key: cord-0973445-e4nxtioh
authors: Blanchard-Rohner, Géraldine; Caprettini, Bruno; Rohner, Dominic; Voth, Hans-Joaquim
title: Impact of COVID-19 and ICU capacity on vaccination support: Evidence from a two-leg representative survey in the UK
date: 2021-05-17
journal: J Virus Erad
DOI: 10.1016/j.jve.2021.100044
sha: 4bdb95765032bef6895a1438f392992622868acf
doc_id: 973445
cord_uid: e4nxtioh

BACKGROUND: Overcoming coronavirus disease (COVID-19) will likely require mass vaccination. With vaccination scepticism rising in many countries, assessing the willingness to vaccinate against COVID-19 is of crucial global health importance. OBJECTIVE: The goal of this study was to examine how personal and family COVID-19 risk and ICU (intensive care unit) availability just before the pandemics influence the acceptance of future COVID-19 vaccines. METHODS: A two-leg survey was carried out for comparing vaccination attitudes pre-and post-COVID-19. UK residents were surveyed in October 2019 about their vaccination attitudes, and again in a follow-up survey in April 2020, containing the previous questions and further ones related to COVID-19 exposure and COVID-19 vaccine attitudes. The study combined survey results with local COVID-19 incidence and pre-COVID-19 measures of ICU capacity and occupancy. Regression analysis of the impact of individual and public health factors on attitudes towards COVID-19 vaccination was performed. RESULTS: The October 2019 survey included a nationally representative sample of 1,653 UK residents. All of them were invited for the follow-up survey in April 2020, and 1,194 (72%) participated. The April 2020 sample remained nationally representative. Overall, 85% of respondents (and 55% of vaccine sceptics) would be willing to be vaccinated against COVID-19. Higher personal and family risk for COVID-19 was associated with stronger COVID-19 vaccination willingness, whereas low pre-COVID-19 ICU availability was associated with lower trust in medical experts and lower COVID-19 vaccine support. Further, general vaccination support has risen during the COVID-19 pandemic. CONCLUSION: Support for COVID-19 vaccination is high amongst all groups, even vaccine sceptics, boding well for future vaccination take-up rates. Vaccination willingness is correlated with health care availability during the COVID-19 crisis, suggesting a powerful synergy between health care system performance during crisis and the general population’s trust in the medical profession – as reflected in vaccination support.

Since January 2020, the coronavirus has infected over 50 million people, leading to more than 3.2 million deaths. As of November 2020, no effective treatment exists; only large-scale vaccination will allow a return to 'normal life'. The novel threat posed by COVID-19 and the need for mass vaccination comes at a time of growing vaccine hesitancya reluctance or refusal to vaccinate despite vaccine availabilityin developed countries (1). For example, only 47% of French citizens, 67% of Germans, and 72% of Americans currently believe that vaccines are safe (2) . Accordingly, vaccination rates have fallen in many rich countries in recent years, such as in the UK, France and the Netherlands (3, 4) . Even before the COVID-19 crisis, the World Health Organization ranked vaccine hesitancy as one of the top ten global health threats (5) . For all these reasons, the National Institutes of Health (NIH) recently expressed concern that vaccine scepticism could undermine COVID immunization efforts (6) .

Vaccine hesitancy creates a major risk of suboptimal health outcomes for the general population (7) and has several causes (8, 9) : lack of confidence, complacency, constraints, calculation, and lack of collective responsibility. Growing vaccination fatigue and dropping vaccination rates have variously been attributed to low public trust in vaccines and health system, lack of perceived risks, wrong beliefs and misinformation, including the spread of fake news through social media, constraints on affordable and accessible vaccines, and insufficient governmental vaccine investments (5, (10) (11) (12) (13) (14) . Significant resistance to measles vaccine can be traced back to a single scientific paper retracted in 2010 (15) .

Further, the limited risk of contracting vaccine-preventable diseases (VPDs) and the perceived low severity of VPDs have arguably reduced vaccination rates prior to COVID-19 (11) . Also, "free-riding" is a distinct possibility, with citizens deliberately benefitting from the vaccination efforts of others while reducing their own, resulting in socially sub-optimal vaccination rates (7) .

In general, few socio-economic variables predict vaccine scepticism. While education and higher income are associated in some countries with higher vaccination rates, in others they are correlated with vaccine hesitancy (16) . Older respondents are generally less vaccinehesitant, while the unemployed tend to be more sceptical (17) . Individual vaccination choices are often strongly influenced by a person's social network and prevailing local social norms (16) . In a representative sample of UK residents, almost a third of respondents were vaccinehesitant for at least half (5 out of 10) of the assessed questions (17) . Lack of awareness of vaccines' benefits and fear of side effects were the main reasons for vaccine-skepticism. In particular, 79 percent of vaccine-hesitant respondents stated lack of trust in immunisation programs as the reason for hesitancy (12) .

To date, few interventions have been successful in overcoming vaccine hesitancy. While information provision can have some effect (such as e.g. educational pamphlets or web-based decision aids), many interventions yield neutral or counterproductive results (18) . In particular, only half (5 out of 10) of the interventions using educational information to boost vaccination rates have led to a significant increase in vaccination intentions (13) . Even in the case of successful interventions (19) , a substantial share of anti-vaccine militants remains who are concerned about autism and bowel disease, despite being confronted with robust J o u r n a l P r e -p r o o f scientific evidence. Proximity to actual disease outbreaksas in the case of measlesappears to have no overall effect on attitudes towards vaccines (20) .

The historic COVID-19 pandemic may affect attitudes towards vaccination for several reasons: First, COVID-19 involves higher personal stakes than many other diseases. In particular, some vaccine-preventable diseases are perceived as relatively mild, and have a low probability of infection (11) . In contrast, because COVID-19 is highly infectious, and has a relatively high case-mortality ratewith no hope for herd immunity anytime soon, incentives for vaccination are greater. Second, suspension of social distancing measures and economic recovery will require successful immunization of large portions of the global population against COVID-19 (21). Third, the success of public health systems in dealing with the historic COVID-19 challenge may affect support for alternative (non-scientifically evaluated) approaches and methods.

Our study aims to assess attitudes towards COVID-19 vaccination, and to examine how the COVID-19 crisis affects support for vaccination in general. The effect of socio-economic characteristics and public health parameters on these attitudes were also considered. Our data come from two surveys and a number of publicly available sources. Quota targets were taken from the most recent official statistics (census and electoral data).

Surveys are a powerful tool for predicting actual vaccination decisions (22) . This first survey mainly asked whether respondents would favor penalties for parents who refuse to vaccinate their children, in the form of fines, child benefit withdrawal, or bans from school. The same 1,653 UK residents were contacted again for a second survey in April 2020, England. Standard errors were clustered at the level of the Local Authority to account for intra-cluster correlation of errors. We reported results for the full sample and for three subsamples which were labelled as "no vax," "hesitants," and "pro vac." Respondents were assigned to one of these three categories based on their answers to a set of 8 questions. Each question presented a statement about vaccines and asked the respondent to rate it on a 4-level scale: "definitely true," "probably true," "probably false" and "definitely false." We generated an aggregate score based on these answers and used it to divide respondents into the three categories (See Section S.2 in the Supplementary Materials for additional details).

All participants have given written consent for survey participation and for results to be published. All the statistical analysis and data storing used anonymized data that did not allow to identify individual participants.

The current study has been approved by the institutional ethics review board of HEC Lausanne, University of Lausanne (assigned acronym: VACAT). This study has been recorded at ClinicalTrials.gov, with identifier NCT04352582. Results 1,653 (i.e. 93% of the respondents registered for participation and directed to the specific survey on vaccine attitudes) completed the questionnaire. 1,194 (72%) of the initial participants responded to the second survey. This follow-up survey in April 2020 was also nationally representative (see Table 1 ). Average socio-demographic characteristics were very similar for the October 2019 and April 2020 samples and the average of the UK population as a whole. When studying if selection bias could be a concern, the non-response rates were not correlated with answers to the October 2019 vaccination questions (Table S1 ).

In our nationally representative sample, 85% of respondents were either definitely or probably willing to become vaccinated against COVID-19 (see Figure 1 , left bar). Only 8% said that they would either probably or definitely not take the vaccine. Attitudes towards vaccination in general correlated with willingness to receive a potential COVID-19 vaccine.

In the group of people generally favorable to vaccination (i.e. the ProVac group), 95% stated that they would like to be vaccinated. However, even among the most skepticalwho believe that vaccines cause autism and have few demonstrable benefits -24% would "definitely" like J o u r n a l P r e -p r o o f to be vaccinated and another 31% would probably do so. Only 29% thought they would probably or definitely not do so. A similar pattern emerged for the question about making COVID-19 vaccinations mandatory, with 36% of the most skeptical respondents favoring a legal obligation to be vaccinated (see supplementary Figure S2 ).

We observed that lower availability of public health resources was associated with lower willingness to become vaccinated during the recent COVID-19 crisis (Figure 2 ). In particular, in the April 2020 cross-section, lower availability of ICU units in an area and a higher occupancy ratio of these ICUs were both associated with more limited support for COVID-19 vaccination, while late-February ICU capacity use was strongly correlated with COVID-19 mortality rates at the peak of the crisis ( Figure S4 and Table S4 with COVID-19, and one using a full set of covariates including age, socio-economic status, education, gender, marital status, and regional fixed effects. In both cases, the effect of ICU occupancy was highly significant. As the disaggregation by subgroup showed, the biggest effect was visible among vaccination hesitantsthose with a somewhat ambivalent attitude towards vaccines. There was also a significant but much smaller effect for the pro vac group.

The no vac group on average appeared to react to ICU occupancy, but estimates were not significantly different from zero.

A similar pattern was apparent for another indicator of public health system capacity -ICU beds per 1,000 inhabitants ( Figure 2 , Panels C and D). Where the NHS had ample hospital beds available before the COVID-19 crisis hit, vaccination unwillingness in general was much lower. The lower the number of ICU units in an area, the more people indicated that they would be unwilling to be vaccinated. The effect was significant at the 90% level without covariates and 95% level with covariates. The average coefficient was largest for vaccination sceptics, but with large standard errors. Hesitants appeared less influenced by ICU provision.

Neither ICU provision per capita nor occupancy rates in October 2019 predicted attitudes towards COVID-19 vaccination, nor did they have predictive power for trust in medical experts and scientists during the COVID-19 crisis (see Table S3 )it was only health care availability during the coronavirus outbreak (i.e. ICU availability in February 2020) that mattered for attitudes.

ICU availability immediately prior to the COVID outbreak was also a significant predictor of trust in science and the medical profession: where the NHS ran out of ICU capacity during J o u r n a l P r e -p r o o f the COVID-19 crisis, trust in science and medical experts was markedly lower (see Table 2 with OLS estimates and Table S5 with Probit estimates). In areas where 90% or more of ICU beds were occupied by late February 2020, only 35% of respondents had a lot of confidence, compared with 42% in the sample overall. The same pattern was truein reversefor ICU provision per capita, with greater numbers of ICU units in the nearest hospital predicting more trust in health experts and scientists.

Personal risk was also an important determinant of attitudes: survey respondents who thought that COVID-19 posed a clear risk to themselves or family members were much more likely to be willing to be vaccinated (see Figure 2 , Panels E and F). The share of respondents unwilling to vaccinate in the at-risk group was a mere 4%, compared with 10% in the no-risk group. The difference was largest among vaccination sceptics. Hesitants and pro vac respondents also react to personal risk, but the effects were smaller; for the hesitants, the effect became insignificant when adjusting for the full set of covariates. 

Can vaccination stop COVID-19? Given the new virus' infectiousness, for vaccination to succeed, take-up rates will have to be high. High support cannot be taken for granted, given the general rise of vaccine hesitancy in recent years.

Our nationally representative survey of UK respondents showed substantial support for vaccination across all socio-economic groups in the UK, and even among vaccination sceptics. Some of this reflects perceived personal risk: Having a close family member at risk sharply increased support for vaccination ( Figure 2 ). Even respondents who believed that vaccines can cause autism, have other severe side-effects, generate few benefits, and are mainly prescribed because of financial interests of the pharmaceutical industry, were overwhelmingly willing to become vaccinated against COVID-19 in April 2020 (Figure 1 ). This implies that vaccine hesitancy is unlikely to impede herd immunity against COVID-19 through vaccination.

Areas with more limited ICU capacity experienced sharply higher case fatality rates. ICU availability was a major concern among the public during the pandemic's peak, in March and April 2020. We found that in areas where ICU availability was limited for exogenous reasons, vaccination support is markedly lowerand so is trust in medical experts and scientists.

In general, vaccination can substitute for health interventions post-infectionhigher immunization rates reduce the need for post-infection treatment. Hence, vaccination willingness should in principle be inversely related to the availability of health care support.

In contrast, we find lower availability of public health resources to be associated with sharply lower willingness to become vaccinated during the recent COVID-19 crisis. As discussed above, in the April 2020 cross-section, lower availability of ICU units in an area and a higher occupancy ratio of these ICUs were both associated with more limited support for COVID-19 vaccination.

We used the occupancy rate in the nearest NHS hospital immediately before the crisis (late February 2020) as an explanatory variable (24) , for three reasons. First, it is an important predictor of mortality rates: As discussed above, late-February ICU capacity use was strongly correlated with COVID-19 mortality rates at the peak of the crisis. Second, late-February ICU occupancy was excludable from the severity of the COVID-19 shock, almost no hospitalizations had occurred by then. Late February occupancy is therefore a plausible shifter of hospital capacity, reflecting the seriousness of other demands on the local health system. Third, ICU capacity was a major concern among the public during the COVID-19 crisis, as evidenced by Google search term frequency: Figure S1 in the appendix shows how search frequency on Google peaked in late March/early Aprilthe time of our follow-up survey.

Our results underline the importance of trust and confidence in mainstream medicine as precondition for positive attitudes towards vaccination, as previously reported (12) . It also highlights the risk of a vicious circle, where negative attitudes towards vaccination lead to lower vaccination rates, greater numbers of severely ill patients in times of crisis, more pressure on ICU units, and in turn, less trust in the medical profession and science in general. This finding in particular highlights an important synergy between public health system performance and the public's support for public health measures like large-scale vaccination.

In conclusion, our results suggest that the availability of sufficient ICU places can contribute to building trust in medical experts and boosts the willingness to become vaccinated against COVID-19. Where Trust is an indicator variable = 1 if respondent reports "a great deal of trust" in health experts and scientists, OR is the occupancy rate of ICU beds in the NHS trust that is closest to the zip code where the respondent lives, ICU is the number of ICU beds per 1000 people in the local authority where the respondent lives, COVID-19 Risk is an indicator variable = 1 if the respondent states that COVID-19 poses a major risk to either himself or someone living in his household, COVID-19 Exposure is an indicator variable = 1 if respondent knows someone infected with COVID-19 and COVID-19 Deaths is the number of COVID-19 deaths per 1000 people in the local authority as of 10 April 2020. Col. 1 includes only these covariates. Col. 2 includes the answers to 3 questions on vaccination attitudes asked in October 2019: "should unvaccinated kids be allowed to attend school?" "should parents who choose not to vaccinate their kids be banned from childcare benefits?" and "should parents who choose not to vaccinate their kids be fined?". For each of these questions, we create an indicator variable = 1 if the respondent stated that he would punish parents who choose not to vaccinate their kids, showing support for measures promoting vaccination. Col. 3 adds a gender indicator variable, 3 age groups dummies (18- Notes: The figure shows responses to the question: "If a vaccine against COVID-19 became available for everyone tomorrow, do you think you would or would not get vaccinated?" The bar on the left reports the breakdown for all respondents of the April 2020 survey (N=1194). The other 3 columns report the breakdown for three categories of respondents: "no vax" (N=148), "hesitants" (N=431) and "pro vac" (N=615). We assign respondents to one of these categories using ther answers to the question on general vaccination attitudes. See Section S.2 in the Supplementary Materials for details on the construction of these categories. Figure 2 . ICU availability, perceived risk and unwillingness to get vaccinated against COVID-19

Vaccine hesitancy: Definition, scope and determinants

Wellcome Global Monitor, How does the world feel about science and health

Drop in vaccination rates in England alarming experts. The Guardian

Vaccines and immunization. Data and statistics

Rabin, others, The public health crisis of underimmunisation: a global plan of action

NIH chief worried vaccine "skepticism" might cause some to skip coronavirus vaccine

Integrating epidemiology, psychology, and economics to achieve HPV vaccination targets

Beyond confidence: Development of a measure assessing the 5C psychological antecedents of vaccination

Current tools available for investigating vaccine hesitancy: a scoping review protocol

The public and the smallpox threat

Larson, others, The benefit of the doubt or doubts over benefits? A systematic literature review of perceived risks of vaccines in European populations

Measuring vaccine confidence: introducing a global vaccine confidence index

A systematic review of interventions for reducing parental vaccine refusal and vaccine hesitancy

The vaccine hesitancy scale: Psychometric properties and validation

Resurgence of Measles in Europe: A Systematic Review on Parental Attitudes and Beliefs of Measles Vaccine

Understanding vaccine hesitancy around vaccines and vaccination from a global perspective: a systematic review of published literature

Assessing vaccine hesitancy in the UK population using a generalized vaccine hesitancy survey instrument

Effective messages in vaccine promotion: a randomized trial

Effects of a web based decision aid on parental attitudes to MMR vaccination: a before and after study

The effect of trust and proximity on vaccine propensity

The case for reopening economies by sectors

The relationship between parent attitudes about childhood vaccines survey scores and future child immunization status: a validation study

From the full sample of respondents living in England we create 20 bins of roughly equal sample size; the last 2 bins have no variation in occupancy rate (100%) and are combined into a single data point. Panel C: unconditional binscatter of February 2020 ICU beds per 1000 people (x-axis) and resistance to COVID-19 vaccine (y-axis)in a local authority without a NHS Trust: these bins are combined into a single data point. Panel E: share of respondents showing resistance to COVID-19 vaccine among those who state that COVID-19 does not poses a major risk to anyone in the household (left bar) and those who state that it does (right bar). The whiskers show the standard errors of the estimates. Panel B, D and F: OLS estimates and 95% confidence intervals from: he would "definitely" or "probably" not vaccinate against COVID-19, and the other variables are defined in the footnote of Table 2. Panel B: estimates of β 1 . Panel D: estimates of β 2 . Panel F: estimates of β 3 . The specification with baseline covariates includes an indicator for whether the respondent knows someone infected with COVID-19. The specification with all covariates includes all explanatory variables in col. 4 of Table 2

Contributorship Statement: All four authors have contributed equally to all aspects of the article (planning, conducting, and reporting the work) and are joint first authors (and listed in alphabetical order). They all affirm that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted;and that any discrepancies from the study plan have been explained.Conflicts of Interest: The authors have no conflicts of interest or funding to disclose.