Tuesday, 31 March 2020

Corona Virus Update: scientific studies on cures for COVID-19 (part 22)

This edition of the Corona Virus Update Podcast focusses on the scientific work on cures for COVID-19. What are the various substances and drugs that are currently being tested, how do they work and how promising are they? But starts with a clarification of yesterday's podcast on lateral-flow antibody tests.

This is part 22 of the Podcast recorded on Wednesday the 26th of March 2020. Science reporter Anja Martini of Germany public radio (NDR Info) talks to Professor Christian Drosten, the head of virology at the at a top German research hospital, the Charité in Berlin. He developed the first test for the virus, which was send to 150 countries by the WHO.

Antibody tests for the public

Anja Martini comes back to yesterday's topic of antibody tests because she received many questions:
Offers are accumulating in doctor's offices: Namely 50 tests at a unit price of 22 euros, please pay one hundred percent in advance. What do you say? Should you be more careful with these things, or what kind of thought comes to mind?
Christian Drosten:
Well, yeah, sure. Careful, definitely. It has to be said, these are lateral flow tests that can be manufactured in large quantities. That's a good thing that it' s technically possible. It's just that the current lateral-flow tests available ... have not yet been validated. So we do not know whether these antibody tests work as well as a real laboratory-based test, i.e. an ELISA test for antibodies, for one thing.

And on the other hand, there is something we already know for sure, namely that antibody tests are too late for acute diagnostics. These antibody tests can only become positive after about ten days of the disease. There are a few patients who have antibodies after only seven days. But in today's situation, when you have a test for the new virus and wants to be tested, then you ask actually always: Did I get infected? Did my symptoms come from this virus maybe? In this situation an antibody test is not useful.
These tests are mostly produced by Chinese companies. Racists abuse this situation to attack China by accusing them of producing bad tests to attack the West. This is Trumpian ignorance of people who should be attacking Xi for being an authoritarian like Trump. They should be attacking China for their concentration camps, for their lack of political freedom. But you do not get to complain that you are a gullible uninformed fool.

Anja Martini:
This antibody test, which might become available for the general public, namely the self-test, how should I imagine it technically? ... You put a prick in your finger and then you can put it on a piece of paper and see if you have antibodies or not?
Christian Drosten:
Yeah, that's pretty much how these tests work. There are several devices that extract a drop of blood from a fingertip. Then they record it. And then it runs from one side to the other in a test strip as a front, just like for a pregnancy test the urine. And at the end there is one stripe or two stripes. And if you see two stripes, the test is positive ... But as I said, all this is not yet technically validated. It will work somehow, maybe better or worse. But of course the normal laboratory-based test will also be widely available.

Polymerase Chain Reaction test

Also Polymerase Chain Reaction (PCR) tests for the virus itself were discussed in yesterday's podcast. There are cases where the patient clearly had COVID-19, which you can see in a lung x-ray, but that the test does detect it. For example patients who stayed at home as long as possible and only arrive in the hospital the second week they are ill. In this case the virus has sometimes disappeared in the throat and is only present in the lungs. Either by sampling sputum the patient coughs up or with a suction catheter a doctor can take a sample from the lungs and detect the virus that way.

Christian Drosten:
This has also caused great concern in China, in Wuhan. So much so that, from a combined impression of this apparent unreliability of PCR from the throat, and also of the laboratories being overburdened, that practically no more PCR capacity was available, they switched to a diagnosis based on the CT image at the peak of the epidemic in Wuhan, because on average, patients were seen relatively late. They stayed at home for a long time, they did not want to go to the hospital. And then they changed the diagnosis.


Anja Martini:
We want to look at drugs today, because there are now several drugs that are tested in hospitals. ... Remdesivir, for example, is a drug that was originally developed for Ebola and is now being tested in two studies on corona patients in Germany. What do you currently know about these studies and how they are going?
Christian Drosten:
In the case of Remdesivir, for the time being, we have here a substance with a plausible and known mechanism. It's an inhibitor of viral RNA polymerase, the virus' replication enzyme. And we have had this substance in the literature for quite some time, it is clear that it is effective against corona viruses in cell culture and also in animal models. That's good. So not for every substance that is currently undergoing clinical trials, we do not have this convincing initial evidence. But for remdesivir it is very good, this initial evidence. There's a real mechanism.
The above paragraph may be a bit too much in the weeds, but I included it to show how a scientist assesses a situation and the likelihood that something will work before the evidence is conclusive. Having two options where test tube tests show they work similarly against viruses, one would first go for the option where one understands why. Even if test tube showed somewhat less good results, I would still go for the one where we understand why. This is one way to protect yourself against problems with purely empirical evidence, which has produced reproducibility problems.
And now the company that distributes Remdesivir, Gilead, has been allowing [[compassionate use]] protocols for quite some time. This means that in certain constellations, the drug is released for a single patient. This is a phase of the disease where the patient already needs oxygen but does not yet need catecholamines, i.e. drugs that support the circulation. This is already a critical phase in the course of the disease. This is the transition where they say soon the patient may have to go into intensive care. It's a critical time when you want to influence [the condition of] the patient.

But the problem is that this is a direct antiviral substance, so we would like to administer it earlier. The virus attacks the respiratory tract in the first week of the disease. In the second week of the disease, when the virus deteriorates, we already have a combination of immune and viral effects that act in the lungs. This suggests that in this later phase you can't do as much if you specifically do something against the virus. You have to be careful that you might also do something against an excessive immune reaction. There are also clinical studies on this. And this is true for Remdesivir as well as for other substances where one could assume that there is a direct effect on the virus.
The rest of this section is background information on how RNA viruses work and how Remdesivir interferes, which you can skip if you just care about your health, but I find it fascinating.

Anja Martini:
How does Remdesivir work in this virus? What does it do?
Christian Drosten:
The virus is an RNA virus. And RNA viruses can't use the replication enzymes in the cell nucleus. Our cell nucleus has DNA. And when cells divide, the DNA has to be replicated.

And some viruses, DNA viruses often, they can use these multiplication enzymes for themselves. So they abuse the duplication enzymes of the cell nucleus for their own genetic material. But RNA viruses cannot do this because our cells do not need to duplicate RNA. Our cells do possess RNA. This RNA is only copied from DNA and is actually the template for proteins. This is the so-called messenger RNA, in the simplest approximation. There are of course other complicated subforms of RNA and so on. But let us now talk about the main case. This messenger RNA is not being replicated. It is simply copied once. But for viral replication, we need proper duplication. And in that process we need to have a step where RNA is copied from RNA. The viral genome consists of RNA, and the product consists again of RNA, we say the replicative intermediate, and from that again RNA has to be copied back again. After all, we have plus and minus and then again a positive sense of the genetic information in this multiplication.

All this leads us to the conclusion that the virus itself must bring along an RNA polymerase, an enzyme that carries out this multiplication, this transcription. There are different ways in which RNA viruses do this. Some RNA viruses have a functioning RNA polymerase in the virus particle. Polymerase is an enzyme that generates a polymer that transcribes. This takes a template, which is the genome of the virus, and makes a copy of it, a mirror image copy in the reading sense, and then takes this mirror image copy again and makes the next generation of genomes from it, which is then packaged. Some viruses bring this as a functioning enzyme, as a protein in the virus particle.

Other viruses simply encode this, they carry the enzyme as genetic information. This is then converted into protein in the cell by ribosomes. The protein that is produced there can then duplicate the viral RNA. Coronaviruses do it in the latter way. Corona viruses bring genetic information with them in order to create an RNA polymerase in the cell do-it-yourself, which is what the cell does, and this enzyme is inhibited with emdesivir. ...

We could also go back into detail here, because it is not quite so clear how things work exactly. We do not know whether the RNA polymerase itself is inhibited in its processivity, or whether the assembly of essential building blocks of the resulting RNA is inhibited, or whether the RNA polymerase continues to work, but makes so many copying errors that the viruses that come out of it are dead.


Anja Martini:
Chloroquine, we still have to say, is an antimalarial drug that is not completely free of side effects, but also against the old SARS virus, at least in cell cultures, has been successful, right?
Christian Drosten:
Right, exactly. In cell cultures and against all kinds of viruses. ...

A lot of people who know about it, including myself, are very sceptical about chloroquine, whether it is really helpful in the end. But I also cannot say what it will look like in the end if a very large study is carried out with a large number of patients. And the up analyse the clinical fate of these patients, what would be the outcome for the patients? So there might be a very small effect.

This effect does not necessarily have to be directly related to the virus, because chloroquine also has a strong influence on inflammatory processes in general. These also play a role in lung damage, so that it is not possible to say exactly what to expect.However, one thing can be said: A resounding effect that really decides the fate of the clinical outcome can hardly be expected with chloroquine. ... Let's put it this way, then it would be very easy to observe it. Then there would be no such contradictory clinical studies. If the effects are quite clear, it is also quite easy to prove the clinical effect.


Christian Drosten:
There is another substance called favipiravir. ... It is approved for use against influenza in several countries. So you can buy it in pharmacies against influenza, for example in Japan. ... This substance is also available in China against influenza. There is now a first study, which has been published, so perhaps we can clarify where we stand. So in case of Favipiravir, we know exactly what the mechanism is.

But I have to say that when it came up to give favipiravir against the new virus, I was surprised, because years ago, when this substance was still in the experimental phase, we did not call it favipiravir, but T-705, which was the short name for a chemical substance at that time. And it did not work well in cell culture. We didn't pursue it further. ...

Favipiravir is now being used in China after all. And a first study has just come out. ...

And in contrast to the French study, which we discussed for chloroquine last week, here it is the case that they really looked at a clinical starting criterion. They simply asked: How is the improvement of the clinical picture seven days after starting the administration of the drug? Clinical picture means for example respiratory rate, fever and other general symptoms of the disease. ...

Most of the cases here are quite normal initial cases, they are not intensive care cases. For example, there were only 18 severe cases here with pneumonia in a total of 116 people who were treated, so the overwhelming majority were not severe cases. And of course they have been included at an earlier stage accordingly.

So now we can say that the difference that can be obtained in this rather optimal situation is that clinical symptoms improve in 56 percent of the cases where no treatment is given, and in 72 percent of the cases where treatment is given. That is a significant difference, a significant difference statistically.

And that's amazing to me. I have to say that in view of the fact that we never actually saw a good effect of this substance in cell culture, I am still skeptical if this is real or if there's some kind of flaw in the clinical study. Now we have to see what other studies indicate. It's certainly not enough, to take one study and even more so one study that has not even been formally reviewed yet.


Anja Martini:
I believe that you yourself are also working with Göttingen researchers on a drug at the moment. How does it work?
Christian Drosten:
That's right. There are studies that we have done together with Stefan Pöhlmann's group in Göttingen, a really absolute specialist in virus entry. Stefan has seen that it is possible to reduce virus entry with a substance called camostat. ...

So it is the case that this virus, this new SARS 2 virus, uses a certain transmembrane protease in a stronger way than the old known SARS virus. And that is, as the name suggests, a protein-cleaving enzyme, but this time it is not an enzyme from the virus but an enzyme from the cell. So the cell itself has this protein on its outer membrane. And with this protein, the cell involuntarily helps the virus enter the cell, by the passage through the membrane. This works in such a way that the surface protein of the virus is cut at one point, is clipped, and this clipping of the surface protein is the first step for the virus to pass through the cell membranes. This virus uses this cellular protein for this purpose.

There is a drug that inhibits this cellular protein and this drug is called camostat. I am deliberately saying drug and not substance, because this substance is approved as a drug for chronic pancreatitis. And it is only approved in Japan. So in Japan you can buy it in the pharmacy. This much we know. We know it works in cell culture, and we know the drug is available in Japan. That's all we know.

But on this basis we can now of course do something that cannot be done with other substances. Namely, we can say that we do not have time for large-scale animal experiments, but we have an approved substance here. In certain cases, we can now test this in clinical controlled trials to see whether patients benefit from it if they get this substance. This is a typical off-label use study. And we're going to do something like this start now.
That sounds promising. Do note that Drosten is here talking about his own research. It is always harder to be just as sceptical about your own work, any scientist will be able to attest.

Other podcasts

Part 28: Corona Virus Update: exit strategy, masks, aerosols, loss of smell and taste.

Part 27: Corona Virus Update: tracking infections by App and do go outside

Part 26: Corona Virus Update on Vaccines: clinical trials, various types, for whom and when.

Part 23: Corona Virus Update: need for speed in funding and publication, virus arrival, from pandemic to endemic

Part 21: Corona Virus Update: tests, tests, tests and how they work.

Part 20: Corona Virus Update: Case-tracking teams, slowdown in Germany, infectiousness.

Part 19: Corona Virus Update with Christian Drosten: going outside, face masks, children and media troubles.

Part 18: Leading German virologist Prof. Dr. Christian Drosten goes viral, topics: Air pollution, data quality, sequencing, immunity, seasonality & curfews.

Related reading

The Corona Virus Update podcast and its German transcript. Part 22.

Nature Magazine on the various possibilities where the virus comes from: The proximal origin of SARS-CoV-2. "This is strong evidence that SARS-CoV-2 is not the product of purposeful manipulation."

European Medicines Agency: COVID-19: chloroquine and hydroxychloroquine only to be used in clinical trials or emergency use programmes. "The European Medicines Agency (EMA) is a decentralised agency of the European Union (EU) responsible for the scientific evaluation, supervision and safety monitoring of medicines in the EU."

Monday, 30 March 2020

Corona Virus Update: tests, tests, tests and how they work (part 21)

The Corona Virus Update Podcast of Wednesday the 25th of March was mostly an interesting lecture about tests. With pain in my heart I cut out the part on the different types of antibodies and the different parts of the immune system. So if you know some German, do read the German transcript. It was a really interesting interview.

The interview with Prof. Dr. Christian Drosten, who studies emerging viruses, was performed by science journalist Korinna Henning. I presume colleagues now call her Corona Henning. Or maybe German public radio is more professional than a university.

The podcast is my main source of Corona science. It is great to listen to someone who knows what he is talking about and is honest about the current limits of our knowledge.

The podcast is about tests for the virus itself, which is what we currently mainly do, and about various antibody tests to detected whether people had the disease, that is had an immune response producing antibodies against the virus.

Polymerase Chain Reaction tests

The main work horse to determine whether people have COVID-19 is a Polymerase Chain Reaction test. This test detects the presence of virus DNA by multiplying it, if present.

Korinna Hennig:
We've already talked a couple of times in this podcast about antibody testing, which has high hopes because it can provide education about the many potentially very mild infections that go undetected. And about how much immunity is already in the population. But we have also repeatedly spoken about the need for large-scale testing for the virus itself. Perhaps we can clear up the large field a bit: The current test for the virus is a PCR test, which means Polymerase Chain Reaction. This is where the genetic information of the virus is multiplied and a colour reaction generated. In simple terms: what exactly is happening actually?
Christian Drosten:
This is a reaction in which the genetic material of the virus is copied and thereby multiplied. This is an invention from the late 1980s that has since then gained an increasing foothold in microbiological and virological diagnostics and in principle has increasingly replaced cultural methods - i.e. the cultivation [breeding] of such a bacterium or virus - in virology in particular; a lot is still being cultivated in microbiology.

This is due to the fact that PCR is both very sensitive and very fast, in terms of the method used, and also has a high specificity. This means that what you find there is really what you are looking for. That is because you have to add small DNA sections, really molecules, physical molecules, to this reaction in order to multiply what exactly matches these molecules.

This means that if, for example, I want to detect this new coronavirus in the polymerase chain reaction, then I have to produce small pieces of this new coronavirus in the laboratory, in the form of RNA, i.e. small snippets of RNA, which are only about 20 bases long, and they will then attach themselves to the genome. And they can only do this if they are really almost or completely identical, i.e. in the sequence of bases, with the genome of the virus you are looking for. And if there is another virus in there that is even a little bit similar, but is actually a different, we won't be able to detect it.
Korinna Hennig:
This means that other corona viruses are not incorrectly identified here?
Christian Drosten:
Right. So, the four cold viruses, human corona viruses, we can't detect them with that. That's not what we want. We only want the test to be positive if this new virus is actually present.

Where and when does the PCR test work?

The test by itself is highly reliable, but it is important that the sample of the patient is taken correctly.

Korinna Hennig:
But the test shows that presence of the virus, you just explained, not the immune response of the patient. In other words, if I do it too late, if I perhaps have symptoms, but the disease is already subsiding, then it is going nowhere, so to speak?
Christian Drosten:
With this disease, it is the case that in the first week of symptoms, the samples from the throat, i.e. the smears, are actually very reliably positive in the PCR. And then, in the second week, they are no longer reliably positive. Then the patient still has symptoms, but in the throat the test might not be able to detect this. This is not because the test is not good, but simply because the virus is no longer present in the throat, but it is present in the lungs.

We now know that even in patients who have very mild symptoms, i.e. who notice almost nothing of their illness, there is still quite a lot of virus in the lungs. And this remains there for about two weeks, or even three weeks, in the uncomplicated cases. That's how long we are able to detect the virus in the lungs with this polymerase chain reaction. However, many patients cannot simply cough up such a sample from the lungs, so throat swabs are actually the most common sample. What can be done, which is not yet well established systematically, is to take a stool sample. The virus is detectable there as well, and for quite a long time actually, as long, or almost as long, as in the lungs.
Korinna Hennig:
But no longer infectious, that was a realization that we also addressed at some point in the podcast: That this contact infection - as is the case with noro-viruses, for example - is not a transmission route for the coronavirus.
Christian Drosten:
Yeah, right. So in our tests, the virus is highly detectable in the stool. So that means it can be used as diagnostic information. But it doesn't look like an infectious virus. We can say this because we simply place the same sample in parallel on cell culture and then see whether a virus grows there and is alive. And it's not.

Random ELISA antibody tests

Sampling patients and people they may have infected makes it hard to interpret the data in terms of how the population is doing. This would require tests of random people. Such information would be very important for public policy.

Germany at the moment has about 50 thousand infections. With a population of about 80 million people, this means that in the order of one in a thousand are infected. Such random tests would thus need to be large to get reliable numbers. Germany already made about half a million tests. So such random tests would be possible, but would reduce capacity for potential patients. The current way of thinking is to wait doing random sampling until we have large scale antibody tests, which does not interfere with patient care.

Korinna Hennig:
You said before that we'll have to take a look around Easter, and then politicians will have to decide whether to take action. Then of course it would be good to know even more. Is it conceivable, as I just mentioned, that random tests could be carried out to detect even more really undetected infected persons?
Christian Drosten:
But what there will be in the very near future, and which is actually much more important and informative, are such random tests for antibodies. This is a completely different test procedure. If we get infected, it takes about ten days for us to produce antibodies in this disease. We have already looked at that; other studies agree. And those antibodies will then become even more apparent and even better in the next few days. So, at first, it's such a low level. And then, two or three weeks after infection, you have a very clear antibody in your blood.

And you can use a blood sample with a technically different kind of test - these are [[ELISA]] tests, enzyme-linked immunosorbent assay - to measure whether a patient has antibodies in his blood, regardless of whether he had a severe infection or a mild infection or a completely unnoticed infection. And that is what we are actually speculating on. If we knew that all infections were symptomatic, then you could actually say: well, we actually record quite precisely what we have in terms of PCR results, and now we do the calculations on that basis.

But what we do not know at the moment, and what unfortunately does not come out well from studies in other countries either - there are unfortunately no convincing data from China on this either - is the rate of those who are really completely unnoticed and infected in the population. So what is the incidental immune activity of this virus? How many people get infected without noticing it or without having taken it seriously because just a little bit of scratching of the throat? But they are still antibody-positive and, as we can assume, immune, and even then contribute to the 60 or 70 percent of the population that must have become immune or infected before the pandemic wave comes to a halt.

Validation of the ELISA antibody tests

Before introducing antibody tests on a large scale, one needs to study how accurate they are. Unreliable tests would be damage patients and help spread the disease.

Christian Drosten:
There is already the possibility of carrying out [a simpler] antibody tests on a smaller scale in the laboratory. ... We have been able to carry out antibody tests for this virus in our laboratory since mid-January - but only on a small scale because it is a lot of work. ...

For these large mass screenings, where we would like to test thousands of patients, we need an automated ELISA test, and they are all just being set up. ...

There is a German manufacturer, a large well-known manufacturer, whom we have helped from the beginning to both set up and, above all, evaluate these test procedures. So to ask: How does what they have set up there compare to our microscope test carried out at home? So when I say "at home", of course I mean in our own laboratory. ...

But [those validations] closed now, they look good. So we can attest to this manufacturer: Go ahead, start producing these so everyone can use them. Such studies are of course now taking place in many other countries. Other manufacturers are now asking us, too, whether we can help validate this. And of course we do that.

When will ELISA tests be available?

Christian Drosten:
We already have one machine here in the institute, and we will build a much larger machine in two weeks here in our laboratory. And other large laboratories elsewhere in Germany are also doing that these days, so it doesn't necessarily have to take another two or three months.

And these tests can then be sent in via a General Practitioner; a blood sample will be taken by the General Practitioner and the results will be available the next day or the day after next. ...

There are considerations on how to use such tests, which doesn't necessarily have to be the planned visit to grandma and grandpa, but also the question: Can this doctor or nurse actually treat patients with the disease again, maybe even with reduced self-protection, i.e. with reduced protective equipment, because there is immunity? That is a very important question, for example. Or outside in outpatient care, whether a nurse has survived the infection, yes or no, is extreme important for the operational capability.

Rapid antibody tests

Very new on the market are rapid tests, lateral-flow-tests, which look like pregnancy tests.

Christian Drosten:
Then, of course, such tests are now coming onto the market, antibody tests that you can buy yourself, perhaps soon in the pharmacy or even now already on Ebay. These tests have not yet been validated, they come mainly from Asia, especially from China. These are tests that also test for antibodies using a different test principle. At the moment, I simply have to say that this should be treated with caution because we do not yet have good validation studies. However, we are currently testing such tests from several manufacturers and are providing the manufacturers with feedback on how well they work. In the next few weeks we will certainly see publications and first publications about how well such rapid antibody tests work.

That these tests are not validated yet is used on social media to attack China for endangering the West with unreliable tests. I would say: if you are so careless to use such a test now, do not blame China for your Trumpian ignorance.

Other podcasts

Part 28: Corona Virus Update: exit strategy, masks, aerosols, loss of smell and taste.

Part 27: Corona Virus Update: tracking infections by App and do go outside

Part 26: Corona Virus Update on Vaccines: clinical trials, various types, for whom and when.

Part 23: Corona Virus Update: need for speed in funding and publication, virus arrival, from pandemic to endemic

Part 22: Corona Virus Update: scientific studies on cures for COVID-19.

Part 20: Corona Virus Update: Case-tracking teams, slowdown in Germany, infectiousness.

Part 19: Corona Virus Update with Christian Drosten: going outside, face masks, children and media troubles.

Part 18: Leading German virologist Prof. Dr. Christian Drosten goes viral, topics: Air pollution, data quality, sequencing, immunity, seasonality & curfews.

Related reading

The German podcast and German transcript of part 21.

All Corona Virus Update Podcasts and their transcripts in German.

Sunday, 29 March 2020

Corona Virus Update: Case-tracking teams, slowdown in Germany, infectiousness (part 20)

The Corona Virus Update, part 20, of Tuesday the 26th of March was about whether the South Korean strategy of case-tracking teams is something for Germany as well, what the data on the number of confirmed infections tells us about the spread of the virus in Germany and about a new study on the infectiousness of the virus and what that tells us about how to contain it.

This podcast is produced by the German public radio broadcaster NDR. Science journalist Anja Martini interviews Professor Christian Drosten, just like every day. He is the head of the Virology Department at the Berlin Charité, which is one of the main research hospitals in Germany. As a scientist he can speak more freely than, for example, the director of the RKI, the German CDC. He specializes in emerging viruses and developed the WHO test for the new Corona virus.

Case-tracking teams in South Korea

Can we learn anything from how South Korea handled the situation? Too long, didn't read: South Korea is very strong in case tracking. They test a lot and have many people working to track contacts of known infected people. This produced good results for fighting one big outbreak, but there are still new infections and we will have to see how well this strategy works in future.

Christian Drosten:
There are case-tracking teams that can follow every infected person and look: Who has there been contact with? Where are the contacts now? The contacts are isolated and monitored and so on. I think that is simply not feasible here, if only for personnel reasons. That's why the question of whether you can learn anything from it is a bit futile.

But it's also true that one shouldn't be fooled. For a very long time, there was the impression in Korea that the outbreak is now actually under control. But what's often forgotten to say is that a big part of the initial outbreak in Korea was a single event. ... And of course you could follow that very well. Of course, you have a list of participants and can say: Okay, they were all there, and we're really going after them now.

But this effect is now over in Korea. This transmission event is now so far in the past that it has been captured. And that has done a lot to bring down the curve in Korea right now. But what I am now hearing from Korea is that individual transmission chains are now starting up all over the country, because of course there have been other cases registered in parallel via multiple channels (let's consider the proximity to China). And that just now in Korea the new infections are clearly increasing, because it's not this very focused measure any more, but suddenly you have to be everywhere. And I could do that I imagine that this will also no longer manageable in Korea, to be everywhere at the same time. But in general, they do case tracking very carefully. And I have a feeling can do this better simply by staffing levels than we in Germany.

Spread in Italy and Germany

Anja Martini:
Italy is now going into the third week of quarantine. The first experts are now breathing a sigh of relief because the number of deaths has not increased any further. How much can we already tell from these figures and this development? How much can one learn from this, and what can we trust?
Christian Drosten:
So the number of cases [infections], which in Italy are apparently not so easy to count because probably not so much diagnosis is made, are the absolute numbers of infections. But of course what has to be counted are the deceased. And it takes on average about three weeks between the onset of symptoms and their passing away. Or between the infection, you have to say correctly, and death. And that's why that's the effect that happens in the statistics. A curfew and other quarantine and isolation measures were put in place three weeks ago. And now you can see the effect, even on the deceased. And that is unfortunately almost a natural constant that can be observed. It just takes three weeks.

And you see it more quickly with [infection] cases in countries that can detect the cases very reliably. ... But in this short period of maybe ten days at the most, we want to see that the increase in new infections is already decreasing, at least in a system, a country where we are already close to reality with diagnostics - which we hope we are in Germany. ... I would be very pleased if this were to be confirmed in the next few days. But one must also say that we will have to wait a little longer. It has to last a few days, this effect, before you can see: So you can see something.
The next paragraph required quite some explanations to interpret the science-speak of Drosten. I hope I did it right. He seems to be saying that already in this early phase the number of cases is decreasing compared to model computations based on the past. That means that the spread of the virus has become less efficient, likely due to all the policy measures taken. Even more summarized: the policies seem to work.
"[The number of infections are] perhaps already now in this early phase, the expected values [from models] are changing compared to the observed values. So there seems to be a difference, which is good. [Because this suggests the model parameters about how the virus spreads are improving] And if it stays like that for the next few days, then you will look at it for a while. And then, for example, in this difference you have a new basis for readjusting models. [Estimating how the policies have affected the spread of the virus] And then mathematicians and modelers in Germany will actually be called upon to take and evaluate this data and then to prepare it for policymakers, for example."
This will also affect the number of deaths, but the coming weeks it will still rise.
We will of course see changes in the deceased with a two or three week delay. Incidentally, we also have to remind ourselves, and I would perhaps like to say this again now, that despite the measures that we already have at the moment, the number of deaths will of course continue to rise, because this effect will continue. And this, too, will be reflected in model calculations. That will of course also be important, because it will give an insight into the seriousness of the cases. And this severity of cases must be taken into account in terms of hospital capacity.

So this kind of epidemiological modelling that is needed here at the moment is not just a pure description of the situation of the cases, but must also take into account when we reach the capacity limit of the medical system. So completely different figures have to be included, such as the number of beds or the number of ventilation places.

And in the very near future, the question will be posed to the scientific community: Where do we stand now? How can we now readjust? Must we leave the current measures as they are? Or can we relax the brakes a little in some places, because it is not just a pure, naked scientific consideration, but also because scientists are well aware that the current measures are of course causing great social and economic damage. And these things have to be weighed against each other.
Around Eastern we should have a better assessment of the situation.

Infectiousness study

Anja Martini asks Drosten about a new study from Hong Kong. He first explains how scientific publishing has changed due to the time pressures of the epidemic. If you are not interested in that you can skip the next long quote.

(The quote provides anecdotal evidence that without peer review scientists would focus much more on studies from well-known groups and that peer review thus helps outsiders gain the credibility they need to have people invest time in their work. I have a blog post on that.)

Christian Drosten:
This is a study that has been published on a preprint server. At the moment we have this very fast situation in scientific publication activity. Normally the review process of a scientific contribution takes weeks or even months. So sometimes it goes from a scientist to a journal. They don't send it to the journal for review. Then you send it to another journal and they send it out. The reviewers need two months, then the comments come back. And then the magazine says: "Fix it, please. And then another month goes by.

And you can't afford that right now with epidemiological research. And that's why at the moment, scientific articles are actually placed in online resources, the so-called preprint servers, as they are written. There are two very big ones, called bio-archives and med-archives.

I always go through them like this. I have to sort a lot of things, because they are not peer-reviewed scientific articles. That means there's also a lot of dead wood. There are a lot of things that you won't see officially appear in this form later on, because they won't survive the review process. That means, what I always do in my free minutes is that I look at the things that appear in a new way. And things that I think are of such a high quality, that will survive any review process, that's really well done, I discuss these things here sometimes. So then I say: this is interesting data. And so it is with this study here.

It comes from Hong Kong from a very well-known epidemiological modeling group, Gabriel Leung.
It is important when a patient is infectious.
[The study was about] when does this disease actually become infectious? Already before the symptoms or with the symptoms or after the symptoms? And this is very important, because with the old SARS corona virus we can briefly sum up: It was so easy to contain because it only becomes truly infectious long after the symptoms start, in the average patient.
Drosten himself also had a small study on this topic. Currently still a preprint (not peer reviewed).
This study has also already shown that the virus replicates in the throat in the early phase of infection and that the virus is clearly detectable in swabs even in the very earliest swabs to such an extent that it is already on the descending branch even on day one and two in swabs. So the further one waits - and if one takes smears from a patient every day - the fewer and fewer, right from the beginning.
The new study found the same result, but with many more patients.
These authors found exactly the same thing in a group of 94 cases in Guangdong, i.e. in southern China near Hong Kong. ... And they saw that from day one the virus was on the decline. That means the peak of the virus must be before the first day.
The new study also quantified how long it takes to be infected.
Then they did something very interesting, something purely epidemiological: they also looked at transmission cases in the same context, namely 77 couples, 77 patients, where it is known that one person infected the other, and they looked closely: How long did it actually take?
I did not understand the explanation of how this works, but the key word is "[[Serial Interval]]". The result was:
The median is 5.2 days, the mean is 5.8, so this is a somewhat skewed distribution, but still with very close averages, so 5.2 to 5.8, you can say is the series inteval.
So the time to infect someone is about the same as the incubation time.
They also calculated the incubation time from their own earlier, very well done study: 5.2 days mean incubation time. This is of course interesting, because we have here a phenomenon where practically the serial interval is almost as long as the incubation time. This tells us that the average patient waits for the symptoms after infection for as long as it takes to transmit the infection between two patients. And if you look at it that way, it means that not only do we have a mean onset of transmission on the day the symptoms start, but probably before that as well. So the average patient is basically transferred [infected] on the day the symptoms start, but this is just the average patient. Some patients are not transferred until after the start of their symptoms and unfortunately some are transferred before the start of their symptoms. ...

it can be said that infectivity starts two and a half days before the onset of symptoms, on average. And the so-called area under the curve, i.e. the area covered by this probability curve, before the onset of symptoms, is 44 percent. In other words, it can be assumed that 44 percent of all infectious events occurred before the infective person was even ill. ...

This also means, of course, that if you lock yourself up at home as soon as the symptoms begin, you have already infected people if you allow a normal social life to continue. So this means that with normal rules of infection protection in case of a noticed illness you cannot contain this disease. There has to be social distancing in a targeted way, where the aim is to change behaviour - and not to identify symptoms and isolate the sufferers. That simply will not work with this disease.

Other podcasts

Part 28: Corona Virus Update: exit strategy, masks, aerosols, loss of smell and taste.

Part 27: Corona Virus Update: tracking infections by App and do go outside

Part 26: Corona Virus Update on Vaccines: clinical trials, various types, for whom and when.

Part 23: Corona Virus Update: need for speed in funding and publication, virus arrival, from pandemic to endemic

Part 22: Corona Virus Update: scientific studies on cures for COVID-19.

Part 21: Corona Virus Update: tests, tests, tests and how they work.

Part 19: Corona Virus Update with Christian Drosten: going outside, face masks, children and media troubles.

Part 18: Leading German virologist Prof. Dr. Christian Drosten goes viral, topics: Air pollution, data quality, sequencing, immunity, seasonality & curfews.

Related reading

Corona Virus Update with Christian Drosten podcasts and transcripts (one day later).

Tuesday, 24 March 2020

Corona Virus Update with Christian Drosten: going outside, face masks, children and media troubles (part 19)

Much of the information on the new Corona virus in Germany, at least for science nerds, is spread by the Corona Virus Update Podcast with Prof. Dr. Christian Drosten. He is the head of the virology department at one of the main university hospitals in Germany and specializes on emerging viruses.

It is a podcast of the German public radio NDR. The half an hour interview number 19, on Monday the 23 of March was held by science journalist Korinna Hennig. Topics were, going out of the house, face masks, children and the media creating trouble.

Walking and jogging

Korinna Hennig:
Meanwhile, many people have come to us who are worried when they walk in the forest and joggers walk very close to them and breathe on them. Safety distance is one of the words of the hour. Is this assumption that the duration of contact plays an important role still tenable despite everything?
Christian Drosten:
Of course, it is very difficult to say anything really tenable about this now. But in principle, when you're outside, what you breathe out naturally dilutes, and the virus also dilutes. Besides, you almost always have a little bit of wind.

And so you have to concentrate more on the situation in closed rooms if you are thinking about such transmission processes.

Face masks

Too Long, Didn't read: Face mask are scarce in Europe. Medical professionals in close contact with infected people need them the most. For normal people wearing a mask outside the house does not provide any protection, the main thing they would do is protect others in case you are infective (without knowing it). However, at the moment we do not have enough masks to use that as containment method.
First of all, of course, it must be said that there is a shortage of these masks in all countries, not only in Germany, but in the whole of Europe and practically all over the world. If we now look at Europe in particular, it really is a nationwide situation in which no country has any stocks or anything like that.

I know that the German Ministry of Health began weeks ago to secure stocks and place orders. So I would think that we in Germany are very well prepared. But things like that are going on, they take time. Orders take time. Production needs a time. And right now in Germany, as in all other countries too, we have a shortage of these masks in the market.

And in fact, hospitals are still being supplied. But it's not like they're in unlimited supply now. That is why purchasing departments at large hospitals are justifiably concerned if the public were to access the same stocks now. You have to imagine that at some point there will be market competition. And supply and demand will then drive up prices. If people in the public think that they can protect themselves from infection by wearing a mask, then of course at some point there will be people who pay lunar prices for something like that, even if it has little or no effect. ...

There must be no market competition. Because in the medical sector, in professions that work close to patients - that's not just the doctor and the nurse, but also in other areas, in nursing homes and so on - it's natural to have very close contact. And different rules apply in this close contact area. And there is definitely data that shows that such respiratory tract disease transmissions are reduced by the masks. ...

For the public, there are two considerations that can be made in this way. One is self-protection: I wear a mask to keep from getting sick. The other is other-protection: I am sick, wearing a mask so that someone else does not get sick, so that the virus is not transmitted further.

And for the latter, there are, let's say mechanical good reasons for doing so. It's easy for anyone to imagine. When I sneeze, I give off tiny droplets. And when I have a piece of cloth in front of my mouth, it can either be a cellulose cloth like a mask I bought, or it can of course be a scarf or something, these big droplets are then caught. ...

The consideration is, the further away you are from this source, the more you are dealing with a finer aerosol. And this is also inhaled sideways into a mask, whether you inhale it from the front of your mouth. Or you have a mask on and you suck it in from the side. That is then simply no longer a difference. That's why: The closer to the source, the better. That's why the mask has to be at the source and not at the receiver.

And that is certainly a perfectly plausible consideration. What is not so plausible is that I cannot protect myself in public with a mask. This is just maybe a little bit difficult to convey. But there is simply either no evidence in the literature or - depending on how you want to interpret it - almost no evidence that this could help.


Do children get infected and ill?
So I think we can say that children do not get severe symptoms. There are simply no known descriptions. Well there are of course individual descriptions of severe, even unfortunately also of a few deceased children, but in view of the mass of cases it hardly seems to occur. Whereby the word hardly means: Just only in a very, very small percentage.

It is, of course, an important consideration, because now it can have two explanations: One may be that the children are not infected at all. That means that they are completely excluded. The other may be that the children are infected, become immune and at some point belong to the circle of those who have already been infected and become immune in society and then do their part to stop the epidemic. ...

Children, school children have a particular network function in society because they interact relatively intensively with several age groups. While other age groups are more in contact with their own age group. Therefore children have a very important function there. And we all want to find out in the next few weeks - by means of antibody tests, also in children - what the background infection rate is, in other words the silent infection. To ask the question: Have children perhaps, without realizing it, already contracted the infection? And may they already become immune unnoticed?
Drosten discusses a preprint (a scientific article that is not peer reviewed) about the early stage of the virus in Wuhan. In this stage all patients went to the hospital for isolation, including children because it was not known yet that they did not get ill.
One can deduce that in this early phase in Wuhan, the Authors speak of thousands to tens of thousands of unrecognized child cases that have occurred. And that, of course, gives hope in a certain way. Namely, on the one hand, if this is such a great effect, it will be possible to correct the actual infectious mortality. And what is even more important, if we know that the children are actually very actively infected, then this means that they also contribute to the [infected part of the population], in other words to this development of herd immunity. That is good news in principle, this [preprint]. What we need now is confirmation of this phenomenon through antibody tests, also in children, but not only.

About the person

Korinna Hennig:
Mr. Drosten, in conclusion, we have already mentioned here in the podcast that you are often accosted, exposed to hate mails and insults. Now we have just seen the opposite effect. Suddenly, all sorts of newspapers have begun to focus on you. You are very much in the spotlight and a hype has developed which has somehow taken on a life of its own. How do you feel about that? Are you coping with it?
Christian Drosten :
I have to admit that it makes me uneasy and I don't like it. I already have the feeling that a legend is being created. ... But of course that has very little to do with reality.

It worries me especially, when I see that that comes along with shortening of statements. For example, what has just happened this weekend is that there was a relatively differentiated interview in a large magazine, where two or three questions were asked about the topic of how can things continue now? So what do we do now? Now these measures are all in force and what does our future look like now? Can you get out of there?

And then I already said for example: Well, if you look at it, whether you fill football stadiums with people or go to school, then going to school is more important. That's why I believe that we won't have full football stadiums any time soon. But that we have to concentrate relatively soon on getting data to decide whether we can perhaps allow the whole school or just a few years of school again. Because that is really important. I was interested in this distinction, what is entertainment and what is essentially important in society? What can you focus on now if you want to get out of these contact measures again?

And then it was shortened, and that was done by the magazine itself on the Internet, of course to attract attention to this article. Basically, all they said was: "Drosten: No more football for a year." And then they wrote, which was not even mentioned in the interview, that this could probably be extended to include holding football matches without spectators. So even that, it is not true that I advised against that. That was not the content. And it wasn't directly described that way, but from the context it sounded that way. Then there is the fact that this article is also behind a paywall. That means, if you go to this internet message and want to watch the interview, then you have to pay. And that annoys me, because it was a whole afternoon of my time that I invested there. ...

It's just bad when media come and still try to make money out of this situation by such contrasts and such incentives. I think the media must stop that now. Otherwise we as scientists can no longer do the kind of things I am doing here. Some of my colleagues are much more cautious. That is of course the main reason why not many other scientists communicate in public, because these things happen all the time. It's just not bearable anymore.

This also scares me as a person because I naturally notice when something like this is announced. It went out on the servers some Sunday afternoon, I noticed it because aggressive comments suddenly appeared in my e-mail inbox that really attacked me. And where I notice, there are people I don't know, who don't know me, but who have found out my email address and who now attacking me. And that is, let's just say, the most harmless consequence. But I also find the misunderstandings that arise very serious. And we have to be clear, for what purpose? Ultimately, only for [newspaper] circulation.
This while the German language press is wonderful compared to the English language press.

Other podcasts

Part 28: Corona Virus Update: exit strategy, masks, aerosols, loss of smell and taste.

Part 27: Corona Virus Update: tracking infections by App and do go outside

Part 26: Corona Virus Update on Vaccines: clinical trials, various types, for whom and when.

Part 23: Corona Virus Update: need for speed in funding and publication, virus arrival, from pandemic to endemic

Part 22: Corona Virus Update: scientific studies on cures for COVID-19.

Part 21: Corona Virus Update: tests, tests, tests and how they work.

Part 20: Corona Virus Update: Case-tracking teams, slowdown in Germany, infectiousness.

Part 18: Leading German virologist Prof. Dr. Christian Drosten goes viral, topics: Air pollution, data quality, sequencing, immunity, seasonality & curfews.

Related reading

Corona Virus Update with Christian Drosten podcasts and transcripts (one day later).

The Robert Koch-Institut (RKI) in Germany is comparable to the US Centers for Disease Control and Prevention (CDC). The RKI publishes a daily summary of the situation: situation reports. First in German, a little later also in English. It has statistics on the number of infections and mortality and on measures take to fight the problem. Something I especially like is that they list cases by date the people got ill (if known), not just raw numbers for one day, which are total based on an unknown number of previous days. This figure (number 3) suggests that the measures work and slowed down the spread of the virus.

This weekend the RKI was looking for staff to help track infected people. They got 10,000 applications and already stopped accepting new applications. Thanks, humanity.

Monday, 23 March 2020

Leading German virologist Prof. Dr. Christian Drosten goes viral (intro & part 18)

Every workday German public radio has an interview with [[Prof. Dr. Christian Drosten]]. In the 50s every family would have sat around their radio receiver and listened with red ears. At least my impression is that nearly everyone listens.

Germany's newspaper for academics called him "Germany's de facto explainer for the current outbreak". Christian Drosten is leader of the virology department of one of the most prestigious hospitals in the world, Berlin's Charité university hospital. He developed the first diagnostic test, which is called the WHO test in Anglo-America, and was shipped to 150 countries.

I guess our university press office would like me to mention that before he went to Berlin Drosten headed the Institute of Virology here at the University Hospital in Bonn.

Because these podcasts are appreciated so much, I thought it would be valuable to translate them into English. The virology is universal, the epidemiology and counter measures will depend on the local circumstances. The latter can be interesting for foreigners in Germany and for inspiration for other countries, so I will also translate such parts.

In one podcast Drosten discussed an article on the influence of closing schools on the Spanish Flu in American cities, where it worked. Some German newspaper made this into a story: "Drosten recommends closing schools", while he also talked about the societal differences between Germany now and America a century ago. As a climate scientist, I will not pretend to know the science good enough to contribute, but I do understand it better than political journalists and that I can promise not to leave out such important context.

If you know German you can find all podcasts with transcripts on the webpage of NDR. The topics of one podcast are all over the map, whatever is current. This podcast science journalist Korinna Hennig asked the questions.

Air pollution

The first interesting question was about the role of air pollution. Drosten answers:
"Yes, there is certainly some speculation about it. But what is perhaps even more important, if you want to talk about something like that, is of course smoking. And we don't even know what the reason for this surplus of male patients is. What is clear, however, is that in China it is mainly men who smoke. And of course it is also clear that in the generation of patients who are now particularly at risk, it is above all men who have smoked a lot throughout their lives. And of course, risk factors for cardiovascular disease are also more prevalent among men in this age group. And I think that all of this plays a role in this pattern. ... It is certainly always a good time to quit smoking, but now is probably a particularly good time."
That sounds like an interesting idea worth pursuing in more detail.

Data quality, comparability

Another interesting question was why "there been so many fewer deaths from SARS-CoV-2 in Germany than in other countries?"

The main explanation is that in other countries far fewer mild cases are recorded because less testing is done and this simply distorts the statistics. "We actually test much more than other countries." He does not expect that these differences between countries will converge when later more people will die in Germany. One reason is that soon the epidemic spreads to much that testing will no longer be able to keep up, even in Germany. Then reporting will change from confirmed cases to suspected cases.

Also data on hospital admissions is hard to compare between countries and even regions:
"There are still hospital admissions because of a diagnosis, with the intention to be rather safe and to admit patients to the hospital in order to isolate them. And in other regions there are already many cases and there one will be rather hesitant about admitting patients who are otherwise healthy."


Drosten and his team work on sequencing the genome of the virus. This used to be to study how the virus is spreading. The main road on which the virus travels to Germany used to be holidaymakers returning from Italy, but this is now changing and soon everything will be very mixed. The next task would be to study whether the virus is changing:
"The real issue then is whether the virus remains stable. And for that you simply have to continue sequencing viruses at a certain frequency. And always look, is the genome still complete? Have mutations crept in at important places? And do these mutations have any significance? In other words, then one has to switch over to the targeted examination of these viruses in the laboratory."


If someone was ill and acquired immunity Drosten estimates that they can no longer carry the disease to others. To be sure large clinical studies are needed, which will be done later. To infect others the virus would have actively replicate in the throat.
"we know from a monkey experiment that once an infection has been overcome, one million infectious viruses can be introduced directly into the trachea of these monkeys and nothing happens. And that is already a very high level of challenge infection, as we call it in such a study. Now, of course, you have to say that these are not people, these are monkeys. Humans can be slightly different in detail. But there are other indications that suggest that we should have a very good immune response. For example, we know that over a long period of time, even in patients who say they have hardly noticed their infection, the virus not only replicates a little bit in the throat, but to a considerable extent in the lungs. And we should then be able to assume that a strong immune response is triggered."

Summer time

Korinna Hennig asked what the biological explanation is for the expectation that higher summer temperatures cannot contain the virus.

Christian Drosten:
"There will certainly be a small effect. A biological explanation, that is, it is just that one can see how endemic viruses decrease in frequency through the temperature effect. By endemic I mean those viruses that occur widely in the population. And these viruses have two problems when it gets warm. First, they have a permanent problem, namely there is population immunity. Then on top of that comes the second problem, let's say of the summer, in other words all the effects that this brings with it. Social distance outside and UV light, heat, dryness, so these things are not good for virus transmission, not conducive. And when that comes together with population immunity, then there is a stop to virus transmission in viruses like influenza. And now you can just look at influenza, for example, an endemic virus, to what extent is that stopped? And then you can compare a pandemic virus with it. To what extent will it be stopped? And it will not be stopped very much, but it will be stopped a bit. This comparative calculation can also be made for coronaviruses, and a study to which I referred before made this comparison. That's what was done there. And the estimate is that there will be a slight slowdown. The estimate is that half a unit of [the basic reproduction value] R0 can be subtracted. ... But at the same time, unfortunately, the estimate that the R0 value won't go below one due to this summer effect alone, that you have to do other things as well."


The podcast ended with a more political question about the effectiveness of curfews. This question is hard to answer because we do not have data on this yet and you cannot study it in isolation: 
"It's all relatively difficult to say, because the curfew itself is one of several measures that are applied in addition to the non-pharmaceutical interventions. There is also something like closing schools, tracing of infected persons and isolation of infected persons at home. Then there is the quarantine of the environment, in the simplest case, for example, the family at home for 14 days. But also the identification of contacts and their isolation at home for 14 days. All these measures come together. And now it is relatively difficult to say, if you add something on top of this, such as a curfew, what difference does it make? There is no data at all for this, either in Germany or elsewhere in other studies, in modelling studies."
Speaking more as a private person he later ads:
"I am not necessarily someone who says that we need an immediate curfew. Especially under the impression that I have that a great many people are now taking this more and more seriously and are also thinking about it, and are staying at home of their own accord. I do think that perhaps we should allow a little more time."
The Sunday after this interview on Friday, the German federal government and the state governments agreed on strong limitations on the freedom of movement.

I hope this English summary is useful. If people enjoy it I am happy to do this for future podcasts are well.

Other podcasts

Part 28: Corona Virus Update: exit strategy, masks, aerosols, loss of smell and taste.

Part 27: Corona Virus Update: tracking infections by App and do go outside

Part 26: Corona Virus Update on Vaccines: clinical trials, various types, for whom and when.

Part 23: Corona Virus Update: need for speed in funding and publication, virus arrival, from pandemic to endemic

Part 22: Corona Virus Update: scientific studies on cures for COVID-19.

Part 21: Corona Virus Update: tests, tests, tests and how they work.

Part 20: Corona Virus Update: Case-tracking teams, slowdown in Germany, infectiousness.

Part 19: Corona Virus Update with Christian Drosten: going outside, face masks, children and media troubles.

Related reading

Translated interview published today (the 21st of March): "We Have To Bring Down the Number of Cases Now. Otherwise We Won't Be Able To Handle It" Published in Die Zeit, the German newspaper for academics. (I had some trouble reading it the first months I was in Germany due to their posh language, but the information is high quality. In German the verb comes at the end of a sentence, in a "well written" sentence that is the moment you understand the sentence. In the "best" Die Zeit articles they do the same for paragraphs and articles. Only after reading the last sentence do you understand the paragraph and only after reading the last paragraph do you understand the article.)

LA Times: Germany’s extensive medical network apparently helped in early stage of coronavirus. The article is actually about all the policy differences with respect to healthcare and the economy that makes handling the situation much easier in Germany. Not mentioned is that the number of infections in Germany is one of the highest in the world, Germany is big, Germans like holiday in Italy and while people still complain did a lot of testing compared to other countries.

Sunday, 1 March 2020

Trend errors in raw temperature station data due to inhomogeneities

Another serious title to signal this is again one for the nerds. How large is the uncertainty in the temperature trends of raw station data due to inhomogeneities? Too Long, Didn’t Read: they are big and larger in America than in Germany.

We came up with two very different methods (Lindau and Venema, 2020) to estimate this and we got lucky: the estimates match as well as one could expect.

Direct method

Let’s start simple, take pairs of stations and compute their difference time series, that is, we calculate the difference between two raw temperature time series. For each of these difference series you compute the trend and from all these trends you compute the variance.

If you compute these variances for pairs of stations in a range of distance classes you get the thick curves in the figure below for the United States of America (marked as U) and Germany (marked as G). This was computed based on data for the period 1901 to 2000 from the International Surface Temperature Initiative (ISTI) .

Figure 1. The variance of the trend differences computed from temperature data from the United States of America (U) and for Germany (G). Only the thick curves are relevant for this blog post and are used to compute the trend uncertainty. The paper used Kelvin [K], we could also have used degree Celsius [°C] like we do in this post.

When the distance between the pairs is small (near zero on the x-axis) the trend differences are mostly due to inhomogeneities, whereas when the distances get larger also real climate trend differences increase the variance of the trend differences. So we need to extrapolate the thick curves to a distance of zero to get the variance due to inhomogeneities.

For the USA this extrapolation gives about 1 °C2 per century2. This is the variance due to two stations. If the inhomogeneities are assumed to be independent, one station will have contributed half and the trend variance of one station is 0.5 °C2 per century2. To get an uncertainty measure that is easier to understand for humans, you can take the square root, which gives the standard deviation of the trends: 0.71 °C per century.

That is a decent size compared to the total warming over the last century of about 1.5 °C over land; see estimates below. This alone is a good reason to homogenize climate data to reduce such errors.

Figure 2. Warming estimates of the land surface air temperature from four different institutions. Figure taken from the last IPCC report.

The same exercise for Germany estimates the variance to be 0.5 °C2 per century2, the square root of half this variance gives a trend uncertainty of 0.5 °C per century.

In Germany the maximum distance for which we have a sufficient number of pairs (900 km) is naturally smaller than for America (over 1200 km). Interestingly, for America also real trend differences are important, which you can see in the trend variance increasing for pairs of stations that are further apart. In Germany this does not seem to happen even for the largest distances we could compute.

An important reason to homogenize climate data is to remove network-wide trend biases. When these trend biases are due to changes that affected all stations, they will hardly be visible in the difference time series. A good example is a change of the instruments affecting an entire observational network. It is also possible to have such large-scale trend biases due to rare, but big events, such as stations relocating from cities to airports, leading to a greatly reduced urban heat island effect. In such a case, the trend difference would be visible in the difference series and would thus be noticed by the above direct method.

Indirect method

The indirect method to estimate the station trend errors starts with an estimate of the statistical properties of the inhomogeneities and derives a relationship between these properties and the trend error.


How the statistical properties of the inhomogeneities are estimated is described in Lindau and Venema (2019) and my previous blog post. To summarize, we had two statistical models for inhomogeneities. One where the size of the inhomogeneity between two breaks is given by a random number. We called this Random Deviations (RD) from a baseline. The second model is for breaks behaving like Brownian Motion (BM). Here the jump sizes are determined by random numbers. So the difference is whether the levels or the jumps are random numbers.

We found in both countries RD break with a typical jump size of about 0.5 °C. But the frequency was quite different, while in Germany we had one break every 24 years, in America it was once every 5.8 years.

Furthermore, in America there are also breaks that behave like Brownian Motion. For these break we only know the variance of the break multiplied by the frequency of the breaks, this is 0.45 °C2 per century2. We do not know not whether the value is due to many small breaks or one big one.

Relationship to trend errors

The next step is to relate these properties of the inhomogeneities to trend errors.

For the Random Deviation case, the dependence on the size of the breaks is trivial, it is simply proportional, but the dependence on the number of breaks is quite interesting. The numerical relationship is shown with +-symbols in the graph below.

Clearly when there are no breaks, there is also no trend error. On the other extreme of a large number of breaks, the error is due to a large number of independent random numbers, which to a large part cancel each other out. The largest trend errors are thus found for a moderate number of breaks.

To understand the result we start with the case without any variations in the break frequency. That is, if the break frequency is 5 breaks per century, every single 100-year time series has exactly 5 breaks. For this case we can derive an equation shown below as the thick curve. As expected it starts at zero and the maximum trend error is in case of 2 breaks in the time series.

More realistic is the case when there is a mean break frequency over all stations, but the number of breaks varies randomly per station. In case the breaks are independent of each other one would expect the number of breaks to follow a Poisson distribution. The thin lines in the graph below takes this scatter into account by computing a weighted average over the equation using Poisson weights. This smoothing reduces the height of the maximum and shifts it to a larger average break frequency, about 3 breaks per time series. Especially for more than 5 breaks, the numerical and analytical solutions fit very well.

Figure 3. The relationship between the variance of the trend due to inhomogeneities and the frequency of breaks, expressed as breaks per century. The plus-symbols are the results based on numerical simulation for 100-years time series. The thick line is the equation we found for a fixed break frequency, while the thin line takes into account random variations in the break frequency.

The next graph, shown below, also includes the case of Brownian Motion (BM), as well as the Random Deviation (RD) case. To make the BM and RD cases comparable, they both have jumps following a normal distribution with a variance of 1 °C2. Clearly the Brownian Motion case (with O-symbols) produces much larger trend errors than the Random Deviations case (+-symbols).

Figure 4. The variance of the trend as a function of the frequency of breaks for the two statistical models. The O-symbols are for Brownian Motion, the +-symbols for Random Deviations. The variance of the jump sizes was 1 °C2 in both cases.

That the variance of the trends due to inhomogeneities is a linear function of the number of breaks can be understood by considering that to a first approximation the trend error for Brownian Motion is given by a line connecting the first and the last segment of the break signal. If k is the number of breaks, the value of the last segment is the sum of k random values. Thus if the variance of one break is σβ2, the variance of the value of the last segment is kβ2 and thus a linear function of the number of breaks.

Alternatively you can do a lot of math and at the end find that the problem simplifies like a high school problem and that the actual trend error is 6/5 times the simple approximation from the previous paragraph.

Give me numbers

The variance of the trend error due to BM inhomogeneities in America is thus 6/5 times 0.45 °C2 per century2, which equals 0.54 °C2 per century2.

This BM trend error is a lot bigger than the trend error due to the RD inhomogeneities, which for 17.1 breaks per century and a break size distribution with variance 0.12 °C2 is 0.13 °C2 per century2.

One can add these two variances together to get 0.67 °C2 per century2. The standard deviation of this trend error is thus quite big: 0.82 °C per century and mostly due to the BM component.

In Germany, we found only RD breaks with a frequency of 4.1 breaks per century. Their size is 0.13 °C2. If we put this into the equation, the variance of the trends due to inhomogeneities is 0.34 °C2 per century2. Although the size of the RD breaks is about the same as in America, their influence on the station trend errors is larger, which is somewhat counter-intuitively because their number is lower. The standard deviation of the trend due to inhomogeneities is thus 0.58 °C per century in Germany.

Comparing the two estimates

Finally, we can compare the direct and indirect estimates of the trend errors. For America the direct (empirical) method found a trend error of 0.71 °C per century and the indirect (analytical) method 0.82 °C per century. For Germany the direct method found 0.5 °C per century and the indirect method 0.58 °C per century.

The indirect method thus found slightly larger uncertainties. Our estimates were based on the assumption of random station trend errors, which do not produce a bias in the trend. A difference in sensitivity to such biasing inhomogeneities in the observational data would be a reasonable explanation for these small differences. Also missing data may play a role.

Inhomogeneities can be very complicated. The break frequency does not have to be constant, the break sizes could depend on the year. Random Deviations and Brownian Motion are idealizations. In that light, it is encouraging that the direct and indirect estimates fit that well. These approximations seem to be sufficiently realistic, at least for the computation of station trend errors.

Other posts in this series

Part 5: Statistical homogenization under-corrects any station network-wide trend biases

Part 4: Break detection is deceptive when the noise is larger than the break signal

Part 3: Correcting inhomogeneities when all breaks are perfectly known

Part 2: Trend errors in raw temperature station data due to inhomogeneities

Part 1: Estimating the statistical properties of inhomogeneities without homogenization


Lindau, R, Venema, V., 2020: Random trend errors in climate station data due to inhomogeneities. International Journal Climatology, 40, pp. 2393-2402. Open Access. https://doi.org/10.1002/joc.6340

Lindau, R, Venema, V., 2019: A new method to study inhomogeneities in climate records: Brownian motion or random deviations? International Journal Climatology, 39, p. 4769– 4783. Manuscript. https://eartharxiv.org/vjnbd/ https://doi.org/10.1002/joc.6105