|The emergence of infectious diseases in an urban area involves a complex interaction between the socioecological processes in the neighbourhood and urbanization. As a result, such an urban environment can be the incubator of new epidemics and spread diseases more rapidly in densely populated areas than elsewhere. Most recently, the Coronavirus-19 (COVID-19) pandemic has brought unprecedented challenges around the world. Toronto, the capital city of Ontario, Canada, has been severely impacted by COVID-19. Understanding the spatiotemporal patterns and the key drivers of such patterns is imperative for designing and implementing an effective public health program to control the spread of the pandemic. This dissertation was designed to contribute to the global research effort on the COVID-19 pandemic by conducting spatial epidemiological studies to enhance our understanding of the disease's epidemiology in a spatial context to guide enhancing the public health strategies in controlling the disease.
Comprised of three original research manuscripts, this dissertation focuses on the spatial epidemiology of COVID-19 at a neighbourhood scale in Toronto. Each manuscript makes scientific contributions and enhances our knowledge of how interactions between different socioecological processes in the neighbourhood and urbanization can influence spatial spread and patterns of COVID-19 in Toronto with the application of novel and advanced methodological approaches. The findings of the outcomes of the analyses are intended to contribute to the public health policy that informs neighbourhood-based disease intervention initiatives by the public health authorities, local government, and policymakers.
The first manuscript analyzes the globally and locally variable socioeconomic drivers of COVID-19 incidence and examines how these relationships vary across different neighbourhoods. In the global model, lower levels of education and the percentage of immigrants were found to have a positive association with increased risk for COVID-19. This study provides the methodological framework for identifying the local variations in the association between risk for COVID-19 and socioeconomic factors in an urban environment by applying a local multiscale geographically weighted regression (MGWR) modelling approach. The MGWR model is an improvement over the methods used in earlier studies of COVID-19 in identifying local variations of COVID-19 by incorporating a correction factor for the multiple testing problem in the geographically weighted regression models.
The second manuscript quantifies the associations between COVID-19 cases and urban socioeconomic and land surface temperature (LST) at the neighbourhood scale in Toronto. Four spatiotemporal Bayesian hierarchical models with spatial, temporal, and varying space-time interaction terms are compared. The results of this study identified the seasonal trends of COVID-19 risk, where the spatiotemporal trends show increasing, decreasing, or stable patterns, and identified area-specific spatial risk for targeted interventions. Educational level and high land surface temperature are shown to have a positive association with the risk for COVID-19. In this study, high spatial and temporal resolution satellite images were used to extract LST, and atmospheric corrections methods were applied to these images by adopting a land surface emissivity (LSE) model, which provided a high estimation accuracy. The methodological approach of this work will help researchers understand how to acquire long time-series data of LST at a spatial scale from satellite images, develop methodological approaches for atmospheric correction and create the environmental data with a high estimation accuracy to fit into modelling disease. Applying to policy, the findings of this study can inform the design and implementation of urban planning strategies and programs to control disease risks.
The third manuscript developed a novel approach for visualization of the spread of infectious disease outbreaks by incorporating neighbourhood networks and the time-series data of the disease at the neighbourhood level. The findings of the model provide an understanding of the direction and magnitude of spatial risk for the outbreak and guide for the importance of early intervention in order to stop the spread of the outbreak. The manuscript also identified hotspots using incidence rate and disease persistence, the findings of which may inform public health planners to develop priority-based intervention plans in a resource constraint situation.