Long-term spatial patterns in COVID-19 booster vaccine uptake

Vaccination is a critical tool for controlling infectious diseases, with its use to protect against COVID-19 being a prime example. Where a disease is highly transmissible, even a small proportion of unprotected individuals can have substantial implications for disease burdens and compromise disease control. As socio-demographic factors such as deprivation and ethnicity have been shown to influence uptake rates, identifying how uptake varies with socio-demographic indicators is a critical step for reducing hesitancy and issues of access, and identifying plausible future uptake patterns. Here, we analyse the numbers of COVID-19 booster vaccinations subdivided by age, sex, dose and location. We use publicly available socio-demographic data, and use Random Forest models to capture patterns of first booster uptake at high spatial resolution, with systematic variation restricted to ∼ 1km in urban areas. We introduce a novel method to predict future distributions using our first booster model, assuming existing trends with respect to deprivation will persist. This provides a quantitative estimate of the impact of changing motivations and efforts to increase uptake. While age and sex have the greatest impact on the model fit, there is a substantial influence of community deprivation and the proportion of residents belonging to a black or minority ethnicity. Changes in patterns from first to second boosters suggest in the longer-term that the impact of deprivation is likely to increase, furthering the disproportionate impact of COVID-19 on deprived communities. Our analysis is based solely on publicly available socio-demographic data and readily recorded vaccination data, and would be easily adaptable to uptake data from countries where data recording is similar, and for aiding vaccination campaigns against other infectious diseases.