Social capital and COVID-19 vaccination in prisons: a survey of attitudes, beliefs, and motivations

Abstract

Background

UK prisons were profoundly affected by the COVID-19 pandemic. Despite this, uptake of the COVID-19 vaccination remained low in these settings. Social roles, relationships, and freedoms influence vaccine uptake. Prisons are isolated by design and may foster mistrust that negatively influences vaccine uptake. Prisons are also unique communities with their own social networks and relationships.

Methods

We undertook a questionnaire survey across all six prisons in Wales, UK, to gather data on vaccine behaviour, attitudes, and other influencing factors. We fitted binomial generalised linear mixed effects models to identify predictors of vaccination.

Results

Surveys were completed by 727 prison residents. We found low vaccination uptake in younger cohorts, those serving short sentences, and those who perceived themselves to be in poorer health. Those reporting low levels of trust towards others and those reporting fewer sources of social support in prison were less likely to be vaccinated.

Conclusion

Our findings suggest that building a prosocial atmosphere in prison and strengthening relationships and trust between prison staff and residents would positively influence vaccine uptake. Specific message framing should be considered to address the beliefs and motivations most prevalent in this population group, rather than focussing simply on enhancing the opportunity for vaccination.

Introduction

UK prisons were profoundly affected by the COVID-19 pandemic, with over 30 000 confirmed cases and 159 deaths from suspected or confirmed COVID-19.¹ Rates of COVID-19 in UK prisons appeared higher than in the general community, with the first wave of the pandemic seeing 7.6 confirmed cases per 1000 people in prison compared to 4.9 outside.² Prisons hold populations with high health needs in crowded conditions, posing challenges for infection control.^3,4 Interventions have been proposed for outbreak management in prisons, including improved hygiene, cleaning, mask-wearing, ventilation, decarceration, and cohorting,^5–7 but vaccination of prison residents remains a public health priority to reduce morbidity and mortality and limit the spread of infection.⁸ Vaccine uptake in prison residents appears lower than in community counterparts globally.^9,10 There is limited evidence of approaches to promote vaccination in prisons.¹¹

COVID-19 vaccination uptake has been linked to social capital^12,13—a multifaceted concept, comprising trust, social connectedness, and norms of reciprocity, all measurable components.¹⁴ Given the breadth of the construct, and how distinct facets can differentially influence health,¹⁵ it is important to be clear about which aspects of social capital are being measured in a given setting.¹⁶ We view prison as a distinct social network with relationships that manifest differently to the wider community. Social organisation in prisons can be influenced by gender, ethnicity, age, criminal experience, geography, gang affiliations, and relationships with staff. Social hierarchies may have violent offenders at the top and those who are considered weak or ostracised at the bottom.^17–19 Prisons isolate by design, challenging bridging (network building) social capital. Prisons can foster deep mistrust,^17,20 threatening bonding social capital (trust and reciprocity) that may undermine health decisions. Whilst social capital in prisons is poorly understood, evidence suggests that it can increase participation in health interventions.^21,22 Social capital, through the role of peer support, has been shown to positively influence hepatitis C treatment adherence in prison.²³ Those in prison who are socially bonded by race, ethnicity, religion, or exercise have been shown to have better health outcomes.²² Research is needed to understand the role of social capital in vaccination decisions. We used a survey design to measure the impact of relationships and trust within the prison on the decision to accept COVID-19 vaccination.

Methods

Ethics and permissions

The study was approved by NHS REC (22/WA/0187) and HMPPS National Research Committee (2022-196). Due to the sensitivity of the data, we regret that we do not have ethical approval to share data.

Setting and fieldwork

There are 5300 residents in six male prisons in Wales.¹³ There are no female prisons in Wales. Prisons in England and Wales are governed by the UK Ministry of Justice, with primary healthcare services provided by the National Health Service and social care provided or commissioned by local authorities.²⁴ Prison residents are entitled to the same vaccines as those in the community.

Between October and December 2022, ~1500 paper questionnaires (enough for 30% of the prison population at each site) were distributed in all six prisons in Wales. The sample size was based on the experience of previous surveys run in these prisons. Participation was voluntary, and responses remained confidential. Participants needed to be 18 years old or over and able to communicate in Welsh or English. Questionnaire distribution was led by nominated custodial staff and prisoner peer representatives who underwent training from the research team. Leaflets describing the survey were distributed across all prison wings by the nominated staff. Peer representatives were available to answer questions about the research and support participation. Participants completed consent forms alongside the questionnaire. Responses were sealed and sent to a research company, Beaufort Research, for collation. Prisons were blinded, identified by a code known only to the chief investigator.

Measures

The questionnaire was based on a community survey on vaccination attitudes, behaviour, and social capital in Wales.²⁵ Content was adapted to ensure appropriateness for a prison population. The full questionnaire is available at https://osf.io/cr4bz/files/osfstorage.

Variables were grouped into three categories: sociodemographic characteristics, vaccination attitudes and behaviour, and social capital and support.

Sociodemographic variables included age, ethnicity, and education. Health-related characteristics included self-reported general health, whether the respondent had been vaccinated against influenza, and whether the respondent had been vaccinated against hepatitis-B.

Sentence-related characteristics were included as a sociodemographic characteristic and a measure of social anchorage in terms of social network analysis,²⁶ and included whether the respondent had lived out of prison since March 2020 (pre-pandemic) and how long they had been in prison during this stay.

To measure vaccination attitudes and behaviour, respondents were asked whether they had been vaccinated against COVID-19. Respondents who reported vaccination were asked why they received the vaccine, selecting all reasons that applied and their main reason. Respondents who reported not being vaccinated were asked why not, again selecting all reasons that applied and a main reason (Table 1).

For the social capital and social support variables, we used four groups of items. First, we measured civic and institutional trust (cognitive social capital)²⁷ using a series of items with the stem: ‘How much do you trust these groups?’, where respondents rated their level on a 0–10 scale, with 0 labelled as ‘Not at all’ and 10 as ‘Completely’. The groups were as follows: ‘Most people in general’, ‘People you know personally in prison’, ‘People you know personally outside of prison’, ‘UK Government’, ‘Welsh Government’, ‘Your doctor or nurse in prison’, Your doctor or nurse outside of prison’, ‘Prison staff’, ‘The media (e.g. TV, newspapers, radio stations)’, ‘Scientists’, ‘Social media’, and ‘Pharmaceutical companies’. Second, we measured relationships with staff (vertical social capital¹⁶) and the extent of institutional engagement by asking ‘Who might you go to for support whilst in prison?’, with the response options: ‘other prisoners’, ‘prison officers’, ‘teachers’, ‘psychology’, ‘chaplaincy’, ‘keyworker/offender manager’, ‘substance misuse/drug worker’, ‘healthcare’, ‘listeners’, ‘telephone helplines e.g. Samaritans’, ‘friends or family’, and ‘prefer not to say’; with respondents asked to tick all that applied. Third, we considered bonding social capital (relationships with friends, family, and fellow prisoners) by asking ‘How often do you phone family or friends who are outside prison?’ We also asked ‘How often do you have social visits from friends or family?’

Analyses

We computed descriptive statistics of the above variables. We fitted binomial generalised linear mixed effects models,²⁸ to test individually whether each variable predicted vaccination. Odds ratios with 95% confidence intervals were used to assess the strength of relationships and whether they exceeded what would be expected by chance. Models contained random intercepts for the prison and wing, to account for nesting of data, and fixed effects for each predictor, with categorical variables dummy-coded as appropriate. A separate model was fitted for each predictor to avoid issues with ‘Table 1 fallacy’.²⁹

Unplanned multivariate versions of the models for social capital variables, adjusting for age and ethnicity, were fitted to assess whether these associations were driven by confounding.

Results

Sample

Surveys were completed by 727 respondents in six prisons. Of these, 496 (68%) reported receiving at least one vaccine, 171 reported being unvaccinated (24%), 34 preferred not to respond, and 26 did not complete the item. Table 2 describes the sociodemographic characteristics of the sample.

Characteristics

Vaccination uptake increased with age and was higher in White respondents, although confidence intervals overlapped with one for all groups except the Asian, other, and not stated groups. The level of education did not affect uptake.

Respondents who had been in prison for shorter periods had lower vaccination rates than those who had been in for 1–2 years, with a trend in the same direction for those who had been in for over 2 years. Those who had spent time outside of prison since the start of the pandemic had similar vaccination uptake to those who had not.

The highest vaccination rates were those reporting ‘good’ general health, with lower uptake above and below this.

Those reporting a confirmed previous infection with COVID-19 had higher vaccine uptake than those reporting no previous infection. Reporting being vaccinated against hepatitis-B and influenza was associated with being vaccinated against COVID-19.

Vaccine attitudes and behaviours

Of those vaccinated, ‘to protect myself’ was the most common reason to accept vaccination (Table 1), but respondents reported other reasons related to protecting others, protecting the NHS, and returning to normal.

Many unvaccinated respondents did not identify the main reason. The reasons given were diverse, likely representing genuine heterogeneity, and the greater range of options was also given.

Respondents gave reasons that suggested COVID-sceptical attitudes, such as ‘I don’t believe in COVID-19’, ‘COVID-19 is not a serious illness’, and ‘There is a hidden agenda behind the vaccine campaign’, with almost 25% giving one of these as the main reason. Over 25% of unvaccinated respondents gave not believing in COVID-19 as a reason for not getting vaccinated.

Many reported concerns about side effects or desire for more certainty about safety: ‘The vaccine might have side effects’, ‘The vaccine might affect my fertility’, ‘I want to wait until there is more evidence about the vaccine’, and ‘I need more information’, with ~22% giving one of these as the main reason. ‘The vaccines don’t work’ and ‘I would rather develop natural immunity’ were also common reasons, suggesting vaccine-sceptical attitudes. Very few cited logistical barriers such as ‘I have been too busy’, ‘I’m not eligible for health reasons’, or ‘I don’t know how to get a vaccine’, and few would rather wait to be vaccinated outside of prison.

Social capital and support

Nearly all trust measures were positively associated with vaccination (Table 4). Confidence intervals for trust in people known personally in prison (95% CI 0.958–1.081) and in social media (95% CI: 0.987–1.150) overlapped with 1, but otherwise, higher trust was associated with being vaccinated across the board. Adjusting for age and ethnicity made little difference to associations, although confidence intervals overlapped with one for ‘People you know personally outside prison’ and ‘Your doctor or nurse outside of prison’, and became significant for ‘social media’, after adjustment.

The number of sources of support identified in response to the question ‘Who might you go to for support whilst in prison?’ predicted vaccine uptake, as did specific sources of support. Adjusting for age and ethnicity attenuated the association for the number of sources of support. Adjustment had a varied effect on the specific support sources. Most changed little, but ‘prison officers’ became non-significant, and ‘Telephone helplines e.g. Samaritans’ became significant after adjustment.

Contact with friends and family from outside the prison was not associated with vaccination (Table 3).

Discussion

Main finding of this study

We surveyed over 15% of the Welsh prison population on COVID-19 vaccination attitudes and behaviour. Respondents were male, younger, more ethnically diverse, held fewer educational qualifications, and reported poorer health than the Welsh population.³⁰

We found predictors of vaccine uptake across several domains, including demography (increasing age and White ethnicity), health (self-perceived good health, previous COVID-19 infection, and previous hepatitis-B or influenza vaccination), and social capital (increasing with sources of support within prison and increasing levels of trust in general).

What is already known on this topic

Rates of having received at least one COVID-19 vaccination were lower (68.2%) than the figure for the Welsh population aged 5+ in July 2022 (86%),³¹ confirming previous evidence that prison populations have lower rates of vaccine uptake than the general community.^9,10

The most common reason for accepting COVID-19 vaccination was to ‘protect myself’. This echoes findings in the United States³² and Spain.³³ Other reasons included protecting others and helping life return to normal, although very few gave these as their main reason.

High levels of disbelief in the COVID-19 vaccine have been previously reported amongst prison populations, and our findings support this. In Pakistan, 60% of prisoners did not believe COVID-19 was a real problem and 50% believed the pandemic was a conspiracy.³⁴ Over a quarter of our respondents stated they did not believe in COVID-19 with 12% giving this as their main reason against vaccination. Almost a quarter gave their main reason for declining vaccination as either not believing in COVID-19 or not believing in vaccines more generally. Twenty percent of respondents felt there was a hidden agenda behind the vaccine campaign. These results suggest the need to address disbelief in COVID-19 before persuading of vaccination benefits.

Concerns about vaccine side effects have been shown elsewhere to be a common reason for declining vaccination, so it is unsurprising that our results supported this. International literature describes high levels of concern about side effects from prisons in the United States, Italy, Canada, and Pakistan.^32,34–36 Just under 40% of our respondents were concerned about side effects, 18% believed the vaccine might affect fertility, and just under 30% wanted to wait until there was more evidence about vaccines. The broad distrust seen in survey responses and wider literature suggests a need to tailor communications to those who are imprisoned, build trust, and involve service users in the design and delivery of health promotional material and messaging. Encouragingly, there is evidence to suggest that this method of increasing education on vaccination improves uptake in prisons.¹¹

Reporting being vaccinated against hepatitis-B and influenza was both associated with vaccination against COVID-19, particularly influenza. This is similar to findings in US prison populations where COVID-19 vaccine refusal was strongly associated with not receiving influenza vaccines,³⁷ and in Italy where COVID-19 vaccine uptake was higher in those reporting influenza vaccination.³⁸ This could suggest that broader attitudes to vaccination help shape attitudes to COVID-19 vaccination but may also be driven by individual health risk status.

What this study adds

This study extends the evidence on social capital as an important determinant of vaccination to the prison setting, which is an interesting and important context in terms of vaccination policy and from a social capital perspective. Prison residents reporting low trust in people and institutions, in general, were less likely to be vaccinated. This supports findings from previous studies that trust is at the core of social capital and should be promoted to support engagement with health in prison.²¹ Social capital can be a double-edged sword when it comes to health, as risky health behaviours can travel along the same social networks as healthy ones do.³⁹ We did not observe evidence of this here, but this may depend on how social capital is measured, as different aspects of social capital have been shown to have different associations with COVID-19 outcomes.^15,40

Likewise, respondents reporting fewer sources of social support in prison were less likely to be vaccinated. This echoes findings in community populations^41,42 that social isolation is associated with vaccine hesitance, highlighting the importance of ensuring that prison residents are not socially isolated. Reporting staff support through chaplaincy, teachers, prison officers, key workers, substance misuse workers, and listeners suggested increased COVID-19 vaccine uptake. These findings suggest that investing in prison staff support structures may reap secondary benefits for health. Engaging staff to support access to vaccines may also help improve uptake. Building social capital through improved social support structures and trusted relationships will improve the health and wellbeing of people in prison. This can be achieved through evidenced methods such as peer support,²¹ or prison social groups to support health interventions.²²

Literature⁴³ suggests that those in prison may draw on sources of social capital from outside the prison, as well as inside it. Interestingly, the frequency of phone calls and visits from friends and family was not associated with vaccination, nor was social support from friends and family, suggesting that social support from outside prison is not a substitute for sources of support inside.

International literature suggests that some prison cohorts may be reluctant to receive vaccination whilst in custody, having more trust in community healthcare.¹⁰ Encouragingly, we did not find this, with only 5% of respondents indicating they would rather receive a vaccine outside of prison. Only 1% of respondents stated they did not know how to get the vaccine whilst in prison, suggesting that logistical barriers are not responsible for the vaccination gap within the community population.

Limitations of this study

The survey was voluntary, and therefore respondents may not be representative of the prison population. More mistrustful residents may have been less likely to respond and less likely to be vaccinated. Survey research in prisons is logistically challenging and compromises were made in questionnaire design and fieldwork—some sensitive topics, such as mental health, were avoided as we could not guarantee confidentiality for respondents. All sites were male prisons; perspectives may differ in female prisons. Results reflect the post-pandemic period in late 2022. Since then, COVID-19 incidence has decreased and COVID-19 booster vaccination in the UK has been limited to those over 65 years of age or at increased clinical risk.⁴⁴ Despite this, prison populations will always be at heightened risk for the spread of respiratory infections, so COVID-19 vaccination will be an important driver of prison health for the foreseeable future. Finally, we measured social capital by asking about sources of social support and trust in various groups. However, prisons represent complex social settings and other forms of social capital that we did not focus on; e.g. membership in organised gangs and relationships with cellmates’ may be important. Future work should develop new measures of prison social capital so that we can better understand its relationship with health. Furthermore, we focused on individual measures of social capital, rather than measuring the social environment on, e.g. specific blocks of prisons. Further work could attempt to assess group-level social capital.

Conclusion

The study supports that social capital is a valuable public health resource. The association between social support and vaccine uptake suggests that policies to improve social support structures for prisoners may pay dividends for health. For infectious diseases, such as COVID-19, these benefits may extend beyond the prisoner receiving interventions if this interrupts transmission chains. Second, the findings may support targeting specific groups within the prison population to improve uptake, namely those who are younger, serving shorter sentences, who consider themselves to be in poorer health, who feel less socially supported, and who feel more distrustful of others. Third, some reasons given for non-vaccination appear amenable to persuasion, so the vaccination gap between the prison and community populations may be able to be narrowed with the right vaccination campaign. Work on specific message framings for prisoners is needed including addressing beliefs about COVID-19 as well as the vaccine. Fourth, vaccination is not enough to prevent prison outbreaks and must be considered alongside other measures to strengthen infection prevention and control.

Acknowledgements

We are grateful to HMPPS staff P.N. and G.M. and staff at each prison for operational support namely A.E., B.H., L.D., K.E., J.C., A.G., and L.P. We would like to express our thanks to the Vaccine Preventable Disease Programme at Public Health Wales, and to three anonymous peer reviewers, for useful comments on the manuscript.

All authors attest they meet the ICMJE criteria for authorship.

Conflict of interest

None declared.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Stephanie E. Perrett, Consultant Nurse

Daniel Rhys Thomas, Consultant Epidemiologist

Christopher W. N. Saville, Clinical Senior Lecturer

References