The Use of Statistics-CDC Version

Yahoo news is reporting the following:

Of the 215 women who delivered babies at New York-Presbyterian Allen Hospital and Columbia University Irving Medical Center in Upper Manhattan from March 22 through April 4, 214 were tested for the coronavirus that causes Covid-19. Thirty-three of them, or more than 15%, tested positive, even though only a few had symptoms. In Gangelt, a German town that makes a big deal out of Karneval (aka Mardi Gras) and had a major coronavirus outbreak after this February’s festivities, 500 residents were tested for evidence of either the virus or the antibodies that indicate one has recovered from it, and 15% of them tested positive as well.

This goes back to my original point- A large random sample of a given population, lets say in NYC, should be tested. This will tell us the prevalence of the virus among people without symptoms. If it showed a large percentage of people already have (or had) the virus, it would show that the virus is not cause for concern. This would dilute the fatality rate and show that a major portion of people who got the virus are not at risk of dying. But because of the way the test are being conducted, what we are going to see is as the number of test rise so will the number of cases.
From the CDC website:

In the beginning of March, as the number of test increased so did the number of cases. When people stop showing up to get tested, we will see the number of cases fall. This tells us absolutely nothing.
This fiasco has reveled at least two things.
1- Epidemiologist are not very bright. They do not know how to use statistics properly. They do not believe there is a trade off between health and commerce.
2- To rely on the federal or state government is foolish.

%d bloggers like this: