Utilização de técnicas de séries temporais na análise de possíveis evidências de subnotificação ou supernotificação do número de mortes humanas no Brasil devido à Covid-19

Abstract

The exchange of digital information in recent years has grown in great scale. Each day people can easily produce and spread content/information on a mass scale, to anywhere in the world with just one click. This easy access to digital media, as well as the production and sharing of information, was crucial since the beginning of the COVID-19 pandemic. On the one hand, people have been using this social-technological media positively to run all sorts of errands as the world entered a quarantine period, on the other hand, it has been used to increase denial regarding the pandemic, mainly in the attempt to influence people to believe that the virus was less dangerous then what was disclosed by the public health authorities, as well as making the public suspicious of the number of deaths, claiming that people were dying from causes unrelated to COVID-19 but were registered as if they were. Based on this premise, this research, using a time series algorithm with non-stationary models (in particular, double moving averages), aimed to check whether there was overreporting/underreporting of the number of deaths recorded by the Brazilian Ministry of Health. For this, the database from the Information System on Mortality (SIM), from 2000 to 2020 was used, with the data grouped by year, number of deaths, and according to a chapter of the International Classification of Diseases, Tenth Revision (ICD-10). After the experiments, the prediction algorithm demonstrated, based on real and predicted data, that even the number of deaths in the ICD-10 chapter that includes those caused by COVID-19 was very high in 2020 due to the emergence of the pandemic; other causes of death practically remained at the same level as the previous year, some having increased and others slightly decreased, but not significantly that it could be confirmed that part of the cases had been wrongly registered as COVID-19. Therefore, from this analysis, it was possible to infer that there was neither overreporting nor underreporting regarding mortality during the period analyzed and that the numbers recorded by the Brazilian Ministry of Health represent, in fact, deaths related to COVID-19.