fact finder
A new preprint study based on health insurance data concludes that the excess mortality in the pandemic years is mainly due to the corona virus.
More people have died in Germany since the beginning of the pandemic than would have been expected based on statistical models – several studies and institutes have already come to this conclusion. Especially in the years 2021 and 2022, a high excess mortality was recorded for individual months. The possible causes for this are sometimes hotly debated. A new preprint study of the BARMER Institute for Health Systems Research (bifg) now provides another piece of the puzzle.
According to the preprint study, there is a clear connection between excess mortality during the pandemic years and Covid-19. According to the study authors, more than three quarters of the excess mortality was associated with previous Covid 19 diagnoses.
“The interesting thing about the study is that it works with a large data set of individual data,” says Sebastian Klüsener, research director at the Federal Institute for Population Research (BiB). Because the basis for the preprint study is the health insurance data of a good ten million people insured by BARMER Krankenkasse. “These data make it possible to take a closer look at relationships at the individual level. This is a great advantage over aggregated statistics, which have often been used in the past.”
excess mortality applies to older age groups
Martin Rößler, research associate at bifg and co-author of the study, also points to this. “We can map the level of individual people: When did the person die? How old was she? What was her gender and what illnesses did she have?” Previous studies on excess mortality have so far worked with publicly available death statistics, so that temporal connections, for example with corona waves, have been used as an indication.
According to the study, about 166,000 more people died in the pandemic years from 2020 to 2022 than would have been expected. This excess mortality was therefore almost entirely attributable to people who were 60 years of age or older. The fact that Covid-19 is associated with around three quarters of the excess mortality is a conservative estimate, says Rößler. “We can only see the infections that doctors have coded. We do not see any Covid-19 disease that has not been diagnosed by a doctor in our data.”
In the case of outpatient diagnoses, a reference to a corona disease can be made due to the billing structure at the quarterly level, with inpatient diagnoses much more precisely. That’s why the authors compared the observed deaths with the excess mortality – i.e. the unexpected deaths – says Rößler. “And you can see very clearly that we can only associate eight percent of the observed mortality with Covid-19, while it is more than three quarters of the excess mortality.”
That is why the observed mortality rate is also such an important reference, because the time reference in outpatient diagnoses could lead to all deaths being associated with Covid-19. “But that’s not the case,” says Roessler. “In the observed mortality, we see very clearly that it is a significantly lower proportion than in the case of excess mortality.”
insurance data can be distorted
Klusener sees the weakness of the preprint study in that the Health insurance data could be distorted, i.e. have a so-called bias. “Health insurance companies often have a certain profile of insured persons. Privately insured persons, for example, are often elitist, while some statutory health insurance companies have a very high proportion of socio-economically less well-positioned persons with increased health risks.”
This could ensure that those insured by a health insurance company are over- or under-represented in certain population groups. An extrapolation of the data to the total population could contain inaccuracies, says Klüsener. The authors of the study tried to use weighting to compensate for such distortions in the data when extrapolating. However, differences in social characteristics such as income or education were not taken into account, according to Klüsener.
In terms of the socio-economic and health profile of the insured, those insured with BARMER Krankenkasse are more in the middle. As a result, certain corona risk groups are underrepresented, says Klüsener. Because according to information, among other things of the Robert Koch Institute For example, mortality was higher among financially poorer people. “This means that the results of the preprint study are to be assessed rather conservatively in terms of the influence of the virus on excess mortality.” The actual number of people who died from or with Corona could therefore tend to be even higher.
Flu not a factor until late 2022
Klüsener believes that the fact that a large part of the excess mortality in the pandemic years can be attributed to the corona virus is well documented. Because especially in the years 2020 and 2021, a clear temporal connection between corona waves and excess mortality can be seen. In 2022, this connection will be less pronounced, but still present. In 2022, however, there was also a strong flu epidemic at the end of 2022, which also resulted in high excess mortality, according to Klüsener.
This increase is also reflected in the health insurance data, says Rößler. While the flu has hardly played a role since the pandemic began, it was diagnosed significantly more frequently towards the end of 2022.
Died of or with Corona?
Even if the preprint study confirms the assumption that the corona virus is mainly responsible for the excess mortality in Germany since the beginning of the pandemic, the restriction remains that the health insurance data does not clearly show whether a person ultimately died of or with corona is.
“In a perfect world, researchers would of course like to know what the main cause of death is for every death,” says Klüsener. “But the human body is a very complex system. Usually there are three or four causes that come together. And you will never be able to record it perfectly. But even without perfect recording, the data is solid enough to provide important insights into trends in the to deduce causes of death.”
The data from the preprint study also comes very close, says Klüsener. “In any case, much closer than is possible, for example, with the simple data on deaths.”