Anti-vaccine attitudes and risk factors for not agreeing to vaccination against COVID-19 amongst 32,361 UK adults: Implications for public health communications

Study design and participants

Data were drawn from the COVID-19 Social Study; a large panel study of the psychological and social experiences of over 75,000 adults (aged 18+) in the UK during the COVID-19 pandemic. The study commenced on 21 March 2020 and involves online weekly data collection from participants for the duration of the COVID-19 pandemic in the UK. The study is not random and therefore is not representative of the UK population. But it does contain a well-stratified sample that was recruited using three primary approaches. First, snowballing was used, including promoting the study through existing networks and mailing lists (including large databases of adults who had previously consented to be involved in health research across the UK), print and digital media coverage, and social media. Second, more targeted recruitment was undertaken focusing on (i) individuals from a low-income background, (ii) individuals with no or few educational qualifications, and (iii) individuals who were unemployed. Third, the study was promoted via partnerships with third sector organisations to vulnerable groups, including adults with pre-existing mental health conditions, older adults, carers, and people experiencing domestic violence or abuse. The study was approved by the UCL Research Ethics Committee [12467/005] and all participants gave informed consent.

For these analyses, we focused on individuals gave data during the month in which the vaccine module was administered (7 September to 5 October 2020). A total of 32,585 participants met this criterion. We then excluded participants with any missing data on vaccine outcome variables (n = 12). Due to insufficient statistical power, we also excluded individuals who had selected “other” in response to gender (n = 134) and “prefer not to say” on ethnicity (n = 95). Seventeen of these individuals selected both responses, leaving a total analytical sample size of 32,361.

Outcome variables

Negative general attitudes towards vaccines were measured using the 12-item Vaccination Attitudes Examination (VAX) Scale. Participants were asked to focus on vaccines in general rather than specifically on vaccines for COVID-19. Responses were rated on a 6-point scale from 1 “strongly agree” to 6 “strongly disagree.” Four subscales which have previously been derived¹⁶ were calculated: 1) mistrust of vaccine benefit, (2) worries about unforeseen future effects, (3) concerns about commercial profiteering, and (4) preference for natural immunity. Adequate convergent validity and internal reliability was established for all 4 subscales in two adult samples (Cronbach’s alphas = 0·77-0·93). Internal consistencies in the current sample were good (Cronbach’s alphas 0·91-0·94). Each of the four subscales was grouped into high (a score of 5-6 on a scale of 1-6), intermediate (score of 3-4), and low (score of 1-2) levels of negative attitudes towards vaccines.

Uncertainty and unwillingness to vaccinate against COVID-19 when available were based on one item (see Supplemental Table S1 for question wording). Response options ranged from “1- very unlikely” to “6 – very likely”. An ordinal variable was coded: (0) intend to vaccinate (responses of 5-6), (1) unsure about whether to vaccinate (responses of 3-4), and (2) unwilling to vaccinate (responses of 1-2).

Predictor variables

Socio-demographic factors included gender (male vs. female), age group (65+, 50-64, 30-45, and 18-29) ethnicity (white vs. BAME groups [i.e. Asian/Asian British, Black/Black British, White and Black/Black British, Mixed race, Chinese/Chinese British, Middle Eastern/Middle Eastern British, or other ethnic group]), education (postgraduate degree, undergraduate degree [further education after the age of 18], A-levels (equivalent to education to age 18) or vocational training, GCSE or lower [equivalent to education to age 16] and no formal qualifications), income (annual household income: >£90,0000, £60,000-89,999,£30,000-59,999, £16,000-29,000, and < £16,000) employment status (not employed vs. employed at the start of the pandemic), living arrangement (live alone vs. not alone), area of dwelling (urban [city, large town, small town] vs. rural [village, hamlet, isolated dwelling]), the presence of a child in the household (no children in the household vs. children in the household), and government’s identified key worker status (not a key worker vs. key worker). The latter included people with jobs deemed essential during the pandemic (e.g. health and social care, education and childcare) and who were required to leave home to carry out this work during the lockdown.

Participant reports of whether they had received clinical diagnoses of a mental illness (e.g. depression, anxiety, or other psychiatric diagnosis) or chronic physical health condition (e.g. high blood pressure, diabetes, heart disease, lung disease (asthma or COPD), cancer, or other physical health condition) were used to create two binary variables (yes/no) to indicate the presence or absence of pre-existing physical and mental health conditions.

Coronavirus anxiety during the past two weeks was measured using the Coronavirus Anxiety Scale (CAS).¹⁷ The CAS contains 5 items with 5-point responses ranging from “not at all” to “nearly every day”. The scale has shown good internal reliability (Cronbach’s alpha = 0·93), construct validity, diagnostic viability, and equivalency of measurement across demographic groups.¹⁷ A CAS score of 9 or more classified adults as meeting (90% sensitivity) or not meeting (85% specificity) the threshold for Generalised Anxiety Disorder.¹⁷ We categorised responses such that participants with one or more COVID-19 anxiety symptom were compared to those who did not report any such symptoms.

Confidence in government and the health service to handle the pandemic were assessed with one question each. Response options ranged from 1 (none at all) to 7 (lots). Two binary variables were created to compare individuals who had a lot of (5-7) versus low (1-4) confidence in the government and health system.

Responses to the question on compliance with government COVID-19 guidelines were on a scale from 1 (none at all) to 7 (very much so). We analyse this as a binary variable reflecting higher (6-7) vs lower (1-5) compliance. Knowledge of COVID-19 was measured with the question: rated on a 7-point scale from 1 (very poor knowledge) to 7 (very good knowledge). Responses of 1-4 were categorised as very poor/poor compared to very good/good (5-7) COVID-19 knowledge. The presence or absence of having had COVID-19 was categorised as a binary variable (yes, diagnosed and recovered, or yes, diagnosed and still ill, or not formally diagnosed but suspected, versus no, not that I know of or no). Prior vaccine behaviour was based on two yes/no questions. See Supplemental Table S1 for a listing of all question wording.

Statistical analysis

Ordinary least squares (OLS) regressions were carried out to examine socio-demographic and COVID-19- related predictors of each of the four negative attitudes toward vaccines subscales. We then fitted one multinomial logistic regression model to examine associations of socio-demographic and COVID-19- related factors, negative vaccine attitudes, and prior vaccine behaviours with intent to vaccinate against COVID-19. The outcome variable in the latter model was coded such that those likely to vaccinate were compared to i) uncertainty about whether to vaccinate and ii) unwillingness to vaccinate. Results are presented as relative risk ratios (RRR) with corresponding 95% confidence intervals (CI). Missing data

The pattern of missing data in the study sample is presented in Supplemental Table S2. The proportion of missing data ranged from 0.01% for having had COVID-19 to 19.66% for the Coronavirus Anxiety Scale. Multiple imputation by chained equations¹⁸ was used to generate 50 imputed datasets for each variable in participants with complete data on all vaccine outcomes (N = 32,361). Imputation models included all variables used in the analysis, as well as additional auxiliary variables (home ownership, anxiety symptoms, depressive symptoms, smoking status). Substantives results using cases without any missing data and the imputed sample were similar (Supplemental Tables S3 and S4). To account for the non-random nature of the sample, all data were weighted to the proportions of gender, age, ethnicity and education obtained from the Office for National Statistics.¹⁹ Analyses were conducted using Stata version 16.²⁰

Role of the funding source

The funders had no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication. All researchers listed as authors are independent from the funders and all final decisions about the research were taken by the investigators and were unrestricted. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Predictors of negative attitudes towards vaccines

Results from ordinary least squares regressions predicting four domains of negative vaccine attitudes are presented in Table 2. The strongest associations with negative vaccine attitudes were for variables representing socio-economic status and suggest a gradient such that lower levels of household income and education were associated with progressively more negative views on vaccines across all four domains. In addition, people from BAME groups, those who reported poor compliance with government COVID-19 precautions, and who had low self-rated COVID-19 knowledge also had more negative vaccine views on all four subscales. Women were more likely to express concerns specifically about unforeseen effects of vaccines and less of a preference for natural immunity, as were people without long-term mental health conditions. Low confidence in the health system to handle the pandemic was also associated with greater mistrust of vaccine safety, more worries about unforeseen vaccine effects, and greater concerns about commercial profiteering, whilst low confidence in government to handle the pandemic was associated with lower scores on worries about unforeseen effects and preference for natural immunity. Finally, there was a relationship between previous experience of COVID-19 symptoms and greater negative attitudes. Young people (ages 18-29) were significantly less likely than older adults (ages 65+) to have negative attitudes towards vaccines on all four domains.

Predictors of uncertainty and unwillingness to vaccinate against COVID-19

Results from the multinomial logistic regression model predicting risk for uncertainty and a lack of intent to vaccinate against COVID-19 are shown in Table 3. Misinformation and mistrust in vaccines across all four domains (most strongly concerns about unforeseen side effects and mistrust in the benefit of vaccines) were associated with a greater likelihood of vaccination uncertainty and unwillingness for COVID-19. Strong and intermediate levels of mistrust of vaccine benefits were each associated with a 5 times higher relative risk of being unwilling to get a COVID-19 vaccine. Similarly, vaccine unwillingness was predicted by strong worries about unforeseen effects (RRR = 4·91; 95% CI: 3·76 to 6·42), intermediate (but not strong) concerns about commercial profiteering (RRR = 1·73; 95% CI: 1·34 to 2·24), and strong preference for natural immunity (RRR = 2·51; 95% CI: 1·78 to 3·53).

Poor compliance with COVID-19 guidelines and low knowledge about COVID-19 also predicted both vaccine hesitancy and vaccine unwillingness. Further, people who did not receive a flu vaccine last year were twice as likely to be unsure about a COVID-19 vaccine (RRR = 1·93; 95% CI: 1·67 to 2·23) and 3·4 times more likely to have decided against having a COVID-19 vaccine (RRR = 3·40; 95% CI: 2·75 to 4·20).

Demographically, groups at increased risk for uncertainty and unwillingness to vaccinate against COVID-19 were women (uncertain: RRR =1·45; 95% CI: 1·27 to 1·65; unwilling: RRR = 1·52; 95% CI: 1·24 to 1·86), those who were keyworkers (uncertain: RRR = 1·18; 95% CI: 1·01 to 1·38), and people living with children (uncertain: RRR = 1·38; 95% CI: 1·13 to 1·70; unwilling: RRR = 1·60; 95% CI: 1·24 to 2·08).

A socio-economic gradient was also evident for uncertainty and unwillingness to receive the COVID-19 vaccine, with people with lower levels of education increasingly more likely to be unwilling and those with lower incomes more likely to be uncertain. Age was unrelated to uncertainty around the COVID-19 vaccine and only slightly related to unwillingness, with adults over 65 more likely to be willing than younger adults (ages 30-49 and 50-64) to get the COVID-19 vaccine.

Finally, ethnicity, long-term mental and physical health conditions, and low confidence in government to handle the pandemic were unrelated to intentions relating to the COVID-19 vaccine.