Modelos para Dados Longitudinais de Contagem: uma aplicação para dados da COVID-19

Abstract

Since December 2019, when the first case of Covid-19 was confirmed in the city of Wuhan, China, many studies on the new coronavirus have been produced. The present work intends to study the behavior of the number of deaths in Brazilian cities with a population between 200 thousand and 1 million inhabitants according to IBGE projections for 2020. Based on the longitudinal data analysis and an application of a Poisson model for longitudinal data, we analyzed the relationship between the number of deaths by Covid-19 in the population and a set of explanatory variables, which are the number of confirmed cases, social mobility measures, estimated Rt, number of beds and available respirators. In this work, we considered as a time variable the epidemiological weeks, which start from the 1st week of 2020 to the 26th epidemiological week 2021. The results indicate that the estimated Rt, considering values from 5 weeks ago, is the predictor variable that has the most influence on the average number of deaths. And as expected, confirmed cases have a positive relationship with the average number of lives lost. Complementary beds, which are adult and pediatric ICU beds, had no significant influence on the number of deaths. It is also verified that the category of residential social mobility has a positive relationship with the cases of deaths, which indicates that when the number of deaths is high, people feel more motivated to stay at home.