RIO DE JANEIRO, BRAZIL – The growth in the number of deaths by Covid-19 in Brazil slowed down, in a pattern similar to that of large European countries, until about 15 days ago. Since then, the rate of increase in mortality has slowed much less than in Europe.
The official number of deaths was rising at 6.5 percent per day on Friday, May 8th, in Brazil. On the same day, in Italy, this increase stood at 3.1 percent. Two weeks ago, the rates in both countries were similar. In France, the rate stands at 2.5 percent. In the United Kingdom, it is three percent. In the USA, it is 8.2 percent. In the state of São Paulo, it now stands at 4.6 percent.
This is not the variation in the absolute number of deaths per day, which was 751 on Friday, for instance. It refers to the percentage increase in the number of deaths on a certain day in relation to the total number of deaths the day before.
If the Brazilian rate had followed that of Italy, as seemed to be the case 15 days ago, the total number of deaths here would have been approximately 2,500 lower than the figure recorded up to Friday, of 9,897: “it is simply an arithmetic exercise”.
“The result for Brazil is quite worrying, more than that recorded in São Paulo. All the progress obtained from the quick implementation of distancing policies may be lost if the measures are relaxed uncontrollably. If this occurs, the expected reduction in the number of deaths, observed in many countries so far, may happen later or to a lesser extent in Brazil,” says Pedro Hallal, epidemiologist and rector of the Federal University of Pelotas.
According to epidemiologist Paulo Lotufo, there is no indication of a downward shift in the death curve in Brazil, quite the opposite (excluding São Paulo’s cases), but in São Paulo this shift seems to have occurred nine days ago. In this calculation, the variation in the number of daily cases was considered.
However, the data precariousness and variation recommend caution, says Lotufo, skeptical in relation to the statistics of the disease. For now, reliable data would only be that of total deaths (for any cause). The lack of detailed examination of fatal cases and the fact that the coronavirus causes deaths in a surprisingly varied manner hinders the classification of causes of death, he says.
Lotufo runs the Clinical and Epidemiological Research Center at the University of São Paulo (USP) and is a professor of medicine at the same university. He also commented on the data prepared by the Folha newspaper report based on information from a software that analyzes the curves and their trends.
The rate of increase in the number of deaths may have ceased to drop more rapidly in Brazil because more tests were conducted on those who died from the disease because the epidemic was uncontrolled or due to a combination of the two factors. It cannot be ascertained for the time being.
One way to speculate on the cause would be to check the general increase in the number of deaths and those due to SARS, but the record of any one death can take up to two weeks to reach the notary and government registries.
Although they believe that the number of deaths is the least inaccurate statistic at present, other epidemiologists consulted would rather wait for sample infection survey data before proceeding to public testing.
Three of them believe that, based on data over the past 15 days, there seems to have been a lack of control of the disease, although not in São Paulo, the South and Midwest, and the effect of a more explosive onset of the epidemic in northern and northeastern states. A month ago, 55 percent of Covid-19 deaths occurred in São Paulo territory. Now, it stands at 35 percent.
The rate of increase in the number of deaths per million inhabitants has also dropped less in Brazil than in large European countries, Canada, Iran, or China, which was not always the case from the 30th to the 40th equivalent day of the epidemic.
After the 40th day, when comparing 12 countries of relevant size and equivalent duration of the epidemic, Brazil was in the worst situation, apart from the USA.
As for the number of deaths per million inhabitants, Brazil, with 36, still registers a lower rate than France (309), Italy (312), Spain (480), Canada (99) and the USA (108); USP’s Lotufo, however, states that such comparisons are very problematic in the case of continental-size countries such as Brazil, China, USA, and Russia.
“It would be appropriate to draw comparisons between regions in those countries.” In São Paulo, the number of deaths per million inhabitants stands at 66, still lower than in European countries of comparable size (but higher than in Argentina).
According to Hallal, this Brazilian result is due to the fact that the country has embraced social distancing at an early stage. For the researcher, the situation has now become more concerning because “most countries began to relax measures when the epidemic curve was already in a downward trend, while Brazil does so even before the numbers begin to drop”.
The number of deaths from Covid-19 is a less uncertain indicator of the advance of the epidemic than that of cases. But it is lagged: today’s deaths point to the progress of the epidemic at least 15 days ago, when people who died were probably infected. The death count is still controversial. Some countries count deaths from Covid-19 even without testing, for instance.
There is still underreporting, although its magnitude is unknown, in Brazil and in each country compared. It is also unclear whether the rate of underreporting is variable (if it is more or less constant, it does not change the assessment of the growth rate).
Finally, even if the rate of increase in the number of deaths drops further (in percentage), this later drop will make the absolute number of deaths much higher. In parallel, the number of severe cases should also be higher, exceeding the ICU care capacity, which is the case in several capitals.
Brazil may have failed to flatten the curve. The number of deaths per day was calculated as an average of seven days (the deaths in one day plus those in the previous six days: a floating average), in order to mitigate variations caused by the large drop in notifications on weekends, for instance.
This data processing, as well as the use of the number of deaths per million inhabitants, is also suggested by “Our World in Data”, linked to Oxford University.
The counting of epidemic days was performed from the fifth death in each country, which can thus be compared in equivalent stages (days).
Source: Folhapress