Vaccination intentions generate racial disparities in the societal persistence of COVID-19

Yanchao Wang, Alina Ristea, Mehrnaz Amiri, Dan Dooley, Sage Gibbons, Hannah Grabowski, J. Lee Hargraves, Nikola Kovačević, Anthony Roman, Russell K. Schutt, Jianxi Gao, Qi Wang, Daniel T. O’BrienScientific 2021 · DOI

TL;DR

This paper uses a simulation model of COVID-19 spread in the greater Boston area to demonstrate how racial disparities in vaccination intentions can lead to significant inequities in infection rates and herd immunity attainment. Even with vaccination, communities of color experienced disproportionately higher infection rates and delayed herd immunity, highlighting the critical need for targeted, compassionate public health messaging.

What Problem Does This Paper Address?

The COVID-19 pandemic highlighted existing societal inequities, and the rollout of vaccines presented a new frontier where these disparities could manifest. While vaccines offer a path to herd immunity, their effectiveness hinges on population uptake. This paper addresses the critical question of how pre-existing racial disparities in vaccination willingness, particularly among Black and Latinx communities, could impact the speed and success of achieving herd immunity and lead to persistent, unequal burdens of infection. The authors aim to quantify these potential disparities and understand the factors that might mitigate them.

Methods

The study employs a sophisticated SIR (Susceptible-Infected-Recovered) epidemiological model, enhanced with mobility data and vaccination parameters. The model simulates COVID-19 transmission across 128 communities (municipalities and ZIP codes) in the greater Boston area. Key methodological innovations include:

  • Mobility Integration: Incorporating daily mobility data derived from anonymized cell phone GPS pings to simulate the movement of individuals between communities, thus modeling disease spread beyond local boundaries.
  • Vaccination Dynamics: Modeling the gradual rollout of a vaccine with a specified efficacy (95%) and a weekly uptake rate. Crucially, the model includes a "persuasion" component, where individuals uncertain about vaccination can be convinced to get vaccinated based on the proportion of vaccinated individuals in their community.
  • Racial Disparity Modeling: Directly integrating survey data on vaccination intentions, stratified by race, to inform the initial willingness to vaccinate within different community demographics.
  • Counterfactual Analysis: Comparing simulation outcomes with and without vaccination to isolate the impact of the vaccine rollout.

Datasets

The study utilizes several key datasets:

  • Population Descriptors: U.S. Census' American Community Survey's 2014-2018 estimates for demographic composition (including ethnic composition) at the census block group level, aggregated to municipal or ZIP code level.
  • Infection Data: Daily and weekly COVID-19 infection case counts for towns in greater Boston and ZIP codes within Boston, obtained from the Massachusetts Department of Health and the Boston Public Health Commission.
  • Vaccination Intentions: Responses from three independent surveys of Massachusetts residents conducted in late 2020, which captured intentions to vaccinate, broken down by race.
  • Mobility Data: Daily "Social Distancing" dataset from SafeGraph, which aggregates anonymized GPS pings from mobile devices to estimate movement patterns between communities.

Key Insights and Ideas

  • Vaccination Bottleneck: The simulation revealed that vaccination programs quickly hit a "bottleneck" where the supply of willing recipients dwindled, and further uptake relied heavily on persuasion. This bottleneck occurred earlier in communities with higher Black and Latinx populations due to lower initial willingness to vaccinate.
  • Persuasion's Dual Role: While persuasion was crucial for closing infection gaps, its effectiveness was limited because a significant portion of Black and Latinx individuals stated they would "never" vaccinate, rendering them unresponsive to persuasion efforts.
  • Mobility's Diminished Role (During Vaccination): During the vaccination period, the impact of mobility-based exposure on infection rates was moderate and largely supplanted by the effect of vaccination itself, suggesting that localized vaccination uptake was more critical than inter-community travel for controlling spread.
  • Compounding Inequities: The study strongly suggests that existing racial disparities in vaccination willingness directly translate into persistent racial disparities in infection rates and the time to achieve herd immunity, exacerbating pre-existing health inequities.

Key Results

  • Infection Rate Disparity: Communities of color experienced infection rates approximately 3 times higher than predominantly White communities.
  • Herd Immunity Delay: Communities of color reached herd immunity on average 45 days later than predominantly White communities.
  • Vaccination Bottleneck Timing: The vaccination bottleneck occurred approximately 2 weeks earlier (mean day 42) in high Black-Latinx communities compared to predominantly White communities (mean day 57).
  • "Never Vaccinate" Proportion: Approximately 1/5th (20%) of Black and Latinx individuals indicated they would "never" vaccinate.
  • Persuasion Impact: Increasing persuasion rates could narrow disparities but could not eliminate them entirely due to the "never vaccinate" segment. Removing persuasion entirely led to a scenario where no high Black-Latinx communities reached herd immunity within the simulated timeframe.

Limitations and Future Work

The authors acknowledge several limitations:

  • No First Responder Rollout: The model did not account for an initial rollout to first responders and at-risk populations, which could have accelerated herd immunity.
  • Vaccinated Individuals as Non-Infectious: The model assumed vaccinated individuals could not carry the virus, a simplification that may not hold true with new variants like Delta.
  • Simplified Persuasion Model: The model did not fully capture the power of leadership in persuasion, particularly in communities of color.
  • Stable Reproduction Rate: The model assumed a stable reproduction rate, whereas vaccination might influence behavioral shifts that alter this rate.
  • Even Vaccination Access: The model assumed equitable access to vaccination sites, which is not reflective of reality, as communities of color often face barriers.

Future work could explore the impact of new variants, incorporate more nuanced persuasion models, and account for varying vaccination access. The authors emphasize the need for well-crafted, compassionate messaging that specifically addresses the concerns of those most resistant to vaccination, particularly in communities of color. The study's findings are particularly relevant given the emergence of more virulent variants and the ongoing challenges in achieving equitable vaccination coverage.

Summary updated

Abstract

We combined survey, mobility, and infections data in greater Boston, MA to simulate the effects of racial disparities in the inclination to become vaccinated on continued infection rates and the attainment of herd immunity. The simulation projected marked inequities, with communities of color experiencing infection rates 3 times higher than predominantly White communities and reaching herd immunity 45 days later on average. Persuasion of individuals uncertain about vaccination was crucial to preventing the worst inequities but could only narrow them so far because 1/5th of Black and Latinx individuals said that they would never vaccinate. The results point to a need for well-crafted, compassionate messaging that reaches out to those most resistant to the vaccine.
BibTeX
@inproceedings{wang2021vaccination,
  author = {Yanchao Wang and Alina Ristea and Mehrnaz Amiri and Dan Dooley and Sage Gibbons and Hannah Grabowski and J. Lee Hargraves and Nikola Kovačević and Anthony Roman and Russell K. Schutt and Jianxi Gao and Qi Wang and Daniel T. O’Brien},
  title = {{Vaccination intentions generate racial disparities in the societal persistence of COVID-19}},
  booktitle = {Scientific},
  year = {2021},
  doi = {10.1038/s41598-021-99248-2},
}
Figure