https://orcid.org/0000-0002-5361-6664
Figures
Abstract
Understanding disparities in COVID-19 preventive efforts among underserved populations requires a holistic approach that considers multiple social determinants of health (SDOH). While disparities in individual COVID-19 risk factors are well-documented, the cumulative impact of these factors on vaccine uptake and testing remains insufficiently quantified. This study applies a polysocial risk framework to assess the combined influence of geo-demographic, economic, and health-related factors on COVID-19 vaccination and testing. Using cross-sectional data from 9,758 participants enrolled in the NIH Rapid Acceleration of Diagnostics – Underserved Populations (RADx-UP) program (February 2020–April 2023), we analyzed associations between polysocial risk and preventive behaviors using multivariable generalized estimating equations (GEE). Overall, 72.5% of participants reported COVID-19 vaccination, and 82.1% reported testing. However, disparities were evident across polysocial risk profiles. Individuals experiencing intersecting geo-demographic (Non-Hispanic Black, age 45, Southern residence), economic (low education, unemployment, financial hardship), and health-related risk factors (substance use, low CVD risk, no flu vaccination) were 43−48 percentage points less likely to be vaccinated compared to groups with higher adoption (p < 0.001). Testing disparities were narrower but remained significant, with differences ranging from 2 to 27 percentage points depending on the specific polysocial risk profiles. The findings underscore the utility of polysocial risk modeling as a predictive tool for identifying populations at highest risk of disengagement from preventive care, informing targeted precision public health interventions. Beyond COVID-19, this approach has broader applicability for understanding disparities in chronic disease prevention, cancer screening, maternal and child health, and health-related social needs (HRSN) interventions. Integrating polysocial risk assessments into clinical and public health settings can enhance data-driven strategies to improve population health outcomes.
Citation: Brown DR, Cyr DD, Wruck L, Stefano TA, Mehri N, Bursac Z, et al. (2025) COVID-19 prevention is shaped by polysocial risk: A cross-sectional study of vaccination and testing disparities in underserved populations. PLoS One 20(7): e0328779. https://doi.org/10.1371/journal.pone.0328779
Editor: Muhammad Muntazir Mehdi Khan, Allegheny General Hospital, UNITED STATES OF AMERICA
Received: February 24, 2025; Accepted: July 2, 2025; Published: July 17, 2025
Copyright: © 2025 Brown et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: This study makes use of data collected by individual RADx-UP projects and shared with the CDCC under the RADx-UP data sharing policy which provides multi-level protections for confidentiality of participants. De-identified data underlying the results presented in this analysis are available by request from the NIH Rapid Acceleration of Diagnostics Data Hub (RADx Data Hub) https://radxdatahub.nih.gov/.
Funding: Research reported in this RADx-UP publication was supported by the National Institutes of Health under Award Numbers U01DA040381, U01MD017423, and U24MD016258. This work was supported by Azure sponsorship credits granted by Microsoft’s AI for Good Research Lab. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Underserved populations experience persistent health disparities [1] driven by interconnected social determinants of health (SDOH)—structural and contextual factors such as economic stability, education, healthcare access, and neighborhood environment [2,3]. These disparities were particularly evident during the COVID-19 pandemic, where variations in healthcare access, economic constraints, and logistical barriers led to differences in vaccination, testing, and mortality rates [4–10]. While traditional models like Behavioral Model of Health Service Use [11,12] and the Health Beliefs Model (HBM) [13] have identified key demographic, socioeconomic, and health-related influences on preventive behavior adoption [14–21], they do not fully account for the complexity of multiple intersecting social determinants.
The polysocial risk framework, analogous to polygenic risk scores in genetics [22], provides a structured method for evaluating how social determinants interact to shape health outcomes and influence health behaviors collectively [23–28]. This approach aligns with Bourdieu’s theory of capital, which conceptualizes economic, social, and cultural resources as key determinants of individual opportunities and health prospects [29,30]. As public health efforts increasingly rely on data-driven approaches, precision public health (PPH) has emerged as a framework for tailoring interventions at a population level [31–34]. However, PPH faces a significant challenge in quantifying the combined effects of multiple social determinants. The polysocial risk framework addresses this gap by providing a systematic method for assessing how intersecting social factors influence health behaviors. This approach is particularly relevant as healthcare systems expand screening for health-related social needs (HRSN) [35] and implement broader strategies to reduce disparities at scale [36–39].
The NIH-funded Rapid Acceleration of Diagnostics – Underserved Populations (RADx-UP) program provides an ideal opportunity to study these relationships in real-world community settings. This national initiative supports community-engaged research addressing gaps in COVID-19 testing and vaccination, enabling examination of how social risk factors influence prevention behaviors across diverse populations and contexts [40–42]. Using RADx-UP data, this study applies the polysocial risk framework to assess disparities in COVID-19 vaccination and testing. Our objectives are to: (1) quantify the prevalence of COVID-19 prevention behaviors, (2) evaluate how various social determinants influence these behaviors, (3) assess the contributions of different social factor combinations, and (4) identify subgroups experiencing multiple barriers to prevention. By examining how multiple social determinants shape COVID-19 prevention behaviors, this study moves beyond individual risk factors to identify high-risk subgroups for targeted interventions. In doing so, it addresses a critical gap in understanding how multiple social factors collectively influence prevention efforts.
Methods
Study design and data source
This cross-sectional study uses data collected by individual RADx-UP projects between February 3, 2020 and April 21, 2023 and integrated by the Coordination and Data Collection Center (CDCC) [41]. The CDCC maintains a standardized data repository with common data elements (CDEs) from RADx-UP projects, allowing for cross-project analysis of factors influencing COVID-19 outcomes in underserved communities [40,43]. The contributing projects varied in geographic location, study design, and population characteristics, with further details available in S1 Table.
This study used data from the RADx-UP Core Analytic Datasets version 1.6 accessed through the CDCC on April 30, 2023 [40,43]. The first available observation for each participant was used for analysis. The components of the CDEs used were: Sociodemographic, Location, Covid Test, Vaccine Acceptance, Testing, Housing/Employment/Insurance, Alcohol and Tobacco, Work PPE and Distancing, Medical History, and Health Status. In this study, we conducted an integrated analysis of geo-demographic context, economic characteristics, and health-related characteristics. This approach aims to provide a more thorough understanding of the barriers and facilitators to prevention engagement.
All participants provided written informed consent for sharing data with the CDCC for analysis. IRB approval was obtained at the institutional level for each individual RADx-UP project and at the Duke IRB (PRo00106873) for the RADx-UP CDCC.
Participants
Participants were adults aged 18 years and older who underwent COVID-19 testing and were enrolled in a RADx-UP project. A series of project-level and participant-level exclusions were applied to derive the final study population, as detailed in Fig 1 (Consort Flow Diagram) and S1 Appendix. Exclusion Criteria.
Measures
The primary outcome measures in this study were COVID-19 vaccination status and COVID-19 testing engagement. Vaccination status was determined based on participant responses to the question, “Have you received a COVID-19 vaccine?” COVID-19 testing engagement was assessed through participants’ history of testing (“Have you ever been tested for COVID-19?”) prior to their enrollment in the RADx-UP study. Prior test positivity rates were also examined among participants who had previously engaged in testing.
In this study, ‘health risk groups’ describe individuals with characteristics affecting their likelihood of engaging in health-promoting activities, such as COVID-19 vaccination and testing. Higher health risk groups are those with characteristics associated with lower uptake of preventive measures, while lower health risk groups have characteristics associated with higher uptake.
Independent variables were selected to represent key social factors influencing prevention behaviors, spanning three domains:
- (1). Geo-demographic context, including age, sex assigned at birth, race/ethnicity, and U.S. geographic region;
- (2). Economic characteristics, such as education, household income, employment status, household description, economic challenges; and
- (3). Health-related characteristics, encompassing insurance status, health status, disability, any illicit drug use, heavy alcohol use, any mental health problem, cardiometabolic risk, other chronic conditions, access to COVID-19 testing, prior flu vaccination history.
Each of these domains captures interconnected aspects of participants’ social environments that may shape their engagement with COVID-19 prevention measures. For example, employment status reflects both economic security and potential exposure settings, while prior flu vaccination history serves as a proxy for healthcare access and established preventive behaviors. Details about the collection of independent variables are provided in S1 and S2 Table. Description of Prevention Behaviors and Population Characteristics.
Statistical analysis
Descriptive statistics summarized participant characteristics at the time of first COVID-19 test (baseline). Continuous variables are presented as the number of non-missing values, mean (SD), and median (25th, 75th percentiles). Categorical variables are presented as counts (percentages). The association between independent variables and the primary and secondary outcomes was analyzed using generalized estimating equations with a logit link. Project was included as a clustering group variable in the models to account for geographic and project-based distributions, as RADx-UP projects are specific to certain populations/regions [44]. Among the 14 projects included in the analysis, sample sizes ranged from 95 to 3,200 participants. The percentage of participants who met the primary vaccination outcome and the primary testing outcome ranged from 26.3% to 93.7% across projects and 46.5% to 98.9%, respectively. Within-project correlation was estimated to be 0.04 and 0.02 for the primary vaccine and testing outcomes, respectively, indicating very low correlation among subject measurements from the same project. The assumption of linearity was assessed for all continuous covariates. Age was found to exhibit a nonlinear relationship with the COVID-19 vaccination outcome and, upon visual inspection, was modeled as piecewise linear with a knot point at 55 years. For the COVID-19 testing outcomes, age was modeled assuming a linear relationship with outcome. Multiple imputation with chained equations was performed to impute missing covariate data. Odds ratios (ORs) and 95% confidence intervals (CIs) were reported. Hypothesis testing was two-sided and Wald tests were used at an alpha level of 0.05. Analyses were conducted using SAS v9.4 (Cary, NC).
To evaluate differences in preventive behavior adoption, we used our estimated models to predict the probability of vaccination and testing across representative population groupings. Three risk groups were considered: 1) geo-demographic risk groups, consisting of age, race, and geographic region of residence; 2) economic risk groups, consisting of education level, employment status, and number of economic challenges present (i.e., ability to get access to food, water, transportation, housing, needed medications, health care); 3) health risk groups, consisting of drug use (i.e., intravenous or any substance abuse), cardiovascular disease risk factors (i.e., smoking, vaping, overweight, diabetes, hypertension, and history of cardiovascular disease), and flu vaccination status. For each grouping, we compared populations with higher versus lower likelihood of engaging in these preventive measures (vaccination and testing). These groupings were selected based on commonly occurring combinations of characteristics in our dataset and were chosen to illustrate key differences in preventive measure engagement across demographic, economic, and health-related factors.
Results
Study population characteristics
The study population included 9,758 participants from diverse racial, ethnic, and socioeconomic backgrounds (Table 1). The median age was 50 years (IQR: 37–60), and 55.0% were female. The sample was racially and ethnically diverse, with 39.6% Non-Hispanic Black, 34.3% Hispanic, 20.1% Non-Hispanic White, and 6.1% Non-Hispanic Other participants. Socioeconomic disadvantage was prevalent, with two-thirds (66.1%) reporting annual household incomes below $25,000 and 62.7% being unemployed at the time of the study.
COVID-19 vaccination, testing, and positivity rates
Overall, 72.5% of participants reported receiving a COVID-19 vaccine, and 82.1% (95% CI, 81.3%−82.9%) reported ever having been tested for COVID-19. Among those tested, 26.0% reported a positive result at some point (Table 1). These rates suggest relatively high engagement with COVID-19 prevention measures in the study population.
Factors associated with vaccination behavior
Multivariable GEE analysis (Table 2) identified several factors significantly associated with vaccination status. Geographic Context: Individuals in the Midwest (OR, 2.06; 95% CI, 1.68–2.51; P < .01) and Northeast (OR, 2.62; 95% CI, 1.18–5.84; P < .01) had higher odds of vaccination compared to those in the South. Race/Ethnicity: Compared to Non-Hispanic White participants, Non-Hispanic Black participants had lower odds of vaccination (OR, 0.62; 95% CI, 0.51–0.75; P < .01), while Hispanic participants had higher odds (OR, 1.34; 95% CI, 1.11–1.62; P < .01). Education: Those with education beyond high school had higher odds of vaccination compared to those with less than high school education (OR, 1.37; 95% CI, 1.17–1.61; P < .01). Employment: Essential (OR, 1.40; 95% CI, 1.26–1.55; P < .01) and non-essential workers (OR, 1.37; 95% CI, 1.21–1.54; P < .01) had higher odds of vaccination compared to unemployed individuals. Economic Challenges: Participants facing 1 or 2 economic challenges had lower odds of vaccination compared to those without economic challenges (OR, 0.82; 95% CI, 0.70–0.97; P = .05). Drug Use: Those reporting drug use had lower odds of vaccination compared to those not reporting drug use (OR, 0.79; 95% CI, 0.67–0.93; P < .01). Flu Vaccination: Participants with prior flu vaccination had higher odds of COVID-19 vaccination compared to those who had never received a flu vaccine (OR, 3.54; 95% CI, 2.78–4.51; P < .01). Cardiovascular Risk: Participants with 3 or more cardiovascular disease risk factors had higher odds of vaccination compared to those with 0−2 risk factors (OR, 1.32; 95% CI, 1.15–1.52; P < .01).
Factors associated with testing behavior
Analysis of testing patterns (Table 2) revealed similar social influences. Geographic Context: Northeast and Midwest residents had significantly higher odds of being tested than those in the south (OR, 6.62; 95% CI, 3.15–13.93; P < .01 and OR, 2.60; 95% CI, 1.35–5.04; P < .01, respectively). Sex: Females had higher odds of being tested compared to males (OR, 1.37; 95% CI, 1.17–1.62; P < .01). Race/Ethnicity: Hispanic participants had significantly higher odds of being tested compared to Non-Hispanic White participants (OR, 1.50; 95% CI, 1.17–1.93; P < .01). Education: Higher education levels were associated with increased likelihood of testing, with those having education beyond high school having higher odds of testing compared to those with less than high school education (OR, 1.70; 95% CI, 1.39–2.08; P < .01). Health Insurance: Uninsured individuals had lower odds of being tested compared to insured individuals (OR, 0.76; 95% CI, 0.64–0.90; P < .01). Access to Testing: Participants with access to COVID-19 testing had higher odds of being tested compared to those without access (OR, 1.84; 95% CI, 1.57–2.16; P < .01). Health Status: Participants reporting fair or poor health had higher odds of being tested compared to those reporting good to excellent health (OR, 1.26; 95% CI, 1.10–1.45; P < .01). Mental Health: Those with any mental health risk had higher odds of being tested compared to those without mental health risks (OR, 1.27; 95% CI, 1.11–1.46; P < .01). Cardiovascular Risk: Individuals with 3 or more CVD risk factors had higher odds of being tested compared to those with 0−2 risk factors (OR, 1.24; 95% CI, 1.12–1.36; P < .01).
Social patterns in test results
Among participants who had been previously tested for COVID-19, several social and environmental factors were associated with positive test results (Table 3): Geographic Context: Individuals in the West (OR, 2.21; 95% CI, 1.46–3.35; P < .01) and Midwest (OR, 1.60; 95% CI, 1.02–2.51; P < .01) were more likely to have tested positive than those in the South. Age: Younger age was associated with higher odds of testing positive (OR per 5-year increase, 0.94; 95% CI, 0.91–0.98; P < .01); Sex: Females had higher odds of testing positive compared to males (OR, 1.22; 95% CI, 1.08–1.37; P < .01); Race/Ethnicity: Hispanics had higher odds of testing positive compared to Non-Hispanic Whites (OR, 1.34; 95% CI, 1.04–1.72; P < .01); Employment: Both essential workers (OR, 1.36; 95% CI, 1.14–1.61; P < .01) and non-essential workers (OR, 1.40; 95% CI, 1.17–1.68; P < .01) had higher odds of testing positive compared to unemployed individuals; Health Status: Participants reporting fair or poor health had higher odds of testing positive (OR, 1.23; 95% CI, 1.13–1.35; P < .01).
Polysocial risk analysis of preventive behavior adoption
Table 4 presents the predicted probabilities of COVID-19 vaccination and testing across geo-demographic, economic, and health risk groups, revealing stark disparities in preventive behavior adoption.
Geo-demographic factors.
Individuals at higher risk for low preventive behavior engagement (age 45, Non-Hispanic Black, residing in the U.S. South) had a predicted probability of vaccination of 0.50 (95% CI, 0.49–0.52) compared to 0.93 (95% CI, 0.90–0.94) for those at lower risk (age 65, Hispanic, residing in the U.S. Northeast), a difference of 43 percentage points. The disparity in testing was smaller but still substantial, with predicted probabilities of 0.69 (95% CI, 0.66–0.72) and 0.96 (95% CI, 0.94–0.96) for the low and high engagement groups, respectively.
Socio-economic factors.
Groups with lower preventive behavior adoption (less than high school education, unemployed, experiencing 3 + economic challenges) had a predicted probability of vaccination of 0.61 (95% CI, 0.59–0.64) compared to 0.82 (95% CI, 0.81–0.83) for groups with higher adoption (beyond high school education, non-essential worker, no economic challenges), a 21-percentage point difference. For testing, the predicted probabilities were 0.80 (95% CI, 0.77–0.82) for the low adoption group and 0.91 (95% CI, 0.90–0.92) for the high adoption group.
Health-related factors.
The largest disparities in vaccination were observed across health-related risk factors. Individuals with lower preventive behavior adoption (any drug use, 0−2 CVD risk factors, never received flu vaccine) had a predicted probability of vaccination of only 0.39 (95% CI, 0.36–0.42) compared to 0.87 (95% CI, 0.86–0.88) for groups with higher adoption (no drug use, 3 + CVD risk factors, ever received flu vaccine), a difference of 48 percentage points. However, testing rates were more similar between the high and low health risk groups, with predicted probabilities of 0.86 (95% CI, 0.84–0.88) and 0.84 (95% CI, 0.81–0.87), respectively.
Discussion
Our findings demonstrate that a polysocial risk framework can capture the complex, compounded effects of geo-demographic, economic, and health-related factors on COVID-19 preventive behaviors with remarkable precision. The data reveal that individuals facing multiple intersecting disadvantages are significantly less likely to be vaccinated, which not only corroborates established theoretical models—such as those related to social capital [29,30]—but also provides actionable insights for public health policy. Specifically, these results advocate for the integration of systematic social risk assessments into clinical practice and community-based interventions, thereby enabling targeted, precision public health strategies.
Interpretation of key findings
The polysocial risk framework revealed dramatic disparities in prevention behaviors across multiple domains:
Geo-demographic factors: Non-Hispanic Black participants aged 45 residing in the South had a 43-percentage point lower predicted vaccination probability (0.50) compared to Hispanic participants aged 65 in the Northeast (0.93). This stark difference highlights how the compound effect of race, age, and region may potentially create barriers to prevention.
Socio-economic factors: Groups facing multiple economic challenges (less education, unemployment, financial hardship) showed 21-percentage point lower vaccination rates compared to more economically stable groups. This gap highlights the association between economic precarity and preventive health behaviors.
Health-related factors: The largest disparities emerged in health-related domains – individuals with substance use, fewer cardiovascular risk factors, and no flu vaccination history were 48-percentage points less likely to be vaccinated than their counterparts. This suggests a strong association between established patterns of healthcare engagement and COVID-19 prevention behaviors.
Notably, testing disparities were narrower than vaccination disparities across risk groups. This likely reflects the success of widespread testing availability through community sites and at-home options [45] and may also reflect the generally higher rates of testing compared to vaccination during the pandemic.
Our results align with prior research demonstrating the profound impact of social determinants on health behaviors [5,46–48], but they also highlight higher-than-expected preventive engagement among underserved groups [49,50]. This may reflect the success of RADx-UP’s community-engaged strategies, emphasizing the importance of trusted, locally tailored interventions.
Public health and policy implications
The study supports integrating social risk assessments in clinical and public health settings [37,51–55] to proactively address compounded vulnerabilities that drive disparities in preventive health behaviors. Our findings highlight significant disparities in COVID-19 vaccination and testing among different polysocial risk profiles. For example, individuals experiencing intersecting high geo-demographic (Non-Hispanic Black, age 45, Southern residence), economic (low education, unemployment, financial hardship), or health-related risk factors (substance use, low CVD risk, no flu vaccination) were each significantly less likely to be vaccinated compared to groups with low geo-demographic, economic, or health risk profiles.
Healthcare systems can embed standardized polysocial risk screening in clinical encounters, and automated risk stratification tools into electronic health records, enabling targeted outreach to high-risk populations, and creation of risk-based care pathways that trigger specific interventions. The utility of polysocial risk modeling as a predictive tool for identifying populations at highest risk of disengagement from preventive care can inform targeted precision public health interventions.
However, implementation challenges—such as standardization of risk assessment tools and provider training—must be addressed to ensure effective integration into routine care. Findings underscore the need for multi-level interventions that address polysocial risk profiles rather than single risk factors. Beyond COVID-19, the polysocial risk framework has broader applicability for understanding disparities in chronic disease prevention, cancer screening, maternal and child health, and health-related social needs (HRSN) interventions
Healthcare delivery models need to address multiple social needs simultaneously and create sustainable funding mechanisms for social care integration. Community-based initiatives should be prioritized in low vaccine-uptake regions, leveraging trusted local organizations to enhance engagement.
Strengths, limitations and future research
Polysocial risk scores have limitations, including potential oversimplification, limited generalizability, and lack of direct policy translation [56]. Rather than modeling outcomes with a single poly-social risk score, we examine predicted probabilities of multiple risks. Thus, our study avoids these pitfalls while still providing valuable insights into the cumulative impact of multiple social factors on COVID-19 disparities. Additional strengths of this study include its large, diverse sample of underserved individuals, and examination of multiple COVID-19 outcomes. The use of data from the RADx-UP program provides unique insights into a population often underrepresented in health research.
While this study provides new insights into social risk stratification, several limitations must be acknowledged. Its cross-sectional design limits causal inference, and reliance on self-reported data introduces potential recall and social desirability bias. Data did not include the frequency of COVID-19 testing, which may influence the positivity rates, or the completion of all recommended vaccine doses. The analysis was also constrained to social factors captured through common data elements across RADx-UP studies, which do not fully encompass the breadth of factors affecting underserved communities. While the RADx-UP sample is diverse, it may not entirely reflect all underserved populations, potentially limiting broader applicability. Selection bias is another consideration, as individuals who engage in community-based research may differ from those with less connection to health services, meaning disparities may be even greater than observed. Future research should focus on longitudinal studies to examine how shifting social and economic conditions influence health behaviors over time. Additionally, refining polysocial risk assessment tools—including validation across different populations—will be crucial for scaling its application in clinical and public health settings.
Conclusions
The polysocial risk framework provides a powerful tool for understanding and addressing disparities in preventive health behaviors. By analyzing COVID-19 vaccination and testing through a multi-dimensional risk model, this study reveals how intersecting social, economic, and behavioral factors drive disparities. Our findings highlight that disparities in prevention are shaped by cumulative social disadvantage, rather than any single factor. Individuals facing multiple overlapping social risks—economic instability, lower healthcare access, and limited preventive health engagement—were significantly less likely to receive vaccines or engage in testing. These patterns emphasize that traditional approaches focusing on singular risk factors miss the broader, systemic forces shaping health behaviors.
Embedding polysocial risk assessment into healthcare and public health systems could fundamentally reshape how we predict, prevent, and intervene in health disparities. By incorporating multi-factorial social risk screening into clinical workflows, public health interventions can become more precise, targeted, and effective—shifting from broad, reactive policies to proactive, data-driven strategies that account for real-world complexity [31,57]. Realizing this potential will require investment in data infrastructure, standardization of risk modeling, and the seamless integration of social risk insights into health interventions.
Supporting information
S1 Table. Study project metadata.
This table provides detailed metadata for various study projects, including geographic location, primary and secondary populations, study design, study setting, mode of data collection, and vaccine availability phase. It also includes exclusion criteria for the study population and descriptions of prevention behaviors and population characteristics.
https://doi.org/10.1371/journal.pone.0328779.s001
(DOCX)
S2 Table. Description of prevention behaviors and population characteristics.
This table describes prevention behaviors and population characteristics, including survey questions and derivation details for primary outcomes such as COVID-19 vaccination, testing, and positive test results. It also covers demographics, economic risk characteristics, health risk characteristics, and access to COVID-19 testing.
https://doi.org/10.1371/journal.pone.0328779.s002
(DOCX)
S1 Appendix. Exclusion criteria.
This appendix outlines the exclusion criteria for the study population, including project-level and participant-level exclusions based on missing data and other criteria. It provides detailed information on the reasons for excluding certain projects and participants from the study.
https://doi.org/10.1371/journal.pone.0328779.s003
(DOCX)
References
- 1.
Barton B, Boonyasai RT, Hahn C, Bonnett D, Fitzell J, Goldstein E. 2023 National Healthcare Quality and Disparities Report. 2023.
- 2. Link BG, Phelan J. Social conditions as fundamental causes of disease. J Health Soc Behav. 1995;35:80.
- 3. Braveman P, Gottlieb L. The social determinants of health: it’s time to consider the causes of the causes. Public Health Rep. 2014;129 Suppl 2(Suppl 2):19–31. pmid:24385661
- 4. Ayorinde A, Ghosh I, Ali I, Zahair I, Olarewaju O, Singh M, et al. Health inequalities in infectious diseases: a systematic overview of reviews. BMJ Open. 2023;13(4):e067429. pmid:37015800
- 5. Williams AM, Clayton HB, Singleton JA. Racial and ethnic disparities in COVID-19 vaccination coverage: the contribution of socioeconomic and demographic factors. Am J Prev Med. 2022;62(4):473–82. pmid:34872772
- 6. Cuadros DF, Gutierrez JD, Moreno CM, Escobar S, Miller FD, Musuka G, et al. Impact of healthcare capacity disparities on the COVID-19 vaccination coverage in the United States: a cross-sectional study. Lancet Reg Health Am. 2023;18:100409. pmid:36536782
- 7. Igoe M, Das P, Lenhart S, Lloyd AL, Luong L, Tian D, et al. Geographic disparities and predictors of COVID-19 hospitalization risks in the St. Louis Area, Missouri (USA). BMC Public Health. 2022;22(1):321. pmid:35168588
- 8. Webb Hooper M, Nápoles AM, Pérez-Stable EJ. COVID-19 and Racial/Ethnic Disparities. JAMA. 2020;323(24):2466–7. pmid:32391864
- 9. Azar KMJ, Shen Z, Romanelli RJ, Lockhart SH, Smits K, Robinson S, et al. Disparities in outcomes among COVID-19 patients in a large health care system in California. Health Aff (Millwood). 2020;39(7):1253–62. pmid:32437224
- 10. De Ramos IP, Lazo M, Schnake-Mahl A, Li R, Martinez-Donate AP, Roux AVD, et al. COVID-19 outcomes among the Hispanic population of 27 large US cities, 2020-2021. Am J Public Health. 2022;112(7):1034–44. pmid:35588187
- 11. Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav. 1995;36(1):1.
- 12. Andersen RM. National health surveys and the behavioral model of health services use. Med Care. 2008;46:647–53.
- 13. Rosenstock IM. The health belief model and preventive health behavior. Health Educ Monogr. 1974;2(4):354–86.
- 14. Salmon DA, Dudley MZ, Brewer J, Kan L, Gerber JE, Budigan H, et al. COVID-19 vaccination attitudes, values and intentions among United States adults prior to emergency use authorization. Vaccine. 2021;39(19):2698–711. pmid:33781601
- 15. Adu P, Poopola T, Medvedev ON, Collings S, Mbinta J, Aspin C, et al. Implications for COVID-19 vaccine uptake: a systematic review. J Infect Public Health. 2023;16(3):441–66. pmid:36738689
- 16. McCabe SD, Hammershaimb EA, Cheng D, Shi A, Shyr D, Shen S, et al. Unraveling attributes of COVID-19 vaccine acceptance and uptake in the U.S.: a large nationwide study. Sci Rep. 2023;13(1):8360. pmid:37225748
- 17. Reiter PL, Pennell ML, Katz ML. Acceptability of a COVID-19 vaccine among adults in the United States: how many people would get vaccinated? Vaccine. 2020;38(42):6500–7. pmid:32863069
- 18. Bonner KE, Vashist K, Abad NS, Kriss JL, Meng L, Lee JT, et al. Behavioral and social drivers of COVID-19 vaccination in the United States, August-November 2021. Am J Prev Med. 2023;64(6):865–76. pmid:36775756
- 19. Latkin CA, Dayton L, Yi G, Konstantopoulos A, Boodram B. Trust in a COVID-19 vaccine in the U.S.: a social-ecological perspective. Soc Sci Med. 2021;270:113684. pmid:33485008
- 20. Olagoke AA, Olagoke OO, Hughes AM. Intention to vaccinate against the novel 2019 coronavirus disease: the role of health locus of control and religiosity. J Relig Health. 2021;60(1):65–80. pmid:33125543
- 21. Lin C, Tu P, Beitsch LM. Confidence and receptivity for COVID-19 vaccines: a rapid systematic review. Vaccines (Basel). 2020;9(1):16. pmid:33396832
- 22. Torkamani A, Wineinger NE, Topol EJ. The personal and clinical utility of polygenic risk scores. Nat Rev Genet. 2018;19(9):581–90. pmid:29789686
- 23. Chen P, Yang Z, Fan Z, Wang B, Tang Y, Xiao Y, et al. Associations of polysocial risk score with incident rosacea: a prospective cohort study of government employees in China. Front Public Health. 2023;11:1096687. pmid:37206873
- 24. Li Y, Zeng Z, Zhuang Z, Zhao Y, Zhang L, Wang W, et al. Polysocial and polygenic risk scores and all-cause dementia, Alzheimer’s disease, and vascular dementia. J Gerontol A Biol Sci Med Sci. 2024;79(3):glad262. pmid:37966923
- 25. Javed Z, Valero-Elizondo J, Dudum R, Khan SU, Dubey P, Hyder AA, et al. Development and validation of a polysocial risk score for atherosclerotic cardiovascular disease. Am J Prev Cardiol. 2021;8:100251. pmid:34553187
- 26. Huang Y, Guo J, Donahoo WT, Lee YA, Fan Z, Lu Y, et al. A fair individualized polysocial risk score for identifying increased social risk in type 2 diabetes. Nat Commun. 2024;15(1):8653. pmid:39369018
- 27. Vest JR, McNamee C, Musey PI Jr. Health-related social needs, methods, and concerns for a polysocial risk score: an expert panel. Am J Manag Care. 2025;31(2):55–62. pmid:39977290
- 28. Figueroa JF, Frakt AB, Jha AK. Addressing social determinants of health: time for a polysocial risk score. JAMA. 2020;323(16):1553–4. pmid:32242887
- 29. Mudd AL, Oude Groeniger J, Bal M, Verra SE, van Lenthe FJ, Kamphuis CBM. Testing conditionality with Bourdieu’s capital theory: how economic, social, and embodied cultural capital are associated with diet and physical activity in the Netherlands. SSM Popul Health. 2023;22:101401. pmid:37123560
- 30. Schneider-Kamp A. Health capital: toward a conceptual framework for understanding the construction of individual health. Soc Theory Health. 2021;19(3):205–19. pmid:32837331
- 31. Roberts MC, Holt KE, Del Fiol G, Baccarelli AA, Allen CG. Precision public health in the era of genomics and big data. Nat Med. 2024;30(7):1865–73. pmid:38992127
- 32. Khoury MJ, Galea S. Will precision medicine improve population health? JAMA. 2016;316(13):1357–8. pmid:27541310
- 33. Khoury MJ, Iademarco MF, Riley WT. Precision public health for the era of precision medicine. Am J Prev Med. 2016;50(3):398–401. pmid:26547538
- 34. Velmovitsky PE, Bevilacqua T, Alencar P, Cowan D, Morita PP. Convergence of precision medicine and public health into precision public health: toward a big data perspective. Front Public Health. 2021;9:561873. pmid:33889555
- 35. Alderwick H, Gottlieb LM. Meanings and misunderstandings: a social determinants of health Lexicon for health care systems. Milbank Q. 2019;97(2):407–19. pmid:31069864
- 36. Bosold A, Singhakiat B, Talwar-Hebert M, Robinson S, Shybut A, Crane G, et al. Sharing social needs data across sectors: lessons from the centers for medicare and medicaid services innovation center’s accountable health communities model. Health Promot Pract. 2025;26(4):624–9. pmid:39219270
- 37.
National Academies of Sciences, Engineering, and Medicine. Investing in interventions that address non-medical, health-related social needs: Proceedings of a workshop. Washington, D.C.: National Academies Press; 2019.
- 38. Centers for Medicare and Medicaid Services, Billioux A, Verlander K, Anthony S, Alley D. Standardized screening for health-related social needs in clinical settings: the accountable health communities screening tool. NAM Perspectives. 2017;7(5).
- 39. Sharma SV, Chuang R-J, Rushing M, Naylor B, Ranjit N, Pomeroy M, et al. Social determinants of health-related needs during COVID-19 among low-income households with children. Prev Chronic Dis. 2020;17:E119. pmid:33006541
- 40. Carrillo GA, Cohen-Wolkowiez M, D’Agostino EM, Marsolo K, Wruck LM, Johnson L, et al. Standardizing, harmonizing, and protecting data collection to broaden the impact of COVID-19 research: the rapid acceleration of diagnostics-underserved populations (RADx-UP) initiative. J Am Med Inform Assoc. 2022;29(9):1480–8. pmid:35678579
- 41. Corbie G, D’Agostino EM, Knox S, Richmond A, Woods CW, Dave G, et al. RADx-UP coordination and data collection: an infrastructure for COVID-19 testing disparities research. Am J Public Health. 2022;112(S9):S858–63. pmid:36194852
- 42. Webb Hooper M, Compton WM, Walsh ER, Hodes RJ, Pérez-Stable EJ. Harnessing the power of community-engaged science to facilitate access and uptake of COVID-19 testing: RADx-UP. Am J Public Health. 2022;112(S9):S854–7. pmid:36446064
- 43. NIH RADx-UP Common Data Elements. In: RADx-UP [Internet]. [cited 13 Oct 2024. ]. Available from: https://radx-up.org/research/cdes/
- 44. RADx® Underserved Populations (RADx-UP): List of all Projects. In: RADx-UP [Internet]. [cited 13 Oct 2024. ]. Available from: https://radx-up.org/research/projects/list/
- 45. Rader B, Gertz A, Iuliano AD, Gilmer M, Wronski L, Astley CM, et al. Use of at-home COVID-19 tests - United States, August 23, 2021-March 12, 2022. MMWR Morb Mortal Wkly Rep. 2022;71(13):489–94. pmid:35358168
- 46. Millett GA, Jones AT, Benkeser D, Baral S, Mercer L, Beyrer C, et al. Assessing differential impacts of COVID-19 on black communities. Ann Epidemiol. 2020;47:37–44. pmid:32419766
- 47. Na L, Banks S, Wang PP. Racial and ethnic disparities in COVID-19 vaccine uptake: A mediation framework. Vaccine. 2023;41(14):2404–11. pmid:36894396
- 48. Wong E, Liu Y, Shier V, Datar A. Heterogeneity in COVID-19 vaccine uptake within low-income minority communities: evidence from the watts neighborhood health study. BMC Public Health. 2024;24(1):503. pmid:38365658
- 49. Kawuki J, Chen S, Fang Y, Liang X, Chan PS-F, Wang Z. COVID-19 vaccine acceptance, attitude and perception among slum and underserved communities: a systematic review and meta-analysis. Vaccines (Basel). 2023;11(5):886. pmid:37242990
- 50. Campbell J, Kaur A, Gamino D, Benoit E, Amos B, Windsor L. Individual and structural determinants of COVID-19 vaccine uptake in a marginalized community in the United States. Vaccine. 2023;41(39):5706–14. pmid:37550145
- 51. American Diabetes Association Professional Practice Committee. 1. Improving care and promoting health in populations: standards of care in diabetes-2024. Diabetes Care. 2024;47(Suppl 1):S11–9. pmid:38078573
- 52. Virani SS, Newby LK, Arnold SV, Bittner V, Brewer LC, Demeter SH, et al. 2023 AHA/ACC/ACCP/ASPC/NLA/PCNA Guideline for the Management of Patients With Chronic Coronary Disease: A Report of the American Heart Association/American College of Cardiology Joint Committee on Clinical Practice Guidelines. Circulation. 2023;148(9):e9–119. pmid:37471501
- 53. Chang R, Amin Z, Cheema NH, Yousefzai S, Gardezi K, Shah A, et al. Integrating the polysocial risk score: enhancing comprehensive healthcare delivery. Methodist Debakey Cardiovasc J. 2024;20(5):89–97. pmid:39525375
- 54.
National Academies of Sciences, Engineering, and Medicine. Integrating social care into the delivery of health care: Moving upstream to improve the nation’s health. Washington, D.C.: The National Academies Press; 2019. Available from: https://www.nap.edu/catalog/25467
- 55.
National Quality Forum (NQF). Social Drivers of Health Data Utilization: Integrating Healthcare and Community Services to Address Health-Related Social Needs. Washington, DC: NQF; 2024. Available from: https://www.qualityforum.org/Publications/2024/3/Social_Drivers_of_Health_Data_Utilization__Integrating_Healthcare_and_Community_Services_to_Address_Health-Related_Social_Needs.aspx
- 56. Kinner SA, Borschmann R. Polysocial risk scores for assessing social determinants of health. JAMA. 2020;324(16):1680–1. pmid:33107934
- 57. Olstad DL, McIntyre L. Reconceptualising precision public health. BMJ Open. 2019;9(9):e030279. pmid:31519678