Explaining COVID-19 contagion in Portuguese municipalities using spatial autocorrelation models

Abstract

This paper investigates the pattern of COVID-19 contagion in Portuguese municipalities from March 23rd to April 5th (the exponential phase). We have recurred to spatial autocorrelation models to discuss how the conglomeration of highly infectious spaces has also contributed to infecting neighbouring municipalities. We have used several indicators for the contagion of COVID-19 from the number of infectious individuals to rates of infectious. As explicative variables, additionally to spatial proximity, we also considered population density, the share of the elderly population as well as the length of municipal perimeter/border. Our results show that highly dense municipalities tended to contaminate close areas. Lengthier perimeters also showed a positive effect on the contagious indicators for a given municipality.

Spatiotemporal Dynamics of COVID-19 Infections in Mainland Portugal

“…Municipal contiguity was also relevant to the contagion during this wave, suggesting that inter-municipal mobility was a determinant factor [79]. We also verified the occurrence of the highest peak of the border effect to be around the more densely populated municipalities, close to the border with Spain [42,81].…”

“…MAs, especially the MAP, and contiguous municipalities emerged as regional epicentres during the first wave, as well as some district capitals and surrounding urban municipalities (e.g., Braga and Aveiro), which were characterized by high population density and concentration of employment in industry [42]. Much of the inland municipalities (less populated) stand out for being relatively free of transmission and being surrounded by other municipalities with low incidence, except for occasional situations where specific circumstances justified a higher rate of incidence.…”

“…The importance of spatial association techniques in the study of the incidence of COVID-19 was reinforced by [4,41,42]. These authors used spatial autocorrelation techniques to measure the spatial dependence of disease incidence.…”

“…Refs. [42,47] used the application of this type of indicator to demonstrate how proximity and contiguity are determining geographic factors in the spread of SARS-CoV-2.…”

Spatiotemporal Analysis of Covid-19 in Turkey

“…The usefulness of the SLM is that it allows a clear distinction to be made between spatial similarity in the dependent variable and spatial similarity in the explanatory variables. The SLM in our study will try to explain the distribution of the Covid-19 variable by adding the prevalence observed in the surrounding provinces (Mourao & Bento, 2021). The study uses the GeoDa software, which provides data visualization, and allows global and local spatial autocorrelation calculations and spatial regression estimations (Anselin, 1995(Anselin, , 1996.…”