Highly Segregated Metropolitan Areas Cause COVID-19 Drivers to Act With Different Weights in Successive Pandemic Stages

Abstract

Background: The Covid-19 evolution has been intensively studied. However, the pathways of the virus spread between interdependent/segregated urban areas, and the impact of mobility restrictions on more comprehensive community transmission remains unexplained. Recent works have indicated a strong correlation between social determinants of health and the pandemic. However, these investigations take a static point of view, not considering that these determinants act differently in distinct urban areas and across the pandemic.Methods: We analyze the dynamic patterns of contagion in the Santiago Metropolitan Area (SMA) by articulating explanatory variables related to urban mobility, socio-spatial characteristics, segregation, and sanitary measures. We estimate mobility within the SMA through indices and correlate them with the contagion patterns, providing data-driven models for insights, analysis, and inferences. To avoid poorly determined coefficients in the applied regressions due to the many and correlated variables, we use Elastic-Net-Penalty. This method provides a feature selection by imposing a size constraint on the variable coefficients, thus allowing us to build a reduced subset of high-predictive variables.Findings: We recognized a three-stage pattern of the pandemic in the SMA, explained by the differentiated influence across the pandemic of different explanatory variables. By applying Elastic-Net-Penalty for the feature selection, the resulting subset of variables was different in each stage. The variables' importance indicators were also distinct, thus highlighting the role of different drivers in each stage. We also found evidence to explain the poor outcomes of the dynamic quarantine deployed by the authorities.Interpretation: The mobility restrictions affect the pandemic context. This context constitutes a dynamic/adaptive complex system. The patterns of contagions are influenced by interrelated drivers that act unequally depending on urban areas and the life rhythms. Our analysis indicates that these interventions were implemented late and without considering the aforementioned complex context, thereby damaging the city’s vulnerable areas.Funding Statement: This research was funded by ANID, grant number FONDECYT/11180569. AGF was partially supported by ANID/FONDAP/13150015.Declaration of Interests: Authors declare that they have no competing interests.