The use of “excess deaths” in comparison to previous years to estimate accurate Covid-19 figures could be doomed in Brazil, suggests research by economist Thomas Fujiwara, an associate professor of economics at Princeton University. According to Mr. Fujiwara, vital data statistics are “not made for real-time analysis,” and researchers therefore must rely on data from Brazilian registry offices to assess the massive underreporting of Covid-19 in the country. Because this method is somewhat new, the researcher suggests data gathering should focus on municipal figures, as they are usually more detailed.
I wrote some slides on estimating excess deaths due to Covid-19 in Brazil – with an application to 17 municipalities with pop above 1 million.
— Thomas Fujiwara (@thomasfujiwara) June 4, 2020
My goal here is to also help other researchers work with a new data source and do their own analysis.
More on the thread below.
1/n pic.twitter.com/fZUi8UyvfU
Mr. Fujiwara’s initial analysis of Brazil’s 17 most populated cities also found specific particularities related to each. In places such as São Paulo, Rio de Janeiro, Manaus, and Fortaleza, “excess deaths” are nearly twice as high as the official numbers. In the southern cities of Porto Alegre and Curitiba, deaths have not varied much in comparison to previous years.
In the Northeast region, Mr. Fujiwara found that notaries could also be struggling with underreporting. Belo Horizonte and Brasilia did not have their data analyzed because, according to the economist, they had “too many issues to be reliable.”
Support this coverage →