Excess Mortality in Brazil: A Detailed Description of Trends in Mortality During the COVID-19 Pandemic

Download in: English & Portuguese

Fatima Marinho¹, Ana Torrens², Renato Teixeira², Elisabeth França³, Ana Maria Nogales⁴, Daisy Xavier⁵, Thomas Fujiwara⁶

Executive Summary

Introduction

The two most common indicators used to monitor the evolution of the COVID-19 epidemic are the number of confirmed COVID-19 cases of and the number of deaths caused by COVID-19. However, identifying COVID-19 cases and deaths is a challenge in Brazil and elsewhere due to limitations in testing. State and municipal health departments and public health laboratories are continuously working to increase their testing capacity. Because testing is frequently limited to symptomatic or hospitalized cases, it is difficult to follow the trends of illness and death directly caused by the virus. Complementary analyses are crucial for monitoring the pandemic’s trajectory.

To address this challenge, WHO and other organizations recommend monitoring and evaluating the total number of deaths regardless of cause. This simple method reveals both the direct and indirect impact of the COVID-19 pandemic, since it includes not only confirmed deaths by COVID-19, but also those assigned to other causes—excluding external as injuries, falls, accidents, etc. The burden on health systems due to severe cases of COVID-19 can lead to higher mortality rates overall due to lack of timely and adequate care for other conditions. Patients’ fear of seeking health care due to perceived risk of COVID-19 infection has increased home deaths, contributing to a growth in population mortality.

Excess mortality is the extent to which the number of deaths within a jurisdiction exceeds the expected toll for the year based on previous years or predictions. This indicator allows us to follow the trajectory of the pandemic and grasp its magnitude. In conjunction with other key indicators, as number of cases and tests, excess mortality can provide a strong foundation for epidemic control measures.

Methodology

The following is a descriptive analysis of deaths from diseases, or ‘natural causes’, registered in Brazil between January 2015 and epidemiological week (EW) 23 of 2020 (ending on 06/06/2020). The data for the construction of the historical series of deaths from 2015 to 2019 (and estimated deaths for 2020) was gathered by the Mortality Information System (SIM) of the Ministry of Health; the data for the current year 2020 was sourced from the Civil Registry Transparency Portal of the Civil Registry Information Center (CRC). To account for the under-registration of deaths in the CR in the year 2020, a correction factor was applied when necessary based on the ratio between deaths registered in the SIM and in the CR in 2019.

Observed mortality trends were used to estimate the number of deaths expected per epidemiological week in the current year 2020 as well as excess mortality. The expected number of deaths was calculated based on a prediction model that followed the trend observed in the last 5 years by location, sex and age. Excess mortality was measured between the 2020 projected point estimate and the point estimate measured in the corresponding epidemiological week of 2020. The period of measurement began with the first confirmed death by COVID-19 in Brazil—registered in EW 12—and ends at EW 23.

Main findings

In the 12 weeks between March 15, 2020 (EW 12) and June 6 (EW 23), 2020 there were an estimated 332,997 deaths from diseases —an excess of 62,490 deaths or 22% more than expected for the time period (Graph 1).

There was a sustained excess of death in relation to expected figures throughout the country from EW 12 to EW 23, with a dramatic increase starting in EW 17 and peaking in EW 19. Up to EW 23, the most affected regions of the country were the Southeast regions, Northeast and North. During this period, deaths were higher than expected by 20% (25,883) in the Southeast region and 31% (22,860) in the Northeast region. In the North region, there were 59% (10,779) more deaths than expected in the same period, an excessively higher proportion compare to other regions in the country, even with a smaller absolute number of deaths (Graph 2).

In state capitals, the excess mortality was proportionally even greater. In the period between EW 12 and 23, there were 40,333 more deaths in Brazilian state capitals than expected for the period, which corresponds to an excess of 48% of deaths (Graph 3). Although other state capitals also have a higher number of deaths than expected for the period investigated, eight Brazilian state capitals represent most of the total excess of deaths in the country: Rio de Janeiro and São Paulo state capitals (Southeast region), Fortaleza, São Luís, Recife and Salvador (Northeast region), and Manaus and Belém (North region) (Annexes, Graph 10).

Deaths of people under 60 increased by 48% in Brazil’s state capitals

In the last 5 years, an average of 74% (4,198,701 / 5,690,917) of deaths due to natural causes in the country occurred among those aged 60 or over (60+). As expected, most of the deaths that occurred in 2020 are concentrated in this age group. In the period from EW 12 to 23, there were 44,546 (21%) more deaths than expected for the period among the elderly. However, there was also a significant excess of deaths among people under 60 years old in the current year, proportionally equal to or even greater than among the elderly in some places, registering 26% (17,943) of deaths above the expected toll in this age group (Graph 6). In the same period from EW 12 to 23, in the state capitals of the country, where excess mortality was highest, there were 48% (29,502) more deaths than expected among the elderly. An equal level of excess mortality was observed among those under 60 years of age, 48% (10,831) higher than expected (Graph 7).

56% more men are dying in Brazil’s state capitals

From 2015 to 2019, an average of 52% (2,967,917 / 5,690,917) of deaths due to natural causes in Brazil were men. The excess mortality observed in 2020 was also greater among men: 26% (38,078) above expectations in the period from EW 12 to 23, while among women this number was 18% (24,412) for the same period (Graph 4). In Brazil’s state capitals, where excess mortality was considerably higher, the excess of deaths in males and females, in the same period was 56% (23,488) and 39% (16,846), respectively (Graph 5).

Conclusions

There is an excess of deaths across Brazil, beginning around the week when the first confirmed death occurred due to COVID-19 in São Paulo, reaching a peak in mid-April/May and persisting until June. From March 15 to June 6, more than 62,000 (22%) deaths from disease (natural causes) would not have occurred but for the presence of the epidemic.

People under 60 also died in excess; more than 16 thousand lives were lost. In the state capitals, it was worse; a 48% (10,831) excess in deaths was observed. Men died at higher rates than women, male mortality was 56% (23,488) above expected levels in state capitals.

Two regions of the country, the Midwest and South, did not show an excess of deaths until June 6, but conditions there are starting to worsen.

The excess mortality in Brazil points to the seriousness of the COVID-19 outbreak in that country through its direct and indirect impacts on mortality. Only strong public policy that coordinates pandemic containment interventions can avert thousands of preventable deaths from continuing to befall this nation.

1. Vital Strategies – Senior Technical Advisor – corresponding author – fmarinho@vitalstrategies.org

2. Vital Strategies – Technical Advisors and Research Consultants

3. Federal University of Minas Gerais (UFMG) – Associate Professor – Faculty of Medicine

4. University of Brasília (UnB) – Statistics Department

5. Federal University of Minas Gerais (UFMG) – Researcher – Epidemiology and Evaluation of Health Services Research Group – Faculty of Medicine

6. Princeton University – Associate Professor of Economics and Brazil LAB Associate Director – Department of Economics