How trust, digital skills, and social factors relate to COVID-19 vaccine hesitancy in Czech adults

An analytical cross-sectional study was conducted using data from the 48th wave of the Czech national panel survey Život během pandemie [Life During Pandemic], collected between 20 and 26 March 2023.³⁹ The panel survey was initiated in March 2020 to monitor the social, economic, and health-related impacts of the COVID-19 pandemic on the Czech population on a regular basis, either monthly or biweekly.³⁹ Participants were recruited through Český národní panel [Czech National Panel] using quota sampling to ensure national representativeness by age, gender, education, and region.⁴⁰ For this study, a single wave (the 48th wave) was analyzed to examine determinants of COVID-19 vaccine uptake and intent.

Data for this study were collected online due to the pandemic's restrictions on physical interactions and the need for rapid acquisition of time-sensitive data. This approach also mitigated potential response biases, as respondents were able to answer sensitive questions about their health behaviors in a private, self-administered format. The questionnaires were developed and validated by researchers from the PAQ Research, z. ú. (Prague, Czech Republic) and the Institute for Democracy and Economic Analysis (IDEA)'s AntiCovid initiative.³⁹

Participants were recruited through the Czech National Panel using stratified quota sampling to obtain a representative sample of adults (≥18 years old) based on region, gender, age, education level, and pre-pandemic employment status. To enhance analytical robustness, the sampling strategy included intentional oversampling of individuals residing in municipalities with 50,000 or more inhabitants. Due to the online mode of data collection, participation was limited to individuals with internet access; thus, the representation of older adults (≥55 years) should be interpreted with caution, given potential limitations in digital accessibility within this subgroup.³⁹

Initially, there were 3,101 target participants in the first wave (March 2020). The response rate in the 48th wave was 55.1/, yielding a final sample of 1,708 respondents, which remained satisfactory given that the minimum target sample size per wave was approximately 1,700. This threshold ensured sufficient statistical precision, with a margin of error of approximately ±2 percentage points.

Besides sociodemographic variables – including sex, age, cohabitation status, household type and size, geographic region, residence size, and education level – this study also incorporated socioeconomic variables, such as employment status, work schedule, contract type, job stability concern (10-point Likert scale), financial satisfaction (10-point Likert scale), and monthly household income. Additionally, anamnestic variables were assessed, including happiness (10-point Likert scale), depression (PHQ-8; discrete score: 8–32), loneliness (discrete score: 3–9), COVID-19 concern (10-point Likert scale), self-perceived health (5-point Likert scale), chronic illness, activity limitations, healthcare registration (general practitioner, dentist, gynecologist), unmet healthcare needs, and recent use of healthcare technologies.^41,42

Political attitudes were assessed using Likert scale items, including trust in constitutional institutions (4-point scale), political views on governance and democratic principles (5-point scale), perceptions of immigrants' positive role (10-point scale), views on Ukrainian refugees' integration into Czech society (5-point scale), and frequency of news consumption (6-point scale).

The 5C model of vaccination psychological antecedents by Betsch et al. was used to assess the influence of these factors on actual COVID-19 vaccination behaviors.²⁷ This model comprises confidence, complacency, constraints, calculation, and collective responsibility. Each construct is assessed using three 5-point Likert scale items, resulting in individual scores ranging from 3 to 15.^27,28

The digital vaccine literacy (DVL) scale of Montagni et al. was used because it is a streamlined instrument of 7 items that measure individuals' ability to understand, trust, and appraise online vaccine-related information.¹⁷ The DVL scale comprises three subdimensions: (1) understanding and trusting vaccine information from official websites, (2) understanding and trusting vaccine-related content from social media, and (3) evaluating and applying online vaccine information in decision-making.

Internal consistency of the multi-item scales was assessed using Cronbach's alpha. Reliability coefficients were high for confidence (α = 0.936), calculation (α = 0.861), political trust (α = 0.842), complacency (α = 0.829), and perceived integration of Ukrainian refugees (α = 0.884). Acceptable reliability was observed for constraints (α = 0.775), DVL – social media (α = 0.756), and DVL – government sources (α = 0.782), while lower internal consistency was noted for collective responsibility (α = 0.490) and DVL – information appraisal (α = 0.483).

The study adhered to the Declaration of Helsinki guidelines for research involving human subjects and complied with the General Data Protection Regulation (GDPR) of the European Union.^43,44 In accordance with Czech legislation and Masaryk University policy, institutional ethics committee approval was not required for this anonymous, survey-based study involving adult participants and no biomedical interventions.⁴⁵ All participants provided their informed consent before completing the survey and had the option to withdraw at any time without negative consequences. The study data were securely stored and processed in compliance with GDPR guidelines, ensuring the anonymity of all participants for researchers who interpreted the data.

Statistical analyses were conducted using SPSS 29 and Jamovi (R-based software). Descriptive statistics summarized sociodemographic, socioeconomic, and anamnestic characteristics. Likert-scale items were analyzed as continuous variables, with mean scores compared across vaccination groups. Bivariate analyses, including the Chi-square test (χ²) for categorical variables and the Mann-Whitney (U) and Kruskal-Wallis (H) tests for numerical variables, examined associations between predictors and COVID-19 vaccination status and intent. Multivariable linear regression assessed determinants of vaccination psychological antecedents and actual vaccination status and intent, controlling for potential confounders. Model fit was evaluated using R² values, and significance was set at p<.05.

Being unvaccinated was significantly more common among females (25.23/) compared to males (16.75/) (p < .001). It was also more prevalent among younger age groups, with 30.96/ of participants aged 18–44 years being unvaccinated, compared to 19.31/ of those aged 45–69 years and 13.85/ of those aged 70 years or older (p < .001). Non-cohabitating respondents were more likely to be unvaccinated than their cohabitating counterparts (24.68/ vs. 20.06/; p < .001). Additionally, the age of partners was significantly higher among participants who had completed the booster series (p < .001). Households with minors were more likely to have unvaccinated members (26.10/) compared to those without minors (17.36/) (p < .001), and the number of minors was higher in these households (1.46 vs. 1.26; p < .001). University degree holders were the most likely to have completed the booster series (67.58/), compared to secondary school graduates both with diplomas (59.12/) and without diplomas (52.77/), and those with elementary education (34.18/). (Table 1).

Vaccine resistance was significantly more common among females (24.65/ vs. 16.27/; p < .001), non-cohabitating respondents (24.68/ vs. 19.68/; p < .001), and households with minors (25.61/ vs. 17.1/; p < .001) compared to their respective counterparts. Younger age groups, including those aged 18–44 years and 45–69 years, exhibited significantly higher rates of vaccine resistance (30.05/ and 18.74/, respectively) than the senior group (aged 70 years and older) (13.85/) (p < .001). Prague (11.9/) and cities with populations over 100,000 (16.69/) demonstrated the lowest rates of vaccine resistance. University graduates had the lowest vaccine resistance (12.77/) compared to secondary school graduates with diplomas (19.71/), secondary school graduates without diplomas (26.68/), and those with only elementary education (36.71/). (Table 1).

Booster series completion was highest among the retired group (70.94/). No statistically significant differences were observed between contract types (p = .161) or employment schedules (p = .789). However, job stability concerns were significantly (p = .012) higher among unvaccinated individuals (4.34 ± 2.83) compared to those who had completed the booster series (4.08 ± 2.67). Conversely, financial satisfaction was significantly (p < .001) lower among the unvaccinated (6.44 ± 2.43) compared to those who had received the booster (6.83 ± 2.28). Net monthly income was also significantly (p < .001) associated with vaccination status, with the highest rate of being unvaccinated observed among those below the poverty line (34.65/), followed by those in the low-income (19.83/), above-average income (17.93/), and high-income (11.61/) categories. (Table 2).

Vaccine resistance was lowest among the retired group (17.14/). No statistically significant differences were observed between contract types (p = .191), employment schedules (p = .615), or job stability concern levels (p = .974). Financial satisfaction was significantly (p < .001) lower among the unvaccinated (6.45 ± 2.44) compared to those who had received the booster (6.89 ± 2.32). The highest rate of vaccine resistance was observed among those below the poverty line (32.67/), followed by those in the low-income (19.56/), above-average income (17.30/), and high-income (11.61/) categories. (Table 2).

Registration with general practitioners and gynecologists had no significant association with COVID-19 vaccine status or intent. On the other hand, registration with dentists was significantly more common among booster series completers (p = .002) and vaccine-confident individuals (p = .004). Generally, the barriers to healthcare access had no significant association with COVID-19 vaccine status or intent. Contrarily, the utilization of some healthcare technologies, i.e., wearables, e-prescription, and insurance applications was significantly more common among booster series completers and vaccine-confident individuals. (Table 3).

Booster series completers and vaccine-confident individuals expressed significantly stronger agreement with the fairness of election campaigns (p = .003, p = .001). They also showed greater disagreement with the notion that "Freedom of speech is restricted in the Czech Republic" (p < .001), reflecting a more positive perception of speech freedoms. Additionally, they were less likely to endorse the acceptability of using violence against politicians they disagree with or harsh personal attacks during election campaigns (p < .001 for both). (Table 4).

Positive attitudes toward immigrants were more prevalent among booster series completers and vaccine-confident individuals who perceived immigrants as enriching Czech culture (mean scores: 5.31 and 5.76, respectively, p < .001). Regarding Ukrainian refugees' integration into Czech society, these groups also expressed higher confidence in their successful integration across key domains, including work, housing, schools, language, and culture (p < .001 for all). These findings suggest a link between vaccination confidence and broader social inclusivity. (Table 4).

Complacency showed a reverse trend, as booster series completers and vaccine-confident individuals exhibited significantly lower scores (6.63 and 6.14) compared with unvaccinated and vaccine-resistant individuals (8.95 and 8.94, p < .001). Similarly, constraints such as personal stresses, unpleasantness of the vaccination process, and discomfort in visiting a doctor were more prominent among unvaccinated and vaccine-resistant individuals (7.37 and 7.34) compared with booster series completers and vaccine-confident individuals (5.91 and 5.47, p < .001). Moreover, calculation scores were significantly lower among booster series completers and vaccine-confident individuals (10.65 and 10.21) compared with unvaccinated and vaccine-resistant individuals (12.01 and 12.04, p < .001). Table 5

Female sex was associated with reduced confidence and collective responsibility (β = −0.73 and −0.27) and greater perceived constraints and calculation tendencies (β = 0.30 and 0.33). Lower education levels were consistently linked to lower confidence, higher complacency, and greater constraints, with individuals holding only elementary education showing the strongest associations compared to those with university education (β = −1.41, 1.74, and 1.08). Other sociodemographic variables, such as age and cohabitation, did not significantly influence psychological antecedents in the regression models. Table 6

While employment status, job stability, and financial satisfaction did not emerge as important determinants of vaccination antecedents, lower income levels were associated with reduced confidence, increased complacency and greater constraints (β = −0.65, 1.31, and 1.67 for income below the poverty line vs. high income). Table 6

Suffering from a chronic illness was associated with improved confidence, reduced complacency and fewer constraints (β = −0.65, 0.63, and 0.63 for respondents without chronic illnesses vs. those with illnesses). Most anamnestic characteristics, e.g., depression, loneliness, and perceived health, and political attitudes, e.g., trust in constitutional institutions and attitudes toward foreigners and refugees, did not have significant associations with vaccination antecedents. Table 6

COVID-19 vaccination uptake was significantly associated with male sex (β = −0.17; p < .001), higher education (β = 0.30 and 0.45; p = .006 and < .001, for secondary school vs. university graduates), higher monthly income (β = −0.27, −0.26, and −0.20; p = .043, .009, and .029, for below poverty line, low, and average vs. high income), and the presence of a chronic illness (β = −0.21; p < .001). Similarly, vaccination intent was significantly associated with male sex (β = −0.26; p < .001), higher monthly income (β = −0.29 and −0.25; p = .011 and .003, for below poverty line and low vs. high income), and the presence of a chronic illness (β = −0.91; p < .001). Other sociodemographic, socioeconomic, and anamnestic characteristics, as well as political attitudes, did not emerge as significant determinants in the regression models. Table 8

The relationship between political trust and vaccine confidence has been a key focus during recent pandemics. During the 2009–2010 H1N1 pandemic, US citizens who showed greater trust in federal and local governments were more likely to accept vaccination.²⁶ Similarly, in the first year of the COVID-19 pandemic, before vaccine authorization, those who doubted the governments' and the CDC's ability to manage the crisis were more hesitant to get vaccinated.⁴⁶ Moreover, political ideology has been found to play a significant role in shaping vaccine confidence, as individuals with conservative orientations tend to perceive greater health risks associated with vaccines, which in turn lowers their vaccination willingness and uptake.^23,24,47 In Europe, the growing support for populist parties has been strongly linked to rising vaccine hesitancy, triggered by deepening distrust in elites and experts, which amplifies skepticism about vaccine safety and effectiveness.^48,49 According to Eberl et al., populist attitudes were found to negatively correlate with trust in both political and health institutions, reinforcing conspiracy beliefs regardless of political ideology.⁵⁰ Consistent with earlier findings from US and Europe, our results reinforce the notion that trust in constitutional institutions is a critical determinant of vaccine confidence, highlighting the universal relevance of political trust in shaping public health behaviors across different political systems and confirming its continued importance in the post-pandemic Czech context.⁴⁶

Our study found that openness to immigrants and refugees was strongly linked to higher COVID-19 vaccine uptake and intent. In agreement with our results, Haro-Ramos et al. reported that immigration-related fears contributed to COVID-19 vaccine hesitancy among Latino communities in the US.⁵¹ This hesitancy is often fueled by populist rhetoric, which promotes discriminatory and xenophobic attitudes that undermine public health efforts in multiple ways.⁵² Historically speaking, infectious diseases outbreaks used to be fertile ground for xenophobic feelings to flourish because of increased fears and uncertainty, perceptions of competition and threat, and feelings of lack of control.^53,54 This phenomenon can be explained by the behavioral immune system (BIS) theory of Schaller et al., which assumes that humans have evolved psychological mechanisms to minimize disease exposure by promoting aversion toward perceived infection sources.⁵⁴ BIS not only influences individual avoidance behaviors but also shapes broader social attitudes, including heightened xenophobia and outgroup prejudice during outbreaks.⁵⁴ A South African study examining the impact of the COVID-19 pandemic on xenophobic sentiments found that COVID-19 vaccination was strongly linked to more positive attitudes toward immigrants.⁵⁵ Therefore, public health messaging during outbreaks should avoid stigmatizing language, as framing diseases as foreign threats and highlighting severe economic consequences both fuel xenophobia and reinforce negative outgroup attitudes.⁵⁶

Education plays a crucial role in shaping vaccine attitudes and behaviors. We found that higher education levels were associated with greater vaccination intent, increased vaccine uptake, stronger confidence, and reduced complacency and perceptions of constraints. Nevertheless, studies from other contexts suggest a more complex relationship; for instance, multiple studies on Chinese parents found that higher education was associated with increased COVID-19 vaccine hesitancy, as it boosted concerns about vaccine adverse events and costs.^57,58 Prior to the pandemic, several Italian studies also revealed a negative correlation between parental education level and children's vaccination rates against other diseases.^59–61 In contrast, studies in Germany, Italy, and Brazil found that parental education level positively influenced COVID-19 vaccine acceptance.^62–64

These conflicting results suggest that broader sociocultural and informational contexts shape education's influence on vaccine hesitancy, underscoring the need for tailored vaccine communication strategies for different education levels. A recent study by Hwang and Jeong found that while higher education enhances critical thinking and reduces misinformation acceptance, lower education increases susceptibility to misinformation.⁶⁵ Moreover, they observed that frequent exposure to misinformation and social media dependence increased misinformation acceptance across all education levels.⁶⁵ Therefore, public health efforts should prioritize media literacy initiatives and ensure that accurate, evidence-based vaccine information effectively reaches diverse educational groups.

A scoping review on COVID-19 vaccine hesitancy drivers in high-income countries, including the Czech Republic, found that lower education, lower income, and female sex were consistently linked to greater hesitancy.⁶⁶ While the impact of education is complex, the relationship between income and vaccine acceptance appears more linear. Yao et al. highlighted the positive influence of annual household income on parental vaccine acceptance in China, and studies from the UK similarly found that lower household income correlated with negative attitudes toward COVID-19 vaccination.^67,68

Another key socioeconomic factor is financial satisfaction, which has been shown to shape beliefs about the pandemic^69,70 and was positively associated with COVID-19 vaccination in Egypt, Poland, and the US.^71–73 In agreement with Kjos et al.'s study on New York City residents, our findings indicate that both objective socioeconomic status (monthly household income) and perceived socioeconomic status (financial satisfaction and job stability) significantly influenced COVID-19 vaccine hesitancy.⁷³

Digital literacy was demonstrated to be a protective factor against vaccine hesitancy, while cyberchondria (excessive online searching for health-related information leading to increased anxiety and misinformation) was associated with increased hesitancy.^18,74 Higher digital literacy levels, denoted by digital information evaluation skills, were strongly correlated with digital health comprehension and decision-making skills, both of which significantly predicted vaccination uptake among university students.⁷⁵ A multi-national study among Malaysian and Filipino adults found that social media use and online information-seeking behaviors were associated with increased odds of COVID-19 vaccine hesitancy.⁷⁶ Among young adults, higher eHealth literacy was linked to increased COVID-19 vaccine uptake, whereas vaccine literacy showed no significant association with vaccination rates.⁷⁷ This highlights the importance of strengthening digital literacy to support informed vaccine decision-making while addressing the risks associated with misinformation and excessive online health searches.

In our study, the reliance on governmental digital sources was significantly associated with COVID-19 vaccine uptake and intent. In contrast, social media use showed no significant association with vaccination behaviors. Governmental websites primarily function as encyclopedic sources of evidence-based vaccine information, linking only to official or pro-vaccine sources with limited interactivity.⁷⁸ In contrast, vaccine-skeptical websites leverage Web 2.0 features to foster online communities, expand their digital reach through diverse hyperlinks, and challenge official vaccine narratives by presenting personal experiences and alternative interpretations of scientific data.^78,79 Studies in Hong Kong, Nigeria, Pakistan, and the UK agreed with our findings on the positive impact of governmental websites as an information source on COVID-19 vaccine confidence.^80–83

Firstly, the use of a quota-representative sampling approach enhances the representativeness of the recruited sample, ensuring the generalizability of the findings. Secondly, the integration of widely validated tools with strong psychometric properties, such as the 5C model and the Digital Vaccine Literacy (DVL) tool by Montagni et al., strengthens the validity of the results, particularly in assessing the psychological dimensions of vaccine hesitancy.^17,27 Thirdly, by examining under-researched determinants of vaccine behaviors – specifically political and economic factors – this study contributes to the ongoing debate on vaccine hesitancy by highlighting the influence of broader societal contexts. Finally, investigating digital vaccine literacy provides valuable insights for developing evidence-based policy recommendations to enhance health communication strategies in the Czech Republic.

The present study has several limitations that should be considered when interpreting the findings. The reliance on digitally collected data introduces a degree of selection bias and limits generalizability, as individuals without internet access, particularly older adults or those in rural or lower-income settings, were excluded. Nevertheless, this limitation is partly mitigated by Czechia's relatively high internet penetration rate (86/), which supports the representativeness of the sample.⁸⁴ Self-reported data present another limitation, as social desirability bias may have led participants to overstate their vaccine confidence or uptake; this could result in an overestimation of population-wide acceptance levels. However, anonymity and online data collection may have reduced this bias. Moreover, the cross-sectional design prevents causal inferences, and while the data are drawn from a well-established national panel, the observed associations cannot be generalized to temporal or longitudinal dynamics. Lastly, the internal consistency of the collective responsibility and DVL – information appraisal subscales was modest, likely due to the limited number of items. Findings involving these specific subdimensions should therefore be interpreted with appropriate caution.

Our findings highlight the need to address political determinants of vaccine behaviors, particularly trust in constitutional institutions and attitudes toward immigrants and refugees, by expanding interventions beyond individual behaviors to take into account the broader societal factors. Strengthening public trust in governmental sources and ensuring that vaccine communication is transparent, accessible, and engaging are also critical measures. Additionally, enhancing digital vaccine literacy is vital to improve the critical evaluation of online health information and counter misinformation. Finally, tailored communication strategies are needed to effectively reach socioeconomically disadvantaged groups and digital-first populations, ensuring equitable access to reliable vaccine information. To counter the influence of vaccine-skeptical Web 2.0 environments, official digital sources should consider adopting interactive and user-tailored communication strategies – such as integrating real-time Q&A features, improving mobile accessibility, and collaborating with trusted online influencers – to enhance credibility and engagement across different population segments.