Data source and participants
Data are from a substudy of Understanding Society (UKHLS), a longitudinal, nationally representative study of UK households, with everyone in the household interviewed annually. It is based on a two-stage stratified random sample of the household population in 2009, with two boost samples of immigrant and ethnic minorities [4]. During the months of April, May, June and July 2020, participants who had taken part in waves 8 or 9 of UKHLS were approached to complete a short web-survey that focused on the experience of COVID-19 symptoms, testing and hospitalisation and other aspects of life under the lockdown including health conditions and shielding [5]. This paper uses data from interviews conducted before the pandemic in 2019 (from waves 10 or 11 of the Study) and four months of the COVID-19 survey, April to July. All participants in our survey gave oral consent at each wave of data collection. Participants were enrolled only after consent was provided. Understanding Society has been approved by the University of Essex Ethics Committee. All methods were carried out in accordance with the approved guidelines and regulations. Further information can be found at https://www.understandingsociety.ac.uk/documentation/covid-19
Procedures
Respondents were asked in the COVID-19 survey to report whether they had any symptoms that could be coronavirus and whether they had been tested for coronavirus (and their test results). Participants were considered positive if they responded positively to any coronavirus symptoms or a positive test result in any of the survey months.
Respondents were also asked if they had received a letter or text from the NHS or Chief Medical Officer saying that they had been identified as someone at risk of severe illness if they caught coronavirus (hereafter referred to as “shielding letter”). As the UKHLS is a household survey with multiple respondents per household, we could identify whether someone was living in household in which either themselves or a co-resident had received a shielding letter. In addition specific to the June survey, an explicit question was asked on whether anyone in the household was shielding. We incorporated this information into our overall measure of anyone in the household shielding from April to July.
Outcomes
Common Mental Disorder (CMD) was measured using the 12-item General Health Questionnaire (GHQ-12) designed to capture depressive and anxiety symptoms and is a widely used measure of non-psychotic psychological distress [6]. Each item has four response categories on a Likert scale ranging from ‘not at all’ to ‘much more than usual’. Respondents who score three or more on the GHQ-12 have probable CMD.
Loneliness was assessed with the question’ In the last four weeks, how often have you felt lonely?’ with responses of ‘hardly ever’/’not at all’ and ‘some of the time’ grouped together and compared with ‘often’.
Co-variates
A range of questions on health conditions and treatments were asked, which allowed us to approximate the group of respondents who were at moderate and high risk of adverse outcomes following COVID-19 drawing on the criteria as used for the NHS shielding policy (https://digital.nhs.uk/coronavirus/shielded-patient-list). Two mutually exclusive groups were created . Medium risk was defined using the following criteria: aged 70 or older, having chronic (long-term) respiratory diseases, such as asthma, chronic obstructive pulmonary disease (COPD), emphysema or bronchitis; chronic heart disease, such as heart failure and coronary heart disease; chronic kidney disease, liver disease; chronic neurological conditions, such as Parkinson’s disease, motor neurone disease, multiple sclerosis, a learning disability or cerebral palsy, diabetes; being seriously overweight (a BMI of 40 or above) or being pregnant. High risk was defined using the following variables: on medication following organ transplant; people with bone or blood cancers or being treated with chemotherapy or targeted cancer treatments, all those with lung cancer being treated by radiotherapy ; people with severe respiratory conditions including cystic fibrosis, severe asthma and COPD; being treated with steroids; those reporting sickle cell anaemia or splenectomy; people on immunosuppression therapies; women who are pregnant with significant heart disease [7].
In addition, a broad range of social and demographic factors were measured before the pandemic from the 2019 interview [7]: Age (16–24, 25–34, 35–44, 45–54, 55–64, 65-74, ≥75 years), sex (women and men), ethnicity (White British, other White, Indian, Pakistani/Bangladeshi, Black Caribbean, African, Chinese/Other), cohabitation (living with a partner or not),number of children under 5 years old in household, housing tenure (house ownership or rent), employment status (employee, self employed, both employee and self employed or not in work). In the COVID-19 survey respondents in paid employment were asked whether they had been furloughed and how often they were able to work from home. We also included region in the UK and if their location was classified as urban or rural.
Statistical Analyses: The distribution of the dependent variables (coronavirus symptoms and tests, CMD and loneliness) by receipt of the shielding letter, COVID-19 risk status and social and demographic factors was examined (Table 1). In addition, we examined logistic regression models (Tables 2-5) to predict the dependent variables, examining the associations with the shielding letter (Model 1), adding in COVID-19 risk status (Model 2) and the interaction between the shielding letter and COVID-19 risk status (Model 3), controlling for a range of social and demographic factors. The interaction effect in Model 3 allows us to examine whether receipt of the shielding letter had a greater effect in those who were at higher risk of COVID-19.
All the standard errors in the regression model analyses were adjusted to take account of the clustered and stratified sample using the svy command in STATA. Models included inverse probability weights to take account of unequal selection probabilities into the study and differential nonresponse at each wave, including to the COVID-19 Survey. These longitudinal weights were constructed from the wave 9 weights and ensure the results are reliable estimates representative of the UK adult population living in private households [7].
Role of funding Source:
The study funders had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. All authors had full access to the data reported in the study and the final responsibility to submit for publication.