Vaccine hesitancy is one of the critical challenges for the implementation of a successful vaccination strategy. Rates of vaccine hesitancy and refusal vary substantially across different socioeconomic groups, and can result in those considered most vulnerable to disease having the lowest levels of uptake. Widespread coverage of COVID-19 vaccination is of particular importance as prevalence remains high, in effort to reduce overall burden from serious disease. Scotland's COVID-19 vaccination programme has progressed to booster vaccinations, however uptake is falling across successive doses, and there is concern that some vulnerable individuals will not have sustained protection.
To this end we analyse uptake in Scotland's first (starting September 2021) booster dose round, as a benchmark for future rounds. We fit a machine learning model to explain variation in uptake across Scotland at fine population scales. The model is able to estimate a neighbourhood's booster uptake with high precision using its population structure and relative deprivation alone, without any knowledge of geographic location. This is indicative of a strong relationship between increasing local deprivation and falling uptake, and specifically in those failing to return for a booster, despite getting a first dose. Geographically, this manifests as clusters of lower uptake, coinciding with communities with higher deprivation.
With an upcoming booster rollout in Autumn 2022, we use first booster uptake as a baseline, to generate a set of plausible distributions for future uptake, if nationwide uptake were to fall. We make the core assumption that as uptake falls, trends with respect to deprivation will persist. Projected uptake declines more rapidly in clusters of more deprived neighbourhoods. If these projected distributions were to manifest, gaps in immunity would emerge in more deprived communities, which have historically had the highest rates of COVID-19 hospitalisation and mortality.