group (3.9%, 99/2,512). This chapter covers a range of causes, including accidents, intentional self-harm, assault and events of undetermined intent (e.g. poisoning). Weighted alcohol- and drug-related causes were higher in the HHDS group (9.8%, 59/600 and 3.6%, 22/600, respectively) than in the IMD5 group (2.7%, 67/2512 and 0.9%, 23/2512, respectively). The percentage of deaths with an amenable cause of mortality was higher in the HHDS group (30.2%, 181/600) than in the IMD5 group (23.0%, 578/2512). This means that after weighting for age and sex, nearly one in three deaths of people in our homeless hospital discharge cohort were due to conditions, many of which are common in the general population (e.g. heart disease), that are amenable to timely health care.

FIGURE 6.

Underlying cause of death by ICD-10 group and subgroup, comparing homeless weighted and IMD5 groups. COPD, chronic obstructive pulmonary disease; CVD, cardiovascular disease; IHD, ischaemic heart disease. Adapted with permission from Aldridge et al. 97 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Adapted with permission from Aldridge et al. 97 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original.

In an unweighted analysis that compared the HHDS group with the ONS homeless causes of death analysis, accidents and suicides contributed to a smaller percentage of all underlying causes of death in the HHDS homeless group (19.5%, 117/600) than in the ONS homeless group (54.9%, 319/581) (Figure 7). Conversely, we observed a larger number of deaths due to cancer (18.2%, 109/600 vs. 4.6%, 27/581), diseases of the liver (13.8%, 83/600 vs. 9.0%, 52/581) and chronic lower respiratory disease (6.0%, 36/600 vs. 2.2%, 13/581) in the HHDS group than in the ONS analysis.

FIGURE 7.

Underlying cause of death across the three comparator groups. Adapted with permission from Aldridge et al. 97 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Adapted with permission from Aldridge et al. 97 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original.

Primary outcomes

Time from discharge to next accident and emergency attendance

In the unadjusted analysis comparing clinically led services with housing-led services for time from discharge to next A&E attendance, sex, age group, number of comorbidities and number of admissions in the year before index admission were all associated with an increase in emergency attendance (Table 10). In the multivariable analyses, older age, larger numbers of comorbidities and larger numbers of admissions in the 1 year before the index admission were all associated with an increase in emergency attendance. We found no evidence in either the unadjusted or adjusted analyses for a difference between the housing- or clinically led services for emergency attendance (HR 1.05, 95% CI 0.83 to 1.32; p = 0.6835) (Table 11). We grouped HHDS sites into those with step-down care and those without step-down care and found no evidence for a difference in emergency attendance between the two groups in a multivariable analysis (HR 1.01, 95% CI 0.82 to 1.26; p = 0.9063) (Table 12). In a multivariate analysis, the hazard of emergency attendance was higher in the combined HHDS group than in the IMD5 group (HR 7.34, 95% CI 6.31 to 8.54; p < 0.0001) (see Table 12).

28-day emergency readmission

The overall percentage of patients with a 28-day emergency readmission in the housing-led group (19.5%, 196/1006) was slightly lower than in the clinically led group (21.2%, 613/2888) (see Table 11). Male patients (21.1%, 600/2840) had slightly higher readmission rates than female patients (19.8%, 209/1054) and readmission rates were highest in the 55- to 64-years age group (22.4%, 134/599). In both the unadjusted and the adjusted analyses, only number of comorbidities and number of admissions in the year before index admissions were associated with 28-day emergency readmissions. In an adjusted analysis, no evidence was found for a difference in readmission rates between housing- and clinically led services [odds ratio (OR) 1.08, 95% CI 0.89 to 1.29; p = 0.4420]. In an adjusted analysis, we found that the HHDS group had higher odds of 28-day emergency readmissions than the IMD5 group (OR 3.60, 95% CI 3.24 to 3.99; p < 0.0001) (see Table 12). When comparing services with step-down care with those without step-down care, the percentage of patients with 28-day emergency readmissions was similar across both groups (20.2% for step-down care vs. 21.1% for no step-down care). There was no evidence in either the unadjusted or adjusted multivariate analysis for a difference between these groups in terms of 28-day emergency readmission (OR 1.05, 95% CI 0.88 to 1.25; p = 0.5821) (see Table 12).

Secondary outcomes

Overall and unscheduled readmission rates

We looked at the 12-month risk of readmission by ICD-10 chapter of index admission (Figure 8). The 12-month risk of unscheduled readmission (often referred to as emergency readmission) was 59% (95% CI 57% to 61%) for homeless patients and 20% (95% CI 19% to 21%) for housed patients, for planned readmission it was 18% (95% CI 17% to 20%) for homeless patients and 28% (95% CI 26% to 29%) for housed patients and for A&E visits it was 92% (95% CI 91% to 93%) for homeless patients and 57% (95% CI 55% to 59%) for housed patients.

FIGURE 8.

Twelve-month risk of readmission, stratified by the ICD-10 chapter of index admission: (a) emergency readmission; (b) planned readmission; and (c) A&E.

Unplanned and A&E readmissions were extremely high for the HHDS group, and these unplanned readmissions are high regardless of the type of index admission. This is in contrast to the IMD5 group, where the readmission risk was lower than in the HHDS group and varied according to the type of index admission. In addition, the readmission risk was particularly high for those admitted for ‘mental and behavioural’ causes. Planned readmissions were generally lower in the HHDS group than in the IMD5 group.

Unadjusted incident rate ratios from a negative binomial model that examined the risk of readmission showed that homeless patients had 5.14 (95% CI 4.71 to 5.62) times the rate of unscheduled readmission, 1.08 (95% CI 0.95 to 1.23) times the rate of planned readmission and 5.38 (95% CI 5.04 to 5.72) times the rate of A&E visits (Table 14). Adjusting for comorbidities and the ICD-10 chapter of the index admission partially attenuated these values, with adjusted rate ratios of 3.77 (95% CI 3.46 to 4.10), 0.71 (95% CI 0.63 to 0.81) and 3.76 (95% CI 3.53 to 4.01), respectively. As an exploratory sensitivity analysis (not prespecified), we fit the model for planned readmissions, excluding ‘series’ of admission (defined above, mainly for renal failure and dialysis), which gave an unadjusted rate ratio of 0.69 (95% CI 0.62 to 0.78) and an adjusted ratio of 0.62 (95% CI 0.55 to 0.69).

TABLE 14 - Incident rate ratios of readmissions and A&E visits, comparing homeless with housed patients (results of negative binomial regression)

Outcome	Unadjusted (matched on age, sex, year and hospital), rate ratio (95% CI); p-value	Adjusted for comorbidities and ICD-10 chapter of admissions, rate ratio (95% CI); p-value
Unscheduled admissions	5.14 (4.71 to 5.62); < 0.001	3.77 (3.46 to 4.10); < 0.001
Planned admissions	1.08 (0.95 to 1.23); 0.236	0.71 (0.63 to 0.81); < 0.001
A&E visits	5.38 (5.04 to 5.72); < 0.001	3.76 (3.53 to 4.01); < 0.001

We then constructed negative binomial regressions models to examine the risks of readmission, comparing homeless patients at sites with an associated step-down service with homeless patients at sites without a step-down service (baseline group). Among homeless patients, the rate of unscheduled admissions was similar when comparing those discharged at a site with or without a step-down service, but the rate of planned admissions was 56% lower (incident rate ratio 0.44, 95% CI 0.35 to 0.55; p < 0.001) and the rate of A&E visits was 21% lower (incident rate ratio 0.79, 95% CI 0.71 to 0.87; p < 0.001) (Table 15). These results were partially attenuated when adjusting for comorbidities and the ICD-10 chapter of the index admission. After adjustment, there was evidence that A&E visits were 18% lower among homeless patients discharged at a site with a step-down service than among those without a step-down service (incident rate ratio 0.82, 95% CI 0.75 to 0.91; p < 0.001).

TABLE 15 - Incident rate ratios of readmissions and A&E visits, comparing homeless patients at sites with an associated step-down service with homeless patients at sites without a step-down service (baseline group) (results of negative binomial regression)

Outcome	Adjusted for age, sex and year of admission, rate ratio (95% CI); p-value	Further adjusted for comorbidities and ICD-10 chapter of admissions, rate ratio (95% CI); p-value
Unscheduled admissions	1.02 (0.92 to 1.13); 0.742	1.08 (0.98 to 1.18); 0.102
Planned admissions	0.44 (0.35 to 0.55); < 0.001	0.88 (0.73 to 1.06); 0.140
A&E visits	0.79 (0.71 to 0.87); < 0.001	0.82 (0.75 to 0.91); < 0.001

Limitations

As described above, when designing the study, we chose the F&T comparator group (for standard care), as we hoped that this would provide a control group of people initially identified as homeless in the community and subsequently admitted to hospitals without HHDSs (i.e. a standard-care comparator). However, with the benefit of hindsight, and our initial descriptive analysis of the data once we received it, we do not feel this to be a fair comparison. For this reason, we do not present results for our primary and secondary outcomes for the F&T comparator group. We were, therefore, not able to undertake a comparative effectiveness analysis that examined the outcomes for people experiencing homelessness admitted to hospitals with a HHDS and those admitted to a hospital without a HHDS.

As noted above, we found no evidence for a difference in our primary outcomes when comparing housing- and clinically led services, or those with step-down service or no step-down service. These findings may be correct, but there remain several important alternative explanations that mean there could still be a difference between these types of service that we were unable to detect. First, we gained data from the sites we planned to and were only slightly under the total number of patients required by our sample size calculation. However, there were more than twice the number of patients in the clinically led sites than in the housing-led sites. Our study is, therefore, underpowered to detect the clinically significant 10% difference in the primary outcomes that we set out to achieve. Taking one of our primary outcomes as an example – readmission owing to any cause – the CI in our final adjusted comparison between housing- and clinically led services (HR 1.07, 95% CI 0.88 to 1.30) is statistically consistent with housing-led services being up to 30% better or up to 12% worse than clinically led services. Our secondary outcome analysis of multiple readmission rates between different types of HHDSs had greater power because of the larger number of events included and it is, therefore, not surprising that this was the one analysis where we found statistical evidence of a difference between different typologies of HHDS for a reduction in A&E admissions. Second, the data we collected were for routine records and, therefore, there may be problems in the recording of outcomes and exposures, which may introduce random measurement error into our analyses, reducing power of the study to detect a difference between the groups further. Third, we were limited in the number of variables we could collect in our analysis on each group and there may be residual confounding in our comparisons. Fourth, there was substantial geographical differences in the way sites were run and the number of patients they saw, with capital city sites seeing larger numbers of patients and being mainly clinically led. This may introduce biases to our analyses in both directions. Finally, we were not always able to identify the date of index admission, therefore limiting our ability identify the exact intervention for each participant, which would further reduce our study power.

Despite these limitations, in being able to compare the effectiveness of different HHDSs, our linkage study had many strengths. The linkage between hospital and death records enabled us to determine long-term outcomes in our groups, which is evidence that has been sorely lacking from this population in England. We present some of the first analysis of the long-term morbidity and mortality outcomes in this socially excluded population and we also describe, in great detail, their health trajectory on admission to hospital and when subsequently admitted or at death.

A further strength of our study is the certainty with which the homeless groups were identified, as these individuals had been seen by services that provide support specifically for people with experience of homelessness. Although ‘no fixed abode’ can be entered as a patient’s address in NHS systems, the use of this coding is inconsistent and, as a result, has limitations as marker of homelessness within HES records. 88

The deaths in this study for the HHDS and IMD5 groups relate to people who have had a hospital admission at centres providing a HHDS service. We were not able to identify patients accessing primary care services, which would be likely to provide an insight into deaths of patients with fewer acute health conditions, but may also include unexpected deaths that occur without prior hospitalisation. The IMD5 comparison group was older and included more women, and we stratified and adjusted our analyses to account for these differences.

The higher levels of deaths owing to alcohol-related causes and drug poisoning in the recent ONS data compared with our study may be as a result of the large number of death records in this study identified from coroners’ reports. Our study is not subject to this potential bias, but we recognise that deaths arising from those admitted under care of HHDSs may also not be representative of all deaths. We have data on deaths for people with experience of homelessness who were admitted to hospital only and, therefore, our results will exclude those people who died having never accessed health care, many of whom may be individuals who die because of external causes. Our results will likely over-represent people with chronic diseases admitted to hospital treatment, compared with the ONS analysis. All methods employed to measure the homeless population have limitations and it is difficult to envisage a method that would provide a fully generalisable population-level estimate for this excluded population. Data linkage for further services that work with homeless people is likely to offer the best opportunities for future research.

Part 1: economic literature review

Part 1 reviewed the economic evidence contained in the individual audit and evaluation reports for the 52 HHDF schemes.

Part 2: evidence for key mechanism resources across 17 homeless hospital discharge schemes

This primary economic evaluation linked HES data to explore the effectiveness and cost-effectiveness of 17 specialist HHDSs (vs. a standard-care control group). The analysis included a comparative focus on HHDSs with different mechanism resources, focusing on two key typologies: (1) clinically led services (i.e. MDTs) compared with housing-led services (i.e. uniprofessional teams) and (2) those with direct access to ‘step-down’ care compared with those without direct access to ‘step-down’ care.

Part 3: focus on three out-of-hospital care configurations

This primary economic evaluation explored the effectiveness and cost-effectiveness of three different ‘out-of-hospital care’ configurations, using three data sources [i.e. published evidence, a survey (see Appendices 7 and 8) and local audit data] in addition to HES data. This considered cost-effectiveness from both NHS and wider public perspectives.

Before introducing the findings linked to each part of the economic evaluation, we outline the methods used for economic modelling. There were some significant protocol changes in the economic evaluation. The main change was linked to the work outlined for part 3 (study of three out-of-hospital care configurations). Although we had planned to collect primary survey data in the seven sites, only one site (site 2) collected sufficient data. In standard-care sites, it was not possible to collect survey data because hospitals without homeless teams did not keep lists of homelessness patients (making their identification for recruitment across the hospital sites difficult). To overcome this problem, we re-recruited the site that had taken part in the pilot for the study. In addition, we drew on primary data from a RCT78 that had studied two HHDSs. We discuss these changes further in Appendices 9 and 10.

Intervention groups (specialist care)

In part 1, 17 HHDSs were recruited and divided into the following intervention groups:

• clinically led (multidisciplinary) schemes (n = 12)/housing-led (uniprofessional) schemes (n = 5) (vs. standard care)

• schemes with access to step-down care (n = 9)/schemes with no direct access to step-down care (n = 8) (vs. standard care).

In part 2, the intervention groups were three different out-of-hospital care configurations:

1. A multidisciplinary clinically led (homeless health-care) team that offers patient in-reach and specialist discharge co-ordination, with no direct access to intermediate care. Two examples of this configuration are considered, one situated in a capital city and one in a seaside resort in the south of England.

2. A multidisciplinary clinically led (homeless health-care) team that offers patient in-reach and specialist discharge co-ordination, with direct access to a 14-bed residential step-down unit. Based in an old industrial city in the north of England (described as site 2 in previous chapters). (Note that here, two separate HHDSs are working together in an integrated way in the same locality.)

3. A uniprofessional housing-led scheme that offers (non-clinically focused) patient in-reach and specialist discharge co-ordination, with housing workers providing floating support in the community for a time-limited period (i.e. community step-down care). Based in an old industrial city in the Midlands. (This was an original pilot site for the study and it did not take part in the qualitative case studies.)

Control group (standard care)

In the study protocol, ‘standard care’ was conceptualised by means of two ‘control groups’. These were the same controls as for the data linkage process, described previously:

1. The first proposed control group comprised individuals who were known to be homeless and were admitted to hospital between 1 November 2013 and 30 November 2016 at any one of the 17 participating research sites, but who were not seen by the HHDS operating in that site.

2. The second proposed control group was a sample of individuals who accessed the F&T service. As noted earlier, F&T is a specialist outreach team that works alongside over 200 NHS and third-sector front-line services to respond to the risk and presence of, among homeless people, drug or alcohol users, vulnerable migrants and people who have been in prison. The F&T database included individuals screened for active pulmonary tuberculosis and patients with tuberculosis managed by the service with record dates between September 2007 and July 2010.

Following analysis of the (homeless) patient characteristics data for control groups 1 and 2, it emerged that the homeless individuals who were not seen by a HHDS were generally healthier and, therefore, they did not constitute a comparator group suitable for analysis. To compensate for this, we considered a third control group that was not originally specified in the protocol.

1. This control group comprised patient-level data that were extracted from the RCT conducted by Hewett et al. 78 The RCT had been conducted by a co-applicant of this study, enabling access to the data. The trial compared ‘standard care’ with ‘specialist care’ across two hospital sites in England (London and Brighton). The primary end point was the reduction in length of stay. Predefined secondary end points included the impact on the quality of life, post-discharge accommodation and cost-effectiveness. Homeless people admitted to hospital were randomised to either standard care or the HHDS (a clinically led HHDS with no access to step-down care). Homeless patients assigned to standard care were visited once by a homelessness health nurse and provided with an information leaflet that described local services. The hospital data system provided health-care usage information and questionnaires were used to assess the quality of life using the EuroQol 5 Dimensions (EQ-5D). EQ-5D is a generic measure used primarily by economists to calculate quality-adjusted life-years (QALYs) (see Appendices 7 and 8). In total, 206 homeless people were allocated to specialist care and 204 to standard care. It was estimated that the service delivery or operating costs were £154,228 per year, with salary costs being added to training costs (£11,120). The impact of the intervention on inpatient activity was assessed by comparing the average cost per inpatient spell or period using all spells for which data were available. The trial found no impact on inpatient costs and it compared operating costs of the intervention with the effect on homeless individuals’ health gain, measured in QALYs. At baseline, health-related quality of life was similar across the arms.

Analysis of the (homeless) patient characteristics data (i.e. age, gender and utility in terms of QALYs) for control group 3 found comparable results with what was reported by the groups of homeless users assigned to the present study interventions. Therefore, control group 378 was considered a suitable ‘standard-care’ comparator.

Data sources for the interventions

Methodology for model 1

The primary outcome chosen for the cost-effectiveness analyses was the cumulative duration of hospital stay (i.e. number of bed-days after the index admission). This followed published literature. 78 As noted above, owing to limitations with the originally proposed comparator groups, data from the control group of the RCT study published by Hewett et al. 78 were used as a proxy for ‘standard care’. Unfortunately, standard-care group data did not present any baseline estimates for either the primary outcome or cost data. For this reason, only annual patient-level estimates for bed-days and readmission costs (relating to the follow-up period) were compared between intervention and control groups. The statistical analyses were consistent with the regression approach used by Hewett et al. 78 A regression model was used to analyse the difference in length of stay between groups at follow-up, adjusting for age and sex. We applied bootstrapping to allow for non-normality in the outcome data. The perspective adopted was that of the NHS. The time frame was 1 year (primary analyses) or 3 years,99 based on the assumption that the individual remains homeless and yearly cost and benefit are constant across time.

Cost data were derived from the use of health-care resources (i.e. number of readmissions) and estimates for the intervention group were extracted from HES data (Table 19). The quantities were then multiplied by a set of national average unit costs (Table 20). The total costs of health services were then summarised at an aggregated level (e.g. according to the type of readmission costs) for the corresponding periods. Economic outcomes used for this analysis included (1) elective readmissions per homeless patient, (2) emergency readmissions per homeless patient, (3) other readmissions per patient and (4) all readmissions. Costs were updated to 2017 figures. A regression model (similar to the one adopted for length of stay) was used to analyse any difference in NHS costs between groups at follow-up and was adjusted according to patient characteristics.

A cost-effectiveness analysis enumerated the additional resources consumed for an improvement in the effects (i.e. avoided bed-days) associated with the intervention groups, compared with standard care. The difference in mean 12-month costs and outcomes were estimated based on the model. Incremental cost-effectiveness ratios (ICERs) were calculated by dividing the difference in mean cost by the difference in mean health outcome.

One-way sensitivity analyses were conducted to assess the impact on the base-case results. With the first sensitivity analysis, the service delivery costs were taken into consideration (along with all readmissions costs). In addition, to test the impact of the intervention when shifting costs from non-elective to elective readmissions, we added a second sensitivity analysis where we separately explored elective readmissions costs (together with the service delivery costs) and non-elective readmissions costs (together with the service delivery costs). Additional scenario analyses for the economic model included the following: bed-days avoided for the comparator were increased to the upper-limit 95% CI and the total costs of the comparator were equal to the lower-limit 95% CI. To investigate the effect of the HHDSs across time, a follow-up period of 3 years was considered. (Three years was used to represent the medium-term follow-up period as reported by Crisis. 99) For the latter, the recommended discount rate of 3.5% was applied to both costs and benefits. A probabilistic sensitivity analysis was performed to assess the uncertainty in the economic model outputs associated with the uncertainty in parameter inputs. A non-parametric bootstrap technique was employed to explore the uncertainty of point estimates of the difference in mean 12-month costs and outcomes from primary analyses. The results from bootstrap resampling were used to construct 95% CIs for incremental costs and incremental bed-days avoided and to plot the cost-effectiveness plane and cost-effectiveness acceptability curve to show the uncertainty surrounding the primary results.

Data derived from the cost-effectiveness analysis were used to estimate the budget impact of replacing usual care when adopting the various interventions. For each intervention, we calculated the rate of return for the investment made:

Intervention groups (specialist care)

The need for both HES and EQ-5D data to facilitate this analysis meant that we could not include all 17 HHDSs, but instead had to focus opportunistically on those schemes for which EQ-5D data were available. This led us to focus on the three out-of-hospital care configurations, as outlined above.

Control group (standard care)

The control group used is the same as for model 1 (i.e. control group 3).

Data sources

In addition to HES data and hospital data (as outlined for model 1), the following additional data sources were employed in model 2:

• Configuration 1 – a member of our research team had recently completed a RCT of two homeless health-care teams employing EQ-5D at two time points. We were able to draw on this EQ-5D data, which conferred access to 206 T1 completions and 56 T2 completions (at 6 weeks post discharge).

• Configuration 2 – in this fieldwork site we were able to meet the target for EQ-5D questionnaire completions. Here we accessed 64 completions at T1 and 54 at T2 (at 3 months post discharge).

• Configuration 3 – this site worked with us in the pilot phase of this present study and was collecting EQ-5D data as part of its own internal monitoring and review procedures. From this site, we accessed 84 completions at T1 and 84 at T2 (at 12 months post discharge).

In total, drawing on these different data sources, we accessed 354 EQ-5D completions at T1 and 248 at T2.

Methodology for model 2

Although Hewett et al. 78 did not collect any (retrospective) baseline data for the length of stay and use of health-care resources (i.e. emergency, elective and other types of admissions), they reported both baseline and 12-month follow-up data for EQ-5D scores. A regression model was used to enable differences in EQ-5D score at follow-up to be adjusted for patient characteristics (i.e. age and sex) and for differences in outcome at baseline (see Hewett et al. 78). The tariff for each combination of the EQ-5D levels for the UK population was applied to calculate utility values. 107 QALYs were then calculated over the duration of the economic model using the area under the curve of utility values from the three time points. 108 For the primary analyses, costs and QALYs were not discounted because the assessing period was 12 months. For the 3-year economic model, a discount rate of 3.5% was applied.

Cost outcomes were the same as reported for economic model 1. Costs were updated to 2017 figures. Cost and effectiveness estimates were calculated for a 1-year follow-up period and a 3-year follow-up period (with costs discounted at 3.5% per annum).

The ICER was calculated by dividing the incremental costs (mean intervention minus mean control) by the incremental QALY (mean intervention minus mean control). 109 The intervention is considered cost-effective if the ICER falls below the NICE threshold of £20,000–30,000 per QALY. The cost-effectiveness acceptability curve presented the impact of uncertainty on the result of an economic evaluation. It was produced by resampling the data 1000 times to generate a mean cost and life-year or QALY gain from each group, using a non-parametric bootstrap approach. 110 The proportion of resampled data sets for which the calculated ICER lies below a given threshold is interpreted as the probability that the ICER of the intervention is below that threshold. Cost of the service delivery was not included in the main analyses, but was included in the sensitivity analyses (and this followed the same approach considered for economic model 1). Both the one-way sensitivity analysis and the probabilistic sensitivity analysis were similar to the above (see economic model 1).

This economic model allowed us to measure the cost-effectiveness of two different out-of-hospital care configurations ‘before and after’ their introduction. Observed differences in performance are assumed to be because of the intervention. This economic model also took a wider public sector perspective (e.g. exploring costs for use of health care, mental health care, drug and alcohol treatment, housing, criminal justice, social care, social benefits and state pension). It replicated the approach used in the Making Every Adult Matter (MEAM) evaluation. 111 In the UK, the MEAM pilot programmes aimed to understand the economic impact of specialist provision targeted at a similar multiple needs cohort where homelessness was the most significant issue. Therefore, the MEAM pilot study provided an already tested methodology for understanding the economic impact of the specialist provision on service use across the whole system.

Intervention groups (specialist care)

The configurations that could be evaluated were limited to two because these were the only sites where it was possible to collect service delivery costs:

• Configuration 2 – multidisciplinary clinically led (homeless health-care team) scheme that offers patient in-reach and specialist discharge co-ordination, with direct access to a 14-bed residential step-down unit. Based in an old industrial city in the north of England. (Note that here, two separate HHDSs are working together in an integrated way in the same locality.)

  • MIR1 + MIR2mdt + MIR3mdt + MIR4res + MIR5.

• Configuration 3 – uniprofessional housing-led scheme that offers (non-clinically focused) patient in-reach and specialist discharge co-ordination, with housing workers providing floating support in the community for a time-limited period (i.e. community step-down care). Based in an old industrial city in the Midlands.

  • MIR1 +  MIR2up + MIR3up + MIR4hf.

Control group (standard care)

No control group was used in this model. This economic model allowed us to measure the cost-effectiveness of the HHDS ‘before and after’ the introduction of the intervention in the same study site(s).

Data sources

• Configuration 2: in this fieldwork site we were able to meet the target for EQ-5D questionnaire completions. Here, we accessed 64 completions at T1 and 54 at T2 (at 3 months post discharge).

• Configuration 3: this site worked with us in the pilot phase of the study and was collecting EQ-5D data as part of its own internal monitoring and review procedures. From this site, we accessed 84 completions at T1 and 84 at T2 (at 12 months post discharge).

Methodology for model 3

The measure of utility was gain in QALYs. Annual cost estimates were reported from health-care and larger public provider perspectives. Outcomes included use/cost of health-care and utility data (QALYs). Costing data included hospitalisation and A&E visits. Costs were updated to 2017 figures. Unit costs are reported in Table 20. For each configuration, effectiveness was expressed as QALYs gained (compared with the care available prior to the project starting). Cost-effectiveness was reported as costs to be invested per QALY gained (ICERs). The time frame was 1 year. Service delivery costs were calculated using costing information collected from the site.

The robustness of the economic model was tested using one-way deterministic sensitivity analyses looking at the following alternative scenarios: no service delivery, service delivery costs increased by 10%, service delivery costs increased by 20%, mean hospitalisation cost at follow-up equal to the upper-limit 95% CI, mean housing costs at follow-up equal to the upper-limit 95% CI and mean QALY at follow-up equal to the lower-limit 95% CI.

In addition, probabilistic sensitivity analyses were undertaken and CIs for the cost per QALY gained were estimated using the non-parametric percentile bootstrap method. The data were resampled 1000 times to generate a mean cost and life-year or QALY gain from each point of interest, and the resulting ICERs were calculated and plotted into the cost–utility plane. A cost–utility acceptability curve was also plotted. The economic model was tested using data from a subgroup of patients of the HHDS in configuration 3 as part of piloting in preparation for the main study modelling (see Report Supplementary Material 1). Summary information providing an overview of the different models is presented in Report Supplementary Material 2.

Overview

In this section, we report the findings of a review of the economic data that were contained in the audits and evaluation reports that had already been published for the 52 HHDSs. Papers for the economic review were identified through the main literature synthesis (reported in Chapter 2). Overall, 39 (57.35%) studies from the main synthesis were relevant for this focused review. Inclusion criteria were that studies had reported on socioeconomic impact and quality-of-life data. Table 21 presents a summary of the papers and considers (1) the quality of the evidence, (2) the economic outcomes measured and (3) the main findings for the cost-effectiveness of different HHDSs. Reports were available for 31 different HHDSs (some schemes were evaluated more than once), representing the following typological configurations:

• discharge co-ordination with no step-down care/limited community follow-up (11 services)

• discharge co-ordination with residential (bed-based) step-down care (three services)

• discharge co-ordination with community (floating support) step-down care (eight services)

• discharge co-ordination with residential and community step-down care (nine services).

One additional paper (reporting the findings of early evaluation of the HHDF) reported on the service delivery costs for all HHDSs. 70 This reported that the average cost per client across all projects that returned expenditure and outcomes data was £2543. This did not consider effectiveness and cost-effectiveness.

Results

The findings from the economic literature review are presented in Figure 9:

• Twenty-one of 39 papers (53.84%) were rated as having good-quality data.

• Nineteen of 39 papers (48.72%) demonstrated economic benefit in terms of decreases in the use of health-care services and NHS economic savings, and societal benefit in terms of return in social value, and three papers demonstrated improved health and well-being of individuals (EQ-5D data).

• In this review, the most methodologically robust evidence is for HHDSs with access to residential ‘step-down’ care. These schemes are shown to be both cost-effective and cost saving:

  • The Economic Evaluation of Homeless Intermediate Care Pilot Project reported that the intervention delivered by Cedars Road Hostel (Clapham) was cost saving. 64 This service is clinically led.

  • One configuration, encompassing the integration of a clinically led HHDS with a residential step-down unit, was able to show that under the most conservative assumptions every £1 invested in each of these services generates at least £1.50 worth of benefit (wider public perspective) and under less conservative (but realistic) assumptions as much as £8 worth of benefit was generated (again, wider public perspective). 117

FIGURE 9.

Economic evidence from the literature: per cent of papers reporting information on the economic evaluation conducted, the data sourced and the economic results. ROI, return on investment.

Implications for realist hypothesis testing

• The review of the published literature alerted us to an emerging evidence base that ‘step-down’ intermediate care is both cost-effective and cost saving. However, there is also strong evidence to suggest that schemes without this component can be cost-effective and cost saving. These findings might, therefore, be taken as a broad indication of the likely benefits of specialist care per se.

• Our theory does not currently distinguish between different types of step-down [residential and home-based (floating support)] care and our findings suggest that this may be an important consideration.

• Clinically led MDT working is a key component in examples of (residential) step-down intermediate care that are cost-effective and cost saving.

Limitations of economic literature review

The main limitation of this preliminary analysis (drawing on grey literature and evaluations of individual schemes) is that many HHDSs and their evaluations are not reporting robust economic outcome data. There is a notable lack of economic evidence for schemes offering community (housing-led) step-down care and schemes with access to both community and residential step-down care. As a result, this may bias the result towards certain schemes.

Overview

This section reports the findings of a primary economic evaluation using linked HES data to explore the effectiveness and cost-effectiveness of 17 specialist HHDSs (vs. a standard-care control group). The analysis includes a comparative focus on HHDSs with different mechanism resources, focusing on two key typologies: (1) clinically led services (i.e. MDTs) compared with housing-led services (i.e. uniprofessional teams) and (2) those with direct access to ‘step-down’ care compared with those without direct access to ‘step-down’ care.

Results

The effect of homeless hospital discharge schemes on health-care resources and costs

Table 19 gives an overview of the number of patients (at index discharge), completed admissions and bed-days in the 1 year after discharge, by HHDS intervention type and for the control group.

Costs per patient for each readmission

The difference in annual NHS costs per patient for a range of readmission types (e.g. elective readmissions, emergency readmissions, other readmissions and costs for all admissions) is shown in Figure 13. Considering all readmissions, homeless patients in the specialist HHDS intervention group used more resources. These patients had higher rates of readmissions and increased hospital costs.

The difference in annual NHS costs per patient (vs. standard care) were greater for clinically led schemes than for ‘housing-led’ schemes (£3112.02 vs. £1295.45; p < 0.01) and for schemes with no ‘step-down’ care than for schemes with ‘step-down’ care (£3271.02 vs. £1325.84; p < 0.01) (Figure 10).

FIGURE 10.

Economic model 1: difference in annual NHS costs per patient (vs. standard care from Hewett et al. 78). All readmissions [i.e. readmitted for the same problem (e.g. within 28 days)] included the sum of elective, emergency and other readmissions. Comparison 1: clinically-/housing-led schemes (vs. standard care) (based on data for 17 schemes). Comparison 2: schemes with/without ‘step-down’ care (vs. standard care) (based on data for 17 schemes).

Clinically led schemes reported higher costs for elective readmissions than housing-led schemes (for which there was a cost saving).

Schemes with no ‘step-down’ care reported higher costs for elective readmissions than schemes with ‘step-down’ care (for which there was a cost saving).

Access to step-down care (vs. no step-down care) results in slightly lower costs associated with emergency readmissions. Clinically led services have higher costs associated with emergency readmissions than ‘housing-led’ schemes.

Bed-days avoided per patient

Table 19 shows that patients using a HHDS used slightly fewer bed-days than homeless people discharged from the hospital without the support of a HHDS.

Figure 11 shows the difference in annual bed-days per patient. The number of bed-days avoided per patient (vs. control group) was comparable for clinically- and housing-led schemes (1.92 vs. 1.55, respectively) and sites with no ‘step-down’ care and with ‘step-down’ care (1.57 vs. 2.34, respectively).

FIGURE 11.

Economic model 1: difference in annual bed-days per patient (vs. standard care from Hewett et al. 78). Comparison 1: clinically-/housing-led schemes (vs. standard care) (based on data for 17 schemes). Comparison 2: schemes with and without ‘step-down’ care (vs. standard care) (based on data for 17 schemes).

Cost-effectiveness (resources to be invested per bed-day avoided)

Figure 12 shows the NHS resources to be invested per bed-day avoided. The analysis includes only the economic consequences on the NHS (readmissions), as service delivery costs were not made available for the 17 HHDSs.

FIGURE 12.

Economic model 1: NHS resources to be invested per bed-day avoided (control group = standard care from Hewett et al. 78). Comparison 1: clinically-/housing-led schemes (vs. standard care). Comparison 2: schemes with and without ‘step-down’ care (vs. standard care). Main analysis: NHS resources included costs for all readmissions (without service delivery costs). Sensitivity analysis 2: NHS resources covered costs for non-elective (emergency) readmissions without service delivery costs.

Bed-days avoided is one measure commonly used to assess the effectiveness of NHS services for homeless people and was adopted here as the primary outcome measure (following Hewett et al. 78). In the main analysis, the NHS resources to be invested included costs for all readmissions (without service delivery costs). In the sensitivity analysis, NHS resources covered costs for non-elective (emergency) readmissions (without service delivery costs). This sensitivity is important because not all costs associated with readmissions are negative. Increasing the number of elective (planned) readmissions is, for example, a positive outcome that can be linked to improved access to health care and potentially better longer-term outcomes.

The cost-effectiveness analysis considered the additional resources to be invested for an improvement in the effects (i.e. avoided bed-days) associated with clinically- or housing-led schemes (comparison 1) and schemes with or without ‘step-down’ care (comparison 2), compared with standard care. The difference in mean 12-month costs and outcomes was estimated based on the model presented earlier. ICERs were calculated by dividing the difference in mean cost by the difference in mean length of stay per patients. ICERs indicated positive costs (NHS costs to be invested) and positive effects (saving in bed-days) (see Figure 12). Given that the number of hospital admissions per homeless person is four times higher than for the general public,128 reduction in homeless-related hospitalisation is a key priority for the NHS in England.

Main analysis

In the main analysis, NHS resources included costs for all readmissions (without service delivery costs).

The hospital costs to be invested by the NHS to save 1 bed-day were greater for clinically led schemes than for housing-led schemes (£2290.36 vs. £1665.10 per bed-day avoided), indicating that housing-led schemes are more cost-effective. If we assume that the NHS cost savings associated with a reduction of 1 bed-day are equal to £447 [we take an average cost per day of stay = £2323/6.86 = £339 (2007 value129 ≈ £447, using Bank of England rates130)] and we consider the NHS money invested by the NHS for readmissions within 1 year, the rate of return for the investment made [i.e. (net profit)/(total investment) × 100] is negative for both groups, meaning a lack of returns within the 1 year on the investment made.

Housing-led schemes were ‘less worse off’ than clinically led schemes (–73% vs. –80%, respectively). The NHS cost invested (per bed-day saved) would need to be lower than £447 to have a positive return on investment.

The ICER was greater for schemes with no ‘step-down’ care than for schemes with ‘step-down’ care (£2902.22 vs. £1115.94), indicating that step-down care is more cost-effective. Similar considerations (to those stated above on lack of returns within the 1 year on the investment) may apply here. Overall, the models suggest that the intervention groups (vs. usual care) are cost-effective. In particular, they can be resource saving (e.g. secure fewer bed-days in the hospital and, therefore, reduce their associated costs). However, from the budget impact analyses, we understand that within the first year the investment costs are greater than the gains.

Sensitivity analysis 2

In sensitivity analysis 2, NHS resources covered costs for non-elective (emergency) readmissions without service delivery costs.

The hospital costs for non-elective (emergency) readmissions to be invested by the NHS to save 1 bed-day were greater for clinically led schemes than for housing-led schemes (£739 vs. £675 per bed-day avoided), indicating that housing-led schemes are more cost-effective. If we consider a threshold of £477, the return on investment is still negative for both groups, with housing-led schemes less worse off than clinically led schemes (–34% vs. –40%, respectively).

The ICER was greater for schemes with no ‘step-down’ care than for schemes with ‘step-down’ care (£856 vs. £544), indicating that step-down care is still more cost-effective. The step-down schemes reported a smaller loss in investment in the first year than schemes with no ‘step-down’ care (i.e. negative return on investment of 18% vs. 48%, respectively).

Other sensitivity analyses (variations in overall costs or effectiveness for the comparator as well as longer follow-up)

More analyses are reported in Report Supplementary Material 3. The analyses looked at alternative scenarios where we varied the overall costs or effectiveness for the comparator. In addition, we looked at longer follow-up. The findings emerging from these alternative sensitivity analyses confirmed those in the main analyses above.

Report Supplementary Material 4 and 5 show the degree of uncertainty around the differences in costs and bed-days avoided between the intervention and control groups. The cost-effectiveness acceptability curves (see Report Supplementary Material 5) show that ‘housing-led’ schemes are more cost-effective than clinically led schemes, and schemes with step-down care are more cost-effective than schemes with no step-down care.

Implications for realist hypothesis testing

Overview

This section reports on the cost-effectiveness of three specialist ‘out-of-hospital’ care configurations using three different modelling techniques. We begin by briefly recapping on the different configurations, their geographical contexts and the data sources used in the modelling.

Results: primary analysis using linked Hospital Episode Statistics data (model 1)

Costs per patient for each readmission

Figure 13 shows the difference in annual NHS costs per patient for a range of readmission types (i.e. elective readmissions, emergency readmissions, other readmissions and costs for all admissions). Considering all readmissions, configuration 1 (clinically led/no step-down care) and configuration 2 (clinically led/residential step-down care) reported considerably greater costs than configuration 3 (housing led/community step-down care) (£7004 vs. £2468 vs. £1353; p < 0.01).

FIGURE 13.

Economic model 1: difference in annual NHS costs per patient across three out-of-hospital care configurations (control group = standard care from Hewett et al. 78).

Configuration 1 has the highest costs for elective readmissions, followed by configurations 2 and 3 (£5711 vs. £598 vs. £285; p < 0.01).

Configuration 2 has highest costs for emergency readmissions, followed by configurations 1 and 3 (£1874 vs. £1316 vs. £1036; p < 0.01).

Bed-days avoided per patient

Figure 14 shows the difference in annual bed-days per patient. Configuration 1 (clinically led/no step-down care) and configuration 2 (clinically led/residential step-down care) reported fewer bed-days avoided than configuration 3 (housing led/community step-down care). It should be noted that configuration 3 performs significantly better than the other two configurations on this measure in terms of the number of bed-days that it is saving (2.56 vs. 4.9 vs. 19.9 days per year per patient; p < 0.01).

FIGURE 14.

Economic model 1: difference in annual bed-days per patient across three out-of-hospital care configurations (control group = standard care from Hewett et al. 78).

Cost-effectiveness (resources to be invested per bed-day avoided)

This section explores the NHS costs that need to be invested per bed-day avoided.

Overall, we see that for all three out-of-hospital care configurations, the ICERs present positive costs and positive effects (Figure 15). Details on the cost-effectiveness and budget impact analyses for the following three separate models are discussed below:

1. main analysis – NHS resources included costs for all readmissions (without service delivery costs)

2. sensitivity analysis 1 – NHS resources included costs for all readmissions (with service delivery costs)

3. sensitivity analysis 2 – NHS resources included costs for non-elective (emergency) readmissions (with service delivery costs).

FIGURE 15.

Economic model 1: NHS resources to be invested per bed-day avoided (control group = standard care from Hewett et al. 78). Main analysis: NHS resources included costs for all readmissions (without service delivery costs). Sensitivity analysis 1: NHS resources include costs for all readmissions (with service delivery costs). Sensitivity analysis 2: NHS resources include costs for non-elective (emergency) readmissions (with service delivery costs).

Results: cost–utility for the NHS (model 2)

In this section, we report on additional analyses using linked HES data and EQ-5D data to establish the cost–utility of three different ‘out-of-hospital’ care configurations for the NHS.

The effect of different out-of-hospital care configurations on health-care resources, costs and health utility

Figures 13 and 16 report the difference in annual NHS costs and QALY outcomes per patient compared with standard care (control group 3: Hewett et al. 78). Costs were commented on above (see Model 1: what is the cost-effectiveness of homeless hospital discharge schemes for the NHS?). All three configurations confer better QALY outcomes than standard care (Figure 16).

FIGURE 16.

Economic model 2: difference in QALY outcomes per patient vs. standard care (Hewett et al. 78). Intervention: QALY data for configuration 1 were extracted from Hewett et al. ,78 whereas local data were used for configurations 2 and 3. QALY data for the control group (standard care) were extracted from Hewett et al. 78 The graph reports on the gains in QALY between intervention and control groups.

Configuration 3 (housing led/community step-down care) achieved the largest number of QALY gains (vs. standard care) followed by configuration 2 (clinically led/residential step-down care) and then configuration 1 (clinically led/no step-down care) (0.29 vs. 0.17 vs. 0.09).

Configuration 1 (clinically led/no step-down care) and configuration 2 (clinically led/residential step-down care) reported greater costs per QALY gain than configuration 3 (housing led/community step-down care) (£7004.14 vs. £2467.52 vs. £1353.22; p < 0.01).

Cost-effectiveness

Configuration 1 (clinically led/no step-down care) and configuration 2 (clinically led/residential step-down care) reported greater ICERs than configuration 3 (housing led/community step-down care) (£55,602 vs. £14,515 vs. £4743) (Figure 17). If a threshold of £30,000 is applied, then configuration 1 may not be cost-effective, whereas configurations 2 and 3 are cost-effective well within the NICE recommendations. It should be noted that when the multidisciplinary ‘clinically led’ homeless health-care team is working in a busy city with higher workload and dealing more frequently with people with more complex needs (and this may apply to configuration 1), it is likely to be costlier to achieve a QALY gain than when sites placed in smaller old industrial cities, dealing with fewer cases and less frequently with complex ones (see configurations 2 and 3).

FIGURE 17.

Economic model 2: additional costs to be invested per QALY gained. Main analysis: NHS resources include costs for all readmissions (without service delivery costs – this is aligned with the analyses in part 2 of this present report). Sensitivity analysis 1: NHS resources include costs for all readmissions (with service delivery costs). Sensitivity analysis 2: NHS resources include costs for non-elective (emergency) readmissions (with service delivery costs). Utility data used here are reported in Figure 16. The cost-effectiveness analyses reported positive costs and positive effects, and trade-offs between costs and effects were considered. This represents the situation where the intervention may be cost-effective compared with the control group, and the value at which the ICER is considered good value for money is provided by NICE (under the threshold of £20,000–30,000 per QALY).

Results: cost–utility for the NHS and broader public perspective (model 3)

In this section, we report on additional analyses using linked HES data and EQ-5D data to establish the cost–utility of different ‘out-of-hospital’ care configurations for the NHS and the wider public perspective. As noted above, this analysis is limited to configurations 2 and 3 because service delivery costs could not be ascertained for configuration 1.

The difference in annual costs for health-care resources (A&E and hospitalisation) showed that there is an annual patient saving in NHS costs [between £856 for configuration 2 (clinically led/residential step-down care) and £2466 for configuration 3 (housing led/community step-down care)] (see Figure 17). For both configurations, the cost saving was greater when considering the broader public provider perspective (£5758 and £4454, respectively). When considering total health-care costs (A&E and hospitalisation plus service delivery costs), there was a cost saving of £2318 for configuration 3, whereas for configuration 2 there was an increase in NHS costs to be invested, equal to £334. QALY gains increased to 0.16 for configuration 2 and 0.23 for configuration 3 (see Figure 17). More information on the sample, patient characteristics, use of resources, costs and health utility is reported in Report Supplementary Material 10.

The cost-effectiveness analyses (i.e. the total costs for the broader public sector perspective with service delivery costs) for configuration 3 (housing led/community step-down care) presented negative costs and positive effects (costs < 0 and effects > 0) (see Figure 18). For configuration 2 (clinically led/residential step-down care), we found positive costs and positive effects (costs > 0 and effects > 0; £2029 cost per QALY) (Figures 18 and 19). Both configurations were cost-effective according to NICE recommendations.

FIGURE 18.

Economic model 3: difference in total annual total costs (broader public perspective) and QALY outcomes per patient.

FIGURE 19.

Economic model 3: ICER and cost per QALY gained (project local survey sites, comparison before and after). NHS costs data covered hospitalisation, outpatient visits and A&E visits and are comparable across sites. ICER estimates based on public sector and total cost estimates are not comparable as the type of data available on the use of resources was inconsistent across sites.

Implications for realist hypothesis testing

Summary of economic evaluation findings

In terms of developing a conceptual road map for commissioners, the economic evaluation supports the development of HHDSs to ensure consistently safe, timely transfers of care for patients who are homeless. Table 23 provides a summary of the cost-effectiveness findings. Indeed, these findings provide much stronger evidence for HHDSs as compared with the previous chapter and it is important to note that the economic analysis has used a different control group. Across the different modelling exercises, the findings confirm the following:

• Specialist care is more effective and cost-effective than standard care.

• Clinically led advocacy (patient in-reach) increases costs for the NHS, but improves access to elective (planned follow-up) care. We interpret this as a positive outcome, indicating that SIHHC is working more effectively than standard care to improve access to health care among a marginalised group for whom one in three deaths are due to conditions amenable to timely health care.

• HHDSs with direct access to step-down intermediate care are more cost-effective than those without.

• When developing complex adaptive systems for out-of-hospital care, it is important not to underestimate the value of housing-led support. This is shown to be as effective and cost-effective as clinically led schemes on a wide range of measures.

• Overall, the evidence points to the potential of well-integrated HHDSs providing continuity of both clinical- and housing-related support, in the hospital and in ‘step-down’ intermediate care.

Strengths and limitations

The main strength of the analysis is that, to the best of our knowledge, the study is the largest evaluation of HHDSs ever conducted in England (and, to our knowledge, internationally). The study involved 17 different schemes (31 in the literature review), covering all England. To the best of our knowledge, no other study in this field has adopted the same multiple data linkage and multioutcome measures approach to assess the impact of hospital discharge service delivery on different aspects of cost-effectiveness and cost–utility for the NHS, as well as for the broader public perspective.

There were, however, several limitations and weaknesses inherent in the study and we summarise these in Table 24. The most significant of these is the lack of primary questionnaire data from the target research sites and the consequent need to use proxy data from an earlier RCT. 78 Proxy data from this RCT were used for both the standard-care control group and for specialist care in configuration 1. The most important learning here is about the importance of not overestimating the capacity of front-line services to support research through their routine monitoring and data collection. This point is also reinforced in a recent study of reablement services in which similar data collection problems were encountered. 133

As we were limited to a subgroup of sites from where QALY data were available (either collected using surveys or extracted from local audit data), this limited the configurations that could be studied. For example, we did not study sites that had access to intermediate care, but no resource for patient in-reach and discharge co-ordination. These standalone services had closed down (before evaluation could be completed), arguably lending some utility to the theory about the need for multiple and integrated mechanism resources.

As QALY data were difficult to collect and were incomplete, it may be biased towards those with ongoing health needs who re-engage with services and it may, therefore, systematically underestimate the impact on QALYs. In addition, the QALY data for the control group were linked to patients seen (randomised) by a clinically led service that may see more complex patients, potentially exaggerating the impact of the non-clinically led models.

Finally, delays in securing initial ethics approvals for the study, and in receiving data from NHS Digital, meant that the research team was able to start preparatory work on the HES data sets only 3 months prior to the scheduled completion date of the project. This left insufficient time to complete all the work that was planned. The project was granted a 12-month unpaid extension to cover the economic analyses. Unfortunately, after completion of the evaluation, further issues with the quality of the data received from NHS Digital put the project on hold for longer while the team was waiting for new data to arrive. In addition, more time was needed to check the new batch of data and make sure that the data received were suitable for analysis. The numerous delays put a strain on the team and also on the programme of analyses. Adjustments were to be made to the analyses of the HES data for the economic model. Use of resource data for the intervention groups were extracted from HES data as aggregate-level estimates instead of individual homeless user-level analyses. According to reasonable practice, the units (i.e. visit to A&E, hospitalisation type) were bundled at a level that matched the price weight. 134

Evidence in support of programme theory

Overall, there was good evidence from across the three WPs to support our programme theory about ‘what works best’ to deliver consistently safe, timely transfers of care:

• Employing a range of different economic modelling techniques, specialist HHDSs were consistently more effective and cost-effective than standard care (WP3).

• NHS trusts with specialist HHDSs had lower rates of DToCs linked to ‘housing’ than standard care (WP1).

• Employing a range of different economic modelling techniques, HHDSs with direct access to specialist intermediate care (step down) were more effective and cost-effective than HHDSs that had no direct access to intermediate care (WP3).

• The data linkage showed that HHDSs with a step-down service were associated with a reduction in subsequent hospital use, with an 18% reduction in A&E visits compared with HHDSs without a step-down service (WP2).

• Clinical advocacy (patient in-reach) provided by hospital-based homeless health-care teams increased access to planned (elective) follow-up care. This is an especially important outcome, as one in three deaths of people in our homeless hospital discharge cohort were due to common conditions (e.g. heart disease) that are amenable to timely health care.

Evidence that troubles programme theory

The data linkage shows that the HHDS cohort were more likely to be readmitted in an emergency, with five times the rate of unplanned hospital readmissions and five times the rate of A&E visits than housed people from deprived neighbourhoods. The data linkage also shows that one in three deaths of those in the hospital discharge cohort are from conditions that are amenable to timely health care. Although these outcome data introduce concerns regarding our programme theory (i.e. raise uncertainties about the efficacy of the HICM to deliver its intended outcomes), they are, we suggest, more indicative of a need to find ways to ensure that HHDSs have more of an impact. Using the metaphor of the lighthouse, we would argue that since the end of the HHDF, the lack of sustained recurrent funding has progressively dimmed their effects.

Need for increased investment in intermediate care

National audits of intermediate care135–137 (carried out in England until 2018) acknowledge a significant lack of investment and suggest that the capacity for intermediate care for all patient groups remains stubbornly stuck at a level below the threshold for whole-system impact. In one study, it is suggested that reablement provision needs to be increased by 200%. 24 One consequence of a lack of funding for intermediate care has been to tighten eligibility for the service. Although earlier guidance confirmed that this service should be accessible to homeless patients, later guidance has been silent on this matter, meaning that there has been little imperative to ‘mainstream’ the services piloted under the HHDF banner.

In Chapter 3, we described how the lack of investment in HHDSs is having an impact on scheme fidelity and the ability to achieve intended outcomes (i.e. the dimmer switch effect). Some HHDSs ceased to operate, whereas others have progressively reduced in reach and scale, sometimes reducing the number of hospitals that they are able to work with or reducing the size of the team (workforce). In one site, the increasing gap between workforce capacity and demand led to a range of problems, including recruitment and retention difficulties (resulting low team morale) and an increased focus on freeing-up hospital beds rather than other aspects of the service (such as patient advocacy). This was associated with poorer outcomes, including discharge to the street.

Need to increase delayed transfers of care (i.e. end discharge to the street)

As touched on above, where services were inadequate or lacking, it remained (tacitly) accepted practice (across both standard and specialist care sites) to discharge homeless patients to the streets rather than delay their transfer of care. Eradicating the ‘poor practice’ that led to discharge to the street was a key objective of the HHDF. However, we found little evidence to suggest that HHDSs had ‘fired’ a change in reasoning with regard to the stigma and cultural distance that positions ‘homeless patients’ as somehow less vulnerable and/or deserving than other groups of patients. We observed that patients who were homeless and using substances were particularly affected by unsafe discharge, especially when their behaviour was perceived to be challenging. Viewed through the lens of adult safeguarding, unsafe discharge may be viewed as neglect (acts of omission and a failure to provide access to appropriate health, care and support). 138 Strengthening adult safeguarding to challenge unsafe discharge is, therefore, an additional MIR necessary to improve outcomes. Certainly, we found no evidence that HHDSs could be deployed as short-term interventions to upskill hospital staff on homelessness issues. Once schemes ended, the high turnover of hospital staff meant that any benefits soon evaporated.

As homeless patients are discharged to the street (unsafely) rather than ‘delayed’ and counted as a DToC, this means that capacity issues linked to housing (i.e. ‘no place to discharge to’) are rendered largely invisible. When DToCs reach crisis point, the government will usually intervene to create extra capacity in adult social care provision. In the winter of 2016/17, DToCs hit an all-time high when, on average, 6660 people were being ‘delayed’ every day in NHS beds. In March 2017, the government introduced a new target to reduce the level of bed-days ‘lost’ to 3.5% and announced an extra £2B of funding for social care services. DToC metrics are, therefore, an important lever for system transformation in that they enable partners to work together to address pinch points in the system. Without this intelligence, the true impact of specialist HHDSs is likely to remain unknown and, as a consequence, appropriate levels of investment in their future unlikely to be forthcoming. From August 2020, new hospital discharge guidance was introduced by NHSE&I,91 including new arrangements for recording delays. The new system measures the number of people who have a length of stay over 14 and 21 days who have been judged as ready to discharge but who continue to reside in a hospital bed. Category xiii covers people who ‘continue to reside’ because of ‘homelessness/no right of recourse to public funds/no place to discharge to’91 (contains public sector information licensed under the Open Government Licence v3.0).

Need for increased investment in housing, care and support

The second dampening effect was shown to be the wider context in which some HHDSs were situated (i.e. where there was a shortage of housing, care and support). The findings of this study strongly support those reported in the national evaluations of intermediate care for older people, where it is reported that interventions that are shown to work well in areas with well-resourced and efficient community support services will have much reduced impact in areas where services are inadequate or lacking. 83 In Chapter 3, we observed how HHDSs could become blocked, as these ‘time-limited’ interventions started to substitute for permanent supportive housing.

Need for investment in chronic care management

Compared with the comparator group (matched housed patients), patients in the HHDS cohort had much higher levels of multimorbidity or combinations of long-term conditions or illnesses, with 8% of patients having five or more conditions (vs. 3% in the comparator group).

The data linkage confirms previous findings that external causes of death are considerably more common in homeless groups than in those from socially deprived areas. However, this investigation also highlights the importance of causes such as coronary heart disease, respiratory disease and cancer in homeless people. Compared with the recent ONS analysis of deaths of homeless people in England and Wales, we found that the proportion of deaths due to cancers and chronic lower respiratory disease was higher, and the proportion of alcohol- and drug-related deaths and suicide deaths was lower. In the comparator group, women and men were nearly two times more likely to die than the general population. In the homeless hospital discharge group, women were nine times, and men seven times, more likely to die than the general population. In the homeless hospital discharge group, 600 patients died between 1 November 2013 and 30 November 2016. Males made up 78% of deaths. The median age of death was 52 years for the homeless hospital discharge group and 72 years for the comparator group.

These findings alert us to the strong possibility that, regardless of scheme typology, the benefits of any type of short-term (time-limited) intervention targeted at this specific population group will quickly evaporate if they are not embedded as part of a fully integrated complex adaptive system that encompasses adequate provision for longer-term chronic care management and, indeed, palliative care. It surely goes without saying that more preventive working is needed to reduce homelessness and prevent these conditions from arising in the first place.

Mixed evidence for multidisciplinary team working

The economic evaluation (WP2) provides robust evidence that ‘clinically led’ (multidisciplinary) HHDSs increase the number of elective readmissions (i.e. access to planned follow-up care). Therefore, there is a strong case to be made that increasing investment (scaling-up) these HHDSs could work to reduce the numbers of deaths due to conditions that are amendable to timely health care.

However, the economic evaluation also finds that (cheaper) uniprofessional (housing-led) schemes are as effective and cost-effective as multidisciplinary (clinically led) schemes on a wide range of other measures (WP2). In this present study, we observed that clinically led teams, because of their primary focus on improving the quality of inpatient care [i.e. patient in-reach and discharge co-ordination (MIR2/3)], do not provide the same level of access to ‘step-down’ care as housing-led schemes (MIR4). A distinctive feature of nearly all the housing-led schemes is that workers ‘in-reach’ into the hospital and will then provide floating support for a time-limited period until community services are in place and working well (MIR4hf). This association with intermediate care (continuity of support post exit from the acute sector) may be the more significant factor in accounting for the comparable performance of housing-led schemes on some key measures (i.e. clouding our understanding of any distinctions to be drawn between multidisciplinary and uniprofessional working). A key finding of the qualitative phase of the study was how quickly the benefits of a hospital stay could evaporate where there was a lack of practical support immediately after discharge. In other words, a lack of intermediate care (MIR4) may dampen the effects of specialist MDT working and discharge co-ordination (MIR3mdt).

Indeed, it is important to acknowledge the value of uniprofessional housing-led HHDSs. Most likely, these positive results are reflecting the value of good-quality ‘floating support’ in bridging the gap between the hospital and the community. This is an important point because housing- and community- (voluntary-) led services are often assumed by commissioners to be the ‘icing on the cake’ and, therefore, the most easily dispensable part of the out-of-hospital care system. 86 Arguably, this evaluation establishes that they are far more important than has hitherto been assumed in terms of adding value for both the NHS and the wider public sector.

Health

• Had suffered a hypoxic brain injury. There is a query regarding his decision-making capacity (because of his age he does not meet the criteria for memory clinic assessment).

• Has diabetes and is forgetful and unable to manage own medications, including insulin. He often refuses to eat. He is having frequent hypoglycaemic episodes and also some high blood glucose levels. He is unaware of the symptoms of hypoglycaemia and is unable to look after himself in this situation. This can be life-threatening.

• He has a mild learning disability.

Co-existing mental health and substance misuse

• Has a complex mental health history, including past episodes of psychosis. Hears voices episodically and these are usually telling him to harm himself.

• Has a history of drug and alcohol misuse. There have been numerous occasions where he has had periods of being heavily intoxicated. This has then had a significant impact on his ability to self-care, including taking vital medication such as insulin. He has also been associated with antisocial behaviour and disturbances to other residents of accommodation services such as hostels.

Care and support needs

• Safeguarding concerns. Neighbours were financially abusing Jason.

• Self-neglect of personal care. Occasional urinary and bowel incontinence.

• Cannot maintain a habitable home environment without support.

• Has difficulties reading and writing and engaging in work, employment and education.

• Is unable to use public transport (make use of community facilities, including GP access) because of being unable to read numbers and letters and has no sense of direction unless he is accompanied by staff.

Housing problem

• Is homeless.

Care planning outcomes to be achieved through Bradford Respite and Intermediate Care Support Service stay

• Improve self-management of diabetes (including medication).

• Maintain personal hygiene/address self-neglect.

• Make use of necessary facilities (e.g. attend hospital appointments).

• Address co-existing mental health/substance misuse.

• Develop family relationships (with children).

• Maintain habitable home environment (including benefits and money management).

Multidisciplinary working co-ordinated through Bradford Respite and Intermediate Care Support Service

Configuration 1

A multidisciplinary ‘clinically led’ (homeless health-care team) scheme that offers patient in-reach and specialist discharge co-ordination, with no direct access to ‘step-down’ intermediate care. Here, support for the patient usually ends at the point of exit from the acute sector (hospital). Two examples of this configuration were considered, one situated in a city and one in a seaside resort in the south of England.

Configuration 2

In this site, two HHDSs work together in an integrated way. The configuration studied here comprised a clinically led (homeless health-care) team that offered patient in-reach and discharge co-ordination. This site had access to a 14-bed residential facility that offered step-down intermediate care. The hospital-based homeless health-care team provided ‘clinical in-reach’ into the residential intermediate care facility, ensuring continuity of multidisciplinary support for patients. The HHDSs are based in an old industrial city in the north of England.

Configuration 3

This site contained a single housing-led HHDS in which a small team of housing support workers visited the hospital to provide patient in-reach and discharge co-ordination. The housing workers then continued to support patients in the community (offering floating support) until longer-term services were in place and working well (i.e. community ‘step-down’ intermediate care). Although a uniprofessional scheme, it had access to a multidisciplinary clinically led homeless health-care team based at the hospital. The HHDS is based in an old industrial city in the Midlands.