The first batch of vaccines arrived in Colombia in February 2021 [16]; in July 2022, approximately 82.5% Colombians had at least one dose of the COVID-19 vaccine and around 70.5% were fully vaccinated [7]. To efficiently apply and distribute vaccines throughout the country, the Ministry of Health prioritized vaccine access according to a person’s risk, considering factors such as work, age, and underlying diseases [16, 17]. Vaccines were offered free of charge to ensure access for economically vulnerable populations and to achieve an equitable distribution [17]. To effectively vaccinate the population, the government designed campaigns that stressed the importance and efficiency of the vaccine to help restore safely to “normality”, [18].

The national government designed numerous campaigns to ensure the distribution of vaccines in all areas of the country. Thanks to an agreement between the regional government and private companies, vaccination posts were established in public transport [19] and some municipalities. For example, in Rionegro, drive-through vaccination stations allowed citizens to receive the vaccine in their cars [20]. Vaccine posts were established in schools or community action offices [17] for people who lived far from health centers. This strategy was intended to allow the entire Colombian population to access state-funded vaccines regardless of where they live. In addition, to increase vaccination rates, the regional government ran a public awareness campaign Vacunarte te da más (being vaccinated gives you more), encouraging its citizens to get vaccinated. This campaign offered discounts and additional benefits in some venues, such as movie theaters and shopping centers, to people who presented their vaccination cards [21].

We conducted a regional multitopic household survey of 800 households (1,592 individuals) from 22 municipalities in rural Antioquia. This constitutes a representative sample for all rural sectors in Antioquia. Our survey collected information related to choices, perspectives, and concerns on six topics: land, ease of access to credit, health, household assets, welfare, and sociodemographic characteristics of the household. Furthermore, we collected a household-level roster, allowing us to construct household-level variables for respondents and the people living with them. We collected several variables, including age, education level, place of birth, and type of employment contract.

To analyze COVID-19 vaccination rates, we built two approaches:

• Share of household members with at least one dose

The average age of the head of household is 46.6 years, the average household size is 3.7 people, and an average dependency rate of 0.7%. Only 4.8% report having a woman as head of household. Regarding household assets, 93.1% of households report having at least one television at home, 92.4% have a refrigerator, 55.7% have a washing machine, 60.1% have access to drinking water, and 95.5% have access to electricity. There is low access to communication technologies, with only 22.6% of households reported having an internet connection, and 15.7% having a satellite connection. Respondents report relatively low levels of trust. We find that 53.1% of households say they do not trust the community action board, 81.4% say they do not trust the police or the armed forces, and 61.2% say they do trust the church. Most survey respondents (68.4%) say they are unhappy with health services.

Vaccination rates were relatively high, but did not reach the same number of households. By 20 February 2022, 76% of household members older than 14 years received at least one dose, and 55 percent received two doses. The analysis considers the share of household members who received at least one dose and were fully vaccinated. Table 1 describes the main characteristics of households with at least one dose of the COVID-19 vaccine. We find that at the time of the survey, households where the head of household has more than a middle school education report higher vaccination rates (4.1% vs. 2.1%). Households with internet access or satellite connection report higher vaccination rates for all household members older than 14, (2.41% vs. 1.79%) and (1.64% vs. 1.33%), respectively.

Households who report trusting the community action board, police, and the armed forces report higher vaccination rates (48.1% vs. 42.9%) and (27.8% vs. 17.3%), respectively. On the other hand, households who report trusting the church and those who report being satisfied with their health services center report lower vaccination rates (60.3% vs. 63.8%) and (30.7% vs. 34.2%), respectively.

A closer examination of the household survey indicates that households that obtain information from newspapers or on the Internet are on average less vaccinated (6.0% vs. 7.7%) and (41.6% vs. 45.9%). Households that receive information through radio or television, social networks, or conversations with friends are, on average more vaccinated, (85.2% vs. 82.7%), (45.5% vs. 45.4%), and (62.8% vs. 59.2%), respectively. Finally, households whose heads of household report being concerned about COVID-19 report higher vaccination rates (64.6%).

Table 2 describes the main characteristics of households with more than one dose of the COVID-19 vaccine (“fully vaccinated”). We find that households where the head of household has more than a middle school education have lower vaccination rates at the time of the survey (2.3% vs. 3.0%). Households with internet access and satellite connection have lower vaccination rates for all household members over age 14, (22.0% vs. 23.4%) and (14.9% vs. 16.8%), respectively.

When we examine trust in institutions, we find that households who report trusting the community action board, the church, and the police and the armed forces are on average, more vaccinated (49.9% vs. 42.5%), (61.6% vs. 60.5%) and (28.3% vs. 21.0%), respectively. However, households who report being satisfied with their health services center are on average, less vaccinated (29.6% vs. 34.4%).

Furthermore, we find that households that obtain information through radio or television, the Internet, social networks, and conversations with friends are on average more vaccinated (85.5% vs. 83.2%), (43.8% vs. 41.0%) (47.4% vs. 42.8%), and (65.2% vs. 57.2%), respectively. Households who obtain information from newspapers are on average less vaccinated (5.7% vs. 7.5%). Finally, households whose heads of household said they were concerned about contracting COVID-19 report higher vaccination rates (65.4% vs. 55.4%).

We estimate the main determinants of remote rural households becoming vaccinated once the availability of the vaccine was guaranteed. We define the linear probability model as follows: V a c i = β 0 + β 1 H H i + β 2 W e a l t h i + β 3 R e m o t e n e s s i + β 4 T r u s t i + β 5 R i s k i + γ i + μ i (1)

The subscripts denote the household i. Vac _i is a vector that contains at least one dose of the COVID-19 vaccine and total vaccination; HH is a vector that contains household characteristics such as sex and education of the head of household, the number of children in the household, the employment rate and the dependency ratio. Wealth _i is a vector that contains a wealth index per household, as well as access to the Internet, a satellite connection, a smartphone, and a computer; Remoteness _i indicates the distance between the household i and the municipal capital; Trust _i is a vector that contains the levels of confidence in the community action board, the church, the police and the armed forces, and sentiment about the health services center; Risk _i is a vector that includes related variables, along with perception of risk about COVID-19; γ _i is a vector of controls that includes fixed effects of stratum and age of head of household; μ _i is an error term with mean zero corrected by sampling design.

In this section, we discuss the main determinants of participation in vaccination for at least one dose. Table 3 and Figure 1 show how the results can change when we add the controls used in Eq. 1 one by one.

In this section, we discuss the main determinants of vaccination take-up for two doses (fully vaccinated), using the results of Eq. 1. Table 4 and Figure 2 show how the results can change when we add the controls used in Eq. 1 one by one.

The willingness to vaccinate is crucial, especially in a global health issue where the only viable solution is to achieve “herd immunity,” which can only be attained through widespread vaccination. This importance is further accentuated in areas where people often face greater economic challenges than the national average, such as rural areas, where the ability to cope with diseases may be limited. In this scenario, more research is needed to address the various factors influencing people’s willingness to vaccinate, especially in rural settings.

This study explored the importance of communication, measured by internet access and satellite connections, in people’s willingness to vaccinate. Therefore, public health investments related to vaccination should consider the availability of vaccines and the places where they are intended to be taken. In other words, it is essential to consider that in rural areas, the likelihood of accessing information through “traditional” media decreases as households move away from urban areas. Therefore, it is necessary to design campaigns that allow households in very remote areas of the capital to stay informed.

Furthermore, it was found that trust in the army and police is a key determinant that leads households to increase their probability of vaccination. Given this, it is important to consider again that it is crucial to have the supply or availability of the vaccine and trust the provider. In rural areas, where the army or police may represent the presence of the state, these figures play an important role in deciding to get vaccinated. Similarly, this occurs with the church, where belonging to a church decreases the likelihood that households are vaccinated. This underscores the ultimate need to include relevant groups or individuals who have representative roles within communities in vaccination campaigns.

We assume that by addressing the preferences or decisions of the households regarding vaccination, we indirectly measure the demand for vaccines from these households. However, it is essential to recognize that the actual availability of vaccines in each municipality is a factor that we cannot directly measure. Although we have reports and figures provided by the national government on vaccine distribution and administration, the quality and periodicity of this information may limit our complete understanding of vaccine supply, especially in rural areas of the country.

It is also crucial to acknowledge that the findings derived from this research are only related to the household level. Therefore, no possible behavioral changes can be observed that may occur within a household, impacting an individual’s inclination toward vaccination. This includes scenarios where individuals coexist with varying vaccination rates. So, future research prioritize examining the dynamics or disparities that may arise in individuals’ willingness to vaccinate within the same household.