EXCESSO DE MORTALIDADE NO BRASIL EM TEMPOS DE COVID-19 EXCESS MORTALITY IN BRAZIL IN TIMES OF COVID-19

Given the growing number of deaths due to the COVID-19 pandemic in Brazil, this study presents an initial and exploratory descriptive analysis of the excess mortality observed from March to May 2020 in capitals and other municipalities. The data source was the death registers from the Civil Registry Offices. The data were disaggregated by gender and capitals and other municipalities of the 26 federative units and the Federal District. The standardized mortality ratio for 2020 was calculated with the 2019 mortality coefficients as standards. The results showed 39,146 excess deaths for the period studied and is higher among men. This increase was more significant among the capitals of the North, Northeast, and Southeast regions. In the other municipalities in these regions, the increase was observed in May, indicating a possible inland-bound COVID-19 transmission. The need to improve the detection and registration of cases is highlighted to enable the efficient monitoring of the pandemic.


INTRODUCTION
The COVID-19 pandemic is so far the most significant health challenge of this century, causing more than 9 million cases and 470,000 deaths worldwide (June 22) 1 . The disease was first described in China in late 2019 and quickly spread worldwide 2 . Since then, many efforts have been made in several countries to control the epidemic, but global growth continues 3 .
In Brazil, the first case of COVID-19 was recorded on February 26 4 . Since then, more than one million cases and 50,000 deaths have been reported, placing Brazil as the country with the second-highest number of cases and deaths in the world 1,5 . However, the number of reported cases and deaths is highly dependent on the testing policy adopted. Some countries only test patients in need of hospitalization, others recommend testing all those with symptoms, regardless of the need for hospital care, and some countries still implement mass testing.
In Brazil, in particular, the outlook for mitigating the epidemic is not favorable due to unfavorable political scenarios and the lack of national planning integrated with states and municipalities. Considering that a significant number of cases of the disease develop unfavorably, understanding the burden of COVID-19 in the country's mortality profile should be among the priorities in facing the various epidemics since the first suspected case of the disease was recorded.
The use of data from existing information systems in Brazil can provide a good benchmark to monitor the epidemic, define prevention and control measures and assess the impact on the weight that this new disease has caused in the country's morbimortality.
This study aimed to describe in an exploratory way the relationship between deaths observed in the months in which the COVID-19 epidemic arrived in Brazil so far, compared with the number of expected deaths concerning 2019 by capital and inland in the federative units and the Federal District by gender.

METHODS
The data source for this study was the registry of deaths from natural causes informed by the Civil Registry of Natural Persons and made available by the National Association of Natural Persons Registrars (ARPEN) on its freely accessible website 6 . The information on this website is updated daily with the sending of the data registered by the notary offices at the Civil Registry Information Center (CRC). Considering the legal deadlines for sending information 7 , the set of deaths between March 1 to May 31, 2019 and 2020 was considered, with data obtained on June 30, 2020. Municipal population projections by gender and age range calculated by Freire et al 8 were employed.
The number of deaths from all the natural causes was analyzed monthly, broken down by capitals and other municipalities in the 26 federative units and the Federal District by gender. The standardized mortality ratio (SMR) for 10-year age groups for 2020 was calculated using the mortality coefficients of the corresponding months and locations in 2019 as standards. The 95% confidence intervals for each SMR were calculated assuming a Poisson distribution, as described by Breslow and Day 9 .
The number of excess deaths was calculated as the difference between the number of deaths observed and the number of expected deaths, from the mortality coefficients in the same months of 2019, applied to the population projected for 2020 in the respective locations. The negative monthly values were defined as zero for the calculation of the total by location. States with extreme values for SMR were excluded by visual inspection for the graphical representation of the data.
The analyses were performed using STATA/MP version 14.2 10 .

DISCUSSION
The total of excess deaths estimated in this study between March and May 2020 (39,146) across the country was 33.5% higher than the number of deaths accumulated by COVID-19 as of May 31, as reported by the Ministry of Health (29,314 deaths) 11 . In the state capitals, the excess deaths for the period studied were 124% higher than in the other municipalities in the country, ranging from 3 in Porto Alegre to 5,527 in São Paulo.
A continued increase was observed in the standardized mortality ratios for each month analyzed in 2020 in the places where increased mortality was noted compared to the same months of the previous year. This increase coincided with the escalating epidemic in the country. The compared standardized mortality ratios for the states, separating capitals from other municipalities, allows observing a great regional variation with higher magnitude in the capitals and more significant excess among men from March to May 2020.
However, in May, the inland region's standardized mortality ratios became more expressive, which may be associated with the inland-bound transmission of COVID-19. This is consistent with the introduction of SARS-COV-2 that started in the capitals but has been spreading inland. The mortality profile changed initially in the capitals of the Southeast, North, and Northeast regions. The epidemic shift allowed perceiving excess deaths in municipalities outside the capitals, which may reflect the arrival of COVID-19.
The findings of this study identify excess deaths among men, compared to what was observed among women, which agrees with what has been reported in the literature 14 . However, attention is drawn to the need to analyze more robust estimates, assessing the absolute and relative differences to understand this issue better 15 . The need to improve the detection and registration of cases is urgent, and health authorities are responsible for providing conditions for this to be done with regularity, daily updating, and transparency to ensure the pandemic's monitoring. The maintenance and constant improvement of our health information systems must be seen as a priority in all spheres of government, lest we lose what has been built for several decades with so much effort and has been the most consistent pillar for disease control in the country.