Back

Assessing COVID-19 pandemic excess deaths in Brazil: years 2020 and 2021

Colonia, S. R. R.; Cardeal, L. M.; de Oliveira, R. A.; Trinca, L. A.

2022-07-29 epidemiology
10.1101/2022.07.27.22278096 medRxiv
Show abstract

We estimated the impact of the COVID-19 pandemic on mortality in Brazil for 2020 and 2021 years. We used mortality data (2015-2021) from the Health Ministry, the Brazilian government, to fit linear mixed models for forecasting baseline deaths under non-pandemic conditions. An advantage of the linear mixed model is the flexibility to capture year-trend while dealing with the correlations among death counts over time. Following a specified model-building strategy, estimation of all-cause excess deaths at the country level and stratified by sex, age, ethnicity and region of residence, from March 2020 to December 2021. We also considered the estimation of excess deaths due to specific causes. The estimated all-cause excess deaths was 187 842 (95% PI: 164 122; 211 562, P-Score=16.1%) for weeks 10-53, 2020, and 441 048 (95% PI: 411 740; 470 356, P-Score=31.9%) for weeks 1-52, 2021. P-Score values ranged from 1.4% (RS, South) to 38.1% (AM, North) in 2020 and from 21.2% (AL and BA, Northeast) to 66.1% (RO, North) in 2021. Differences among men (18.4%) and women (13.4%) appeared in 2020 only, and the P-Score values were about 30% for both sexes in 2021. Except for youngsters (< 20 years old), all adult age groups were badly hit, especially those from 40 to 79 years old. In 2020, the Indigenous+East Asian population had the highest P-Score (26.2%), and the Black population suffered the greatest impact (34.7%) in 2021. The pandemic impact had enormous regional heterogeneity and substantial differences according to socio-demographic factors, mainly during the first wave, showing that some population strata benefited from the social distancing measures when they could adhere to them. In the second wave, the burden was very high for all but extremely high for some, highlighting that our society must tackle the health inequalities experienced by groups of different socio-demographic status.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.