Assistive technology and telecare to maintain independent living at home for people with dementia: the ATTILA RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 10/50/02. The contractual start date was in January 2013. The draft report began editorial review in September 2019 and was accepted for publication in December 2019. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2021 Gathercole et al. This work was produced by Gathercole 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 adaption 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.

2021 The authors

Scientific background

There are approximately 850,000 people with dementia in the UK3,4 and an estimated 700,000 people who provide unpaid care for them. Many of these people with dementia will require accommodation in nursing or residential care homes when their illness has progressed to the point at which they can no longer live safely and independently in their own homes. It has been estimated that, over the next two decades, the number of people aged ≥ 85 years will increase by two-thirds. 5 Over half of all users of adult social care services are aged ≥ 65 years,6 and a steep rise in the numbers of people living with dementia is expected over the next few decades. The financial cost of caring for people with dementia is considerable,3 as is the social and psychological cost to unpaid caregivers (generally a relative or friend, subsequently referred to as ‘caregivers’). Caregiver breakdown is a common reason for the unplanned admission of older people (many of whom will have dementia) to permanent nursing or residential care. 7

Living Well with Dementia, the theme of the 2009 National Dementia Strategy for England,8 involves helping people with dementia to retain their independence while living in their own homes, and to maintain their quality of their life. People living with dementia who move from their own homes into institutional care often experience a loss of independence and quality of life. To minimise this possibility, the NHS and councils with adult social services responsibilities (CASSRs) in England aim to support people with dementia to live safely in their own homes for as long as possible.

Assistive technology and telecare (ATT) offer relatively new means of delivering care and support to people with social care needs by helping to manage the risks facing older people with dementia who wish to remain living independently at home. It is claimed that sensors (e.g. to detect falls, floods or the presence of gas from an unlit appliance in someone’s home), passive monitoring that uses sensors placed in a home environment to detect movement, and alerting devices to relay information from the person’s home to a remote site such as a call centre support the independence of people with social care needs,9,10 reduce the burden on caregivers11–15 and save money for CASSRs. 16 By addressing risks associated with independent living for people with dementia, it is claimed that ATT help reduce the need for community care, prevent unnecessary hospital admissions and delay or prevent admission to residential or nursing care. 12,17–19 The evidence to support such claims is limited, and based largely on qualitative evidence or uncontrolled quantitative studies. 20,21 There is, therefore, an urgent need to provide evidence to inform decisions about whether or not to provide ATT in the homes of people with dementia.

The first use of electronic ATT in the UK, in the 1990s, was to provide support for people with dementia and their caregivers. 17,22–25 Within a decade, interest in ATT has developed from a fringe interest for a handful of enthusiasts to a multimillion-pound industry commanding government support, a Department of Health and Social Care (DHSC) strategy26 and, increasingly, the use of ATT in CASSR settings as a mainstream service (see Woolham et al. 25). However, as interest in ATT has increased, the specific focus on its application for those living with dementia has diminished. 13 The performance indicators that followed the Preventative Technology Grant (given to CASSRs in 2008 by the DHSC26) were intended to promote the widest possible use of telecare. The DHSC did not, however, offer a clear indication of what this grant was supposed to ‘prevent’. Although Woolham12 has drawn attention to the cost-effectiveness of telecare for people with dementia by closely matching ATT with assessed need, thereby preventing the need for more expensive forms of care, Poole27 has argued that CASSRs should see ATT as a long-term investment, deploying it at an early stage without expecting immediate savings. This has contributed to a situation in which CASSRs have implemented ATT across several different care groups without always referring to the needs of the specific groups, such as people with dementia. The current economic situation, and a significant reduction in government CASSR funding, has led to increasing numbers of CASSRs developing an interest in ATT. Some have developed local strategies to use it, whereas others already have well-developed ATT services that can be deployed alongside, or instead of, non-institutional forms of support, often known as ‘community care’ in the UK.

Despite growing ATT use, the evidence to support its use is limited. The Whole System Demonstrator (WSD) study was funded by the DHSC in 2008 to investigate the impact and effectiveness of ATT in England. 28–35 However, individuals with dementia were not specifically included. This, together with the relative dearth of dementia-specific studies relating ATT, means that a significant gap in the evidence remains. Although there are relatively large numbers of qualitative studies, audits and service evaluations, there are few studies with sufficient rigour and appropriate design to offer any degree of generalisability21 or agreement about how ‘success’ can be measured. 36 One study25 has suggested that, when used appropriately, ATT are highly cost-effective, but limitations in design and methodology constrain the generalisability of this study’s findings. A Cochrane review in 201737 found no research evaluating assistive technology for people via a randomised controlled trial (RCT), with the exception of this Assistive Technology and Telecare to maintain Independent Living At home for people with dementia (ATTILA) trial, which was in progress at the time of publication of the Cochrane review. 37 Evidence from a well-designed trial such as ATTILA is clearly needed to guide future policy direction.

The DHSC’s ‘Building Telecare’ strategy in 200513 provided generic advice to CASSRs. As part of this strategy, a Preventative Technology Grant, which CASSRs were required to spend on developing local ATT services, was made available. England’s 2009 National Dementia Strategy8 recommended a ‘watching brief’ for emerging evidence of the impact of telecare, stating:

However, with respect to more recent innovations, this is not an area where the strategy is able at this time to make specific recommendations. Instead, central, regional and local teams should keep in touch with initiatives in the areas of housing and telecare and make appropriate commissioning decisions as data become available, for example from the Department’s large-scale field trials of telecare and assistive technology.

Reproduced from DHSC. 8 © Crown copyright 2009. Contains public sector information licensed under the Open Government Licence v3.0

Aims and objectives of the trial

The ATTILA trial was designed to answer questions about the efficacy and cost-effectiveness of ATT, with relevance for those who commission and provide care for people with dementia.

The aims of the ATTILA trial were to test the following hypotheses:

• that the application of ATT will significantly extend the time that people with dementia can continue to live independently and safely in their own homes

• that ATT interventions are cost-effective in the management of risk and maintenance of independence for people with dementia living in their own homes

• that provision of ATT interventions to people with dementia living at home will significantly reduce the number of incidents involving serious risks to safety and independent living, particularly those involving acute admissions to hospital

• that ATT interventions will reduce burden and stress in family and other caregivers and increase quality of life for people with dementia.

These hypotheses were tested by the following primary and secondary objectives:

• Primary objective – to establish whether or not ATT assessments and interventions extend the time that people with dementia can continue to live independently in their own homes and whether or not this is cost-effective.

• Secondary objectives –

  • to establish whether or not these technologies can significantly reduce the number of incidents involving serious risks to safety and independent living, including acute admissions to hospital

  • to reduce burden and stress in family and other caregivers, and increase quality of life for people with dementia and their caregivers

  • to collect qualitative and quantitative data from people living with dementia, their paid and unpaid caregivers, and members of the NHS and CASSR teams about their experience of using these technologies.

Structure of the report

In Chapter 2, we provide a summary of the trial methods. In the chapters that follow, we set out the methods and results of the research, beginning with work carried out to describe the intervention (see Chapter 3). In subsequent chapters, we report on participant outcomes, cost-effectiveness, caregiver outcomes and ethnographic research with participants and caregivers. In Chapter 8, we summarise the findings, reflect on the strengths and weaknesses of the research and make recommendations for future research and practice.

Deviations from the protocol

We did not conduct one proposed cost-effectiveness analysis. It was proposed that an analysis of the change in EuroQol-5 Dimensions, five-level version (EQ-5D-5L), score over 2 years would take into account the costs of permanent care home and hospital stays of those admitted to care homes over that period. This analysis was not conducted because no outcomes and costs data were collected from caregivers when participants were permanently admitted to care during the 2-year follow-up.

Trial objectives

The ATTILA trial was a pragmatic RCT comparing outcomes for people with dementia who received a full ATT package with the outcomes for people with dementia who received equivalent community services but ATT were limited to a pendant alarm, non-monitored smoke and carbon monoxide detectors, and key safes. Assistive technology is defined as ‘any item, piece of equipment, product or system, whether acquired commercially, off the shelf, modified or customised, that is used to increase, maintain, or improve functional capabilities of individuals with cognitive, physical or communication difficulties’. 22 Telecare can include many different interventions, including those aimed at delivering care and monitoring remotely. 38 For the purposes of this trial, these devices had to be provided to support challenges related to memory problems and recommended by a health or social care professional.

The primary objectives of the trial were to establish whether or not:

• ATT assessments and interventions can extend the time that people with dementia can continue to live independently and safely in the community

• ATT interventions are cost-effective in the management of risk and maintenance of independence in people with dementia living in their own homes.

The secondary objectives were to:

• establish whether or not these technologies can significantly reduce the number of incidents involving serious risks to safety and independent living, including acute admissions to hospital; reduce stress in family and other caregivers; and increase quality of life for those with dementia and their caregivers

• collect qualitative and quantitative data from people living with dementia and their formal caregivers about their experience of using these technologies.

Trial design

The ATTILA trial was a multicentre, pragmatic RCT, conducted over 260 weeks, that took place in the homes of people living with dementia who were eligible to receive a package of care. The trial compared outcomes in two groups of participants randomised to one of the two trial arms: (1) receiving an assessment of needs followed by the installation of appropriate ATT devices and response services to be deployed by the CASSR or NHS (a full ATT package) or (2) receiving an assessment of needs followed by the installation of an ATT package restricted to smoke and carbon monoxide detectors, a key safe and a pendant alarm if indicated, also arranged by the CASSR (a basic ATT package). The co-primary outcomes were time to institutionalisation and cost-effectiveness of the intervention.

The trial was not funded to source, assess for, or deploy ATT. Our approach was to work alongside CASSRs who were charged by the Department of Health and Social Care with responsibility for establishing and developing local ATT services.

Site identification and recruitment

Recruitment of local authority Adult Social Care Department (ASCD) sites to the trial was opportunistic because of anticipated difficulties in securing permissions. Sites known to have well-established telecare services were identified, as well as those located in geographical areas close to the place of employment of research team members. This also meant that participating ASCDs were geographically widely spread across England, and that all types of local authority were represented. Telephone contact was usually first made with a telecare manager in identified sites. In most cases, this person referred our request to the departmental senior management team or to the director. In some sites, repeated visits were needed to discuss the request. Several ASCDs declined to take part in the main trial, either because they did not feel that they had the resources to do so or because of lack of fit between the trial aims and the strategic priorities of the service.

Participants

Participants were people with any dementia diagnosis, or suspected dementia, who were living in the community and were from one or more of three constituencies:

1. people who sought help or support from local authority social care services in the areas that had agreed to support the trial (Barnsley, Blackburn, Blackpool, Cambridgeshire, Croydon, Lambeth, Lancashire, Nottingham, Norfolk, Oxfordshire, Southwark and Suffolk) and who met local ASCD eligibility criteria for their support

2. people who were supported by the services of the NHS and were referred to an ASCD and met local ASCD eligibility criteria

3. people who were recruited from the caseload of NHS services for older adults and referred to local social services and met local ASCD eligibility criteria.

Those referred from the NHS usually had to meet eligibility criteria for social care because this often determined if ATT could be provided.

Most participants (n = 431, 87.1%) did have a dementia diagnosis, but some of those referred by LAs had not yet had a formal diagnosis. In these cases, clinical judgement was used by the research worker and health or social care staff involved to decide whether or not the cause of the potential participant’s memory impairment was dementia. If necessary, the research worker could discuss further with the local principal investigator.

Screening for eligibility and the preliminary information visit

All trial procedures, including the initial visit and consent visit, took place in participants’ homes. At the first appointment, participants were assessed for eligibility based on the following inclusion and exclusion criteria.

Selection and recruitment

There were several routes for participant recruitment. Most participants were referred from local authority social services, but other referring services included not-for-profit organisations providing ATT; charitable organisations such as Age UK and the Alzheimer’s Society; and NHS mental health, community care and primary care services. Those referred from the NHS had to meet eligibility criteria for social care provision.

After assessing eligibility of new referrals for both social care support and the ATTILA trial, participating services asked if a potential participant’s contact details could be made available to a named individual in the local research team. Once identified, the research worker contacted this person and arranged to visit them and a caregiver who knew them well. Those meeting the eligibility criteria had the possible benefits and risks of participation in the trial explained to them. Following this, the participant was given a general outline of three possible options: (1) taking part in the ATTILA trial with the intervention (i.e. ATT package or regular support package without ATT) decided by randomisation, (2) declining to participate in the ATTILA trial and (3) taking more time to consider their decision about whether or not to participate. Those who were interested in taking part in the trial were provided with participant and caregiver information leaflets to find out more about the trial before deciding whether or not to participate. After a full explanation of the intervention options and the manner of treatment allocation, all suitable participants were invited to take part in the randomised component of the trial. If urgent provision of support services was required, then consent was sought at that visit so that they could be immediately randomised. Otherwise, information about the trial could be left with the prospective participant and, if they required more time to consider participation, the researcher would return at a later date to take consent and subsequently randomise the participant. Consent was also obtained from the caregiver using the caregiver consent form in the trial folder. If a participant lacked capacity, a professional or personal consultee was involved to ensure that participation in the trial was in the person’s best interests (according to guidelines established in the Mental Capacity Act 2005: Code of Practice39). When appropriate, data-sharing agreements were agreed with the CASSRs and health services concerned to ensure that the transfer of personal data from the local authority to the research team was lawful. If consent was not given, the participant was not included and any personal data were removed from research team records and destroyed. Reasons why those who were potentially eligible did not consent to take part were recorded on a screening log in the ATTILA trial folder. After randomisation, assessment for ATT and provision of ATT services (within limits set by randomisation) were left entirely up to the local authority or health service operational teams concerned.

Outcome measures

The co-primary trial outcomes were (1) time to institutionalisation and (2) the cost-effectiveness of the ATT intervention. Table 1 shows the schedule of all assessment points and the measures used at each.

Secondary efficacy parameters

Sample size

The two primary outcome measures were time to transition to institutional care and cost-effectiveness. It was anticipated that 50% of participants with a Bristol Activities of Daily Living Scale (BADLS)47 score of ≥ 15 would transition to institutional care after 24 months, based on observed institutionalisation rates in participants from the AD2000 (Alzheimer’s disease) cohort. 48 A reduction in the estimated 24-month transition to care home rate by 30% (i.e. 50% institutionalised at 2 years reduced to 35%) would require the involvement of 500 participants, allowing for 10% attrition due to death while still community resident. This equated to an average of 55 days more of independent home life for each participant who received ATT. The trial would therefore be powered to detect a mean institutionalisation delay of just under 8 weeks. Expert opinion suggests that 8 weeks is close to the minimum clinically important difference in delaying institutionalisation.

Trial interventions

As the trial design was pragmatic, all aspects of the intervention (ATT assessment, funding, choice of devices, or ordering and installation of devices) were determined by staff from participating LAs or telecare providers. We worked alongside these teams, which have been charged by the Department of Health and Social Care with responsibility for establishing and developing local ATT services. The trial was not funded to source, assess for or deploy ATT.

Each participant underwent an assessment with the ATT provider to determine the level of need and what services that they required. The intervention involved the installation of simple, battery-operated, standalone technologies and/or telecare (a range of devices and sensors that communicate and relay messages to an external call centre where an appropriate response is arranged). The installation and selection of the technology to be deployed was the responsibility of the LAs involved. Those in the control arm were limited to a pendant alarm, non-monitored smoke and carbon monoxide detectors, and a key safe, as recommended by the health or social care professional assessing their needs. Both arms could use additional support services, such as paid care, meals on wheels and day centres.

Randomisation

Participants were randomised in a one-to-one ratio, using telephone-based randomisation and a computerised data entry portal provided by the Oxford Clinical Trial Service Unit. Treatment allocation was via a minimised randomisation procedure stratified by the following variables to reduce the risk of chance imbalances between arms. This information was obtained by the local trial team following consent and during the screening process. Variables were as follows:

• sex

• age (< 65, 65–80 or > 80 years)

• risk of wandering or leaving the home inappropriately (low, moderate or high risk)

• safety risk in the home (low, moderate or high risk)

• level of caregiver support available (live-in caregiver, caregiver visits at least once daily or caregiver visits less often than daily).

This stratification procedure was reviewed by the Trial Steering Committee after the first 100 randomisations.

Blinding

Blinding was not undertaken for participants or trial staff collecting data as it was not practicable or ethical to conceal allocation of ATT. The staff who entered the data were unaware of which arm a participant had been allocated to.

Patient and public involvement

The Alzheimer’s Society was involved in devising the research question and in the production of the trial materials. Two service user representatives sat on the Trial Steering Committee.

Ethics approval

The trial was conducted and designed in compliance with the principles of the Declaration of Helsinki49 and the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use – Good Clinical Practice (ICH-GCP). 50 All researchers working on the trial received training in ICH-GCP guidelines. The integrated form for both site-specific information and research and development approval at all participating NHS sites was approved prior to recruitment at each site. Annual progress and safety reports and a final report at conclusion of the trial were submitted to the Multicentre Research Ethics Committee and the Health Technology Assessment programme within the timelines agreed.

The trial was approved by the NHS Health Research Authority National Research Ethics Committee (reference number 12/LO/186) and is registered with the International Standard Randomised Controlled Trial Number registry (ISRCTN86537017).

Sponsorship

The trial was sponsored by South London and Maudsley NHS Foundation Trust.

Introduction

A detailed exploration of the intervention under investigation is needed to give insight into the fidelity of the intervention and to allow for replication. 52 The aim, therefore, was to provide an overview of routine ATT practice and the systems in place to deliver ATT for people with dementia.

Method

We adhered to the Template for Intervention Description and Replication (TIDieR)53 checklist in describing the components of the ATTILA trial intervention, in terms of what happened; who was involved; how, where and when the intervention happened; how much ATT was provided; and whether or not it was tailored to participants.

Results

Assessment

Of the 451 documented ATT needs assessments available, 413 contained an ATT recommendation. Of the 248 participants recruited to the intervention arm, data from 209 participants were available for analysis of ATT installations.

In total, 60% of assessment responses identified an ATT need, with 4.4 ATT needs (range 0–12) identified per participant (Tables 3 and 4). The mean number of ATT needs identified varied, ranging from two to six per site (p < 0.001). The areas of concern most frequently identified as triggering the need for ATT were daily activities (93%), memory (89%) and problem-solving (83%). Health and social care professionals identified more ATT needs than ATT assessors (p = 0.047). More ATT needs were identified through in-person, at-home assessments than through telephone assessment methods (p < 0.001). There was no significant difference between ATT needs in men and women (p = 0.337). The number of ATT needs identified for each participant differed depending on the risk of wandering (p = 0.005): a medium risk of wandering was associated with more ATT needs than a low risk of wandering (p = 0.016). ATT needs varied by category of SMMSE score (p < 0.001): participants with severe dementia had more ATT needs than those with mild (p < 0.001), those with moderate (p = 0.002) or those with questionable dementia (p < 0.001).

TABLE 3 - Fidelity with ATT assessment standard and identified ATT needs, by assessment areas

Site ATT assessment areas/standard	Fidelity with ATT assessments standard		ATT needs (i.e. responses rated as ‘at risk’)
	n/N	%	n/N	%
1. Insight	241/451	53	151/241	63
2. Values	245/451	54	100/245	41
3. Wandering/disorientation	284/451	63	219/284	77
4. Daily activity	321/451	71	298/321	93
5. Conversation	226/451	50	100/226	44
6. Express needs	175/451	39	24/175	14
7. Memory	320/451	71	284/320	89
8. Problem-solving	218/451	48	181/218	83
9. Mobility	335/451	74	224/335	67
10. Grip/dexterity	147/451	33	18/147	12
11. Space	140/451	31	47/140	34
12. Resources	128/451	28	26/128	20
13. Social support	325/451	72	183/325	56
14. The way the activity is completed	162/451	36	118/162	73
Total responses	3267/6314	52	1973/3267	60

TABLE 4 - Fidelity with ATT assessment standard and identified ATT needs, by participant and assessment characteristics

Characteristic	Fidelity with ATT assessments standard			Number of ATT needs
	Median	Mean	%	Median	Mean	%
Participant characteristic
Gender
Female	8	7.67	62	4	4.46	60
Male	5	6.65	38	3	4.25	40
	p = 0.027			p = 0.337
Risk of wandering
Low	7	6.93	70	4	4.10	68
Medium	9	8.37	23	5	5.04	23
High	6	7.24	7	4	5.24	9
	p = 0.038			p = 0.005
SMMSE score of 18 points
Questionable dementia (26–29)	7	7.22	13	3	3.38	10
Mild dementia (21–25)	7	6.6	27	4	3.9	27
Moderate dementia (11–20)	7	7.38	45	4	4.27	44
Severe dementia (0–10)	8.5	7.96	15	5.5	5.79	19
	p = 0.309			p < 0.000
Assessment characteristic
Assessors
Health and social care professionals	8	7.85	68	4	4.66	67
ATT assessor	5.5	6.51	29	3	3.86	29
	p = 0.051			p = 0.028
Assessment method
In person, at home	10	9.14	85	5	5.06	82
In person, not at home	5	6.43	8	3	3.38	8
Telephone	2	3.42	6	2	2.71	9
Case notes	3	3.33	1	2.5	3	1
	p < 0.000			p < 0.000
Service structure
Public telecare provider	7	7.59	73	4	4.41	70
Not-for-profit telecare provider	6	6.41	25	4	4.31	28
	p = 0.026			p = 1.00
	Mean fidelity with ATT assessment standard per participant: 7.2 assessment areas addressed (range 0–13)			Mean number of responses per participant rated as an ATT need: 4.4 ATT needs (range 0–12)

Fidelity of assessment

The local ATT assessment fidelity with the ATT assessment standard was 52% (7.2 assessment areas were addressed per assessment) (see Tables 3 and 4). Of 451 ATT assessments reviewed, 99 (22%) addressed 0–2 areas of assessment. There was higher fidelity to assessment areas relating to ‘mobility’ (74%), ‘social support’ (72%), ‘daily activity’ (71%) and ‘memory’ (71%). Fidelity varied across sites: the mean number of assessment areas addressed ranged from two to 13 per site (p < 0.001), with public telecare providers addressing more assessment areas than not-for-profit telecare providers (p = 0.026). Health and social care professionals addressed more assessment areas than ATT assessors (p = 0.046). Fidelity varied across assessment methods (p < 0.001), with the in-person, at-home assessment method addressing more assessment areas than in-person, not-at-home (p = 0.003), telephone (p < 0.001) and case notes assessment methods (p = 0.003). Women had more assessment areas addressed than men (p = 0.027). More assessment areas were addressed for participants at medium risk of wandering than for participants at low risk of wandering (p = 0.028).

Assistive technology and telecare delivery system

Value networks

Networks delivering services (offering value) to participants in ATTILA trial sites were classified into two types (Figures 1 and 2). First were ‘public telecare provider networks’ (n = 4), for which two assessor roles were identified: the ATT assessor and the authorised (or trusted) assessor (health or social care professional). ATT assessors were employed by public agencies (NHS or CASSRs); their primary role was to assess for a full range of ATT devices [‘networked’ (monitored by a telecare call centre or caregiver) or ‘standalone’]. Authorised assessors could offer first-generation telecare (pendant-only systems) or straightforward ATT (e.g. adding on an additional sensor or providing a memo minder), depending on their level of experience and local permissions; they performed ATT assessment as a secondary role. In these networks, most or all of the ATT infrastructure for procurement, installation, stock control and maintenance of ATT devices fell to units in the CASSR. The second type of networks were ‘not-for-profit provider networks’ (n = 3). Three assessor roles were identified across these ‘not-for-profit telecare networks’. Telecare assessors working for not-for-profit telecare providers assessed for AT that was networked to providers’ call-monitoring centres. Assessment for standalone assistive technology fell to assessors in the CASSR. A ‘social care ATT assessor’ role was also identified; these assessors could assess for ATT (networked or standalone) and work with a choice of suppliers to procure and arrange the installation of ATT devices. Private companies offered combinations of procurement and stock control, installation, and maintenance services to the not-for-profit telecare providers.

FIGURE 1.

Public assessor value networks. a, Featured in some but not all networks.

FIGURE 2.

Not-for-profit value networks. a, Featured in some but not all networks.

Summary

The findings are the first to describe assistive technology for people with dementia. The components of the ATTILA trial intervention have been described, in terms of what happened, who was involved, how, where and when the intervention happened, how many devices was provided and whether or not the intervention was tailored to participants. Value networks operating in ATTILA trial sites were characterised as either public or not-for-profit telecare provider network types. The ATTILA trial intervention is summarised in Table 7, using the TIDieR format.

Recruitment

Eleven sites in England were opened for recruitment to the ATTILA trial. The first participant was randomised on 14 August 2013; recruitment ended on 26 October 2016. The ATTILA trial randomised over a period of 38 months, with an average recruitment rate of 13 participants per month. Yearly recruitment per site is shown in Table 8.

A total of 1411 people were assessed for trial eligibility; of these, 495 were randomised across the 11 sites: 248 were randomised to receive ATT and 247 were randomised to the control arm. The Consolidated Standards of Reporting Trials flow diagram of participants through the ATTILA trial is shown in Figure 3, and Table 9 shows the participant status at the end of the trial.

FIGURE 3.

The Consolidated Standards of Reporting Trials flow diagram.

Those people who declined to consent did so because they wanted ATT (n = 53), because they did not want ATT (n = 83) or because they did not want to participate in research (n = 162). Other reasons for being excluded were, primarily, that the researcher was unable to contact the potential participant (n = 131) or because participation was deemed inappropriate (n = 48).

During the follow-up of the ATTILA trial, in total, 200 participants were admitted to care, 89 participants died, 42 withdrew from further follow-up and 18 were lost to follow-up. This resulted in 146 participants finishing the trial living independently in the community: 85 in the intervention arm and 61 in the control arm. Relatively few participants (3.6%) were lost to follow-up, as, once randomised, every effort was made to follow up participants throughout the trial to obtain all follow-up forms and outcome assessments. Of the 18 lost to follow-up, 10 were in the intervention arm and eight were in the control arm. Once a participant was admitted to care, the follow-up was terminated and no outcome assessments were collected. Table 10 displays a comparison of the key participant characteristics of the 495 participants randomised.

All characteristics appear to be reasonably well balanced across the two randomised arms. The average age was 80.9 years, 59% (290/495) were female and 48% (240/495) had a live-in caregiver. The majority of participants, 72% (358/495), were classified as being at low risk of wandering or leaving their home inappropriately. Half of the participants (249/495) were deemed to have low safety risks identified within the home. The average SMMSE score was 18.7 points in the intervention arm and 16.9 points in the control arm, so participants in the intervention arm had a slightly higher baseline SMMSE score than those in the control arm. The average BADLS score was 19.5 in the intervention arm and 20.4 in the control arm. The missingness of these data was similar between the two arms.

Primary outcomes

Time to institutionalisation

The primary analysis of admission to care was defined as a permanent transition from living in a participant’s own home to living in nursing or residential care, or admission to an acute care facility that resulted in permanent placement. The end point was compared between the intervention and control arms using survival analysis. Kaplan–Meier curves are for graphical representation of time to an event. Statistical significance was determined through log-rank test. The primary analysis was conducted according to intention to treat, and participants who have died, withdrawn from follow-up or who were lost to follow-up were censored at the date of withdrawal. The time to admission to care, split by randomised arm, is shown in Figure 4.

FIGURE 4.

Kaplan–Meier survival curve: time to admission to care by randomised arm, unadjusted analysis.

The intervention and control arms showed a similar pattern of time to admission to care over the 3-year period plotted. Comparing the ATT arm with the control arm, the hazard ratio is 0.76 [95% confidence interval (CI) 0.58 to 1.01; p = 0.054]. This unadjusted analysis showed a borderline significant difference in slowing the time to admission to care with ATT use when compared with the control. At 2 years, the admission to care rate for the ATT arm was 65.6% (95% CI 58.8% to 71.5%), compared with 63.4% (95% CI 56.3% to 69.7%) for the control arm.

The rates of admission to care can be affected by participants’ functional ability. This was measured using the BADLS. BADLS scores range from 0–60, with higher scores indicating greater impairment. Figure 5 shows the time to admission to care split by BADLS scores 0–4, 5–14, 15–29 and 30–60.

FIGURE 5.

Kaplan–Meier survival curve: time to admission to care by baseline BADLS score.

There was a highly significant difference in the time to admission to care when comparing baseline BADLS scores. Participants with a higher baseline BADLS score were more likely to be admitted to care (p < 0.0001). Baseline BADLS scores are presented in Table 10 and shows an imbalance in baseline scores. More participants in the intervention arm than in the control arm had a lower baseline BADLS score. As Figure 5 showed that participants with a higher baseline BADLS score are more likely to be admitted to care; this difference at baseline was adjusted for in the primary analysis. A forest plot split by baseline BADLS score is shown in Figure 6.

FIGURE 6.

Forest plot: admission to care by randomised arm, adjusted for baseline BADLS score. NS, not significant; O/E, observed over expected; Var., variance.

When adjusting for baseline BADLS scores, there is no significant difference in the time to admission to care between those in the intervention group and those in the control group (hazard ratio 0.84, 95% CI 0.63 to 1.12; p = 0.20).

The reasons for admission to care are usually multifactorial. To determine whether or not ATT might have helped prevent admissions to care, the reasons given for institutionalisation have been categorised as having any mention of safety, then any mention of wandering and then falls, with others classified as inability to perform activities of daily living (ADL), behaviour, other medical condition, deterioration (unspecified), caregiver health, other and unknown. This can give only an approximate classification, given the complexity of Alzheimer’s symptoms, but the breakdown of the most likely causes is given in Table 11.

TABLE 11 - Reasons for admission to care categorised

p-value from Mantel–Haenszel test (ignoring time to event).

Categorised reason	Trial arm (number of participants)		Total number of participants (N = 495)	p-valuea
	Intervention (N = 248)	Control (N = 247)
Safety concern	12	4	16	0.043
Wandering	5	13	18	0.054
Falls	13	13	26	0.990
Loss of ADL	14	29	43	0.016
Behaviour	8	10	18	0.630
Other medical condition	7	6	13	0.790
Deterioration (unspecified)	14	11	25	0.540
Caregiver health	9	3	12	0.081
Other	6	8	14	0.580
Unknown	5	10	15	0.190
Any cause	93	107	200

The most common reason for admission to care is the inability to perform ADL. Institutionalisation for safety concerns, which might have been expected to be reduced by ATT, is actually more common in the intervention group (12 vs. 4 participants; p = 0.043). By contrast, the risk of wandering, which might, again, be mitigated by appropriate ATT, was reduced in the intervention group (5 vs. 13 participants; p = 0.054). There was also a significant reduction in the number of participants moving into residential care because of the inability to perform ADL (14 vs. 29 participants; p = 0.016). A total of 15 admissions to care were classified as being for an unknown reason.

Subgroup analysis

To investigate whether or not ATT use varied by baseline characteristics, we did subgroup analyses of admission to care in the ATT group compared with admission to care in the control group by gender, age, risk of wandering from home and safety risk within the home (Figure 7). As there were no significant differences seen, there is no indication of any benefit from ATT use in any of these subgroups.

FIGURE 7.

Subgroup analyses of admission to care for the ATT group vs. the control group, by baseline characteristics. O/E, observed over expected; Var., variance.

Deaths while in the community

Eighty-nine participants died while in the community. Figure 8 is the Kaplan–Meier graph of time to death while community resident and Table 12 shows the categorised reasons for cause of death. In the Kaplan–Meier analysis, participants who had been admitted to care, withdrawn from follow-up or lost to follow-up were censored at the date of withdrawal.

FIGURE 8.

Kaplan–Meier survival curve: time to death while community resident, by randomised arm.

TABLE 12 - Causes of death categorised

p-value from Mantel–Haenszel test (ignoring time to event).

Cause of death	Trial arm (number of participants)		Total number of participants (N = 495)	p-valuea
	Intervention (N = 248)	Control (N = 247)
Health/dementia deterioration	8	4	12	0.25
Pneumonia/respiratory failure	4	10	14	0.10
Heart attack/heart failure	3	8	11	0.13
Stroke	7	5	12	0.56
Cancer	7	4	11	0.36
Infection	6	4	10	0.53
Other	2	4	6	0.41
Unknown	4	9	13	0.16
Total	41	48	89

There were no significant differences seen overall (p = 0.14) or in the grouped categories for cause of death (see Figure 8 and Table 12).

Serious adverse events

Serious adverse events have been grouped into broad categories and are summarised in Tables 13 and 14. The categories were decided on by members of the ATTILA trial team with clinical expertise and were categorised separately by two members of the team, then assessed for consistency. Any differences were discussed, and input sought from a clinical expert in the team. Raters were unaware of treatment allocation of the participants involved. Table 13 presents the counts of SAEs recorded and Table 14 presents the number of participants reporting the SAEs, as participants can report multiple SAEs. Similarly to the reasons for admissions to care, SAEs could be multifactorial. The categories are any mention of safety concerns, wandering, falls, dementia progression, behaviour, other medical condition, caregiver related, accidents, health deterioration, other and unknown.

TABLE 13 - Counts of categorised SAEs

Categorised SAE	SAE count (n)
	Intervention arm	Control arm	Total
Safety concerns	15	5	20
Wandering	36	71	107
Falls	182	187	369
Dementia progression	37	46	83
Behaviour	5	21	26
Other medical condition	214	220	434
Caregiver related	11	10	21
Environmental/accident	14	21	35
Health deterioration	6	3	9
Other	2	1	6
Unknown	10	18	28
Total count of SAEs	532	603	1135

TABLE 14 - Number of participants reporting categorised SAEs

Categorised SAE	Number of participants			p-value
	Intervention arm	Control arm	Total
Safety concerns	13	5	18	0.06
Wandering	25	36	61	0.13
Falls	86	88	174	0.83
Dementia progression	37	43	80	0.45
Behaviour	5	16	21	0.01
Other medical condition	107	109	216	0.83
Caregiver related	11	10	21	0.83
Environmental/accident	13	15	28	0.69
Health deterioration	5	2	7	0.26
Other	2	1	3	0.57
Unknown	10	16	26	0.22
Total	195	201	396	0.45

Participants reported multiple SAEs. Overall, 1135 SAEs were reported from 396 participants. The most common SAE was ‘other medical condition’, which was reported by 216 participants. The second most-reported SAE was related to falls (369 falls were reported by 174 participants). Figure 9 plots the number of participants experiencing each SAE type, ordered by hierarchy of classification, with a test of significance between intervention and control participants.

FIGURE 9.

Forest plot of the incidence of SAEs. p-values from Mantel–Haenszel tests (ignoring time to event). OR, odds ratio.

Figure 9 shows that there was a significant reduction in the number of participants experiencing behavioural-related SAEs in the intervention group, compared with the control group (p = 0.01). More participants in the intervention group than in the control group reported SAEs related to safety concerns (p = 0.06).

Methods

The economic evaluation included cost-effectiveness and cost–utility analyses.

Analyses

Descriptive analyses were produced for service and other resource use items available at each time point, presented in terms of the proportions of each treatment group using the service and the mean use of the service in each group. Group means and standard errors (SEs) were calculated for categories and totals of costs and for outcomes at each time point, as were mean differences and SEs of the difference between groups.

For the outcome days in the community until admission to a care home, the number of days was estimated in a survival analysis accelerated failure time model using the Weibull distribution, adjusting for baseline BADLS score. A further step was involved in the QALY outcome: to quality-adjust the days lived in the community, taking a population (or group-based) approach to produce a quality-adjusted survival curve, using the EQ-5D index scores at each assessment point to estimate the average utility per treatment group. 69,70 Costs were partitioned and estimated in a generalised gamma accelerated failure time model with a square root link; the probability of not being censored in each time interval was estimated by accelerated failure time models (generalised gamma model at 104 weeks and a Weibull model at 52 and 104 weeks, as the generalised gamma model did not converge), generating inverse probability weights for costs at each interval. 70–72 Cost regressions were adjusted for the treatment allocation, BADLS score and stratifying variables. Cases without a baseline BADLS score and cases for whom the dyad had never participated in assessments were excluded from the analyses. Bias-corrected bootstrapped CIs of regression estimates of cost and outcome differences were produced (25,000 resamples).

For the change in EQ-5D-5L index outcome, multilevel linear difference-in-difference models were fitted to costs and EQ-5D index scores data. The models estimated the difference in the change in scores from baseline to follow-up in the intervention group, less the difference in the change in scores from baseline to follow-up in the control group (i.e. the difference between groups in the difference between baseline and follow-up costs/outcomes in each group). Models were adjusted for stratifying variables and the three-category BADLS variable for dependency (see Chapter 4). Multilevel mixed-effects linear models were estimated by maximum likelihood, assuming that data were missing at random on the response variable. In other words, missingness was assumed to be dependent on model covariates or on previous or following responses, had they been observed, but not on the missing responses. 73,74 Cases at 52 and 104 weeks were considered available for analysis if baseline outcome/cost data and at least two follow-up data points were available; at 24 weeks, cases were considered available if baseline outcome/cost data and at least one follow-up data point were available. Bias-corrected bootstrapped CIs of the model estimates of cost and outcome differences were obtained (5000 resamples).

Cost-effectiveness analyses

Incremental cost-effectiveness ratios (ICERs) were calculated for each outcome. ICERs were calculated separately for the difference in the EQ-5D outcome at the 24-, 52- and 104-week follow-ups, and for QALY-adjusted institutionalisation-free days. The ICER was defined as the difference in mean costs incurred by the intervention and control groups (ΔC), divided by the difference in mean outcome (ΔE) between the treatment groups. The ATT intervention can be interpreted as representing value for money if the ICER is below some threshold of WTP for a unit of additional effectiveness, λ:75

A full package of ATT can be considered cost-effective if (1) the package is significantly more effective and less expensive than a basic package of ATT or (2) ATT is significantly more effective and more expensive, but the payer is willing to pay the additional cost (up to λ) to achieve the additional effectiveness or, possibly, (3) ATT is significantly less effective and less expensive, but the payer considers the sacrifice of some effectiveness worth making because of the savings that could be achieved. ATT can be considered, unambiguously, to be not cost-effective if it is both significantly less effective and more expensive.

Incremental net monetary benefit75 can be expressed as a rearrangement of the decision rule in (1):

This is the monetary value of gains in outcomes associated with the treatment at a given WTP threshold, net of (less) the additional cost of providing the treatment. 76

Cost-effectiveness acceptability curves (CEACs) were produced when the intervention strategy was more effective and costs were lower. Estimates of cost and outcome differences were obtained by non-parametric bootstrapping of regression estimates, producing 25,000 replicates in the case of the institutionalisation-free days and QALY outcomes and 5000 replicates in the case of the change in EQ-5D index outcome. The proportion of replicates for which the net monetary benefit was positive was graphed over a series of WTP values from £0 to £50,000 to produce CEACs. The current NICE WTP threshold for the adoption of new technologies is between £20,000 and £30,000 per QALY. 77

A sensitivity analysis explored the impact on societal costs of valuing time spent by caregivers in providing care to the person with dementia at replacement cost (the hourly cost of a home care worker) at 104 weeks.

Results

Sample numbers

The flow of dyads who completed assessments is given in Figure 10. As can be seen, some dyads declined to participate in full assessments involving the completion of participant-/proxy-/caregiver-reported measures, but agreed to remain in the study and provide more limited information on community residence and SAEs or AEs (by telephone follow-up). There were 412 dyads who participated in the baseline and at least one follow-up assessment. A small number (intervention, n = 11; control, n = 14) did not participate in an assessment at any point and 45 (intervention, n = 19; control, n = 26) participated only at baseline. A substantial proportion of the 12-week follow-up assessments were not conducted: 20% in the intervention group and 17% in the control group. The numbers of dyads contributing data to the cost-effectiveness analyses varied depending on the measures and the analysis; valid numbers of observations associated with each measure are presented with the results of the analyses (see Cost-effectiveness analyses). Demographic characteristics of the sample participating in full assessments at baseline are given in Appendix 4.

FIGURE 10.

The ATTILA trial flow of dyads. Delayed: researcher noted that the assessment had been delayed enough that the date of the assessment was closer to that of the next scheduled assessment than to that of the previous assessment or randomisation screening visit.

Outcomes

The mean EQ-5D participant-reported index scores (Table 15) were higher than the proxy-reported scores at all time points. The mean participant-reported baseline and week 12 scores were similar between groups, but at 24, 52 and 104 weeks, the mean scores for the intervention group were significantly lower than the scores for the control group (p < 0.05 in each case). The mean scores of the proxy-reported measures did not differ between groups.

Cost-effectiveness analyses

Sensitivity analysis: replacement cost of unpaid care

The impact on societal costs and the ICER of valuing unpaid care at replacement cost at 104 weeks was explored in a sensitivity analysis. The costs of unpaid care and societal and intervention costs using replacement costs are presented in Appendix 7, Table 36. Cost-effectiveness analysis results are presented in Appendix 7, Tables 37–39.

Discussion

The additional 3-month cost of a full ATT package over a basic package at each time point was modest (ranging from £11 at baseline to £21 at week 12). Although the cumulative cost estimated for the full package was significantly greater than that of the basic package (an additional £119), the average costs of ATT were modest in both the intervention and control groups (£408 and £288, respectively). Mean (unadjusted) societal costs were in the order of three times that of health and social care costs. Total health and social care and societal costs over 104 weeks, taking censoring into account, did not differ between groups. Average 3-month costs estimated in difference-in-difference analyses did not differ between groups at 24, 52 or 104 weeks. Change in proxy-rated and participant-rated EQ-5D index scores estimated in difference-in-difference analyses also did not differ between groups at those time points. For the institutionalisation-free days outcome, there was no statistical difference between the intervention and control groups in days spent in the community. Groups did not differ on proxy-reported QALYs. The mean participant-reported QALY was lower in the intervention group than in the control group at the 52- and 104-week time points. The probability of cost-effectiveness on a measure of change in proxy-rated health-related quality of life was greater in the short term than in the long term from the health and social care and societal perspectives, at a WTP of £20,000. A similar pattern was seen in proxy-rated QALYs from the health and social care perspective, but the probability of cost-effectiveness was higher in the long term from the societal perspective.

Some limitations must be acknowledged. Costs were measured by asking caregivers to retrospectively report service use, so the results could be subject to recall bias. The costs accrued during the intervals between periods with CSRI data available were estimated by carrying forward most costs, although the costs of inpatient and emergency department use were estimated from the SAE data available over these times. Costs of regularly used services were thus assumed to be constant over these intervening periods. Participant-reported EQ-5D data were missing for one-quarter of intervention group participants and for one-third of control group participants who had participated in assessments at 104 weeks. Although difference-in-difference analyses controlled for baseline covariates (stratification variables and dependency, as measured by the BADLS) that might have accounted for differences in data availability between groups on this measure, QALY analyses drew on group mean utilities at each time point and did not adjust for baseline characteristics. Therefore, the finding that the intervention group had fewer QALYs derived from participant-reported EQ-5D must be interpreted with some caution as substantial rates of missingness in that measure could be a concern. Generalisability of these findings may also be limited because all assessment data were missing from some participating dyads at the baseline and follow-up points (8% of the sample at baseline and between 16% and 19% of the community-dwelling sample still participating in the study over the follow-ups did not participate in an interview).

The study sample size was powered to detect a delay to entry to permanent care home residence of 55 days over 24 months. As can be seen from the measures of uncertainty around all point estimates of cost differences, there was substantial variability in the cost data (as might be expected). It is possible that, because the variance around the mean difference in costs was of a greater magnitude than the variance around the survival outcome, a larger sample size would have been required to detect a between-group difference in costs than in days.

The mean EQ-5D proxy-rated index scores were consistently lower than index scores derived from the participant-rated version of the measure. This pattern is in line with that reported in other studies involving people with dementia and caregiver proxy raters. 81 The EQ-5D has been reported to be more reliable than some dementia-specific quality-of-life measures in terms of agreement between caregiver proxy-rated and self-rated total scores, in a population with mild and moderate dementia participating in a psychosocial intervention. 82 The divergence we observed between proxy and participant raters’ EQ-5D and QALY outcomes may relate to selection effects, validity of the instrument when used for people with more advanced dementia and genuine differences in perspectives between raters. 81 The participant–proxy inter-rater reliability of the EQ-5D index scores from this study merits further investigation, as does the response rate of participants at different stages of dementia.

Conclusions

The results suggest that the health and social care and societal costs of a full package of ATT were not greater than those of a basic package of ATT for people with dementia. There was little evidence that the intervention influenced the length of time lived in the community over 104 weeks, improved health-related quality of life as rated by participants and proxies or increased participants’ QALYs based on quality of life as rated by proxies. The results suggest that the intervention reduced participants’ QALYs based on quality of life rated by people with dementia themselves. Generalisability of findings based on the dyad-reported outcomes data is limited by the extent of missing data at the individual outcome measure and at the assessment level.

Aims

We aimed to assess and compare the use of ATT with usual care for people with dementia on caregiver psychosocial and health outcomes, namely caregiver burden, depression and anxiety.

Methods

Results

The findings reported here consist of the first three waves of data collection in the ATTILA trial: at baseline and at 12 and 24 weeks. All LMM analyses in each section were adjusted for each of the demographic variables presented in Tables 21 and 22. The tables below (see Tables 23–26) provide the adjusted means of both groups on each outcome variable. We conducted a whole-sample analysis and two secondary subgroup analyses. The results are broken up into three sections:

• section 1 – all caregivers in the study

• section 2 – caregivers who were living with the person with dementia (co-resident)

• section 3 – caregivers who were partners of the person with dementia (spousal care partners).

Section 1: findings for all caregivers

Section 2: spousal caregivers who were living with the person with dementia (co-resident)

The sample of caregivers in the ATTILA trial were diverse in their living arrangements with the person with dementia. Some lived in the same home as the person with dementia. Others were resident in proximity, but living separately, and some were located some distance from the person with dementia. As there is some evidence that co-location of the caregivers with the person in receipt of care implies a more involved role in caring for the person with dementia,120 this group was selected for secondary analysis, whereby effects of the telecare intervention on caregivers’ outcomes were examined only for those who were living in the same household. Details of the group in this substudy are presented in Table 27.

The analyses presented in the following sections mirror those conducted for all caregivers in the previous sections.

Section 3: caregivers who were the spouse or partner of the person with dementia

Much of the literature on caregivers has selected spousal care partners for special attention, and there is evidence that spouses are particularly vulnerable to burden and poor psychological well-being. 121–123 Spousal caregivers of people with dementia have to live with an altered relationship with the person they elected to partner and live with as the mental faculties of the person with dementia alter. Furthermore, spousal carers may be of similar age to the care recipient and, therefore, may be experiencing physical or cognitive difficulties themselves. A secondary analysis was conducted on spousal caregivers of the person with dementia in the sample, so as to keep the relationship between the caregivers and the person with dementia constant. Details of their age and the SMMSE score of the person with dementia are provided in Table 28.

Discussion

The impact on caregivers’ health and well-being of caring for someone with dementia has led to the development of interventions to reduce the burden they face. It has also been recognised that these interventions may have a broader impact, as the alleviation of caregiver burden may reduce the likelihood of the cared-for person being institutionalised, thereby reducing social and health-care costs. In this study within the ATTILA trial, we investigated the effects of using ATT for people with dementia on the outcomes of the participants’ caregivers, namely caregiver burden and psychological well-being in the first 24 weeks following installation of the ATT intervention.

No effects of the installation of ATT compared with usual care were found on caregiver burden and its subcomponents, or depression and state anxiety. We also conducted post hoc subgroup analyses among live-in caregivers, and among caregivers who were the spouse or partner of the cared-for person, in whom we might expect poorer psychological well-being and levels of burden. Neither of these subgroup analyses revealed differences between the two groups in any of these outcomes. It is notable that the levels of caregiver burden, depression and anxiety remained stable during the course of the trial. Although this is not a non-inferiority trial, the data suggest no negative impact of receiving the ATT interventions.

One explanation for the lack of impact on these outcomes is the relatively low levels of burden, depression and state anxiety reported in the sample at baseline. 124 The mean levels of burden across both the intervention and control groups for the overall sample and the examined subgroups were in the mild to moderate range. Similarly, mean levels of depression in this sample were below the clinically relevant threshold on the CES-D-10 scale, for which a score of > 10 indicates depression. Mean scores on the STAI-6 at baseline were 9.6 and 8.7 for control and intervention group participants, respectively. These data suggest that there may have been limited scope to reduce burden, depression and anxiety among the sampled caregivers. A previous study125 indicated higher levels of depression and anxiety at baseline in its study population, and a 2016 study,126 using the same instrument for assessing depression as used in this study, found higher scores, above the clinically relevant threshold, in its sample.

Alternatively, it is possible that the effects of ATT interventions may be limited in effecting change in these outcomes, and that interventions specifically targeting caregivers’ well-being may be more effective than those aiming to support the cared-for person. Meta-analyses indicate that caregiver-directed interventions have demonstrated effectiveness in reducing burden, depression and anxiety in this population. Effective interventions include cognitive–behavioural therapy, cognitive reframing and educational interventions. 127–130 Therefore, to optimise the benefits of the installation of ATT for both the care recipient and the caregiver, it may be important to provide additional caregiver-directed support. Effective and potentially low-burden and low-cost interventions include the use of telephone- and internet-based interventions to improve caregiver outcomes. 131,132

In the current sample, the mean scores on the SMMSE indicated moderate levels of cognitive impairment in the cared-for participant sample (mean 10–19). There is some evidence indicating that the severity of the dementia is related to levels of depression and anxiety, with only severe dementia leading to caregivers having high levels of depression and anxiety,133 although this relationship has not always been supported. 134 It is possible that the level of dementia in the current study was not sufficiently severe to produce high scores at baseline in the caregivers such that they may have been reduced by the intervention.

Strengths, limitations and suggestions for future research

This study provides the first insight into the potential impact of up-to-date technological interventions for people with dementia on outcomes for their informal caregivers within 6 months of deployment. A large data set to assess impact of ATT is available; however, the extent of missing data at follow-up precluded investigation of longer-term effects of the technology on caregiver outcomes. Furthermore, loss to follow-up in the caregiver data set was non-random, introducing some degree of bias. This is because dropout among some caregivers was partly due to the care recipient moving into residential care or dying. Furthermore, power analysis was conducted on the study primary outcome (time to institutionalisation), rather than on caregivers’ outcomes. Therefore, it is possible that our analyses were statistically underpowered.

A general limitation of the trial should also be recognised, as we have reported elsewhere51 that there was limited fidelity of technology deployment in relation to the recommendations arising out of the needs assessment. This mismatch may have also contributed to the lack of any impact on caregiver burden and mood.

Future work will need to determine the minimum sample size required to detect an effect of the ATT intervention based on expected effect size for caregiver outcomes. It may well be that longer follow-up times with additional caregiver support are necessary to produce any intervention effects for caregivers’ outcomes.

Conclusions and implications for practice

This study provides a first insight into the potential impact of ATT on caregiver burden, psychological well-being and quality of life. No impact of ATT on caregiver burden, depression and anxiety was identified. As key drivers of transfer to long-term residential care, interventions aiming to specifically target these aspects of caregiver psychosocial well-being along with the deployment of ATT may be important for delaying institutionalisation and preventing the costs associated with this, both for social care services and for the individual and their family members. Evidence-based effective interventions to prevent negative impact of caregiving may include care partner-directed psychological and educational techniques, as well as ensuring that caregivers have an appropriate understanding of the role of ATT.

Aims and research questions

The study aimed to exemplify and examine how and why people with dementia and their caregivers used or chose not to use ATT in their lives and the ways in which ATT use affected their environments and relationships.

To address this research aim, the study team sought to answer three research questions:

1. How and why do people with dementia and their caregivers use ATT at home?

2. How does ATT fit into people’s lives and care in their homes?

3. How does ATT affect people’s lives and care in their homes?

Methods

This study used a qualitative ethnographic observational, longitudinal design to investigate how and why people with dementia in the intervention arm of the ATTILA trial used or did not use the ATT offered to them. This section describes the study design, sampling strategy, data collection and analysis methods.

Design

Ethnographic approaches have commonly relied on sustained fieldwork, whereby a researcher takes part in a group’s practices while observing them, so as to rigorously interpret how people make sense of their everyday lives and social systems. 135 The ethnographic approach used in this study, however, drew on recent multidisciplinary research in trials to design a study ‘embedded’ in the ATTILA trial136,137 to collect focused138,139 observational data on situated practices of people with dementia and their caregivers when using ATT in their everyday lives. We used these data to construct in-depth extended cases140,141 of how people with dementia and caregivers used (or chose not to use) these technologies, to help explain how and to what extent specific technologies for supporting people with dementia may be relevant in the context of their everyday places and interactions in home and community settings.

This study demonstrates how technology-enabled dementia care systems work with and for people, how they attempted to fit ATT into their lives and homes and how ATT-related processes affected their activities and care relationships.

To provide ‘trustworthy’142,143 findings and interpretations, we formulated credible, transferable, dependable, and confirmable processes for data collection and analysis through:

• Extensive and intensive data collection from participants (see Participant observation and Analysis) and critically discussing anonymised data (credibility) findings within the study teams (credibility, dependability).

• Noting in contextualised detail participants’ situated practices and how noting these may affect interpretation of findings to other contexts (transferability).

• Reflecting on researcher role in the process and how this could affect interpretations. Detailed fieldnotes and accounts of the research context to allow future research to confirm, challenge or otherwise build on findings from this study (confirmability).

Ethics approval for the study was granted by the National Research Ethics Services Committee East of England, Norfolk (reference number 15/EE/0015) on 3 February 2015.

Sample

A purposive sampling strategy was used to select potential participants from the wider ATTILA trial population who were able to provide data relevant to examining care practices and specific reasons for, and ways of, their uptake (or not) of diverse technological interventions.

This strategy, therefore, had to incorporate the ATTILA trial inclusion and exclusion criteria (see Chapter 2), but also ethnographic-specific purposive sampling criteria that would provide contextually relevant and diverse types of participants’ experiences from three characteristics:

1. severity of the person’s dementia, as recorded by ATTILA trial research workers

2. type of family relationship between the caregiver and person with dementia

3. types of ATT equipment provided to the person with dementia.

Settings

To recruit participants, we collaborated with local ATTILA trial research workers, who collected data across three distinct authorities in east and south-east England: ‘Shire’, ‘Metropolitan’ and ‘Coast’. They included a mix of urban and rural populations in areas of economic wealth and areas of economic deprivation. 144 These are pseudonymised to ensure anonymity and confidentiality:

• ‘Shire’ – participants from this area lived in large villages or small market towns with independent vendors and occasional high-street shops.

• ‘Metropolitan’ – a major city divided into different districts. Trial participants lived in two adjoining districts with historical reputations of poverty. More recent gentrification is transforming these areas.

• ‘Coast’ – two counties with a seaside border, sharing service responsibilities for adult services. Participants from this area lived predominantly in large market towns or in villages near major regional city hubs.

These brief setting descriptions highlighted distinct features of the wider areas where people with dementia lived to contextualise wider relations with their environment.

Recruitment

The ethnographic team recruited prospective participants alongside three ATTILA trial research workers responsible for recruiting and collecting data from ATTILA trial participants in Metropolitan, Shire or Coast. This process to embed ethnographic recruitment in the ATTILA trial involved four steps:

1. The fieldworker (ML) collaborated with an ATTILA trial research worker to identify prospective ACCOMMODATE participants from the existing ATTILA trial sample. The fieldworker selected people based on how they aligned with the purposive sampling criteria for ACCOMMODATE. He selected participants to ensure a maximum variation across all three purposive sampling criteria.

2. The fieldworker attended the pre-arranged ATTILA trial follow-up visit with the area’s local research worker to meet with prospective participants to discuss participation in the ethnographic substudy.

3. The fieldworker sought substudy informed consent from people with dementia and their caregiver to take part in the ethnography substudy, or a consultee declaration from the caregiver for the person with dementia to participate.

4. The fieldworker continually renegotiated informed consent or consultee declaration for each subsequent monthly visit during independent fieldwork.

Participant observation

The study included nine ethnographic cases, each consisting of at least one person with dementia (n = 10) and their caregiver (n = 10). Two cases included more than one caregiver (the Campbells included two caregivers) or more than one person with dementia (the Stewarts included two people with dementia). Data collection involved up to six visits over 6 months (one per month) to the home of each person with dementia. These monthly visits lasted between 1 and 5 hours (mean 3.5 hours). A total of 208 hours of observation took place over 60 visits. Each visit involved Matthew Lariviere observing the practices of people with dementia and their caregivers, focusing particularly on whether or not and how they were using ATT and unstructured ethnographic interviews with participants to elicit their reasons for using or not using ATT.

Analysis

Initial notes from observations were written in a journal during or immediately after the visit and maps were drawn to illustrate objects and people’s places in settings. The initial notes were later transcribed and developed into field notes with any accompanying audio-recordings from conversations. Additional researcher memos identified where participants were interacting with or discussing the technologies.

We analysed each case using situational analysis of longitudinally extended cases. 8,11 The main themes identified from early memos highlighted how people with dementia and caregivers attempted to fit ATT into their everyday practices, leading to placing, replacing and displacing care and uses of spaces inside and outside the home. Focused coding was used to identify instances of these themes during different visits for each ethnographic case, which informed comparisons within and across cases and contextualised specific instances of these analytical themes. The findings are presented in the following section as extended cases that highlight everyday features seen as common to several cases. These are depicted further through indicative maps.

Findings

Each of these ethnographic cases provides in-depth representations of how participants enacted their practices with ATT in their homes. Table 29 provides descriptions of the nine ethnographic cases to highlight the diversity of the nine cases based on their case name, location, the severity of the person’s dementia, the nature of participant’s family or care relationship, and the types of assistive technologies and/or telecare devices in place. Ethnographic cases were evenly distributed across different levels of dementia severity (i.e. three mild, three moderate and three severe cases). Participants in the ethnographic substudy most commonly received a falls detector (n = 6) and door sensors (n = 4). Keysafes (n = 3) and calendar-clocks (n = 3) were also relatively common. In most cases, an adult child was caring for an older parent with dementia (n = 6). Caregivers rarely co-resided with the person with dementia under their care (n = 3).

This diverse range of characteristics across all cases supported robust interpretation of how this diversity shaped differences and similarities across the nine ethnographic cases. Three key themes were identified as relevant to understanding the ATT-relevant and care-relevant everyday practices, routines and relationships of study participants:

1. Placing technology in care.

2. Replacing care with technology.

3. Technology displacing care and everyday life. Displacing everyday life refers to how people’s everyday routines and built environments become disrupted and altered.

Placing technology in care

The theme ‘placing technology in care’ refers to instances when people with dementia and/or caregivers fit ATT products into their existing care arrangements. It addresses participants’ processes and practices in incorporating and adapting technologies into their everyday lives with varying degrees of success.

The Drapers showed initial troubles in how they placed a falls detector within their pre-existing everyday practices. The person with dementia, Violet Draper, received a neck-worn falls detector pendant, after experiencing several falls. Around the when time research visits began, she had another fall, but the alarm did not trigger. She decided not to trigger the alarm manually as she ‘did not want to be a bother’ to her son and caregiver, Thomas, or to the emergency response services. Thomas also commented that his mother frequently forgot to wear the pendant or took it off in the evening with her bed having been moved down to the sitting room (Figure 11 shows a map). After this fall, Thomas changed two elements of his mother’s care. First, he asked the local ATT provider to swap the pendant-style detector for one worn around the wrist. Second, he reminded Violet every day to ‘press the button’ if she ever fell again. A couple of months later, Violet had another much more serious fall in which she broke her leg; this time, the falls detector did not detect the fall automatically, but Violet did remember to press the falls detector button to contact first responders and Thomas.

FIGURE 11.

Map of sitting room in Violet Draper’s home.

This case highlights that both Thomas and Violet acted to make the falls detector fit into their lives. Their case illustrates the work of caregivers to instruct and to reinforce any such instruction to ensure ‘successful’ implementation of ATT that would otherwise be invisible to care workers and designers of the technology. Such work was even more important here because the falls detector did not activate automatically. This raises further questions about the different types of technologies designed to enhance the safety of people with dementia. The key distinction here is between ‘passive’ devices that automatically trigger, not requiring the user to perform any actions, and devices that require an action to be performed to activate them. It seems important to consider what use of passive devices is appropriate in the case of people with dementia, especially where they are either reluctant to trigger alarms or have memory difficulties that lead them to fail to remember to trigger a device. This case demonstrated how social connections and support were important even for enacting ‘technology-enabled’ care systems.

In contrast, the Stewarts’ case illustrates how people with dementia or caregivers may appropriately place assistive technology, yet find that other objects in the home may suitably address problems when they arise. Mary and Michael are a married couple, both of whom have dementia. Their daughter, Sally, moved them into an annex of her home to support them full time. During one fieldwork visit, Sally asked her parents for the date. Neither Michael nor Mary knew the date. They also did not appear to notice the nearby calendar-clock that had been provided to display this information but did perceive wooden calendar blocks across the room, which helped orientate them. Michael told Sally the correct date.

This case highlights the importance of appropriately placing ATT in a person’s home. Here, a wooden calendar may be seen to serve as more ‘assistive’ than the ‘formally provided’ calendar-clock. Although the calendar-clock did not disrupt or disorientate Michael or Mary, the technology did not actively facilitate them to be orientated to date and time. Someone still had to interact with the regular calendar to make it work by changing the date each day, but its familiar location and design may have more easily supported their orientation because it relied in part on older memory and was perhaps more readily recognisable than more recent digital counterparts.

The Browns’ case raises further questions about how researcher and practitioners come to define ‘use’ of ATT. Sam Brown, a person with mild cognitive impairment, had a memo minder in his house entrance. A recording of his daughter’s voice reminded him to lock the front door whenever anyone walked in front of the infrared motion sensor. Sam told the researcher that he always remembered to lock the door because of it. Sam also shared his home with his wife and with his adult daughter and son-in-law, who lived in a converted garage annex. The other household residents became annoyed with the memo minder repeatedly going off whenever they went to put on or remove their shoes and outerwear. Sam decided to turn off the recording but leave the memo minder in its place next to the front door. He insisted, and the researcher observed and confirmed, multiple times, that seeing the now-silent memo minder, still beside the door, reminded him to close and lock it when he left the house to go back home.

Such practices blur ‘use’ of ATT and its ‘non-use’. Although the person with dementia switched off this device, its co-location with him in its ‘appropriate’ place provided the prompt he needed to remember to lock up. This case, alongside others, illustrates both the importance of ‘place’ in sustaining appropriate technology-enabled care for people with cognitive impairments, and how people are able to actively accommodate technology to work in their shared spaces and their relationships with others, including their caregivers.

Technology replacing care

This theme, ‘technology replacing care’, particularly addresses how caregivers, through ATT, replaced or reconfigured their practices of caring for people with dementia.

The Clydes’ case involved Arthur Clyde, an older person with dementia, who received a falls detector from his local authority. His son and daughter-in-law, Mark and Cathy, visited his home every weekday to work from the front room of his house, which they had converted into an office for Mark’s business. Mark also used to visit his father at least 1 day over the weekend to see whether or not he remembered to heat up and eat his pre-prepared meals. However, after Arthur started to wear the falls detector around his wrist, Mark visited his father less frequently. Mark told the researcher that he had ‘peace of mind’ that the call centre would notify him if his father had a fall. Mark decided instead to telephone Arthur on Saturdays and Sundays to ask him whether or not he had eaten his meals instead of visiting to confirm this.

This case illustrates how caregivers may change their care practices for a person with dementia after they introduce ATT into their arrangements. Here, the caregiver visited his father less frequently and relied on the falls detector and telephone to monitor his father with dementia. Falls detectors and telephones reconfigure monitoring practices to be mediated through technologies rather than face-to-face interactions. Caregivers’ sense of security, often articulated as their ‘peace of mind’, was a common response across cases, including the case that follows.

In the Smiths’ case, Lauren had the local service provider install an activity monitoring system in the living room of her father’s bungalow. Lauren thought her father, Christopher Smith, frequently got up from his favourite chair to walk around the house based on activity reported on the monitoring system’s accompanying application for her tablet. Lauren told the researcher that she had ‘peace of mind’ that her father remained active even when home alone, especially as she lived in another county, distant from her father. However, the researcher rarely saw Christopher move from his chair during his visits (Figure 12 shows a map). During one visit, Lauren and the researcher noticed the dog jumping on the couch. The activity monitoring system’s motion sensors happened to be at the same height as the dog. The researcher asked Lauren whether or not she possibly monitored the dog instead of her father. She looked at the application and noticed recent activity when no one else appeared to have moved in the room except for the dog. She wondered aloud how frequently her father really left his chair.

FIGURE 12.

Map of Christopher Smith’s bungalow.

The Smiths’ case highlights how people use the devices for reassurance and peace of mind. In this case, the poorly placed product led to inaccurate information and misguided reassurance. Once this was established, the caregiver’s ‘peace of mind’ became replaced by concern as she could no longer be certain whether the activity monitoring system monitored only her father’s movement or also that of other people or animals.

In contrast to the other two cases exemplifying this theme, the case of the Campbells demonstrates how people can independently adopt ‘other technologies’, and how this will also shape their care practices. Kenneth Campbell shared his home with Lillian, his mother living with dementia. They were offered door and bed sensors, but Lillian tore out the cable of the bed sensor from under the mattress. They did not have a telephone line for the door sensor to connect with. Instead, Kenneth independently purchased and used a closed-circuit television (CCTV) system to monitor the downstairs rooms of the house, where his mother lived, through monitors in his living room upstairs.

The Campbells’ case displays marked changes in how caregivers may provide care with the addition of technologies. Notably, here, the ATT did not appear to fit into the lives of Kenneth or Lillian. Rather than reconfigure their practices or home, Kenneth adapted security equipment, namely CCTV, as a means to monitor his mother in their home. This case raises further questions about how we characterise means to monitor people with dementia in their home as appropriate yet still ensure dignity and safeguard them against harm. It also calls into question, as noted in Chapter 6, whether or not carers’ work here may have changed rather than diminished.

Discussion

These findings illustrate how technological mediation through ATT with associated applications and screens could replace, displace and disrupt co-located, face-to-face interactions. The reported findings suggest that how policy-makers and industry imagine community care through ATT service provision may not always reflect actual practices in technology-enabled dementia care. As previous research has also suggested, people with dementia and caregivers may make ATT work for them in contingently, rather than systematically. 145–147 This study, however, highlights not only the role taken by caregivers to fit these technologies into care practices, but also how their use of ATT can change the spaces and placement of care and everyday life. It shows how people’s practices with ATT may shift dependencies in care arrangements as a result of actual or perceived changes or disruptions in care. 147,148 These findings identify limitations for ATT to enable people with dementia to ‘live independently in the community’ if they do not also have added support from caregivers to help adapt technologies to fit into their lives.

Conclusions and implications

These ethnographic findings flag up unintended and unanticipated consequences for ATT implementation and uptake in ‘real-world’ community-based dementia care contexts. The ethnographic approach details how people’s use of ATT shifted over time. Nonetheless, even temporary use of ATT may have deferred more complex and more acute care crises for the person with dementia or caregiver. Transient effects or limited engagement with technology should not necessarily be interpreted, therefore, as a failure in its uptake or effect. It underlines the need to identify and map the context of ATT provision over time within the changing lives of people with dementia and their caregivers, and relative to service provider organisations, as revealed in these cases.

These findings have relevance for technology developers and providers in indicating how to consider more appropriate products and services that can support more sustainable technology-enabled care for people with dementia and their partners in care. Designers and service provider organisations should also work with caregivers and people with dementia, including charities representing these groups, such as Carers UK and the Alzheimer’s Society, to co-produce suitable technological interventions.

This study suggests that we must more fully appreciate the importance of people’s activities and relationships in continuously shaping care, including dementia care involving the implementation and uptake of technology in community settings to improve their effectiveness and sustainability.

Main outcomes

To the best of our knowledge, this study is the first RCT of ATT for people living with dementia in the community. Despite the large increase in available products and their promotion as effective, cost-saving ways of helping people with dementia remain at home, there has been a paucity of robust research on how beneficial these products and services can be. 37 The main outcomes of this trial were whether or not ATT could help keep people living at home longer and whether or not this was more cost-effective than a basic ATT package and alternative support. The answer to both questions was no: there were no significant differences found between the ATT and groups on any outcomes.

Home-based dementia care is a policy prioritised across the world,149,150 including in the UK. 151 Remaining at home to maintain a higher quality of life is preferred by people with dementia, is more cost-effective than residential care and can provide continuity and familiarity that is beneficial to those experiencing a decline in cognitive and functional abilities. 152 ATT has been marketed as an aid to keeping people with dementia safely at home by providing monitoring of movement, safety and well-being; reminders for events and medication; and social support, among other things.

There have been many studies conducted around ATT, but all have been very small and/or have had poor methodology. The results of these have often been conflicting, with some viewing ATT very positively,153 others finding no benefits154 and some even regarding it negatively, for example although surveillance can be enabling, it can also be an invasion of privacy. 8 A 2017 Cochrane Review37 found that there were no studies sufficiently robust to make a judgement and cited our study as being the only ongoing research that met their criteria. The lack of evidence to date makes the contribution of this study to the evidence base vital.

Dementia is an increasing challenge for health-care systems across the world, with 43.8 million cases in 2016;11 this is predicted to rise to 75 million by 2030. 155 With no treatment for dementia, or many of its causes, currently available, and medication to slow development showing only marginal clinical improvement,156 it is increasingly important to help people with dementia and their families to live well in the community for as long as is safe for them to do so. Caregivers provide 42% of the estimated US$604B spent globally on dementia,37 so supporting their role is also vital.

There are a number of possible reasons for the lack of efficacy shown in the current study. Previous research has suggested that ATT are often introduced when the cognitive decline is too great and the ability to adapt is too impaired for the user to incorporate it into their routines, and, at all stages of dementia, devices are seen as most useful when they are simple. 157 Aids that do not need adjusting have been considered most useful by caregivers, whereas complex interfaces and connection problems reduce the perceived utility of ATT. 158 However, other studies have found that a lot of ATT devices lack user-centred design approaches,159 and user experience can be improved by optimising specific details of a product. 160

Limitations

There was great variation across sites in the assessment and provision of ATT, and the study was not powered or intended to examine the impact of local configurations of ATT. The trial was designed to embrace the heterogeneous nature of local ATT assessment and equipment provision for people with dementia. As discussed in Chapter 3, the way in which ATT was delivered locally varied considerably: two types of value networks were found to be operating in our relatively small sample of local authority areas. A mixture of models of providing and financing ATT was encountered across trial sites. The extent to which councils and providers charged user fees varied depending on the nature of local markets and on the needs profiles of ATT users. The extent of information on locally available ATT services and charges is limited in the UK; most ATT services, including ASCDs, charge fees. 161 We did not investigate the impact of deficiencies in service information or the presence of user charges, although such barriers might have affected the uptake of ATT and reduced the use of potentially beneficial devices.

Delivery systems were investigated and described as planned (see Chapter 3 and Appendix 2), but it proved difficult to collect data from sites in a consistent manner sufficient to calculate a unit cost of ATT. The unit costs of ATT devices, installation, monitoring and response were drawn from databases of prices set by public sector procurement frameworks. This approach allowed a realistic costing of ATT services and equipment provided by the public sector (e.g. in comparison to pricing devices via telecare providers’ and online technology retailers’ websites); prices paid for ATT devices and services by private individuals might be higher, and so ATT costs, which made up a very modest proportion of participants’ total costs, might have been somewhat higher. Because control participants were also in receipt of at least a basic ATT package over the course of the study, their costs could equally be higher and differences in costs between groups little affected overall. We explored the feasibility of collecting providers’ data on ATT devices used by participants; however, as the study was not resourced to negotiate data-sharing arrangements with all the providers involved, we relied on researcher-collected technology checklist data.

The CSRI was reliant on participant recall, which may have affected the precision and size of cost estimates. Costs were measured by asking caregivers to retrospectively report service use, so the results could be subject to recall bias. Costs of the intervals between periods with CSRI data available were estimated by carrying forward most costs, although costs of inpatient and emergency department use were estimated from the SAE data available over these times. Costs of regularly used services were thus assumed to be constant over these intervening periods. Participant-reported EQ-5D data were missing for one-quarter of intervention participants and for one-third of control participants who had participated in assessments at 104 weeks. Although difference-in-difference analyses controlled for baseline covariates (stratification variables and dependency, as measured by the BADLS) that might have accounted for differences in data availability between groups on this measure, QALY analyses drew on group mean utilities at each time point and did not adjust for baseline characteristics. Therefore, the finding that the intervention group had fewer QALYs derived from participant-reported EQ-5D data must be interpreted with some caution, as substantial numbers of missing data in that measure could be a concern. Generalisability of these findings may also be limited because all assessment data were missing from some participating dyads at the baseline and follow-up points (8% of the sample at baseline and between 16% and 19% of the community-dwelling sample still participating in the study over the follow-ups did not participate in an interview).

Future research

An official national collection of data on local authority ATT provision, the size of contracts with provider partners and total expenditure and unit costs would shed light on the variety of delivery and financing models in place, and the relationship with expenditure. A study on the nature and size of the private ATT market (including internet of things and smart home applications) would shed light on the relative importance of public and private sectors in ATT provision. 161

An observational study of telecare for dementia could be designed, which could employ a larger sample than would be achievable with an experimental design, and observe longer and continuous durations of service use. Such research could also identify processes and mechanisms of implementation (procurement, clinical and care decision-making, maintenance of technologies, call centre activities, etc.) of ATT in health and care systems to support people with dementia. If routine data from health and social care providers could be linked162 to telecare providers’ data, further research could examine associations between ATT (activations and device types) and hospital and care home admissions.

Implications

This study found that a full package of ATT did not result in a significant increase in the length of time a person with dementia can remain living in the community, nor did it achieve decreases in caregiver burden, depression or anxiety. The findings of Chapter 3 suggest that providers of assessments and providers of ATT do not always work in tandem. Work to understand why local delivery systems were producing such mismatches should be conducted with local authorities, not-for-profit providers and the NHS in England. Meanwhile, the use of ATT that extend beyond basic devices (including pendant alarms) as support for people with dementia should not be unconditionally adopted and should be subject to the same rigorous evaluation as other interventions. However, ATT recommended on the basis of need and installed and used as intended might be very useful for people with dementia and their caregivers. Designers and service provider organisations should work with caregivers and people with dementia and their advocates to co-produce suitable technological interventions. Alternatives to technological solutions to difficulties should also be actively sought.

The ATTILA trial triallists group

Contributions of authors

Rebecca Gathercole (https://orcid.org/0000-0002-6380-2655) (Trial Manager, Old Age Psychiatry) ran the trial, contributed to the paper, interpreted the data and wrote the initial draft paper.

Rosie Bradley (https://orcid.org/0000-0002-0758-4905) (Medical Statistician, Medical Statistics) analysed the survival data, contributed to the paper, interpreted the data and wrote the initial draft paper.

Emma Harper (https://orcid.org/0000-0001-5651-6258) (Clinical Trials Co-ordinator, Population Health) ran the trial.

Lucy Davies (https://orcid.org/0000-0001-6632-7215) (Medical Statistician, Medical Statistics) analysed the survival data.

Lynn Pank (https://orcid.org/0000-0001-6398-6565) (Administrative Assistant, Medical Statistics) and Natalie Lam (https://orcid.org/0000-0001-8591-444X) (Data Manager, Medical Statistics) provided data management support and data entry at the Clinical Trials Unit.

Anna Davies (https://orcid.org/0000-0003-0743-6547) (Senior Research Associate, Health Psychology) analysed the caregiver outcome data, contributed to the paper, interpreted the data and wrote the initial draft paper.

Emma Talbot (https://orcid.org/0000-0002-7614-9683) (Research Practitioner, Mental Health Research), Emma Hooper (https://orcid.org/0000-0002-4059-6035) (Research Practitioner, Mental Health Research), Rachel Winson (Research Practitioner, Mental Health Research), Bethany Scutt (https://orcid.org/0000-0003-0456-2556) (Research Worker, Old Age Psychiatry), Victoria Ordonez Montano (https://orcid.org/0000-0002-3500-1922) (Research Practitioner, Mental Health Research), Samantha Nunn (https://orcid.org/0000-0002-2853-9652) (Research Practitioner, Mental Health Research) and Grace Lavelle (https://orcid.org/0000-0003-3768-1797) (Research Worker, Old Age Psychiatry) recruited and retained participants.

Matthew Lariviere (https://orcid.org/0000-0001-6901-3115) (Research Fellow, Anthropologist of Care, Ageing and Technology) designed and analysed the ethnographic study.

Shashivadan Hirani (https://orcid.org/0000-0002-1577-8806) (Senior Lecturer, Health Services Research) and Stefano Brini (https://orcid.org/0000-0002-1909-1796) (Research Fellow, Health Services Research and Management) analysed the caregiver outcome data and contributed to the paper.

Andrew Bateman (https://orcid.org/0000-0002-2547-5921) (Clinical Manager, Neuropsychological Rehabilitation), Peter Bentham (https://orcid.org/0000-0002-6443-3353) (Consultant, Psychiatry), Alistair Burns (https://orcid.org/0000-0002-9837-0645) (Professor, Old Age Psychiatry) and Barbara Dunk (https://orcid.org/0000-0002-6363-5009) (Senior Occupational Therapist, Older Adult Memory Services) were principal investigators.

Kirsty Forsyth (https://orcid.org/0000-0002-6732-1699) (Professor, Occupational Therapy) analysed the needs assessment data and contributed to the paper.

Chris Fox (https://orcid.org/0000-0001-9480-5704) (Clinical Professor, Psychiatry) designed and analysed the ethnographic study, contributed to the paper and was a principal investigator.

Catherine Henderson (https://orcid.org/0000-0003-4340-4702) (Assistant Professorial Research Fellow, Health Policy) analysed the health economic data, contributed to the paper, interpreted the data and wrote the initial draft paper.

Martin Knapp (https://orcid.org/0000-0003-1427-0215) (Professor of Social Policy, Health Policy) analysed the health economic data and contributed to the paper.

Iracema Leroi (https://orcid.org/0000-0003-1822-3643) (Professor, Psychiatry in Ageing and Dementia) contributed to the paper and was a principal investigator.

Stanton Newman (https://orcid.org/0000-0001-6712-6079) (Professor, Health Psychology) analysed the caregiver outcome data and contributed to the paper.

John O’Brien (https://orcid.org/0000-0002-0837-5080) (Professor, Old Age Psychiatry) was a principal investigator.

Fiona Poland (https://orcid.org/0000-0003-0003-6911) (Professor, Social Research Methodology) designed and analysed the ethnographic study and contributed to the paper.

John Woolham (https://orcid.org/0000-0003-3128-7756) (Senior Research Fellow, Social Care Research) designed the trial and contributed to the paper.

Richard Gray (https://orcid.org/0000-0003-4440-574X) (Professor, Medical Statistics) designed and ran the trial, analysed the survival data, contributed to the paper, interpreted the data and wrote the initial draft paper.

Robert Howard (https://orcid.org/0000-0002-3071-2338) (Professor, Old Age Psychiatry) designed and ran the trial, contributed to the paper, was a principal investigator, interpreted the data and wrote the initial draft paper.

All authors assume responsibility for the accuracy and completeness of the data and for the overall content and integrity of the paper.

Publications

Leroi I, Woolham J, Gathercole R, Howard R, Dunk B, Fox C, et al. Does telecare prolong community living in dementia? A study protocol for a pragmatic, randomised controlled trial. Trials 2013;14:349.

Forsyth K, Henderson C, Davis L, Roy AS, Dunk B, Curnow E, et al. Assessment of need and practice for assistive technology and telecare for people with dementia – The ATTILA (Assistive Technology and Telecare to maintain Independent Living At home for people with dementia) trial. Alzheimers Dement 2019;5:420–30.

Howard R, Gathercole R, Bradley R, Harper E, Davis L, Pank L, et al. The effectiveness and cost-effectiveness of assistive technology and telecare for independent living in dementia: a randomised controlled trial. Age Ageing 2021:afaa284.

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 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, NETSCC, the HTA 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, NETSCC, the HTA programme or the Department of Health and Social Care.

The ATTILA trial: proposal for costing the assistive technology and telecare intervention

In this appendix, we set out the proposed plan of methods to calculate the costs of providing the ATT intervention, written while the trial was still in progress. The proposal begins with an overview of the trial as a whole and of the economic evaluation in particular, and lists the data collections that supported the economic evaluation. The costs that we proposed to collect are then summarised, and the methods for these data collections are outlined.

Trial overview

The ATTILA trial is evaluating the impacts of ATT on people with dementia and their carers. ATT is defined for the purposes of the trial as ‘simple, battery-operated, standalone technologies and/or telecare (a range of devices and sensors that communicate and relay messages to an external call centre where an appropriate response is arranged)’ (see Additional interviews and correspondence to support the economic evaluation: sample selection and recruitment of key informants). The trial is designed as a pragmatic RCT. Participants randomised to the intervention will receive an ATT needs assessment, followed by the installation of ATT devices and response services deployed by the host local authority, in addition to usual health and social care services. Control group participants will receive an ATT needs assessment followed by a package of ATT that is limited to smoke/carbon monoxide detectors and pendant alarms, in addition to usual health and social care services. Thirteen sites in England (Lambeth, Southwark, Croydon, Lancashire, Oxford, Suffolk, Norfolk, Cambridgeshire, Nottingham, Blackpool, West Sussex, Barnsley and Blackburn) are currently involved in the trial. As the trial design is pragmatic, all aspects of the intervention (ATT assessment, choice of devices, or ordering and installation of devices) are determined by staff from the participating local authorities or telecare providers.

Trial data collections

Current data collections include:

• Participants and their identified carers will be visited by study research assistants to complete a battery of questionnaires covering self-reported quality of life, acceptability of the technology, ATT in situ (using a technology checklist) and service use and costs (using the CSRI44). These visits will take place at five time points: baseline and 12, 24, 52 and 104 weeks thereafter. Questionnaire packs will be sent to the Oxford Clinical Trials Service Unit for data entry or data will be entered directly into an electronic database.

• Data on ATT assessments are gathered by Kirsty Forsyth. Forms completed by assessors in order to recommend a package of ATT are rated against the MOHOST.

Proposed activities

• Seek NHS Research Ethics Committee permission for interviews with key informants.

• Local researchers to provide the economic evaluator with a set of current contact details for potential key informants, with their permission:

  • local authority operational/middle managers in adult services

  • local authority commissioners of telecare

  • telecare providers managers.

• Researchers from the project team to set up and carry out interviews with key informants in each site (after Research Ethics Committee approval). Contact to be made by e-mail or telephone and a list of the relevant questions sent to the potential participants, who can then decide whether or not to participate. Interviews to be carried out via site visits or over the telephone. Interviews will not be recorded but handwritten notes will be taken.

• Project team to request relevant budget/expenditure data from key informants (local authority and telecare providers’ managers).

• Project team to update data-sharing agreements as required prior to requesting participants’ equipment data from relevant local authority and telecare providers; project team to establish procedures for transferring these data to the project team.

Additional interviews and correspondence to support the economic evaluation: sample selection and recruitment of key informants

Motivation

Routine

Communication

Cognitive skills

Physical skills

Physical environment

Social environment