Renowned German Mathematician and Professor of Statistics Slams Dramatization of Covid-19
A horde of scientists are pointing out the obvious — infection rates are far higher and lethality far lower than is being reported
The media reports daily the new «number of infected», while everyone involved actually knows that they are far too low.
The next part of the sentence then deals with “death rates”, while everyone involved knows that they are far too high.
There is a confusion of terms, and politicians have to make decisions based on highly uncertain assumptions, with serious effects for society. Jens Berger spoke with professor in statistics Gerd Bosbach about definitions, numbers and the lessons that we have not drawn from past crises and hopefully will draw from this crisis in the future.
Jens Berger: Last Friday, the German Society for Epidemiology warned that in one hundred days we will have more than a million patients in Germany who will need intensive care. A horror scenario that was immediately picked up by many media, but was clearly put into proper perspective by the epidemiologists a few hours later. This forecast was made on the basis of assumptions that, let’s say politely, are not really scientifically reliable. Can you explain to our readers how such reports come about?
Gerd Bosbach: I can only guess. Such societies and associations want to get public attention. Perhaps this unconsciously prompts the acceptance of statistical models that result in extreme statements. Such statements are happily received and disseminated by the media.
Naturally, I also looked at this warning. In my opinion, the entire simulation model used is inconclusive, and then the report also mentioned that two percent of those infected would need intensive medical care. However, this appeared at six percent in graphics and calculations, without any previous mention. Even as a trained statistician, this was not easily apparent to me. I don’t want to assume that this was done with bad intent, but of course this is a catastrophic procedure. Then they corrected themselves and called this literally an «updated version with adjusted model parameters», instead of admitting the error. No journalist would be able to quickly identify the previous day’s mistake from that. And so the threatening number will continue to have effect in some minds.
— When we speak of two, four or even six percent of those infected, there is also a conscious or unconscious certainty in this statement, that this is a reliable figure. The number of intensive care patients is indeed measurable. But what about the total number of people infected?
It is now known as a fact that we do not even know the total number of infected people. If you have no symptoms, you will not be tested, and others will only be tested to a very limited extent. We know this.
But as soon as numbers are mentioned again, we pretend that we know them exactly. What we know is the number of people who tested positive. The number of infected people is definitely much higher, but no one can say by what factor. To answer this, we would need a representative sample from the population. This is currently not feasible in Germany due to the lack of test capacity.
— Wouldn’t it make more sense to talk about the number of people who tested positive than the number of people infected?
You are absolutely right, yes!
— Another number that appears in the media, again and again, these days is death rate. As a statistician, you will certainly not be able to get much out of this term in this context. You prefer to instead speak of lethality and mortality, and point out that these two terms are often mixed up. Perhaps you can briefly explain the difference to our readers and explain why it is so important to differentiate here.
There are different ways to define the death rate. A common definition is the number of deaths per 100,000 inhabitants. But that is not what is meant in the current discussion. The question here is how many of the people with corona die. Now we don’t know how many people are sick at all. Therefore, in the medical statistics there is the size of the lethality, which takes into account that we have to estimate the number of infected from the number of known infected people. As a result, lethality is an inaccurate quantity that depends on one’s own estimate. More precisely, it becomes arithmetic if one only takes into account the known positive infections, mostly referred to as the mortality rate. This method significantly overestimates mortality from the disease, but by how much is unknown. So we have a muddle of terms that ultimately explains why we keep talking about infected people instead of people who tested positive. The high numbers remain in our memory, such as the WHO mortality rate of 3.4%. And that creates fear.
My request, far beyond corona, is that you have to understand how such a percentage is defined and then look closely at the numerator and the denominator. Which one of these do we really know? In which direction is the percentage influenced by an estimate? With corona, at the denominator this means: are they using the number of infected people or the number of those who tested positive? There are also uncertainties in the numerator, for example infected people who in the future will die from corona. They are counted as infected in the denominator, but not included in the numerator.
— What about causality when it comes to deaths? Is a terminally ill person who, for example, suffers from multi-organ failure and then becomes infected and dies with the corona virus, counted as a person who died from corona? If you look at the average age of the deceased and the facts about previous illnesses, this is not an abstract question.
Many individual cases of this kind are known. Some of them are counted among the corona dead. That depends on the person who recorded it. The President of the [German federal disease control agency] Robert Koch Institute recently admitted this – unfortunately only when asked by a journalist: «We consider a corona fatality to be someone who has been diagnosed with a coronavirus infection.» So what the person died of, apparently doesn’t matter then.
Certainly there are also cases of people who died of corona but who have not been tested for the virus and who are therefore not counted. Most of these cases seem to include people who did not die of corona, but would otherwise have died. This is especially true for intensive care units or end-of-life units, where the patients are crammed together. There are corona cases among the patients and others who, terminally ill and immunodeficient, very quickly catch the virus. This is another error in the nominator, in this case too many deaths are registered.
— Could you even go so far as to say that the very mild flu season this year will have a negative effect on the number of corona deaths already reported and still to be expected?
In Germany, the flu wave has so far been relatively harmless and many very old people survived the winter, which was otherwise very warm and without much snow. Of course, these people are particularly at risk, especially if they are already ill. It is not for nothing that we speak of this risk group. In this respect, it may be that corona will not increase the overall mortality rate at all, or only minimally. After all, an average of 2,600 people died every day in Germany in 2018. From a statistical point of view, the just over 200 deaths from corona in 14 days so far are hardly noticeable. As I said, this is statistically speaking. Humanly, every dead person naturally counts.
— That changes, of course, when the dynamic progresses.
Right. We are currently analyzing past figures. As for the future – and that’s a clear warning – we simply don’t have numbers. We can extrapolate the numbers and developments from the past few days, but we do not know whether this will actually happen so.
The growth rates of those who tested positive have declined since Saturday, but that can still go in another direction. We just don’t know. The statisticians should be silent, conventional growth models will not fit and the developments from China are only partially transferable.
A big problem is that we have to conduct the current debate without reliable assumptions. Corona is currently the dominant political issue. Draconian measures are being taken worldwide to contain the infection. The side effects, especially in business life, are profound. There is already talk of an upcoming global economic crisis. Let’s play with the thought. What happens if in a year or two if the statisticians come to the conclusion that the situation had been assessed too negatively and that the draconian measures were actually not necessary? Wouldn’t this be the worst-case scenario for science? Could a politician then stand in front of his people and call for unpopular measures in climate protection, for example, with reference to scientific forecasts?
We don’t have to look into the future. Many were very afraid of swine flu in 2009. That is forgotten today because it was not dealt with after the crisis failed to materialize. Swine flu was completely overestimated and ended up being milder than many seasonal flus in previous years. It should have been necessary to work out why the swine flu was presented this way in media and why the politicians reacted with drastic and at that time unpopular measures in the vaccine strategy. From this, one could have learned the lesson not to listen to a few whisperers in your ear. And that you need clean data as soon as possible.
It is almost too late now, but at the beginning of the corona epidemic, for example, you could have tested a district, a village or a large company in a representative manner and would then have had important data as a basis for decision-making.
Nationally and internationally, we will discover a lot after the crisis, including the economic and social problems caused by today’s decisions. This should be relentlessly clarified in order to gain experience for the next crises.
–And what are your conclusions from this problem?
Our health care system must not be geared entirely towards profit. There must be enough reserves, even if it is uneconomical in normal times. Intensive care units, which are already 80% full in normal times, are not sufficient. Even if this costs money, it is urgent to invest here. And if private corporations cannot or do not want to do so, then they have to be nationalized again. Health is just not a commodity, as attac stated long ago.
My second conclusion is that we should make sure that the media does not use the power of images to create emotions that affect our judgment. If you get pictures of coffins and end-of-life units from Italy or pictures of completely empty shelves, then their effects exceed the facts.
If we pick out only a small part of the whole thing with a magnifying glass, we lose the big picture. But that is probably just a dream of an enlightened society with an enlightened media.
Speaking of the media, how do you assess the role that individual scientists play in communicating the crisis and even more in political solutions? In Great Britain there was a debate about whether the country should pursue a strategy of herd immunity. This model was then rejected based on a single study from Imperial College. Politicians today rely randomly on believing the assumptions of individual scientists. Science should have replaced belief since the Enlightenment. Do you see a problem here?
The problem is fundamental. If we want to know something for the future, we have to look forward with many assumptions. That is a dilemma. Appropriate experts must make these assumptions. Neither politics nor the media are able to do this. How to get out of this dilemma is not so easy to answer.
In the case of less urgent questions, I advise the broadest possible debate about the assumptions and calculations with various assumptions, so-called scenarios. For urgent questions, scientists who have proven themselves in the past should actually be consulted.
However, that is exactly where I want to express criticism. One often asks advice exactly from those who have been wrong in the past and some you also know have ulterior interests. At that time, the Robert Koch Institute was already noticed negatively. In other policy areas, such as the pension debate, even highly discredited people who have been proven to have made massive mistakes and to have ulterior interests, can appear again as so-called experts.
Here I almost feel sick when I think of the negative social consequences. Here the cleanliness of the advisory institutions must be checked and used as a basis for decision-making. «Fool me once, shame on you, fool me twice, shame on me», as one say. By this, I do not mean mistakes one admits oneself. It happens to everyone, but then honestly, please.
This is a fundamental requirement that I ask politicians to check afterwards. Check whether the selected experts have said something sensible or simply nonsense. In connection with corona, I noticed the head of the Robert Koch Institute early on.
–… whose President Lothar Wieler had announced in mid-February that the corona virus would probably not spread outside of China.
Later I noticed Lothar Wieler when he apparently wanted to give the all-clear with the phrase «80% of all cases are light». However, he also said that 20% of all cases are not light, and created fear. His percentages were again only related to those who tested positive, i.e. not to all infected.
And early Monday this week, he sent a message to the media that the growth numbers were declining, suggesting that the measures were already working. I clasped my head and asked myself: What kind of ignorant person is this? We only have the data from the weekend. On a weekend, less people are tested and less reported. Going public with these numbers on Monday morning and spreading «cautious optimism» frankly shocked me. A glance at the numbers from the weekend before should have warned Wieler. I would like to remove the camera or microphone from such scientists. The institute has more capable people than this.
But finally something positive, if I may. A lot of solidarity and community spirit has emerged during the crisis. Many think about what’s really important. We have to implement that in politics afterwards. We should already start thinking about what shape it should take.