Widening or narrowing inequalities? The equity implications of digital tools to support COVID‐19 contact tracing: A qualitative study

Abstract Background As digital tools are increasingly used to support COVID‐19 contact tracing, the equity implications must be considered. As part of a study to understand the public's views of digital contact tracing tools developed for the national ‘Test and Protect’ programme in Scotland, we aimed to explore the views of groups often excluded from such discussions. This paper reports on their views about the potential for contact tracing to exacerbate inequalities. Methods A qualitative study was carried out; interviews were conducted with key informants from organizations supporting people in marginalized situations, followed by interviews and focus groups with people recruited from these groups. Participants included, or represented, minority ethnic groups, asylum seekers and refugees and those experiencing multiple disadvantage including severe and enduring poverty. Results A total of 42 people participated: 13 key informants and 29 members of the public. While public participants were supportive of contact tracing, key informants raised concerns. Both sets of participants spoke about how contact tracing, and its associated digital tools, might increase inequalities. Barriers included finances (inability to afford smartphones or the data to ensure access to the internet); language (digital tools were available only in English and required a degree of literacy, even for English speakers); and trust (many marginalized groups distrusted statutory organizations and there were concerns that data may be passed to other organizations). One strength was that NHS Scotland, the data guardian, is seen as a generally trustworthy organization. Poverty was recognized as a barrier to people's ability to self‐isolate. Some participants were concerned about giving contact details of individuals who might struggle to self‐isolate for financial reasons. Conclusions The impact of contact tracing and associated digital tools on marginalized populations needs careful monitoring. This should include the contact tracing process and the ability of people to self‐isolate. Regular clear messaging from trusted groups and community members could help maintain trust and participation in the programme. Patient and Public Contribution Our patient and public involvement coapplicant, L. L., was involved in all aspects of the study including coauthorship. Interim results were presented to our local Public and Patient Involvement and Engagement Group, who commented on interpretation and made suggestions about further recruitment.


| BACKGROUND
The COVID-19 pandemic has deepened pre-existing health inequalities. [1][2][3] The risk of contracting COVID-19 and experiencing worse outcomes, such as hospitalization, intensive care unit admission and mortality, is greater for those living in poverty and for minority ethnic groups. [3][4][5] Essential workers, including lowerpaid occupations, and those living in overcrowded housing are at greater risk. 6,7 Mitigations, in particular, social distancing and working from home, and the ability to self-isolate are all more difficult for marginalized populations, with intersections between poverty, multiple disadvantage and ethnicity. 2,[8][9][10] A key strategy for population control is testing and contact tracing to break chains of transmission, but this also has the potential to exacerbate inequalities. In England, reporting of positive cases and contact tracing was lower in areas of deprivation. 11 Nevertheless, there has been little consideration of the inequities that might occur with contact tracing interventions. A rapid review of contact tracing interventions for a range of infectious diseases (including  found that no intervention considered the potential for increasing health inequities and there were no attempts to mitigate potential inequities in their design. 12 Others have also commented on the potential for digital technologies such as proximity apps to disadvantage certain groups. 13 This digital divide is well recognized, with certain groups including minority ethnic populations, older people and those on low incomes all likely to be disadvantaged by technological approaches to health and health care. 14 It is this potential for contact tracing and its associated digital interventions to exacerbate inequalities that is the focus of this paper. Contact tracing is a mainstay of public health surveillance. 15 Following the identification of a positive case, contact tracing seeks to rapidly identify all their contacts within a predefined period; those identified are tested and/or self-isolate. 16 Contact tracing has been used in the surveillance of many communicable diseases. 17,18 However, its history is problematic, with different responses to different population groups, breaches of trust in data use and confidentiality and lack of privacy for contacts. 15,18 The COVID-19 pandemic presents particular challenges for contact tracing. Cases may be asymptomatic, population spread can be rapid and the prevalence of infection high. Contact tracing has to occur at speed and scale to cut transmission chains. 16,19 Previously, contact tracing relied on manual approaches, with healthcare professionals or others collecting information directly from individuals. More recently, digital approaches have been developed and used in outbreaks of Ebola, tuberculosis and in sexual health services. 17,[20][21][22] COVID-19 has accelerated this shift towards digital tools to enable large-scale contact tracing. [23][24][25] Attention has focused on the development and use of COVID-19 contact tracing proximity apps. 26 Downloaded onto smartphones, these use Bluetooth to identify when another phone has the app active in close proximity. Close proximity is determined by distance (generally ≤2 m) and contact time (usually ≥15 min), to rule out passing contacts, for example in a shop or in the street. If a user then tests positive for COVID-19 and uploads that information onto the app, close contacts are automatically notified. There are, however, challenges to their use. First, a substantial proportion of the population has to download and use the app-in the United Kingdom, estimated to be about 80% of smartphone users (37 million people). 19 Second is storage of contact data. Decentralized approaches retain a record of device interactions on the smartphone itself; centralized systems hold data on cloud servers owned by commercial companies (commonly Google or Apple), raising concerns about data privacy and confidentiality. 19 Third, there are concerns about increased digital inequality in terms of access and use of apps. 13 Studies have explored public views of contact tracing apps.
While many recognized that, in a pandemic, such apps may be necessary to control infection, there were reservations about data storage, privacy and ethics. 27 Qualitative studies have explored people's concerns in more detail. Over 300 qualitative interviews conducted across nine European countries in April and May 2020 identified a spectrum of views regarding COVID-19 contact tracing apps, ranging from opposition, through scepticism about their feasibility, to support. 35,36 Similar views were reported in focus groups held in May 2020 with UK adults. Those concerned about privacy, uptake of the app and stigma were most likely to say that they would not download a contact tracing app. 37 While this study offers important insights, we believe that there are limitations especially in relation to the study population. Some studies were conducted before such apps were available, leaving the public to think about their use hypothetically. While study populations were broadly representative in terms of gender and age, they were less likely to consider ethnicity and, where information was given, participants tended to be well educated, indicating more socioeconomically affluent populations. These are all important considerations when trying to develop approaches that will not further exacerbate COVID-19 health inequalities. 38

| Digital contact tracing NHS Scotland
In the summer of 2020, NHS Scotland developed a proximity app for contact tracing. 39  inequalities. These are the findings that we present in this paper.

| METHODS
This was a qualitative study involving semi-structured interviews and two focus groups with members of the public, and semi-structured interviews with key informants representing community organizations. All participants were aged over 18 years. Data collection occurred in June and July 2020, while the proximity app and contact tracing tools were in development.

| Recruitment
We began with key informant interviews.

| Data collection
Interviews and focus groups were conducted via Zoom at a time suitable to participants; most participants were interviewed in their own home. Topic guides and information materials were developed in advance by the research team, including our patient and public involvement (PPI) coapplicant L. L., drawing on available literature at the time but also including questions and topics suggested by the developers of the digital tools, to assist them in developing the tools.
While the topic guide was not piloted before use, we discussed it throughout data collection and added prompts where necessary, for example, in relation to data privacy. During the interviews and focus groups, we shared screenshots of the digital tool with participants. For example, we shared and discussed privacy statements to determine their clarity and understandability and showed participants a set of screenshots with the information and layout that would be contained in the app. Interim results were shared with our Patient and Public Involvement and Engagement group for comment, who made suggestions on data interpretation and on future recruitment strategies.
Participant information sheets and consent forms were emailed to those wishing to participate before data collection. Completed consent forms were emailed back to the research team; however, verbal consent was also recorded at the beginning of each interview or focus group. Interviews and one focus group were conducted by S.

| Data analysis
Analysis utilized the framework approach to enable rapid feedback to the app developers and to facilitate comparison across participants. 42,43 A lack of resource meant that the interviews and focus groups were not transcribed; analysis instead used the sound recordings as the primary source of data. [44][45][46] Using the framework approach and regular coding discussions within the team ensured that rigour and transparency were maintained during the coding process. The recordings were also augmented by fieldnotes made at the time by those conducting data collection. 47 Analysis followed the five steps of the framework approach: First, we became familiar with the data by listening repeatedly to the recordings, both those who collected the data (authors C. O'D., S. B. and A. A.) and other members of the team. A thematic framework was developed, drawing on the areas covered by the topic guide but also issues that arose during the interviews (e.g., in relation to information sharing). These formed a set of eight charts in Excel. Each broad theme was broken down into subthemes. The thematic framework was then applied to each interview and focus group, with data extracted into the Excel sheets, along with researchers' interpretation of the data. Extracted data included a timestamp to allow researchers to identify the data in the sound recording, to facilitate data checking and extraction of quotes. These became our main data source for report and paper writing. As our particular interest was the potential for contact tracing to exacerbate inequalities, we used the PROGRESS framework as a lens to inform the interpretation of our findings. 48

| Views and knowledge of contact tracing
Public participants generally viewed contact tracing and proximity apps as a good idea, even if they were not clear about the process.
There was no one who said it was not a good idea. Some framed use as a responsibility to their family, friends and wider community, and others as a 'civic duty' (P2, older male).

| Language and digital access
Using these tools requires both language and digital access skills.
Missing one or both of these, it was suggested, would widen inequalities. For example, although information on the Test and Protect programme was available in several languages, the webbased interface for contact information was only in English, disadvantaging those whose first language was not English. There were also concerns that some of the information required a reasonable degree of literacy.
So language wise, there needs to be more than just English or its going to miss an entire swathe of the  Interlinked with the issues of digital access and affordability was trust in how data would be collected, stored and used and trust in the organizations involved. This is discussed next.

| Data governance and privacy
Public participants were, in general, trusting of the contact tracing process and the digital tools to support that process. Although contact tracers do not say who has passed on contact details, there was also a fear that this would happen informally within social networks or because the population size was small, for example, in rural areas. A few participants, especially amongst key informants, discussed the possible stigma of testing positive.
The age group 45 and above will be much less likely to share contacts. Stigma in these [minority ethnic] communities will play a big role. (K9) Thus, inequalities could be perpetuated in different waysthrough socioeconomic position, ethnicity/ race or through geography.

| Barriers to self-isolation
At the time of data collection, Scottish Government advice was to isolate for up to 14 days if one had COVID-19 or were a close contact of a positive case. It was recognized that there were barriers to self-isolation and these affected people differently according to their circumstances. While most public participants interviewed felt that they had strong support networks that could assist them if they required groceries and essential supplies, there was recognition that others may not have such support. One participant commented that successful isolation required reliance on local communities for help and talked about being reticent to ask strangers for help. The impact of self-isolation on people's mental health and well-being was also raised, particularly if they were asked to self-isolate more than once.
Self-isolation was also recognized to have financial implications including loss of earnings and a lack of sick pay, particularly for people on zero hours contracts and precarious employment.  49 Trust, however, is essential in adopting the measures designed to control the spread of COVID-19. 50 We, therefore, suggest that targeted information campaigns using trusted members of communities identified as marginalized should be implemented to give people the opportunity to ask questions, raise concerns and-crucially-have those concerns addressed.

| Strengths and limitations
A strength of this study was our ability to involve public participants and key stakeholders who were from, or represented, a diverse group of people who are often at the margins of society, including those experiencing severe and multiple disadvantage, homelessness or those from minority ethnic communities. Applying an equity lens using the PROGRESS framework identified the clear potential for widening and deepening the inequalities that are already present in our society and that COVID-19 has exacerbated. 3 Limitations included that we recruited key stakeholders rather than people with lived experience of some situations, for example, living with domestic violence. As interviews and focus groups were all conducted by Zoom, our sample had at least a degree of digital literacy. While participants reflected on the likely impact of having poorer digital literacy or lack of access, they could not fully represent the views of people in that situation. Lack of formal transcription was another limitation and could result in more superficial coding and analysis; however, we used recognized methods to approach analysis and interpretation of findings. [44][45][46] In addition, the use of the PROGRESS framework aided interpretation of our data. Finally, given the timeframe of data collection, we did not include anyone who had actual experience of using Test and Protect and its associated digital tools, which could identify other barriers, and facilitators, to contact tracing that we did not uncover.

| Comparison with previous literature
Previous work on the development of contact tracing tools for COVID-19 focused on public views of proximity tracing apps, often before such apps were fully developed. [27][28][29][30][31] However, few studies included people from marginalized populations. Our work is the first to actively recruit people, or their representatives, from a range of marginalized groups. While the digital tools had not yet been launched for public use, we were able to share the prototypes with participants, so they were able to see-and reflect on-the tools that would be implemented. Our work shows that the inequalities associated with COVID-19 infections and adverse health outcomes -namely, poverty, multiple disadvantage and ethnicity-also play a role in people's ability to participate in contact tracing 1-3 and mirror well-known barriers to accessing and using digital technology more generally. 13

| CONCLUSIONS
The impact of contact tracing and its associated digital tools on marginalized populations requires monitoring. This should include not only contact tracing itself but also the ability of people to self-isolate. Support for self-isolation is a vital part of contact tracing. Without that, people will make difficult decisions not to participate or may withhold information on contacts. Our work showed that it was possible to seek the views and opinions for those in marginalized communities and use the information to improve the design of the digital tools developed; this participatory approach allows the system to be 'sense-checked' with as diverse a group of people as possible.