*3.4. Factors Associated with COVID-19 Vaccination up to 1 October 2021*

As of 1 October 2021, immediately after the peak of COVID-19-related deaths in Guatemala, 2.58 million persons (15.1% of the total population) had completed a primary COVID-19 vaccine course [11]. In the bivariate analysis of all COVID-19 vaccination data up to 1 October 2021, significant factors negatively associated with COVID-19 vaccination coverage by municipality (α = 0.05), adjusting for departmental level differences, were similar to the overall analysis and included the proportion of the municipality identifying as Mayan, the proportion living in a rural residence, and proportion experiencing poverty, but did not include the departmental-level under-five childhood mortality rate and the department's Gini coefficient (Table 5). Factors positively associated in the bivariate model were the same as in the overall analysis.

After adjusting for all covariates and departmental effects in the full model, the proportion of the municipal population (1) having received at least a primary school education, (2) experiencing poverty, (3) aged 60 and above, and (4) tested for SARS-CoV-2 remained significantly associated with complete vaccination coverage (Table 5, Section "Full multivariable model"). In the simplified multivariable model (Table 5, Section "Simplified multivariable model"), when adjusting for covariates and departmental level differences, a 10% higher proportion of people experiencing poverty within a municipality was associated with 1.1% lower COVID-19 vaccination coverage (95% CI: −1.90–−0.39); a 10% increase in the proportion of the municipality having received at least a primary school education was associated with 1.6% higher COVID-19 vaccination coverage (95% CI: −0.12–3.28); a 10% increase in the proportion of the municipality in the 60 years and older age group was associated with 13.9% higher COVID-19 vaccination coverage (95% CI: 8.93–18.73); and a 10% higher proportion of the municipality tested for SARS-CoV-2 infection was associated with a 2.6% higher COVID-19 vaccination coverage (95% CI: 1.46–3.80) (conditional R<sup>2</sup> = 0.615).





**Table 4.** Association between sociodemographic factors (by municipalities and departments) and two-dose vaccination coverage (%) among patients aged 60 years or older by municipalities in Guatemala (N = 336). SARS-CoV-2 case and vaccination data are from 13 February 2020 to 30 November 2022.



**Table 5.** Association between sociodemographic factors (by municipalities and departments) and two-dose vaccination coverage (%) by municipalities in Guatemala(N = 336). SARS-CoV-2 case and vaccination data are from 13 February 2020 to 1 October 2021.


#### **4. Discussion**

In this cross-sectional analysis of COVID-19 vaccination coverage among Guatemalan municipalities, we provide population-level information on sociodemographic and health system variables associated with vaccination. In the adjusted multi-level model evaluating vaccination data as of 30 November 2022, municipalities with higher proportions of people experiencing poverty had lower COVID-19 vaccination coverage. Municipalities with higher proportions of people who had received at least a primary school education, children, people aged 60 years or older, and testing for SARS-CoV-2 infection had higher COVID-19 vaccination coverage. In our subanalyses, the timing of the country's response to the pandemic did not appear to notably affect the results, as factors associated with vaccination coverage in the overall model were similar to the point at which Guatemala had just passed its highest daily death rate. Additionally, poverty, educational level, and prevalence of testing for SARS-CoV-2 infection remained significant factors associated with COVID-19 vaccination coverage among Guatemalans aged ≥60 years.

Our findings are generally consistent with those from previous studies. Prior studies have largely focused on factors related to the intent to vaccinate rather than the completion of COVID-19 vaccination. A 2022 global, population-based analysis noted that participants identifying as female, in older age groups, with a higher level of education, and with health insurance reported being more willing to get vaccinated [32]. In a study of Latin American and Caribbean countries, those with a university education, residence in an urban area, and a higher perceived likelihood of contracting COVID-19 had higher intentions to be vaccinated [31]. With regards to age, our model showed that municipalities with more children and people aged 60 years or older had higher vaccine coverage when adjusting for other covariates and variations at the departmental level. This is expected given that older populations were prioritized under national vaccination planning given their elevated risk of severe COVID-19 [8]. The role of children is less clear, especially as vaccines for children under 12 years of age only became available in 2022, and there are no vaccines for children under six years of age at the time of analysis [38]. Possibly, concern over the well-being of children motivated parents' vaccination, or that when vaccines were available for children, other family members were also vaccinated. The role of children in the community could be an area of further investigation.

It is also expected that municipalities with more SARS-CoV-2 testing had a higher proportion of the population that was vaccinated for COVID-19 as these municipalities may have more access to health facilities or services. The MSPAS provided free testing to people with symptoms or to those who were COVID-19 contacts, however, these services were generally offered at health facilities that were not always accessible to rural populations [9,21,39]. It is possible that the presence of more testing resources could positively influence individuals to receive COVID-19 vaccinations. This would support public health interventions, such as making SARS-CoV-2 testing more accessible in rural areas through mobile health units. In our model, higher proportions of deaths due to COVID-19 in the municipalities were not significantly associated with increased vaccination. While Guatemala has reported less excess mortality compared with other countries within the region [40], the excess mortality was found to be 46% higher compared with confirmed COVID-19 death counts, according to a study by Martinez-Folgar and colleagues [41], indicating that mortality and case estimates are likely underestimated, which may have affected our analysis. Moreover, their study showed that most deaths appeared to occur at home, further highlighting barriers to healthcare access that are likely reflected in low COVID-19 vaccination coverage and possibly higher mortality.

In unadjusted models, we observed significant associations between lower COVID-19 vaccination coverage and Indigenous identity, rural residence, poverty, and self-reported difficulty accessing healthcare. In the adjusted model, of these sociodemographic variables, only the proportion of the municipality experiencing poverty remained negatively associated with vaccination coverage. It is possible that poverty partially explains the observed associations between COVID-19 vaccination and other sociodemographic variables, such

as Indigenous identity, rurality, and healthcare access. We found a few municipalities with high rurality and Indigenous populations that reached at least 70% vaccination coverage. There were only three municipalities with 50% or more of their population experiencing poverty that reached 70% vaccination coverage. Poverty remained significantly associated with low vaccination coverage when the analysis was restricted to the time that COVID-19 deaths peaked, or to coverage among those aged 60 years or older. Even as the risk of mortality due to COVID-19 has been shown to be higher among those in the lowest socioeconomic strata [23], there is evidence that economic insecurity was associated with fear of adverse effects from the vaccine in Latin America [27]. Additionally, the monetary and opportunity costs of accessing vaccination sites, missing work, arranging childcare, etc., have been described as potential barriers to vaccine access [5,21]. Therefore, while Indigenous and rural communities are at a higher risk for low vaccine access, it may be particularly effective to use poverty indices when designing community-wide vaccination interventions in Guatemalan municipalities, and to focus on interventions such as transportation, childcare, and alternative hours of service that can overcome cost-related barriers. Further, research to better understand the structural determinants of poverty, including class, gender, and race, can help guide future interventions [42]. Additional research on the monetary and opportunity costs of accessing vaccination within Guatemala may be needed. Lastly, outreach specifically to areas with lower access to primary school education and vaccination programming that accommodates potential literacy issues may be considered.

There are limitations to this study which should be considered. Our analysis was conducted at the municipal level as we did not have access to community estimates or individual-level data that could possibly provide more complex explanations of low vaccination coverage among certain sociodemographic groups. Our conclusions at the population level may not be applicable to specific sociodemographic groups within municipalities. Secondly, factors such as poverty, Indigenous identity, and rurality are complex and interrelated, and it is difficult to assess their relationship to vaccination and healthcare access in isolation. The proxies we used for healthcare access, such as testing for SARS-CoV-2 infections and vaccination program reach, such as childhood Pentavalent vaccination coverage, may not capture the intricacies of the political, economic, and historical reasons for low COVID-19 vaccination coverage. Additionally, our analysis may differ depending on alternative definitions of vaccination coverage, such as partial vaccination with one dose or coverage with booster doses. Lastly, as we relied on data from the most recent national census, some of our findings may not reflect the population during the COVID-19 pandemic.
