In many regions of Russia, data on deaths from coronavirus are systematically underestimated. And this is how it goes
Journalists from Mediazona and Meduza were able to familiarize themselves with the data of the operational interdepartmental base on the development of the coronavirus epidemic in Russia. The Coronavirus Information Center (ICC) database contains daily data for each of 85 Russian regions on hospitalizations, critically ill patients, availability of free hospital beds and deaths. The mortality data from the ICC database in many regions is very different from the figures from the publicly available Stopcoronavirus.rf database. This is not news to Moscow: a government official admits that federal authorities often simply do not have accurate data on the state of affairs in the regions. Meduza journalists Dmitry Kuznets and Svetlana Reiter tell the details of this story; the material was prepared with the participation of Mediazona.
Abstract : The real state of affairs with deaths from coronavirus can be learned – but with a very long delay – from Rosstat data on excess mortality. These data do not coincide with the number of deaths indicated on the Stopkoronavirus.rf website, they are 4-5 times more on average in the country. But such a discrepancy, as it turned out, exists not only with public data, but even with official databases closed from the public, which are used by the officials themselves. It seems that the central authorities themselves often do not have reliable operational information about the situation with the coronavirus in many regions, where they simply do not show real numbers even in closed databases.
Mediazona journalists gained access to data from the Information Center for Monitoring the Coronavirus Situation (ICC) and released an investigation into the underreporting of mortality statistics from coronavirus in different regions. The data for “Mediazone” was provided by a user of the ICC, who wished to remain anonymous: he demonstrated to the journalists that he had access to the ICC system, and shared with the publication an archive with files from this system. We are talking about “Operational monitoring of the readiness of regional health systems for hospitalization of patients with pneumonia” from March 31 to November 22, as well as “Summary information on the readiness of the bed fund”, but only for November 2020.
Meduza studied data from the Mediazona investigation and compared it with data on deaths from coronavirus from other sources. It turned out that in the reports of the ICC, the total number of deaths associated with COVID-19 from April to November 22, 2020 is more than two times higher than indicated on the public website Stopkoronavirus.rf. But if we compare both of these databases with data on excess mortality (we can confidently say that they reflect the real mortality from the virus), it becomes clear that in many regions, data on deaths from coronavirus are systematically and deliberately underestimated – both in reports for the population and for higher officials from the center.
How is COVID-19 mortality data collected?
In addition to Rosstat, which collects information on mortality from registry offices, there are two more databases that are updated daily with data directly from hospitals and other medical institutions in each region.
- Base on the site ncov.ncmbr.ru. It is on the basis of this base that the ICC prepares buried reports for regional and federal authorities.… In this database, those responsible in hospitals every morning enter the most recent (but not always verified) data on the number of admitted patients with suspected coronavirus and community-acquired pneumonia, the number of those discharged and the number of deaths. The database also contains data on the availability of resources (beds, medicines) in the hospital.
- The database of the “Federal Register of Persons with Novel Coronavirus Infection” – covid.egisz.rosminzdrav.ru… This is the main database, which includes all the personal data of those infected, those who contacted them, and – in recent months – those who were vaccinated with the coronavirus vaccine. The data are entered into the register by the same responsible people from hospitals and other medical institutions who fill out the ICC database. But this is already different data – more detailed and, ideally, more verified. There are test results, pathological reports, concomitant diseases, etc. From this register, regional representatives of Rospotrebnadzor obtain information, who maintain a public database “Stopcoronavirus.rf” – for it, they select data on patients with a confirmed diagnosis of coronavirus infection. It is this data that is reported every day in the news.
- Finally there is Rosstat data about monthly mortality in the country (first of all, about excess mortality compared to last year), as well as deaths from the coronavirus. Excess deaths in 90-100% of cases are in one way or another associated with coronavirus infection, as evidenced by reports from few regions where excess mortality data have been analyzed in detail (for example, the Moscow Department of Health has been doing this since April ). The data on mortality themselves go to Rosstat from the registry offices, which issue death certificates to the relatives of the deceased. It is believed that this is the most accurate data, but they are published 35–40 days after the end of each month – this is a very long delay in an epidemic.
How does the data in these databases differ?
If we compare the death data for these three sources in different regions, it becomes clear that they fall into three broad categories.
- Some regions seem to provide underestimated data on mortality from coronavirus to both bases – both in the ICC and in the Federal Register. In these regions, excess mortality, according to Rosstat from April to September, is 20-120 times higher than mortality from coronavirus according to the Stopcoronavirus.rf website and the ICC database. Such a difference can only mean deliberate distortion of information at the level of data collection in hospitals or during their processing in regional governments (according to the instructions, for example, regional epidemiologists can edit the Register to “eliminate duplicate personal files”). Outstanding representatives of this group: Bashkortostan, Tatarstan, Chuvashia, Lipetsk, Bryansk, Chelyabinsk, Ryazan, Yaroslavl, Leningrad regions. Some of these regions report more deaths to the ICC than to the Stopkoroanavirus.rf website; others have the same data in their databases; some have fewer deaths in the ICC database. But everyone has a big gap between the data on the databases of the Ministry of Health and the Rosstat data on excess mortality.
- There are regions reporting to the ICC operational data on large quantity deaths of patients with suspected coronavirus, but their data on the public website Stopkoronavirus.rf is noticeably lower… However, in general, the gap between the databases and Rosstat data in these regions is relatively small. Meduza was unable to verify how these deaths are reflected in the closed Federal Register, which is not only the basis for Stopkoronavirus.rf, but also a base of information for decision-making officials. Outstanding representatives of this group: Kaliningrad, Saratov, Samara, Orenburg, Sakhalin, Kaluga regions, Mordovia, Karachay-Cherkessia, North Ossetia and Krasnodar Territory. Some regions underestimate the data for the Stopcoronavirus.rf website not very much, but constantly – with the same “coefficient”.
- Regions where data from all sources – databases of ICC, “Stopkoronavirus.rf” and Rosstat reports on excess mortality – coincide relatively exactly with each other… Here the best are the Tula and Murmansk regions, Moscow and Sevastopol, Tyva and Adygea, as well as the Krasnoyarsk Territory.
What do the authorities think about it?
Federal officials responsible for the fight against the epidemic consider the CPI database to be the least accurate of all, especially when it comes to mortality. Its main advantage is efficiency.
As a source for Meduza at the federal headquarters for combating COVID-19 explains, the ICC appeared in the spring of 2020 as a single information center on coronavirus; all press conferences, online and interviews were held at the ICC site . It was supposed to provide interaction between regional structures and the federal headquarters, but was not intended to collect statistics – and the function of exchanging operational data appeared in the ICC for the same coordination of regional and federal officials, says Meduza’s source at the federal headquarters. Data for the ICC are entered into the database on the website ncov.ncmbr.ru.
Meduza was unable to find a data collection method for this database. A federal official familiar with the problem explained that all the dead with suspected coronavirus and with community-acquired pneumonia go there “in theory”, but the main purpose of collecting this information is to have data on the availability of covid beds in hospitals; deaths of patients in this database are secondary. “[For the analysis of mortality data, the database] is not suitable: it is the most operational, but also the most inaccurate,” says a federal official. “This is, in fact, a dashboard for interested organizations, including government bodies, for a general understanding of the situation.”
The federal register of COVID-19 patients (from which public data for the Stopcoronavirus.rf website is taken) is much more accurate, the official said in a conversation with Meduza. His words are also confirmed by a source at the federal headquarters, who is familiar with the work of the Stopkoronavirus.rf website. “The data on the Stopkoronavirus.rf website on the number of deaths caused by COVID-19 as the main cause of death comes from the Federal Register, which keeps a personalized record of all cases of disease and outcomes. In accordance with the 373rd government decree (of March 31, 2020), medical organizations enter the data into the register . Selective calls are also made to the regions to clarify the data, ”he explains.
According to the federal official, the data from the Register, which are entered by the regions, also sometimes raise doubts in Moscow. If these doubts are significant, the data on specific personal cases of patients are checked. “This is a very laborious process,” the official explains. Meduza’s source did not specify exactly when and where such reconciliations were carried out, saying only that there were “many” of them.
If you want really accurate death data, you need to look at Rosstat data, a federal official told Meduza.
Why do we need so many different bases at all?
If different databases have different methodology, then maybe this is not an understatement of data?
Doubtful – the difference is too big. We know the result from different databases:
- According to website “Stopkoronavirus.rf”, from April to 22 November died 36 516 people… It is claimed that these are only cases carefully selected by the Federal Register where infection with the virus became the leading cause of death.
- According to ICC base, on “covid” beds in hospitals during the same time, almost 75 thousand people… Probably, some of them were not sick with coronavirus, some are infected, in whom the virus did not become the main cause of death. Finally, one more part got into the database by mistake of responsible persons due to the fact that officials do not attach much importance to this database – and it has a deliberately low quality of data. However, this cannot explain either the twofold difference in deaths with the public base, nor the fact that in many regions both bases coincide to within one deceased.
- According to Rosstat base – it is collected according to the diagnoses indicated in the death certificates in the registry offices (and the registry offices receive “final” – after all the studies – diagnoses from the morgues with a long delay), – only from April to September died in the country 55.6 thousand peoplediagnosed with coronavirus. Both those for whom the virus was the main cause of death and those for whom it led to an exacerbation of other diseases are taken into account, and they led to death. Obviously, by the end of November, there will be no fewer such deaths in the Rosstat database than it appears in the CIC database. This was the case before: as of the end of September (that is, as of the date on which Rosstat counted 55.6 thousand deaths), only 41.3 thousand deaths were listed in the ICC database. Most likely, Rosstat’s data on deaths diagnosed with coronavirus are similar to the sum of all coronavirus deaths from the Federal Register. It is these data of the register, according to the sources, that the authorities use; however, they are not available to the general public.
According to the Rosstat database excess mortality, from April to September died – from all reasons – on 116.3 thousand peoplemore than during the same time in 2019. Rosstat data for October will be published in the coming days. However, already now, the total number of deaths at the end of October associated with COVID-19 is estimated at at least 124 thousand people. Deaths for November cannot yet be estimated.
How exactly do regions underestimate mortality?
We don’t know for sure.
In the spring, many regions enjoyed a discrepancy in the rules for making a diagnosis and indicating the cause of death if a person was sick with coronavirus.
The federal center then decided that all deaths from covid should be carefully investigated. And the final cause of death must be established after autopsy on the basis of nosological rules. Simply put, doctors and pathologists should have established causal relationships between different diseases – primary and concomitant – and their interactions leading to death. Coronavirus is often accompanied by an exacerbation of other diseases, which leads to death – which means that it is wrong to always prescribe COVID-19 as the main cause of death of a patient. The Ministry of Health issued guidelines for pathologists, in which it required the most severe (and most resource-intensive) disease to be indicated by the initial cause of death.
As Meduza found out in the spring, the broad interpretation of the rules allowed local medical bosses in many regions – often not of their own free will, but at the behest of the authorities – to manipulate data on deaths associated with coronavirus. Deaths from exacerbations of other diseases, as a rule, did not get into the Stopkoronavirus.rf database.
In the latest – July – version of the rules for making the final diagnosis, the wording has been significantly changed. All deaths associated with COVID-19 are divided into three groups:
- cases where COVID-19 is chosen as the initial cause of death; it is these deaths that are published on the Stopkoronavirus.rf website;
- cases when COVID-19 is chosen as another cause of death – and it “is essential in the development of the underlying disease and its fatal complications”;
- cases when COVID-19 did not cause such complications (for example, as indicated in the manual for pathologists, in case of stomach cancer in a person who contracted mild coronavirus or recently).
Now, pathologists are required to identify the coronavirus as the original cause or as the condition that significantly influenced the course of the disease almost always. However, in practice, even in Moscow (one of the leaders in the “correct” registration of cases and deaths), a third of the deaths diagnosed with COVID-19 in October were classified as “died from other causes, but with a positive test for coronavirus,” and the latter did not affect the course of the disease of significant influence.
In many regions, deaths from coronavirus do not seem to fall into any of the above groups at all. So, in Bashkortostan, with an excess death rate in September of more than a thousand people (according to Rosstat from the registry office), three people died from coronavirus as the main cause, and one person died with a diagnosis of COVID-19, which did not affect the disease. In the spring, sources in medical organizations told Meduza that they were often simply forbidden to record the dead as covid.
Why exactly the regions underestimate the mortality rate in the messages for the population and for the authorities in the capital can be judged only from fragmentary messages. On the one hand, the regional authorities may have picked up the signals from Moscow in this way: in May, many regional bosses felt that coronavirus cases should be recorded as “pneumonia” so that they would not be forced to reintroduce tough restrictions, sources told Meduza at the time.
How dangerous is this data corruption for Russia?
Death data is the main indicator for assessing the speed and intensity of the spread of the epidemic.
What data help predict the epidemic?
Without accurate data on mortality, it is impossible to predict the development of the epidemic and to capture the moment when it is necessary to introduce or relax restrictions. Apparently, the federal authorities do not really have accurate data from most regions. True, it is possible that the regions themselves (they bear the main responsibility for the fight against the epidemic) have these data, since they maintain their own epidemiological registers.
However, even if the regions have all the information, but do not publish it, no one can double-check their conclusions about the situation. As the authors of the monograph “Society and Pandemic” from the RANEPA point out, the disclosure of data (for example, on hospitalizations and deaths not only by region, but also by large cities) is also necessary for the expert community, which could build its own models and forecasts.
Misinformation of the population can have even more devastating consequences: the discrepancy between the official data and the observed reality can cause persistent distrust of people in the authorities. Mistrust, in turn, leads to massive refusal to comply with even well-founded restrictions, interest in fake news and conspiracy theories, etc.
According to one of the polls conducted in July 2020, Russian medical workers, many of whom are involved in filling the databases of the Ministry of Health, do not believe the official data on the spread of the epidemic.
Authors: Dmitry Kuznets and Svetlana Reiter with the participation of Mediazona