Coronavirus (COVID-19) support study experiences of and compliance with self-isolation: main report

2. Methods

2.1 Study design

In order to fully understand the knowledge, attitudes, experiences and support needs of those being asked to self-isolate, a mixed method design was required. A quantitative online survey (with the option of telephone completion instead) and qualitative telephone/video interviews were identified as the most appropriate and feasible means of meeting the study objectives and to generate a robust evidence base to inform policy.

Both the survey and qualitiative interviews involved collecting data from adults aged 16 and over from three key groups:

• Those asked to self-isolate after testing positive for COVID-19 (Index Case).
• Those who had been in close contact with someone who tested positive, either within their own household or from a separate household (Contact Cases).
• Those who had returned from international travel (International Travellers) (see below for details on changes to the self-isolation rules for this group during wave 3 of the survey).

More detail on each of the study strands can be found under the relevant headings below. A glossary of some of the key terms used throughout the report can be found in Appendix A.

2.2 Survey

2.2.1 Sample

The aim was to invite potential participants in each of the three case types (Index, Contacts and International Travellers) on or as close as possible to day eight of their self-isolation. This allowed more time for these individuals to have experienced self-isolation but was also intended to reduce the potential burden on those recovering from COVID-19 who may wish to respond to the survey.

Eligibility

Eligibility was established separately for Index and Contact Cases and International Travellers.

Index and Contact Cases' eligibility

The following outlines the sequence of steps taken to identify Index and Contact Cases in the Test and Protect system eligible to take part in the study. A case was eligible if:

1. They agreed, when asked by Test and Protect, to future contact about follow-up research^[12]
2. They were aged 16 or over
3. They did not travel in the last 14 days^[13]
4. The onset of their symptoms (or, for Contact Cases, the onset of symptoms for the person they were in close contact with) fell within the relevant fieldwork period
5. They were on/close to Day 8 of their official self-isolation period
6. Still met criteria outlined at steps 2 and 3 after steps 4-5 had been completed^[14]

Cases that met these criteria were deemed suitable to receive a survey invitation.

International Travellers' eligibility

The following outlines the sequence of steps taken to identify International Travellers in the Test and Protect system eligible to take part in the study. A case was eligible if:^[15]

1. They were identified as an International Traveller in the Test and Protect System
2. They agreed, when asked by Test and Protect, to future contact about follow-up research
3. They were aged 16 or over
4. They did not provide a Home Office email address^[16]
5. They were on Day 8 of their official self-isolation period on the day invitations were issued

Cases that met these criteria were deemed suitable to receive a survey invitation.

Recruitment

In accordance with data protection regulations and consent given to uses of the personal data of those in the Test and Protect database, invitations to take part in the survey were issued by Public Health Scotland. Due to the way different case type information is held it was decided that all eligible Index and Contact Cases would be invited to take part via an SMS invitation, while International Traveller Cases were sent an email invitation.

Survey invitations were designed in conjunction with and approved by the Scottish Government research team that included:

• A link to the online survey.
• A unique eight digit access code for every individual invited.
• The survey freephone number and option of undertaking a telephone interview.

ScotCen created the survey link and generated unique access codes for the survey. The survey link and access codes were provided to Public Health Scotland (PHS) who issued invitations daily by SMS or email to those individuals on the Test and Protect system who met the eligibility criteria on a given day.^[17] No reminders were issued.

To maintain confidentiality and in accordance with the study privacy policy, anything that could link access codes to completed surveys were not provided by ScotCen to any external organisations including the Scottish Government, Public Health Scotland or any other third party.

In order to ensure that participants were fully informed and supported during the research process, a website containing further details on the study, the privacy notice and a list of relevant organisations that may be useful to potential participants were also made available, along with the ScotCen freephone number and email address for any queries or additional feedback. A Scottish Government email address was also provided. Links to these resources were included within the email invitations, however, due to character limitations for text messages and to avoid the participant burden of receiving too many messages, these details were also provided at the start of the online/telephone surveys and several of the useful organisations were also listed at the end of the survey (see questionnaire in Appendix B, available in the Supporting Files).

2.2.2 Fieldwork

Survey fieldwork took place over three distinct fieldwork periods:

Wave 1: 19^th – 31^st March 2021
Wave 2: 12^th April – 5^th May 2021
Wave 3: 10^th May – 2^nd June 2021

As anticipated at the design stage, to ensure a sufficient sample size the fieldwork periods for waves two and three were extended from the original timings as the falling case numbers during those periods meant a reduction in the eligible population. An increase in the numbers eligible to take part, along with a higher than anticipated response among International Travellers during wave two, allowed the invitation window for this group to be reduced by one week in wave three.

2.2.3 Questionnaire

The questionnaire was designed to take an average of 15 minutes to complete online and covered several topics under three broad themes of interest:

Figure 2.1 Survey topics

Experiences	Support/guidance	Understanding and opinion
Current isolation status & reason for self-isolating	Contact & guidance offered	Understanding of self-isolation guidelines
Experiences of self-isolation	Whether applied for Self-isolation Support Grant	General opinions on self-isolation guidelines and behaviour
Experiences of being tested for COVID-19	Support offered/ accessed/ needed from Local Authorities

Additional demographic details were also collected to aid analysis. A copy of the full questionnaire can be found in Appendix B in the Supporting Files for this report. The questionnaire was identical for those who opted to take part over the telephone.

The questions were converted into web and telephone survey programs with in-built routing and filtered question wording appropriate to the individual participant's circumstance (for example, whether an individual was still isolating or not). Participants were informed that they could skip any questions that they did not wish to answer.

The questions were identical across all the three waves with the exception of two new questions introduced in wave 2 and also included in wave 3: shielding status and indoor close contact with non-household members.

2.2.4 Survey response

The overall response rate for each of the survey waves ranged from 7% in wave 1 to 9% in wave 3. Reponse was highest among International Travellers (in the range 12%-14%) and Index Cases (11% in all waves) and lowest among Contact Cases (4%-5%). [Figure 2.2]

Figure 2.2 Survey response

Activities and Response	Total Wave 1	Index Cases Wave 1	Contact Cases Wave 1	International Travellers Wave 1	Total Wave 2	Index Cases Wave 2	Contact Cases Wave 2	International Travellers Wave 2	Total Wave 3	Index Cases Wave 3	Contact Cases Wave 3	International Travellers Wave 3
Invitations issued (n)	14705	3655	9135	1915	23403	2592	12458	8353	21626	2327	11382	7917
Failed delivery (n)	1954	140	1803	11	2049	136	1873	40	2204	108	2058	38
Invitations delivered (n)	12751	3515	7332	1904	21354	2456	10585	8313	19422	2219	9324	7879
Total completes (n)	917	385	267	265	1748	276	439	1033	1,660	254	455	951
Response rate* (%)	7%	11%	4%	14%	8%	11%	4%	12%	9%	11%	5%	12%

Survey questionnaires were classifed as 'completes' and included in analysis if the participant answered questions up to, and including, the reason for their self-isolation or beyond.

2.2.5 Sample profile (Index and Contact Cases)^[18]

The sex, age and area deprivation profile of the survey sample was well-balanced and appeared to match the Test and Protect system profile, over the same period, reasonably well. This is encouraging given the opt-in nature of the survey recruitment. The proportions of 25-44 year olds in the survey sample were similar to those in the Test and Protect system over the same period. Similarly, the proportions of participants living in SIMD quintiles 2, 3 and 4 were similar to those observed in Test and Protect. Nevertheless, some groups were under-represented in the survey, notably: male Index and Contact cases, those aged 16-24 and those living in Scotland's most deprived SIMD quintile. It is not uncommon for these groups to be under-estimated in social surveys, even those drawn using random probability sampling. [Figure 2.3]

Figure 2.3 Sex, age and area deprivation profile of Index and Contact Cases in the survey sample and in the Test and Protect system during survey fieldwork

Demographics	Test & Protect system	Survey sample
Sex
Female	53%	67%
Male	46%	32%
Age group
16-24	25%	14%
25-34	21%	18%
35-44	21%	23%
45-54	16%	24%
55+	17%	21%
SIMD Quintile
1 – Most deprived	26%	20%
2	22%	20%
3	17%	18%
4	18%	20%
5 – Least deprived	17%	22%

2.2.6 Analysis

Analysis of the survey data was undertaken in SPSS. The survey data has not been weighted as data on the key characteristics of the population from which the sample was from was not available at the time of analysis.

Frequencies and bivariate analysis have been used to describe the survey data and to identify associations between variables of interest. A wide range of independent variables were included across the bivariate analysis although the exact variables explored varied slightly depending on the dependent variable under investigation. Socio-demographic variables explored included: sex; age group; area deprivation (SIMD quintiles); household income; household composition; perceived financial hardship; ethnicity; carer status and long-standing illness. Some of the COVID-19 specific independent variables included in analysis were: case type; booking type (for International Travellers only); shielder status; vaccination status; offer of Local Authority support during self-isolation; acceptance of Local Authority support and isolation history. Variables on knowledge of, and attitudes towards, self-isolation were also included.

Throughout the report, the terms 'statistically significant' and 'significant' are used to imply significance at the 0.05% level.^[19] It should not, however, be assumed that all observed differences in Excel tables are statistically significant.

A limitation of bivariate analysis is that while it can identify the existence of an association between two variables, it cannot explain the direction of that relationship and nor does it mean that all within a particular sub-group (e.g. all aged 16-24) think or behave in the same way. Similarly while it gives an indication of potential association between two variables, it cannot account for the other associations at play.

When making inferences from the results it should be borne in mind that as the sample was opt-in, and not randomly drawn from the eligible population, the results describe the patterns and associations found among those who took part in the survey as opposed to all adults asked to self-isolate. It should also be borne in mind that some groups in particular appeared to be under-represented in the sample including: male Index and Contact Cases; Index Contact Cases aged 16-24 and Contact Cases from Scotland's two most deprived quintiles.

To ensure that the disproportionately large number of International Traveller participants' did not unduly bias overall totals presented, their data has been analysed and presented separately to Index and Contact Case data. In addition, differences in the self-isolation requirements for the International Traveller group warranted some separate questions and analysis for this case type. Where appropriate, findings for International Travellers have been compared with Index and Contact Case participants combined.

2.2.7 Data tables

The figures quoted in this report are drawn from data tables available as supplementary files to this report. Tables display frequencies for survey responses and derived variables (such as the behavioural measure of compliance). Data that is discussed in the main report will be referred to as, for example, Figure 1. Where data is only available in the supplementary files it will be referred to as, e.g.Table 1.

2.3 Qualitative interviews

Sample and recruitment

Survey participants were asked during the questionnaire if they were happy to consent to be approached to participate in an in-depth telephone/video interview, in which their experience of self-isolation would be explored in more detail. All of those who consented to be recontacted from wave one of the survey formed the sampling frame for the qualitative component.

Qualitative purposive sampling was conducted in order that survey participants were contacted to be interviewed based on the criteria outlined below. The main sampling criteria were: level of compliance with self-isolation requirements (self-reported and observed (see Chapter 3 for more on this)); case status; and receipt of local authority support. It was also important to include a range of participants across demographic variables such as age, sex, SIMD, ethnicity and employment status. Figures 2.4 – 2.6 demonstrate that the research team was successful in interviewing 30 participants across a range of purposive sampling requirements.

Figure 2.4: Rates of compliance of final qualitative sample (30 interviews)

Rate of compliance:	Fully Compliant	Mostly Compliant	Partially Compliant	Not Compliant
Number of interviewees:	5	4	19	2

Figure 2.5: Age range of final qualitative sample (30 interviews)

Age	16-24	25-34	35-44	45-54	55-64	65-74	75+
Number of interviewees:	2	7	5	6	4	5	1

Figure 2.6: Scottish Index of Multiple Deprivation ( SIMD) quintiles of final qualitative sample (30 interviews)

SIMD	1 (most deprived)	2	3	4	5 (least deprived)	No data
Number of interviewees:	10	4	3	2	6	5

The range of other purposive sampling requirements were also met. Out of the thirty interviewees:

• 10 were index cases; 7 were household contact cases; 6 were non-household contact cases; and 7 were international travellers
• Fifteen were male and 15 were female
• Sixteen were working^[20]; 3 were in education^[21]; and 11 were neither working nor in education^[22].
• Six had accepted Local Authority support to self-isolate; 15 had not accepted it; and 9 said they had not been offered it (including all the international travellers).

The interview sample was made up of a majority of White participants (n=25).

Those who consented to be recontacted were asked to provide an email adress and/or a contact phone number. In total, 67 individuals were approached by ScotCen in order to achieve the 30 interviews. All potential interviewees were provided with a study information sheet outlining the nature of the qualitative research, what participation would mean and contact information for members of the research team. The participants were then able to opt in to the qualitative research directly, or to contact the research team with any questions.

Qualitative fieldwork

The interviews took place between the 15^th of April and the 14^th of May 2021 and lasted, on average, 45 minutes. Verbal consent was sought at the start of the interview, and the participants were also asked for their agreement that the interviews would be digitally recorded and fully transcribed. Interviewees were given an incentive of a £30 gift card or online e-voucher on completion of an interview.

Qualitative topic guide

The interview topic guide was developed by the research team and agreed with the Scottish Government prior to fieldwork commencement. The interviews covered the following broad areas in detail:

• Covid-19 beliefs and other contextual factors
• Experiences of self-isolation
• Experiences and perceptions of support received
• Factors that helped and hindered self-isolation
• Future changes that would enable self-isolation to operate more smoothly.

The nature of qualitative interviewing meant that while the topics above were all covered, this mode offered flexibility in the order in which they could be addressed so that the conversation flowed naturally for each interview rather than following a set order. The full topic guide is outlined in Appendix C in the Supporting Files for this report.

Analysis

Completed transcripts were checked, edited and entered into qualitative analysis software NVivo 12 to aid the thematic analysis. A coding framework was developed and agreed by the research team. After coding was completed, a detailed analysis was undertaken in which the main emergent themes were identified, explored and developed. Comparison and cross-referencing within and between groups was also conducted, for example, by examining the responses of Contact and Index Cases. The main emergent themes were summarised, developed and written up, and are included in the main results section of the report, along with verbatim, anonymised quotations.

2.4 Changes in self-isolation guidance during the research

For Index and Contact Cases the requirements and guidance around self-isolation did not change during the life of the research project.

With regards international travel, in mid-May 2021, Scotland was aligned with the rest of the UK with the introduction of a traffic light system, designed to reflect the current situation in international countries with regards to the pandemic and transmission risk of COVID-19 and any variants. During the latter phase of wave 3 fieldwork this system, updated every three weeks, provided information on countries that should not be travelled to for holiday or leisure purposes and which require self-isolation in managed accommodation (red countries), those it was advised not to travel to and that require self-isolation at home upon returning (amber countries) and those it was possible to travel to without self-isolation but with the appropriate testing (green countries). Prior to the introduction of this system, all International Travellers returning to Scotland were required to enter managed isolation in a hotel, whereas those that arrived via another UK country, were required to self-isolate and test at home. Since fieldwork completed further changes have been made to the the traffic light system and whether people need to isolate once fully vaccinated when returning from an amber country, but the above were the regulations in place over the course of this reseach.

2.5 Ethics

Prior to commissioning, the study was subject to the Scottish Government's ethical review process. A steering group comprised of representatives from across the Scottish Government, Test and Protect, NHS Scotland and Public Health Scotland had oversight of the study, while the day-to-day management and oversight was undertaken by the Scottish Government Health and Social Care Analysis team. The study was also reviewed and approved by NatCen's Research Ethics Committee in February 2021.

2.6 Strengths and limitations of the study

Study Strengths

• The study included all case types in the Test and Protect system allowing comparisons across case types to be made.
• To measure actual behaviour and minimise recall bias, participants were invited to take part in the study during their official self-isolation period.^[23]
• By asking questions regarding individual components of self-isolation compliance (e.g. how quickly someone started isolating, length of the isolation period etc.), as well as for the participants' own assessment of their level of self-isolation compliance, the survey was able to measure and report in more than one way, and at more than one point in time, in the self-isolation period.
• Survey fieldwork was carried out over more than one wave across a three month period. This meant that the data was able to capture the changing nature of self-isolation experiences and developments in the pandemic response, such as the widespread availability of free lateral flow tests and changes to the rules for international travel.
• Carrying out fieldwork over more than one wave also enabled for interim data to be delivered that allowed the survey findings to be reported and acted upon in real time before the final report was delivered.
• The survey method was developed to be as inclusive as possible with the option of telephone surveying for anyone who was invited but was unable or unwilling to complete the survey online.
• The broad nature of the survey allowed data to be captured that enabled analysis based on a range of demographic, geographic, behavioural and attitudinal sub-groups. This not only allowed a picture of self-isolation compliance, attitudes and support needs to be built up by case type but also for different sub-groups to allow the nuances of experiences, attitudes and needs to be explored, to ultimately inform policy development and implementation.
• The survey allowed for a broad range of topics to be investigated across large numbers of participants, while the qualitative interviews provided a depth and additional understanding of personal experiences of self-isolation that added a greater richness to the survey data.
• Statistical testing was used to determine if observed differences across different groups of participants were statistically significant or not, adding weight to the survey findings reported and clarity regarding factors that are impacting on the self-isolation experiences of individuals.

Study Limitations

• The self-selecting nature of the sample was a limiting factor for the study. The opt-in approach required for this survey means that there was potentially higher representation among participants who were more likely to be compliant with the requirements to self-isolate, while those that did not take part in the survey were potentially less compliant with the requirement to self-isolate. While efforts to emphasise the value of the study and its non-judgemental and confidential nature were made, the nature of the recruitment process, particularly the use of a single SMS for Index and Contact Case invitations issued by Test and Protect, limited the opportunity for these assurances. It is important to note, that while qualitative sampling was undertaken to interview a range of case types, the sample of those available for this was also drawn from those who agreed to take part in a follow-up qualititative interview. It is therefore possible that those who did not consent to follow-up research were somehow different in terms of their compliance than those who did participate in qualitative interviews.
• The agreed recruitment process dictated that participants could only receive one piece of communication inviting them to take part. A reminder strategy may have yielded a higher response rate among groups typically harder to reach in surveys (e.g younger people, those in areas of greater deprivation) and which we know were under-represented in the survey sample.
• While invitations were issued during a case's self-isolation period, it is likely, particularly among those that chose to complete the survey after their self-isolation period had ended, that the accuracy with which some participants were able to recall past events and experiences was subject to a degree of recall bias.

While every care was taken, at the design-stage, to ensure non-judgemental language was used across the survey and interviews, it is likely that the nature of the topic will have resulted in social-desirability bias. Given participants were explicitly asked about activites the Scottish Government has requested they do not do, it is likely that some participants did not accurately report carrying out such behaviour in case it was viewed unfavourably. The most likely impact on the data is that the figures in this report over-state compliance.