The development and implementation of COVID-19 remote home monitoring services (phase 2)

Overview

What was already known

• COVID-19 remote home monitoring services were implemented internationally during wave 1 of the pandemic.
• During our phase 1 evaluation (wave 1 of the pandemic), some facilitators and barriers to COVID-19 remote home monitoring services were identified.
• It is important to explore local variation to fully understand implementation and to identify factors that might explain variation in effectiveness and cost outcomes across service models.

What this chapter adds

• Rapid development of national remote home monitoring services took place in three phases: (1) local development, (2) national development and roll-out, and (3) local implementation.
• Patient enrolment to COVID-19 remote home monitoring services across England was lower than expected.
• There was large variability in the models implemented for COVID-19 remote home monitoring services.
• Variation was influenced by patient, workforce, organisational and resource factors, and the rapid and evolving COVID-19 situation.
• These factors might help to explain variation in outcomes/effectiveness of different service models.

Introduction

It is important to understand the impact of local factors on adoption when implementing national policy initiatives81–83 as evidence suggests local factors affect service design,84–86 ability to meet service aims,87 sustainability of the service and can also hinder the accurate assessment of efficacy and cost-effectiveness.57,81,83,84,88

To increase our understanding of how local conditions affect rapid implementation of national policy during a public health crisis89–92 and support future development and implementation of services,93–96 it is necessary to explore the development, enrolment rates and implementation of COVID-19 remote home monitoring services (see Research question 6, Chapter 1).

This chapter provides:

1. an overview of the development of COVID-19 remote home monitoring services including the national roll-out and the enrolment rates of the services
2. a description of how services have been implemented and used locally
3. an analysis of the local factors that help explain why the services varied across sites
4. a backdrop for findings presented in Chapters 5–11.

Methods

A summary of the methods is provided below; further details can be found in Chapter 2, Report Supplementary Material 1.

Design

This study draws upon findings from all four workstreams (see Figure 3).

FIGURE 3 A summary of how data from each of the four workstreams contributed to Chapter 4. (Figure not reproduced here; see the original article.)

Sample and recruitment

Twenty-eight sites were recruited to our study. Sites were included from all regions across England, represented CCGs and trusts, varied in urban/rural, deprivation scores, ethnicity, and size (see Chapter 2, Report Supplementary Material 2).

Data collection

To identify when new services started and to estimate the proportion of new diagnosed cases who were onboarded (enrolment rates), we drew on a selection of data sources (see Report Supplementary Material 1), including Kent, Surrey and Sussex AHSN (KSS AHSN) data (see Chapter 6), bespoke onboarding data and data on diagnoses of COVID-19 provided by PHE (now the UK Health Security Agency).

To describe the development and roll-out of services we conducted interviews with national leads (see Chapter 2). To explore implementation, we conducted a documentary analysis of service documentation, surveys with patients and staff and interviews with patients, carers and staff, and a cost survey (see Chapter 2).

Data analysis

Survey and interview data

Inductive thematic analysis was used to analyse interview data inputted into RAP sheets.70 Survey data were analysed using descriptive statistics (using SPSS, version 25). To allow for comparison, quantitative and qualitative data sets were mapped against the SOP components (referral, triage, onboarding, monitoring, discharge).

National data on enrolment

For the CO@h programme, we defined rates of enrolment as the proportion of new cases diagnosed who were onboarded. For each CCG, we estimated these rates by dividing the numbers onboarded each fortnight by the number of new cases of COVID-19 diagnosed in the same fortnight. We could only do this for CCGs where the onboarding data were judged to be complete, and to make this judgement we used a combination of management information provided by the programme and data collected by our own costing survey (see Report Supplementary Material 1).

For the CVW programme, data quality issues left us unable to use NHS situation reports (SITREPs) data on the number of patients with COVID-19 discharged from hospital to virtual wards. We were able, however, to carry out a limited analysis of enrolment rates for seven hospital trust-based sites by comparing the number of patients monitored on a post-discharge pathway (as reported in our costs survey, see Chapter 7) to the number of patients with COVID-19 discharged alive, according to national HES Admitted Patient Care (APC) data.

Findings

Within this section, the following findings are presented: (1) a summary of interview and survey data (n = 28 sites), (2) an overview of the development of COVID-19 remote home monitoring services (including (a) development of these services, and (b) coverage of these services nationally and rates of enrolment), (3) a description of the different typologies and implementation of services between sites, and (4) factors influencing the implementation of these services.

Participant characteristics

We received 292 staff surveys (39% response rate) across 28 sites and 1069 patient and carer surveys (18% response rate) across 25 sites (see Report Supplementary Material 2 for demographic characteristics). Interviews were conducted with national leads (n = 5), staff (n = 58 across 17 sites), and patients or carers (n = 62 across 17 sites). See Report Supplementary Material 2 for demographic characteristics.

An overview of the development of COVID-19 remote home monitoring services including the national roll-out and the uptake/coverage of the resultant service

Within this section, we first present an overview of the three phases of development and implementation of these services. Secondly, we present the uptake/coverage of the final service.

Development of COVID-19 remote home monitoring services

The development of COVID-19 remote home monitoring services occurred rapidly, in three stages: (1) local development and implementation, (2) national development and roll-out, and (3) local implementation (see Figure 4 for timeline and key groups involved in each phase).

FIGURE 4 Development of COVID-19 remote home monitoring services. (Figure not reproduced here; see the original article.)

Local development and implementation – wave 1 and the first use of pulse oximetry in patients’ homes

Local development and implementation

The use of oximetry to monitor patients with COVID-19 in their home began in wave 1 as several, initially disconnected, local services were established, motivated by attempts to mitigate silent hypoxia (very low oxygen saturations, often without breathlessness). Pulse oximetry was initially used to monitor patients in the community.36 During phase 1, local clinical leaders within CCGs and secondary care trusts had identified a need for this service and were integral in facilitating the initial development, set-up and implementation of services, and instigated the development of an informal community of practice. NHSX supported a few of the pilot sites to use tech-enabled models by putting them in touch with tech companies. The local development and implementation of services gained interest from national stakeholders.

An evaluation of phase 1 was commissioned and conducted by researchers from NIHR RSET and NIHR BRACE (see Chapter 3). The purpose of this evaluation was to provide lessons for the ongoing development/roll-out of the services.

Supporting local implementation

Two learning communities were set up prior to the national roll-out, to support local implementation: a community of practice and a national learning network.

The community of practice was an informal group, led by key national clinical and policy leads and attended by national and local clinical leaders for local services adopting these pulse oximetry services. The community of practice held regular online meetings (every couple of weeks) to share best practice quickly from national leads to local teams, and between local teams, offering mutual support and leadership. This community of practice grew rapidly.

A national learning network (the National Deterioration and COVID-19 Forum) was set up by the AHSN patient safety collaboratives to support clinical leads and provide resources to facilitate implementation of these services. The set-up of the national learning network was prompted through NHS @home17 (via the national team and regional medical directors). Additionally, regular webinars were held to share learning.97 The learning network was hosted within the Future NHS collaboration platform (an online platform designed to help services to co-develop services, share materials and work together).98

The conversations and shared learning from both these networks played a key role in shaping the basic pathway underpinning these services. These communities were set up to support local clinical leads and share experiences.

National development and roll-out

National development

Between wave 1 and 2, there was a period of gathering evidence (research and clinical consensus), discussions around how services should be developed and development of safety netting guidance. This evidence included our evaluation findings from phase 1 (see Chapter 3).

During the summer and autumn of 2020, NHSEI began to develop SOPs for CO@h, with leadership and governance from the NHS @home team. The National Incident Response Board approved operationalisation and national roll-out of the CO@h service as the appropriate next step.

The SOP for CO@h38 (see Appendix 1, Table 24) that emerged was informed by the learning gained from previous services which provided care at home, experience gained from managing wave 1 in the community including the various pilots and case studies. It was shared through the NHS Futures website98 and the pre-existing evidence base.32,36 It was intended that services would be inclusive by design and enabled by, but not dependent on, technology (to remain accessible to the most vulnerable patient groups).

In January 2021, a SOP for CVW (early discharge from hospital) was published and these services were rolled out.39 Virtual ward models were designed to be run by acute trusts, with responsibility for services being with secondary care services39 (see Appendix 1, Table 24).

National roll-out

Following service development, the AHSNs were commissioned through the AHSN’s patient safety collaboratives to facilitate regional roll-out.97 A programme board with all key stakeholders was established and met weekly to discuss a number of workstreams including data management, workforce/training, patient safety, operations, pathways, evaluation and oximeter distribution. The programme was monitored via comprehensive data collection for CO@h which aimed to track how the programme was developing/performing with a focus on using the data as shared learning. As part of this process any safety incidents were recorded, using patient safety groups where appropriate.

National roll-out of CO@h and virtual ward services was supported by 20 highly attended regional launch events, which included question and answer sessions. To support national roll-out, national clinical leads and seven regional clinical leads were funded to support local implementation. NHSEI agreed to purchase pulse oximeters. NHSEI purchased 706,000 oximeters across waves 1 and 2. These pulse oximeters were CE marked (as recommended by MHRA guidance)37 and were available to local sites within three days through an ordering system set out in the SOP. NHS Digital supported data management.99 NHSX were a partner in the roll-out and implementation of tech-enabled platforms within COVID-19 remote home monitoring services. NHSX committed to supporting (funding) up to 50 organisations to use a tech-enabled model for CO@h and/or virtual ward models during wave 2 (sites bid for this funding and then chose and procured the platform they wanted to use). CO@h and virtual ward services were rolled out between November 2020 and February 2021.

Local implementation

NHSEI were integral in supporting the local implementation of COVID-19 remote home monitoring services, including providing financial support for primary care to establish and implement these services,40 resourcing NHS Digital to supply data to service providers to identify people who may benefit from these services41 and the commissioning and funding of AHSN support.

National leads described the importance of local leads assuming responsibility and accountability for their services, guided by the SOPs for CO@h and CVW.

While there was some variation in responsibility and accountability, local CCGs were primarily responsible and accountable for delivery of CO@h and secondary care trusts were responsible and accountable for delivery of CVW. Each local service had its own local governance and management structure, which were expected to minimise national input and allow local services to be more responsive to and reflective of the needs of local populations. This local governance was overseen and supported by the AHSNs. NHSX managed data reporting on tech-enabled deployment.

With local services responsible for the day-to-day running of CO@h and CVW, services allocated resources, designed staffing models, and distributed equipment and educational materials. The management teams particularly of integrated services (i.e. where CO@h was combined with CVW) consisted of colleagues drawn from across multiple settings including primary and secondary care, but also the ambulance service, mental health trusts, care homes, and patient safety collaboratives. The precise mix depended upon the scope of services within the local environment. Regional ‘clinics’ were established by the local AHSNs to facilitate regular engagement and support the initial set-up. Two-way relationships were developed between policymakers and strategic management and front-line NHS workers as a way of sense-checking the work completed. Significant support and buy-in from senior management within acute trusts and across CCGs helped to set up COVID-19 remote home monitoring services. There was, however, a need to have supportive local authority and public health teams to integrate CO@h and CVW services with existing ways of data sharing.

Supporting local implementation

Throughout the national roll-out, local services were supported by the community of practice and the national learning network (as described earlier).

National coverage of COVID-19 remote home monitoring services and enrolment

COVID Oximetry @home

We judged that the national onboarding data for CO@h were complete for 37 of 135 CCGs (see Report Supplementary Material 1), so our analysis of coverage and enrolment was restricted to these areas.

The dates sites within each of the 37 CCGs started onboarding patients are plotted in Figure 5. Three of these CCGs were onboarding patients in October and all 37 were onboarding patients in the fortnight beginning 11 January 2021.

FIGURE 5 Number of CCGs with sites operating the CO@h programme by fortnight (data for the 37 CCGs where the onboarding data were assumed to be complete). (Figure not reproduced here; see the original article.)

Once services started operating and up until 4 April 2021, enrolment across the 37 CCGs among people aged 65 years or over is shown in Figure 6. The CCG with the highest enrolment rates had values that were more than twice the next highest; 22 CCGs (59%) enrolled fewer than 10%. The overall enrolment rate over the period for this age group across all 37 CCGs was 8.7%.

FIGURE 6 Enrolment of CO@h services among people aged 65 years or over by CCG after implementation started up until 4 April 2021 (enrolment is defined as the ratio of numbers onboarded to new cases over the period and calculated from 19 October 2020 for any services in operation before then). (Figure not reproduced here; see the original article.)

Virtual wards

Dates acute trusts opened their virtual wards are plotted in Figure 7. A majority of services (62%) launched in January and February 2021. For seven trust-based sites included in our costs survey (see Chapter 7), we were able to estimate an approximate enrolment proportion: that is, the proportion of patients with COVID-19 discharged who were monitored on a virtual ward. Two sites had fewer than 10% of patients monitored, one site had 20% and four had more than 50% (but below 65%) of patients monitored. Otherwise, with national SITREP data not useable because of data quality issues, we had no further information on virtual ward enrolment.

FIGURE 7 Start dates of CVW services in England’s non-specialist acute hospital trusts (n = 123 trusts); 14 trusts had no CVW by 22 February 2021. (Figure not reproduced here; see the original article.)

A description of the different typologies and implementation of services between sites

We created a typology (classification system) and categorised sites according to seven domains (see Table 1): (1) type of model (CO@h/CVW/Integrated); (2) sector leading services (primary care/secondary care/both); (3) type of monitoring (analogue-only (paper and telephone)/tech-enabled and analogue); (4) admission criteria (age and clinical vulnerability); (5) workforce (number and type of staff); (6) date service started; and (7) enrolment rates within the CCGs where the sites were located. As above, enrolment rates were defined as the ratio of patients onboarded to new cases among people aged 65 or over by CCG after implementation started up until 4 April 2021. It was calculated from 19 October 2020 for any services in operation before then.

Different models of COVID-19 remote home monitoring services were implemented across the 28 local sites. Below, we describe variability in relation to the stages of services outlined in the national SOPs (see Report Supplementary Material 4).38,39

Entry criteria

In terms of entry criteria, most services used age criteria of either 18 years or over, or 50 years or over, rather than the age criteria recommended within the SOP of 65 years or over. Many sites adapted and reduced their enrolment age throughout the period. Most sites used risk factors alongside age to enrol patients, but each site varied in the risk factors used.

Workforce

As shown in Table 1, there were large differences across sites with regard to the number of staff involved in setting up (n = 2–20+) and running (n = 2–70+) the service, and with regard to the type of staff involved in monitoring. Most sites used clinical staff only to support monitoring.

Triage

As recommended by national guidance, most sites had processes in place for patient triage (n = 24/25). However, interview findings indicate that triage processes varied across different models (e.g. in terms of who checks against admission criteria, and the processes used). See Chapter 11 for a description of models categorised by risk-stratification.

Onboarding

In line with national guidance, all of the sites provided patients with oximeters. However, their means of distribution varied, including delivery (n = 27 sites), given to patients at GP/hospital (n = 9 sites), patients/family members collecting oximeters (n = 7 sites). Some patients described difficulties in collecting oximeters due to their poor health and isolation.

In line with national guidance, most sites provided patients with information (n = 22/25). Most sites provided patients with written information or provided information over the telephone.

Monitoring

Sites differed in the type of options they offered for patients submitting data (e.g. tech-enabled and/or analogue) and the level of interaction with staff (see Appendix 2, Table 25). Most services were tech-enabled with telephone options offered when tech-enabled modes of submission were not possible. Members of staff were involved in taking and reporting readings at all sites; none of the services solely asked all their patients to self-monitor and self-escalate care (see Appendix 2, Table 25). However, interview findings indicated that two sites offered some of their patients (exact number not known) a self-monitoring option. Self-monitoring is defined as patients or carers (where appropriate) monitoring their readings and escalating their own care as necessary (see Report Supplementary Material 4).

Escalation

Escalation processes varied, with some tech-enabled solutions identifying patients for escalation (with the addition of phone or face-to-face assessment in some cases) and some escalation processes being manually initiated (with the addition of phone or face-to-face assessment in some cases) (see Report Supplementary Material 4).

Discharge

In line with national guidance, many but not all sites had explicit processes in place for patient discharge (n = 19/25). Our findings on the flexibility of discharge were in line with the SOP, in that most sites reported patients were discharged after 14 days, but that patients could be discharged earlier or kept on if needed. However, only 73.4% (n = 769) of patients were aware of discharge processes (range across sites: 33.3–90.9%). Oximeter return processes were also not apparent at every site with only 69% of patients being asked to return the oximeter (range across sites: 21.5–100%).

Factors influencing implementation of COVID-19 remote home monitoring services

We observed differences in the design and delivery of services between local sites and between the national guidance and local implementation. Implementation may have been influenced by a number of site-specific factors relating to patients, staff, the organisation delivering the programme, and the resources at their disposal (see Report Supplementary Material 4).

Patient-related factors

Patient factors relate to demographic characteristics, disease prevalence, digital access, and patient engagement. Sites were charged with responding to the needs of local populations and the degree to which they did so was shaped by the specific needs of the local population, influenced by factors such as socioeconomic status, ethnic and cultural background, prevalence of comorbidities, the ability of patients to access or utilise digital technologies and their level of engagement with the programme (see Box 1). However, sites varied in the time and resources dedicated to attempt to engage with the vulnerable and disadvantaged groups disproportionately affected by COVID-19.

Although all of our sites used telephones to collect readings, some sites would also use these calls to offer comfort and reassurance to isolating patients or otherwise signpost them to other services or community groups. Some sites reported that staff would visit patients in-person if particularly concerned about their wellbeing or if the patient was visually impaired. This additional support was at the discretion of sites and not prescribed by the national guidance.

In site O, they developed materials in a range of languages. They also used community-based partnerships and voluntary organisations to help onboard and monitor patients. Patients were drawn from a range of vulnerable populations leading them to successfully target minority ethnic groups, those who were morbidly obese and members from the local homeless population.

Staff factors

A number of staff-related factors influenced the way in which services varied between sites. These related to their level of training and experience, whether they worked remotely or in a shared workspace with more direct clinical oversight, and the overall workload they experienced (Box 2). For example, those that were redeployed to services tended to be nurses working in secondary care, whereas those asked to take on the responsibility of delivering these services alongside their current roles tended to be drawn from primary care.

The knowledge, skills and confidence of the staff charged with delivering services varied between sites, and these differences in confidence and knowledge were particularly evident when services relied on non-clinical staff. Across many sites, the routine availability of senior clinical oversight was deemed important.

In site M, where staff worked remotely from other staff, they described how their training was minimal and this gave rise to a number of concerns. These included their ability to use the software through which they managed their caseload, but also more fundamental issues around their lack of experience of teleconsultations, remote monitoring of respiratory conditions, and how this left them uncertain about whether they were offering the correct advice to patients or their non-clinical colleagues with patient-facing roles.

Organisational factors

A range of organisational factors impacted on the design of individual services including staffing models, the culture and work practices that supported the learning of individuals and/or the organisation,100 and the engagement of senior staff in changing practice and supporting innovative delivery.101 The timing of the introduction of remote home monitoring services in each local area may have influenced variation in services observed between sites. The SOP produced in November 2020 contained the caveat that those that had already established a service would not be expected to adhere to the national guidance.

A key factor in shaping services was the level of successful collaboration between the organisations involved at each site (Box 3). The importance of robust communication pathways between senior managers at local sites was recognised, as was the value of collaborating with other teams and specialities, particularly in refining and improving services and for supporting patients with additional needs (e.g. those with existing health conditions or requiring mental health support).

At site K, there was close collaboration between different sectors. Their service was run by primary care and secondary care organisations together with each organisation/sector operating within clearly defined roles (identification of patients, running hot hubs/assessments, monitoring). Staff within this service reported a history of integrated working relationships between secondary, primary care and voluntary/community sectors. These long-standing relationships were felt to better facilitate the implementation.

Resource factors

Resource factors related to the availability of reliable pulse oximeters and the availability and functionality of the software used to support the service. One major investment in these services was the provision of pulse oximeters by NHSEI. However, the perceived quality of these oximeters reportedly varied and not all sites had the links with the voluntary sector that many used to deliver the instruments (in workable order) to patients in the community.

The tech-enabled platforms selected by local systems varied across regions and local systems, with each possessing various levels of functionality and interoperability. The more readily service data could be integrated with existing databases the more effective the programme could operate. In its absence, valuable information on patients would be unavailable, for example, current medications, prior history of respiratory illness, or susceptibility to anxiety. Several sites reported that technology platforms were not always ready for use at the onset (due to the speed at which services were developed and implemented). This meant switching mid-pandemic without the necessary time for training or familiarisation (Box 4).

At site J, the site originally used (platform 1) but this did not meet their requirements to automatically send three reminders to patients per day. Therefore, mid-way through the second wave the service switched over to (platform 2). However, this presented its own challenges of moving from one system to another while the service was under enormous pressure and they had to temporarily return to using (platform 1) due to fears that some patients were being omitted.

Discussion

Key findings

The rapid development of COVID-19 remote home monitoring services in England took place in three phases: (1) local development and implementation (in wave 1 and 2), (2) national development and roll-out (in wave 2), and (3) local implementation. The proportion of new diagnosed cases who were onboarded on to these services was lower than expected. The implementation of services varied between sites and from national guidance. A range of service typologies emerged across our evaluation including noticeable variation in relation to enrolment rates, the nature of onboarding, referral criteria, monitoring processes, and discharge and pulse oximeter return procedures. Patient factors (e.g. needs of the local population), staff factors (e.g. workload and work environment), organisational factors (e.g. staffing models and collaboration) and resource factors (e.g. software) influenced implementation and enrolment.

How findings relate to previous research

The implementation of COVID-19 remote home monitoring services in England can be seen as an example of major system change or ‘large system transformations’, defined as ‘coordinated, systemwide change affecting multiple organisations and care providers, with the goal of significant improvements in the efficiency of health-care delivery, the quality of patient care, and population-level patient outcomes’ (Best et al.,102 p. 422). A conceptual framework proposes five main stages of major system change: (1) decision to change influenced by drivers to change, governance and leadership of decision-making, (2) decision on which model to implement, (3) the implementation approach, (4) implementation outcomes, and (5) intervention outcomes.103 There are different ways of implementing major system change (top-down and bottom-up approaches). Previous research suggests that a combination of both top-down and bottom-up approaches may be beneficial.104 Our findings indicate that rapid implementation processes may follow a bottom-up (local services make decision to change and which model to implement), top-down (national service makes decision to change and provides guidance on which model to implement), then bottom-up process (local services interpret guidance from national service and decide which model to implement) (as described in Chapter 4). Therefore, rapid implementation may vary from major system change in non-rapid contexts (e.g. Fulop et al.).103 However, our findings support previous research which indicates that this combination of approaches may impact on consistency of implementation approach and uptake.105

Prior to this evaluation, there was limited evidence regarding how these services have been implemented. Our findings indicated that most remote home monitoring services studied had low rates of enrolment and that there was considerable variation in which model was implemented and how, according to local context. These findings support previous research demonstrating likelihood of variation when implementing national services generally81–83,89–92 and in response to the COVID-19 pandemic.106–108 Previous research has highlighted differences in the models implemented and the ways in which services are implemented must be considered when interpreting effectiveness and cost-effectiveness of services.57,81,83,84,88

Strengths and limitations

This study used mixed-methods data from a wide range of sources, including national quantitative data, cost data, national surveys and case study interviews. Additionally, this evaluation is a large rapid scale evaluation, including 28 health-care services throughout England. Therefore, the findings provide a comprehensive picture of the development, coverage and implementation of COVID-19 remote home monitoring services through phase 2 of the pandemic.

Given the rapid implementation of services, we are unable to say whether the degree of variation we witnessed is a result of the rapid roll-out or whether it would have been observed if lengthier implementation processes had been followed.

Different data sources were triangulated to analyse the findings presented in this chapter. Sometimes, different data sources or respondents disagreed and consensus needed to be reached. For example, across several sites, more than one clinical lead/service manager completed the staff survey and there were several instances of responses being inconsistent (to overcome this problem, we sought further information from services as necessary and selected one clinical lead per site using relevance of job title).

Our assessment of service enrolment within CCGs relied on their onboarding data being complete. To judge completeness, we relied on assessments made by individual CCGs, although as they were not necessarily the service providers, and therefore removed from the data entry process, we were not sure about the accuracy of this information in all cases. We were able to cross-check against the data provided by the 28 study sites themselves and found a reasonable match for most but not all, and we were unable to verify the data from the CCGs that were not study sites in the same way.

Implications

Our findings have implications for interpretation of quantitative outcomes of the evaluation. Our findings on low rates of enrolment and variation in implementation can be used to help interpret findings relating to effectiveness and cost of the services (as we discuss in Chapters 5–7). However, we were not able to look at how specific variation linked with particular service outcomes or enrolment more generally.

Several valuable lessons for service development have emerged from our work. Findings indicate that when developing and rapidly implementing national roll-out of services, service planners must consider the context and priorities of local organisations (e.g. their patient population characteristics and resources). It may be helpful to consult with patient groups to ensure that strategies to ensure reach are appropriate and feasible, and that services meet the needs of the local population. Flexibility of workforce is needed for those delivering the service.

Future research

Future research should explore how similar nationally implemented programmes vary from their intended design when not implemented at such speed and the degree to which local variation acts favourably in sustaining the service. Further evidence is needed on self-monitoring models of remote home monitoring for COVID-19, including the groups of patients that this might be appropriate for and whether its use would enable services to onboard more patients. Further research is needed to explore why enrolment was lower than expected, and how to improve enrolment to these services in future.

Conclusions

Local and national leadership were integral in the development and implementation of COVID-19 remote home monitoring services. Enrolment to these services was lower than expected. There was large variability in the models implemented and that variation may have been influenced by patient, workforce, organisational and resource factors. Together, low enrolment and variation in implementation indicate that it is difficult to evaluate the true extent of effectiveness of COVID-19 remote home monitoring services.