Overview
This chapter draws on a manuscript by Crellin et al.190 uploaded as a preprint to medRxiv. This is an open access article under the term of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0).
What was already known
- Evidence shows that COVID-19 has a disproportionate impact on certain population groups, such as ethnic minority groups, older adults and those with comorbidities.
- Literature exploring the effectiveness of strategies to reduce health disparities for remote monitoring services and the impact of remote monitoring services on health disparities is scarce.
- The rapid adoption and spread of remote home monitoring services in England must be accompanied by evaluations at a local level to monitor the impact on health disparities.
What this chapter adds
- Increased understanding of how remote home monitoring services can be designed and delivered to address local population needs to increase accessibility and facilitate engagement with the service.
- We identified a number of local service adaptations, including prioritising vulnerable groups, creating referral pathways, offering translation services, offering non-digital options, and providing face-to-face assessments, and we discuss their potential impact on health disparities.
- We found that despite efforts to adapt services, disparities across patient groups in their experience of, and engagement with, the service were reported (related to age, health status, ethnicity, level of education, employment status and gender).
Introduction
The COVID-19 pandemic has shone a spotlight on existing health disparities in the UK. The pandemic has disproportionately affected the most marginalised communities, particularly people living in more socioeconomically deprived areas, those from minority ethnic communities, older adults, people with a learning disability and those residing in care homes, and people in key worker occupations.12,130,191
The pandemic has also led to changes to the way that health services are delivered at an unprecedented pace,192,193 with an acceleration of the move towards remote monitoring models of care. Yet, little is known as to whether and how this shift might affect existing health disparities.194–196
The benefits brought about by remote methods of delivering health care are not experienced by all: for tech-enabled remote methods, digital exclusion may negatively impact access to or quality of care.197–199 Other factors may exacerbate disparities in access, engagement or quality of care. Some patient groups are more likely to experience inequities (e.g. older adults, people with coexisting conditions, those with difficulties communicating, and those isolated or experiencing language barriers, those with lower health literacy or understanding of how health systems work) and may need additional support to engage with services.200
Inequities can be introduced at all stages of the planning and delivery of health interventions, including access, diagnostic accuracy, patient uptake and engagement.201 Given the recent rapid shift towards remote models of care, it is important to explore whether disparities exist in how patients access, engage with and experience services, and to understand whether and how remote services can be designed and delivered to promote inclusion. This study addressed the following questions:
- Were COVID-19 remote home monitoring services adapted at a local level to increase service reach and patient engagement for different groups?
- Were there disparities in patients’ reports of their ability to engage with the service, and were these moderated by the modality of the service? What were the potential impacts of service adaptations on health disparities according to patients?
- Were there disparities in patients’ reports of their experience of the service?
Methods
A summary of the methods is provided below; further details can be found in Chapter 2 or Report Supplementary Material 1.
Design
This chapter draws on staff and patient/carer surveys and staff and patient/carer interviews. Data collection took place between February and June 2021 (patient surveys retrospectively included January).
Sample
Patients/carers who received these services and staff leading and delivering services (see Chapter 2 and Chapter 4 for sample details).
Measures
We developed the survey and semistructured topic guides specifically for this study (see Chapter 2 and Project website documentation – Study instruments).
Data collection
We conducted a survey with patients/carers from 25 sites and with staff from 28 sites. We conducted interviews with staff, patients and carers from 17 sites (see Chapter 2 for details).
Analysis
Descriptive statistics were calculated to explore staff survey responses relating to perceived patient groups facing barriers. Open text responses relating to local service adaptations and patient groups facing barriers were analysed; responses were triangulated with service lead interview data, coded and grouped into themes.
For patient survey data, non-parametric univariate analyses (i.e. Mann–Whitney U-tests and Kruskal–Wallis tests) were conducted to explore patient engagement with services and experience of services across patient characteristics. Where statistically significance differences were detected (at the p < 0.01 level) across more than two groups, post hoc analyses were conducted. For both patient and staff survey responses, where data were missing for specific questions, available case analysis was performed and the denominator reported.
We conducted logistic regression modelling to examine whether modality of the service, age, education, health status and ethnicity were associated with likelihood of patients reporting at least one problem with the service. Patient open text survey responses providing feedback about services and any suggested changes or improvements were triangulated with patient interview data, coded and grouped into themes.
Service lead and patient interviews
To understand whether and how local services were adapted to address health disparities and their potential impacts (RQ1), a thematic analysis was carried out on primary data from service lead interviews.202,203 Interview transcripts were coded inductively and organised into higher-order themes.
To examine the potential impacts of adaptations from patients’ perspectives, RAP sheets70 were analysed. Data was deductively coded using the themes from the service lead interviews.
Results
Participant characteristics
Surveys were received from 292 NHS staff and 1069 surveys were received from patients and carers. We conducted 23 interviews with service delivery leads and 62 interviews with patients/carers (see Chapter 4 and Report Supplementary Material 2 for demographic characteristics).
Comparison of the characteristics of patient survey respondents with those of more than 26,000 patients onboarded to CO@h services114 indicate the survey sample to be relatively representative of patients engaging with the service. However, several differences should be noted. The survey sample is under-representative of patients over 80 years (5% vs. 10%) and under 50 years of age (21% vs. 33%), and over-representative of patients aged 50–80 years (74% vs. 57%). The survey sample comprises a higher proportion of patients from white ethnic groups (91%) compared to those onboarded to services (76%). Finally, our sample is over-representative of patients in the least deprived deciles of the IMD (18% vs. 13%) and under-representative of patients in the most deprived deciles (24% vs. 28%).
Service adaptations to increase service reach and patient engagement
Staff reported their views of patient groups facing barriers to engaging with the service. Some 39% (n = 113/292) reported patients for whom English was not their first language, 33% (n = 97/292) patients with a visual or hearing impairment, 31% (n = 91/292) cognitive impairment, 22% (n = 63/292) patients with a learning disability, 21% (n = 61/292) those who are digitally excluded, 13% (n = 37/292) older adults, 7% (n = 20/292) ethnic minority groups, and 5% (n = 16/292) unpaid carers. Other patient groups that staff reported in surveys or interviews included those living alone without support, patients unable to collect the oximeter, people with mental health problems, asthmatics, pregnant patients, patients whose GP had not engaged with the service, those with difficulty communicating via phone, and patients without a fixed abode.
Two-thirds of service leads (16/24 sites) reported that their service had been adapted at a local level to accommodate specific service user needs or requirements. Examples included providing information in different languages, offering translation services, offering non-digital monitoring options, face-to-face assessments, and flexibility of monitoring. Service-lead interviews and survey data indicated there was considerable local variation in the number of strategies adopted by sites. For example, several services made considerable efforts to standardise the coverage of the service, invest resources into setting up special pathways targeting hard-to-reach, vulnerable or at-risk groups, and monitor uptake. National strategies and local service adaptations reported by staff are set out in Appendix 4, Table 27.
Disparities in patients’ reports of their ability to engage with the service
In this section, we present findings from patient-reported engagement with these services in terms of the accessibility of information, the achievability of tasks, and problems reported via the survey (drawn from the survey analysis). See Report Supplementary Material 8 for further details. We also use findings from patient interviews and open-text survey responses (referred to as ‘qualitative data’) relating to patient experiences of the impact of service adaptations aimed at addressing inequalities.
Accessing the service
In interviews, patients reported being referred to services by many different organisations and pathways. However, sites were not always consistent in the admission criteria applied to access the service and there was sometimes a lack of clarity in how patients were referred to the service; some patients reported inconsistencies across local areas and several patients from sites that used active case finding reported they did not know how they were referred. Collaboration between services was perceived to have facilitated more rapid access to care; several patients highlighted that they were supported quickly because of close liaison between services and their GP practice, hospital or ambulance services. Once enrolled into the service, there were challenges relating to obtaining the appropriate equipment; some patients reported it was difficult to collect the oximeter when they were unwell and relied on family members or friends.
Accessibility of the information about the service
Interviews and survey responses indicated variation across and between sites in the information available to patients and how it was provided. We found evidence of a difference in understanding information provided by services with patient age (p = 0.005) and ethnicity (p = 0.001). There were also some marginally non-significant differences with health status (p = 0.014).
Patients aged 80 years and over reported less ease in understanding information; a lower proportion of patients reported understanding the information to be ‘very easy’ (40%) compared with all other age categories (60–67%). However, the number of patients aged 80 years and over was small and when ‘very easy’ and ‘easy’ responses were combined there was little difference across age categories. Patient ratings of helpfulness of the information was also statistically significantly lower for older patients (p = 0.005), with similar patterns of responses to understanding the information. Patients from minority ethnic groups reported less ease in understanding information provided; 81% of patients from minority ethnic groups reported understanding the information to be ‘easy’ or ‘very easy’ compared to 96% of patients from white ethnic groups.
Interviews identified several other factors affecting the accessibility of patient information. Several patients noted it was particularly difficult to take on the information when they were unwell, while those with visual impairments would have welcomed written information in a larger font, and several patients would have liked additional information about the service, their condition and recovery (e.g. how to interpret oxygen readings, how to manage their recovery, and where to seek help).
Patients’ ability to complete tasks as part of the service
We found evidence of differences in patient-reported ease of completing monitoring activities (i.e. using an oximeter, recording and providing readings) and escalation activities (i.e. seeking further help) with age, gender, health status, ethnicity, and level of education (see Report Supplementary Material 8 for further details).
Monitoring activities
Overall, our analysis found that most patients reported monitoring activities easy to engage with. The proportion of patients reporting difficulties was low; however, some patient groups might require additional support to engage with the service. This is consistent with qualitative data; some patients reported receiving additional support beyond the monitoring protocol or national guidance, such as receiving face-to-face visits.
Survey analysis found evidence of a difference in patient-reported ease of recording readings with health status (p = 0.004). Fewer patients with a health problem or disability rated recording readings as ‘very easy’ (70%) compared with those without a health problem (79%), however when this response option was combined with ‘easy’ there was little difference between groups. A similar trend was found for health status and providing readings to services, however, was marginally non-significant (p = 0.017).
There was evidence of differences between ethnic groups for recording readings (p < 0.001). Fewer patients from minority ethnic groups reported recording readings (91%) to be ‘easy’ or ‘very easy’ compared with white groups (98%). A similar pattern was found for providing readings to services however was marginally non-significant (p = 0.013). While translated patient materials were available to services, the extent that these were used by services is unclear.
Staff interviews indicated translation services were often needed and not always available for patients for whom English was not their first language. Services often relied on liaising with friends or family to support communication (when translation services were not available) and this was not always possible.
Although not statistically significant, there was some evidence of differences in ease of recording (p = 0.022) and providing readings (p = 0.016) with age. There was little difference in the proportion of patients reporting tasks as ‘easy’ or ‘very easy’ across age categories, but fewer patients aged over 80 years reported recording readings (55%) and providing readings (57%) to be ‘very easy’ compared with those under 50 years (80%, 79%) and in the 50–64 years age category (77%, 78%). Older patients reported support from friends, family and health-care professionals was more important for their engagement with services: 26% (n = 11/43) of patients over 80 years reported they had support from family and friends to use equipment and 49% (n = 21/43) reported support from health-care professionals helped them to engage with the service. This compares with 21% (n = 40/195) of patients under 50 years of age reporting support from family and friends, and 43% (n = 83/195) reporting support from health-care professionals.
A statistically significant difference in providing readings to services (p = 0.001) was also identified for gender: fewer male patients reported providing readings (70%) to be ‘very easy’ compared with female patients (80%). A higher proportion of males (25%, n = 97/385) also reported the importance of friends and family in supporting their engagement with services compared with females (19%, n = 102/531).
There were no statistically significant differences (at the p < 0.01 level) in engagement with monitoring activities with employment or living situation. In interviews several patients in paid employment reported that they would prefer to reduce the number of daily readings or that they preferred to receive calls at a specific time each day. Patients living with others reported greater reliance on support networks, friends and family; 10% (n = 14/137) of patients living alone reported support from family or friends helped them to engage with the service compared with 23% (n = 175/754) of those living with others. Many patients reported the support from family members, friends or carers to take and provide readings, and liaise with the remote monitoring team was crucial. This was particularly important for patients who were tired or more acutely unwell.
Escalation processes
As for escalation processes (i.e. seeking further help when needed), we found evidence of a difference with employment status (p = 0.007). Fewer patients who were not in employment reported seeking further help to be ‘easy’ or ‘very easy’ (84%) compared with those who were employed (89%). There was also some evidence of differences in ease of engagement with escalation processes across ethnic groups (p = 0.015) and with level of education (p = 0.015), however these were marginally non-significant (at the p < 0.01 level). Patients with higher education attainment and those from minority ethnic groups were less likely to report seeking further help to be ‘easy’ or ‘very easy’.
Disparities in patient engagement by the mode of services
Overall, 25% (n = 228/936) of patients reported at least one problem completing tasks that were part of services (e.g. using the oximeter, providing or recording readings, or seeking further help). Logistic regression analysis found mode of monitoring was not related to whether patients reported a problem, nor were health status or ethnicity (see Report Supplementary Material 8). Increasing age and a higher level of educational attainment were associated with an increased likelihood of reporting a problem. The odds of reporting at least one problem with services were approximately 7.6 times higher for older adults (80 years and over) than younger adults (under 50 years of age), for patients aged 65–79 years the odds were 2.9 times higher and for patients aged 50–64 years the odds were 2.3 times higher compared with patients under 50 years. The odds of patients who were educated to A level or equivalent reporting a problem were three times higher than patients who had received no formal qualification.
Disparities in patients’ reports of their experience of the service
We found evidence of differences in patient reported experience and helpfulness of services with age and employment status, however reassurance provided did not differ markedly across patient groups.
Older patients reported services to be less helpful (p = 0.004); fewer patients aged 80 years and over reported services to be ‘helpful’ or ‘very helpful’ (88%) compared with those in the under 50 years (92%) and 50–64 years age categories (92%). Older patients also reported a less positive experience of services (p = 0.034), although this was not statistically significant (at the p < 0.01 level). The less positive experiences of patients aged 80 years and over might be explained by differences in understanding information about services and ease of completing monitoring tasks; however, they might also be due to the smaller sample size of the 80 years and over age category.
Patients in employment rated services more positively (p = 0.001) and found services more helpful (p = 0.003); 96% rated it ‘good’ or ‘excellent’ and 93% ‘helpful’ or ‘very helpful’ compared with 92% and 87%, respectively, among patients not in employment. The differences in patient experiences based on employment status might be related to other factors, such as health status, level of education or age, with older patients more likely to be retired.
Living situation was not identified in the survey data to be a significant factor in patient reported experience, or reassurance provided. However patient interview data indicated living situation and strength of social network as important factors; provided by services was reported to be particularly important for those socially isolated or living alone.
Discussion
Key findings
At a local level many sites designed and adapted their service to be more inclusive to the needs of their local population and to expand their reach. Strategies included: providing information in different languages or formats, offering translation services, offering analogue and tech-enabled options for relaying readings, face-to-face assessments, delivering oximeters, providing additional training or support, flexible monitoring processes, and liaising with family or friends. There was considerable local variation in the adaptations employed by services. The adoption of such strategies was often dependent on local leaders as to how patient groups were approached and supported,90 and on the resources available to deliver the service.
Despite efforts to adapt services to meet local needs, there were disparities across patient groups in their experience of, and engagement with, services including age, employment status, ethnicity, level of education, health status and gender.
How findings relate to previous research
Despite previous evaluations investigating models of remote home monitoring and implementation of such services for COVID-19 patients,36,109 to our knowledge, this is the first research that focused on strategies adopted by services to address disparities. Our findings are consistent with previous research of strategies that can be used to reduce disparities in access to and use of health services more broadly,204–207 for example, expanding admission criteria and referral routes, and providing information in a range of formats (although support was provided at a national level, not all services provided information in formats accessible to all patients). Our findings extend research by outlining some strategies not previously identified; for example, staff providing additional support beyond monitoring (linking with social support, or more contact time), amending protocols to patient need and liaising with family/carers. Some possible areas for improvement included the need to routinely collect and analyse data on relevant populations, co-developing platforms with patients and staff (there was limited opportunity due to the rapid implementation) and working with communities to promote the service.
Critiques of remote models of care often focus on possible harms arising from digital exclusion. Having analogue options available as part of COVID-19 remote home monitoring services is, therefore, crucial to prevent increasing health disparities.208 This is particularly important given the overlap between patient groups at greater risk of severe illness from COVID-19 (e.g. older adults) and those who are at greater risk of digital exclusion. Our finding that mode of service did not impact on patient engagement or experiences might be explained by patient satisfaction or the fact that all sites offered analogue modes (see Chapter 11).
Implications
Our evaluation generates a range of recommendations for the design and delivery of remote monitoring services for patients with COVID-19, and other health conditions. Services must consider the needs of their local population and be adapted accordingly, monitor the impact of services on population groups at risk of health disparities and work with existing local systems (e.g. community groups) to engage harder to reach groups.
Given that many of the groups of patients (e.g. clinically vulnerable, older adults, those without support) that services aimed to reach found it more difficult to engage with the service, patients’ needs and circumstances should be assessed upon referral in a standardised way and services must be tailored to provide appropriate levels of support for patients and carers (see Chapters 9 and 10 results and Miles and Huberman,203 Cooper et al.209 and Centre for Ageing Better210). Examples of appropriate tailoring could include patient information in a range of formats,211 additional information, education or support in navigating services,212 and flexible monitoring processes.).
When implementing remote models of care, adequate resources and infrastructure should be provided to allow patients to be involved in the design and implementation. Co-design can help to ensure that services are better suited to patient needs and can facilitate adherence and engagement.213 Holistic health care and collaboration or linking between services have previously shown to be important for reducing inequalities, particularly for those who are unwell or have comorbidities.209,213
Strengths and limitations
The analysis of experience and engagement across patient groups could have been subject to false positives due to the number of comparisons made, although a more stringent p-value was used in reporting significant results (p < 0.01). When making comparisons between patient groups, the sample size of several groups was relatively small (e.g. patients aged over 80 years), limiting generalisability. The planned multivariate regressions investigating whether mode of monitoring and patient characteristics were associated with patient experience of the service were not possible due to the skew within the dependent variable (and therefore insufficient power to detect any differences). The interviews with patients focused on their experiences of services more broadly and not necessarily the identification of inequalities in access or quality of care, so much of our analysis of the latter draws on staff perspectives or patient survey data.
Future research
There is little published literature on the implementation of remote monitoring and health disparities. There are other populations at risk of health disparities that were beyond the scope of this evaluation and should be considered in any future evaluations (e.g. the homeless community, people with disabilities and unpaid carers).
The scope of our evaluation also did not allow us to examine the impact of specific strategies used to increase accessibility or patient engagement, or the impact of specific adaptations made by local services. We are therefore not able to make any conclusions regarding the effectiveness of strategies in relation to health outcomes for particular patient groups. This is an area for future research.
Conclusions
When health services undergo such a rapid transformation, as we have seen with the shift towards remote models of care, evaluations of the effectiveness of such services must include an impact assessment to ensure services are accessible and available to all patients, and to monitor the impact on health disparities. Addressing health disparities must therefore be a key focus in the planning, design and delivery of remote monitoring models for COVID-19 and other conditions, both nationally and locally, so that remote models of care can be of value to all population groups. Staff and patients from groups typically experiencing disparities in their health care must play an active role in service design to ensure their needs, experiences and expectations are accounted for.