Spatio-temporal modelling of COVID-19 infection and associated risk factors in Dakar, Senegal

Infectious diseases are a major threat to global health and economy and the recent COVID-19 pandemic is a perfect example of this. Appropriate modelling and accurate prediction of the outcome of disease spread over time and across space are critical steps towards informed development of effective strategies for public health interventions. In low and middle-income countries, however, the scarcity of spatially disaggregated time-series infectious diseases data often limits the analysis of the burden of infectious disease at a broad-scale, and the effects of the contextual risk factors is not often fully captured. In this study, we investigate the spatio-temporal patterns of COVID-19 infection in Dakar at the neighbourhood level, and evaluate the impact of potential risk factors. Geostatistical models based on COVID-19 infection data were used to explain and predict the spatio-temporal distribution of infections between June 2020 and June 2021. We specified a Bayesian regression model that incorporates a spatio-temporally autocorrelated random effect in order to quantify the evolution of the spatial patterns of the COVID-19 infection overtime. Results show significant strong spatial heterogeneity but relatively small temporal variations of the COVID-19 distribution, and a positive association between adjusted population density (mean of the posterior probability: 0.29, credible interval: 0.24-0.34) and residential areas (mean of the posterior probability: 1.25, credible interval: 0.66-1.83) with COVID-19 infection. Western areas are at higher risk of COVID-19 infection compared to eastern and less densely populated peripheral neighbourhoods. Measuring the role of contextual risk factors and mapping the at-risk areas can provide valuable insights for policymakers, enabling more targeted public health interventions. These efforts also support the management of endemic diseases and preparedness for future outbreaks.