An Ecological Model of COVID-19 Space-Time Disparities in Vaccination Rates for Texas

Keywords: COVID-19 Vaccine, Bayesian spatio-temporal model
Abstract Type: Paper Abstract

Abstract

Although being fully vaccinated is one of the most effective means to prevent the spread of COVID-19 as well as to mitigate any infection complications, vaccination coverage lags behind the goal of developing herd immunity with noticeable spatial and temporal inequalities. The propensity to become fully vaccinated may be caused by local and temporal variations of vaccine hesitancy and acceptance and to a lesser degree accessibility issues. Relevant risk, as well as protective factors, were drawn from the cultural, political, public health, demographic, risk perception, regional, and socio-economic domains in each county in Texas. Furthermore, control variables linked to specific demographic groups are incorporated into this study. These spatial disparities are affected by availability, eligibility age, and COVID-19 variants.
The main objectives of this study are to identify the underlying determinants of the space-time vaccination disparities and to evaluate any shifts in the relevance of the risk factors over time. An ecological Bayesian mixed effects model, which accounts simultaneously for the interaction among both the temporal and spatial dimensions, is built. The weekly vaccination prevalence – having received at least two rounds of inoculations – for the eligible population residing in the 254 counties of Texas is evaluated within the time frame from mid-April, 2021, to the end of January 2022. The results of the model indicate that a conditional autoregressive spatial random effect as well as a first-order autoregressive serial random effect specification – due to the cumulative nature of the target variable – are highly relevant.