Abstract
Objectives:
The COVID-19 pandemic demonstrated how vaccination decisions are influenced by misinformation, disinformation, and social pressures, leading to varied and inequitable uptake rates. In this study, we examined how COVID-19 vaccine messages received via social networks were associated with vaccine uptake in rural Alabama.
Methods:
From November 2021 through March 2022, we collected 700 responses to a telephone survey administered in 4 rural Alabama counties. We asked respondents to indicate whether certain social relationships (eg, family, businesses) tried to influence them to (1) obtain or (2) avoid a COVID-19 vaccine. We used χ2 tests, Kruskal–Wallis tests, Mantel–Haenszel χ2 tests, and Fisher exact tests to examine the associations between vaccination status and survey responses.
Results:
Respondents in majority–African American counties were significantly more likely than those in majority–White counties to have received ≥1 dose of COVID-19 vaccine (89.8% vs 72.3%; P < .001). Respondents who received ≥1 dose had a significantly higher mean age than those who had not (58.0 vs 39.0 years; P < .001). Respondents who were encouraged to get vaccinated by religious leaders were more likely to have received ≥1 dose (P = .001), and those who were encouraged to avoid vaccination by family (P = .007), friends (P = .02), coworkers (P = .003), and health care providers (P < .001) were less likely to have received ≥1 dose. Respondents with more interpersonal relationships that encouraged them to avoid vaccination were more likely to be unvaccinated (P < .001).
Conclusions:
Interpersonal relationships and demographic characteristics appeared to be important in COVID-19 vaccine decision-making in rural Alabama. Further research needs to identify how to facilitate vaccine-positive interpersonal relationships, such as peer mentoring and trusted messenger interventions.
The COVID-19 pandemic has highlighted the essential role of vaccination in mitigating morbidity and mortality from infectious diseases. Moreover, the ability to rapidly develop vaccines means that new vaccines can be introduced on relatively short notice. However, vaccine hesitancy is a major challenge, particularly among socially and medically marginalized populations, such as African American people and people living in rural communities. 1 Therefore, creating effective information campaigns is essential if communities are to successfully mitigate infectious disease morbidity and mortality. 2 However, the spread of misinformation about vaccines through social networks and distrust of mainstream sources of medical information are growing concerns. 2
People in rural communities are more likely than people living in urban areas to be hesitant about getting immunized against COVID-19 and other health conditions.1,3-6 Lack of information or misinformation is associated with vaccine hesitancy.7,8 Specifically, hesitancy in rural communities is linked to perceptions that the COVID-19 vaccines are unsafe and not trustworthy.7,8 However, the delivery of public health messages, including messages that support vaccine uptake, can be challenging in rural communities. For example, rural residents, compared with urban residents, have less access to health information from physicians, use internet search engines less frequently, 9 and are less likely to have access to preventive care services, in part due to underlying demographic and socioeconomic factors. 10 Additionally, mistrust of medical and public health institutions may be heightened among the African American population because of structural racism and historical violence, such as the Tuskegee experiments.11,12
Social networks, or the friends, family, and colleagues a person interacts with, can be major sources of health information, provide opportunities for health education and health promotion, and serve as mechanisms for peer support. Social networks can have important influences on health behaviors, including vaccine uptake. 13 For example, a study published in 2019 showed that 80% of African American parents trusted family or friends “some” or “a lot” for advice on vaccinating their children against human papillomavirus (HPV). 14 In another study, high levels of exposure to both anti–HPV vaccine viewpoints and limited exposure to pro–HPV vaccine viewpoints were associated with HPV vaccine refusal. 14 And, in another study, parents who reported that most or all of their friends would not get the COVID-19 vaccine were significantly more likely to express unsure intention to vaccinate their children. 15
In addition to these interpersonal relationships, research has shown that organizations (eg, businesses, schools) and professional relationships (eg, religious leaders, health care providers) affect a person’s vaccine decision-making. An article published in 2023, for example, demonstrated how patients in Ukraine prefer COVID-19 vaccine communications to come from health care providers, and a US-based survey in 2021 showed that a belief in God was negatively associated with vaccine uptake.16,17 Additionally, trust in authority and in the scientific community is positively associated with COVID-19 vaccine uptake.18,19 A strong theoretical basis exists for the influence of social networks at multiple levels (family, friends, and coworkers) on health behaviors. The socioecological model, for example, notes an interpersonal level of influence, as well as community- and organizational-level influences, on vaccine uptake. 20
The objective of this study was to explore the role of social networks in COVID-19 vaccine uptake in rural Alabama and examine the ways in which social networks affect vaccine hesitancy. Although the crisis of the 2020-2022 COVID-19 pandemic has passed to some extent, lessons from the pandemic remain applicable across other initiatives designed to increase vaccine uptake. Findings from our study can enhance public health officials’ approaches to future vaccination campaigns in rural locales by focusing on the importance of social networks.
Methods
We conducted a survey of rural Alabama residents in 4 counties from November 2021 through March 2022 as part of a COVID-19 testing initiative funded by the National Institutes of Health. 21 While the survey comprehensively measured attitudes and beliefs about COVID-19 testing and vaccinations, this study focused on which social relationships (ie, interpersonal relationships) were likely to influence respondents’ vaccination decisions. This project received expedited approval from the institutional review board at the University of Alabama at Birmingham (protocol number IRB-300006055) on September 16, 2020, and all respondents provided verbal informed consent.
We purposely sampled participants from 2 counties that were majority Black or African American and 2 counties that were majority White (Table 1). Community health workers (CHWs) administered surveys via telephone to participants. To reduce participant burden and facilitate collection of high-quality data, the survey was web-based and data were collected via REDCap. 22 Inclusion criteria were residence in 1 of the 4 counties and an age of ≥19 years. To collect data, CHWs read the survey to participants and then recorded their responses. In October 2021, we provided 6 hours of live virtual training to 16 CHWs on study protocol and consenting processes, including live virtual instruction on the aims, purpose, and methods of this study; CITI Program training on the ethics of human subjects research (https://about.citiprogram.org); and interactive role-playing. Of the 16 interviewers, 12 identified as Black or African American and female. Each CHW completed a certification for survey administration with an experienced interviewer. CHWs placed recruitment flyers at local businesses, churches, and other community organizations. Survey participants directly contacted CHWs working in their communities or were contacted by CHWs after submitting a survey interest card. Survey responses were collected from November 2021 through March 2022. Survey completion took approximately 30 minutes, and respondents received a mailed $50 gift card upon survey completion.
Racial and ethnic characteristics of 4 rural counties in Alabama from which a sample of respondents (N = 700) was drawn for a survey on social networks and COVID-19 vaccine uptake, November 2021–March 2022 a
Data source: US Census Bureau. 23
The survey asked respondents whether they had received ≥1 dose of the COVID-19 vaccine (yes or no). We created ordinal variables from 2 other survey questions about who influenced respondents to receive or not receive the COVID-19 vaccine:
• [Do/did] you feel any of the following (family, friends, employer, coworkers, schools, businesses, religious leaders, doctors or other health care providers, government officials) tried to influence you to get a COVID-19 vaccine?
• [Do/did] you feel any of the following (family, friends, employer, coworkers, schools, businesses, religious leaders, doctors or other health care providers, government officials) tried to influence you to AVOID getting a COVID-19 vaccine?
For both questions, we combined types of influencers into 3 groups. One group combined family, friends, and coworkers into the domain “interpersonal relationships.” A second group combined businesses, employers, and schools into the domain “organizational relationships.” The third group combined religious leaders, health care professionals, and government officials into the domain “professional relationships.” The scores of each of the 6 new variables (3 per question) ranged from 0 to 3, with 0 indicating that the respondent was not influenced by any individual or organization in that domain and 3 indicating that the respondent was influenced by every type of individual or organization in that domain (eg, family, friends, and coworkers). We used univariate demographic characteristics to describe the sample and bivariate data to demonstrate the relationship between vaccine uptake and a respondent’s social network.
We compared the rate of vaccination (≥1 dose) among types of influencers (eg, family) with the overall rate in the sample and compared the rates of vaccination between the number of influencers within each domain (eg, 1 influencer in the interpersonal domain vs 3 influencers in the interpersonal domain). We used Pearson χ2 tests of association to evaluate differences between patients receiving ≥1 dose of the COVID-19 vaccine when the outcome was categorical and Mantel–Haenszel χ2 tests when one of the categorical variables was ordinal (ie, relationship domain). We used modified ridit scores in all Mantel–Haenszel χ2 tests to avoid the assumption that the categories of the ordinal variable were equally spaced. If the expected counts of categorical variables violated the test assumptions, we used Fisher exact tests. We performed all calculations in SAS version 9.4 (SAS Institute, Inc); P values < .05 were considered significant.
Results
Our sample consisted of 700 respondents, most of whom were Black or African American (n = 573; 81.9%), not Hispanic (n = 694; 99.1%), and assigned female sex at birth (n = 589; 84.1%) (Table 2). The median (IQR) age was 56.0 (41.0-64.0) years.
Demographic characteristics of a sample of respondents (N = 700) to a survey on social networks and COVID-19 vaccine uptake administered in 4 rural counties in Alabama, November 2021–March 2022
All values are number (percentage) unless otherwise indicated.
Some demographic characteristics and relationships were associated with a respondent’s decision to receive ≥1 dose of a COVID-19 vaccine (Table 3). Respondents who received ≥1 dose had a higher median age than respondents who had not received ≥1 dose (58.0 vs 39.0 years; P < .001). Proportions receiving the vaccine differed across counties: 93.4% of respondents from Bullock County reported ≥1 dose, whereas 65.2% of respondents in Marion County reported ≥1 dose (P < .001). Respondents who lived in a majority–African American county had a significantly higher vaccination rate than those who lived in a majority–White county (89.8% vs 72.3%; P < .001).
Bivariate comparisons between demographic characteristics and vaccine uptake: results of a survey on social networks and COVID-19 vaccine uptake administered to residents (N = 700) of 4 rural counties in Alabama, November 2021–March 2022
Abbreviation: —, not applicable
Unless otherwise indicated.
Determined by Pearson χ2 test of association; P < .05 considered significant.
Determined by Kruskal–Wallis test; P < .05 considered significant.
Each additional year of age was associated with 1.07 times increased odds of having received ≥1 dose.
For each domain, we assigned a score of 0 to 3 to each respondent, indicating whether the respondent reported they were encouraged to obtain a COVID-19 vaccine by 0, 1, 2, or 3 of the individuals or institutions in that domain (eg, a respondent scoring 3 in the interpersonal domain reported being encouraged to obtain a vaccine by family, friends, and coworkers).
Determined by Mantel–Haenszel χ2 test; P < .05 considered significant. Modified ridit scores were used to avoid the assumption that the categories of the ordinal variable were equally spaced.
Determined by Fisher exact test; P < .05 considered significant.
Particular kinds of relationships appear to have had a substantial influence on the decision to obtain or avoid COVID-19 vaccination (Table 3). Specifically, respondents who were encouraged to be vaccinated by religious leaders were significantly more likely than the overall sample to have received ≥1 dose (P = .001; 92.4% vs 83.1%). Additionally, respondents who were discouraged from being vaccinated by family (P = .007; 72.8%), friends (P = .02; 77.7%), coworkers (P = .003; 65.8%), and health care providers (P < .001; 36.4%) were less likely than the overall sample (83.1%) to have received ≥1 dose. Interestingly, respondents who were discouraged from vaccination by government officials were significantly more likely than the overall sample to have received ≥1 dose (93.6% vs 83.1%; P = .047), as were respondents who had not been discouraged from vaccination by any of the individuals or institutions listed (86.0%; P = .04).
For the most part, the total number of business and professional relationships was not significantly associated with the decision to obtain or avoid a vaccine. The number of influencers within the interpersonal domain expressing an opinion on vaccines was associated with uptake rates. As the number of interpersonal relationships discouraging vaccination increased, so did the likelihood of not obtaining the vaccine (P < .001). For example, 42.9% of respondents who had family members, friends, and coworkers (a score of 3) discouraging vaccination received ≥1 dose, whereas 86.2% of those who had only 1 of those influencers discourage vaccination received ≥1 dose.
Discussion
Survey results in a sample of adults in 4 rural Alabama counties indicate that social networks are powerful influencers of COVID-19 vaccination decisions and that discouragement from friends, family, coworkers, and health care providers negatively affects receipt of COVID-19 vaccine. Notably, as the number of types of interpersonal relationships discouraging receipt increased, the likelihood of receiving ≥1 dose of vaccine decreased substantially (57.1% vs 13.8% for 3 vs 0 types of relationships). Moreover, the overall population characteristics of the county of residence were also associated with receipt of ≥1 dose, with respondents in majority–White counties being less likely than respondents in majority–African American counties to be vaccinated. Taken together, these findings suggest that public health campaigns need to be mindful of the heterogeneity and variability of rural counties when developing communications and outreach programs, and the central role of interpersonal relationships in informing vaccine decisions, particularly the powerful role of discouragement. Interventions that leverage endogenous social capital and relationships, such as CHW programs, could prove valuable in widely disseminating trusted information that dispels misinformation among social networks as a means to encourage health promotion behaviors, such as vaccination.
Previous studies have also shown that vaccine attitudes are influenced by one’s social network. 24 For example, Hayashi et al 25 found subjective norms (beliefs that others either approve or disapprove of behaviors) predict intentions to accept COVID-19 vaccination. Prior studies have also noted that people who received COVID-19 vaccinations believed that “those important to them” supported their decision and that levels of trust in friends and family members predicted COVID-19 vaccine hesitancy.24,26 In addition, trust in government and health care professionals improves COVID-19 vaccine acceptance. 19 In an interesting deviation from the literature, our survey showed that respondents were less likely to have received a vaccine if a health care provider advised them to avoid vaccination but were not significantly more likely to be vaccinated if a health care provider encouraged them to get vaccinated. However, this finding could reflect the small number of respondents (n = 11) who reported being advised to avoid vaccination by a health care provider. It suggests that in rural Alabama, at least, more work is needed to build public trust in medical professionals and public health officials to offset social influences and improve preventive social norms.26,27
Our analysis of demographic data showed that older respondents were more likely than younger respondents to have received a vaccine, which mirrors the results found in other studies.28,29 Contrary to many other studies, however, our demographic analyses demonstrated that respondents living in majority–African American counties had higher rates of COVID-19 vaccination than respondents living in majority–White counties; other studies found lower vaccination rates among racial and ethnic minority populations than among non-Hispanic White populations.30,31 The relatively high vaccination rate in the 2 majority–African American counties included in our study may have been due to community-based efforts and engagement with trusted messengers to improve vaccination rates. For example, our study results showed that respondents encouraged to get vaccinated by religious leaders were more likely to receive ≥1 dose than those who did not indicate that they were encouraged by religious leaders. At least 1 study conducted prior to COVID-19 vaccine availability found that, in Alabama, African American people were not significantly more hesitant than non-Hispanic White people to receive a vaccine, although hesitancy was relatively high among all racial groups. 32 Additionally, whereas previous studies found that women expressed more vaccine hesitancy than men, we did not find an association between sex at birth and vaccination status in our sample, which could have been due to the high percentage of women in our sample.33 -35
Studies examining the perceptions of community members show that barriers to COVID-19 vaccine acceptance, such as trust among African American people, can be overcome through community-based interventions and through information from trusted messengers, including religious leaders.36 -39 In future health emergencies, public health officials should focus more efforts on engaging community members, religious leaders, and other trusted messengers, especially those trusted by young people, to increase participation in interventions. 34
Limitations
Some limitations of our study should be highlighted. Our sample may not have accurately represented the entire population or community being studied: our convenience sampling approach may have led to the oversampling of women and Black people. Additionally, the $50 incentive could have led to participation bias but not necessarily social desirability bias. Furthermore, caution must be exercised when attempting to generalize the findings of our research to a broader context or community beyond the populations included in this study. The particular characteristics and circumstances of the counties in our study could have considerably influenced outcomes, making it challenging to apply the results of our study universally.
Conclusion
Our study showed that social networks influence decisions on obtaining a COVID-19 vaccine and that vaccine behaviors differ across age and race. In addition, we found that social networks can have positive and negative influences on vaccine hesitancy or uptake. Notably, discouragement of vaccination by friends, family, and coworkers was significantly associated with not receiving the COVID-19 vaccine. Public health efforts designed to increase vaccine uptake should consider how social networks can be leveraged to communicate the importance of vaccination, in particular, to dispel misinformation, mistrust, and distrust that might lead to discouragement from vaccination. Geographically tailored, culturally relevant, and age-appropriate communication, delivered in a local context via trusted messengers, is needed to build public trust and confidence in public health interventions in the United States and may be particularly salient to socially and medically marginalized rural communities.
Footnotes
Acknowledgements
The authors acknowledge their community partners on this project: the Alabama Area Health Education Centers, the Alabama Quality Management Group, ConnectionHealth, and Acclinate. In addition, the study team thanks its Human Subjects Unit and its Community and Scientific Advisory Board for providing guidance to ensure this project was ethical, culturally competent, and scientifically robust, as well as Lynn Matthews, MD, MPH, and Aadia Rana, MD, from the University of Alabama at Birmingham Heersink School of Medicine for their leadership on this project.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is part of Rapid Acceleration of Diagnostics–Underserved Populations, a National Institutes of Health–funded initiative to increase access to SARS-CoV-2 testing among vulnerable populations. This study’s National Institutes of Health award is a supplement to the University of Alabama at Birmingham Center for AIDS Research, 3P30AI027767-33S1.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.