“18 instead of 78 – why the incidence values are calculated incorrectly”. This is the title of the YouTube video that math student Patrick Schönherr uploaded a little over a week ago. In it, the 23-year-old from Berchtesgadener Land accuses the Robert Koch Institute (RKI) of making a gross error in calculating the 7-day occurrence.

*Note: The video has since been deleted. FOCUS Online explains what theory the math student sets out in it.*

The incidence rate indicates the corona cases detected per 100,000 inhabitants. According to Schönherr, however, the RKI omits an important point: **There are not the same amount of tests everywhere in Germany.**

If, on the other hand, the incidence were compared with the tests, a completely different value would arise. This would have serious consequences, especially in counties that test significantly more than average.

“It’s enough to have math knowledge from middle school to understand the problem,” Schönherr introduces his video.

He gives an example:

County A and County B each have 50,000 inhabitants and one percent infected each.

**County A:**tests 5,000 residents, finds about 50 infected. that**occurrence**are here**100**.**County B:**tests 2000 residents, finds about 20 infected. that**occurrence**are here**40**.

The countries are therefore in the same pandemic situation, but have different occurrences. “There are more tests. More positive cases are being discovered. The incidence is increasing,” mathematics students conclude, “so this incidence value does not currently allow any conclusions to be drawn about the pandemic.”

Schönherr therefore proposes to include the ratio between the tests and the total population in the incidence calculation, “ie a kind of test-positive rate”.

For this, it would only be necessary to standardize the number of tests and orient them to the national average. Since November, the average would **about 1.5 percent of the population** tested per week. A new, normalized incidence rate could then be calculated from the “test positive rate”.

## In Berchtesgadener Länd the occurrence would be only 18

As an example for his calculations he uses the occurrence in Berchtesgadener Land. This was included at the time of publication in week 9 **78.3**this week, 83 people tested positive.

In all was **6817 of the 106,000 inhabitants **tested for Sars-CoV-2, which corresponds to 6.4 percent of the population. 83 tests were positive, i.e. one **“Test positive rate” of 1.2 percent.**

that **previous incidence calculation** er: 83 x (100,000 / 106,000) = 78.3

The calculation for the new occurrence works as follows:

- Normalization: 1.5 percent of the population corresponds to 1590.
**new “case number**with only 1590 tests would therefore**at 19.08**(equivalent to 19 positive tests). - These positive tests are now used as usual to calculate the incidence. 19.08 x (100,000 / 106,000) = 18.0
**The incidence would therefore be 18 instead of 78**.

## According to the math student, the calculation method would have several advantages

For this great discrepancy there is, according to the student of mathematics **for two reasons:**

- Berchtesgadener Land tests significantly more than the rest of Germany, on average four times as much.
- The share of positive tests of 1.2 percent is well below the German average of 6.1 percent most recently.

According to Schönherr, this method of calculation would have **several benefits:**

- Negative tests will also be included in the “test positive rate”.
- The national German average would hardly change.
- Higher test numbers would not increase the incidence.
- The counties were easier to compare with each other.

## Physicist Priesemann comments

According to “Traunsteiner Tagblatt”, the math student hopes that “the problem will be discussed more in politics and media”. He therefore sent his calculations to politicians in the Bavarian state parliament. However, he does not want his invoices to be “misused”. He is not concerned with criticism of the corona measures, but only with a correct statement of the incidence values, which are comparable for all districts.

Schönherr’s video was shared by many and did not go unnoticed by top German researchers. For example, physicist Viola Priesemann of the Max Plank Institute commented on Twitter.

According to her, the video “addresses an important topic”. Increased testing is “punished” in the short term, as more chains of infection are then detected. In the long run, however, it will be worth it, the physicist points out. “Because it stops the chains.”

But Priesemann opposed Schönherr’s theory: “The video sounds logical, but assumes a wrong assumption. So the conclusion is wrong.”

In his video, Schönherr assumes that tests were done randomly. “That’s not the case,” says the physicist. “People are not tested randomly, but mostly because there is a suspicion.” These include symptoms, contacts or a positive quick test.

## “Follow-up results are not relevant”

According to Priesemann, one should in between **two causalities** distinguish:

**Case A:**Several cases have been found because more have been tested.**Case B:**More tests are being done because there are more suspected cases.

“Ultimately, both contributions play a role,” she explains. “But in the video, it is assumed that it is really only A that applies.” All further calculations after this slide are based on this incorrect assumption. So the subsequent results are not correct.

“According to the calculation in the video, the occurrences in the district can be easily reduced,” Priesemann also warns. For each suspect test, do a test on people who are most likely negative (or a random test). “The incidence is already (almost) halved.”

Priesemann then also suggests a solution: “It would be best if we, like the UK, had screening, that is to say around 100,000 samples, which gives an objective picture of the outbreak each week. – Then we should not discuss here.”