Understanding the COVID-19 pandemic through bayesian spatio-temporal modeling of several outcomes

Abstract

Understanding the spatio-temporal dynamics of past pandemics and the factors driving these patterns can enhance preparedness against future pandemics. This study aimed to investigate the COVID-19 pandemic by analyzing the spatio-temporal variations in infections, hospitalizations, deaths and reinfections.

We conducted a retrospective analysis of data from the adult population of the Basque Country at Primary Care Unit level from March 1, 2020 to January 9, 2022. Using a Bayesian hierarchical spatio-temporal model, we estimated relative risks for each outcome, accounting for the effects of a deprivation index, urbanicity, and COVID-19 testing rates.

SARS-CoV-2 infections and mortality followed similar risk patterns, with a strong clustering in highly populated areas. Hospitalization risks were influenced by proximity to hospitals, revealing potential access barriers in remote areas. High reinfection risks were predominantly localized in the northwest coast of our region. Increased testing rates were associated with higher risks across all outcomes. Urbanicity showed positive associations with hospitalizations (relative risk, [95 % credible interval]: 1.22, [1.12–1.33]) and infections (1.34, [1.15–1.57]). Similarly, deprivation was positively associated with hospitalization risks (1.09, [1.04–1.15]) and mortality risks (1.07, [1.02–1.12]), reflecting the increased vulnerability of socioeconomically disadvantaged populations.

This comprehensive analysis of various COVID-19 outcomes provides valuable insights into the pandemic’s spatio-temporal dynamics and highlights key improvement areas. Addressing healthcare access disparities in rural areas and focusing on the deprived populations could help mitigate the impact of future pandemics. This approach could be extended to other regions to inform specific public health strategies.