Abstract

Background

The COVID-19 pandemic resulted in the rapid development and distribution of vaccines as a critical strategy to control the spread of the virus. This paper explores COVID-19 vaccine uptake in the state of Ohio, with a particular focus on the difference between metro and non-metro residents.

Method

Survey data collected as part of the IMPACT-Ohio Project were used for this study. From August 2021 to February 2023, 3,806 individuals who resided in 12 Ohio counties (six metro and six non-metro counties) responded to the survey. Chi-square tests compared the relationships between various demographic, socio-economic and clinical characteristics among metro and non-metro region respondents. Binary logistic regression modeled the probability of receipt of COVID-19 vaccine and compared those Ohioans who lived in metro (RUCC codes 1鈥�3) vs non-metro (RUCC codes 4鈥�9) counties with adjustment of various covariates.

Results

Participants residing in metro counties were almost two times more likely to receive the COVID-19 vaccine compared to those living in non-metro counties adjusting for demographic, socioeconomic and clinical characteristics (aOR: 1.89, 95% CI: 1.38鈥�2.58, P鈥�&濒迟;鈥�0.0001). Lower COVID-19 vaccine uptake was associated with younger age (less than 65听years old), lower education level, having no health insurance or public insurance and being food insecure.

Conclusion

This study provides a comprehensive understanding of the barriers and determinants associated with vaccine uptake which can inform future interventions and public health policies aimed at improving vaccination rates in Ohio.

Introduction

The COVID-19 pandemic highlighted health disparities globally, particularly those affecting minority, underserved, and non-metro populations [1]. Lower rates of vaccine uptake within these communities have been well-documented historically, with factors such as limited healthcare access, higher levels of vaccine hesitancy, and broader socioeconomic challenges contributing to these disparities [2]. Non-metro regions, including Appalachian areas, face unique barriers to healthcare, including geographic isolation, fewer healthcare facilities, higher poverty rates, and infrastructural limitations [2, 3]. These factors amplify risks in public health crises, such as COVID-19, where vaccine uptake can play a crucial role in reducing transmission and severe health impacts.

Previous studies indicate that non-metro residents face distinct challenges, including limited access to vaccination sites, lower trust in healthcare systems, and fewer health promotion resources tailored to non-metro settings compared to their metro counterparts [3, 4]. Further complicating vaccine acceptance is the interaction between cultural beliefs and community norms that sometimes magnify misinformation and mistrust among specific populations [5]. Demographic variables like age, sex, income, educational attainment, and prior vaccination experiences, have also been shown to be indicators of COVID-19 vaccine uptake [3, 6,7,8]. However, few studies have directly compared metro and non-metro populations within a single state like Ohio, which has a mix of metro, non-metro, and Appalachian counties, each with unique health needs and barriers to vaccine access.

This study seeks to deepen understanding of the factors influencing COVID-19 vaccine uptake in Ohio by comparing metro and non-metro populations and examining how demographic, socioeconomic, and clinical variables interact with vaccine behaviors. By examining the unique challenges and characteristics of these geographic areas, this study seeks to provide actionable insights that can inform public health policies. The findings will be valuable for policymakers, healthcare providers, and public health organizations in designing interventions that ensure equitable access to future vaccines and enhancing vaccination coverage in non-metro regions will be vital for improving overall public health outcomes in Ohio.

Methods

Data was collected from a survey conducted by IMPACT-Ohio (Improving Access to Covid Testing in Ohio, part of the national initiative, RADx-UP). RADx-UP was created by the National Institutes of Health with a focus on communities most affected by the pandemic [9]. IMPACT-Ohio tested an adapted, customizable multi-level intervention (MLI) to improve COVID-19 testing uptake, consisting of both clinic-based in-reach and community-based outreach using evidence-based interventions (EBIs), to address the underuse of COVID-19 testing in minority, underserved, and vulnerable populations (MUVPs) in 12 counties in Ohio. There were six metro counties (Jefferson, Franklin, Cuyahoga, Lucas, Trumbull and Butler) and six non-metro counties (Ross, Meigs, Muskingum, Williams, Hardin and Scioto) participating, including six Appalachian counties (Jefferson, Meigs, Muskingum, Ross, Scioto, and Trumbull), and activities occurred within the top ten most vulnerable census tracts of each county (except for Hardin, Meigs and Williams鈥攆or these three counties all their census tracts were utilized). The primary goal was to address and reduce the disparities in COVID-19 education, testing, contact tracing, follow-up and treatment (COVID-19 testing components) among MUVPs.

To better understand the impact of COVID-19 on an individual level, a survey was developed to solicit information from individuals living in certain census tracts in the 12 project Ohioan counties. The survey contains RADx-UP common data elements and the Ohio State University (OSU) questions. Two surveys were developed, one that participants completed from August 2021 to February 2023 and a second from March to August of 2023. Survey questions assessed participants鈥� use of COVID-19 testing, COVID-19 exposure, infections and symptoms experienced. The survey also assessed participants鈥� COVID-19 vaccine receipt and vaccine hesitancy, housing and employment, food insecurity, medical conditions, and participant demographics. The second survey asked about participants鈥� experiences with Long COVID as well as asked to detail the actions they took after receiving a positive COVID-19 test result.

Within the 12 project counties, information from purchased lists was used to distribute 410,000 survey postcards to individuals via mail with information as to how to complete the survey. Community members were given the opportunity to complete the online survey via a QR code or link and could complete a survey via mail or in-person. Community Health Workers (CHWs) also distributed additional postcards at various community events, with the self-testing kits which were part of the overall IMPACT-Ohio program. However, they did not actively recruit participants. Social media (i.e. Facebook) was also used to recruit individuals from the specified counties to complete the survey.

Participant zip codes were used to determine which Ohio county respondents lived in at the time they completed the survey, and based on the county of residence, they were determined to be metro or non-metro residents according to the 2013 Rural鈥揢rban Continuum Codes (RUCC) [10]. The RUCC codes of 1鈥�3 were defined as metro residents while 4鈥�9 were used to define non-metro residents (Fig.听1).

贵颈驳.听1

IMPACT-Ohio project counties with number of respondents

Variables

The dependent variable was COVID-19 vaccine uptake, defined as receipt of a COVID-19 vaccine (yes/no). The analysis was done comparing metro to non-metro residents with respect to the receipt of COVID-19 vaccine with adjustment of various demographic, socioeconomic and clinical variables. These demographic variables included: age (<鈥�30, 30鈥�49, 50鈥�64, 65鈥�+), sex (Male, Female), race (White, Black/African American, Other, Multiple Races), ethnicity (Non-Hispanic, Hispanic) and Appalachian residence was defined based on Appalachian Regional Commission (yes/no) [11]. Income was defined as total household income before taxes in 2019 (<鈥�$35,000, $35,000-$74,999, $75,000 or higher), education defined as highest level of education a respondent achieved (Less than HS, HS/GED, Some College, College Graduate or higher) and health insurance was defined as primary kind of health insurance or health care plan respondent has currently (No health insurance, Private, Public).

The measure of socioeconomic status (SES) used in this analysis was based on modified Hollingshead scale and was derived by combining information on employment status, education and income. Scores ranged from 0 to 6, with higher scores indicating better SES. All three variables were categorized as the following: employment status (currently working (+鈥�2), retired (+鈥�1), other (0)); education (more than high school (+鈥�2), high school or GED (+鈥�1), less than high school (0)) and income (over $50听K (+鈥�2), $25听K-$50听K (+鈥�1), under $25听K (0)). For the analysis, SES was categorized into three groups: SES scores of 0鈥�1, 2鈥�3 and 4鈥�6 representing low-, middle- and high-level SES, respectively [12].

Food insecurity was coded as 鈥榶es鈥� if respondent indicated 鈥榦ften true鈥� or 鈥榮ometimes true鈥� to either of these two questions: 鈥淭he food that (I/we) bought just did not last, and (I/we) didn鈥檛 have money to get more. Was that often, sometimes, or never true for the household in the last 12听months?鈥� and 鈥�(I/We) could not afford to eat balanced meals. Was that often, sometimes, or never true for the household in the last 12听months鈥�. Also, if the respondent indicated 鈥榶es鈥� to the following questions: 鈥淚n the last 12听months, did respondent/other adults in the household ever cut the size of the meals or skip meals because there wasn鈥檛 enough money for food?鈥� or 鈥淚n the last 12听months, were respondent ever hungry but didn鈥檛 eat because one couldn鈥檛 afford enough food?鈥�, he/she were coded as 鈥榶es鈥� for food insecurity. Otherwise, food insecurity was coded as 鈥榥o鈥�.

Stable housing was defined as whether the respondent had been staying in the same place in the past two months (yes/no). Among those who indicated they were currently working; the respondents were asked if they were considered as an essential worker (yes/no). Two clinical variables were used for previous vaccination history (received flu vaccination in the past but not in past year, received flu vaccination in the past year, has never received flu vaccination). Cancer diagnosis and/or treatment within the past 12听months was defined (yes/no). The 2020 Presidential election results at the county level in Ohio were used to determine county political affiliation [13]. Respondents who indicated 鈥渘ot too likely鈥�, 鈥渘ot at all likely鈥�, or 鈥渄efinitely not鈥� to the following question: 鈥淗ow likely are you to get COVID-19 vaccine?鈥�, were asked about their concerns. They were asked: 鈥淲hy would you not get a COVID-19 vaccine?鈥� and could select their reasons for being hesitant.

Statistical analysis

Descriptive statistics were displayed with frequencies and percentages for binary or categorical variables. Chi-square tests (or Fisher鈥檚 exact tests when appropriate) compared the relationships between various demographic, socio-economic and clinical characteristics among metro and non-metro region respondents. Binary logistic regression modeled the probability of receipt of COVID-19 vaccine and compared those Ohioans that lived in metro vs non-metro counties with adjustment of various covariates. Covariates that were statistically different (p-value鈥�<鈥�0.05) between metro and non-metro respondents were further included in the multivariable regression. Two multivariable regressions were utilized for this analysis to compare metro to non-metro respondents. One adjusted for age, sex, race, education, insurance, previous vaccination history, health insurance coverage, food insecurity and stable housing (omitting income due to its collinearity with insurance) while the second adjusted for age, sex, race, SES category, previous vaccination history, food insecurity and stable housing (omitting education, insurance and income due to its collinearity with SES group). Both bivariate and multivariable analyses included participants with non-missing responses to each variable of interest.

Additional analyses included displaying the types of concerns among the respondents who did not receive the COVID-19 vaccine and were not likely to receive it using frequencies and percentages, as well as using Chi-square test to evaluate the relationship between presidential voting results at the county-level and receipt of COVID-19 vaccine. All these analyses were performed using SAS v9.4 (SAS Institute; Cary, NC; ) and statistical significance was defined as two-sided alpha鈥�<鈥�0.05.

Results

Upon the distribution of 410,000 surveys, 3,940 surveys were completed. 3,806 (96.6%) of these surveys were used for analysis. The other 134 surveys were excluded due to missing zip codes, invalid zip codes, non-Ohio zip codes, zip codes outside of the 12 target counties and responses outside of the survey timeframe from August 2021 to February 2023. The eligible survey respondents were 3,806 Ohioans, among which 2,584 (67.9%) resided in metro areas and 1,222 (32.1%) resided in non-metro areas (Fig.听2).

贵颈驳.听2

Consort diagram

The metro respondents resided in six counties, with 40.1% in Cuyahoga, 30.4% in Franklin, 12.6% in Lucas, and the remaining 16.9% from Jefferson, Butler and Trumbull Counties. Non-metro respondents lived in six different counties, with 26.2% in Ross, 18.6% in Williams, 16.7% in Muskingum, and the remaining 38.5% from Hardin, Scioto and Meigs Counties (Table听1).

Table听1 Number of respondents by each of the 12 counties with Metro or Non-Metro residence

Table 2 depicts the characteristics of the sample overall and by metro/non-metro residence. Among all included participants, 3,309 (90.2%) received the COVID-19 vaccine while 360 (9.8%) individuals did not. In the sample population, 37.0% were 65听years of age or older, 61.3% were female, 81.1% were of white race, 97.0% were of non-Hispanic ethnicity, 47.9% had an education level of at least college graduate and 49.6% had private insurance that was purchased directly or through employment. Most respondents (78.1%) were classified as food secure, 95.9% had stable housing (have been staying in the same place in the past two months as of survey date), and 2.7% had a past cancer diagnosis and/or received cancer treatment within the last 12听months.

Table听2 Demographic summary of respondents by Metro vs Non-Metro residence

Respondents in metro areas compared to non-metro areas were significantly younger (p鈥�<鈥�0.0001), more likely to be female (63.6% vs 56.5%, p鈥�<鈥�0.0001) and less likely to be white (74.4% vs 95.3%, p鈥�&濒迟;鈥�0.0001). Those with metro vs non-metro residence were more likely to have lower total household income before taxes of less than $35,000 (34.4% vs 29.8%, p鈥�=鈥�0.0234), at least college education (51.5% vs 40.4%, p鈥�<鈥�0.0001) and more likely to have private health insurance (51.7% vs 45.0%, p鈥�=鈥�0.0009). More respondents in metro compared to non-metro areas reported food insecurity (23.7% vs 18.0%, p鈥�=鈥�0.0001). Among respondents who indicated they were currently working (n鈥�=鈥�1712), those who considered themselves as essential workers were more likely to reside in non-metro vs metro areas (59.8% vs 46.3%, p鈥�&濒迟;鈥�0.0001).

In the first multivariable analysis (Table听3), those participants residing in metro counties were almost two times more likely to receive the COVID-19 vaccine compared to those living in non-metro counties (aOR: 1.89, 95% CI: 1.38鈥�2.58). Participants who were younger than 65听years of age were at lower odds of receiving the COVID-19 vaccine compared to those who were 65 or older, specifically those who were younger than 30听years old were at 66% lower odds to have received the COVID-19 vaccine compared to those who were 65听years or older (aOR: 0.34, 95% CI: 0.18鈥�0.62). Those participants with some college education, were at 34% lower odds of receiving the COVID-19 vaccine compared to those with education of college graduate or higher (aOR: 0.66, 95% CI: 0.48鈥�0.92).

Table听3 Multivariable binary logistic regression for receipt of COVID-19 vaccine

Respondents who had no health insurance or had public health insurance were less likely to have received the COVID-19 vaccine compared to those with private insurance. Individuals who had received a flu vaccination in the past, but not in the past year, had 90% lower odds of receiving COVID-19 vaccine compared to those who received a flu vaccination this past year (aOR: 0.10, 95% CI: 0.07鈥�0.15). Residents who struggled with food insecurity were at 30% lower odds of receiving the COVID-19 vaccine, compared to those who were not struggling with food insecurity (aOR: 0.70, 95% CI: 0.50鈥�0.96). Sex, race, and stable housing were not statistically significant predictors of receiving COVID-19 vaccine in the first multivariable model.

In the second multivariable model, the results were very similar to what was seen in the first model (Table听4). The respondents residing in metro counties had 85% higher odds of receiving COVID-19 vaccine compared to those residing in non-metro counties (aOR: 1.85, 95% CI: 1.33鈥�2.57). The respondents with low level SES (scores of 0 or 1) were at 55% lower odds of receiving COVID-19 vaccine compared to respondents with high level SES (scores 4 through 6) (aOR: 0.45, 95% CI: 0.24鈥�0.85). Sex, race, and stable housing were not statistically significant predictors of receiving COVID-19 vaccine in this multivariable model as well.

Table听4 Multivariable binary logistic regression for receipt of COVID-19 vaccine

The concerns among respondents who did not receive the COVID-19 vaccine and indicated they were not likely to get it were documented (Table听5). The most common reasons for not likely in getting COVID-19 vaccine had to do with the potential side effects (69.3%) and safety of the vaccine (67.6%).

Table听5 Respondents concerns about the COVID-19 vaccine

The relationship between county presidential voting results and receipt of COVID-19 vaccine was also investigated (Table听6). Out of twelve counties in the study, three counties (Cuyahoga, Franklin and Lucas) voted for Democratic candidate while the remaining nine counties voted for Republican candidate. Participants living in counties that voted for the Democratic candidate were more likely to be vaccinated (91.6%) than those residing in counties that voted for the Republican candidate (88.3%).

Table听6 County-level presidential voting results and receipt of COVID-19 vaccine

Discussion

This study aimed to explore differences in COVID-19 vaccine uptake between metro and non-metro regions. Using data from a survey that was administered to Ohioans during the COVID-19 pandemic, our analysis found that COVID-19 vaccine uptake was significantly lower in non-metro counties compared to metro counties. Lower COVID-19 vaccine uptake was also associated with younger age (less than 65听years old), lower education level, having no health insurance or public insurance, receipt of flu vaccine in the past but not in the past year or never receiving flu vaccine and reporting food insecurity, as well as lower SES.

Individuals of higher SES often have better access to quality care, transportation, and have greater social support, all of which facilitate vaccination [1]. Moreover, higher SES is associated with higher levels of education, income, and health literacy, which often increases the willingness to receive vaccinations based on informed decision-making [1, 3]. Previous research also found higher vaccination rates in metro populations [3, 8, 14]. Residents of metro areas have more proximate to healthcare facilities and locations that offer vaccines [6]. Whereas fewer healthcare and transportation options and longer travel time to care for rural residents create barriers to vaccine access, particularly among individuals who lack reliable transportation or have limited mobility [15]. Understanding this association is crucial for future public health strategies to ensure equitable vaccine distribution and access among different SES groups.

The findings that COVID-19 vaccine uptake varied based on rurality, age, health insurance status, and education level are consistent with previous studies [6, 15]. However, these studies did not assess the role of food insecurity on COVID-19 vaccine uptake, a significant finding in this study. Food insecurity and income are heavily intertwined because limited financial resources force individuals to prioritize basic needs such as food and housing over healthcare. Individuals who are food insecure often work multiple jobs and are afforded less flexibility to schedule time-off for medical appointments to get vaccinated. Comprehensive efforts are needed to address both food insecurity and healthcare access to ensure that vulnerable populations can access essential services like vaccination. Public health interventions should consider the broader socio-economic context to effectively reach and serve marginalized communities, especially to combat vaccine hesitancy.

The most reported reasons for not getting vaccinated were concerns about potential side effects and safety of the vaccine (69.3% and 67.6%, respectively). These reasons reported are in line with findings from other studies that have found that mistrust and concern over health information can spread quickly by word of mouth within communities [16]. The behavior of fellow community members, particularly within the same age group, is a strong determinant of individual vaccination decisions, which leads to further disparities in vaccination rates across counties [16]. This misinformation can lead to confusion and mistrust in public health directives, making it difficult to combat vaccine hesitancy in these communities.

County-level presidential voting results were also found to be an indicator of COVID-19 vaccine uptake. For the 2020 Presidential election, Republican-voting counties had lower COVID-19 vaccine uptake compared to Democratic-voting counties (88.3% vs 91.6%, p鈥�=鈥�0.001). There is a longstanding relationship between voting patterns and vaccination coverage. Historically, vaccinated individuals in the United States are more likely to be living in a county where the majority voted for the Democratic candidate [17]. This study also looked at previous flu vaccination history among respondents and found that it was an indicator of COVID-19 vaccine uptake which aligns with the longstanding correlation between voting patterns and vaccination coverage in the United States.

Limitations

One limitation of the study was that self-reported vaccine status was used rather than medical record validation, which might be more accurate in assessing vaccination status. As of September 2023, nearly 75% of Ohioans have received at least one dose of COVID-19 vaccine, indicating a heightened vaccination rate (90%) within this sample population [18]. Additionally, the RADX-UP survey data was collected over 18听months during the pandemic, when there were drastic shifts in COVID-19 guidelines and disease control efforts in Ohio. The varying vaccine distribution plans, changing guidelines, and discrepancies in prioritizing different population groups may have decreased COVID-19 vaccine uptake among Ohioans due to confusion regarding eligibility status during this time. Another limitation was the study focused on 12 selected counties in Ohio and is not generalizable to the entire state of Ohio as well as other states with different demographic and socioeconomic characteristics.

This study, however, is the first to explore the relationship between food insecurity and COVID-19 vaccine uptake in Ohio, filling a critical gap in understanding how socioeconomic factors affect health behaviors. Additionally, the finding that there are more essential workers in non-metro counties highlights the demographic disparities in job roles, which can influence access to healthcare and vaccination resources in the state.

Recommendations

Given the high mortality among non-vaccinated individuals in pandemics, interventions should be introduced to non-metro areas to better educate residents about the importance of vaccination, combat the spread of misinformation, and reduce financial and access-related barriers that result in lower vaccine uptake. Specifically, addressing food insecurity and healthcare access for individuals residing in rural areas is important as it will promote more favorable health behaviors, and therefore outcomes. Vaccine hesitancy, which is influenced by misinformation, distrust in healthcare systems, cultural beliefs, and social norms, should also be addressed. CHWs who worked on the IMPACT-Ohio project were instrumental in providing educational information about the efficacy of COVID-19 vaccines and specific details on vaccination sites in the community. Furthermore, close-knit non-metro communities often rely on community ties, which can amplify the influence of vaccine hesitancy among peers [16]. Working with trusted primary care physicians and religious leaders in non-metro communities to implement vaccine education programs is a great place to start.

Conclusions

As this study highlights, non-metro areas experience limited health insurance coverage and lower education levels, which can hinder vaccine acceptance. Tailored interventions should be developed to increase health literacy, provide financial assistance, and implement culturally sensitive communication efforts in non-metro areas. Social determinants of health, including education and economic stability, were strong indicators of COVID-19 vaccination uptake. This underscores the importance of community-based interventions, targeted educational campaigns, mobile vaccination clinics, and efforts to address vaccine hesitancy in the future. Since vaccinations save lives, these efforts should be a priority.