RIO DE JANEIRO, BRAZIL – Two months into the first quarantines in Brazil, politicians, health authorities, and researchers are mostly in the dark. The weeks of advantage Latin America had on Asia and Europe seem wasted. The tests to detect the coronavirus, which officials, scientists, and analysts said in mid-March were crucial for isolating patients and minimizing damage to health and the economy, never quite reached the required level.
The tests conducted by the most affected country – and the most populous in Latin America – are so few that the magnitude of the epidemic is unknown. There is an official balance: almost 19,000 deaths among the 271,000 cases recorded by Wednesday, May 20th. But there is also consensus that this is a somewhat underestimated figure. Yet Brazil is the most affected country on the continent after the US.
El País newspaper has estimated that only one in 20 cases is reported. This means that the number of infected individuals in Brazil is most likely to be 3.7 million when applying the method used to calculate underreporting by a group of mathematicians and epidemiologists from the London School of Hygiene and Tropical Medicine.
The calculations were made based on last Saturday’s official balance. By dividing confirmed deaths in Brazil on that day (15,633) by the cases detected (233,142), the result is that 6.7 percent of infected people die. But this figure is well above the most current studies on Covid-19 mortality, which place it around one percent.
London School researchers are investigating this difference between reported mortality and real mortality to determine underreporting. Estimating the actual number of cases is possible using the approximate real mortality rate (around one percent, according to current studies), comparing it with the rate observed in Brazil (6.7 percent) and taking into account the time between infection and death.
As it is likely that the disease is not equally lethal in all of the country’s regions or locations (due to factors such as the average age of those infected or access to hospitals, ICUs, or ventilators), the estimate assumes a wide range of uncertainty. It considers two other mortality rates: 0.6 percent in the least aggressive assumption and 1.4 percent in the most aggressive one. Hence, applying the mentioned model to official mortality data, it is estimated that the range of infected individuals varies from 1.8 million to 6.3 million, with 3,760,000 as the central reference number.
A study published a few days ago by a team of Brazilian specialists, employing a similar methodology, found it likely that approximately nine out of ten infected individuals would go unnoticed. The numbers provided here suggest higher underreporting: about 19 out of 20 cases. This would be consistent with the low rate of diagnostic tests in Brazil. According to data from Oxford University’s Our World In Data platform, Brazil is among the countries in the region with the lowest rate of tests per capita.
Most likely, the vast majority of these undetected infections have developed with mild clinical conditions.
The track of death certificates
Faced with the difficulty in screening all those infected through diagnostic tests, authorities and specialists try to supplement the estimates with data from death certificates. The Civil Registry of Brazil publishes daily the latest available data on deaths reported as a consequence of the coronavirus and those reported as suspected, due to pneumonia or respiratory failure. By May 16th, the National Civil Registry recorded 16,396 confirmed or suspected Covid-19 deaths throughout Brazil.
Should the total number of cases be estimated using the above-mentioned London School method, but based on the Civil Registry, the number of probable cases is significantly higher, and in any case, very close to that obtained using the official death data for the disease.
However, even the Civil Registry estimate may involve underreporting. The reason is that each local registry collects and enters information into its database at its own pace. That is why the excess deaths this year compared to the average recorded in previous years has become one of the reference data for assessing the incidence of the epidemic in a given place.
Researcher André Ricardo Ribas Freitas, a specialist in assessing excess mortality and professor at the São Leopoldo Mandic Medical School in Campinas (São Paulo), explains that data on the most populous cities are accurate since they are inserted virtually in real-time. Backlog issues may occur in municipalities with fewer than one million inhabitants.
Based on this, he prepared with other researchers a weekly estimate of the number of deaths in some of the most populous cities in Brazil: they compare Civil Registry data since the start of the pandemic with an expected number of deaths for that period. This statistical calculation uses the mortality observed in previous years as a reference. For most of these cities, Ribas Freitas notes that the figures for 2020 are significantly higher than the official death count by Covid-19.
The figures for Manaus, Fortaleza, São Paulo, and Rio de Janeiro were analyzed, four of the most active outbreaks of the pandemic in Brazil.
The difference between the official and excess death toll in these four cities totals 3,300 deaths since the start of the pandemic until the end of epidemiological week 19 (May 10th). If Covid-19 caused most of these deaths, the figure would far exceed not only the Ministry of Health’s data for each of the cities but also the estimate made by the Civil Registry of unaccounted suspected cases (about 600-700 by May 16th) for the entire country.
Given its scale, Brazil could hold in its territory twice the area of the European Union. That is why in times of pandemic, regional differences also matter. Cases are focused in the Southeast (around São Paulo and Rio de Janeiro), the Northeast (mainly Recife, Fortaleza, and São Luis) and the North (Manaus and its periphery).
Manaus, four times more deaths
The situation in Manaus, the central city in the Amazon, is particularly critical. In the last weeks of April, the Civil Registry data pointed to aggregate mortality that multiplied by up to four compared to previous years. The curve is now on a downward trajectory.
The State of Amazonas, three times the size of Spain, has a population of 3.5 million, most of them located in Manaus. The registries recorded a number of deaths much higher than the 1,413 confirmed coronavirus deaths. The Amazon health network includes 450 ICU beds, according to the Ministry of Health.
Fortaleza, 189 percent more in one week and upward curve
Fortaleza is located in an impoverished region and has one of the highest incidences of Covid-19 deaths in relation to the total population. The rate of excess deaths recorded in one week in early May is 189 percent, with the curve rising. Ceará, with Fortaleza as its capital, has a population of 8.4 million and a little more than a thousand beds in intensive care units in hospitals.
Rio de Janeiro, double in one week
Rio de Janeiro registers lower, but rather pronounced, increases: in the transition week from April to May, mortality multiplied by two compared to previous years. The State has approximately 4,000 beds for 17 million inhabitants.
São Paulo, small increases involving thousands of people
São Paulo shows less dramatic data. However, the city’s very dimension, where the metropolitan area is the most populous in the country, implies that even an excess of deaths in the range of 20 to 30 percent represents thousands of additional deaths compared to what is typical for this time of year. The capital’s ICUs have a 90 percent occupancy, and the State’s around 70 percent. The State, with a population of 45 million, has over 9,200 ICUs, half of them in the public network.
Source: El País