Introduction and context

Box 1 provides a summary of Chapter 1 of this report.

Summary of key points

• The social care system in England is under significant pressure, and there are funding gaps and workforce challenges that make it difficult to keep up with rising demands. The exit of the UK from the EU, and the COVID-19 pandemic, have created additional pressures.
• Technology has been suggested as one way to improve care and address pressures in the health and social care system. It was specifically highlighted as a priority within recent social care reforms as a way to support people living independently within their homes.
• Sensor-based technology with artificial intelligence capabilities is one type of technology that may be useful in some contexts. There is evidence to suggest that this type of technology can potentially improve some aspects of care and care planning, although there are key gaps in evidence that need to be addressed.
• The uptake and sustainability of technology is influenced by wider factors, including implementation processes, support available for front-line staff, and engagement with people who draw on care and support and their carers.
• This study is based on scoping work identifying a key gap in evaluations of artificial intelligence in social care. It focuses on decision-making and implementation processes.

Social care in England

‘Social care’ is a term which refers to the practical assistance which is given to people with disabilities, people with mental health problems, people with learning difficulties, frail older people and unpaid carers to manage activities of daily living (getting washed and dressed, eating, going to the toilet, etc.) as well as support to have choice and control over their own lives. This can take the form of direct services (home care, day care, residential care, etc.), or a direct payment (where the person receives funding with which to design their own care and support, potentially employing their own personal assistants). It is funded and organised by local councils (known as ‘local authorities’ in England, sometimes known as ‘municipalities’ internationally), as well as by people who are funding and/or organising their own care, and direct services are delivered by a mixed economy of private, voluntary and public sector services.

Even prior to COVID-19, the social care system in England was under significant pressure. There is an increasing demand for support, with a total of over 1.9 million requests for adult social care in 2019/2020, a number which has increased by 120,000 since 2015–6.1 A growing number of older people nationally, as well as increases in working-age adults with care and support needs, has contributed to this trend. In the same time period, the number of adults receiving long-term care has fallen,1 indicating growing unmet and undermet need. In recent years, it has been estimated that there are 1.2 million older people in England with unmet care needs.2

Funding gaps are an important issue in social care. While spending increased in 2019–20 compared with previous years, this should be understood within a context of years of low funding/cuts followed by only a very modest increase from 2015 to 2020. Compared with 2010–11, spending in 2019–20 was higher in real terms, but lower on a per-person basis. Increasing demands and rising costs (even adjusting for inflation) have contributed to this funding gap, along with general financial pressures on local authorities that commission social care.1 This highlights the need for innovation and renders the long-term sustainability of social care a pressing issue in policy and practice.3

The social care workforce is also facing challenges. Adult social care is characterised by large numbers of unpaid carers, and workforce issues among paid carers, including high staff turnover and poor conditions.2 The vacancy rate in social care decreased slightly in 2019–20 compared with 2018–9, although it was still high compared with 2012–3 and with the overall unemployment rate, potentially made worse by low pay, low status and poor levels of public understanding.1 In order to meet projected demands for social care, a 2021 report estimated that an extra 627,000 social care staff would be needed, representing a 55% growth over the next decade. This far exceeds the growth in the social care workforce that has occurred over the past 10 years.4 The exit of the UK from the EU may also impact on the social care workforce [16% of whom are non-British (7% from European Economic Area countries)], which relies heavily on migrant workers from EU and non-EU countries.5

The fragility of the social care sector has been exacerbated by the COVID-19 pandemic.6 Social care workers are at an increased risk of infection and death from COVID-19, and the pandemic has likely exacerbated issues in social care around underfunding, unmet social care needs, the burden on unpaid carers, and widespread social and economic inequalities.7 Although there were fewer requests for social care during the COVID-19 pandemic, we do not yet know what impact there will be in the long term (e.g. in terms of a ‘backlog’ of need to be addressed, an exacerbation of need that was temporarily unmet during lockdown, and/or the long-term implications for public finances).1 The impacts of ‘long COVID’ are also still unclear and may impact social care needs in the future.

Digital technology in health and social care

As social care in England faces growing pressures, there has been a turn towards new technologies as a potential solution that might ease pressures across a number of public services and promote independence. While this has often focused more on the health service, there has been increasing awareness of the potential of digital technology for social care in recent years.8 In 2018, the Health Secretary pledged £500 million to transform care technology within both health and social care. The announcement of this money included references to a ‘tech transformation’, with funding set aside for hospital- and home-based technology, electronic systems within the NHS, finding new technology with applications in health and social care, and driving culture change.9

For people not from a technology background, the language used in different policy announcements and in the broader literature can often be confusing, with lots of different terms used, sometimes interchangeably. Before we explore recent policy and the underpinning evidence in more detail, Box 2 sets out some key definitions.

The NHS Long Term Plan, published in 2019, promoted digitally-enabled care as one of its central tenets. In particular, the Plan mentions that technology can help automate some aspects of care, improve the quality of care and free up staff to do other tasks. The Long Term Plan also highlights that home-based monitoring equipment can play a key part in preventing hospital admission, and identifies AI-based technology as a practical solution to drive digital transformation in the NHS.10 In line with this, the Government Digital Service and the Office for Artificial Intelligence have published recommendations for public sector organisations to: assess whether AI has the potential to assist services to meet the needs of their users; describe how the public sector may best use AI; support the safe, fair and ethical adoption of AI; and guide organisations on how to plan and prepare for the implementation of AI.11

In the same year, NHSX was created to take forward digital transformation within the NHS, signalling further commitment to digitally-enabled care within the health and social care system. It was recently announced that NHSX and NHS Digital (a non-departmental body created in 2013 to support information, data and information technology (IT) systems within the NHS) will merge into NHS England. According to representatives from the Department of Health and Social Care, this merger will help to accelerate digital transformation within the NHS.12 Digital Social Care was also created in 2019 to support and advise the social care sector on technology.8

In 2019, Eric Topol led an independent review on how to prepare the health-care workforce to deliver digitally-enabled care. This review recognised that the deployment of AI and other technologies can free up more time to deliver care. Recognising the direction of travel towards digital health and social care, the review identified a need for improved digital literacy of staff and targeted support to implement digital technology. While other technologies are also discussed in the report, AI was highlighted as one of the central approaches to meeting the financial challenges of delivering high-quality care.13

The most recent reform of adult social care, as reflected in the White Paper People at the Heart of Care, also pledged at least £150 million of additional funding to drive the adoption of technology and achieve digitisation to support independent living and improve the quality of care.14 The paper also explores how technology fits into a broader 10-year vision for adult social care, setting out various commitments to accelerate the adoption of technology, ensure that individuals will be able to adapt their homes and access practical tools and technology to live independently and well in their homes, and ensure that staff working in social care will have the confidence to use technology to support care needs and free up time in social care.14 The paper also mentions the potential for technology to be fully utilised to ‘enable proactive and preventative care and to support people’s independence’, including by using technology to identify risks, prevent incidents and ensure quick responses to care needs.14 To achieve this, the White Paper identifies a need for practical guidance and training on technology and digital skills among the social care workforce, additional investment and funding models, and data infrastructure to support technology in care.14

Although COVID-19 has caused additional pressures for the health and social care system in England, it has also accentuated awareness of the potential of technology to deliver or facilitate the delivery of care and support.8,15 Since the pandemic, more people who draw on care and support have needed to receive care remotely, and more technology has been implemented across health and social care to allow care to continue as best it can, for both COVID and non-COVID patients.

AI-based technology for social care

One type of technology that may be particularly relevant for adult social care is sensors with AI capabilities. These can be thought of as a type of telecare, which includes technology and devices used to monitor people who draw on care and support, generate and analyse data about them, and connect them to or provide them with health and care services.8

There are often said to be three generations of telecare, which increase in complexity. In the first, a button is pressed by the user during an emergency, and an alert is sent to a nominated individual via a telephone connection. Second-generation technologies do not rely on intervention by the user (i.e. by requiring the user to press a button). Instead, an automatic alert is sent when a potential breach to health and safety (e.g. a fall) is detected by the sensors. The third generation uses more innovative systems that focus on functionality, with the provision of remote support and lifestyle monitoring to pre-empt, detect and reduce problems relating to activities of daily living. This is the type of technology we are focused on in this evaluation, and we refer to it as ‘AI-based technology for social care’ or ‘new and emerging technology’. The aim of these technologies is to improve quality of life for people who draw on care and support and their carers (rather than exclusively providing reassurance to the person or their carer, or to ensure that help can be available after a fall, as was often the case with previous generations of technology).16–18 They aim to maintain or increase independence by generating alerts when problems occur and by spotting emerging issues early to prevent problems or crises before they arise.19–21

Other mechanisms by which AI-based technology for social care is intended to improve health and well-being can include better personalisation of care, greater control and choice for people who draw on care and support (whether at home or in care homes), and a reduction of costs for health and social care services.22,23 There are also more novel applications of new and emerging technology in social care, such as interactive robots placed within people’s homes,24 although these are less well developed and less widely used in the UK.

While these technologies are not new in social care, technological advances are leading to the development of more sophisticated AI devices that use machine learning to identify patterns in people’s daily activity, recorded by multiple sensors around a person’s home.3 Machine learning (or computational learning) is defined as ‘the set of techniques and tools that allow computers to “think” by creating mathematical algorithms [read as a set of instructions] based on accumulated data.’3, p. 6 The machine learning aspect of AI can allow such technologies to utilise data collected through sensors for offline learning with the aim of improving care.

Evidence around technology in social care

While it cannot replace personal care, technology has the potential to help people with care and support needs to live independent and active lives, and to assist those providing support. However, in order for technology to be successfully embedded into care, it needs to be ‘based on a persuasive evidence base; useful to those who are caring; and – first and foremost – help achieve the outcomes users desire’.25, pg. 1

Research supporting the effectiveness of these types of technology in the context of social care is, however, mixed. A 2008 randomised controlled trial by the then Department of Health [the Whole Systems Demonstrator (programme)] aimed to explore the cost-effectiveness of technologies for health and social care.8,20,26–29 The WSD evaluation did not look specifically at AI, which was less well developed at that time, yet findings from this study provide key learning relevant to other types of technology. The aim of the WSD evaluation was to provide local health and social care commissioners with knowledge on which to base investment decisions, and to provide suppliers of the technology with an understanding of potential business models. Results were mixed, and neither telehealth nor telecare was found to be cost-effective compared with usual care.8,30 The trial identified barriers to the use and acceptance of telecare among people drawing on care and support, including concerns around a lack of technical competence to operate equipment, misunderstandings around what skills are needed to use technology, and fears associated with ageing and self-reliance with respect to using telehealth and telecare resources.31 Despite early indications from the Department of Health that telehealth and telecare had resulted in reductions in emergency and elective admissions, bed-days, costs and mortality rates,32 it has been argued that the findings of the study were misinterpreted based on political pressure and that the actual findings of the study were less positive.33

There are some indications that home sensors with AI technology may be effective in some contexts. A recent randomised controlled trial of an AI- and machine learning-based technology using home sensors for people with dementia (the Technology Integrated Health System study) found that the technology led to mental health and well-being benefits for people who draw on care and support, and that it prompted early intervention to avoid hospital admissions. Furthermore, the technology was well accepted by people drawing on care and support and carers.34,35 However, some people raised concerns about the reliability and user-friendliness of the technology, highlighting internet connectivity issues and the need for more passive forms of data collection that do not require input from the person with dementia, who may become annoyed by having to use devices to collect data.34 It is worth noting that this evaluation focused upon health rather than social care outcomes, specifically: urinary tract infection, agitation, irritation and aggression.34,35

Despite these relatively positive results, other studies more focused on social care outcomes found that similar technology was not effective. A health technology assessment (the ATTILA randomised controlled trial) of another assistive technology for people with dementia who continue to live independently at home found that the technology did not allow people to live in the community for longer, nor did it decrease carer burden, depression or anxiety relative to a basic care programme.36

What influences the success of technology in health and care?

The wider literature around the adoption and implementation of technology in health and social care points to the importance of considering wider contextual factors and how these influence the success of technology. For instance, Greenhalgh et al. (2017) argued that many previous studies have tended to focus on the short-term adoption of simple innovations by individual adopters and on individual barriers and facilitators, failing to theorise the adoption of new technology.37 In contrast, they propose a new framework for theorising and evaluating the adoption/non-adoption, abandonment and challenges to the scale-up, spread and sustainability of health and social care technologies: the NASSS (adoption/non-adoption, abandonment, scale-up, spread, sustainability) framework. This framework includes elements relating to: the condition or illness of the person/population; the technology itself; the value proposition in terms of what the technology will accomplish; adopters of the technology (e.g. patients, staff and carers); organisational considerations; elements of the wider system (e.g. policy, regulation); and how the technology is embedded and how the system around it is adapted over time (see Box 3).37 This framework (informed by theory and evidence) describes the barriers to successful uptake of assistive technology innovations, and provides a guide to the type of issues that should be considered by commissioners and providers when deciding which technologies to adopt, as well as by researchers interested in evaluating the implementation of assistive technologies in health and social care.37 Chapter 10 of this report draws on this framework (alongside insights from the ‘rational model’ of policy implementation) to draw out key learning from our findings.

The framework proposed by Greenhalgh et al. (2017) incorporates the following influences:

• Condition (the nature of the condition or illness of users, comorbidities, sociocultural influences)
• Technology (material features; type of data generated, knowledge needed for use, technology supply model)
• Value proposition supply-side value (to developer), demand-side value (to patient)
• Adopters for staff (role, identity), for users (simple vs. complex input), for carers (availability, nature of input).
• Organisation (capacity to innovate – leadership etc, readiness for technology/change, nature of adoption/funding decision, extent of change needed to routines, work needed to implement change)
• Wider system (political/policy, regulatory/legal, professional, sociocultural)
• Embedding and adaptation over time (scope for adaptation over time; organisational resilience).

Greenhalgh et al. recommend that evaluators of assistive technologies for health and social care consider each of these groups of influences on the success of the implementation effort.

Reproduced with permission from Greenhalgh T.37 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The text above includes minor additions and formatting changes to the original text.

The wider literature also points to some well-known problems associated with the adoption of new technologies. For example, there is often a mismatch between the value that developers of technology see in creating a business case and sales model for their products (the supply side), as opposed to the value that providers and decision-makers within the health and social care system (the demand side) expect from the technology in terms of benefits to people who draw on care, staff and the health and social care system.38,39 Training needs have been highlighted as an issue in the increasing presence of AI-based technology in social care settings. For example, there is a need for training for staff in how to interpret the data collected by AI technologies.3 There are also key considerations such as internet connectivity, particularly in certain parts of the UK, along with ethical questions and data privacy and security concerns, which need to be taken into account when commissioning technology for adult social care.40

Lastly, the implementation process matters greatly in terms of how successful a digital technology may be in health and social care. Findings from the NHS England Test Bed programme suggest a number of key lessons in supporting implementation, adoption and spread of technologies in health and social care,41 as described in Box 4.

• Dedicate sufficient time and resource to engage with end-users.
• Co-design or co-production with end-users is an essential tool when implementing technology.
• Identify the need and its wider impact on the system, not a need for a technology.
• Explore the motivators and barriers that might influence user uptake of an innovation.
• Ignore information governance requirements at your peril.
• Don’t be afraid to tailor the innovation along the journey.
• Ensure adequate training is built in for services using the technology.
• Embedding the innovation is only half the journey; ongoing data collection and analysis is key.
• Ensure there is sufficient resource, capacity and project management support to facilitate roll-out.
• Recognise that variation across local areas exists and adapt the implementation accordingly.

Similar factors were identified in a report looking specifically at assistive technology in adult social care. This pointed to the importance of engaging people who draw on care and support and using language they can understand around assistive technology, engaging with staff to encourage culture shifts and greater acceptance of technology, ensuring that technology meets identified needs when commissioning it for social care, having clear protocols around the use of data, and ensuring that digital infrastructure will support technology.42

Gaps in the evidence

It has been recognised that technology is not being used to its full potential in social care, with a range of stakeholders (from national voluntary organisations such as Carers UK to the government’s Industrial Strategy) identifying the need to increase investment in this area.23,43 There are substantial gaps in understanding the use of emerging technologies to deliver public services, including for assistive technologies in health and social care.3,19,20,44–46 Specifically, there is a lack of evidence on the expected or achieved impact of these technologies on people drawing on care and support and carers, hindering the wider use of technology in social care.20 This lack of evidence has been acknowledged by commissioners of services, who have also expressed concern about this in terms of how to implement technology alongside existing systems, the potential impact on future service use, and cost-effectiveness.20 In addition, although there is a wide range of devices and systems commercially available to support people with care and support needs, there is very little validated information to help select the most suitable technologies.19

Research and evaluation of AI-based technologies for social care in the literature are underdeveloped and limited by methodological issues, and a call has been made for more robust research in this area.3,20 The evidence base is characterised by gaps in knowledge across several themes, including the attitudes and perspectives of different stakeholders (people who draw on care and support, carers and care staff) and the impact on the broader workforce and services. Little is known about how the care workforce may be affected by the adoption of AI-based technology, or what is required from them to facilitate acceptance by people drawing on care and support and carers.3 A lack of involvement of the social care workforce in the design and development of AI for social care has also been acknowledged as a barrier to its implementation.3 The value of exploring the experience of people who draw on care and support and carers using qualitative methods has been acknowledged in the literature, for instance via input into the design and implementation of assistive technology.47 However, research and evaluation specifically on the experiences and attitudes of people drawing on care and support and of carers towards AI-based technology is sparse,3 and there is a lack of evidence around the perspectives of technology providers and innovators and those commissioning care. Lastly, although there is a clear interaction and overlap between health and social care,28 most literature and empirical research on assistive technology focuses on health outcomes.

As a result, discourse on the promise of technological innovation (e.g. from policy-makers and technology companies) is contradicted by evidence showing a poor track record of assistive technologies. They are not widely used in social care currently as they often fail to be scaled up at a local level, or spread widely, and are not sustained in the long term at a system or organisational level.37 This dissonance has been identified in the literature around remote care technologies for social care in the UK, pointing to the need to fill gaps in the evidence before promoting technology as a ‘silver bullet’ for existing challenges in social care.48

Context for study

Interest from policy-makers, commissioners and care providers in developing and using technology for social care makes it a current priority for evaluation.11,22,49 In July–November 2019, an NIHR-funded national prioritisation exercise was held, during which organisations and individuals with knowledge of adult social care and social work identified promising innovations.50 As part of this, a prioritisation workshop using James Lind Alliance principles was held with 23 members, including people who draw on care and support, carers, care staff, researchers, commissioners and policy-makers.51 This prioritisation exercise resulted in a shortlist of the top innovations which would benefit from a rapid evaluation, several of which related to new and emerging technology.

This study builds on the prioritisation exercise by evaluating how one example of new and emerging digital technology (a system of sensors with AI which we have anonymised using the name ‘IndependencePlus’) was implemented in case study sites across England. In particular, we focus on the decision-making process for and implementation of this technology, including consideration of the outcomes that local authorities/care providers are trying to achieve by adopting new and emerging technology, and the way in which this is experienced by people who draw on care and support, carers and care staff.

As a result of this scoping work, the resulting study seeks to answer the following core research questions:

• RQ1. How do commissioners and providers decide to adopt new and emerging technology for adult social care? (Decision-making)
• RQ2. When stakeholders (local authorities and care providers, staff, people who draw on care and support and carers) start to explore the potential of new and emerging technology, what do they hope it will achieve? (Expectations)
• RQ3. What is the process for implementing technological innovation? (Implementation)
• RQ4. How is new and emerging technology for adult social care experienced by people who draw on care and support, carers and care staff? (Early experiences)
• RQ5. What are the broader barriers to and facilitators of the implementation of new and emerging technology in addressing adult social care challenges? (Barriers and facilitators)
• RQ6: How has the COVID-19 pandemic influenced responses to the questions above? (Impact of COVID-19)
• RQ7: How can the process of implementing new technology be improved? (Making improvements)

IndependencePlus

Below (see Box 5), we provide a brief overview of the features of the IndependencePlus technology that was piloted at case study sites involved in this study. While our study is not an evaluation of this specific technology, this is meant to provide context that will be useful in understanding this report and the study findings.

• Sensors: Sensors placed in key locations in individual homes collect information about an individual’s daily activities, without video or audio recording. Information that is collected includes: opening and closing of doors; getting out of bed; opening the refrigerator; using the kettle; and flushing the toilet.
• AI capabilities: Once data are collected on an individual for a sufficient period of time, a baseline can be established, allowing for automated processing of data indicating whether a measure has increased or decreased from the baseline, indicating potential improvement or decline of the individual drawing on care and support. To establish a baseline, continuous data collection is required over a period of time, which may be disturbed if sensors become disconnected or if they run out of battery.
• Individuals living alone: Since the technology cannot distinguish between individuals within the living space, it is best used for individuals living alone, rather than in group living situations.
• Connectivity and Wi-Fi: Individuals using IndependencePlus must have Wi-Fi connection for their data to be collected, and sensors must be plugged in or have a charged battery in order to operate.
• Data dashboard: Data from sensors for each individual are displayed on a data dashboard. In case study sites, only social care workers had access to this data dashboard, although it could theoretically be made available to individuals drawing on care and support or their carers. Social care staff then needed to interpret the data, and understand what increases or decreases in different parameters meant for the individual person.

Part way through local pilots, IndependencePlus evolved the data it collected to include a range of health data (see Chapter 7 for further discussion of this development).