The result of a graduation study? Just a percent, indicate treasures

A large test of the occurrence of oppositions among the population of the Czech Republic b, the result of the study should be available during May. What to show me? Probably we have a lot of work to do in front of us.

The main goal of the testing is to find out how many people in the Czech Republic have encountered the SARS-CoV-2 virus. The result may be in the course of May, a clear thermal eclipse has not been determined. But it is very likely that there will not be many who have encountered the new coronavirus: I will certainly not say that it would be the percentage, said Roman Prymula, a member of the Ministry of Health.

If the situation could be such a scene with its positives and negatives, it indicates (but does not prove) the course of the epidemic in the Czech Republic with low sweat hospitalized with the virus. Confirm this similar study conducted abroad.

In comparison, it is a bag for many reasons. days from public studies did not change the scope of the experiment. Due to the large differences in the epidemiological situation, their results cannot simply be penetrated into our conditions. But in many ways, including illustrations of monch errors and inaccuracies, for our inspiration may be.

Due to doubts about the data from it and other circumstances (it has not been done for a long time in the large part of the test room for counterparts), we will limit ourselves to the results from Europe and the USA.

V epicenter

From Germany, we chose a study conducted in Gangelt for its benefits, problems and a certain media response. It has about 12 thousand inhabitants and was severely affected by the disease. Due to the carnivals of the 15th of November festivities, the virus inn has spread throughout the city. Several city dwellers with a proven infection with the new coronavirus have died, and many have health problems. Thus, the situation was far from as critical as, for example, in the most affected part of Lombardy; in Germany, however, the least affected area was at least during the burrow.

Therefore, in Gangelt, a group of vdc from several German states conducted a large pre-April study at the turn of the day and in April, which includes a total of a thousand people. The study is not completed, only the results of the test were published to about 500 people (a brief summary of the preliminary results in PDF is available in both German and English).

About 15 percent of them tested positive. Tinct percent had a counter test. Two percent of those tested then tested positive for an acute test (i.e., the specific presence of RNA viruses in the body).

If, after the publication of the remaining samples, it is not changed for other reasons, a general statement of affairs would be confirmed in Gangelt, that in the case of the availability of a medical death, only a small number of people will suffer. If the first salt went out, it would mean that about 0.4 percent of all nations died in the city.

The result would be a relatively good signal for the future. Today, virtually no one considers it possible to eradicate the virus, the second best option is so much inflected promoovn. The authors in their first report on the preliminary results of the Gangelt began the process of professing central immunity. Unfortunately, this is greatly simplified, because it is not entirely certain that the actual facts are protected from the next flow of the virus. Of course, this is generally assumed, but it is not reliably proven. The publication is therefore somewhat optimistic, it should not be objective.

But if that were the case, and the counteracts actually meant that immunity was obtained, a fifteen percent prominence would lead to a reduction in the severity of the disease (that is, the so-called reproductive salt, referred to as R). According to the model, it should disappear significantly faster than seasonal lace, especially as a pandemic lace (ie with R you not 2), but the furnace only slower than during the first wave of the disease. I recall that for the Czech Republic the ZIS model assumed R about 2.6.

Horn extrm

Gangelt, however, is a unique example of such a high degree of promotion in some (relatively small) in the flat test area of ​​the occurrence of antibodies. At first glance, the data makes two sense, because the area was affected by relatively high. Mon is a delusion, but it could be revealed soon. In Germany, various state and federal studies of SARS-CoV-2 antibodies are currently being launched, and comparisons should be available relatively soon.

In the world, there are very likely areas with a high proportion of people in the population, not Gangelt, the result is minimal. A similar result had, for example, a test run of a randomly selected thousand New Yorkers. In New York City alone, roughly 21 percent of respondents had a positive effect on counterattacks. Outside the city, according to the positive results, it fell: in Long Island it was 17 percent, in other suburban areas (Westchester, Rocklad) about 12 percent, in the rest of the state it was three percent.

Unfortunately, we don’t know much about the survey. For now, Governor Andrew Cuomo only presented him at a press conference on Thursday, April 23. Therefore, we do not only have detailed results, but neither data on the exact progress of the beer brewers nor the quality of the test. According to the information provided by the laboratory that developed the test (see this document), it should first know people without antibodies in 93 and 100 percent cases (this is the so-called specificity of the test). This is a good stop, but it is basically impossible to know from how accurate the result itself is – perhaps because it does not matter what percentage of the case the test knew correctly that someone had counterparts (so-called sensitivity).

In Lombardy, which is the best-known example of a severely affected area, the test for counter-impacts has only just started. As of April 21, there should be a large test room on a sample of 20,000 people. According to the available information, the bag will not be over the population, because health care providers will be the first participants in the study. The tests should be used so that those who have counterparts in the body can get to work (assuming, not quarantined). As a result, it is not easy to judge how many people in the area of ​​the disease have passed it, but it is not sufficiently representative.

When health is marked as ill

Lack of detail about the studies is not trivial. In such large tests, for example, the quality of the pouitch test plays a huge role. For example, there have been repeated concerns that some tests are not sensitive to antibodies to white coronaviruses – and thus do not show that the patient did not actually develop a completely common infection.

In addition, good tests that, for example, do not detect infection with another virus can, under certain circumstances, provide results that are different from reality. The previous article tried to explain the problem in detail, now let’s look at one simple example: look for one patient with a certain disease in a million inhabitants.

There is a perfectly reliable test that, for simplicity, can be found by any real patient (this is called sensitivity, in this case in 100%). He then mistakenly identifies a healthy person as ill in only one percent of cases (so-called specificity, in this case 99 percent, which is good for the day of counter-tests).

What does this mean in practice? If you test each of the millions selected, you will have to look for a sick among the 10,000 people who mistakenly report the test as sick.

A slightly sharp example shows that if you are looking for a needle in a haystack, something relatively rare in a large group, it is very, very difficult to achieve an accurate result. For example, in a large study, according to published information, tests were used which, in 95 percent of cases, correctly indicate that they have not (probably) previously resumed infiltration and coronavirus infection.

Let’s assume for the idea that in Gangelta they used tests with the same parameters, ie with 95% specificity (according to the KICs, the tests used in R are the best available). Five percent in fact of negative samples therefore test the test as positive. What in the worst imaginable case means that, in fact, about nine percent of people have counterparts. The rest were mistakes.

In the case of the New York survey, the tests had a specificity of 93 and 100 percent. Which means that we have to leave seven percent of the result – and I don’t. Otherwise, yes, the results are interesting, but let’s not talk enough about them. Not even how accurate they could actually be.

At the same time, it is certainly certain that Gangelt is – like the changed New York – are cases of cities where the disease really spread. If you test in an area where you can expect more, the error rate of the test may change the result more significantly. You can try it in our interactive tools, we can show it in other, from the point of view of strategy promo seemingly relatively optimistic hearing from the USA.

American uncertainty

The results of two studies from the late period, specifically from California, were published in the United States this April last week. Altogether, they had about three thousand participants, we will focus only on those for which detailed information was available on the MedrXiv.org server.

A team of California hygienists and a clerk at Stanford University took samples from more than 3,000 volunteers in Santa Clara County on April 3 and 4. The result seems to be in terms of a gradual reduction in immunity among the population.

1.5 percent of those tested had a positive result. After switching to the entire population (ie taking into account, for example, in the test there were a small number of people of uritch age groups, etc.), the authors came to the conclusion that about 2.5 and 4 percent of the population of Santa Clara County encountered the virus. That’s about 50x and 80x more than the official number of positive people.

If there could be much more than a person with a test proven disease, there is nothing surprising. But if I really was her and 80xvce, it would be great at first. Santa Clara has not been affected in any way, so the people are gradually gaining immunity, including the health care system.

Unfortunately, in essence, a positive result at a close glance. The reason is more (for example, the method of selection involves ads on social networks), but the main one is related to the parameters of the pouitch test. The test used by the cars was very good.

Only a completely reported report of someone who has not encountered the virus as positive (of course, in the sense of the presence of antibodies, not an acute illness). It was ground only in 0.5 percent ppad; Thus, 99.5 percent of people who did not have the virus should determine the right one. Well done, it has a sensitivity of 99.5 percent. There were 3,330 samples in the study. When done, this means that the test should inaccurately mark 17 samples as positive, which are in fact negative.

There were a total of 50 positive results in the study. So if we take the data on the accuracy of the test for the light, the aunt of positive results is probably a mistake. The result is not fully accurate (it does not even), but basically stand. The people who encountered the virus are quite a lot. For example, the 2008 pandemic Mexican chip H1N1 lasted about a year, not infecting 11 and 21 percent of the world’s population. Therefore, coronavirus would be faster and unprecedented in many parts of the world.

Only the situation is complicated. The properties of the test, in particular its tendency to false positives, were tested on blood samples which SARS-CoV-2 could not contain. The manufacturer and the authors of the experiment tested it on 401 samples, their positive result in two cases. Therefore 0.5 percent.

Number of persons with covid-19 by region

191705
234486
101293
99796
45952
124692
81347
105272
94279
80843
157749
93798
92232
181390

As many observers have noticed, this is a naive assumption. The attempt for the test parameter was relatively small, so as a result, an accident could play a big role. Inaccurate results could be both you or zero, even est. If there were more attempts, the result would be more accurate, but with such a small sweat, it can not be guaranteed that the false positive is exactly 0.5 percent.

We will not go into detail, you will find a detailed explanation (in English) for example in the article on Medium.com. Briefly, it would be explained from the point of view of statistics that the positive results could be somewhere between 0 and 1.2 percent.

And because 1.5 percent of the results in the study were positive, it means that the vast majority of them can be fully falenched. It is very long that these actual positive results were only 0.3 percent. And that’s a big problem.

As we have changed, the study also had problems that together mean that the actual number of people who have encountered the disease will probably be in a much larger range of monch results, not the cars said. Mon is a bit of you, but so be me. Mon je nakaench tenkrt vce ne captenchench ppad, mon padestkrt, mon osmdestkrt. This simply cannot be recognized as a result, despite the author’s assertion.

What does this mean for esko?

The known results can also be expected from the results of the Czech study on counter-effects. Although it has a much larger scope (and with regard to better design) does not change the study of California and research in Gangelt, but absolute accuracy cannot be expected from them. So do not fix yourself on whether the positive result will be one, two or even five percent.

So there was no reason to expect that the result would be much more and somehow, according to the inhabitants of the Czech Republic, he encountered the virus. It is very, very unlikely that the result would be at the bottom of the percentage, only at the lower end of the range (ie around ten percent). This may be the case in the most walled municipalities in the Olomouc Region, but it is practically impossible that it would apply to the entire republic.

Thus, standard immunity does not exist in the case of SARS-CoV-2. Therefore, experts repeatedly say that from a biological point of view, it is easy for the order to be re-established at virtually the same rate as at the end of a pandemic.