Integration, effectiveness and costs of different models of primary health care provision for people who are homeless: an evaluation study

Article history

The research reported in this issue of the journal was funded by the HSDR programme or one of its preceding programmes as project number 13/156/03. The contractual start date was in April 2015. The final report began editorial review in September 2021 and was accepted for publication in June 2022. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HSDR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2023 Crane et al. This work was produced by Crane et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2023 Crane et al.

Background

Since 2010, homelessness has increased substantially across England. Contributory factors include high housing costs and a shortage of affordable housing; the ending of assured shorthold tenancies in the private rented sector; welfare benefit changes and sanctions, including the capping and freezing of Local Housing Allowance; and cuts to social support budgets. 1 A 2018 report suggested that approximately 200,000 single people experience homelessness each year. 2 Many stay in hostels, bed-and-breakfast hostels or with friends or relatives, and move from place to place. Others ‘sleep rough’ on the streets, in vehicles or parks, or in other public places. The number of rough sleepers in London increased from 3673 in 2009/10 to 11,018 in 2020/21. 3,4 Of the 2020/21 number, 7531 were described as ‘new’ rough sleepers.

Physical health, mental health and substance misuse problems are common among people who are homeless. 5–7 Their health needs are greater than those of the general population, and many have multiple long-term conditions and die earlier. 8–10 People sleeping rough are exposed to damp and the elements, are at risk of exposure and hypothermia, and are susceptible to infestation. Chronic respiratory disorders and circulatory and gastrointestinal problems are common. Physical health problems are aggravated by alcohol use, drug use and malnutrition, and injuries from accidents and assaults are common. Homelessness is also associated with demoralisation and depression. Health problems among people who are homeless are exacerbated by their unsettled lifestyle and sometimes disorganised behaviour, which can reduce their engagement with treatment programmes. Many also face barriers to accessing health care, including inflexible services, negative attitudes from some staff, and the challenges of treating complex and multiple needs. 11 They make unusually high demands on emergency health services, such as accident and emergency (A&E) departments. 12 A 2010 Department of Health (DH: known as Department of Health and Social Care (DHSC) since 2018) study estimated that this group consumes around four times more acute hospital services than the general population, costing at least £85M each year, and hospital stays are, on average, three times longer than those of the general population. 13

Since the 1980s, specialist primary health care services for homeless people have been established in several UK towns and cities. Their development took various forms, including dedicated ‘walk-in’ health centres and mobile health teams visiting hostels and day centres. 14 The National Health Service (Primary Care) Act 199715 provided the statutory framework for the development of Personal Medical Services (PMS). Through flexible contractual arrangements, health professionals were encouraged to deliver primary health care to underserved groups, including people who were homeless. According to Wright,16 this was ‘the most significant favourable piece of legislation for homeless people since the start of the NHS’. There have, however, been very few evaluations of these services, and their success in engaging people who are homeless in health care is unknown.

The 2010 DH study13 grouped specialist primary care provision for people who are homeless into four models: (1) mainstream general practices providing special services for people who are homeless, (2) outreach teams of specialist homelessness nurses, (3) full primary care specialist homelessness teams and (4) a fully co-ordinated primary and secondary care service. The analysis was unable, however, to demonstrate whether or not the provision was fully meeting the needs of people who were homeless. The study reported lack of systematic data on the use of health services and on costs, and lack of research evidence of the potential to improve primary care and health outcomes, and reduce secondary costs.

There are long-standing debates about whether primary health care for people who are homeless should be provided by mainstream or specialist services. Several researchers and clinicians believe that some targeted provision is necessary to reach people on the streets, but the aim should be integration into mainstream general practice services. 16–18 A survey of 86 people who were homeless found that 84% preferred specialist primary health care services. 19 A 1999 survey in England of managers of services for people who were homeless found that the majority favoured integration into mainstream primary health care services for their clients, believing that separate services were divisive. 20

Study proposal and aims

In 2013, the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research Programme issued a call for studies on the effectiveness and cost-effectiveness of innovative and integrated health and care services for homeless people. In response to this call and to address the knowledge gap identified in the DH study,13 this research proposal was submitted and funded.

The overall aim of the HEARTH study was to evaluate the effectiveness and costs of different models of primary health care provision for people who are homeless, with special reference to their integration with other services, and how this affected a range of health, social and economic outcomes. The objectives were as follows.

• To identify (1) the prevalence of specialist primary health care services for single people who are homeless and their geographical distribution, (2) types of models found in different NHS regions and key characteristics of these services and (3) areas with a homeless population but no specialist health care service.

• To examine the characteristics and integration of different models of primary health care services for people who are homeless with dental, mental health, secondary health, substance misuse, homelessness sector, housing and social care services.

• To examine the effectiveness of different models in (1) engaging people who are homeless in health screening; (2) responding to the physical health, mental health and social care needs of people who are homeless; and (3) providing continuity of care for health problems including long-term and complex conditions.

• To evaluate the impact of different models over time on service users’ health and well-being, and their use of other health and social care services including dental, emergency and secondary care.

• To investigate the resource implications and costs of delivering services for the various models.

• To compare the various models across a range of outcomes, reflecting service user and NHS perspectives, using a cost–consequences framework.

• To provide evidence to NHS commissioners and service providers regarding cost-effective organisation and delivery of primary health care to people who are homeless.

It was proposed that four Health Service Models would be evaluated, including a ‘usual care’ model for comparison.

1. Health centres specifically for homeless people, comparable to the DH’s full primary care specialist homelessness team, but located at a fixed site.

2. Mobile Teams that run sessions in homeless services such as hostels, comparable to the DH’s outreach team of specialist homelessness nurses.

3. Mainstream general practices that also provide specialist services for people who are homeless.

4. Mainstream general practices that provide ‘usual care’ services to the general population, which by default include people who are homeless. This type of provision was not included in the DH models, but is commonly used by people who are homeless if there are no local specialist services.

The research questions that the study would address were as follows.

• Which models or service elements are more effective in engaging people who are homeless in health screening and health care?

• Which models are more effective in providing continuity of care for long-term or complex health conditions?

• What are the associations between integration of the models with other services and health outcomes for people who are homeless?

• How satisfied are service users, primary health care staff and other agencies with the services?

Layout of this report

Chapter 2 presents literature reviews undertaken during the study, and Chapter 3 describes the study design and methodology. Chapter 4 summarises the findings of the mapping exercise. Chapters 5 and 6 set the scene, by describing the CSSs and the case study participants. Chapters 7–13 focus on primary and secondary outcomes. Chapter 14 examines ways in which contextual factors and mechanisms of health care delivery are likely to have had an influence on outcomes. Finally, the conclusions and implications for NHS commissioners and primary health care managers and practitioners are discussed in Chapter 15.

Review A: primary health care for people who are homeless – evidence-based practice

This scoping review examined evidence-based practice of the delivery of primary health care to people who are homeless, focusing on models of provision and methods of delivering interventions. The inclusion criteria were single people aged 18 years and older who were homeless, but not homeless families or children. It involved the delivery of general medical services (GMS) by general practitioners (GPs), primary care physicians, practice and community nurses, and specialist primary health care teams. It did not include studies of the prevalence of health conditions or clinical features of illnesses, nor studies of specialist services that were not part of primary health care teams.

Review B: health policy developments in England relating to people who are homeless

Several policy developments in England since the 1990s addressed the delivery of primary health care to people who are homeless. A Royal College of Physicians’ working party on homelessness and ill health in the 1990s17 recommended that the DH should introduce systematic monitoring of the health of people who are homeless and their use of health services, and that the government should fund special primary care practices for people who are homeless, and restructure deprivation payments to GPs. The National Health Service (Primary Care) Act 199715 enabled the development of PMS, which stimulated the development of primary health care services to people living in deprived communities, and to underserved and disadvantaged groups, including people who are homeless. Local Development Schemes were introduced by the DH in 1998, enabling additional payments for GPs and allied staff to provide services in deprived areas (later known as ‘enhanced services’). The extra funding incentivised GPs, for example, to register and provide medical care to people staying in hostels.

Underpinning these developments, Addressing Inequalities: Reaching the Hard-to-reach Groups was published by the DH in 2002 as a practical aid for primary care. 42 Among its recommendations were primary care trusts [replaced by Clinical Commissioning Groups (CCGs) in 2013] should encourage GPs and nurses to focus on hard-to-reach groups via PMS and/or Local Development Schemes. The Royal College of General Practitioners (RCGP) produced a statement on homelessness and primary care in 2002, with recommendations for practices and primary care trusts. 43 In 2004, primary care organisations could commission new Alternative Provider Medical Services (APMS) to provide essential primary care services, additional services where GMS/PMS practices opted out, enhanced services and out-of-hours services. They could contract these services from providers, including commercial and not-for-profit agencies, and NHS foundation trusts. 44

Influential reports, such as Securing Good Health for the Whole Population45 and the 2010 Marmot Review,46 ensured that equalising health outcomes across society gained prominence within national policies. In 2010, the Social Exclusion Task Force launched Inclusion Health, a framework for driving improvements in health outcomes for socially excluded groups. A DH report published alongside Inclusion Health acknowledged that health care for people who were homeless may have been historically underfunded due to inaccurate population data. 13 A National Inclusion Health Board was established to lead the Inclusion Health agenda. Just 3 months later, however, there was a change of government and the Social Exclusion Task Force was disbanded.

The Health and Social Care Act 201247 transferred NHS commissioning responsibilities to CCGs prompting greater general practice control of service provision. Under the Act, CCGs were required to reduce health inequalities and provide integrated services. 48 Health and well-being boards were established by local authorities to act as forums whereby health and social care commissioners and providers could address the health and well-being of local populations and promote integrated services. 49 These boards were required, with CCGs, to produce a Joint Strategic Needs Assessment and a Joint Health and Wellbeing Strategy for their local populations. Leng50 produced a guide to help local authorities improve health and well-being among people who are homeless, and reduce health inequalities. Healthwatch was developed in 2013 as a patient and service user champion at local and national levels. This has taken a role in monitoring some services for people who are homeless.

Public health responsibilities moved from primary care trusts to local authorities in April 2013, and Public Health England (PHE) was established. PHE brought together public health specialists into a single public health body responsible for protecting and improving the public’s health and reducing health inequalities. Its call for action in 2015, All Our Health: Personalised Care and Population Health,51 urged health care professionals to use their skills and relationships to maximise impact on avoidable illness and promote well-being and resilience. PHE produced a framework and issued guidance on homelessness, both of which were updated in 2019. 52,53 It recommended that health and well-being boards should ensure that homelessness is addressed in joint strategic needs assessments and health and well-being strategies, and that the relationship between health and homelessness is acknowledged in local housing authorities’ homelessness reviews. PHE also produced guidance on tackling TB, a disease disproportionately affecting people who are homeless. 54

In 2016/17, Sustainability and Transformation Plans were developed in 44 areas across England as a new planning process for health and, to some extent, social care. Renamed Sustainability and Transformation Partnerships, they required NHS organisations to develop plans for local health services, including working with local authorities and other partners. They represented an important shift in DH policy: although the Health and Social Care Act 201247 sought to stimulate competition within the health care system, NHS organisations were asked to collaborate, rather than compete, to plan and provide local services. 55 The Health and Care Act 2022 was passed in April 2022. This legislation required Integrated Care Systems (ICSs) to be established in England on a statutory basis from 1 July 2022. ICSs replaced Sustainability and Transformation Partnerships as outlined in the White Paper Integration and Innovation: Working Together to Improve Health and Social Care for All. 56 Forty-two ICSs have been established across England with a responsibility to plan and fund health and care services in their area. Each ICS is made up of an integrated care board (ICB) and an integrated care partnership (ICP). ICBs replaced CCGs and are responsible for planning NHS services, including ambulances, primary care, mental health care, hospital (acute services), community and specialist care, and are accountable to NHS England. A representative from ICBs is required to participate in local Health and Wellbeing Boards, in place of CCG members. ICPs cover public health, social care and wider subjects impacting the health and wellbeing of their local populations, and operate as a statutory committee between the ICBs, local authorities, and voluntary and community organisations. The establishment of ICSs represents the first large-scale structural change to the NHS since 2012 (NHS Confederation 2022). Further details are in Chapter 15.

Linked to such initiatives have been several policy moves to tackle homelessness. The Ministerial Working Group on Homelessness was formed in 2010, and 1 year later a ‘No Second Night Out’ scheme was launched to ensure that people who were sleeping rough received help quickly. 57 Around the same time, ‘Housing First’ pilots were introduced into the UK. This model originated in New York in 1992, with the premise that stable housing is key to tackling chronic homelessness, and should be secured before problems such as substance misuse and mental illness can be addressed. 58,59 Significant modifications to the model have, however, made it difficult to assess the influence of the model on programme outcomes. 60,61 Several researchers have concluded that Housing First provides improvements in housing stability, but, apart from the use of emergency health services, there is little evidence as yet to suggest that it produces better outcomes for physical health, mental health and substance misuse than treatment in the community. 62,63

The Homelessness Reduction Act 201764 placed many public authorities, such as emergency departments (EDs), urgent treatment centres and in-patient hospitals, under a duty to refer people at risk of homelessness to the local authority. The Act’s focus on prevention, as well as on developments of planned individual support, is arguably the most important policy development across the NHS and local authorities for people who are homeless or insecurely housed since the Housing (Homelessness Persons) Act 1977. 65 In 2018, the Ministry of Housing, Communities and Local Government published a new Rough Sleeping Strategy, with a commitment to halve rough sleeping by 2022 and end it by 2027. 66 In the first year, it allocated up to £2M to enable access to health and support services for people sleeping rough.

Published in 2019, the NHS Long Term Plan sets out new action the NHS will take to help tackle health inequalities, with £4.5B over the next 5 years. 67 Up to £30M is being invested into the needs of people sleeping rough, and ensuring areas most affected have better access to specialist NHS mental health support, integrated with existing outreach services. New primary care networks are intended to bring together general practices to work at scale together to focus on local patient care.

Although many of these developments reflect primary care services and their interface with local authorities, policy attention has also focused on improving services for people who are homeless and in hospital. Pathway, a charity founded in 2009, introduced the ‘Pathway’ model of integrated care to bridge the gap between primary and secondary care. This involves specialist primary health care services collaborating with secondary care services to improve care in hospital for people who are homeless. In 2020, The Royal College of Emergency Medicine produced a best-practice guide, Inclusion Health in the Emergency Department: Caring for Patients who are Homeless or Socially Excluded. 68 A 2021 evaluation concluded that specialist approaches to hospital discharge for people who are homeless are more effective and cost-effective than standard care involving discharge to the streets without support. 69

Over the decades, policy attention to the health of people who are homeless has been driven by organisations such as The Faculty for Homeless and Inclusion Health (supported by Pathway), a multidisciplinary network of health care workers and experts by experience. It produced a set of standards for commissioners and service providers in 2011 covering health care planning, commissioning and provision for people who are homeless (revised in 2013 and 2018). 48,70 The London Homeless Health Programme, formed in 2015 as part of the Healthy London Partnership, issued guidance in 2016 (updated in 2019) for London’s CCGs on improving health outcomes for people who are homeless. 71 It recommended there should be a Homeless Health Lead in every CCG to champion the homeless health agenda and engage with London’s homeless health clinical networks. In partnership with Healthwatch London and Groundswell (a charity supporting people who are homeless), it produced ‘my right to access healthcare’ cards to help people who are homeless register with a general practice. The Queen’s Nursing Institute has developed a Homeless and Inclusion Health Programme.

Efforts to improve health care for people who are homeless are continually advocated by voluntary sector homelessness organisations, including Centrepoint, Crisis, and St Mungo’s. In 2009–10, Homeless Link was funded by the DH Third Sector Investment Programme to pilot a Homeless Health Needs Audit Tool, to help health service commissioners and providers, and local authorities, gather data about the health needs of local people who are homeless and their use of health services. 11 An updated audit tool, with funding from PHE, reflected new local commissioning environments and other changes. 72

As a result of the COVID-19 pandemic, the NHS and English housing authorities implemented the ‘Everyone In’ initiative, to mitigate the effects of the pandemic on people experiencing homelessness. Under the initiative, £3.2M was made available to local authorities in March 2020 to accommodate people sleeping rough or in accommodation where it was difficult to self-isolate, such as night shelters. 73 By the end of November 2020, more than 33,000 people had been helped to find accommodation, and, according to the National Audit Office, COVID-19 infections and deaths among this population had been relatively low. 74 In November 2020, the ‘Protect Programme’ backed by £15M, was announced to ensure that vulnerable people, including those sleeping rough, were to be protected during the second period of national restrictions and throughout the winter. 73 In April 2020, NHS England and NHS Improvement also produced clinical advice and guidance on delivering health care to people who are homeless during the COVID-19 pandemic. A new National Institute for Health Protection has been formed with a focus on biosecurity and other elements of public health; the implications of this for homelessness policy and services will no doubt emerge. 75

Theoretical and conceptual framework

It has long been recognised that those who are in most need of health care are least able to access services, a phenomenon termed the ‘inverse care law’. 76 In terms of people who are homeless, the belief was that their complex needs could not be met by mainstream GPs; therefore, specialist primary health care services were established in some areas (see Chapter 1). Some theorists associate the exclusion of people who are vulnerable with problems of discrimination and the ‘bureaucracy’ and regimes of formal services, which result in them being inadvertently or deliberately excluded. Merton77 associated the exclusion from mainstream society of people who are homeless with ‘retreatist’ behaviours, and an inability or unwillingness to comply with society’s norms and values.

Using a case study approach, the HEARTH study examined the effectiveness of different models of primary health care services for people who are homeless to determine what works, for whom and in what circumstances. This approach allows researchers to investigate a contemporary phenomenon within its real-life context, and enables multiple sources of evidence to be gathered. 78 The role of contextual factors and mechanisms in the delivery of health care to people who are homeless, and how these influence outcomes, were examined at CSSs. Contextual factors included the wider health and care system, financing, staff and physical resources, and the availability and accessibility of other relevant services. Mechanisms included strategies used by CSS staff to engage with people who were homeless and provide health screening and treatment (Table 1). The theoretical framework for this was informed by Andersen’s79 behavioural model of health service use by vulnerable populations, and applied by Gelberg et al. 80 to people who are homeless. The model has three domains: (1) population characteristics, such as demographics, personal and family resources, community and health services resources, and perceived health needs; (2) health behaviour, such as lifestyle factors and use of health services; and (3) outcomes, such as satisfaction with care, and the availability and accessibility of health services.

Integration between each CSS and relevant services were measured, drawing on measures adopted by Browne et al. 81 and Joly et al. 82 Distinctions were made between (1) types of services, for example health, housing and social care; and (2) organisations involved at different stages of care such as hospital services, and those that provided complementary services. The ‘depth’ of integration between the CSS and each service was scored by staff, according to the extent to which they were involved with a service, and the extent to which they should be involved (see Chapter 11). Similar information was collected from external health and social care agencies.

Study approvals and management

The study started in April 2015. It received ethics approval from the London – Bloomsbury Research Ethics Committee on 5 October 2015 (reference number: 15/LO/1382) and study-wide governance approval from the lead Clinical Research Network (CRN) on 19 October 2015, and local NHS Research Governance approval was granted for each study site as they were recruited. As the study progressed, three substantial and seven non-substantial amendments were approved by the London – Bloomsbury Research Ethics Committee and the Health Research Authority, and four changes were made to the original research protocol. The non-substantial amendments concerned the addition of CSSs to the study, and the substantial amendments concerned changes to the research protocol (see following sections).

A Study Steering Committee (SSC) was formed; it met annually and guided the research team throughout the study.

Phase 1: mapping of specialist primary health care services

The mapping exercise aimed to (1) examine the prevalence, geographical distribution and characteristics of specialist primary health care services in England for single people who are homeless; (2) determine the extent to which temporary accommodation and day centres for single people who are homeless had access to specialist primary health care services; and (3) collect information from those not linked to specialist health services about accessing primary health care.

Using semistructured questionnaires, two complementary surveys were undertaken. The first collected information from specialist primary health care services about their service such as opening hours, staff composition and patient groups (see Report Supplementary Material 2). The second survey collected information from managers of hostels and day centres for single people who are homeless about their project, access to primary health care for clients and general practices used, clinics run by doctors or nurses at their project, and whether or not the primary health care needs of their clients were being met (see Report Supplementary Materials 3 and 4). Details of the methods are available,21 and summary findings reported in Chapter 4.

Phase 2: evaluation of models of primary health care provision

An evaluation of four models of primary health care provision for people who are homeless was undertaken:

1. Two specialist health centres primarily for people who were homeless (Dedicated Centres).

2. Two mobile homeless health teams that held clinics in hostels or day centres for people who are homeless (Mobile Teams).

3. Two general practices with special services for people who were homeless (Specialist GPs).

4. Four generic general practices providing ‘usual care’ services to the local population, including to people who were homeless (Usual Care GPs). The intention was to have two sites, but because of difficulties recruiting general practices and insufficient numbers of patients who were homeless at these sites, two additional practices were added.

Fieldwork ran from January 2016 to June 2019. Data were collected through interviews with CSS managers and staff, and local health, social care and welfare agencies. People with a current or recent history of homelessness and registered with the CSS were recruited as ‘case study participants’, and information was collected from them and from their medical records about their health and service use over 12 months.

Primary outcome: health screening of people who are homeless

The primary outcome was the extent of health screening among people who are homeless. Six Health Screening Indicators (HSIs) were selected: body mass index (BMI), mental health, alcohol use, TB, smoking and hepatitis A. The HSIs are a set of minimum standards or ‘markers’ from a clinical perspective, and extend beyond screening alone, as evidence of an intervention was sought if a problem was identified. Screening for BMI, mental health, alcohol use and TB covered the preceding 12 months, smoking covered the preceding 24 months and hepatitis A required that a vaccination programme was in progress or had been completed in the preceding 10 years. The HSIs were derived from existing guidelines,83–87 and from the expert opinion of Ford (retired GP), who consulted two generic GPs, two GPs specialising in homelessness and a hospital physician. The six HSIs were included in a list of health screening indicators for people who are homeless in the Faculty for Homeless and Inclusion Health’s standards for primary care. 48 They were selected for our study because smoking, mental health and alcohol problems are common among single people who are homeless (see Chapter 1), and their diets are often poor. 88 Moreover, compared with the general population, hepatitis A and active pulmonary TB are relatively common. 89–91

Data for the HSIs came solely from the participants’ medical records, which were accessed at the end of the study. A score of 1 was given to each HSI if there was evidence in the medical records of both screening having taken place and an intervention being offered, if applicable, thus giving a total range of 0–6. Further details are in Chapter 7.

Secondary outcomes

Recruitment of Case Study Sites

Drawing on the mapping exercise, primary health care services working with people who were homeless were identified for the three specialist models. Their selection depended on whether or not they responded to the mapping exercise, the number of patients who were homeless, and staff resources or imminent changes to the service that affected their ability to participate.

Early on it became apparent that recruiting general practices for the Usual Care GP model would be exceptionally difficult. The mapping exercise revealed that many large hostels were linked to specialist primary health care services. Among those that were served by a mainstream general practice, some hostel managers declined involvement, and some agreed, but the general practices failed to respond or declined as they were too busy. General practices serving smaller hostels were therefore considered, although this meant fewer potential case study participants per site.

Substantial time and effort were spent liaising with general practices and primary care leads of CRNs to try and recruit Usual Care GPs. It was agreed with the SSC that, instead of two sites for this model, attempts would be made to recruit additional practices to reduce the number of participants required at each CSS. In collaboration with five CRNs, attempts were made to recruit a cluster of hostels and general practices in an area. This proved successful in two locations, and eventually four Usual Care GPs participated. Recruitment of participants at the final site did not commence until April 2018.

Case studies of health and service use over 12 months for people who are homeless

At each CSS, patients who were currently or recently homeless were recruited as case study participants and information was collected about their health and service use over 12 months.

Interviews with Case Study Site staff and sessional workers

Face-to-face interviews were conducted by the fieldwork research team with 65 CSS staff and sessional workers, including practice managers, primary health care nurses, GPs, mental health nurses, drug and alcohol workers, case managers, health care assistants (HCAs) and receptionists. Using a template (see Report Supplementary Material 9), operational and performance data were collected from CSS managers about (1) the development of the CSS; (2) current operation, including staffing, client groups served, opening hours, registration, funding and types of services provided; (3) integration with local health, dental, welfare and social care services; and (4) involvement of the CSS in local strategy and service development.

Using an interview schedule adapted for different job roles (see Report Supplementary Material 10), interviews with CSS staff and sessional workers collected information about (1) length of time with the CSS, role within the team, qualifications and experience of working with people who are homeless; (2) services provided to people who are homeless, and strategies to encourage engagement; (3) collaboration with local agencies and services; and (4) perspectives of the CSS’s strengths and limitations. All consented for their interview to be recorded. They also rated their actual and expected levels of integration with other services (see Chapter 11). Each interview lasted approximately 60 minutes. They were sent a participant information sheet prior to the interview explaining the study and their participation.

There were delays in arranging some interviews due to staff workload. Where sessional workers were employed by an organisation that was not the CSS, additional research and development approvals were required. Our intention had been to hold a focus group with CSS staff towards the end of the study to gather their reflections on their work as a team. However, owing to their work pressures, individual interviews were more appropriate.

Interviews with local service providers and stakeholders

To examine the wider context in which the CSS delivered care, interviews were also conducted with 81 local service providers and stakeholders. All except three were face-to-face, and all but two were recorded. Service providers included street outreach workers, hostel and day centre managers, and drug and alcohol workers who were not part of the CSS. Using an interview schedule (adapted for different job roles), information was gathered about (1) their work with people who are homeless, (2) their awareness of the CSS and referral procedures, (3) use of the CSS by their clients or reasons for non-use and (4) their perspectives of the CSS (see Report Supplementary Materials 11 and 12). Each person was asked to rate their actual and expected level of integration with the CSS. Interviews lasted approximately 45 minutes. They also were sent participant information sheets prior to the interview.

Interviews with three local authority service commissioners and four Healthwatch directors collected information about local strategies and plans for health care delivery to people who are homeless, and the role of the CSS in local health provision. It was not possible to arrange interviews with some stakeholders.

Interviews with people who were homeless and not using the Case Study Sites

In eight CSS areas, 107 interviews were conducted with people who were homeless and not registered with the CSS. Using a short, structured questionnaire (see Report Supplementary Material 13), information was gathered about their awareness of the CSS, reasons for not using it, and their health needs and use of health care services. They were recruited at hostels and day centres, and were offered £5 as thanks for taking part. Staff at these sites handed out participant information sheets to potential participants, and those interested were introduced to the research team. The research team explained the study in more detail, checked the potential participant’s eligibility and obtained their written consent to participate. Peer interviewers conducted 49 of these interviews (described in Patient and Public Involvement).

Observations

Observations were undertaken of ways in which CSS staff engaged with people who were homeless, and worked with other agencies. This mainly involved observations in CSS reception areas and in day centres while trying to recruit or find participants for follow-up interviews. Observations focused on the ways in which people who were homeless presented to the CSS, their behaviour in the reception areas and how this was managed by the CSS staff, and whether they were seen by clinicians or left prematurely. It was not feasible to undertake observations in reception areas of Usual Care GPs, as it would not have been possible to identify whether patients were housed or homeless.

Observations were undertaken at day centres of the systems in place for service users to see a CSS nurse, and the ways in which nurses engaged with those who were reticent. Other opportunities to undertake observation work were sought. At one CSS, two researchers attended a multidisciplinary staff meeting. At another, a researcher accompanied a CSS staff member and outreach worker while they conducted street outreach work early in the morning. At a third site, research staff attended a drop-in service for rough sleepers run by CSS staff. Field notes were maintained of the various observations.

Data from medical records

After the interview at 8 months, printouts of all medical record data held by the CSS for the 12-month study period, including consultations, letters, reports and referrals, were requested for each participant. This included GP records that were shared with the Mobile Teams. Data were also requested dating back 2 years regarding smoking and dating back 10 years regarding hepatitis A (for scoring the HSIs). Organising data extraction and ensuring that complete data were obtained proved intricate and time-consuming. Each site identified a person responsible for data extraction and Joly (research team) instructed them accordingly. They were provided with the requirements for data extraction for each participant. The extraction process was undertaken by the CRN at one CSS, and elsewhere by CSS administrative staff. Data were then checked by Joly for completeness, and missing documents requested. At some sites, this process was repeated several times before complete data were obtained.

Medical records were obtained for 349 of the 363 participants. They were used to score the primary outcome and some SHC outcomes, and to provide service use data. Medical records were not obtained for 14 Usual Care GP participants, including both participants from one site (UC4). This was because participants were no longer registered with the CSS at the end of the study and the records were unavailable. As shown in Table 4, medical records for the entire study period (12 months) were available for most participants, with no statistically significant difference by Health Service Model.

Although the research team checked with CSS staff that a person had been registered at least 4 months prior to recruitment, once medical records were obtained, it was found that 50 participants had been registered less than 4 months prior to recruitment. For 26 of these participants, it was possible to collect additional data from the medical records at the end of the study period to provide 12 months’ medical data on service use.

Data management

All data collected during the study were stored in locked filing cabinets in our department at King’s College London (KCL), which itself is locked. Only the KCL team had access to these filing cabinets, and to the database on the university server (which was password protected). Identifiable participant information was not disclosed beyond the KCL research team. Names were not entered into spreadsheets or databases created for the analyses.

The recorded interviews of CSS staff, service providers and stakeholders were transcribed by a professional company used by KCL. A confidentiality agreement was drawn up and audio files were transferred to the company using the KCL File Transfer Protocol system (a secure method of transferring data). Staff were informed of this on the participant information sheets and consent forms. Once interviews were transcribed, all personal identifiers were removed.

Data analyses

A descriptive picture of the context and mechanisms of each CSS was built from interviews with CSS staff and other agencies. Using the templates created by each CSS manager, information was entered into an IBM SPSS Statistics version 26 (IBM Corporation, Armonk, NY, USA) database about the service, including staffing, types of services provided and accessibility of the CSS. Similarities and differences between and within models were examined.

Information from the case study participants’ interviews was entered into an SPSS database, alongside data relating to (1) primary and secondary outcomes extracted from the medical records, (2) characteristics of the CSS and (3) service use over 12 months. Summary statistics relating to both the background characteristics of participants and quantitative outcomes were produced, along with histograms to enable assessment of normality so that appropriate statistical tests could be employed.

A range of indicators was used to measure the relative effectiveness of the four Health Service Models, and each indicator was analysed separately. First, the models were compared with respect to the primary outcome variable (the six-item HSI, score 0–6) using analysis of variance (ANOVA). An in-depth comparison was then performed using appropriate regression techniques to explore associations between the HSI and demographic, background and health profiles of participants. The model type was entered as a dummy variable. The four models were then compared for each of the secondary outcomes. The prevalence of each of the SHCs was compared across the models using the chi-squared test. The analysis of each SHC then proceeded using just the subgroup having the relevant condition. Each of the five dichotomous SHC outcomes within each SHC was compared across the four models using the chi-squared test.

Each of the continuous outcome variables [physical component score (PCS) and mental component score (MCS) of the SF-8, the SWEMWBS, and service user satisfaction with the CSS] was compared between the four models at baseline, and changes from baseline to 8 months were calculated and compared by model type. Stepwise linear regression modelling was performed to adjust for other factors (such as personal characteristics, length of time using the CSS and service features of the CSS) when comparing the four models. This was carried out initially at baseline to include as many service users as possible, and at 8 months using changes when available.

Differences in outcomes between models were investigated in relation to the contexts and mechanisms of care to seek understanding of the reasons underlying the patterns observed. Quantitative information was triangulated with data from qualitative interviews with case study participants and staff about accessing health and other services. Satisfaction of the case study participants with the CSS was compared with satisfaction of the general population with their GP, using data from the GP Patient Survey.

Comparisons were made between oral health status and receipt of dental care across the four models, and their extent of integration with primary dental care services. Comparisons were also made between access to dental care and impact on self-reported oral health status and OHRQoL, dental anxiety and satisfaction. Receipt of dental care by the four models was compared with local and national populations’ access to primary dental care, using national NHS statistics.

Patient and Public Involvement

Patient and Public Involvement (PPI) was an integral part of the study from the outset and was led by two research team members. An initial PPI group of four people who had experienced homelessness worked with the research team in preparing the study proposal. Some had worked with team members on other studies. They made valuable contributions about research questions that should be addressed, and ways information should be collected. Their suggestions were incorporated into the study design. They commented on draft questionnaires and the content and phrasing of participant information sheets. They also contributed to SSC meetings, and commented on the study’s findings.

Besides an advisory role, PPI members undertook interviews with people in hostels and day centres who were homeless but not accessing the CSSs. A training day covered the role of peer interviewers, the interview process, obtaining consent, managing difficult or sensitive topics, personal safety and data security. Completion of training and a Disclosure and Barring Service check were required before a peer interviewer could commence fieldwork. The research team arranged interview dates with service providers, and accompanied PPI members to all sites. The peer interviewers explained the study to potential participants, checked eligibility and gained consent to take part. Using a structured questionnaire, they then collected the necessary information, working alone or accompanied by a research team member for support.

Feedback during and after each fieldwork session and three half-day reflection sessions allowed the interview process to be reviewed, and allowed discussions of what went well or was challenging. Peer interviewers were enthusiastic and committed. They welcomed learning about homelessness and services, spending time with the research team and contributing to the study. Early into interviewing at the first site, they requested more information for additional context about the study sites in which people were being interviewed. They were remunerated and expenses were covered.

Models of specialist primary health care services

The mapping exercise identified 123 specialist primary health care services in England for single people who are homeless. Some operated from a ‘fixed’ site, whereas some were mobile teams that ran clinics in hostels and day centres. Some were primarily for people who were homeless, whereas others delivered targeted services to this group, but also provided health care to the general population. Most involved a team of workers, although, in a few instances, a single nurse worked at a hostel or day centre. Some were reliant on volunteers and charitable funding.

The 123 specialist services were classified into six models, plus a seventh which encompassed ‘other medical/nursing arrangements’ (Table 5). Some of the services in this seventh model were run by social enterprises and commissioned by local CCGs. Nearly three-quarters of services were specialist health centres, or general practices with homeless services. The former worked primarily with people who were homeless and tended to have far fewer patients than the latter. They were also more likely to provide on-site specialist services such as drug and alcohol clinics, and housing or financial advice sessions. Specialist health centres were less likely, however, to offer extended opening hours. These centres and general practices with homeless services offered GP registration, but this was not generally the case for other models.

The 123 specialist primary health care services were spread across the then-five NHS England regions (North, Midlands and East, London, South East, and South West). There were clusters in London and in the North region around Greater Manchester and Merseyside, and a single such service was found in several coastal areas in the South West and South East regions. Relatively few services were identified in the northern part of the NHS North region, and in parts of the NHS Midlands and East region. They were mainly located in urban areas where there were concentrations of people who were homeless, and hostels and day centres for this population. Few specialist primary health care services were identified in rural areas, although a few mobile homeless health teams were based in urban areas and covered market towns.

Summary

This chapter summarises the types of specialist primary health care services identified during the mapping exercise. Their distribution varied greatly, and most were in urban areas. The majority fitted into one of three broad models, namely specialist health centres, general practices with homeless services and mobile homeless health teams. These are comparable to the models described in the 2010 DH report,13 although several other smaller services were revealed that were not mentioned in the DH report.

Regardless of the model, the specialist primary health care services shared some common characteristics. They provided easily accessible and flexible services that reached out to people who were homeless. Most offered either drop-in sessions at their service and/or ran clinics at hostels and day centres. Many provided integrated care together with other health and social care providers. The mapping exercise informed the selection of Health Service Models and CSSs for the HEARTH study, which are described in the next chapter.

Dedicated Centres

The Dedicated Centre model was represented by two CSSs [Dedicated Centre (DC) 1 and DC2]. Both had existed for many years and had developed substantially over time. They were located in large cities that had a substantial problem of homelessness, and were relatively close to several hostels and day centres for people who are homeless. Their catchment area was citywide. Both had a practice manager, GPs, practice nurses, and reception and administrative staff. Most staff had considerable experience of working with people who were homeless. DC1 had a full-time primary mental health nurse, whereas the community mental health team worked part time at DC2. Substance misuse workers held sessions most days at the sites, but were not employed by the CSS. Both sites had additional specialist workers, but these cannot be identified to maintain anonymity.

Both sites were part of an NHS trust and operated in similar ways. DC1 held a contract as an APMS; DC2 also held a special contract. They were open Monday to Friday, and were covered by a GP out-of-hours service. Both offered permanent GP registration. In contrast to many mainstream general practices, the caseloads of the CSSs were relatively small.

DC2 mainly focused on single people who were homeless, whereas DC1 also provided care to asylum seekers and refugees. Neither site provided registration to the general population, and once patients were rehoused and settled, they were assisted to register with mainstream general practices services. The sites offered booked appointments at the practice and a same-day drop-in service. The duration of appointments tended to be longer than at mainstream general practices, for example DC2 allocated 20-minute appointments, instead of the customary 10 minutes.

Dedicated Centres provided GMS, including management of acute and chronic illnesses, smoking cessation, immunisations and vaccines, treatment for infestations and referral to secondary health care services. They also provided mental health care and counselling, and access to community alcohol detoxification programmes and drug treatment, including opioid substitution treatment (OST). CSS staff conducted street outreach at least monthly to engage with rough sleepers and encourage them to access services. Staff at DC1 also visited drop-in centres regularly to encourage registration. At both sites, nurses or health support workers visited day centres and hostels at intervals to deliver influenza vaccination programmes and health promotion. Staff accompanied patients to hospital appointments if necessary, and helped them access housing and welfare benefits services. Daily staff meetings were held at both sites to discuss patients, and regular staff training took place.

Mobile Teams

The Mobile Team model was represented by two CSSs [Mobile Team (MT) 1 and MT2], which had been in operation for several years. They were based in urban areas with a substantial number of people who were homeless, and several hostels and day centres for this population. Both teams were funded by the NHS to run clinics and deliver health care in these settings. They worked Monday to Friday.

The Mobile Teams mainly comprised specialist nurse practitioners with considerable experience of working with people who were homeless, and some were non-medical prescribers. They received some administrative support, and input from mental health practitioners. They worked closely with local general practices, and some joint work in hostels was undertaken by MT1. They provided health care to people who were homeless, to asylum seekers and refugees, and to people with no recourse to public funds. MT1 also offered health care to people who were housed, but who attended homelessness sector day centres for support.

Neither team provided GP registration, and patients were encouraged by the CSS nurses to register with local general practices. This meant that patients could receive health care from both a GP and the Mobile Team. Those registered with a GP received an out-of-hours service through the general practice. Caseloads were reviewed periodically by the Mobile Teams, and cases were closed once care was complete or if contact had been lost. Their caseloads tended to be small (see Table 6).

Services provided by the Mobile Teams differed from those provided by other models. Much of their work concerned assessing the health needs of patients, and linking them to local services, including general practices and drug or alcohol agencies. Occasionally, nurses accompanied a patient to appointments if there was an urgent need, or if the person was not engaging with services. They were less involved in acute disease management other than ensuring that patients were seen by a GP or at a hospital if necessary. They undertook health promotion, longer-term care for conditions such as leg ulcers, and routine blood tests.

The Mobile Teams held drop-in nursing clinics in hostels or day centres on set days and at fixed times. The frequency of clinics ranged from 1 to 5 days a week, and each lasted 2–4 hours. The offices where the Mobile Teams were based did not have facilities for patient care. The spaces available to see patients in hostels and day centres ranged from well-equipped clinical rooms to rooms used by other services, or simply a screened area in a drop-in centre or hostel sitting room. Access to a consultation with a nurse was primarily through a list held by hostel or day centre staff, or by the nurse actively seeking out individuals during a clinic. The Mobile Teams conducted frequent street outreach to engage with people sleeping rough.

Specialist GPs

The Specialist GP model was represented by two CSSs [Specialist GP (SP) 1 and SP2]. Both were in urban areas with high rates of homelessness and had been operating for several years. They were mainstream general practices delivering health care to the housed population, as well as additional or enhanced services to people who were homeless. Hence, the total number of patients registered at these practices was considerably higher than that of the Dedicated Centres and Mobile Teams, but the number of patients who were homeless was comparable. One site had more than 10,000 registered patients, of whom an estimated 850 were homeless. The other had between 5000 and 10,000 patients, of whom around 300 were homeless.

Unlike other models, there were noticeable differences between the two CSSs in the Specialist GP model. Both had GPs; practice nurses; HCAs; and reception, administrative and information technology (IT) staff. Two GPs at each site had a special interest in homelessness, although they worked with all patients. At SP1, some practice nurses worked primarily with patients who were homeless, and non-clinical team members provided case management to those with complex health and social care needs. At SP2, nursing care was delivered by practice nurses who worked with both patients who were housed and those who were homeless.

Both CSSs were funded by the NHS, and held additional contracts for their work with patients who were homeless. They were open Monday to Friday, and were covered by a GP out-of-hours service. Besides booked appointments, SP1 offered drop-in sessions for people who were homeless three times a week. At SP2, drop-in sessions were provided daily for all patients. The CSSs provided general primary health care, management of acute and chronic conditions, wound care, and immunisations and vaccines. Both had input from sessional mental health workers. At SP1, drug and alcohol workers also held clinics on a sessional basis, and provided community alcohol detoxification and drug treatment, including OST and needle exchange. At SP2, drug and alcohol treatment was not available on the premises; patients were referred to local substance misuse services.

Both sites held clinics in hostels and day centres for people who were homeless, although the frequency of these varied. SP1 held several nursing clinics each week at these services, and visited other sites to engage with service users who were not accessing health care. Staff also undertook street outreach with other agencies. At SP2, clinics were held twice weekly by GPs at homelessness services (see Table 6).

Usual Care GPs

The Usual Care GP model was represented by four CSSs [Usual Care GP (UC) 1, UC2, UC3 and UC4]. They were mainstream general practices delivering primary health care to the general population, which, by default, included people who were homeless. Unlike Specialist GPs, they did not provide special services or have dedicated staff for patients who were homeless. Three sites had caseloads of more than 15,000 patients, the fourth had slightly fewer. The proportion of patients who were homeless ranged from 0.04% to 5.4%. All had a PMS or APMS contract, and all except UC1 received additional funding to work with patients who were homeless (UC1 received funding for work with patients with substance misuse problems).

All were open Monday to Friday, and three also opened on Saturdays (see Table 6). When closed, a GP out-of-hours service provided cover. All provided permanent GP registration, and pre-booked and same-day appointments. They employed a range of staff associated with the running of a general practice, including GPs; practice nurses; HCAs; and reception, administrative and IT staff. Unlike the other models, they either employed a non-dispensing pharmacist or had a pharmacist providing sessional services. The pharmacists dealt with prescription queries, undertook medication reviews with patients and reviewed prescribing practices. The use of clinical pharmacists in general practice was introduced as a pilot scheme in 2015, and became part of the funding framework for general practices in 2019. Hence, they were not in operation when Dedicated Centres and Specialist GPs were recruited.

All four CSSs provided general primary health care, including management of acute and chronic diseases, smoking cessation services and routine blood tests. Sessional workers provided specialist services at the CSSs, although this varied between sites. Drug and alcohol services, including OST, were provided at UC1 and UC3 by GPs through shared care with local drug services. At UC2, a similar long-standing arrangement had ended because of funding cuts, and the drug service took over all care. UC1 and UC4 employed a mental health nurse; UC3 had input from mental health practitioners on a sessional basis. Only UC2 offered no on-site specialist mental health care.

All except UC1 held registers of patients who were homeless. UC2 and UC3 used a computer-based ‘homeless template’, which prompted clinicians to ask about specific issues when a patient attended an appointment, such as alcohol consumption. No sites held clinics in hostels or day centres for people who were homeless, but UC3 visited local hostels annually to encourage residents to attend a homeless health check. UC2 was close to a hostel and in regular communication with hostel staff.

Summary

This chapter summarises the various ways in which the four Health Service Models delivered health care to people who were homeless. There were marked differences in the sizes of the practices, the ways in which they operated, the services provided and their patient populations. These differences will be considered when evaluating the effectiveness of the models in delivering health care to people who are homeless. The next chapter describes the patients of these models who agreed to be case study participants.

Sex, age and ethnicity

Four-fifths (80.2%) of participants were male and 19.8% were female. A slightly higher proportion (90.7%) of Usual Care GP participants were male; some were recruited at hostels specifically for men. The average age of participants was 41.6 years, with little difference by Health Service Model. Just 13.2% were aged 18–29 years, and only 5.2% were aged 60 years and over.

Almost three-quarters (73.4%) were White British or Irish; the remainder were mainly black or Black British, or other white background. Of the 74 who were born abroad, most had been living in the UK for years: 51% for more than 10 years, and just 10.8% for less than 5 years. Nineteen participants born outside the UK had gained British citizenship; 32 were European migrants, and eight were asylum seekers. There were statistically significant differences by Health Service Model (Table 7). Mobile Teams served a markedly lower percentage of UK-born clients, which partly reflects the CSS locations.

Employment status, income and social support

At baseline, 14.6% of participants were in education, training or employment, with no difference by Health Service Model. This included 2.5% employed full or part time. Most were in receipt of an income, nearly all through state benefits (see Table 7). Seven-tenths (70.7%) received Employment and Support Allowance, for people below state pension age with a health condition or disability affecting their ability to work. One-fifth (20.2%) received Personal Independence Payment, payable to people with a health condition or disability for 3 months, who have difficulties with daily living activities or mobility, and expect these difficulties to continue for at least 9 months. A small percentage (5.8%) were seeking work and receiving Jobseeker’s Allowance.

Thirteen per cent had no income when first interviewed, including a significantly higher percentage (29.2%) of Mobile Team participants (see Table 7). Many without an income were European migrants or asylum seekers. Overall, 6.4% had no recourse to public funds, including 15.2% of Mobile Team participants. For others, state benefits had stopped because they did not comply with regulations associated with receipt of benefits, or because their circumstances had changed which had led to administration problems. Participants without an income were mainly sleeping rough or staying in night shelters that do not require rent.

At baseline, 48.9% of participants had received support from family or friends during the last 4 months, with no difference by Health Service Model. This support was mainly provided by a friend, mother or partner.

History of homelessness and accommodation at baseline

It is difficult to collect accurate details of the length of time people have been homeless, as ‘homelessness’ is often not a continuous state. Many people who are homeless frequently move between sleeping rough, hostels, night shelters, bed-and-breakfast hotels, sofa surfing and other insecure housing. Some have experienced repeated episodes of homelessness interspersed with stays in prison. At baseline, the median length of time participants had been homeless continuously or intermittently was 5.5 years, but this ranged from less than 1 year (9.2%) to 25 or more years (5.8%). There were no significant differences by Health Service Model (see Table 7).

At baseline, 52.1% were staying in a hostel or supported accommodation with 24-hour or part-time staff, and 21.2% were sleeping rough on the streets, in tents or vehicles, or in buildings not designed for habitation, such as squats and night shelters. The night shelters were mainly church halls, which opened only at night and required people to leave during the day. Among the remaining participants, 7.2% were sofa surfing, 11.8% were in bed-and-breakfast hotels or other temporary accommodation and 7.7% had a tenancy. There were significant differences by Health Service Model (see Table 7). Participants of Specialist GPs and Usual Care GPs were more likely to be in hostels or supported accommodation with staff. Those using Dedicated Centres and Mobile Teams were more likely to be sleeping rough. More than two-fifths (43.3%) had been in their last accommodation 3 months or less, including 22.9% for 1 month or less. Usual Care GP participants had been in their accommodation the longest; the difference was statistically significant (see Table 7).

Physical and mental health problems

At baseline, 94.8% of participants reported physical health problems. This high proportion is to be expected as the inclusion criteria meant that participants needed to have been a patient of the CSS for at least 4 months, and had at least one consultation with a CSS doctor or nurse during this time. Common problems included respiratory disease, gastric or intestinal problems, and arthritis or other musculoskeletal problems, and just over one-fifth (21.9%) had hepatitis C (see Appendix 1, Table 46). There were strong associations between alcohol misuse and problems such as gastritis, liver problems, blackouts and fits. Likewise, there were strong associations between injecting illegal drugs and hepatitis C, thrombosis and leg ulcers.

Nine-tenths of participants reported mental health problems at baseline, most commonly depression, anxiety and panic attacks (Table 8). Personality disorder or behaviour problems, such as anger management problems, were mostly reported by those aged 18–29 years (47.9%, compared with 21.1% overall). Usual Care GP participants were more likely than other participants to report bipolar disorder and schizophrenia or hearing voices. This may be related to sampling: some Usual Care GP participants were recruited at hostels exclusively for people with mental health problems.

Alcohol use

Alcohol consumption among participants varied greatly. When asked at baseline about their frequency of drinking alcohol over the preceding 4 months, 35.1% reported no alcohol, whereas 28.5% had consumed alcohol almost daily. Some drank heavily for a few days when they received their state benefits, followed by days of abstinence when they had no money, or had episodes of binge drinking lasting several days or weeks. A small percentage (5.2%) had no alcohol in the previous 4 months, but were recovering from an alcohol problem. Slightly more men than women (30% vs. 22.2%) reported drinking alcohol almost daily.

One-third of participants drank super-strength lagers or strong cider, including some that had 9% alcohol by volume, and 17.1% drank spirits. Using participant information about the type and amount of alcohol consumed on a typical day when drinking, the weekly number of units of alcohol consumed was estimated. Twenty-one per cent were drinking more than 100 units weekly. The National Institute for Health and Care Excellence (NICE)113 describe (1) ‘lower risk drinking’ as up to 14 units weekly for men and women; (2) ‘hazardous drinking (increasing risk)’ as 15–34 units weekly for women and 15–49 units weekly for men; and (3) ‘harmful drinking (high risk drinking)’ as 35 or more units weekly for women and 50 or more units weekly for men. By these guidelines, 11.3% of the HEARTH study participants had hazardous drinking patterns and 31.9% were harmful drinkers. Participants of Usual Care GPs were more likely than other groups to be hazardous or harmful drinkers (54.2%), although the findings were not statistically significant (Table 9).

Drug misuse

At baseline, 60.1% of participants (62.5% of men and 48.6% of women) reported drug misuse in the preceding 4 months, that is the use of illegal drugs or misuse of prescription or over-the-counter drugs. A further 7.2% were recovering from a drug problem, but had been abstinent in the previous 4 months. One-quarter (25.6%) had taken drugs most days over the preceding 4 months. The most used drugs were cocaine or crack cocaine (36.7%), heroin (32.1%), cannabis (29.9%) and spice (8.3%); 16.1% had used at least three different types of drugs, and 40.4% had used one or more Class A drugs [e.g. heroin, morphine, cocaine, lysergide (LSD) or ecstasy] (see Table 9). Just over one-third were receiving OST, mainly methadone. A further 4.7% commenced OST as the study progressed.

Mobile Team participants were less likely than the other three cohorts to report drug usage (see Table 9). Usual Care GP participants were more likely to have used cannabis only, whereas higher percentages of Specialist GP and Dedicated Centre participants reported using Class A drugs, injecting drugs and receiving OST.

Registration at Case Study Sites

The date of participants’ registration with the CSS, or the date they started using the service (for Mobile Team participants), was obtained from the medical records. This included the date of the most recent registration or use of service for those who had been registered or a service user previously. Overall, 54% had been registered or using the CSS for more than 1 year at baseline, and 26.4% had been a patient for less than 6 months. There were no statistically significant differences between models (Table 10).

Accommodation during study

The potential study period for each participant was 12 months, that is the 4 months preceding their baseline interview, and the subsequent 8 months. Information about their accommodation during this time was available for most (90.6%), with no statistically significant differences by Health Service Model (Table 11).

Information about accommodation during the study period was assembled for each participant to determine (1) the number of changes of sleeping settings, for example move from the streets to a hostel, or from a hostel to a bed-and-breakfast hotel or another hostel; (2) the percentage of time living in a building designed for habitation, such as a hostel, as opposed to the streets or a squat or church hall not designed for habitation; and (3) the percentage of time living in accommodation with 24-hour or part-time staff. Change of accommodation included episodes of 7 or more days in hospital, prison or a rehabilitation unit.

Just over one-quarter of participants (26.2%) remained in the same accommodation throughout the study, including 46.7% of Usual Care GP participants. In contrast, 26.2% moved four or more times, including 3.9% who changed accommodation at least eight times. There were strong significant differences between accommodation experiences during the study and Health Service Models (see Table 11). Usual Care GP participants spent a higher percentage of time in staffed accommodation, whereas the Dedicated Centre cohort had the highest average number of moves, and spent considerably less time in staffed accommodation. They were also more likely than other groups to have been in prison during the study (18.8%, compared with 11.4% overall; χ² = 8.36; p = 0.039). Mobile Team participants spent longer than other groups sleeping on the streets or in makeshift arrangements, such as church halls.

Differences between Case Study Sites within a Health Service Model

There were some marked variations in participants’ characteristics by CSSs within the same Health Service Model (Table 12). MT2 participants were much less likely than those of MT1 to be British born

or have British citizenship, and were much more likely at baseline to be sleeping rough or in settings not designed for habitation. Such marked differences are likely to be related to MT2 participants’ lack of entitlement to welfare benefits, and thus inability to access hostels. At baseline, 24.4% of MT2 participants, compared with just 6.4% at MT1, had no recourse to public funds. There was little difference in overall reports of mental health problems, apart from across the Mobile Teams: such problems were reported by 95.6% of MT1 participants, compared with 73.5% from MT2. UC3 participants were more likely than the other two Usual Care GP sites to report having schizophrenia or hearing voices.

There was little difference in reports of alcohol drinking among CSSs within the same model, although ‘harmful’ drinking was more prevalent among MT1 than MT2 participants. Drug use varied greatly within models. Exceptionally high prevalence rates of heroin or cocaine usage were reported at DC1, MT1 and UC1, compared with other sites within these models (see Table 12). During the study, UC1 participants were much less likely than the other two Usual Care GP site participants to have spent time living in staffed accommodation. There were no marked differences between CSSs within a Health Service Model regarding number of years homeless, number of moves during the study or percentage of time in a building designed for habitation.

Summary

This chapter summarises the characteristics of case study participants. Self-reports of physical health, mental health and substance misuse problems were prevalent. This partly reflects the study sample, which was drawn from users of health services, although it is well documented that health and substance misuse problems are common among people who are homeless. There were several differences in participants’ characteristics between the four Health Service Models, and between CSSs within the same Health Service Model. This amplifies the importance of performing multivariate analyses when assessing the performance of each Health Service Model.

The chapter also demonstrates the unsettled histories and makeshift living arrangements of many participants, which are likely to affect their health. One-third had been homeless 10 years or more; although some were living in hostels or temporary accommodation, others were sleeping on the streets. When interviewed at baseline, many had been in their latest living arrangements for just a few weeks or months, and had been accessing the CSS for just a short while. The unsettledness and mobility of some participants, and changing of health service providers, are likely to have implications for their engagement in health care.

Health Screening Indicators and the scoring process

Six HSIs were selected for the primary outcome: BMI, mental health, alcohol use, TB, smoking and hepatitis A (see Chapter 3). The first four required evidence of screening at some point during the 12-month study period; smoking required screening in the previous 24 months, and hepatitis A required a vaccination programme in progress or completed in the previous 10 years. Each was given a score of 1 (positive) or 0 (negative), depending on evidence in the medical records of screening having taken place and an intervention offered if a problem was identified (Table 13). The total score ranged from 0 to 6.

A Microsoft Excel file was created for each CSS, with separate worksheets for each HSI. Searches of the medical records for each HSI were conducted, and relevant information documented, including dates of observations. Checks were carried out by a second coder. Working papers were circulated among the research team as queries arose, and subsequently discussed at SSC meetings. Uncertainties or discrepancies regarding individual cases were reviewed by Ford (GP retired).

Occasionally, information pertaining to the HSIs was found in documents such as A&E reports or outpatient letters, rather than a CSS consultation. Mobile Teams could see general practice records for most of their participants, as they shared records. It was agreed with the SSC that a systems-wide approach should be adopted and scoring would include screening by other health professionals if this could be seen by CSS staff.

Questions arose as to what constituted ‘an intervention’. An intervention generally included, for example, advice about weight or smoking, medication for depression or referral to mental health services. In some instances, however, CSS staff ‘signposted’ a patient to a service without a formal referral. For example, following screening for mental health problems, one participant was encouraged to attend a mental health drop-in centre, and another was handed a leaflet about a counselling service. Likewise, a participant who reported an increase in alcohol consumption was given details of a harm reduction service. Hence, the onus was on the patient to contact the service. As some services encourage self-referrals, it was discussed with the SSC and agreed that signposting should be scored as an intervention. Likewise, interventions offered by CSS staff, but declined or ignored by participants, were scored positively, as the CSS had tried to intervene.

In many instances, formal assessment tools for screening mental health problems were not used. Instead, records indicated mental health status had been observed, albeit sometimes with limited descriptions, for example ‘mood low’ or ‘mood good’. Pre-coded text was used several times by MT1, stating ‘mental health fully performed’ without elaboration. It was agreed with the SSC that interpretation of mental health screening should include situations in which it is assumed that questions must have been asked about a person’s mental state. A similar approach was adopted for screening alcohol use.

Some medical records stated that a person was ‘hepatitis A immune’, meaning that they had had hepatitis A and developed antibodies. Ford confirmed that, in this situation, vaccination is unnecessary, and so this should be scored positively. Another query concerned the maximum interval between the first and second injections for a vaccination programme to be considered ‘in progress’. Ford checked with a specialist, who clarified that one dose of the hepatitis A vaccine could be effective for up to 10 years. This cut-off point was used in the scoring.

Primary Outcome Score and screening by Health Service Model

Figure 1 summarises the number of HSIs per participant that were screened, and an intervention offered if applicable. Sixty per cent of participants received screening for three or four HSIs.

FIGURE 1.

Health screening for primary outcome: total score.

Overall, the mean Primary Outcome Score was 3.30, although there were statistically significant differences by Health Service Model (Table 14 and Figure 2). Participants of Dedicated Centres and Usual Care GPs had been screened for the highest number of HSIs, with little difference between the two models (means of 3.54 and 3.51, respectively). Participants of Mobile Teams had been screened for the fewest (mean 2.90). The highest median score was achieved by Usual Care GPs.

FIGURE 2.

Primary Outcome Score (mean with 95% CI) by Health Service Model. CI, confidence interval.

Around 80% of participants had been screened for smoking status, mental health and alcohol use; 45.8% had been screened for BMI, 33.8% for hepatitis A and just 8.6% for TB (Table 15). There were statistically significant differences between Health Service Models. Usual Care GPs were more likely than the other three models to have recorded BMI and smoking status, but were least

likely to have vaccinated against hepatitis A. Staff of UC1 and UC2 explained that they did not routinely screen for hepatitis A unless a patient presented with symptoms or it was required for travelling abroad. Mobile Teams were less likely than other models to have assessed mental health and alcohol use. Screening and vaccinations for hepatitis A were more common among participants of Dedicated Centres and Specialist GPs. They tended to have higher numbers of injecting drug users. Even so, only 41.8% of participants who injected drugs had been screened for hepatitis A.

Regression modelling of Primary Outcome Score

To isolate the effect of the four Health Service Models on the Primary Outcome Score, an initial backward stepwise linear regression was run with Primary Outcome Score as the dependent variable and the variables listed as predictors (Tables 16 and 17). Variables were drawn from participants’ reports, medical records and service use data. Statistically significant and near statistically significant (p < 0.15) predictors were used in a second backward stepwise linear regression.

A final model was run using only the statistically significant predictors resulting from this regression, to maximise the number of observations used. The fitted parameters for this final model are shown in Table 18. No statistically significant difference in effect was seen between Dedicated Centres, Specialist GPs and Usual Care GPs. Only Mobile Teams differed, with a highly statistically significant negative effect identified [–0.603, 95% confidence interval (CI) –1.015 to –0.191]. Other statistically significant effects in the model were reporting depression at baseline (eliciting an improvement of 0.415 in the Primary Outcome Score), drug use at baseline (eliciting an improvement of 0.295 in the Primary Outcome Score), being British born or a British citizen (eliciting a decrease of 0.852 in the Primary Outcome Score), having seen a non-accommodation-based key worker (eliciting a decrease of 0.012 in the Primary Outcome Score), the percentage of the study period living in accommodation with full- or part-time staff (each additional 10% elicited an improvement of 0.055 in the Primary Outcome Score), the number of consultations at a CSS with a GP, a nurse or a HCA (each additional 10 contacts elicited an improvement of 0.099 in the Primary Outcome Score) and number of years homeless (each additional 10 years elicited an improvement of 0.281 in the Primary Outcome Score).

The effect on the Primary Outcome Score of the number of consultations the person had with a GP, a nurse or a HCA at the CSS is likely to be caused by increased opportunity for staff to undertake health screening the more a patient is seen. Likewise, participants who reported depression or drug use at baseline had more consultations during the study with GPs or nurses at the CSS. The positive association between the percentage of the study period in accommodation with full- or part-time staff and the Primary Outcome Score may be due to staff encouraging residents to access health care, or may be an indicator of the stability of participants and their willingness to engage with housing and health services. The negative association between contact with an external key worker and the Primary Outcome Score may be an indicator of the unsettledness of some participants. Many key workers were street outreach and day centre workers supporting participants who were sleeping on the streets. The effect on the Primary Outcome Score of the number of years a person was homeless may be due to the person being more accustomed to, and accepting of, services over time.

Primary Outcome Score and screening by Case Study Site

There were marked variations in screening when the CSSs within some models were examined (Table 19 and Figure 3). There was little difference in the Primary Outcome Score between the two Dedicated Centres, but a noticeable difference between the two Specialist GP sites. SP1 had the second highest score (mean 4.16), whereas SP2 had the lowest score (mean 2.42). Part of the reason may be that the number of GP/nurse/HCA consultations at the CSS during the study varied greatly between the two sites: an average of 12.76 for SP1 and 7.13 for SP2. Another reason may be that SP1 had dedicated nurses for patients who were homeless, and case managers for those with complex health and social care needs. This was not the situation at SP2 (see Chapter 5). It is likely that nurses would have undertaken elements of the screening, and case managers would have encouraged attendance at health appointments.

FIGURE 3.

Primary Outcome Score by CSS.

There were also substantial differences in scores between Usual Care GP sites, ranging from a mean of 2.65 for UC1 to 4.20 for UC3 (the highest of all CSSs). BMI was documented in the medical records for 93.3% of UC3 participants, compared with just 29.4% of UC1 participants (see Appendix 2, Table 47). Once again, the number of consultations during the study period with a CSS GP/nurse/HCA varied greatly between sites: an average of 5.96 for UC1, 8.33 for UC2 and 8.53 for UC3. In addition, UC2 and UC3 used a computer-based ‘homeless template’ to assess the needs of patients who were homeless (see Chapter 5). UC3 also proactively visited local hostels annually to encourage residents to attend a homeless health check, and UC2 was near a hostel and in regular communication with hostel staff.

There was slight variation between Mobile Teams, with MT2 scoring higher than MT1. There was little difference between screening of BMI, mental health and alcohol use, but MT2 participants were more likely than those of MT1 to have been screened for TB, smoking status and hepatitis A (see Appendix 2, Table 47). This reflects different working practices. MT2 proactively screened for TB and hepatitis A, whereas MT1 explained that it did not offer this service and relied on drug agencies and other health services to undertake such screening.

Summary

This chapter describes the extent to which screening for the six HSIs comprising the primary outcome had been undertaken with case study participants. Screening for smoking status, mental health and alcohol use was most common, although around one-in-five had not been screened for these conditions. Just over half had no BMI recorded during the study period, yet the diets of many people who are homeless are poor. 88 Even fewer participants were screened for hepatitis A and TB, yet reports indicate high prevalence rates of these diseases among people who are homeless, and among groups at risk of homelessness, namely people who inject drugs or have been in prison. 115,116 Since 2000, outbreaks of hepatitis A have been reported in several parts of England among people who inject drugs, or who are homeless and living in hostels and shelters. 89,90,117 Likewise, people who are homeless have higher rates of active pulmonary TB than the general population, and often present late for health care or are lost to follow-up care. 91,118

There were differences in the Primary Outcome Score and in screening of individual HSIs by Health Service Model and by CSSs within models. Regression analyses revealed a highly statistically significant negative effect for the Mobile Team, but no difference for the other three models. Given that most Mobile Team participants were also registered with a general practice, it appears that neither the Mobile Teams nor the general practices regularly undertook screening for the HSIs.

The chapter also found that screening differed across CSSs within the same model, indicating the importance of considering working practices of individual CSSs. Among the Usual Care GPs, UC2 and UC3 used ‘homeless templates’ to assess needs and had a Primary Outcome Score comparable to Dedicated Centres. Likewise, SP1 had designated nurses and case managers to work with patients who were homeless and scored much higher than SP2. Some CSSs routinely screened HSIs, but others did not. Staff of DC2 and SP2 said they did not routinely screen for TB unless it was indicated, and MT1, UC1 and UC2 said the same for hepatitis A. This is discussed further in Chapter 15, which outlines recommendations from the Faculty for Homeless and Inclusion Health and from NICE. 70,119 The next chapter examines the effectiveness of the Health Service Models in the management of SHCs.

When interpreting the extent of screening at the CSSs, consideration must be given to various ‘rules’ adopted by the research team while scoring. In many instances, formal assessment tools for screening mental health and alcohol consumption were not used; instead, positive scores were given when text in the medical records indicated that questions had been asked about these conditions. For example, some participants were seen frequently by CSS doctors and nurses and asked about their mood or drinking habits. These were, however, less evident for MT1 when pre-coded text was used. ‘Signposting’ a person to another service was scored positively, as were cases when interventions such as help for smoking were offered by CSS staff but declined by participants. Our method of scoring may have inflated the Primary Outcome Score in some instances. However, it was similarly applied across all Health Service Models.

Criteria for determining eligibility as a Specific Health Condition

Eligibility criteria were created for each SHC (Table 20). Only those both reported by participants when interviewed at baseline and identified by the CSS at this time (through documentation in medical records) were included. SHCs not reported by participants or reported for the first time at 4 or 8 months were excluded. Such participants would not have completed a rating scale at baseline (outcome 5), and there would have been insufficient time to assess ‘continuity of care’ by the CSS (outcome 2). Likewise, SHCs first documented in the medical records after the first 4 months of the study period were excluded.

Many SHCs were listed in the medical records as a ‘current active problem’ or documented in a GP consultation. In some instances, however, participants described a possible SHC, but it was not explicitly mentioned in the medical records, although medication associated with the condition was prescribed. In such cases, information from medical records and participants was reviewed by Ford (retired GP), and included if there was sufficient evidence that the treatment provided was for the SHC.

In a few instances, asthma or depression were listed in the medical records as a ‘current active problem’, but diagnosed many years ago. There was no mention of monitoring or treatment during the study, and participants said they had not required treatment for several years. Such cases were reviewed by Ford and, if the diagnosis was more than 3 years ago, and the participant had not required medical intervention in the interim, the problem was regarded as ‘inactive’ and not a SHC.

Outcome measures

Five outcomes monitored the effectiveness of the CSS in managing each SHC. Outcomes 1 and 2 assessed processes of care, outcomes 3 and 4 involved patient perceptions of the quality of care, and outcome 5 assessed control or change in health status over the study period.

Scoring of Specific Health Conditions

To score the SHCs, a Microsoft Excel file was created for each CSS, with separate entries for each SHC. Information from participants about help received was recorded in the data file. Details were extracted from medical records regarding (1) diagnosis or mention of the SHC and (2) action taken concerning outcomes 1 and 2. Joly and Crane (former registered nurses) undertook this work and checked each other’s entries, with Ford reviewing uncertainties. All decisions were documented.

In total, 905 potential SHCs were reported by participants; medical records were available and were searched for 864 of these. Of these 864, 629 SHCs (72.8%) were reported during the first 4 months of the study and met the inclusion criteria, 173 were not mentioned, 38 were classified as inactive problems and 24 were first mentioned after 4 months. The following sections describe outcomes for each SHC. Each table summarises the screening process, then the severity of the condition at baseline for participants with a SHC, and outcome scores. For some outcomes, response rates are lower than the number of participants with the problem; this was because either interviews could not be conducted at 8 months or participants did not access the CSS for the entire study period.

Hypertension

At baseline, 51 participants (14%) reported hypertension, although several were unclear whether or not their symptoms were linked to anxiety and panic attacks. Hypertension was documented in the medical records for just 20 (5.5%): four from Dedicated Centres, 11 from Mobile Teams and five from Usual Care GPs. Owing to insufficient numbers, it was unfeasible to examine outcomes.

Chronic respiratory problems

At baseline, 119 participants (32.8%) reported chronic respiratory problems, mainly asthma. Medical records were obtained for 111 participants, and chronic respiratory problems were documented during the first 4 months for 74 participants (Table 23). Eleven mentioned asthma, but had had no treatment or monitoring for several years, and the problem was regarded as inactive. Of the 36 participants for whom chronic respiratory problems were not documented in their medical records or it was an ‘inactive’ problem, 24 (66.7%) had scores indicating ‘high’ or ‘very high’ impact when completing the baseline COPD Assessment Test.

Of the 74 participants for whom chronic respiratory problems were a SHC, 74.3% had symptoms at baseline suggestive of high or very high impact. A treatment plan had been started for all except one of the 74 participants. Continuity of care was achieved for 70.8% of participants, ranging from 87.5% at Dedicated Centres to 46.7% for Mobile Team participants. However, continuity of care for the latter increased to 60% when GP interventions were included. In terms of outcomes 3 and 4 (explanation about the problem and views of treatment), positive responses were given by most participants, with slightly lower rates for Specialist GP and Usual Care GP participants regarding outcome 3. Nearly two-fifths of participants reported symptoms indicating a low impact on their health at 8 months or a positive change between baseline and 8 months (outcome 5), with only slight variation between Health Service Models and no statistically significant differences (see Table 23).

Depression

At baseline, 293 participants (80.7%) reported depression, with 173 meeting the eligibility criteria (Table 24). Depression was most likely to have been identified as a problem by Dedicated Centres, followed by Specialist GPs. Among participants who reported depression, it was not documented in the medical records for 30.6%, and it was classified as ‘inactive’ for 5.3%. Just over two-fifths of Mobile Team and Usual Care GP participants were in these categories. Of the 101 participants for whom depression was not documented in medical records or it was ‘inactive’, 48.5% had PHQ-9 scores at baseline indicating ‘moderately severe’ or ‘severe’ depression, and 22.8% had scores indicating ‘moderate’ depression. Just eight of this group had scores suggesting minimal or no depression.

Among the 173 participants for whom depression met the criteria as a SHC, 44.5% had PHQ-9 scores at baseline indicating severe depression, and 24.3% had PHQ-9 scores indicating moderately severe depression (see Table 24). Severe depression was most common among Usual Care GP participants, followed by those of Specialist GPs. Part of the reason may be that the former included participants living in hostels exclusively for people with mental health problems.

A treatment plan was initiated for most participants identified by the CSSs as having depression (outcome 1). Outcome 2 (continuity of care) included regular reviews and monitoring by CSS staff of mood; prescribing antidepressants; consultations and reviews by mental health teams; and discussions of the relationship between depression and substance misuse. In some instances, medication for depression could not be prescribed because of heavy drinking. Six participants were admitted to a mental health unit during the study. There were highly significant differences between models for outcome 2: continuity of care was most likely to have been achieved by Dedicated Centres and Specialist GPs (70% and 64.6%, respectively), and least likely to have been achieved by Mobile Teams (34.3%). The result for Mobile Teams increased to 68.6% when GP services were included. Exceptionally high scores were recorded for outcomes 3 and 4 for Mobile Teams, suggesting good rapport with participants and an ability to explain depression.

For outcome 5, 42.1% of participants who completed the PHQ-9 at baseline and at 8 months reported minimal depression at 8 months or a positive change over time. Although there were no statistically significant differences in outcome 5 across the Health Service Models, Usual Care GP participants were more likely, and Mobile Team participants least likely, to report minimal severity level or a positive change. This is likely to be due to several factors. As mentioned in the previous paragraph, Usual Care GP participants included people living in hostels exclusively for people with mental health problems, with staff on site, and many remained in the same accommodation throughout the study (see Table 11). In contrast, Mobile Team participants were less stably housed and changed accommodation several times during the study and spent more time sleeping rough. As mentioned previously, continuity of care for depression was also less likely to have been achieved with Mobile Team participants.

Alcohol problems

Just over half (55.4%) of participants reported an alcohol problem at baseline. This was confirmed for most (85.1%), with 165 meeting the inclusion criteria. Of the 150 participants who could specify the amount of alcohol consumed, 63.3% reported ‘harmful’ drinking; 19.3% reported ‘hazardous’ drinking; and 17.4% had reduced or stopped drinking but were still being treated or monitored by health or substance misuse services (Table 25).

A treatment plan (outcome 1) was initiated for 96.4% of participants identified by the CSSs as having an alcohol problem. There were, however, statistically significant differences for continuity of care (outcome 2) by Health Service Model (see Table 25). It was achieved for 72.7% of participants from Dedicated Centres, compared with only 29.7% from Mobile Teams. Continuity of care for the latter increased to 48.6% when interventions by GPs were added, bringing it to a level similar to that of Usual Care GPs. Interventions during the study included reviews of alcohol intake by CSS staff; case management by substance misuse or dual diagnosis teams (many also had mental health problems); alcohol reduction programmes; and prescribing of thiamine, vitamin B compound and nutritional supplements. During the study, 18 participants were admitted to an alcohol detoxification unit or undertook detoxification while in hospital, prison or a community setting; just three were still abstinent at the end of the study. Hence, for most, detoxification had been a temporary, rather than a long-lasting, intervention. Some participants declined referral to an alcohol service or did not feel ready for a detoxification programme; the CSS continued to monitor their care.

Overall, 77.1% of participants believed that they had received sufficient explanation of their condition (outcome 3), with a slightly lower rate for Specialist GPs. A higher percentage (86.3%) was satisfied with their treatment (outcome 4). For outcome 5, 40.7% reported abstinence over the study period or had reduced their alcohol consumption by 8 months. This was slightly higher among Specialist GP participants, and lower among Mobile Team participants, although the findings were not statistically significant.

Drug problems

Two-thirds of participants (66.4%) reported a drug problem at baseline. This included the use of illegal drugs or misuse of prescription or over-the-counter drugs. Self-reports were confirmed by medical records for most (86.4% of those for whom medical records were obtained) (Table 26). Hence, a drug problem was eligible for inclusion as a SHC for 197 participants. Of these, during the preceding 4 months, 43.9% had used illicit drugs 5 or more days a week, and 67.3% had used one or more Class A drugs. Class A drugs included heroin, morphine, cocaine, LSD and ecstasy. Of the 14 participants who reported drug use but it was not mentioned in their medical records, nine used only cannabis.

A treatment plan was initiated for 91.4% of participants identified by the CSSs as having a drug problem, with statistically significant differences by Health Service Model. A plan had been started for 98.3% of Dedicated Centre participants, compared with 78% from Mobile Teams (see Table 26). There were highly significant differences in relation to continuity of care: it was achieved for 84.9% of Dedicated Centre participants, compared with 15% of Usual Care GP participants. Continuity of care for Mobile Team participants increased from 25.7% to 48.6% when GP services were included. Continuity of care included shared care between GPs and substance misuse workers, maintenance on OST (55.9% were receiving such treatment at 8 months), involvement in recovery and rehabilitation programmes, and joint reviews with mental health and alcohol services (94.9% also reported mental health and/or alcohol problems). In relation to OST, some participants engaged with a programme and received treatment throughout the study, whereas some had spells when they failed to comply and treatment was intermittent.

There were fairly high scores across the models for outcomes 3 and 4. For outcome 5, 34.8% reported abstinence or a reduction in frequency of drug use during the study. The total score for all five outcomes was lower for Usual Care GPs, and the finding was statistically significant. Outcome 5 concentrated on changes in frequency of drug use. There were also small reductions in the numbers of participants who reported using Class A drugs by 8 months and who reported injecting drugs.

Management of Specific Health Conditions by Case Study Sites

There were differences by CSSs within the same model in the recognition of some SHCs. The percentage of participants who reported depression at baseline but it was not identified in the medical records varied by CSS (Figure 4). This applied to a higher percentage of cases at DC2 than at DC1. One reason might be that DC1 employed a full-time primary mental health nurse, whereas DC2 had input from community mental health practitioners at the CSS for just 15 hours per week. Hence, there may have been more opportunities for the DC1 mental health worker to engage with patients over time and recognise depression. Likewise, the percentage of cases of depression reported by participants at baseline, but not identified in the medical records, was higher for SP2 than for SP1. The latter had on-site sessions with mental health workers, whereas SP2 had input from a counselling service which relied on CSS staff to detect a problem and make a referral. SP1 also had full-time nurses and case managers dedicated to patients who were homeless (not the situation at SP2), who would have been able to arrange assessments for patients with low mood and possible depression.

FIGURE 4.

Depression reported by participants at baseline, but not identified in medical records, by CSS. Includes only participants who reported depression, and refers to the percentage of participants for whom depression not mentioned in medical records in previous 3 years.

There were also differences in the management of SHCs by CSSs within the same model, although, because of small numbers, it was possible to examine outcomes 1 and 2 for depression, alcohol and drug problems only (Table 27). MT1 was more likely than MT2 to provide continuity of care for all three SHCs, although the findings were not statistically significant. They also should be treated cautiously because of the large amount of Read-coded text used by MT1 that was coded as continuity of care.

Specialist GP 1 was more likely than SP2 to provide continuity of care for depression and drug problems, and the latter was statistically significant (chi-squared test: p = 0.013). This may be due to drug workers holding clinics at SP1 and providing OST, whereas this was not the case at SP2. In addition, as described previously, SP1 had full-time nurses and case managers working primarily with patients who were homeless, and staff regularly visited hostels and day centres, undertook street outreach work and worked closely with a local hospital. These are all factors that are likely to have a positive impact on continuity of care. Overall, continuity of care for alcohol problems was achieved for 61.6% of participants at CSSs with on-site alcohol services, compared with 44.9% where such services were unavailable (chi-squared test: p = 0.036). The respective figures for drug problems were 70.6% with, and 32% without, on-site drug services (chi-squared test: p = 0.0005).

There were differences for Usual Care GP sites, but the findings should be treated cautiously because of small numbers. UC3 was least likely to initiate a treatment plan for patients identified as having depression. As part of its health assessment for patients who were homeless, the PHQ-9 was completed, but it appeared that no further action was taken when scores indicated depression.

Summary

This chapter describes the management of SHCs by Health Service Models. Apart from hypertension, all other SHCs were common among participants and each was examined separately. The prevalence of hypertension may have been higher if all medical records were searched, instead of just those where the problem was first reported by participants. For the majority of participants who reported alcohol and drug problems, these were mentioned in their medical records, although they were less likely to be documented for Usual Care GP participants. This may be because participants felt less confident in admitting such problems to Usual Care GPs. As noted in Chapter 7 (see Table 15), screening for alcohol problems was highest among Usual Care GP participants. Given that several of this group of participants reported alcohol problems when interviewed by the research team, but they were not mentioned in their medical records (see Table 25), this suggests that some may have downplayed their use of alcohol when asked by general practice staff. Under-reporting of the use of alcohol and illicit drugs is common in health care settings and general population surveys. 123–126 Many participants reported chronic respiratory problems and depression, which were not documented in the medical records, despite baseline instruments indicating substantial problems. This applied to all Health Service Models, suggesting that staff were failing to recognise these conditions among some people who were homeless.

For each SHC, treatment plans were initiated for most participants identified in their medical records as having a problem. There were, however, differences with regard to continuity of care. It was most likely to have been achieved for chronic respiratory problems, whereas, for depression, alcohol problems and drug problems, it was attained for just over half of the participants. Dedicated Centres, followed by Specialist GPs, were most likely to have maintained continuity of care for each of the SHCs; for depression, alcohol problems and drug problems, the findings were significant. Mobile Teams were least likely to have provided continuity of care for all SHCs apart from drug problems, for which Usual Care GPs scored lower. However, when interventions by GP services were added to the Mobile Teams’ scores, continuity-of-care rates reached levels comparable to or above those of Usual Care GPs. As discussed previously, caution is needed when interpreting findings from MT1 as the medical records contained a large amount of pre-coded text with no added details. This was scored positively.

Continuity of care and treatment for the SHCs was not straightforward. Many participants had multiple conditions, namely depression, heavy drinking and the use of illicit drugs, which affected each other and presented challenges to health care providers. For example, medication for depression could not be prescribed in some instances because a person was drinking heavily. Treatments such as alcohol detoxification or OST maintenance were sometimes offered but were declined, or were introduced but not sustained. For many, care involved joint working between mental health and substance misuse services.

For each SHC, positive changes were reported by some participants between baseline and 8 months, or their condition remained stable. This applied to around two-fifths with depression or alcohol problems, slightly fewer with chronic respiratory problems, and just over one-third with drug problems. These findings must, however, be treated cautiously. They concerned only participants who were interviewed at 8 months, and were thus more likely to have sustained contact with the CSS. Several who could not be interviewed were in hospital or prison, and, in many instances, substance misuse had been a contributory factor to this outcome. As mentioned previously, people tend to under-report usage of alcohol and illicit drugs, and, unlike respiratory problems and depression, it was not feasible to use validated instruments to detect changes in their use. Moreover, there are many dimensions to assessing drug misuse over time, including frequency of use, types and multiplicity of drugs taken, toxicity of drugs taken and mode of administration (oral, inhaling or injection). No instrument could be found that captured these various dimensions. Adding to the complexity, the frequency and mode of drug administration differed according to the type of drug, and by factors such as changes in personal circumstances and access to finances. As a result, there were fluctuations in drug misuse among many individuals throughout the study period.

Each SHC was scored on five dimensions, each being allocated a score of 1 or 0. It can be argued, however, that achieving continuity of care over at least 8 months (outcome 2) is harder to achieve than starting a treatment plan (outcome 1) or positive feedback from participants about help received (outcomes 3 and 4). Likewise, assessing stability or change in health status from baseline to 8 months (outcome 5) covers a relatively short period. The scoring of the management of SHCs is not a validated tool. It is simply a way of assessing how the CSSs responded to various health conditions.

As with the primary outcome, there were differences across CSSs within the same Health Service Model in relation to recognising depression, and providing continuity of care for depression and substance misuse problems. The findings suggest that these may be related to service provision and staff at the sites. Dedicated Centres and SP1, characterised by features such as designated nurses and case managers for patients who are homeless, on-site specialist mental health and substance misuse workers, street outreach, clinics in hostels and day centres, and involvement in hospital inpatient work, were significantly more likely to have maintained contact with patients. The next chapter examines the impact of the Health Service Models on health status and well-being.

Smoking

At baseline, the majority (88.1%) of participants smoked cigarettes or tobacco (not including cannabis or other drugs). A further 4.4% used to smoke but had stopped. In comparison, just 13.9% of adults in England in 2019 were smokers. 127 It is difficult to collect precise information about smoking habits, although 21.1% of participants reported having more than or equal to 20 cigarettes or ‘roll-ups’ daily and so were classified as ‘heavy smokers’. Participants of Mobile Teams were least likely, and those of Specialist GPs most likely, to report current smoking; the findings were statistically significant (Table 28). Of those who smoked at baseline, 34.8% had received advice or help in the preceding 4 months with reducing or stopping smoking from CSS doctors or nurses, including 15 participants who had been given nicotine patches. Usual Care GP participants were more likely to have received help, although the findings were not statistically significant.

Thirteen participants who were smokers at baseline had stopped smoking by 8 months: four from Mobile Teams and three from each of the other models. Nineteen people who smoked, but not heavily, at baseline reported being heavy smokers at 8 months (four from Dedicated Centres, eight from Mobile Teams, five from Specialist GPs and two from Usual Care GPs). This may reflect an increase in their tobacco intake, or they were more candid over time about smoking habits. In contrast, 21 people who were heavy smokers at baseline smoked less than 20 per day at 8 months. Of those who did not smoke at baseline, none had started or resumed smoking at 8 months.

Nutrition

At baseline, the median number of days per week that participants had a meal containing protein, such as meat, fish or a vegetarian equivalent, was five. This varied greatly: 48.5% had such a meal 6 or 7 days a week, and 20.5% had such a meal once a week or less (see Table 28). Likewise, 25.1% had a portion of fruit or vegetables once a week or less. Just over one-third of participants cooked for themselves, 23.5% had food provided at their hostel and 45.7% had food at day centres. Other sources of food included soup runs and handouts on the streets (17.7%), takeaway food (13.9%) and meals cooked by relatives or friends (10.8%). Overall, 53.6% reported difficulty getting meals or eating healthily, mainly because of insufficient income, poor appetite or problems with cooking. Three-tenths had used a food bank at least once in the preceding 4 months. Participants in their own tenancy, followed by those sleeping rough, were more likely to report poor food intake. Among the former, 34.6% had a meal once a week or less, 63% described difficulty getting meals or eating healthily, and several relied on day centres for food. Participants who were sleeping rough mainly obtained food at day centres and street handouts.

There were no statistically significant differences by Health Service Model in nutrition habits among participants, although a slightly higher percentage of Usual Care GP participants reported difficulty getting meals or eating healthily. Nearly three-tenths of participants (28.5%) had received help in the preceding 4 months regarding nutrition from a CSS doctor or nurse, including advice about weight and food intake, vouchers for food banks or a prescription for nutritional supplements. Dedicated Centres and Specialist GPs were most likely, and Usual Care GPs least likely, to have provided help; the findings were statistically significant (see Table 28). Hence, Usual Care GP participants were most likely to describe nutritional problems and least likely to have received advice from the CSS.

By 8 months, the median number of days per week that participants had a meal had increased to seven, with 17% stating once a week or less. There were no statistically significant differences by Health Service Model in nutrition habits at 8 months, although slightly fewer (40.8%) described problems obtaining food or eating healthily.

Health status

The SF-8 is a short, self-administered instrument that produces a health-related quality-of-life profile. It measures eight ordinal items over the preceding 4 weeks: general health, physical functioning, ability to carry out physical activities, bodily pain, vitality, social functioning, mental health and emotional roles. Scale means, based on the same standard metrics as for the Short Form 36 Health Survey, are assigned to each response, creating a continuous outcome for each of the eight items. Regression coefficient weights are assigned to each item to produce a PCS and a MCS, both of which are continuous variables and have a mean of 50 (SD 10) in the general population of the USA. 128 Higher scores represent more favourable health states, and lower scores indicate worse health states.

The SF-8 was self-completed by the HEARTH study participants at baseline and at 8 months. At baseline, the mean PCS was 39.8 and the mean MCS was 35.4, with no statistically significant difference by Health Service Model. There was a slight increase at 8 months in the mean PCS and MCS across all models, but the findings were not statistically significant. When changes in scores are examined for participants who completed the SF-8 at both baseline and 8 months, there were no statistically significant differences by Health Service Model, although Mobile Team and Usual Care GP participants experienced a negative change in mean for the SF-8 MCS (Table 29).

The average scores in the general population of the USA for the PCS and the MCS are 50. 128 Of the HEARTH study participants, at baseline, 81.9% scored less than this for the PCS and 83.6% scored less than this for the MCS. No data could be found of SF-8 use in England among either the general population or people who were homeless to compare scores. However, a recent study in Belfast used the SF-8 to assess the mental health of people in different deprivation areas, drawing on the 2010 Northern Ireland Multiple Deprivation Measure. 129 The MCS score for those in the most deprived area (n = 215) was 45.6, which is considerably higher than that of the HEARTH study participants. Likewise, during a survey in Alameda County, California, 292 people who had been homeless for 12 months or more in the previous 3 years (i.e. chronically homeless) completed the SF-8. Their scores were slightly higher (PCS 42.1, MCS 39.6) than those of the HEARTH study participants. 130

Mental well-being

The SWEMWBS was developed in 2009 as a shorter version of the Warwick–Edinburgh Mental Wellbeing Scale. It is a validated instrument for use by the general population, and has seven items relating to psychological functioning, such as feelings of usefulness and optimism, thinking clearly and dealing with problems. 131 Each item allows five responses, from ‘none of the time’ to ‘all of the time’ (scored 1–5), and people rate their experiences over the preceding 2 weeks. Scores range from 7 to 35, with higher scores indicating greater positive mental well-being. The total raw scores from the SWEMWBS are transformed into metric scores using a conversion table.

The SWEMWBS was self-completed by the HEARTH study participants at baseline and 8 months. At baseline, the mean score was 18.5, with no statistically significant difference by Health Service Model. There was a slight increase in scores for all models at 8 months, but the findings were not statistically significant. Likewise, there were no statistically significant differences by Health Service Model in changes in scores between baseline and 8 months when the scores of the 248 participants who completed the SWEMWBS at both baseline and 8 months are examined (see Table 29).

When the scores are compared with responses in the Health Survey for England 2010–13 of 27,169 people aged 16 years and older,131 the mental well-being of the HEARTH study participants was considerably lower than that of the general population. In the HEARTH study, the mean baseline scores were 18.4 for men and 18.7 for women. In contrast, the mean scores in the Health Survey for England were 23.7 for men and 23.2 for women. 131

Regression modelling of Short Form 8 Health Survey and Short Warwick–Edinburgh Mental Wellbeing Scale scores

Dichotomous and continuous predictors, similar to those applied to the modelling of the Primary Outcome Score, were used to isolate the effect of the four Health Service Models in the modelling of two outcomes for each of the following: the SF-8 PCS, the SF-8 MCS and the SWEMWBS. The first concerns baseline scores, and the second concerns changes in scores from baseline to 8 months.

Summary

This chapter has examined the impact of the various Health Service Models on health status and well-being over time. A very high percentage of participants were smokers, and many had poor nutrition. There were slight improvements over time in their smoking and eating habits, but no statistically significant differences by Health Service Model in the provision of help for smoking. However, Dedicated Centres and Specialist GPs were most likely, and Usual Care GPs least likely, to have provided help with nutrition: a statistically significant disparity.

The SF-8 PCS, the SF-8 MCS and the SWEMWBS scores of the HEARTH study participants indicate low levels of physical and psychological functioning, compared with the general population and other marginalised groups. There were no significant differences in their scores at baseline by Health Service Model, and most models experienced a slight improvement in scores over time. Further interrogation of the data through regression modelling, however, revealed a positive improvement in changes in SF-8 PCS from baseline to 8 months by Dedicated Centres and Specialist GPs, and a positive change in SWEMWBS scores for Specialist GPs. However, several participant characteristics had strong significant effects on scores at baseline, and on changes in scores from baseline to 8 months. These mainly involved negative effects associated with older age, physical health problems, depression, heavy drinking, drug use and injection of drugs. A strong positive effect concerning involvement in education, training or employment was associated with change in the SWEMWBS scores from baseline to 8 months. The next chapter focuses on oral health and participants’ use of dental services.

Measures of oral health and dental access

Two proxies were used to assess oral health. The first was a single-item global self-rating of dental health (1 = very bad, 5 = very good),132 collected from participants at baseline and at 4 and 8 months. The second was the shortened Oral Health Impact Profile (OHIP-14), used to measure oral health-related quality of life (OHRQoL),101 completed by participants at 4 months. They rated the frequency (never to very often) of 14 dental-related impacts (problems) in the previous 8 months. In addition, at 8 months, participants were asked if their dental health had changed over the previous 12 months (1 = worsened a lot, 5 = improved a lot).

Features of access to dental care were hypothesised to involve registration with a dentist, dental attendance, completion or continuation of a course of dental treatment and satisfaction with dental care. The first three were recorded at each interview, and satisfaction with dental care was recorded at 8 months using a single-item rating scale. 133 Dental access was considered effective when urgent (pain) and felt dental needs were resolved. Experience of dental pain was recorded at baseline and 4 months,134–136 and felt need for dental treatment was recorded at baseline and 8 months. 104

Dental access is strongly affected by dental fear, and levels of dental anxiety were recorded at baseline and 8 months using the MDAS. The MDAS rates level of anxiety (1 = not anxious, 5 = extremely anxious) in five situations: anticipating dental treatment the next day; in dentist waiting room; about to have a tooth drilled; about to have teeth scaled and polished; and about to have a local anaesthetic injection in the gum. The MDAS has a maximum score of 25, with scores of 19 or more denoting dental phobia. 102

Oral health, dental anxiety and use of dental services at baseline

At baseline, 21.9% of participants rated their dental health as ‘very good’ or ‘good’. Participants of Dedicated Centres and Mobile Teams described their oral health more positively than those of the other two models (Table 32). Just over one-fifth (22.1%) had experienced dental pain ‘often or very often’ in the previous 4 months. By comparison, in 2009, 71% of the general population in England rated their dental health as very good/good,135 and 8% of dentate adults reported dental pain in the previous 12 months. 136

Nearly two-fifths (38.7%) of participants were registered with a dentist at baseline. This included a slightly higher proportion of Usual Care GP participants, although the findings were not statistically significant (see Table 32). Of those who responded, 6.8% were registered with a dentist at their CSS, 28.2% were registered with a dentist in the same town as that of the CSS and 3.1% were registered with a dentist in another town. By comparison, adult registration rates with NHS dentists in England over the study time frame ranged from 50.2% to 51.8%. 137–140 Of the HEARTH study participants, 56.9% had seen a dentist in the previous 2 years, including 21.7% in the previous 4 months. There were no statistically significant differences in having seen a dentist at least once in the previous 2 years, or in the previous 4 months, by Health Service Model.

A high proportion (82.2%) of the HEARTH study participants reported felt need for dental treatment. Of 177 participants who reported felt need and were not registered with a dentist, only 9% had discussed their dental need with CSS staff. Dental anxiety is an important predictor of dental attendance. 141 One-third of participants scored 19 or more when completing the MDAS, indicating dental phobia, with Specialist GP participants reporting the highest anxiety levels (see Table 32). In contrast, just 12% of the general population in 2009 described having dental phobia. 135 Several reasons were given by the HEARTH study participants for not seeing a dentist in the previous 24 months, most commonly dental fear (32.9%), ‘not got around to it’ (23.7%) and too embarrassed about state of teeth (17.1%). Just 12.5% said they could not find an NHS dentist, and 11.8% could not afford NHS charges (see Appendix 4, Table 52).

Oral health and use of dental services at 4 months

At 4 months, 25.2% of participants rated their dental health as very good/good, representing a slight improvement from baseline, but no difference by Health Service Model. The percentage who reported dental pain was similar at baseline and at 4 months. The proportion registered with a dentist had increased slightly from baseline, although the rate (43.4%) remained low compared with that of the general population. There were no changes in dental attendance rates over the preceding 4 months. Participants of Usual Care GPs were more likely than the rest of the sample to be registered with a dentist, and to have seen a dentist within the preceding 4 months. The latter was statistically significant (Table 33).

At 4 months, 273 participants completed all 14 questions of the OHIP-14. This covers seven domains of dental health that may affect a person’s quality of life. Each question was scored from 0 to 4, with a maximum total score of 56. Lower scores were more favourable. The overall mean score was 14.77 (SD 12.65), with higher scores for Dedicated Centres and Specialist GPs (17.08 and 16.51, respectively; see Appendix 4, Table 53). A contributory factor is likely to be that these two models had higher rates of heroin and cocaine use among participants, which have significant adverse effects on oral and dental health. 142 Among the HEARTH study participants, the total OHIP-14 mean score was 19.88 (SD 12.84) for people who used heroin at baseline, compared with 12.81 (SD 12.03) for those who did not.

To examine the number of impacts (problems) on oral health, responses from the OHIP-14 of occasionally/often/very often were scored 1, and responses of never/hardly ever were scored 0, with a possible maximum score of 14. The mean number of problems per participant was 4.89, with 81% reporting at least one problem, and 19% describing 10–14 problems (see Appendix 4, Table 53). By comparison, using similar scoring for the general population, just under two-fifths (39%) experienced at least one problem in the previous 12 months, and the mean number of problems per person was 1.2. 132 Dedicated Centre participants reported the most problems (28.2% described 10–14 impacts), followed by Specialist GP participants. The most common problems concerned self-consciousness and embarrassment about one’s dental health, followed by physical pain relating to aching in the mouth and chewing difficulties (see Appendix 4, Table 54).

The rates of behaviours described by PHE as promoting good dental health143 were suboptimal. Although 92.6% of participants who had their own teeth said they had a toothbrush, just 30% brushed their teeth twice or more a day, and 14.8% never brushed their teeth. Consumption of sugary drinks was common, with 79.5% of participants saying they put sugar in hot drinks, and 41.3% drank fizzy drinks six or more times a week. No differences were observed by Health Service Model.

Oral health, dental anxiety and use of dental services at 8 months

At 8 months, 25.3% of participants rated their dental health as very good/good, the same as at 4 months. Nearly half (48.8%) were registered with a dentist, and 29.5% had seen a dentist in the previous 4 months. Rates of registration and dental attendance had increased over the preceding 4 months, resulting in registration rates just slightly lower than those of the general population (around 51%). At the 4-month interview, Usual Care GP participants were most likely, and those of Dedicated Centres and Mobile Teams were least likely, to have seen a dentist in the previous 4 months, but the opposite was true at 8 months, and the findings were statistically significant (see Table 33). A high proportion of participants (77.3%) reported felt dental need at 8 months, and 34.3% had MDAS scores indicating dental phobia. These are comparable to baseline findings. Mobile Team participants reported a slight increase in dental anxiety over time.

Dental outcomes over study period

Over the 12-month study period, 49.4% of participants believed that their dental health had stayed the same, 15.1% said it had improved ‘a little’ or ‘a lot’ and 35.5% that it had worsened ‘a little’ or ‘a lot’. Mobile Team participants were least likely to report improvement, but the findings were not statistically significant (Table 34). There were, however, associations by participant characteristics. Those who reported depression at 8 months were more likely to say their dental health had worsened (40.1%, compared with 17% without depression; chi-squared test: p = 0.004), as were those who drank more than 14 units of alcohol weekly (40.2%, compared with 32.1% not exceeding guidelines; chi-squared test: p = 0.049). Likewise, 43% who reported drug misuse at 8 months said their dental health had worsened, compared with 26.4% not misusing drugs (chi-squared test: p = 0.024).

The number of participants registered with a dentist increased from 38.7% at baseline to 48.8% at 8 months. The latter included 16 people (14.7%) not registered at baseline (see Table 34). The majority, however, were either registered at both time periods (34.5%), or remained unregistered throughout the study (44.4%). Overall, 51.6% of participants saw a dentist at least once during the study period. This applied to fewer Specialist GP participants, although the findings were not statistically significant. When dental attendance over 32 months is examined (24 months prior to baseline interview and 8 months from baseline to the 8-month interview), 27.9% had not engaged with dental services, which is longer than national guidelines of no more than 24 months between dental visits. 144

Of the 104 participants who received dental treatment over the study period, most completed their course of treatment (70.2%) or it was still in progress (18.3%). Of those who received dental care, 69% rated it as ‘very or fairly good’, although Specialist GP participants were much less satisfied (see Table 34). Satisfaction levels were lower than national levels reporting very or fairly good (85%) during the study time frame. 145,146

Availability of dental services and integration with Case Study Sites

Information was collected about the availability of dental services and their integration with CSSs, drawing mainly on interviews with CSS staff and with two dentists. Dental services varied greatly and were not uniform within each Health Service Model (Table 35). Dental treatment clinics were held at the CSS or at homelessness services in the locality for MT2, DC2, SP1, SP2 and UC4, with weekly clinics specifically for people who were homeless available at DC2 and SP1. Dental treatment clinics were also held for people who were homeless or vulnerable at other locations run by community dental services (DC1, DC2 and UC2). Oral health promotion was undertaken by dental services in day centres used by people who were homeless in four localities. Such sessions included advice on oral hygiene and oral cancer screening. Two were day centres where a CSS also delivered services. At all locations where information was collected, CSS staff were involved in signposting patients to NHS dentists and/or referring to community dental services.

The extent to which a CSS was integrated with dental services varied, with four reporting close collaborative relationships. At DC2, for example, dental staff attended CSS staff meetings to discuss patients and CSS staff managed dental appointment bookings. The relationship between MT2 and dental services had developed through shared office space and they had used this opportunity to improve services. They discussed patients in common, CSS staff could refer patients and easily make appointments, and dental staff contacted CSS staff to help encourage patients to attend appointments. DC1 reported less integration, but had knowledge of dental services and assisted service providers working with homeless people to link with dentists so they could arrange oral health promotion sessions.

No associations were found between a dental treatment clinic being held at a CSS or within services specifically for people who are homeless and (1) rating of dental health over the study period, (2) registration with a dentist at 8 months and (3) dental attendance over the study period (see Appendix 4, Table 55).

Summary

Regardless of Health Service Model, most participants had poor oral health and experienced extensive dysfunction, discomfort and disability arising from this. Dental anxiety was relatively high, and access to dental care was suboptimal. Although dental registration rates increased during the study, many participants did not seek help or treatment, and much dental need, including dental pain, was left unaddressed. The availability of dental services and their integration with the CSSs varied within Health Service Models. No associations were found between the location of dental services (i.e. provided at a CSS or at homelessness service sites) and dental outcomes.

Integration with community mental health services

All CSSs reported high levels of integration (scores of 4 or 5) with one or more mental health service, and these scores were largely reciprocated by the external agencies. Many had links to community mental health teams. In addition, all except UC1 had connections to a specialist mental health team for people who were homeless, or for those with enduring mental health problems or complex needs. DC1, DC2, MT1, SP1 and UC4 also had links (mostly brief communication) to voluntary sector mental health services, such as Mind.

All CSSs except UC2 either employed mental health practitioners or had external workers hold regular clinics at the CSS. At DC2, for example, one such worker held sessions three times a week at the CSS, attended staff meetings on those days and met regularly with the GPs to discuss patients. Likewise, MT2 staff met regularly with a mental health team for people who were homeless to discuss patient care. MT1 and UC2 staff described ways in which they worked closely with mental health services:

We’ve quite a lot of patients we refer to them [mental health services]. [They send us] weekly updates ... they will tell us who has been referred, who is waiting for a mental health assessment, who has had their assessment, what the plan is, and who’s been taken to the [psychiatric hospital].

MT1 staff

We’ve a long working relationship with [mental health team for people who are homeless] ... one patient had been sleeping rough on the street for some time and recently got into accommodation. The [mental health] worker asked to meet me with the patient [and] his key worker ... we sat down together with the client and they handed over ... [The workers] then made themselves available for ongoing contact afterwards, although his medical care had been taken over by us.

UC2 staff

Some CSS staff discussed difficulties in working relationships with mental health services. Although a mental health team for people who were homeless was based at SP2, there was little contact between this service and CSS staff. According to MT1 staff, not having a GP in the team resulted in problems referring patients to secondary mental health care, as this required a GP or mental health professional referral. MT1 staff also explained that they would have welcomed more information-sharing with mental health services regarding safety aspects:

We’ve seen somebody and then afterwards been told, ‘You shouldn’t have lone-worked with that person’. Sometimes we don’t find out until a few weeks afterwards. So it would be really helpful to have proper collaborative working with mental health services so we can support one another.

MT1 staff

Integration with substance misuse services

Most CSSs reported integration levels of 4 or 5 with at least one substance misuse service, and the scores were largely reciprocated by the external agencies. No CSS directly employed substance misuse workers, but DC1, DC2, SP1, UC1 and UC3 had shared care arrangements in place, whereby substance misuse workers held regular clinics at the CSS, and GPs trained in substance misuse prescribed OST. At all of these sites except UC3, substance misuse workers were at the CSS at least 4 days per week. DC2 and MT1 staff also described collaborative working with residential detoxification services. The former, for example, provided primary health care to patients of a local detoxification unit while they were undergoing treatment at the facility.

At many CSSs, substance misuse workers who held on-site clinics attended staff meetings to discuss patient care, and worked jointly with the GPs. They described the benefits of such relationships:

It’s a high level of joint working because we literally talk to the GPs on a daily basis and it can be several times a day. We can see two or three patients together with the GP. We meet with GPs once a month to discuss more high-risk patients and any safeguarding issues.

Substance misuse worker, DC2

Joint working with a lot of the staff [at CSS], especially with the mental health team ... and obviously the GPs. We’ll have discussions about clients before I’m due to see them or after I’ve seen them. We’ll have quite complex discussions ... we might set up case conferences here with external agencies.

Substance misuse worker, SP1

Where on-site substance misuse services were not in place, CSS staff referred patients to external services. However, communication with these services varied, and integration levels of 3 and 4 were reported. Some MT2 and SP2 staff described being unaware of whether or not their patients attended appointments with substance misuse workers, and were not provided with updates on their patients’ progress (this was apparent to the research team when the medical records were searched). This reflects the findings of SHC outcomes: MT2 was less likely than most other CSSs to have achieved continuity of care for alcohol and drug problems (see Table 27). The following extracts illustrate the difficulties experienced by MT2 and SP2 staff:

We refer people to [substance misuse services] ... but we don’t know if they’re being seen. We don’t know if they’re going ... I can think of one person I referred that was seen but [I only know that] because the patient told me.

MT2 staff member

[The drug service] is quite separate and we don’t do a lot of joint working. I think communication between us and them could be better. [Difficulties occur] when we have to ask for updates on what they are doing with people ... there isn’t great communication, I don’t think, in either direction, and I’m sure that’s something that could be improved.

SP2 staff member

Integration with hostels and temporary housing schemes

All three specialist Health Service Models reported high levels of joint working with several hostels and temporary supported housing schemes. For example, MT1 scored integration levels of 4 and 5 with 11 hostels, and the remaining CSSs, apart from UC1 and UC4, gave similar high scores for at least three hostels. A few hostels in the locality of the Dedicated Centres and MT1 were given low integration scores (0 or 1) by CSS staff. This was partly because the hostels were a distance from the CSS and residents therefore registered at local general practices. SP2, MT2 and UC1 reported similar low scores for four or more hostels, with some CSS staff indicating that insufficient resources prevented them from doing more collaborative work. UC1 in particular had very little involvement with hostels.

Regular clinics were held in one or more hostels by nurses or GPs from DC1, MT1, SP1 and SP2. Instead of clinics, MT2 had weekly meetings with staff at one hostel to discuss concerns about residents, and provided training at hostels on the health needs of residents. The aim was to reduce inappropriate out-of-hours service use. Strong integration links between CSSs and hostels enabled flexible working and prompt responses between the two services. Hostel staff, for example, telephoned the CSS and sought guidance if they were concerned about a resident, without having ‘to go through a lot of red tape’. Likewise, CSS staff liaised with hostel staff if a resident required encouragement to attend a health appointment. The Mobile Teams were sometimes under pressure from hostel staff to hold (additional) clinics at their services, although the former believed that training for hostel staff was more feasible and a better use of resources. The following examples exemplify the positive links between some CSSs and hostels:

We do teaching sessions at the hostels [with staff] to try to help with reducing ambulance call-outs ... there’s a hugely disproportionate use of ambulances at homeless hostels and we try and reduce frequent attendance at accident and emergency ... so we give health promotion talks at hostels, [such as] if one of the hostel residents has a seizure what would you do, when would you call the ambulance.

MT2 staff member

If we’re in a crisis situation, we’ve rung up [the CSS], explained what the situation is and I think because of the working partnership that we have with them, we have the credibility (or they give us the credibility) of not saying it’s a crisis if someone sneezes twice. I think building up that working relationship has paid dividends. It works really well.

Hostel staff member, DC1 locality

In some CSS areas, the closure of several hostels meant that residents were dispersed into small housing units across a wide geographical area, making it difficult for CSS staff to maintain contact with vulnerable patients and support staff. Low staffing levels in the hostels and the use of agency staff also led to challenges in maintaining relationships.

Integration with day centres and street outreach teams

All CSSs within the specialist Health Service Models gave scores of 4 or 5 for integration with one or more day/drop-in centres for people who are homeless. Indeed, SP1 staff reported high integration levels with seven centres. Apart from UC3, there was much less contact between Usual Care GPs and day centres, and some CSS staff had very little knowledge of such services. Day centres were the settings for much of the work undertaken by Mobile Teams, where they held regular health clinics, worked jointly with staff to plan care for service users with complex needs, and assisted with medical emergencies. As one MT2 staff member explained, day centres are now doing a great deal of work with people who have mental health problems that used to be undertaken by health and social services. SP1 and SP2 also held regular clinics at a day centre. DC1 and SP1 visited a centre regularly to encourage registration with the CSS, and DC2 undertook periodic health promotion activities at a centre, such as giving influenza vaccinations. Some day centre staff expressed a wish for closer integration with CSSs as they believed that this would benefit service users. The following examples illustrate the different relationships between CSSs and day centres:

We do some close working with day centres, because they now do a lot of the work that health and social services used to do. A lot of day centres will now do the mental health work with the patient.

MT2 staff member

We’re just another person who might ring [the CSS]. There are no formal links. If we ring up, we get in the queue just like everyone else; there’s no secret passage for our clients ... no one from [the CSS] will ever ring here ... and ask about someone. I think it would be good for them to link with agencies like us who are spending time with patients and clients – we know them and I think that goes a long way towards solving someone’s health issues.

Day centre staff, UC1 locality

All specialist Health Service Models worked closely with street outreach teams. Staff of Dedicated Centres and SP1 accompanied outreach workers on the streets, and the outreach workers accompanied clients to appointments at the CSSs. SP2 worked closely with the street team, but CSS staff explained that they did not have the resources to conduct outreach themselves. Both Mobile Teams undertook regular outreach with local teams. Among Usual Care GPs, only UC4 reported regular communication with a street outreach team.

Other integration and strategic work

The level of integration between the CSSs and housing departments and social care services varied, although most reported some communication, mainly around patient referrals. Dedicated Centres, Mobile Teams and SP1 also had some involvement with local hospitals. For example, DC2 advised hospital doctors about medication for drug dependency and communicated around patient discharge, and DC1 and SP1 were involved in hospital inpatient care and visited patients on wards. The Mobile Teams reported integration mostly at level 5 with some general practices in their area. They encouraged patients to register with a GP, and, in most cases, shared patients’ medical records. MT1 ran joint clinics in a few hostels with a GP, whereas MT2 had weekly meetings with a GP to discuss patients with complex needs and occasionally held clinics in that general practice.

The specialist Health Service Models, but not the Usual Care GPs, were involved in local strategic meetings regarding provision for people who are homeless. The former also attended multiagency meetings concerning people sleeping rough, frequent attenders at EDs, safeguarding concerns and people with complex needs. SP2 highlighted the benefits of multiagency meetings:

One patient has significant physical health problems and substance misuse ... so we had a safeguarding meeting that was led by someone from the substance misuse team. We now have a communication group regarding that patient where we’re all kept up to date. [Multiagency meetings] work for individual cases where there’s an increased level of concern.

SP2 staff member

Strengths and limitations of the work of the Case Study Sites

This section presents the qualitative analysis of the interviews with CSS staff, other service providers and stakeholders. In each CSS locality, these participants were asked for their views of: (1) the strengths of the CSS and what aspects of the service worked well; (2) whether or not the CSS was meeting the health, social and welfare needs of local people who were homeless; (3) the limitations of the CSS and what aspects of the service did not work well; and (4) the ways in which the CSS should change or improve its services. Their opinions are summarised in Appendix 5, Table 56. Six themes relating to the mechanisms and contextual factors listed in this evaluation’s framework emerged from the data (see Table 1): (1) understanding of homelessness and the development of specialist services, (2) CSS staff members’ attributes and delivery of tailored (person-centred) health care, (3) partnership-working and the delivery of holistic care, (4) flexibility and accessibility of the CSS, (5) outreach and engagement and (6) unavailability of supporting services. Each of these themes is discussed in the subsequent sections. UC4 has not been included as only one interview was conducted.

Summary

All three specialist models appeared to be well integrated with mental health, substance misuse and homelessness sector services. Relationships were founded on a shared client group, and an understanding of the problems experienced by people who are homeless and the complexity of their health and social care needs. Although this work often took place at the individual-client level, CSSs that worked collaboratively with services (not solely around patients) had opportunities to become involved in strategic approaches to tackle unmet need. Usual Care GPs were less integrated with homelessness sector services, and did not have the same extensive networks as specialist models. However, some achieved high integration levels with a small number of hostels, suggesting that positive working relationships can be built between mainstream general practices and homelessness sector services. Overall, for all CSSs, the extent and depth of integration with services largely depended on the resources available, as most expressed a wish to maximise partnerships.

The benefits of a hub of services in the same building were acknowledged both by CSS staff and external agencies. It enhanced partnership-working and enabled easy access to services. Although it would not be feasible or practical to introduce such an arrangement in many locations, more collaboration between services is likely to be beneficial. Poor communication between substance misuse and mental health services and a few CSSs indicates the need for improvement regarding information-sharing. Homelessness sector staff valued greatly the relationship they had with some CSSs, citing ease of access to the practice and staff willingness to advise and provide support as important factors.

The CSS staff and agency interviews highlight the intricacies of engaging people who are homeless in health care, and the importance of designing services that are appropriate and accessible. Staff of the three specialist models in particular had the experience and capacity to develop tailored approaches to health care that were acceptable to patients, could be delivered opportunistically and had some positive health outcomes. Although both CSS staff and external agencies stressed the need for more outreach work by the CSSs, there appeared to be different expectations and opinions as to its aims and how and where it should be delivered. Such interventions also require consideration of the issues raised in this chapter around delivering health care in non-NHS settings.

Caution has to be taken when interpreting the integration scores. In a few instances, staff gave unusually high scores for their involvement with a particular service. They may have overstated the level of integration as they believed they should have been working more with the service (although ‘expected’ scores were collected as well). There were also difficulties identifying a few services mentioned by staff as they could not name the service and were unclear about its purpose.

Participants’ ratings of Case Study Sites

Case study participants were asked at baseline about their views of the relevant CSS, drawing on questions from the GP Patient Survey. 147 Most rated the CSS positively, with overall experience as ‘very good’, ranging from 82.3% of Mobile Team participants to 38.6% of Usual Care GP participants (Table 36). The latter received a higher (poorer) median score than the three specialist models, and the findings were statistically significant (Kruskal–Wallace test: p < 0.0005). Compared with the general population’s rating of their general practices in the 2017 GP Patient Survey,147 the HEARTH study participants of the three specialist models were much more likely to describe their CSS as very good, whereas Usual Care GP participants were more likely to give an unfavourable score.

Participants were also asked to rate the care they received the last time they saw a doctor or nurse at the CSS, according to how good the person was at:

• giving them enough time

• listening to them

• explaining their condition and treatment

• involving them in decisions about their care

• treating them with care and concern

• providing or arranging treatment.

They could respond as follows: 1 = very good, 2 = good, 3 = neither good nor poor, 4 = poor and 5 = very poor. An average response was calculated from the total scores, creating a six-item ‘Quality of Care’ rating, with lower scores being more positive. The mean score was 1.60 (SD 0.84); it was 1.20 (SD 0.39) for Mobile Teams, 1.60 (SD 0.79) for Dedicated Centres, 1.65 (SD 0.82) for Specialist GPs and 2.08 (SD 1.07) for Usual Care GPs (ANOVA: p < 0.0005).

Using data from the 2017 GP Patient Survey,147 comparisons of care by a GP were made between the general population and HEARTH study participants from Dedicated Centres, Specialist GPs and Usual Care GPs. Dedicated Centre and Specialist GP participants were much more likely to rate aspects of care from a doctor as good or very good, whereas participants from Usual Care GPs were less likely to give positive ratings (Table 37). Comparisons of care by a nurse were made between the general population and HEARTH study participants of Mobile Teams. The latter were substantially more likely than the general population to score their care as good.

Similar findings emerged when participants were asked if they had confidence and trust in the doctor or nurse. Participants from the specialist models were more likely than the general population to respond ‘yes, definitely’. Nearly one-quarter (23.1%) of Usual Care GP participants, compared with 4.3% of the general population, said they had no confidence or trust in their doctor (see Table 37). Participants of the three specialist models were more likely than those of Usual Care GPs to describe easy access to the CSS, saying they could ‘drop in’ to the CSS when they wanted to see a doctor or nurse and be seen that day. This applied to 94.7% of Mobile Team participants, 85.1% of Dedicated Centre participants, 87.5% of Specialist GP participants and just 10.8% of Usual Care GP participants (chi-squared: p < 0.0005).

There were differences across CSSs in the Usual Care GP model in ratings of the CSS. Half of UC2 and UC3 participants described their overall experience of the CSS as ‘very good’, which is slightly higher than the general population’s score (42.9%). In contrast, this applied to just 21.4% of UC1 participants. There was no marked difference in GP confidence and trust. Regarding Quality of Care, UC2 scored higher than UC1 and UC3, and its scores for each item were slightly higher than those of the general population. The scores for UC1 and UC3 were lower than those of the general population.

Regression modelling of participants’ ratings of Case Study Sites

Dichotomous and continuous predictors were used to isolate the effect of the four Health Service Models in the modelling of participants’ ratings at baseline of (1) overall experience of the CSS and (2) Quality of Care when last seen by a doctor or nurse at the CSS. The predictors were similar to those applied to the modelling of the Primary Outcome Score and the SF-8 and SWEMWBS scores, with the addition of variables concerning overall experience of the CSS, Quality of Care ratings, trust in doctor or nurse and ability to ‘drop in’ to CSS to see a doctor or nurse.

For each outcome, an initial backward stepwise linear regression was run with each dependent variable at baseline and the variables in Tables 48 and 49 (see Appendix 3) as predictors. Statistically significant and near statistically significant (p < 0.15) were used in a second backward stepwise linear regression. A final model was run using only the statistically significant predictors resulting from this regression to maximise the number of observations used.

The salient fitted parameters, namely those for Health Service Model (whether or not statistically significant), and for other statistically significant predictors for overall experience of CSS and Quality of Care quality outcomes, are shown in Table 38. With regard to overall experience of the CSS, a highly statistically significant beneficial effect is seen for the Mobile Team and Specialist GP models in comparison to the Usual Care GP model. Other predictors that showed strong statistical significance were overall rating of Quality of Care when last seen by a CSS doctor or nurse, and having trust in that person.

With regard to the Quality of Care outcome, the final regression model is not an especially impressive fit (R² = 0.176; see Table 38). Nevertheless, it clearly demonstrates that the Dedicated Centre, Mobile Team and Specialist GP models receive a superior Quality of Care rating by participants, compared with the Usual Care GP model, having adjusted for other statistically significant factors (i.e. involvement in education/training/employment, recent drug misuse and participant age). Substituting these three Health Service Models with Usual Care GP in the regression model elicits a highly statistically significant (p < 0.0005) corresponding parameter of 0.600 (95% CI 0.397 to 0.803), which represents a sizeable deterioration in Quality of Care on a scale with a width of just 4 (from 1 = very good to 5 = very poor).

Participants’ accounts of the Case Study Sites

This section presents qualitative findings of the participants’ perspectives of the CSSs. It draws on their interviews at baseline and 8 months when they gave their opinions of different aspects of the CSS, including accessing the service; opening hours; experiences of the waiting room; the attitudes of receptionists and clinical staff; the delivery of health care by doctors, nurses and other staff; and help given for housing, welfare and social problems. Six themes emerged from the data that pertain to some of the mechanisms identified in the framework for the evaluation of the CSSs (see Table 1): (1) the attitude of staff, (2) provision by the CSS of a welcoming and acceptable environment, (3) provision by the CSS of a safe and non-threatening environment, (4) flexibility and accessibility of the service, (5) practical aspects of health care delivery and (6) a holistic approach to care that addresses non-medical needs.

People who were homeless and not using Case Study Sites

One of the HEARTH study’s secondary outcomes was to examine whether or not the CSSs were addressing the health needs of the local population who were homeless, and whether or not there were unmet health needs. Interviews were conducted with 107 people who were homeless but not registered with the CSSs (29 in Dedicated Centre areas, 14 in Mobile Team locations, 35 in Specialist GP areas and 29 in Usual Care GP areas).

Most interviewees were male (76.6%) and White British (72%), with an average age of 40 years. Almost two-thirds (64.4%) had been homeless for more than 1 year. When interviewed, 42.1% were staying in hostels, 17.8% were in supported housing, 13.1% were sofa surfing and 18.7% were sleeping rough. Most reported physical and mental health problems.

Just over four-fifths received health care from a GP surgery or medical centre, and several also used walk-in clinics and A&E. Those in Dedicated Centre areas were most likely to use A&E for health care (see Appendix 6, Table 57). Only 6.5% reported no health care. Nearly three-quarters (72.4%) had seen a doctor or nurse in the previous 3 months, whereas for 18.1% it had been more than 12 months. Just over half (55.1%) had attended A&E in the previous 12 months, including 9.3% who attended at least four times.

One-fifth had used the CSS in the past, mainly those in Dedicated Centre or Specialist GP areas. The main reasons for not using the CSS were that they were already registered with a GP and were satisfied with the care they were receiving, or they had not heard about the CSS and what it had to offer. A few said it was too far from where they were staying. Seven people said they were deterred from using the CSS by the presence of people who use drugs also accessing the service or in its vicinity, and two said they did not want to associate with people who were homeless or with homelessness services. This group of nine were in Dedicated Centre or Specialist GP areas. Two additional people mentioned that they had been removed from the CSS practice list because of their aggressive and demanding behaviour. The following comments illustrate their mixed reasons for not using the CSS:

I’ve been with my GP for a long time. It is clean and convenient to the chemist, and I feel safe and secure.

UC3 area

It is easy to access services, easy to get a sick note. I’ve been with my current GP for 4 years and am happy with the service.

DC1 area

[The CSS] is on an ‘intimidating street’ with lots of ‘horrible people’ hanging around. If you go there, it’s like admitting you’re a loser who can’t manage to get over problems and move forward. You get tarnished by association if you are seen there by other people.

DC2 area

[CSS] horrible place – full of druggies.

DC1 area

Summary

This chapter summarises the views of people who were homeless about the CSSs, the care they received, and their reasons for using or not using the service. The case study participants of the three specialist models rated the service and care they received considerably more favourably than participants of Usual Care GPs and the general population’s ratings of their GP surgery. Usual Care GP participants rated the service less favourably than the general population, although there was some variation between individual CSSs in this model. The findings indicate that people who are vulnerable or have complex needs find it hard to manage mainstream GP systems, but are very positive about specialist services with staff who are understanding of homelessness and respond flexibly to needs.

As described in Chapter 1, many single people who are homeless face difficulties accessing primary health care services because of factors such as mistrust of health care professionals, an unsettled lifestyle and competing needs, the inflexibility of services, and negative attitudes from some staff. In this study, the majority of case study participants from all Health Service Models believed they were treated with respect by CSS staff, but those from the specialist models were more likely than those of Usual Care GPs to mention welcoming and friendly gestures by staff, the flexibility of the service, ease of getting appointments with GPs and nurses, and confidence and trust in the staff. These findings largely corroborate the views of CSS staff and external agencies described in the previous chapter.

Most people interviewed in this study who were homeless but not using a CSS received health care elsewhere, mainly at general practices. Only a minority reported not having health care. The findings should be treated cautiously. The sample was obtained opportunistically at hostels and day centres, and no interviews were conducted with people who were sleeping rough and not using these services. Some external agencies expressed concern about the health needs of people sleeping rough who were reluctant to engage with services. To better understand their health needs, further investigation would be needed.

Aim

The aim of the health economic analysis was to compare four models of primary care provision for people who are homeless with respect to: (1) the resources assigned by services to the care of registered patients who are homeless and (2) the use and costs of health and social care services over a 12-month period by the case study participants. It also sought to evaluate outcomes for participants in relation to service use costs. The four models of primary care provision, as described previously, were (1) Dedicated Centres providing care only to people who are homeless and other marginalised groups, (2) Mobile Teams, (3) Specialist GPs providing targeted services (e.g. specific clinics) within their practice and (4) no special provision by practices (Usual Care GPs).

Data sources

People who were homeless were recruited through the services/practices (CSSs) that were involved in the study. Two sites and 96 participants were included for each of the three non-usual care models; 75 people were recruited from four Usual Care GPs; overall recruitment was 363.

Information on the resourcing of the CSSs was requested through interviews with managers and other personnel. Questions were asked about organisation and financing, the patient population and proportion who were homeless, staffing, facilities, dedicated clinics, other services provided (e.g. for mental ill health, drug misuse and lifestyle), and integration with health and social care systems.

Service use data were collected for each participant starting 4 months before the date of baseline interviews to the date of the 8-month interviews, giving 12 months of data by two methods. Self-report data were collected retrospectively by recall through questions embedded in the interview schedules (at baseline and at 4 and 8 months) covering contacts with the primary health care centre (GPs and nurses) and other health and allied health professionals, out-of-hours services (A&E, NHS walk-in/urgent care clinic, NHS 111, ambulance call-out) and hospital admissions. Primary care medical records of participants were also accessed (subject to separate consent) at the end of the 12-month follow-up period. These were taken as the primary source of health service use data as they were more comprehensive and complete than information obtained from participants, and they were also considered to be more reliable. GP records provided service use data of participants in the three GP-led models; both Mobile Team and GP records (when shared) were used for participants in the Mobile Team model. Use of accommodation (e.g. nights in hostels) was not included.

Self-report data were used to provide information on items that were not routinely available from medical records, such as contacts with key workers, substance misuse services, pharmacists for methadone, local authority housing and social workers, personal care workers (e.g. home care workers) and volunteer supporters. Dates were noted when participants were not using local services during the 12-month observation period, for example because they were in prison, were in rehabilitation, had left the area, or had died. Data were recorded initially by 4-month periods, but were amalgamated for analysis. Service use data were costed in 2020 Great British pounds using national validated tariffs;107 costs of tests were accessed from a variety of sources (see Appendix 7, Table 58).

Service level

Information reported by managers and other personnel regarding patient populations, resourcing and organisation of services is reported in Chapter 5. All sites were commissioned through the NHS and received local enhanced payments for providing services to people who were homeless (UC1 received payments for providing services to patients who had substance misuse problems). The Dedicated Centre, Specialist GP and Usual Care GP sites were GP led and provided care primarily at general practice premises; the Dedicated Centres and Specialist GPs also offered some outreach. Mobile Teams were run by specialist nurses through clinics in day centres and hostels. They provided more limited services around wound dressing, monitoring of long-term conditions and health promotion, referring patients on to other services as needed, including to local GPs with whom they worked closely, and shared medical records. A GP out-of-hours service was available in all sites.

The patients receiving care from Mobile Teams also differed significantly in some respects from those registered with GPs in the other models (see Chapter 6). Mobile Team patients were more likely to have been born outside the UK, to be European Union migrants and to sleep rough; they also were less likely to use drugs.

Participant service use

Costs

The cost of contacts with services and professionals is shown in Table 41. Considering all health and social care use (except dentist and optician, and the extensive personal care support received by two participants), the highest-cost items were hospital stays (other than for detoxification) (mean £2164, median £0), out-of-hours care (i.e. A&E, walk-in, urgent care, NHS 111 and ambulance) (mean £792, median £296) and contacts with GPs (mean £686, median £462). The Grand Total Costs of service use were highest in the Dedicated Centre model.

Average costs were compared between models for six main service use groups using the Kruskal–Wallis test, and between pairs of models using the Mann–Whitney U Test (Table 42). Statistically significant differences existed between models for the following: GP, nurse and HCA contacts (higher for Dedicated Centres than for the other three models; lower for Usual Care GPs than for Mobile Teams and Specialist GPs); use of smoking, drug and alcohol services, including methadone collection from pharmacists (higher for Specialist GPs than for Mobile Teams and Usual Care GPs; higher for Dedicated Centres than for Mobile Teams);

social care contacts with care co-ordinators/managers, key workers, and housing and social care officers (higher for Mobile Teams than for the other three models); Grand Total Costs (higher for Dedicated Centres than for Mobile Teams and Usual Care GPs; higher for Specialist GPs than for Usual Care GPs). No significant differences were found between models for out-of-hours services or for hospital nights (including psychiatric, excluding detoxification). Distributions were right skewed for all items and models. Differences are illustrated in box plots (Figure 5).

FIGURE 5.

Comparison of costs by Health Service Model. (a) GP, nurse and HCA contacts; (b) smoking, drug and alcohol services, pharmacist delivery of methadone; (c) social care contacts (care managers/co-ordinators, key workers, housing and welfare officers); (d) out-of-hours (A&E, walk-in/urgent care, NHS 111, ambulance); (e) hospital nights (including psychiatric, excluding detoxification); (f) total cost of all service use (excluding dentist, optician and personal carer). Boxes denote the IQR and show median bar; whiskers are minimum and maximum, with outliers 1.5 to 3 IQRs from the end of the box shown as small circles and extreme outliers more than 3 IQRs from the end of the box shown as asterisks. IQR, interquartile range.

Comparisons of sites within models were also conducted for each of the six main cost groupings. No statistically significant differences were found except between the two Dedicated Centre sites for costs of participant use of smoking, drug and alcohol services, which were significantly higher in DC1 than in DC2 (data not shown) due to the large numbers of contacts with pharmacists for collection of methadone.

Costs in relation to outcomes

The rates of use and the costs across all services were significantly higher among participants in the Dedicated Centres than in the other three models, and significantly lower in the Usual Care GP model than in the Mobile Teams and Specialist GPs. The specialist interest of staff for care of people who are homeless, and high levels of integration with relevant services (mental health, substance misuse) and local homeless hostels and day centres, may contribute to the higher service use of participants registered with Dedicated Centres. In contrast, Usual Care GP sites are not primarily set up to deal with the special needs of people who are homeless; they reported lower levels of integration with other providers, and staff may have been less proactive in identifying issues and referring on. Drawing on findings from previous chapters, there is some evidence that higher levels of service use are reflected in better outcomes, but this does not hold for all measures.

Screening activity (the primary outcome) was not significantly different between the three GP-led models, but was significantly lower in the Mobile Team model. Similarly, performance on care for SHCs (secondary outcome, overall score) did not differ between models, except that the Usual Care GP model recorded a lower score than the other three models for people who used substances. Dedicated Centres did, however, score significantly better on the continuity of care component for all four of the SHCs analysed, with Mobile Teams performing least well on this outcome. Controlling for other factors, participants in the Dedicated Centre and Specialist GP models showed small improvements in PCS of the SF-8 over an 8-month period; the well-being of participants (measured using the SWEMWBS) in the Specialist GP model also improved. However, this finding should be viewed with caution: approximately 100 participants were not available for follow-up, and the 250 participants for whom follow-up data were available may not have been representative of the entire group. Although relatively low proportions of participants in all four models reported receiving nutrition advice (mean 28.5%), this was lowest among those in the Usual Care GP model (13.3%). In terms of satisfaction with the primary care service, participants in all three specialist services (Dedicated Centres, Mobile Teams and Specialist GPs) rated their experiences highly (more so than is recorded for the general population); participants in Usual Care GPs rated their experiences worse than the general population did, but this was largely a result of low scores in just one of the sites. Costs and outcomes across the four Health Service Models are summarised in Appendix 11, Table 65.

Summary

The service used most frequently by participants was the GPs (nurses in Mobile Teams) with which participants were registered. The number of contacts with GPs over the 12-month period was considerably higher among the study sample than has been recently observed among the general population [mean 10.7 (SD 11.4) vs. mean 3.74 (SD 1.24), respectively], and similarly for nurses [mean 8.3 (SD 15.2) vs. mean 1.3 (SD 0.78), respectively]. 149 Management of dependency problems (smoking, alcohol and drug misuse) was also associated with relatively high average costs, largely attributable to pharmacist time distributing OST, including methadone and buprenorphine, on a daily basis (n = 142, 39.1% of participants). out-of-hours services were accessed by 65% of participants over the 12-month period, with many having multiple contacts. Modelling suggests that out-of-hours service use is associated with having many accommodation changes; other participant characteristics (demographic or health) and model of care were not predictive.

On average, Dedicated Centre participants incurred higher overall service use costs than participants in other models of care. Having more changes of accommodation and spending more time in accommodation with on-site staff also added to cost; people who were black/Black British or in education, training or employment were associated with lower service use costs. The higher service use in the Dedicated Centre model was associated with significantly better outcomes for some, but not all, indicators measured in this study. Participants reported low levels of satisfaction with Usual Care GP provision and high levels of satisfaction with the other three models.

Limitations

The analysis is limited in various ways. It was not possible to identify resources devoted to care of patients who were homeless within services for calculating an average cost of provision in different models, but all services received NHS payments in recognition of their care of this group. Some service use data were not available from medical records for some participants for all or part of the follow-up period, resulting in differential reliance on self-reported information, which may have been less reliable. We did not adjust for missing information or undertake imputation because the distribution of number of days that participants had access to services was similar across models, which means service use may underestimate actual annual service use across the board. The uneven distribution of missing medical records across sites (higher for Mobile Teams and Usual Care GPs) may have affected the comparison of Grand Total Costs across models. Differences between the mode of delivery of care and characteristics of the patient populations of the nurse-led Mobile Team model, compared with the three GP-led models, need to be borne in mind in the interpretation of comparisons. Similarly, some sites provided more in-house services (e.g. support for smoking and drug and alcohol misuse), which were included in GP and nurse contacts, rather than appearing as a contact with a specialist service. Although this would not affect the analysis of Grand Total Costs, it may have affected comparisons of costs between models. Granularity was lost in the interests of parsimony in the combining of variables for the cost analysis.

Dedicated Centres

Dedicated Centres are specialist primary health care centres, mainly for people who are homeless. This model was represented by two CSSs and 96 case study participants. Outcomes for Dedicated Centres were favourable, even though participants of this model tended to be less settled and had more complex needs than those of other models. At baseline, they were significantly more likely than Specialist GP and Usual Care GP participants to be sleeping rough and less likely to be staying in hostels or other staffed accommodation. They exhibited the highest mean number of accommodation changes during the study, spent significantly less time in staffed accommodation and spent almost twice as much time in prison (see Table 11). Dedicated Centre participants were also significantly more likely than those of Mobile Teams and Usual Care GPs to use heroin or cocaine, inject drugs and be receiving OST.

There was no statistically significant difference between this model and the Specialist GP and Usual Care GP models in the number of HSIs screened for the primary outcome (Mobile Teams scored significantly lower). Dedicated Centres had the highest total outcome score for each SHC (chronic respiratory problems, depression, alcohol problems and drug problems), and scored significantly higher than other models in maintaining continuity of care for patients with depression, alcohol problems and drug problems. In relation to health status, there was a statistically significant positive improvement in the SF-8 PCS scores from baseline to 8 months for Dedicated Centre participants (p = 0.005; paired t-test). They also reported the highest level of help or advice from CSS staff regarding nutrition.

Compared with the general population, Dedicated Centre participants were more likely to rate their overall experience of their general practice and the care they received as ‘good’ or ‘very good’, and were more likely to report a higher level of confidence and trust in their GP. Many described the CSS as a welcoming and accessible environment, and found the staff to be non-judgemental, friendly and helpful. Compared with other models, Dedicated Centre participants had substantially more GP contacts during the study and, apart from Mobile Team participants, had many more contacts with nurses (see Appendix 10, Table 64).

The relative success of this model is likely to be attributable to several contextual and service delivery factors (summarised in Appendix 12, Table 67). Both Dedicated Centres had been delivering health care to single people who were homeless for many years, and most staff had considerable experience of working with this patient group. The CSSs received NHS funding exclusively for this work, and the number of registered patients was smaller than that of Specialist GPs and Usual Care GPs. This enabled Dedicated Centres to provide flexible, holistic and well-integrated services. For example, they operated ‘drop-in’ clinics whereby patients could be seen by a doctor or nurse without an appointment, offered consultations that were longer than the customary 10 minutes, and engaged in some outreach on the streets and at homelessness services. Multidisciplinary working was prominent at both Dedicated Centres, with on-site mental health and substance misuse services, and daily staff meetings attended by CSS staff and sessional workers. The CSSs were well integrated with local hospitals, street outreach teams, and hostels and day centres, and staff attended multidisciplinary case management meetings. These factors are likely to have contributed to participants’ higher levels of service use in Dedicated Centres than in other models, and aided continuity of care with a patient group who tended to be unsettled or hard to engage. Although Dedicated Centre participants incurred higher overall service use costs than other models, this is likely to reflect the complexity of their health problems and the intensity of health care delivered.

The model operated relatively successfully in the context of a changing environment. Both CSSs were in cities with large populations of single people who were homeless, and in recent years there had been increases in the numbers of people sleeping rough and using the drug ‘spice’. At the same time, restructuring of homelessness services and the closure of several local hostels had resulted in some people being dispersed throughout the city in small temporary housing schemes, making it harder for CSS staff to engage with them and liaise with their key workers. According to the staff, funding cuts and insufficient resources had affected the number of outreach activities and amount of collaborative work that could be undertaken. Nonetheless, their expertise in this field had enabled them to adjust to changing needs and deliver an effective service. Furthermore, practice managers and other staff were involved with local NHS trusts and CCGs in the planning of local services for people who are homeless.

Specialist GPs

Specialist GPs provide GMS to the local population, and targeted services for patients who are homeless. This model was represented by two CSSs and 96 case study participants. Like Dedicated Centres, participants of Specialist GPs were significantly more likely than those of Mobile Teams and Usual Care GPs to use heroin or cocaine, inject drugs and be receiving OST. In contrast to Dedicated Centre participants, however, Specialist GP participants were more stably housed during the study. They spent more time in staffed accommodation and had fewer accommodation changes. SP2 participants were more unsettled than those of SP1; otherwise, there were no substantial differences in accommodation status between the two sites.

With regard to the Primary Outcome Score, there were no statistically significant differences between Specialist GPs, Dedicated Centres and Usual Care GPs. There were, however, noticeable differences between the two sites within the model: SP1 had the second highest Primary Outcome Score of all CSSs, whereas SP2 scored lowest. In relation to SHCs, Specialist GPs scored less than Dedicated Centres in achieving continuity of care for all four SHCs, but had higher scores than Mobile Teams and Usual Care GPs. SP1 scored considerably higher than SP2 regarding continuity of care for depression and drug problems, and reached levels comparable to those of Dedicated Centres.

For well-being outcomes, there were statistically significant positive improvements in both SF-8 PCS and SWEMWBS scores from baseline to 8 months for Specialist GP participants. Specialist GPs also scored favourably with regard to providing advice or help around nutrition, with levels comparable to those of Dedicated Centres. Specialist GP participants were more likely to rate their overall experience of the CSS and the care they received as ‘good’ or ‘very good’, compared with the general population’s rating of their GP service. There were no noticeable differences in ratings between SP1 and SP2. However, SP1 participants were more likely than SP2 participants to say it was easy to access a doctor or nurse. Overall, SP1 participants had more contacts with GPs and nurses at the CSS than SP2 participants, whereas the latter used A&E nearly three times more than SP1 participants (see Appendix 8, Table 61).

There were noticeable differences in service delivery factors between SP1 and SP2 (see Appendix 12, Table 68), which are likely to have contributed to variations in ease of access to doctors and nurses, and in health screening and continuity of care. SP1 operated in a way similar to that of Dedicated Centres, with designated nurses and other staff for patients who were homeless, drop-in clinics at the CSS and on-site substance misuse services. SP1 staff also conducted street outreach, ran several weekly clinics at homelessness services, worked closely with the local hospital and attended multidisciplinary case management meetings. SP2 had no staff working specifically with patients who were homeless, ran clinics only twice weekly in homelessness services, did not conduct street outreach work, and on-site substance misuse services were unavailable. Drop-in clinics at SP2 were available for all patients, with a restricted number of slots per day. However, SP2 participants reported difficulties getting a slot and long waits for a booked appointment. The more intense and integrated work provided by SP1 staff to patients who were homeless is likely to have contributed to its more favourable outcomes.

Contextual factors are also likely to have contributed to differences in performance between SP1 and SP2 (see Appendix 12, Table 68). Both CSSs were in cities with large populations of single people who were homeless. According to staff, in recent years there had been an increase in the number of people who were homeless and in the complexity of their health and substance misuse problems, with many using the drug ‘spice’. Restructuring or closure of several local hostels had resulted in residents being dispersed outside the city centre, making it harder for CSS staff to engage with them. Although some staff at both CSSs had considerable experience of working with people who were homeless, funding restrictions had limited their work, particularly for SP2 staff. According to the SP2 staff, insufficient resources prevented them from holding additional clinics in hostels and day centres, undertaking street outreach and working more collaboratively with other agencies. SP1 was involved with local healthcare commissioners and the local authority in planning services for people who are homeless, whereas this was less the case for SP2. The latter described lack of co-ordination between local authority and health services, and delays and lack of support from the CCG when trying to introduce new healthcare services for people who are homeless.

Mobile Teams

Mobile homeless health teams hold clinics in hostels or day centres for people who are homeless. In this study, the model was represented by two CSSs and 96 case study participants. The teams mainly consisted of nurse practitioners; neither had a GP. Patients were therefore encouraged to register with a local general practice, in addition to accessing Mobile Teams. Compared with other models, greater percentages of Mobile Team participants were born outside the UK, and did not have British citizenship or recourse to public funds. They were significantly less likely to report misuse of drugs, and reports of mental health problems were also lower. During the study, they spent longer than other groups sleeping on the streets or in makeshift shelters such as church halls, although they had fewer accommodation changes and spent more time in staffed accommodation than Dedicated Centre participants. MT1 participants were more likely to be British born and to have alcohol and drug problems. MT2 participants were more likely to be born outside the UK, to have no recourse to public funds and to be sleeping rough.

Compared with other models, the Mobile Team model had less favourable outcomes. It had a lower Primary Outcome Score, with a highly statistically significant (p = 0.001) screening deficit. Regarding SHCs, Mobile Teams received relatively high scores for outcomes 3 and 4 (explanations to patient of condition and treatment), but were least likely of all models to have maintained continuity of care for all SHCs apart from drug problems, for which Usual Care GPs scored lowest. When interventions by GPs were included, the Mobile Teams’ scores for continuity of care reached levels comparable to or above those of Usual Care GPs. In most instances, however, they remained considerably lower than those of Dedicated Centres and Specialist GPs. MT1 was more likely than MT2 to have maintained continuity of care with participants, which may reflect the greater percentage of MT2 participants who slept rough.

As with Dedicated Centres and Specialist GPs, Mobile Team participants were more likely to rate their overall experience of the CSS and the care they received as ‘good’ or ‘very good’, compared with the general population’s rating of their GP service and the care they received from nursing staff. They were more likely to say they had confidence and trust in the nurses. Unlike Dedicated Centres and Specialist GPs, however, there were no statistically significant improvements in participants’ well-being over time, and Mobile Teams were significantly less likely to have provided advice or help around nutrition.

Although Mobile Teams worked with a relatively large percentage of patients who were sleeping rough, given that mental health and drug misuse problems, which might contribute to poor engagement and disorganised behaviour, were less prevalent, their less favourable outcomes suggest that service delivery factors were attributable (see Appendix 12, Table 69). The teams mainly comprised specialist nurse practitioners with considerable experience of working with this patient group, but neither team contained a GP. Participants were therefore registered with a local general practice, and hence received primary health care from both Mobile Team nurses and a GP. In effect, the model was a ‘Mobile Team plus separate GP model’. The Mobile Teams worked closely with some GPs: some joint work was undertaken, and, in most cases, medical records were shared. They ran drop-in clinics in hostels and homelessness sector day centres, and much of their work concerned assessing health needs and linking patients to GPs or other agencies, rather than acute disease management. Although the mean number of nurse consultations during the study was considerably higher among Mobile Team participants than in all other models, the number of GP contacts was lower than that of participants of Dedicated Centres and Specialist GPs (see Appendix 10, Table 64).

The less favourable scores of the ‘Mobile Team plus separate GP’ model concerning health screening and continuity of care for long-term health conditions suggest poor co-ordination between the two services. The model involved health care being delivered by multiple primary health care providers. For both Mobile Teams, their patients were not registered at a single GP, and hence CSS nurses had to forge links with several GPs. Moreover, Mobile Team participants had the option to access health care from CSS nurses and/or GPs, which could lead to uncertainty and confusion among patients, particularly as the availability of health care by Mobile Team nurses was limited. As explained by a PPI group member in a discussion of the study’s findings, the accessibility of Mobile Team nurses at hostels and day centres is restricted to specific days and times, and is reliant on the homelessness service. Contextually, funding restrictions placed on some homelessness sector services by local authorities had, for example, resulted in the opening hours of some day centres being reduced. In contrast, Dedicated Centres and Specialist GPs operated from a ‘fixed’ site that was accessible throughout the day and on most days. Such services are likely to be more suitable for patients who are homeless, move around a great deal and find it hard to keep to a schedule.

The services offered by the ‘Mobile Team plus separate GP’ model are comparable, in some ways, to those of the Specialist GP model, particularly SP1: namely they operated in urban areas with many people who were homeless, health care was provided by GPs and nurses, and designated staff reached out to patients in hostels and day centres. However, the Specialist GP model performed better in terms of study outcomes. One important difference is that health care by Specialist GPs was delivered by GPs and nurses from the same practice, and patients were registered with a single primary health care provider. This is more likely to have enhanced collaborative working among staff and aided patients’ understanding of where to seek health care. Moreover, the overall service use costs of Mobile Team participants were similar to those of the Specialist GP participants.

Usual Care GPs

Usual Care GPs provide GMS to the local population, which, by default, includes people who are homeless. They do not offer special or targeted services to patients who are homeless. In this study, the Usual Care GP model was represented by four mainstream general practices and 75 case study participants. Usual Care GP participants tended to be more settled than those of the other three models. Most were in staffed accommodation at baseline, with few or no accommodation changes during the study. Many had mental health problems, but they were less likely than Dedicated Centre and Specialist GP participants to use heroin or cocaine or inject drugs. UC1 participants tended to be less settled than those of UC2 and UC3; they were also more likely to use Class A drugs, they changed accommodation more often during the study and they spent considerably less time in staffed accommodation (see Table 12).

Compared with other models, the Usual Care GP model scored relatively low for some outcomes. However, it had the second highest Primary Outcome Score, achieving the highest of all models for the screening of BMI, alcohol use and smoking (see Table 14). There were, however, differences in scores by individual CSSs. UC2 and UC3 used a computer-based ‘homeless template’ to screen patients who were homeless and had a significantly higher Primary Outcome Score than UC1 (see Appendix 11, Table 65). Compared with the three specialist models, Usual Care GPs were less likely to have detected alcohol and drug problems among their participants, and were less likely than Dedicated Centres and Specialist GPs to have achieved continuity of care for participants identified as having depression, alcohol problems and/or drug problems. The continuity of care score for drug problems was exceptionally low at all Usual Care GP sites. There were no statistically significant differences in outcomes for chronic respiratory problems between Usual Care GPs and the other three models.

Staff of the Usual Care GP model were less likely than those of other models to have discussed nutrition with participants, despite a significantly higher proportion of their patients reporting problems at baseline with getting meals or eating healthily (see Table 28). Participants’ ratings of their Usual Care GP and the quality of care they received were significantly lower than those of the participants of the other three models and of the general population’s ratings of their GP services. However, ratings differed between individual CSSs, with UC2 scoring slightly higher. This may be because UC2 staff had a relatively close working relationship with some hostels. During the study, Usual Care GP participants had considerably less contact with GPs than Dedicated Centre and Specialist GP participants (see Appendix 10, Table 64), and their overall service use costs were considerably lower than those of other models (see Table 41).

Usual Care GPs operated very differently to the other three models, and service delivery and contextual factors are likely to have been crucial in contributing to their relatively low performance for some outcomes (see Appendix 12, Table 70). Their practice list sizes were large, while the proportion of patients who were homeless was small. They had no dedicated staff or targeted services for patients who were homeless, no drop-in clinics at the CSSs or in hostels or day centres, and very little contact with outreach teams. Indeed, some CSS staff had little knowledge of homelessness services in their area. UC2 and UC3 received some additional NHS funding for their work with patients who were homeless, and UC1 received some additional NHS funding for its work with patients who have substance misuse problems. According to the CSS staff, however, this was insufficient to allow more proactive and integrated work for patients who were homeless, such as providing drop-in sessions or collaborating closely with hostels, day centres and street outreach teams.

The Usual Care GPs were in cities with relatively large populations of single people who were homeless, and in areas with several hostels and other temporary accommodation. According to CSS staff, in recent years there had been an increase in the number of people who were homeless, although very few of their patients were sleeping rough. UC2 and UC3 were a few miles from specialist primary health care services for people who are homeless, so it is likely that those sleeping rough were drawn to areas where specialist services were located. Given that the Usual Care GP model performed relatively low for some outcomes, this strongly suggests that the model in its current form would struggle to provide quality health care to a considerable number of patients who are homeless, particularly if they had complex needs. Furthermore, the low satisfaction scores of the CSS given by participants, and the low rate of GP use, compared with specialist models, suggest that people who are homeless and vulnerable find it harder to access care from mainstream general practices. However, the positive health screening scores for UC2 and UC3, which used a ‘homeless template’, and the higher service satisfaction ratings for UC2 suggest that more can be done by mainstream general practices to accommodate patients who are homeless.

Summary

This chapter has examined ways in which contextual factors and mechanisms of health care delivery are likely to have had an influence on outcomes for each Health Service Model, and highlighted differences between some CSSs within the same model. Common contextual factors on service delivery were apparent across all Health Service Models. For example, all had experienced an increase in the number of single people who were homeless in their locality. As mentioned in Chapter 1, cuts to housing, health and social support budgets, a shortage of affordable housing, and welfare benefit changes and sanctions have contributed to this rise. CSS staff also reported an increase in the complexity of health and substance misuse problems among this population, such as an exacerbation of their use of the drug ‘spice’, which can contribute to serious health problems. In addition, in many areas, CSS staff have found it harder to engage with people who are homeless because of changes to homelessness service provision, such as the closure of hostels, resulting in the dispersal of people away from city centres (and the CSSs), and a reduction in the opening hours of day centres. At the same time, funding cuts, restrictions and insufficient resources have limited the services the CSSs are able to provide. At the Dedicated Centres and SP1, where more comprehensive and integrated services were available, outcomes for the study participants were more favourable.

Overview of findings

The framework for this study was informed by Gelberg et al. ’s80 behavioural model of health service use by people who are homeless. The model has three domains: (1) population characteristics, such as demographics, personal and family resources, community resources and perceived health needs; (2) health behaviour, such as lifestyle factors and use of health services; and (3) outcomes, such as satisfaction with care, and the availability and accessibility of health services (see Chapter 3). Our study similarly found that many factors were influential in determining outcomes, namely contextual factors, service delivery mechanisms and population characteristics (see Chapter 14). More favourable outcomes were achieved if models were well integrated with other health and social care providers and homelessness sector services. For example, primary health care teams were more likely to have maintained contact with participants who had drug and alcohol problems if on-site substance misuse services were available. Where this was not the case, some CSS staff reported being unaware of whether or not their patients were attending appointments with substance misuse workers, as the latter sometimes failed to update GPs on patients’ progress. There was a positive link between overall service use costs during the study period (incurred by more service use) and spending more time in staffed accommodation. This suggests that hostel staff were instrumental in encouraging residents to use health services; helping them to book appointments; and reminding them of, and accompanying them to, appointments, if necessary.

Outcomes for Dedicated Centres and SP1 were relatively favourable. They had staff working exclusively with patients who were homeless, held regular drop-in clinics, conducted outreach work, and were well-integrated with hospitals. These are all factors likely to have encouraged people who were homeless to engage with primary health care services, which supports the findings of similar studies in the USA (see Chapter 2).

With no GP in the Mobile Teams, patients received health care from both Mobile Team nurses and local GPs. The less favourable scores associated with this arrangement for health screening and continuity of care for health conditions suggest poor co-ordination between the services. Although the mean number of nurse consultations was considerably larger for this model, the number of GP contacts was smaller than those of Dedicated Centres and Specialist GPs. The mapping exercise identified a Mobile Team that employed GPs as part of the team and provided GP registration. 112 The effectiveness of this arrangement in relation to the HEARTH study’s outcomes is unknown.

Usual Care GPs operated very differently to other models, and service delivery factors are likely to have been crucial in contributing to their relatively low performance for some outcomes. Their practice list sizes were large, they had no dedicated staff and no targeted or outreach services for patients who were homeless, and they were not well-integrated with homelessness services. However, favourable scores for health screening at two sites, and higher satisfaction ratings at one site, suggest that some positive aspects of health care delivery to patients who are homeless can be achieved by mainstream general practices, given the right circumstances. Although, in this study, the Usual Care GP sites received additional NHS funding for their work with this population, according to CSS staff this was insufficient to allow more proactive and integrated work.

Across all models, the number of contacts with GPs over the study period was considerably higher among the study participants than the general population (see Chapter 13). There was also high use of out-of-hours services. The number of such contacts positively correlated with the number of GP and nurse contacts, indicating that out-of-hours services were not necessarily a substitute for GP or nurse consultations. Regression analyses found that out-of-hours use was associated with more changes of accommodation during the study period. Total cost of service use was also linked positively to number of accommodation changes.

Four areas of concern affected all models. The first related to health screening. This was most common, but not universal, for mental health, alcohol use and smoking. Less than half of participants had their BMI recorded during the study, and relatively few were screened for hepatitis A or TB. Variations in screening were partly related to different working practices. Some CSSs did not routinely screen for TB or hepatitis A (see Chapter 7). However, screening rates for some HSIs were higher for participants of all HEARTH models than for the general population. In Scotland, a review of general practice medical records found that just 31% of adult patients had had BMI recorded in the previous 2 years. 150 This compares to 45.8% of HEARTH study participants in the previous 12 months. Likewise, 51.9% of UK adults registered with GPs in January 2018 and followed up for at least 1 year had a code in their medical records related to alcohol use in the previous 5 years. 151 In comparison, 79.4% of HEARTH study participants had had alcohol use documented in the previous 12 months.

Second, several participants reported chronic respiratory problems or depression at baseline, completing instruments that indicated severe problems, yet these were not documented in the medical records. This applied to all Health Service Models, suggesting a failure, at times, by primary health care staff to identify or record these problems. Although the PHQ-9 formed part of the ‘homeless template’ at UC3, it appeared no further action was taken when scores indicated depression. Both chronic respiratory problems and depression are common in the UK. More than 6 million people are reported to have asthma or COPD,152 and more than one-third of GP visits are estimated to involve a mental health component. 153 Recognising depression in primary health care settings can be challenging, as some patients who are depressed report only non-specific symptoms or somatic complaints, or share concerns about depression only at the end of a GP consultation. 153,154 Chronic respiratory problems and depression are common among people who sleep on the streets, and are exacerbated by their living conditions (see Chapter 1). One of the HEARTH study’s PPI team members suggested that these problems may not be given highest priority by patients or GPs during a consultation if there are more acute and pressing health needs. The PPI member believed that GPs were ‘powerless to tackle such problems’ until the living conditions of people who are homeless improve.

Third, across all Health Service Models, poor oral health was common: many participants did not seek dental care, and dental pain and other dental needs were unaddressed. Compared with the general population, dental anxiety and dental phobia were more common among HEARTH study participants. Dental services specifically for people who are homeless or vulnerable were available at or near several CSSs, but many participants did not access these. This suggests that dental services were not effectively engaging with people who were homeless, and primary health care teams were failing to screen for oral health problems and to link participants to local dental care providers.

Finally, mental health problems are common among single people who are homeless. Yet, as reported by many staff and external agencies from all CSSs except UC3, the availability of mental health treatment services was poor in their area (see Chapter 11). It included long waits for people to be assessed and start treatment, insufficient services for people with mild to moderate illness, long waits or barriers to services for people with combined mental health and substance misuse problems, and lack of community mental health nurses and hospital beds.

Implications for NHS commissioners

The reporting of this study’s findings comes at a time of considerable change within the NHS (described in Chapter 2). ICSs were established in July 2022, and have four strategic aims: (i) to improve population health and health care; (ii) to tackle unequal outcomes and access; (iii) to enhance productivity and value for money; and (iv) to help the NHS to support broader social and economic development. Likewise,155 Primary Care Networks build on the core work of current primary health care services with an emphasis on more co-ordinated and integrated health care services. Their focus for 2022/23 includes tackling health inequalities in primary health care delivery, supporting better patient outcomes through proactive primary care, and supporting improved patient access to primary care services. 156 A 2021 study in England of ED attendees suggested that some hospital admissions among people who were homeless may have been preventable with improved access to primary health care. 157 In March 2022, NICE produced guidance on integrated health and social care for people experiencing homelessness. 158 This section describes implications of our study’s findings for NHS commissioners.

• In all NHS regions, and particularly in areas without specialist primary health care services, the mapping exercise revealed difficulties accessing primary health care for people who were homeless, which need to be addressed. Health and Wellbeing Boards have a statutory duty, with local ICBs, to produce a Joint Strategic Needs Assessment, which identifies health needs in the locality (see Chapter 2). Such an assessment should include the scale and nature of homelessness in an area over time; the characteristics and needs of people who are homeless; and the availability, accessibility and performance of primary health care provision for this patient group. Furthermore, people’s health needs are closely intertwined with their housing and support needs, and the solutions require the involvement of the NHS, local authorities and community sectors. Our proposals are reiterated by NICE in its 2022 guidance: (1) commissioners of health, social care and housing services and other sectors should work together to plan and fund integrated health and social care services for people experiencing homelessness; and (2) local homelessness health and social care needs assessments should be conducted, which include quantifying and characterising the population experiencing or at risk of homelessness, and assessing the quality, capacity and accessibility of existing service provision. 158

• In locations where the health needs of people who are homeless are not being met, NHS commissioners need to consider what models of primary health care services are most appropriate. The HEARTH study found that participants in Dedicated Centres had the highest service use, and outcomes were consistently among the best. However, this model may be financially viable only in locations with a considerable number of people who are homeless or have other vulnerabilities, such as asylum seekers. The Specialist GP model was found to perform similarly to Dedicated Centres when targeted services were in place. Most specialist primary health care services that have been established since 2010 for people experiencing homelessness involve mainstream general practices that are funded to provide enhanced services to this patient group.

• Questions arise as to the function of Mobile Teams and their collaboration with GPs, and whether a more effective service could be delivered if they operated as part of a general practice, rather than as a separate service. The feasibility of such arrangements requires further exploration. It would mean patients would be registered with a single primary health care provider and have a fixed site from which to obtain health care. There would likely be improved collaboration between GPs, nurses and other practice staff, which undoubtedly would have a positive effect on patient care. Mobile Teams often cover a wide geographical area and maintain contact with people who are unsettled and move around. They sometimes see people who are reluctant to attend general practices, or are without GP registration or are registered with a GP in another location. Hence, it would be important that the outreach element of their work continued. Ways of achieving this if Mobile Team nurses were based at a local general practice need investigation. In terms of costs, the HEARTH study found that the overall service use costs of Mobile Team participants were similar to those of Specialist GP participants.

• Usual Care GPs are the main primary health care provider for many people who are homeless, particularly outside large cities. Their relatively poor performance with regard to continuity of care and patient satisfaction, and poor integration with homelessness services, raises questions about their suitability for managing patients who are homeless, and if or when commissioners should consider introducing additional support. A GP may be able to adequately provide health care to a local hostel if residents are relatively settled, but may struggle to deliver a service to a large number of people with complex needs and unsettled lifestyles. Commissioners should be guided by effective homelessness health needs assessments as to when a more enhanced service, such as that provided by a Specialist GP model, is required.

• Across all Health Service Models, CSS staff reported funding cuts or insufficient resources, which limited the extent and intensity of care they could provide. Where more flexible and comprehensive services were available, such as drop-in clinics, longer GP appointments and outreach into hostels and day centres, outcomes were more favourable. The benefits of these interventions should be taken into consideration when commissioning services. NICE also recommended that commissioners plan and design services for people experiencing homelessness that encourage engagement with health care, such as reduced caseloads, lengthening contact time with patients, low-threshold services and outreach work. 158

• Oral diseases are largely preventable,159 yet HEARTH study participants experienced considerable oral health neglect and poor access to dental services. Most local NHS dental services were separate and had little integration or established formal networks with CSSs, making it difficult to support participants’ access to dental care and oral health promotion. This is particularly important for people with substance misuse issues, who appear to experience increased oral disability and dysfunction. Oral health care should be integral to health and social care commissioned for people experiencing homelessness. Furthermore, dental services need to be commissioned in a way that they are fully embedded with health and social care, enabling networking and more responsive and accessible dental services. Different configurations of dental care delivery need to be explored. The Faculty for Homeless and Inclusion Health70 produced detailed guidance on standards for dental services that should inform integrated commissioning.

• Consideration should be given to the poor availability of mental health services and the impact this is having on both the health of patients and the work of primary health care providers. The NHS Long Term Plan renewed a commitment that mental health services will grow faster than the overall NHS budget, with a ringfenced investment worth at least £2.3B a year for mental health services by 2023/24. 160

• The inclusion of effective monitoring and evaluation of services in the commissioning process is critical. According to a report published by the RCGP,161 commissioners should be able to determine which services are working effectively, and those that are not meeting their objectives. The Faculty for Homeless and Inclusion Health proposed locally designed key performance indicators to cover health screening, and access to treatment, and the management of long-term health conditions to QOF standards. 70 NICE also recommended that commissioners should define and measure health and social outcomes and service use when developing services for people experiencing homelessness. 158

Implications for primary health care service managers and practitioners

Drawing on evidence from the four models and from individual CSSs, this section describes the implications of our study’s findings for primary health care service managers and practitioners.

• There needs to be improved health screening for people who are homeless, and it is essential that this leads to an intervention when indicated. The Faculty for Homeless and Inclusion Health recommended that primary health care services working with this patient group should offer screening and access to treatment for mental health, alcohol and drug problems, smoking, and TB and blood-borne viruses, and should offer vaccination against hepatitis A and B, particularly for people who inject drugs or have hepatitis C. 70 NICE further recommended active TB case-finding among people who are homeless and sleeping rough or in hostels,119 and that alcohol screening should be routine practice in primary care. 162 PHE and the RCGP recommended hepatitis A vaccinations for people who inject drugs or have hepatitis C or chronic liver disease. 117,163 Decisions need to be reached between primary health care service managers and clinical staff about screening for particular health conditions, taking into account national guidelines and service contract agreements.

• Two Usual Care GPs that used a computer-based ‘homeless template’ to assess the needs of patients who were homeless had higher Primary Outcome Scores. This suggests that there are benefits of using a ‘screening plus intervention template’ in all mainstream general practices with patients who are homeless, and in specialist models without a current screening process. It is important that ‘alerts’ are placed on medical records as reminders of follow-up actions or further screening. Such a template should be disseminated widely by organisations such as the RCGP and the Royal College of Nursing to ensure that mainstream primary health care practitioners are reached. Pathway has developed a digital health screening template, which is available free of charge to health care providers using the EMIS Web clinical system (EMIS Health, Leeds, UK). It is very detailed, however, and the practicalities of its usage, particularly in mainstream general practices, need further investigation. A 2019 systematic review of health assessments and screening tools administered to people who were homeless identified no validated tools that assessed chronic conditions, nutrition, functional decline and frailty, or vision and hearing. 164

• Given the large number of cases of depression and chronic respiratory problems reported by participants but not documented in their medical records, awareness needs to be raised by organisations such as the RCGP and Faculty for Homeless and Inclusion Health of the links between these conditions and homelessness, and the need for their inclusion in the screening of patients. Scheduled or opportunistic reviews need to be undertaken among patients who are homeless to detect these conditions and initiate treatment if indicated.

• Several working practices were beneficial in enabling staff to engage with patients who were homeless and retain them in health care. These included easily accessible and flexible services; designated staff for patients who are homeless; close working with mental health, substance misuse and homelessness sector agencies; and outreach work. Consideration should therefore be given as to how these features can be built into existing primary health care services where applicable.

• The introduction of a full- or part-time role within mainstream general practices, whereby a clinical staff member is responsible for patients who are homeless, would enable more targeted work to be undertaken with this patient group, links to be forged with other health practitioners and homelessness sector staff, and engagement in local initiatives and strategic work concerning this patient group. Additional training for that person would be required. NICE similarly proposed ‘homelessness leads’ in mainstream services to co-ordinate care for patients who are homeless. 158 UC1 also acknowledged the likely benefits of this (see Chapter 11).

• The running of regular health clinics in hostels and day centres needs to be reviewed. They are invaluable in certain circumstances, such as encouraging people who are homeless to register with and attend a general practice, or delivering basic health care if a person declines to access a general practice. As described by a nurse from a Specialist GP who ran health clinics in a day centre, the arrangement worked well ‘as an outpost of the GP ... and as a conduit into primary care’. It is questionable, however, whether such clinics should be available to all hostel residents or day centre users, or whether it deters use of general practices. Furthermore, the health care that can be provided in such settings is limited, and some settings lack suitable facilities for clinical work. In our study, CSS staff and external agencies had mixed views about the role and frequency of outreach clinics by health workers (see Chapter 11).

• People who are homeless have multiple needs, and an all-inclusive, holistic approach to their care is required. In the HEARTH study, more frequent changes of accommodation were associated with increased use of out-of-hours services. Attention should be given to the housing, social care and welfare needs of people who are homeless when health assessments are undertaken, and multiagency working between health, housing and support services is essential. Working relationships with mental health and substance misuse services could be improved in some primary care settings, including providing regular updates to GPs of patients’ progress. Having the staff capacity to attend multiagency case management meetings is also crucial.

• There needs to be greater awareness of oral health among health and social care teams working with people experiencing homelessness. Practitioners should proactively ask patients about dental problems, and provide help and advice about maintaining oral health and accessing local dental providers. It is essential that efforts to promote dental health and access to dental care should be integrated and embedded into local health and other services for people experiencing homelessness.

Further research

This study examined a few selected health screening activities undertaken by primary health care staff, and data about physical and mental health conditions pertaining to the SHCs were extracted from medical records. Given the wealth of information collected during the study and the availability of comprehensive medical records for 12 months, there is the potential for much more interrogation of the data. This could include analyses of (1) other screening activities, such as breast and cervical cancer screening, and influenza immunisations; (2) the prevalence and treatment over time of illnesses, such as cardiovascular disease, gastrointestinal problems, hepatitis C, HIV and psychosis; and (3) hospital admissions and use of out-of-hours services, including reasons and outcomes.

The study has demonstrated that participant characteristics, service delivery mechanisms and contextual factors were influential in determining outcomes, and that these differed between and within models. There is the potential to explore further these relationships with attention to participant characteristics, such as age, sex, ethnicity, disability, multiple health conditions and frequent A&E attenders or users of other out-of-hours services.

This chapter has described several aspects of the delivery of primary health care to people who are homeless that need further review. These include the function of Mobile Teams and the feasibility of such nursing staff operating from general practices, the advantages and disadvantages of holding regular health clinics in hostels and day centres, the suitability of individual mainstream general practices to care for patients who are homeless, and the introduction of a ‘screening and intervention’ template and homelessness lead in general practices for this patient group. Further research is also needed into the delivery of primary health care to people who are homeless since the start of the COVID-19 pandemic, and its impact on practices such as ‘drop-in’ consultations with GPs or nurses, on-site mental health and substance misuse services, and administration of OST by pharmacists.

This study is a starting point for further research into aspects of primary health care provision for people who are homeless, namely (1) methods of integration between primary care GPs and nurses, and mental health and substance misuse services, and arrangements that are most effective; (2) the effectiveness of various approaches for engagement and retention in primary health care services; and (3) measures and instruments to capture the provision of primary health care to this patient group.

Strengths and limitations of the study

To our knowledge, this is the first UK study to evaluate and compare different models of primary health care provision for people who are homeless. Most previous such studies were conducted in the USA and focused on one aspect of service delivery or a single intervention, rather than on a model or service in its entirety (see Chapter 2). Among the unique features of the HEARTH study are the inclusion of a ‘usual care’ model for comparison, analyses of different aspects of health care delivery, and analysis of service use and costs by participants over 12 months.

Information for the study came from many sources, enabling triangulation of data. Longitudinal data over 12 months were obtained through three interviews with case study participants. Recruitment of participants reached the target number of 96 at each of the three specialist models, and 75 participants (78% of the required number) were recruited at Usual Care GP sites. Follow-up interviews were achieved with 272 of the 363 participants (74.9%) at 4 months, and with 263 (72.5%) at 8 months. The latter was slightly higher than the expected number at 8 months after allowing for attrition (n = 256), despite fewer people being recruited to the study. Full medical records held by the CSSs for the entire study period were collected for each participant, covering service use for more than 10 months (and up to 12 months) for 81.9% of participants (see Table 4).

From a statistical perspective, the study proceeded well. A sample size (size = 5%, power = 80%) of at least 64 participants in each of the four Health Service Models was required to detect a 1-point difference in the six-item primary outcome tool between any two models. This was assuming a worst-case scenario of the tool having a SD of 2. In fact, the SD was only 1.24 (see Table 14), which retrospectively required a minimum of 26 participants per model. This was comfortably achieved for each model. Indeed, retrospective calculations indicate that 95% power was exceeded for comparing any two models, affirming the significantly lower Primary Outcome Score reported for Mobile Teams than for each of the three other models.

With regard to assessing the efficacy of the models in relation to SHCs, as anticipated in the protocol, a scarcity of cases did occur, but only for self-reported hypertension. The other four conditions were able to be compared soundly, eliminating any need to construct a valid overall combination of conditions that bore varying levels of complexity.

The successful retention of participants in the study enabled incisive comparisons between Health Service Models of many aspects of participants’ health and well-being, and the care they received. This facilitated extensive regression modelling of the primary outcome and several secondary outcomes to identify various factors associated with each outcome, and, most importantly, to adjust for such factors and gauge more precisely the impact of each of the four Health Service Models.

Service use data were obtained assiduously with a high level of detail, which contributed to in-depth analyses of health service staff commitments, and a thorough breakdown of costs for the four Health Service Models. Specific cost groupings comprised primary and community care, social care, out-of-hours services, hospital admissions and outpatient services, tests and investigations, and allied health professionals (e.g. dentists, opticians), with further breakdowns within each. The cost of out-of-hours services was of relevance, given recent findings of the high rates of A&E visits among people who are homeless, compared with the general population. 12

There were limitations to the study, which took much longer than planned. One of the main difficulties was recruiting mainstream general practices with enough patients who were homeless for the Usual Care GP model. Substantial time and effort were spent liaising with general practices and with primary care leads of CRNs, and fewer participants than intended were recruited for this model. At UC4, just two patients were recruited; therefore, this site had to be omitted from many of the analyses. Moreover, several mainstream general practices that fitted the study criteria declined to take part. Likewise, the selection of specialist sites depended on their willingness to participate, the number of patients who were homeless, and staff resources or imminent changes to the service. In addition, the recruitment of case study participants relied on their willingness to participate. As demonstrated in this study, there are many differences between models, and between CSSs within a model. The findings of this study pertain to the CSSs and their patients who participated, and may not necessarily apply to the performance of other primary health care services for this patient group.

Given the innovative nature of this study, various measures were used for the first time to assess the performance of the CSSs. Screening for the primary outcome and the management of SHCs did not rely on validated tools for scoring (as none could be found). Instead they depended on the expertise of the research team and other clinicians, and recommendations by the Faculty for Homeless and Inclusion Health in its standards for primary care. 70 As described in Chapters 7 and 8, scoring was intricate and lengthy, and relied on information from case study participants and medical records; many challenges occurred during the process. Various ‘rules’ were adopted by the research team after preparation of several working papers and agreement with the SSC. This included allocating positive scores when pre-coded text with no further details was used in the medical records, when CSS staff ‘signposted’ patients to other services, and when a service was offered by the CSS but declined by the patient. Undoubtedly, such rules had an influence on the study’s outcomes. As mentioned previously, further development is needed into measures and instruments to assess the performance of primary health care services for patients who are homeless.

Extracting data from medical records was very time-consuming, partly because the quality and clarity of the records varied greatly. Some clearly stated health conditions and treatments, but some were pre-coded to such an extent that it was not straightforward to make judgements. Medical records for the 12-month study period were also very extensive for some participants (one set of records comprised 283 pages), with numerous consultations, missed appointments and A&E attendances. To avoid error or oversight, each set of medical records was searched several times to retrieve data for different outcomes, and checks were made by a second, and sometimes third, coder. Other limitations to the study, including the unavailability of some medical records and the variability of in-house services at the CSSs, are discussed in Chapter 13.

Summary

To our knowledge, this is the first UK study to compare and evaluate different models of primary health care provision for people who are homeless. New and much-needed evidence has been generated about the effectiveness of different ways of delivering health care to this patient group. Contact was maintained for the entire study period with the majority of case study participants, thus enabling the study’s aims and objectives to be met. Innovative approaches were employed to evaluate outcomes and produce findings that are of importance to NHS commissioners, primary health care service managers and practitioners.

Many of the HEARTH study’s findings corroborate data from the USA (see Chapter 2), in which tailored services, nurse case management, integrated care provision and outreach on the streets and at homelessness services had positive health outcomes for people who were homeless in terms of improving screening uptake, encouraging the use of primary health care services, treating health conditions and engaging people in specialist care.

Contributions of others to the study

We send special thanks to Sarah Coward, who managed the fieldwork and data collection at some CSSs and conducted many of the interviews in these areas. She assisted with the coding of data and attended SSC meetings. We appreciate the help given by Ruby Fernandez-Fu with the mapping exercise. We thank Janet Robinson, Robert Calder and Carl Purcell, who assisted with data searching and coding; Nicole Steils, who helped with the literature review; and Jess Harris and Caroline Norrie, who conducted a few interviews with stakeholders. Thanks are also sent to Morro Touray for his assistance with the health economic analysis, and to Laura Cole and Nicole Steils for taking minutes at the SSC meetings. We also appreciate the advice given by Dr Patrick White, Reader of Primary Care Respiratory Medicine, KCL, and the medical professionals who were contacted by Dr Chris Ford.

We are grateful to the contributions throughout the study of members of the PPI group who assisted in the study design, study implementation, participant interviews and the interpretation of findings, and were involved in the SSC meetings. Special thanks are sent to Gordon Chaston, who, with another member of the PPI team, undertook some interviews and willingly accompanied the research team to different locations, which sometimes involved overnight stays. They remained extremely enthusiastic about the study even when there were substantial delays to the timetable.

Special thanks are sent to Homeless Link, which provided databases of accommodation and day centre services in England. These were very useful as a starting point for the survey of homelessness services. We are grateful for the input of Vic Rayner, formerly of Sitra and former co-investigator in the study, who assisted in its early months; of Burcu Borysik, formerly of Sitra and Homeless Link; and of other staff at Homeless Link who assisted with the PPI work. We appreciate the involvement of St Mungo’s apprenticeship scheme in the PPI work. An agreement was drawn up between OptumInsight Life Sciences Inc and King’s College London for use of the SF-8 in the study (Licence number QM033370). Written permissions were also granted by the University of Warwick for use of the SWEMWBS scale, and by the CAT Governance Board of GlaxoSmithKline for use of the COPD Assessment Test.

Finally, thanks are sent to members of the NIHR CRNs who helped with recruitment of general practices for the Usual Care GP model, and to our lead CRN staff member who undertook the extraction of data from medical records at one CSS.

Study Steering Committee membership

The research team appreciate the guidance and advice provided by members of the SSC:

• Liddy Goyder, Professor of Public Health, School of Health and Related Research, University of Sheffield (chairperson)

• Jennifer Beecham, Professor of Health and Social Care Economics, Personal Social Services Research Unit, University of Kent

• Caroline Bernard, Head of Communications and Advocacy, Homeless Link

• Andrew Casey, Health Strategy Manager, St Mungo’s

• Mohammed Ismail, Research Director, Analytical Research Ltd

• Gill Leng, Health and Homelessness Adviser, PHE

• Jeremy Porteus, Managing Director, Housing, Learning and Improvement Network

• Rebecca Rosen, GP and Senior Fellow, Health Policy, Nuffield Trust

• Sara Shaw, Senior Researcher, Nuffield Department of Primary Health Care Services, University of Oxford.

Contributions of authors

Maureen Crane (https://orcid.org/0000-0002-7364-7855) (former Principal Research Fellow, now Visiting Senior Research Fellow) was the principal investigator; was involved in the study design, data collection, analyses and interpretation of findings; and co-authored the report.

Louise Joly (https://orcid.org/0000-0002-1615-2987) (former Research Fellow, now Visiting Research Fellow) was a co-investigator; was involved in the study design, data collection, analyses and interpretation of findings; was responsible for PPI and organisation of the collection of participant medical records; and co-authored the report.

Blánaid JM Daly (https://orcid.org/0000-0001-7748-6940) (former Senior Lecturer Special Care Dentistry at KCL, now Professor of Special Care Dentistry, Trinity College Dublin) was a co-investigator; was responsible for the dental features of the study, dental analyses and reporting; wrote Chapter 10; and contributed important intellectual content to revisions of the final report.

Heather Gage (https://orcid.org/0000-0002-2049-9406) (Professor of Health Economics) was a co-investigator, was involved in the study design, was responsible for the economic aspects of the study, undertook analyses of service use and costs, wrote Chapter 13 and contributed important intellectual content to revisions of the final report.

Jill Manthorpe (https://orcid.org/0000-0001-9006-1410) (Professor of Social Work) was a co-investigator; contributed to study design and management, including budget responsibility; undertook policy review; co-wrote Chapter 2; and contributed important intellectual content to revisions of the final report.

Gaia Cetrano (https://orcid.org/0000-0003-1532-0480) (former Research Associate, KCL; now Social Work Lecturer, Middlesex University) was involved in data collection, coding and the study’s implementation; assisted with PPI and the literature review on evidence-based practice; co-wrote Chapter 2; and contributed important intellectual content to revisions of the final report.

Chris Ford (https://orcid.org/0000-0001-5842-8518) GP for more than 30 years before retirement, working with people who used drugs and/or alcohol and had multiple health conditions. With colleagues, she founded Substance Misuse Management in General Practice, which is a successful support network for doctors and other health professionals working with people who have alcohol and drug problems; started the RCGP Management of Drugs and Alcohol Problems in Primary Care conference (now in its 26th year); and was a key member in starting the RCGP Certificate in Drug Dependence. Chris was involved in study design; provided clinical expertise throughout the study; reviewed cases for inclusion in study outcomes; advised on measures to detect changes in alcohol intake and drug use; and contributed important intellectual content to the revisions of the final report.

Peter Williams (https://orcid.org/0000-0001-7885-0564) (Statistics Consultant) was involved in the study design, was responsible for the statistical analyses and production of statistical models and outputs, drafted statistical sections of the report and contributed important intellectual content to the revisions of the final report.

Publications

Crane M, Cetrano G, Joly L, Coward S, Daly B, Ford C. et al. Mapping of Specialist Primary Health Care Services in England for People who are Homeless. London: NIHR Health and Social Care Workforce Research Unit, King’s College London; 2018.

Crane M, Cetrano G, Joly L, Coward S, Daly B, Ford C. et al. Mapping of Specialist Primary Health Care Services in England for People who are Homeless: Summary of Findings and Consideration for Health Service Commissioners and Providers. London: NIHR Health and Social Care Workforce Research Unit, King’s College London; 2018.

Crane M, Cetrano G, Joly L, Coward S, Daly B, Ford C. et al. Inventory of Specialist Primary Health Care Services in England for People who are Homeless. London: NIHR Health and Social Care Workforce Research Unit, King’s College London; 2018.

Joly L. Involving People with Experience of Homelessness as Peer Interviewers. Southampton: NIHR; 2019.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the HSDR programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, the HSDR programme or the Department of Health and Social Care.