Spatial analysis of COVID-19 risk factors: a case study in Bangladesh

Abstract

In 2019, the world grappled with an unexpected and severe global health crisis—the Coronavirus disease (COVID-19) outbreak, which significantly impacted various aspects of human life. This case study, focusing on Bangladesh, aimed to uncover the complex spatial patterns and potential risk factors influencing the virus’s uneven spread across 64 districts. To analyze spatial patterns, two techniques, namely Moran I and Geary C, were employed to study spatial autocorrelation. Hotspots and coldspots were identified using local Moran I, while spatial hotspots were pinpointed using local Getis Ord G. Exploring spatial heterogeneity involved implementing two non-spatial models (Poisson–Gamma and Poisson-Lognormal) and three spatial models (Conditional Autoregressive model, Convolution model, and Leroux model) through Gibbs sampling. The Leroux model emerged as the optimal choice, meeting criteria based on the lowest values of deviance information criterion and Watanabe–Akaike information criterion. Regression analysis revealed that factors such as humidity, population density, and urbanization were associated with an increase in COVID-19 cases, while the aging index appeared to hinder the virus’s spread. The research outcomes provide a comprehensive framework adaptable to the evolving nature of COVID-19 in Bangladesh. It categorizes influential factors into distinct clusters, enabling government agencies, policymakers, and healthcare professionals to make informed decisions for controlling the pandemic and addressing future infectious diseases.