Skip to main content
PMC home page
PLOS Medicine logoLink to PLOS Medicine
. 2022 Jul 28;19(7):e1004069. doi: 10.1371/journal.pmed.1004069

Disparities in distribution of COVID-19 vaccines across US counties: A geographic information system–based cross-sectional study

Inmaculada Hernandez 1,*, Sean Dickson 2, Shangbin Tang 1, Nico Gabriel 1, Lucas A Berenbrok 3, Jingchuan Guo 4
Editor: Elvin Hsing Geng5
PMCID: PMC9333439  PMID: 35901171

Abstract

Background

The US Centers for Disease Control and Prevention has repeatedly called for Coronavirus Disease 2019 (COVID-19) vaccine equity. The objective our study was to measure equity in the early distribution of COVID-19 vaccines to healthcare facilities across the US. Specifically, we tested whether the likelihood of a healthcare facility administering COVID-19 vaccines in May 2021 differed by county-level racial composition and degree of urbanicity.

Methods and findings

The outcome was whether an eligible vaccination facility actually administered COVID-19 vaccines as of May 2021, and was defined by spatially matching locations of eligible and actual COVID-19 vaccine administration locations. The outcome was regressed against county-level measures for racial/ethnic composition, urbanicity, income, social vulnerability index, COVID-19 mortality, 2020 election results, and availability of nontraditional vaccination locations using generalized estimating equations.

Across the US, 61.4% of eligible healthcare facilities and 76.0% of eligible pharmacies provided COVID-19 vaccinations as of May 2021. Facilities in counties with >42.2% non-Hispanic Black population (i.e., > 95th county percentile of Black race composition) were less likely to serve as COVID-19 vaccine administration locations compared to facilities in counties with <12.5% non-Hispanic Black population (i.e., lower than US average), with OR 0.83; 95% CI, 0.70 to 0.98, p = 0.030. Location of a facility in a rural county (OR 0.82; 95% CI, 0.75 to 0.90, p < 0.001, versus metropolitan county) or in a county in the top quintile of COVID-19 mortality (OR 0.83; 95% CI, 0.75 to 0.93, p = 0.001, versus bottom 4 quintiles) was associated with decreased odds of serving as a COVID-19 vaccine administration location.

There was a significant interaction of urbanicity and racial/ethnic composition: In metropolitan counties, facilities in counties with >42.2% non-Hispanic Black population (i.e., >95th county percentile of Black race composition) had 32% (95% CI 14% to 47%, p = 0.001) lower odds of serving as COVID administration facility compared to facilities in counties with below US average Black population. This association between Black composition and odds of a facility serving as vaccine administration facility was not observed in rural or suburban counties. In rural counties, facilities in counties with above US average Hispanic population had 26% (95% CI 11% to 38%, p = 0.002) lower odds of serving as vaccine administration facility compared to facilities in counties with below US average Hispanic population. This association between Hispanic ethnicity and odds of a facility serving as vaccine administration facility was not observed in metropolitan or suburban counties.

Our analyses did not include nontraditional vaccination sites and are based on data as of May 2021, thus they represent the early distribution of COVID-19 vaccines. Our results based on this cross-sectional analysis may not be generalizable to later phases of the COVID-19 vaccine distribution process.

Conclusions

Healthcare facilities in counties with higher Black composition, in rural areas, and in hardest-hit communities were less likely to serve as COVID-19 vaccine administration locations in May 2021. The lower uptake of COVID-19 vaccinations among minority populations and rural areas has been attributed to vaccine hesitancy; however, decreased access to vaccination sites may be an additional overlooked barrier.

Inmaculada Hernandez and colleagues investigate the disparities in early-phase distribution of COVID-19 Vaccines across U.S. Counties.

Author summary

Why was this study done?

  • Equity in the distribution of Coronavirus Disease 2019 (COVID-19) vaccine is of major relevance.

  • It is unknown whether there were differences in the distribution of COVID-19 vaccines to healthcare facilities depending on the demographic composition of the population.

What did the researchers do and find?

  • We tested whether healthcare facilities serving minority or disadvantaged neighborhoods were less likely to administer COVID-19 vaccines in the early phase of the COVID-19 vaccine rollout process.

  • We found that healthcare facilities in counties with higher Black composition, in rural areas, and in hardest-hit communities were less likely to administer COVID-19 vaccines in May 2021.

What do these findings mean?

  • There were disparities in the early distribution of COVID-19 vaccines to healthcare facilities across the country.

Introduction

The equitable and timely distribution of Coronavirus Disease 2019 (COVID-19) vaccines was the public health priority in 2021, after the successful race to develop vaccines against Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) in 2020. In the United States, the distribution of COVID-19 vaccines followed the framework proposed by the National Academies of Science, Engineering, and Medicine, which prioritized in early phases of vaccine distribution healthcare workers and long-term care residents, followed by older adults, frontline workers, those with high-risk conditions, and essential workers [1]. For each population group, the framework called for a prioritization of vulnerable areas that have been disproportionately impacted by the COVID-19 pandemic. Although the US public generally agreed with the prioritization of disadvantaged communities hardest hit by COVID in vaccination [2], most COVID-19 distribution plans were created without input from health equity experts [3].

Consistently across the phases of distribution of COVID-19 vaccines, vaccine uptake has been lower among racial/ethnic minority groups than non-Hispanic White individuals [4–6]. Mistrust, misinformation, and access to COVID-19 vaccines have been identified as the key determinants of uptake of COVID-19 vaccinations among underrepresented groups [7]. However, most discussions on vaccine uptake among minority populations have focused on mistrust and misinformation [7–11]. Soon after the approval of COVID-19 vaccines, we reported that spatial access to healthcare should be a major consideration in ensuring an equitable access to COVID-19 vaccines because underrepresented minorities are less likely to live near healthcare facilities than non-Hispanic White individuals [12]. Rader and colleagues recently reported that COVID-19 vaccine deserts were more likely to be located in rural and lower-income areas [13]. Similar findings on the distribution of vaccine sites were noted by Williams and colleagues for the borough of Brooklyn, New York [14]. In the state of Florida, Kim and colleagues also observed lower accessibility of underserved communities to COVID-19 vaccination sites [15].

These past studies adopted a population focus in comparing the accessibility to COVID-19 vaccination sites across sociodemographic subgroups. Such approach is, however, limited in that it does not differentiate whether lower access in underserved neighborhoods is a product of the lower concentration of healthcare facilities in these areas or of inequities in the distribution of COVID-19 vaccines to facilities. To answer this question, we tested whether the likelihood of an eligible healthcare facility administering COVID-19 vaccines in May 2021 varied with the county-level racial/ethnic composition and degree of urbanicity. This approach enabled us to quantify equity in the early distribution of COVID-19 vaccines to facilities. We hypothesized that healthcare facilities serving rural areas and underrepresented populations would be less likely to administer COVID-19 vaccines.

Methods

Data sources

In a prior analysis, we identified and mapped open-door healthcare facilities eligible for the delivery of COVID-19 vaccinations, including community pharmacies licensed to provide immunizations, federally qualified health centers (FQHCs), rural health clinics (RHCs), and hospital outpatient departments (HODs) [12]. Long-term facilities were not included because they are not available for the vaccination of community dwelling individuals. Nontraditional settings for vaccination such as stadiums or convention centers were not included since they are not healthcare facilities.

Data from community pharmacies were obtained from the National Council for Prescription Drug Programs. Addresses of FQHCs were obtained from the Health Resources and Services Administration [16]. Coordinates of RHCs and addresses of HODs were obtained from the Centers for Medicare and Medicaid Services [17,18]. To identify actual COVID-19 vaccine administration locations, we extracted VaccineFinder data from the Centers for Disease Control and Prevention (CDC) as of May 10, 2021 [19]. These data capture the healthcare facilities involved in the early distribution of COVID-19 vaccines, since the federal government required states to make vaccines available to the general public by May 1, 2021 [20].

Data from community pharmacies were obtained by a restricted license from the National Council for Prescription Drug Programs, but the remaining data sources are publicly available as detailed in the Data Statement. Our study was exempt from human subjects regulation because no human data were used. The study did not have a registered protocol, but it is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 STROBE Checklist). Analyses were performed June to September 2021.

Outcome

The outcome was whether an eligible vaccination facility was registered as actually administering COVID-19 vaccines to the public in May 2021, and was defined by spatially matching locations of eligible and actual COVID-19 vaccine administration locations. Two locations were considered a match if (1) they had the same placekey (a universal standard identifier for any physical location); or (2) they were closest in proximity to each other after geocoding, and their addresses and names matched. Geocoding was performed using ArcGIS, version 10.7 (Esri).

Independent variables

Independent variables included racial/ethnic composition, urbanicity, median household income, social vulnerability index, COVID-19 mortality rate, 2020 presidential election results, and availability of other COVID-19 vaccine administration locations. Independent variables were defined at the county level as opposed to the census tract level because first, counties are the political unit with most control over local vaccine distribution, and second, service areas of HODs, FQHCs, and RHCs span multiple census tracts.

County-level measures of racial/ethnic composition included proportion non-Hispanic Black and proportion Hispanic and were obtained from the University of Wisconsin Population Health Institute County Rankings and Roadmap data [21]. Counties were categorized in 3 levels based on racial/ethnic composition: (1) below US average; (2) between US average and 95th county percentile; and (3) above 95th county percentile, as previously done by CDC [22].

Urbanicity was categorized in 3 levels using US Department of Agriculture Rural–Urban Continuum Codes (RUCC): (1) metropolitan (RUCC code 1, accounts for 168.5 million people); (2) suburban (RUCC codes 2 to 3, accounts for 93.9 million people); and (3) rural (RUCC codes 4 to 9, accounts for 46.3 million people) [23].

Social vulnerability index was obtained from the CDC and represents the degree to which a community exhibits certain conditions that affect its ability to prevent human suffering and financial loss in the event of a disaster [24]. The social vulnerability index includes 4 components: socioeconomic, household composition and disability, minority status and language, and housing type and transportation. We created indicator variables for counties in the top quintile of the socioeconomic, household composition and disability, and housing type and transportation components. These indicator variables represent most vulnerable communities. We did not include a variable for the minority status and language component of the social vulnerability index because minority status is already captured by the measures of racial/ethnic composition defined above.

COVID-19 mortality was obtained as of May 10, 2021, from the John Hopkins University Coronavirus Resource Center [25]. We created an indicator variable for counties in the top quintile of COVID-19 mortality, which represent communities that were hardest-hit by COVID-19 prior to the date of our analysis.

Results of the 2020 election were obtained from the MIT Election Data and Science Lab [26]. We created an indicator variable representing the Presidential candidate that won the 2020 election in each county. This variable was included in analysis to capture the majority political affiliation of the area served by a facility.

Healthcare facilities in counties relying on nontraditional settings for vaccination such as stadiums or convention centers may have been less likely to administer vaccines. To account for this, we adjusted analyses for the density of alternative vaccination locations. Alternative vaccine administration locations were defined as vaccination locations reported by CDC in VaccineFinder data that did not match to pharmacies, HODs, RHCs, or FQHCs. Finally, a healthcare facility may have been less likely to administer COVID-19 vaccines if there was another COVID-19 vaccination location nearby. Thus, we adjusted for the availability of a COVID vaccine administration facility within 500 m of a given facility.

Statistical analyses

The outcome—whether an eligible vaccination facility was registered as actually administering COVID-19 vaccines in May 2021—was regressed against covariates listed above using logistic regression. All independent variables specified above were included in the final model. We applied generalized estimating equations to account for clustering of facilities within a county. We tested interactions between racial/ethnic composition and urbanicity. We constructed 1 set of analyses for all 4 types of healthcare facilities and a second set for pharmacies because some FQHCs, RHCs, and HODs only offered COVID-19 vaccines to registered patients. Analyses were conducted using SAS 9.4 (Cary, North Carolina) and Stata 17 (College Station, Texas).

Results

The sample included 50,806 community pharmacies, 11,619 FQHCs, 3,187 HOPDs, and 1,255 RHCs distributed across 2,942 counties in US states (territories were not included). Across the US, 61.4% of eligible healthcare facilities and 76.0% of eligible pharmacies provided COVID-19 vaccinations in May 2021 (Table 1).

Table 1. Proportion of facilities serving as COVID-19 vaccine administration locations.

% Served as COVID-19 Vaccine Administration Location
Variable—n(%) Pharmacies (n = 50,806) All Healthcare Facilitiesa (n = 66,867)
All 76.0% 61.4%
County-Level Proportion of Black Populationb
     Facility in County with Proportion Black Population <12.5% 76.3% 60.5%
     Facility in County with Proportion Black Population 12.5%–42.2% 75.9% 63.9%
     Facility in County with Proportion Black Population >42.2% 71.8% 55.5%
County-Level Proportion of Hispanic Populationc
     Facility in County with Proportion Hispanic Population <18.5% 75.8% 62.2%
     Facility in County with Proportion Hispanic Population 18.5%–38.7% 76.4% 61.5%
     Facility in County with Proportion Hispanic Population >38.7% 76.4% 56.4%
Urbanicityd
     Metropolitan 76.4% 64.4%
     Suburban 77.4% 62.7%
     Rural 71.7% 51.1%
Facility in County in Bottom Quintile for Median Income 71.8% 45.5%
Facility in County in Top Quintile for Vulnerability Index—Socioeconomic Component 73.5% 48.8%
Facility in County in Top Quintile for Vulnerability Index—Household Composition and Disability 72.6% 51.2%
Facility in County in Top Quintile for Vulnerability Index -Housing Type and Transportation 74.5% 55.8%
Facility in County in Top Quintile for COVID Mortality 71.4% 54.6%
Facility in County where Trump Won the 2020 presidential election 75.3% 60.0%
Open in a new tab

aIncluded pharmacies, FQHCs, RHCs, and HODs.

bThe proportion of Black population at the county level was obtained from the University of Wisconsin Population Health Institute County Rankings and Roadmap data[21] and was categorized in 3 levels, following prior methodology used by CDC [22]: (1) below US average (12.5%); (2) between US average and 95th county percentile (42.2%); and (3) above 95th county percentile (42.2%).

cThe proportion of Hispanic population at the county level was obtained from the University of Wisconsin Population Health Institute County Rankings and Roadmap data [21] and was categorized in 3 levels, following prior methodology used by CDC [22]: (1) below US average (18.5%); (2) between US average and 95th county percentile (38.7%); and (3) above 95th county percentile (38.4%).

dUrbanicity was categorized in 3 levels using US Department of Agriculture RUCC: (1) metropolitan (RUCC code 1); (2) suburban (RUCC codes 2–3); and (3) rural (RUCC codes 4–9).

CDC, Centers for Disease Control and Prevention; COVID-19, Coronavirus Disease 2019; FQHC, federally qualified health center; HOD, hospital outpatient department; RHC, rural health clinic; RUCC, Rural–Urban Continuum Codes.

In the early phase of the vaccine rollout process, healthcare facilities were less likely to serve as COVID-19 vaccine administration locations when located in counties with >42.2% Black population (i.e., > 95th county percentile of non-Hispanic Black race composition) compared to counties with <12.5% non-Hispanic Black population (i.e., lower than US average), with OR 0.83; 95% CI, 0.70 to 0.98, p = 0.030 (Fig 1). Location of a facility in a rural county (OR 0.82; 95% CI, 0.75 to 0.90, p < 0.001, versus metropolitan county) or in a county in the top quintile of COVID-19 mortality (OR 0.83; 95% CI, 0.75 to 0.93, p = 0.001, versus bottom 4 quintiles) was associated with decreased odds of serving as a COVID-19 vaccine administration location. There were no differences in the likelihood of facilities serving as COVID-19 vaccine administration locations by voting results in the 2020 Presidential election (OR 0.96; 95% CI, 0.89 to 1.03, p = 0.240 for counties with majority vote for Trump, compared to counties with majority vote for Biden). Results were similar for analyses conducted for community pharmacies.

Fig 1. Adjusted odds ratios of facilities serving as COVID-19 vaccine administration locations, main effects.

Fig 1

Open in a new tab

The figure shows the results of logistic regression models fitted with generalized estimating equations for the primary outcome of a healthcare facility (or a pharmacy) serving as a COVID-19 vaccine administration location. The model only included main effects. All healthcare facilities included pharmacies, FQHCs, RHCs, and HODs. The circles represent the point estimate for the odds ratio, and the whiskers represent the 95% confidence interval. COVID-19, Coronavirus Disease 2019; FQHC, federally qualified health center; HOD, hospital outpatient department; RHC, rural health clinic.

We identified a significant interaction of urbanicity with non-Hispanic Black race composition: In metropolitan counties, facilities in counties with >42.2% non-Hispanic Black population (i.e., > 95th county percentile of Black race composition) had 32% (95% CI 14% to 47%, p = 0.001) lower odds of serving as COVID administration facility compared to facilities in counties with below US average Black population (Fig 2). This disparity was not observed in rural or suburban counties. Moreover, we detected a gradient for pharmacies: Pharmacies in metropolitan counties with above-average non-Hispanic Black composition had 12% (95% CI, 1% to 20%, p = 0.030) lower odds of serving as COVID-19 vaccine administration locations than pharmacies in metropolitan counties with below-average Black composition; in counties above the top 95th percentile of non-Hispanic Black composition, odds were 30% (95% CI, 11% to 41%, p = 0.003) lower.

Fig 2. Adjusted odds ratios of facilities serving as COVID-19 vaccine administration locations, interaction for proportion non-Hispanic Black population and urbanicity.

Fig 2

Open in a new tab

The figure shows the results of logistic regression models fitted with generalized estimating equations for the primary outcome of a healthcare facility (or a pharmacy) serving as a COVID-19 vaccine administration location. All healthcare facilities included pharmacies, FQHCs, RHCs, and HODs. The model adjusted for all covariates listed in Fig 1. Additionally, the model constructed for all healthcare facilities included an indicator variable for facility type (pharmacy vs. others). The circles represent the point estimate for the odds ratio, and the whiskers represent the 95% confidence interval. COVID-19, Coronavirus Disease 2019; FQHC, federally qualified health center; HOD, hospital outpatient department; RHC, rural health clinic.

We also identified a significant interaction of urbanicity with Hispanic composition: In rural counties with above-average Hispanic composition, healthcare facilities had 26% (95% CI, 11% to 38%, p = 0.002) lower odds of serving as COVID-19 vaccine administration locations compared to facilities in counties with below-average Hispanic composition (Fig 3). However, this disparity was not observed in metropolitan or suburban counties.

Fig 3. Adjusted odds ratios of facilities serving as COVID-19 vaccine administration locations, interaction for proportion Hispanic population and urbanicity.

Fig 3

Open in a new tab

The figure shows the results of logistic regression models fitted with generalized estimating equations for the primary outcome of a healthcare facility (or a pharmacy) serving as a COVID-19 vaccine administration location. All healthcare facilities included pharmacies, FQHCs, RHCs, and HODs. The model adjusted for all covariates listed in Fig 1. Additionally, the model constructed for all healthcare facilities included an indicator variable for facility type (pharmacy vs. others). The circles represent the point estimate for the odds ratio, and the whiskers represent the 95% confidence interval. COVID-19, Coronavirus Disease 2019; FQHC, federally qualified health center; HOD, hospital outpatient department; RHC, rural health clinic.

Discussion

To our knowledge, we present the first nationwide study to quantify disparities in the early distribution of COVID-19 vaccines to healthcare facilities in the US. Our study demonstrates that healthcare facilities in counties with higher non-Hispanic Black composition, in rural areas, and in hardest-hit communities were less likely to serve as COVID-19 vaccine administration locations in May 2021. We observed, however, significant interactions between urbanicity and demographic composition: The county-level proportion of Black population was associated with decreased odds of a facility administering COVID-19 vaccines in metropolitan, but not in suburban or rural counties. The county-level proportion of Hispanic population was associated with decreased odds of a facility administering COVID-19 vaccines in rural, but not in metropolitan or suburban counties. Our findings based on May 2021 data represent the early distribution of COVID-19 vaccines and may not be generalizable to later phases of the COVID-19 vaccine rollout process.

A study by Kim and colleagues examined disparities in access to COVID-19 facilities in Florida and found that racial/ethnic disparities in access differed across rural and urban counties [15]. Consistent with our findings, this study reported that, compared to non-Hispanic White individuals, Black individuals had lower access to COVID-19 vaccine administration sites in urban counties; however, disparities in access for the Hispanic population were concentrated in rural counties [15]. Our findings are also consistent with the results by Rader and colleagues, who found that COVID-19 vaccine deserts are more likely to be located in rural and low-income areas [13]. Nevertheless, our analysis is a major contribution to the existing literature because, to our knowledge, it is the first to analyze equity in the distribution of vaccines to healthcare facilities across the nation. This is important because when studies evaluate population access to COVID-19 administration facilities, it is not possible to differentiate whether lower access in rural or underserved areas is a product of the lower concentration of healthcare facilities or of inequities in the distribution of COVID-19 vaccines to facilities. To answer this question, we tested instead whether the likelihood of healthcare facilities serving as COVID-19 administration sites in May 2021 differed with the sociodemographic composition of the population served. In doing so, we demonstrate that facilities in counties with higher non-Hispanic Black composition, in rural areas, and in hardest-hit communities were less likely to serve as COVID-19 vaccine administration locations in the early phase of the vaccine rollout process. In other words, we demonstrate that underrepresented, underserved, and rural areas have lower access to COVID-19 vaccines not only because of the lower concentration of healthcare facilities, but also because the facilities serving these areas were less likely to administer COVID-19 vaccines, at least in the early phase of the vaccine distribution process.

Our analyses were conducted at the facility level, but covariates were defined at the county level for 2 reasons: First, counties are the political unit with most control over local vaccine distribution; second, facility service areas can span hundreds of census tracts [27]. Our sample did not include nontraditional vaccination sites because our objective was to measure the likelihood of existing healthcare facilities serving as COVID-19 vaccine administration locations. Nevertheless, we adjusted for the density of alternative vaccination locations. In other words, we accounted for the fact that healthcare facilities in counties heavily relying on nontraditional settings may have been less likely to administer COVID-19 vaccines. Our evaluation of racial/ethnic disparities focused on non-Hispanic Black and Hispanic individuals due to the low proportion of other minority groups in most counties, which yielded unstable estimates.

An equitable distribution of vaccines would imply that vaccines were distributed in a timely manner across all individuals. This is the reason why we used data from May 2021, when every US adult became eligible for vaccination. The association between vaccine distribution and sociodemographics of the population served by facilities could have changed over time. In other words, our findings based on May 2021 data may not be generalizable to later phases of the COVID-19 vaccine rollout process. Nevertheless, our findings based on the initial rollout process are of prominent relevance, because analyses based on more recent data could mask differences in the timing of vaccine distribution across facilities.

Our nationwide quantitative evaluation of health equity in the early distribution of COVID-19 vaccines is limited by the data available. We operationalized the definition of health inequity as differences in the likelihood of a healthcare facility administering vaccines associated with the sociodemographics of the population served. This binary outcome does not account for important variables not available in the data such as the number of vaccines distributed in each healthcare facility, the volume of vaccines distributed in other facilities within the county, or the area of service of each facility. One could argue that the likelihood of a facility administering vaccines is not relevant because an area could be well covered by a smaller fraction of facilities, if such facilities provided enough vaccines to cover the population and were accessible enough. Nevertheless, regardless of whether a smaller share of facilities is able to adequately cover the population through higher volume, the variation in the proportion of available healthcare facilities used associated with the sociodemographic characteristics of the population suggests structural inequities in the design of the early COVID-19 vaccine rollout. The differential use of available healthcare infrastructure by the demographics of the population served may suggest prioritization of convenience for some residents rather than the actual distribution volume needed.

Four additional limitations of our analyses are worth noting. First, locations not registered in VaccineFinder were not identified, which could have led to an underestimation of the proportion of healthcare facilities serving as vaccine administration locations. Second, we did not evaluate variation in the types of vaccines distributed to each healthcare facility. Third, the data compiled only contained information on the addresses of vaccine administration sites, so it was impossible to define whether institutional or structural reasons led some facilities to not become COVID-19 administration sites. Finally, our data and analyses were at the facility level rather than at the person level, so we were not able to assess individual preferences in accessing healthcare facilities for vaccination. Our study, however, presents important strengths, including the use of spatial matching methods and the national coverage of the results. The innovative approach used to measure health equity should also be noted, since we used facilities as observations, instead of individuals or census tracts.

The discussion on the lower uptake of COVID-19 vaccines among racial/ethnic minority groups has mostly focused on mistrust and misinformation [7–11]. However, our analysis suggests that systematic barriers play an important role in differential rates of COVID-19 vaccination across racial/ethnic groups, which are often omitted in conversations around mistrust. In addition to the lower concentration of healthcare facilities in underserved and rural areas [12], the facilities that serve these vulnerable populations were less likely to administer COVID-19 vaccines in the early phase of the vaccine rollout process, even when the framework proposed by the National Academies of Science, Engineering, and Medicine called for a prioritization of these areas precisely [1]. In tandem with community engagement efforts to address vaccine hesitancy and interventions to improve spatial access to vaccines, public health authorities should review COVID-19 distribution plans to identify the reasons why these processes resulted in an inequitable distribution of COVID-19 vaccines in early 2021. Identifying the drivers of the lower involvement of facilities serving underserved and rural areas in the early distribution of the COVID-19 vaccine is crucial to improve equity in the distribution of booster shots and of future public health prevention programs.

We evaluated whether the likelihood of healthcare facilities serving as COVID-19 vaccine administration sites in the early distribution of COVID-19 vaccines varied with the sociodemographic composition of the population served. We found that healthcare facilities in counties with higher Black composition, in rural areas, and in hardest-hit communities were less likely to serve as COVID-19 vaccine administration locations in the early phase of the vaccine rollout process.

Supporting information

S1 STROBE Checklist. STROBE checklist for cross-sectional study.

(DOC)

Click here for additional data file. (83KB, doc)

Acknowledgments

We thank the Data Science team at the West Health Institute for their help extracting VaccineFinder data.

Abbreviations

CDC

Centers for Disease Control and Prevention

COVID-19

Coronavirus Disease 2019

FQHC

federally qualified health center

HOD

hospital outpatient department

RHC

rural health clinic

RUCC

Rural–Urban Continuum Codes

SARS-CoV-2

Severe Acute Respiratory Syndrome Coronavirus 2

Data Availability

Data from community pharmacies were obtained from the National Council for Prescription Drug Programs under a license that does not allow for data sharing. This is the reason why the data cannot be shared publicly without restrictions. Requests for community pharmacy data should be addressed to the National Council for Prescription Drug Programs (http://dataq.ncpdp.org/) The remaining data sources are publically available: Addresses of federally qualified health centers are available from the Health Resources and Services Administration website: https://data.hrsa.gov/data/reports/datagrid?gridName=FQHCs Coordinates of rural health clinics are available from the Centers for Medicare and Medicaid Services website https://www.cms.gov/Research-Statistics-Data-and-Systems/Downloadable-Public-Use-Files/Cost-Reports/Rural-Health-Center-222-2017-form Addresses of hospital outpatient departments are also available from the Centers for Medicare and Medicaid Services website https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Downloads/Outpatient_Data_2017_XLSX.zip Addresses of COVID-19 vaccine administration locations are available from the Centers for Disease Control and Prevention website https://www.cdc.gov/vaccines/covid-19/reporting/vaccinefinder/about.html.

Funding Statement

The author(s) received no specific funding for this work.

References

Decision Letter 0

Beryne Odeny

27 Oct 2021

Dear Dr Hernandez,

Thank you for submitting your manuscript entitled "Disparities in Distribution of COVID-19 Vaccines across U.S. Counties" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by Oct 29 2021 11:59PM.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Beryne Odeny

PLOS Medicine

Decision Letter 1

Beryne Odeny

17 Mar 2022

Dear Dr. Hernandez,

Thank you very much for submitting your manuscript "Disparities in Distribution of COVID-19 Vaccines across U.S. Counties" (PMEDICINE-D-21-04494R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Apr 07 2022 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Beryne Odeny,

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

1) Please revise your title according to PLOS Medicine's style. Your title must be nondeclarative and not a question. It should begin with main concept if possible. Please place the study design in the subtitle (i.e., after a colon), e.g., a retrospective cohort study.

2) Please include line numbers in your next draft.

3) Is there a chance you can obtain more recent data from this setting?

4) Abstract:

a) Please ensure that all numbers presented in the abstract are present and identical to numbers presented in the main manuscript text.

b) Please quantify the main results (please present both 95% CIs and p values).

c) In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

5) At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

6) Introduction: Please address and cite past research and explain the need for and potential importance of your study. Indicate whether your study is novel and how you determined that. If there has been a systematic review of the evidence related to your study (or you have conducted one), please refer to and reference that review and indicate whether it supports the need for your study.

7) Please conclude the “Introduction” with a clear description of the study question or hypothesis. The description of the database can be moved to the methods section

8) Did your study have a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

9) Please temper claims of primacy of results (e.g. “In the first nation-wide study…)” by stating, "to our knowledge" or something similar.

10) Please ensure that the study is reported according to the STROBE and include the completed STROBE checklist as Supporting Information. Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)."

a) When completing the checklist, please use section and paragraph numbers, rather than page numbers.

11) How was race/ethnicity defined and by whom?

12) You examined (as a binary outcome, yes/no) whether a facility serves as a covid-19 vaccination site; is there any data on the number of vaccines actually delivered?

13) Do you have any insights into the reasons why particular sites might be signed up as vaccination sites or not

14) Please discuss your analyses based on a hypothetical “Trump win” in 2020. For example, add a sentence to the effect that there were “no differences by voting preferences in the 2020 Presidential election”, or similar, assuming this is the case.

15) Please provide p values in addition to 95% CIs in the main text and tables

16) Please indicate in the figure caption the meaning of the bars and whiskers in Figure 1.

17) Please present and organize the Discussion as follows: a short, clear summary of the article's findings; what the study adds to existing research and where and why the results may differ from previous research; include relevant citations; strengths and limitations of the study; implications and next steps for research, clinical practice, and/or public policy; one-paragraph conclusion.

18) References:

a) Please provide access dates for all references with a weblink

b) Please select the PLOS Medicine reference style in your citation manager. In-text reference call outs should be presented as follows noting the absence of spaces within the square brackets,

c) Please ensure that journal name abbreviations consistently match those found in the National Center for Biotechnology Information (NCBI) databases. https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references.

Notes from the Academic Editor:

The message and finding are important – that access is an issue.

The major limitation is the ecological analysis, and I would think important to ensure you have addressed things like Social Vulnerability Index (SVI). Adjustment for SVI may attenuate the race-access relationship, but I think it would not necessarily undermine the argument – it could be a mediator of the relationship with racialization.

Comments from the reviewers:

Reviewer #1: The authors have applied geographic information systems methods to test whether the likelihood of an eligible healthcare facility administering COVID-19 vaccines varied with the county-level racial composition and degree of urbanicity.

Comments:

"The outcome was regressed against county-level measures for racial/ethnic composition, urbanicity, income, COVID-19 mortality, 2020 election results, and availability of non-traditional vaccination locations using generalized estimating equations."

Did the authors consider including covariates in the models in order to adjust for potential confounding, such as county- or facility- level age distribution, comorbidity prevalence, and gender ratio?

"The outcome -whether an eligible vaccination facility was registered as actually administering COVID19 vaccines—was regressed against covariates listed above using logistic regression. We applied generalized estimating equations to account for clustering of facilities within a county."

The authors have applied a technically appropriate modelling technique for the data and research question in hand.

"We tested interactions between racial/ethnic composition and urbanicity. Because some FQHCs, RHCs, and HODs only offered COVID-19 vaccines to registered patients, we constructed one set of analyses for all four types of healthcare facilities and a second set for pharmacies only."

The authors have undertaken several additional analyses that help to demonstrate the robustness of the study findings.

Can the authors please present statistical findings (i.e. exact p-values and CIs) for all stated study inferences throughout the Results text?

Did the authors consider exploring 'County-Level Proportion of Black Population' and 'County with Proportion Hispanic Population' as continuous variables within the analysis?

Reviewer #2: I enjoyed reading this manuscript and I believe it is worthy of acceptance and publication in PLOS Medicine. I liked how the authors side-stepped the limitations regarding what data the federal government has regarding reported vaccination administrations with regards to race and ethnicity. However, before publication, I believe the manuscript can be strengthened. Most, if not all, of these suggestions should not require reanalysis, but may require some reorganization and a few more sentences. I think there is enough to call it a major revision, but hopefully not one that is a heavy lift:

The authors state their goal is to measure health equity in the "actual distribution of COVID-19 vaccines." If that is the ultimate goal, then the portion that is pharmacies or alternate facilities is less important than the overall totality of vaccination sites and how those impact minorities. If the goal is also to highlight that, after COVID-19 ends, there may be pullback and these variations in how vaccines are administered will be more important for 'normal' vaccinations, this could be stated more clearly. I suggest this is a secondary focus because of a sentence in the discussion that states the objective was to measure the likelihood of existing healthcare facilities providing vaccinations. This objective is different from the "actual distribution of vaccines" and makes the later discussion of improved geographic access to vaccination fall somewhat flat. For example, yes, there is a continuum for pharmacies for the Black population (Figure 1a), but for whatever reason this only holds significantly for the greater than 42.7% group for all health care facilities (and here that confidence interval is a 'squeaker.') The same issue for pharmacies versus all four types of facilities does not seem to exist for either the Rural or the COVID mortality. Pragmatically, if the only goal is to get COVID vaccines into arms, then who is delivering those vaccines does not matter. If you want to draw conclusions about what it looked like at the beginning of the pandemic, what the current coverage says about how the system is responding to inequities (and here you'd need to spell out why a pharmacy vaccination has a different connotation than a federal facility, etc.), or how vaccinations will look after, then who is giving the vaccines matters more.

Continuing on this theme, the authors do acknowledge non-traditional vaccine distribution sites and availability of other vaccine distribution locations, but as a reader I don't think I fully understood the implications in the text. Here, I'm trying to highlight reader confusion, rather than give exact instructions on how to address that confusion. What do these variable these mean along the rural to urban spectrum. What does 500m mean in a rural location? Is there some kind of relationship for rural locations having non-traditional sites? This is not primary to the manuscript, but I didn't feel comfortable that I understood what these variables really meant. They appear again in Figure 1A. Since I don't know what they mean, when I read their odds ratios, I am guessing that having them in a county means more people are getting vaccinated than one would expect? Does this imply they are likely geographic overcompensation? Does the philosophy of vaccine distribution sites differ for low density rural areas, where one would expect long drives to pretty much everywhere? Or does it not differ because you can't expect everyone to have cars?

Another issue that should be addressed, if not accounted for possibly, is that CDC and Census report "Black" as a race and "Hispanic" as an ethnicity. This means that Hispanic includes Hispanic Blacks and Black includes Hispanic Blacks. I am guessing that the authors ignored this nuance because this group of people is not large compared to the total for either Black or Hispanic. Or because, for example, Florida doesn't report its data like Census, etc. This Race/Ethnicity detail should probably be acknowledged before being ignored (caveat: this is a pet peeve).

The results, conclusions, and Figure 1 should also focus on Hispanics, as they are mentioned in the text. There should be a paragraph in results about the Hispanics that mirrors that for Blacks, at least a sentence in the conclusions, and probably a 1C section for Hispanic data that mirrors 1B. My assumption is that the data showed no significantly different odds ratios for Hispanics, so all of this was stricken for not showing a relationship. If so, the data cannot reject lack of a health equity issue (null hypothesis). If one goes looking for a health equity issue for Hispanics and there isn't a visible relationship that supports one's assumptions for Hispanics, that should probably be explicitly stated or there is a risk of appearing to "cherry pick" to meet expectations coming into the scholarship. It might also be worth noting why other groups were not considered (e.g. Non-Hispanic Asians, which might flip the other direction). I'm assuming it was an issue of sample size, rather than an a priori assumption that only Blacks and Hispanics would be significant.

This dovetails into the bigger point regarding the Discussion. The implications of Figure 1B should guide more of the conclusions of the manuscript. I think the existing first two sentences at the start of the discussion are fine. But it should be spelled out that, when combining Black composition and urbanicity, only >42.5 was significant. The existing sentence points out there is a continuum for metropolitan areas, but it does not indicate that neither Suburban nor Rural were significantly different for Blacks. Nor does it note, apparently, that the authors could not find such a relationship for Hispanics. This should be noted, but the authors then have the opportunity to explain why a relationship was not found (brainstorming rather than statements of fact: data limitations, underreporting of Hispanics in Rural areas that use migrant workers, Blacks really are the special case here and being Hispanic/Latino doesn't matter, there's a difference between being Hispanic in some counties rather than others—kinds of work, illegal vs. citizen, etc.—that is obscuring a real relationship). The significant mortality odds ratio is almost self-explanatory (less people vaccinated means more die), but this is not explicitly stated. Nor is the meaning of the non-traditional vaccine distribution sites and availability of other vaccine distribution locations. Are non-traditional sites appearing in counties that otherwise would be 'vaccine deserts' and therefore indicate that decision makers realized they had a problem in these locations and shored up vaccine coverage? Does the less than 500m calculation matter at all in rural locations? All of this also feeds into your abstract conclusion, which could be more specific.

Less pressing suggestions:

Please include line numbers on the next draft. I can't give you specific feedback about typos, etc, because it is hard to direct you.

Please provide the reader with a stage setting of what kinds of vaccine sites are seen nationally and then dive into the data you have for those sites. For example, what about pop-up clinics, what about vaccines at long term care facilities? Indicate to me that you have considered/know about all of the options and how they fall into your categories.

PLOS Medicine may want outcomes first in the abstract and methods (I am guessing), but if not, it would be better to set the stage before going straight to them.

Was it really defined as "indicator of Trump Win?"

The second paragraph of the results could be more explicit that it is discussing 'all 4 facility types', if that is what it is doing.

You might want to give more basic numbers (ideas rather than exact guidance follows): How many counties were you looking at? if there are some you had to remove, state that or throw up a map showing your data gaps to assure the reader that you are being close to comprehensive. Are you just looking at the US States? What about USVI, Puerto Rico, American Samoa, etc.? How many Blacks, how many Hispanics?

Other potential data issues that might be worth addressing: I can't think of exactly how it would affect your data overall, but people jump counties for vaccines (go into a city for work, the clinic is one county over, but close to home, etc.). Also, people jump channels (pharmacy to federal facility, etc.). Maybe this noise is self-limiting in your data, I don't know if this is worth addressing or not. If you have taken data from the beginning until now, do you need to account for individuals getting up to three doses at this point or successively wider age ranges for vaccination? Is there some kind of significant relationship for the J and J one shot vaccine going into a particular population or urbanicity that skews your results?

This is probably beyond your scope, but it would be interesting in a figure showing vaccines over time for Blacks and Hispanics. As COVID-19 has gone on, has the country done a better job at addressing inequities at the county level? Is there a way to display that?

Reviewer #3:

Disparities in Distribution of COVID-19 vaccines across US Counties

Hernandez et al.

The authors present a study of COVID-19 vaccination sites, with the goal of evaluating fairness of vaccine access.

Major comments

Introduction

Please include a summary of or at least references to one of the many frameworks that were created prior to COVID-19 vaccine availability to support equitable distribution of vaccines, e.g., PMID: 3237895. In the discussion section, consider commenting upon our success at following the goals laid out in one or more of these frameworks. Additionally, it's important to recall that the US allowed COVID-19 vaccine eligibility in phases; this should be mentioned in the introduction as it likely affected placement of vaccine sites.

The authors state, "to our knowledge, no nationwide studies have measured health equity in the actual distribution of COVID-19 vaccines to healthcare facilities;" however, studies taking a different approach to evaluation of health equity in vaccine distribution have been published and are easily accessible, e.g., PMID: 34213561. The authors should consider a more thorough literature review to provide the reader with a succinct summary of the work that has already been done on health equity and vaccine distribution in the US.

Methods and Results

As this study only evaluated vaccination sites in place as of May 2021, it's important to consider how eligibility for and availability of vaccines changed over time (https://www.cdc.gov/vaccines/imz-managers/downloads/Covid-19-Vaccination-Program-Interim_Playbook.pdf, see page 12), and to note that this varied by county. If possible, it would be helpful for the authors to compare their data to similar data later in 2021 or in early 2022. If this is not possible, the study should be clearly characterized as an early evaluation of vaccine sites, e.g., in the first sentence of the Results section should read, "…provided COVID-19 vaccinations as of May 2021."

Consider showing the outcome(s)/independent variables stratified by the population types that were eligible for vaccination during their study period, which ended in May 2021 (therefore, only 3-4 months after COVID-19 vaccines were available to only certain portions of the population [e.g., healthcare workers who likely were vaccinated at their places of employment, and the elderly, who were likely vaccinated at their physician's office). In other words, during the time when vaccines were only available for healthcare workers and people >65, were vaccines easily accessible to the corresponding elderly portion of the population?

Additionally, it would be helpful for the authors to comment upon the dates of their independent variable data. E.g., did they obtain the COVID-19 mortality data from the Johns Hopkins dashboard in or before May 2021 so that it temporally corresponded with their vaccine site data?

The table suggests that proportions of vaccine sites available to counties with varying levels of Black/Hispanic population did not vary significantly (e.g., for counties with black population <12.5%, 12.5%-42.2%, and >42.2%, the percent of healthcare facilities that served as vaccine sites was 60.5%, 63.9%, and 55.5%, respectively. It would be helpful for the authors to include a statistical comparison of the differences in these proportions and comment upon any significant findings, rather than merely mentioning the proportions of select groups in the Results section.

Discussion

Again, the authors' results should be characterized as "early" if they are unable to show later analyses for comparison. E.g., in the first sentence of the Discussion, they should write "...disparities in the early distribution…"

Consider offering suggestions as to why the findings may be true. E.g., what barriers challenged availability of COVID-19 vaccination centers in the "hardest hit" communities? Were early COVID-19 vaccination sites targeted at locations with concentrated populations of people >65yo regardless of race because only people >65yo were eligible for vaccination at the time of the study?

It doesn't make sense to evaluate availability/accessibility of vaccine sites when most of the county population was not eligible for vaccination. For most of the study period, which ended in May 2021, only a relatively small proportion of the population was eligible in most counties (e.g., healthcare workers, the elderly, people with certain medical conditions). I think it's important that the authors discuss (and hopefully address by updating their data/analysis!) these limitations in the discussion section. In the last paragraph of the Discussion they provide a call to action that may not be necessary if vaccine equity did in fact improve after the study period. Ideally the authors would apply the same methods to evaluate health equity at vaccine sites at a later period as well, e.g., in late 2021 and/or early 2022.

Reviewer #4: OVERALL COMMENTS:

This is an important and well-written study that examined the equity in vaccine districubtion across health care facilities and pharmacies in the US. The study examined equity in the distribution of COVID-19 vaccines across healthcare facilities and pharmacies in the US. They found the distribution of vaccines differed by geographic and racial composition.

INTRODUCTION:

The introduction is too brief and does not fully expand on the scope and significance of the problem. Even a descriptive report on who received COVID-19 vaccines to date would be helpful. There had been a recent study on public opinion of vaccine distribution prioritization in 2021 that may be useful to cite.

Would also be helpful to state why measuring health care facilities is important beyond who received the vaccines. You state this later in the manuscript, but this is important to describe in the introduction.

In addition, the paper is focused on the health facilities that were registered to distribute. It may also be helpful to briefly mention the barriers to the health care facilities getting registered and how this may exacerbate disparities?

METHODS:

Under data sources:

Please operationalize vaccine distribution equity, this will help ground the results.

Please define the acronyms for HOD and RHC.

The authors indicate the study was exempt, was this determined by and IRB? If so, please state that.

Under independent variables:

It is not clear why election results were included. A sentence or two to explain this was included.

Can you provide an example of an alternative vaccine distribution location? The authors define what they were not (HOD, RHC and FQHCS), but not what they were.

RESULTS:

In the figure, a key for the shapes (triangle, circle, square) would be helpful.

DISCUSSION:

In the first sentence of the Discussion section. I believe it should read "…quantify disparities in the distribution locations of COVID-19 vaccines, …".

The interpretation of the interaction terms should be mentioned.

The data is limited to the first quarter of vaccine distribution to the general public (May 10, 2021). This should be discussed in the limitations. Distribution may have improved over time.

Also worth mentioning the role of preference for where to access, resources in rural and safety-net locations. A location to distribute and resources to distribute are both important.

The odds ratios are helpful but not sufficient. It is not clear what the impact is. Ground the results in how many lives are impacted by the odds.

Reviewer #5: Overall, this article poses an interesting research question but lacks details and justification about their Methods which would be used to judge their Results. Specific comments include:

The Introduction is very sparse on why the variables they chose to investigate are important and relevant. This should be built out more.

This sentence need a citation: "County-level measures of racial/ethnic composition included proportion Black and proportion Hispanic, and were categorized in three levels following CDC methodology:(6) 1) below US average; 2) between US average and 95th county percentile; 3) above 95th county percentile."

Authors do not include their variable selection methodology for the final logistic regression model.

Authors do not include counts of facilities, etc. in the Results section write-up. This is important for readers to understand how big these sample sizes are and to determine if the Methods are sound.

There are no p-values in the text - unless excluding them is part of the journal style, this is essential to include.

What proportion of the U.S population falls into metropolitan vs. suburban vs. rural counties? This is an important comparator.

Reviewer #6: General impression: This is a brief report of findings on variability in distribution of vaccinates from eligible to distribution centers/pharmacies. The written presentation is clear and presents the data well. The data presentation is difficult to interpret without the scale (how many pharmacies and facilities were examined and considered eligible). The ability to post the publicly available data on the number of eligible pharmacies and their distribution would strengthen the ability to replicate the findings.

Abstract: The abstract is written clearly and coveys the key findings well. One point that is not clear is the interpretation of the significant interaction between urbanicity and Black area-level composition - the interpretation of that interaction term should be described more clearly.

Objectives: The study objectives are to understand the "likelihood of an eligible healthcare facility administering COVID-19 vaccines varied with the county-level racial composition and degree of urbanicity." However, the introduction reads that the study motivation is to better understand the "actual distribution of COVID-19 vaccines to healthcare facilities." These are related but different concepts with different policy implications. What led to the distribution of vaccines (structural problem)? Why did specific institutions or facilities administer vaccines (institutional or structural issue)? The objectives in the introduction might be helped by additional information on the hypothesis, and any evidence supporting the interpretation of the hypothesis.

Methods:

The methods do not describe how vaccination sites that were not listed on VaccineFinder were found. If these clinics were not included, this should be noted as a data limitation in the discussion section. The language for some of the concepts should be reviewed and clarified for an international audience (what is a Trump win? Is that to indicate Republican, Democratic or Independent candidate for elected office/president?) The limitation section should discuss some of the assumptions made calculating non-pharmacy locations (for example, Gillette Stadium, a large football field, was used to distribute vaccines in Massachusetts).

Results: Throughout the results presentation, the Ns for the number of pharmacies/vaccine sites should be presented in addition to the percentages. Without a sense of the number of eligible pharmacies/vaccine sites, the data are difficult to interpret. The data presentation would be strengthened by the ability to verify the vaccine distribution sites (a link to these publicly available data).

Discussion: The limitations of the data should be described in greater detail as indicated in the methods section of this review.

Any attachments provided with reviews can be seen via the following link:

[LINK]

Attachment

Submitted filename: Review PLOS Medicine_2022 COVID vaccine distribution inequities.docx

Click here for additional data file. (14.5KB, docx)

Decision Letter 2

Beryne Odeny

9 May 2022

Dear Dr. Hernandez,

Thank you very much for submitting your manuscript "Disparities in Distribution of COVID-19 Vaccines across U.S. Counties: A Geographic Information System-Based Cross-sectional Study" (PMEDICINE-D-21-04494R2) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Please pay particular attention to reviewer #3's second concern regarding comparison of data at multiple pandemic time points and extracting the required data for later pandemic time points from the same sources you have used for the primary analysis. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by May 30 2022 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Beryne Odeny,

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

Comments from the reviewers:

Reviewer #1: The authors have responded to each comment in turn, and have suitably included social vulnerability index into the analysis.

Reviewer #2: Thanks to the authors for addressing my comments and those of others. Some of them may have seemed exasperating, but I think this version is clearer for the adjustments.

Line Edits:

Line 81, Should White be non-Hispanic White? "among racial/ethnic minority groups than White individuals" This occurs again later. It's fine to use short-hand after first use, particularly if you spell that out, like you do for "Blacks" being "non-Hispanic Blacks," but if there is ambiguity, I'd at least define it the first time. You could check all your race references for this. For instance, I might explain blacks again when first mentioned in the discussion. Line 218 could use some precision for White and Black too. Or change throughout so that "Black" and "White" never stand alone (and consider defining what Hispanic means?).

Line 105-106.When you are talking about LTCFs and how they aren't included, you could talk about the other types (like stadiums) that are not included? Perhaps move up Line 159 -165.

Line 113. Add the month and year that adults got access to the vaccine.

Line 156-157 Rather than fixating on one of the parties, you could say you created a variable that said whether Trump or Biden won the election, unless there was someplace another candidate one. This would keep the audience from reading between the lines and getting tangled up in politics of COVID-19 and health equity.

Line 172 I try to avoid sentences that start with dependent "because" clauses.

Line 212 comma after metropolitan (trying to leave the line-editing to the line editor, sorry). Line 214 needs a comma after rural too.

Line 245 "Non-Hispanic Black and Hispanic" These could use a noun, like "people" or "minority groups."

Line 248 identification identify

Line 251 delivered "at" each site. Same issue for line 253, up to you.

Line 255 suggested adjustment: "so it was impossible to define whether institutional or structural reasons led some facilities to not become COVID-19 administration sites". Sentences that lead with negatives always sound a bit awkward, but I don't how to fully remove it here.

Line 258 level","

Line 259-262. This sentence could use a rewrite. Too many clauses, too many words.

Line 263 The discussion "regarding the causes of" lower uptake. I hope the discussion isn't focused on mistrust. Makes peer review more difficult, if so.

Line 265 starts with almost a pun, I like it.

Line 266 Does this journal allow references in the middle of sentences? if so, never mind.

Reviewer #3: The authors responded adequately to many of the reviewers' comments, and specifically to many of my own previous comments. However, two remaining glaring concerns are as follows:

1. Is it correct to draw conclusions about (and quantify) vaccine equity based only on number of sites eligible vs number of sites that actually distributed vaccines without the context of how many vaccines were distributed? Eg, even if 10 sites in a given area are eligible to distribute vaccines but only 5 of the sites actually distribute the vaccines, as long as those 5 sites distribute enough vaccines to the area's population then equity is achieved despite the fact that all 10 eligible sites did not distribute vaccines. As the authors used county-level measures for race/urbanicity, etc, why not additionally evaluate county level measures for number of COVID vaccines distributed during the study period? These county level data certainly exist, though they may exist outside the vaccine finder dataset.

2. The authors have made clear that they are unable to compare data at multiple pandemic time points, though the reason why they are unable to extract data for later pandemic time points is unclear (it should be available from the same sources the authors used to evaluate the May 2020 time point, eg from CDC's vaccine finder). Longitudinal evaluation of the data that the authors have put forward is essential to determine whether geographic disparity in vaccine distribution existed during the pandemic and whether or not it improved. Given the number of unknown variables that may have influenced the authors' finding that all eligible vaccine distribution sites did not distribute vaccines early in the pandemic, the utility of this stand-alone/cross sectional finding is unclear. If this finding persisted over time--particularly in the context of comparatively fewer vaccines being distributed to underserved populations living in a given county over time--a more meaningful/impactful conclusion could be drawn.

Any attachments provided with reviews can be seen via the following link:

[LINK]

Decision Letter 3

Beryne Odeny

23 Jun 2022

Dear Dr. Hernandez,

Thank you very much for re-submitting your manuscript "Disparities in Distribution of COVID-19 Vaccines across U.S. Counties: A Geographic Information System-Based Cross-sectional Study" (PMEDICINE-D-21-04494R3) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.  

We look forward to receiving the revised manuscript by Jun 30 2022 11:59PM.   

Sincerely,

Beryne Odeny,

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

1) Please temper claims of primacy in the discussion (i.e., “… because it is the first to analyze equity in the distribution…”) by stating, "to our knowledge" or something similar.

Notes from Academic Editor:

I think the reviewers' remaining (reviewer #3) concerns are justified and I also think the authors’ responses are reasonable. Nevertheless, I do think the authors should acknowledge some of these issues in their limitations.

On my read, the reviewer is questioning in part how meaningful the fraction of eligible facility delivering vaccines is, because this does not account for the volume of vaccines given per site, and a geography could be “covered” even if fewer facilities offered if those that did offered enough and were accessible enough. The authors respond that they are interested in the “distribution of vaccines” — I find this reasonable but not entirely convincing for two reasons. First, they say that whether a facility distributes at all is a more meaningful indication of equity in the supply, but it is not clear to me that regional demand does not influence whether a facility distributes at all. In other words, do we know that distribution as has been measured here is not influenced by demand? I do, nevertheless agree, that the “distributed or not” outcome is still valuable.

Second, I do think that the authors need to be clear that they are looking at a particular moment in time. The equity of distribution of COVID related services (whether testing or vaccines) changed dramatically over time. Regardless of how it ended up, the initial geographical distribution of vaccine supply is important, but it should be noted if this was during a particular swath of time. May 10, 2021 was very early in the vaccine roll out. In fact the whole paper needs to be clear that this is disparities in the initial roll out.

Any attachments provided with reviews can be seen via the following link:

[LINK]

Decision Letter 4

Beryne Odeny

6 Jul 2022

Dear Dr Hernandez, 

On behalf of my colleagues and the Academic Editor, Dr. Elvin Hsing Geng, I am pleased to inform you that we have agreed to publish your manuscript "Disparities in Distribution of COVID-19 Vaccines across U.S. Counties: A Geographic Information System-Based Cross-sectional Study" (PMEDICINE-D-21-04494R4) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process. 

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper. 

Sincerely, 

Beryne Odeny 

PLOS Medicine

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 STROBE Checklist. STROBE checklist for cross-sectional study.

    (DOC)

    Click here for additional data file. (83KB, doc)
    Attachment

    Submitted filename: Review PLOS Medicine_2022 COVID vaccine distribution inequities.docx

    Click here for additional data file. (14.5KB, docx)
    Attachment

    Submitted filename: response letter_3-30-2022.doc

    Click here for additional data file. (189.5KB, doc)
    Attachment

    Submitted filename: responses.doc

    Click here for additional data file. (50KB, doc)
    Attachment

    Submitted filename: response to comments.docx

    Click here for additional data file. (20KB, docx)

    Data Availability Statement

    Data from community pharmacies were obtained from the National Council for Prescription Drug Programs under a license that does not allow for data sharing. This is the reason why the data cannot be shared publicly without restrictions. Requests for community pharmacy data should be addressed to the National Council for Prescription Drug Programs (http://dataq.ncpdp.org/) The remaining data sources are publically available: Addresses of federally qualified health centers are available from the Health Resources and Services Administration website: https://data.hrsa.gov/data/reports/datagrid?gridName=FQHCs Coordinates of rural health clinics are available from the Centers for Medicare and Medicaid Services website https://www.cms.gov/Research-Statistics-Data-and-Systems/Downloadable-Public-Use-Files/Cost-Reports/Rural-Health-Center-222-2017-form Addresses of hospital outpatient departments are also available from the Centers for Medicare and Medicaid Services website https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Downloads/Outpatient_Data_2017_XLSX.zip Addresses of COVID-19 vaccine administration locations are available from the Centers for Disease Control and Prevention website https://www.cdc.gov/vaccines/covid-19/reporting/vaccinefinder/about.html.

    Articles from PLoS Medicine are provided here courtesy of PLOS

    RESOURCES