Thread

One thing that came up in discussions was how small the number of Corona related fatalities in Germany seems to be. So I plotted the number of fatalities vs. total cases in loglog, and indeed, the data seems very weird.

This is based on the data from github.com/CSSEGISandData…

I can understand that the mortality rate goes up in places like China because new infections go down, but it takes a time since everyone is recovered.

You could also argue that you can see when the health system in Italy was at capacity and mortality rate started to increase (about 1500 cases or so).

Still, the differences are so huge, they don't make a lot of sense and (I suspect) rather point to huge biases in terms of how much testing is done and who is tested.

The US mortality rate is also over all going down, which is good, I think. Let me know what you make of this.

Ah, minor correction, this is not log log anymore, of course, that was the first version... .

I should also add that this is not a proper mortality rate (where you would wait for cases to recover and then backpropagate that properly), but number of fatalities / number of new cases. Given how fast things are evolving, this is a proxy.

But at least the error is systematic and applies to all countries.

Short follow up (in German), reasons why the numbers are so low in Germany: morgenpost.de/vermischtes/ar… People aren't tested post mortem, so those who die and are undiagnosed don't enter the statistics.