Analysis of Regression Models for State Mortality Rates from Covid-19 in Brazil
COVID-19, GJS regression, PL regression, rates and proportions
The Ministry of Health received the first notification of a confirmed case of COVID-19 in Brazil on February 26, 2020. Shortly thereafter, on March 20, 2020, community transmission of the disease by COVID-19 was declared throughout the national territory. Brazil was one of the countries most impacted by the COVID-19 pandemic, with a cumulative total of 712,090 deaths. The country's federative units were affected unevenly by the pandemic and adopted combat measures with different proposals and intensities. Regression models of the distribution classes GJS, Generalized Johnsons SB, and PL, power logit, introduced, respectively, by Lemonte and Bazán (2016) and Queiroz and Ferrari (2023) are presented as alternatives to the model proposed in Cribari-Neto (2023) for estimating mortality rates due to COVID-19 in the Brazilian Federative Units. This work aims to present a new contribution applied to the study of comparing regression models for rates and proportions with the GJS and PL model classes. In addition, it presents a good model for estimating the proportion of deaths from Covid-19 in Brazilian states, which is extremely important in decision-making and in the measures taken by a state government not only during a crisis such as the Covid-19 pandemic, but also in preparing for future crises.