Calculating real Slovak COVID death rate (and a bit on Sweden)

Note: all links given on this page are archived at web.archive.org If I discover a link is down, or has been substantially changed in content from when I accessed it, I will replace the links with archives.

Jump to calculations.

From the beginning of the pandemic, until at least the end of September, the Slovak government only counted deaths as being from the respiratory form of COVID19 on the basis of an autopsy establishing that as the cause of death. A press release on September 26th 2020 stated that to date 247 deaths had been investigated in connection with suspected SARS-CoV-2 infection. Of those 44 had been judged to be caused by COVID19.(1) In that time period, people who died with a SARS-CoV-2 infection were not counted in any statistics on COVID19 deaths, if other diseases were found to be the cause, and COVID19 was just a contributing factor. See for instance the elderly lady from the village of Staré, who was infected along with many other villagers during the feast day of the church in the village (when Slovaks with connections to a place return from the big city and aside from a church service also have feasts at their family homes and visit friends), was admitted to the hospital and died. She however had many co-morbidities, therefore her death was not recorded in the statistics. (2) The approach used for autopsies was graphically described in an interview with the chief pathologist around the same time (3). At the time, I was somewhat shocked, because I realized that while, on the one hand, a detailed determination of who really died from COVID19, rather than just with a SARS-CoV-2 infection is useful information, and helps put the real risks in perspective, on the other hand, Slovaks were using these deaths that passed a very strict criterion to compare themselves to other countries.

One of these other countries, of course, was Sweden. In the fall (October 30th), I contacted through Twitter a doctor working in Sweden, to inquire what standards are used there to determine if a death is counted in the statistics. He replied:  "Sweden is more generous in its definitions. It is very rare for a patient to be autopsied, and antibody tests are usually not taken. The diagnosis on the death report is usually based on a combination of pcr test result and symptoms before death." Recently, I have found the current definition on the Swedish Public Health website:

  • The statistics show the number of people with laboratory-confirmed covid-19 who have died, regardless of the cause of death, and are reported as dead in the SmiNet database.
  • The deaths have either been reported dead by the treating doctor or have, according to the population register, died within 30 days after a covid-19 diagnosis. In a limited number of cases, it may be known by the Infection Control Unit that the cause of death was non-covid related and then the death is removed from the statistics.

https://www.folkhalsomyndigheten.se/smittskydd-beredskap/utbrott/aktuella-utbrott/covid-19/statistik-och-analyser/bekraftade-fall-i-sverige/ Text is translation from Google Translate. This page has the latest statistics, the number of deaths is the same as is reported on e.g. Worldometers. Also, data is downloadable as an Excel spreadsheet. The spreadsheet is in Swedish, but only a few keywords to figure out in order to use that data.

At some point, I think in December, the Slovak National Center for Health Information COVID19 data portal changed the format in which they presented information on deaths to include number of deaths from COVID19 as well as those with COVID19. At the very end of December, to no fanfare, the Institute for Health Analysis released files with detailed data on deaths to their github site, including five year age group of the deceased, information that before had not been available in a coherent form to the public, only sporadically and incompletely being released as Facebook posts by the Ministry of Health. From the file OpenData_Slovakia_Covid_DeathsCumulative.csv we can see that recording of statistics on "deaths with COVID" began on 3rd of October, though this data was not, as far as I know, made available to the public till December.

Calculations:

Calculating comparable death rate for Slovakia

As discussed above, the definition of a COVID19 death in the statistics that the Slovak government presents to the world and gets reported on Worldometers and other sites (even ECDC) are those deaths from COVID19, "deaths of a SARS-CoV-2 positive individual, whose main cause of death was the disease COVID19", but exclude deaths with COVID19, "deaths of a SARS-CoV-2 positive individual, whose main cause of death was another disease". This remains a stricter criteria than that in other countries, e.g. Sweden as described above, or states in the USA, where the tendency is to include deaths of individuals who had recently tested positive and SARS-CoV-2 infection could have contributed (ideally excluding things like car crashes, though there are notorious examples from parts of the US where these deaths got included). Personally, I think the more precise approach taken by the Slovak government is better, but for the sake of valid international comparisons, we should compare numbers that count the same thing. So here are the steps to get the total number of deaths from and with COVID19:

  1. Go to https://covid-19.nczisk.sk (use a computer or switch your mobile browser to show the desktop site)
  2. There is a language selector, it only translates the summary page.
  3. Summary block 03 is "COVID-19 Number of Deaths", click the expanding arrow icon to see the detail view.
    Image of summary block
  4. See image for description of each field. BTW, the "per 1000 positive cases" number is the Case Fatality Rate (at least, by that definition of "case" that now means "tested positive"). So as of writing this, CFR in Slovakia is 2.05% or 2.49%, depending on which deaths you count.
    Detail display of deaths in SK.
  5. So now we have the total number of deaths, 6707 as of February 9th 2021.
  6. The population of Slovakia as of Nov. 2021 (latest of writing) can be found on the Statistical Office Datacube  and is 5,463,916
    Population of Slovakia 2021
  7. Rate of deaths per million is 6707 / 5.464 = 1227

And now about Sweden...

  1. Above I've given a link to official statistics of Sweden on COVID deaths. However, they just report that one number of everyone who died "with laboratory-confirmed covid-19" and this number can be found all over the place, like at Worldometers. Today it is 12,326.
  2. Population is misreported as 10.2 million on many popular sites. The Swedish Statistics Office gives the up to date number, as of this writing latest is November, 10,380,245
  3. Rate of deaths per million is 12326 / 10.380 = 1187

Dr. Rushworth discusses on his website why he thinks Sweden had more deaths than its neighbors.

Conclusions

Whatever Slovakia is doing, isn't working. The level of restriction here has at times been excessive and the guidance given by the government (to this day) is heavy on people wearing masks, but they are otherwise told that keeping 2m of distance is needed just between strangers. Personally, I suspect that the masks give people a false sense of security. Older people especially seem to get really close to you to talk, like 40cm. Any time the government has a press conference here, the politicians are wearing masks while standing shoulder to shoulder.

At the moment, my home US state of Georgia has, depending on whether you include "probably cases", 1270 or 1441 cases per million. Given the average daily number of deaths in GA vs SK, adjusted by population, I would not be surprised if in a couple of weeks Slovakia reaches the same level as Georgia.