Intro

In May 2023, the Centers for Disease Control and Prevention (Centers for Disease Control and Prevention (CDC) September 8, 2022) recommended that everyone aged 5 years and older receive at least one booster following the COVID-19 primary series. As of October 2022, 62.85% of the Texas population (6 months+) was fully vaccinated (FV), falling below the US national average (Texas Department of State Health Services (DSHS), n.d.; Centers for Disease Control and Prevention (CDC) March 28, 2022). Among those eligible, only 42% had received at least one booster (Texas Department of State Health Services (DSHS), n.d.). Uptake rates have been previously shown to differ across demographic groups. For example, individuals who were non-Hispanic Black or Hispanic/Latino and fell within a lower income bracket were less likely to receive the booster as compared to their respective reference groups (Lu et al. 2023; Marus et al. 2022; Nguyen et al. 2022).

We investigated the demographic trends in booster uptake among the eligible population 5 years+ in 3 areas in Texas: Harris County, Cameron County, and Northeast Texas (NETX) counties. Our objective was to identify disparities in uptake to help support vaccination outreach efforts.

Methods & Models

Methods Overview

COVID-19 vaccination data up to February 2023 was obtained from the Texas Immtrac2 registry (Harris = 3,005,630; Cameron = 346,669; NETX = 283,747). Positive case data was obtained from local public health departments and merged with vaccination records. The likelihood of receiving a booster dose (Yes/No) was assessed using multivariable logistic regression including: age, race/ethnicity, gender, previous COVID-19 infection, and months between vaccine approval to first dose (a measure of vaccination hesitancy). To investigate interactions, we fit stratified models by age, previous infection, and both age and infection, and present the main effect results (odds ratios, OR) for race/ethnicity.

Models

The response variable was binary, indicating if an individual had received a booster dose (1) or had not (0). We included only individuals eligible to receive a booster dose.

List of Models:

  1. Model 1: Booster ~ Age + Race/Ethnicity + Gender + Previous Infection + Months to First Dose
  2. Model 2: Booster ~ (Age x Race/Ethnicity) + Gender + Previous Infection + Months to First Dose
  3. Model 3: Booster ~ Age + (Race/Ethnicity x Previous Infection) + Gender + Months to First Dose
  4. Model 4: Booster ~ (Age x Race/Etnicity x Previous Infection) + Gender + Months to First Dose

Data was then stratified by Age, Previous Infection, and both Age and Previous Infection. Models were refit with the remaining variables and the Race/Ethnicity coefficients were inspected and visualized in forest plots.

List of Stratified Models:

  1. Model 2: Booster ~ Race/Ethnicity + Gender + Previous Infection + Months to First Dose.
    1. Model 2a: 40 and over
    2. Model 2b: Under 40
  2. Model 3: Booster ~ Race/Ethnicity + Age + Gender + Months to First Dose.
    1. Model 3a: Previous infection
    2. Model 3b: No previous infection
  3. Model 4: Booster ~ Race/Ethnicity + Gender + Months to First Dose
    1. Model 4a: 40 and over with a previous infection
    2. Model 4b: 40 and over with no previous infection
    3. Model 4c: Under 40 with a previous infection
    4. Model 4d: Under 40 with no previous infection

Results

Overall vaccination and booster uptake varied by region (Harris: 64.96% & 29.7%; Cameron: 81.9% & 40%; NETX: 47.2% & 20.11%). Individuals under 40 years (Harris: OR 0.399, 95% CI [0.397, 0.401]; Cameron: 0.284, [0.279, 0.288]; NETX: 0.334, [0.327, 0.420]), with more months between approval to first dose (Harris: 0.708, [0.707, 0.709]; Cameron: 0.791, [0.789, 0.794]; NETX: 0.714, [0.711, 0.716]), and with a previous COVID-19 infection (Harris: 0.764, [0.758, 0.769]; Cameron: 0.719, [0.702, 0.737]; NETX: 0.530, [0.512, 0.549]) were less likely to have received a booster dose across all regions. Males and Hispanics had lower booster uptake in Harris (Male: 0.885, [0.880, 0.889]; Hispanic: 0.798, [0.792, 0.803]) and Cameron counties (Male: 0.828, [0.815, 0.840]; Hispanic: 0.693, [0.679, 0.797]). Black (0.665, [0.635, 0.696]) and American Indian/Alaskan Native (0.587, [0.451, 0.763]) individuals had lower booster uptake in NETX counties.

Stratified Models: Being over 40 and having a previous COVID-19 infection increased the likelihood of booster uptake in Black and Hispanic individuals in Harris and NETX counties (A, B).

Figures

Model 1

The forest plot of estimated odd ratios (OR) for Model 1 with no interactions. This figure is presented supplementary to the APHA poster.

Forest plot of the odds ratio estimates and associated 95% confidence intervals for each of the covariates in Model 1, including age, gender, race/ethnicity, previous infection, and months until first dose. The reference levels are 40 and over, female, Non-Hispanic White, and no previous infection, respectively. Odds ratio estimates are displayed for all three regions: Harris (yellow triangle), Cameron (blue circle), and NETX (green-gray square).

Figure 1. Forest plot of the odds ratio estimates and 95% CIs for Model 1

Forest Plot for Model 2 and 3 Interactions

A combined forest plot of estimated odd ratios (OR) for stratified models (2a, 2b, 3a, 3b) investigating two interactions: Age and Race/Ethnicity (left two panels) and Previous Infection and Race/Ethnicity (right two panels).

A four-paneled forest plot of estimated odds ratios for stratified models. All panels present the main effect results for the race/ethnicity covariate with Non-Hispanic White as the reference level. Odds ratio estimates are displayed for all three regions: Harris (yellow triangle), Cameron (blue circle), and NETX (green-gray square). The two left panels investigate the age times race/ethnicity interaction by presenting two stratified models. The leftmost panel presents main effects results of race/ethnicity for those 40 and over and the second left presents main effects for those under 40. The two right panels investigate the previous infection times race/ethnicity interaction by presenting two stratified models. The second right panel presents main effects for those with a previous COVID-19 infection and the rightmost panel presents main effects for those with no previous infection. Notably, in Harris County, Non-Hispanic Black individuals under 40 were less likely to receive a booster, but those 40 and over were more likely to receive the booster as compared to Non-Hispanic Whites. Similarly, in NETX, Non-Hispanic Black individuals with a previous infection were more likely to receive the booster while those without a previous infection were less likely to receive the booster as compared to Non-Hispanic Whites.

Figure 2. Paneled forest plot of odds ratio estimates and 95% CIs for Models 2a (left), 2b (second left), 3a (second right), and 3b (right)

Forest Plot for Model 4 Interaction

A forest plot of estimated odd ratios (OR) for stratified models (4a, 4b) investigating the three-way interaction between Age, Race/Ethnicity, and Previous Infection.

A four-paneled forest plot of estimates odds ratios for stratified models. All panels present the main effect results for the race/ethnicity covariate with Non-Hispanic White as the reference level. Odds ratios estimates are displayed for all three regions: Harris (yellow triangle), Cameron (blue circle), and NETX (green-gray square). The four stratified models presented investigate the three-way interaction between age, race/ethnicity, and previous COVID-19 infection. The leftmost panel presents main effects for those 40 and over with a previous infection. The remaining panels, from left to right, present main effects for those 40 and over with no previous infection, those under 40 with a previous infection, and those under 40 with no previous infection. Notably, in Harris County, Hispanic individuals who were 40 and over with a previous infection were more likely to receive a booster dose than Non-Hispanic Whites 40 and over with a previous infection. The opposite was observed in the other three stratified models. Similarly, in Harris County, among Non-Hispanic Black: those under 40 were less likely to receive the booster, regardless of infection status as compared to Non-Hispanic White individuals.

Figure 3. Paneled forest plot of the odds ratio estimates and 95% CIs for Models 4a (left), 4b (second left), 4c (second right), and 4d (right)

Conclusions

Significant disparities exist in COVID-19 booster uptake across Texas. Further, the specific observed patterns of disparity in uptake may differ by specific sub-regions within the state. This work highlights the need for tailored outreach and education efforts to ensure equitable and efficient vaccination efforts for COVID-19 and pandemics more broadly.

Acknowledgments

We acknowledge the important contributions of counties and staff for data cleaning and other work: Catherine Martin & Daniel Pinon for vaccination data, Gabriela Saucedo, Ashley Ruiz, & Raquel Castillo for testing and case data, and Samantha Thomas, Kendrick Boddie, & Russel Hopkins for additional work. This study was supported by NIH/NCATS 3UL1TR003167-02S1 and 3UL1TR003167-03S3, HRSA 6 U3SHS45319-01-03.6, and Cameron County Public Health Contract #0017402.

References

Centers for Disease Control and Prevention (CDC). March 28, 2022. “COVID Data Tracker.” Accessed: 2022-09-22 https://covid.cdc.gov/covid-data-tracker.
———. September 8, 2022. “COVID-19 Vaccination.” Accessed: 2022-09-22 https://www.cdc.gov/coronavirus/2019-ncov/vaccines/stay-up-to-date.html.
Lu, P, A Srivastav, K Vashist, CL Black, JL Kriss, MC Hung, L Meng, et al. 2023. “COVID-19 Booster Dose Vaccination Coverage and Factors Associated with Booster Vaccination Among Adults, United States, March 2022.” Emerg. Infect. Dis. 29 (1): 133–40.
Marus, J, N. Holtkamp, A Kolbe, and T Beleche. 2022. Demographic Characteristics of Adults Receiving COVID-19 Booster Vaccinations. Washington, DC: Office of the Assistant Secretary for Planning; Evaluation, U.S. Department of Health; Human Services.
Nguyen, KH, E Anneser, A Toppo, JD Allen, J Scott Parott, and L Corlin. 2022. “Disparities in National and State Estimates of COVID-19 Vaccination Receipt and Intent to Vaccinate by Race/Ethnicity, Income, and Age Group Among Adults > 18 Years, United States.” Vaccine 40 (1): 107–13.
Texas Department of State Health Services (DSHS). n.d. “Workbook: COVID-19 Vacccine in Texas (Dashboard).” Accessed: 2022-10-17.