Conclusions
By analysing the evolution of the COVID-19 epidemic in Minas Gerais, Brazil, we showed the importance of considering sub-notification not only of deaths but also of infection cases. It was shown that the largely used criteria of a historical all-deaths series did not work in this case, where most of the deaths are associated with causes that should decrease under social distancing and reduction of economic activities. Other criteria for better estimating deaths, infected cases, and high mortality rate during the pandemic in Minas Gerais, discussing possible contributions and their implications were then presented. It is shown that the excess deaths and infected cases are related to the strong increase of SARS deaths and cases reported in 2020 compared to 2019. A new methodology to estimate excess deaths and cases was developed. A quite simple and intuitive model based on the Gompertz function was also used to well fit the data and to predict the evolution of the epidemic. Based on these analyses an excess of 21.638 deaths and 557.216 infected cases is predicted until the end of 2020, with an upper bound of the case fatality rate of around 2.4 % and a prevalence of 2.6 %. The importance of ethnic and geographical distribution of cases and deaths is demonstrated, and the discrepancies between cases and deaths in black and white patients are discussed. Government and private organizations need to work together to increase public awareness, stimulate social distancing, and keep a surface disinfection routine to curb the viral infection, especially in critical places. Particular attention should be given to people under social vulnerability, living in underdeveloped regions.