[ad_1]
More than 500 thousand death records, referring to different years, were excluded from the Transparency Portal of the Civil Registry, a database fed by the registry offices. The system is used by researchers and the media to estimate, in the absence of evidence, the volume of deaths from the coronavirus.
The change in information in the database, even for previous years, makes it difficult to analyze the impact of Covid-19, since it undermines the base of comparison with 2020.
Like sheet showed on Wednesday (13), the portal has food problems that make it impossible to use it to analyze the true dimension of Covid-19 deaths in the country.
After this report, the sheet He re-analyzed all death records from the last 5 days and compared them with those filed on Thursday (15) in the system. When comparing the files, it is observed that 546,490 include deaths from coronavirus itself.
This represents 11.5% of total death records from January 2015 to April 2020, including all causes of mortality.
Most of these records that are no longer in the system come from the state of Rio de Janeiro, which lost 71% of occurrences, and refer to the years 2016, 2017 and 2018.
However, if we consider the data before the review, the estimate of the number of deaths due to Covid-19 in April would be 11,000. With the review, it rises to 26,000 (comparing the average number of deaths between 2015 and 2019 with 2020).
That’s because the month’s average, before the review, was 88,919 deaths. Then it went to 78,349. Currently, the platform indicates that there were 104,664 deaths in April.
Arpen (association of registry offices) confirmed that there was a review of data on Rio.
The problem was initially highlighted by Lucas Lago, developer of @ projeto7C0, through his Twitter account. For about two weeks he has been posting his findings on the portal.
Lago monitoring shows that the large variation in the number of deaths occurred between what was recorded in the system on Wednesday and Thursday. In other words, in just one day, more than 593 thousand records were excluded (referring to the years 2016 to 2019), according to the Lago survey.
In theory, the data presented by the registry offices could give a dimension of the real impact of Covid-19, when comparing the number of deaths in 2020 with the same period of previous years. What deviates from the pattern could be attributed to the disease. But changing the data from previous years is detrimental to this survey.
The database maintained by registry offices proposes to report the death records that arrive at these establishments, from all over Brazil, almost in real time.
Other systems that monitor mortality in the country, such as DataSus’ SIM, have a delay of at least one year. This time is largely used to verify and consolidate data.
Even the civil registry statistics, published by IBGE and which are also based on data from notaries public, are disclosed to the public with an interval of one to two years.
For this reason, the portal of the registry office would have the advantage of providing tools that would allow us to estimate the actual size of deaths from coronavirus in Brazil in an updated way.
However, the system has inconsistencies in the data source. In addition to the more than 500,000 missing records in the span of a week, the report found very large fluctuations in the monthly number of deaths from natural causes in each city, compared to previous years.
Since homicides and accidents do not count in this type of analysis, in a normal scenario, events should show some stability; there would be an increase only amid epidemics, as is the case now. Other mortality bases, such as that of the IBGE, show this trend of stability in previous years.
This is not what is seen in the numbers of almost half of the 518 cities that appear on the notary portal.
Porto Velho, for example, had 11 death records in February 2019 and 204 for the same month in 2020. Five other capitals had a variation in the volume of deaths of over 20% in January and February, when Covid-19 was still it had not spread throughout the country.
Arpen, the national association of registry offices that maintains the system, says there is indeed a delay in updating. There are more than 7,500 records in the country, and they have different rhythms to feed the base. There is a lag in the data even during years like 2017 and 2016.
In big cities, the trend is for the system to work better, and São Paulo is cited as an example of regularity in data entry. Even so, the entity suggests that the May figures are not taken into account, since there is a certain delay in the entry of records.
Regarding the disappearance of records, the entity said, in a note, that there were compatibility problems in the platform that supplies the portal in Rio de Janeiro, with duplication of old records. With the correction, the occurrences were eliminated, which caused the reduction in the numbers.
The entity indicates that these data are not used on the website to which the general public has access, so the review would not alter the conclusions present there. However, the altered information is detrimental to researchers seeking to calculate excess deaths from Covid-19 from the comparison with the historical series.
