I saw some data that makes me question the Stanford study. First, the accuracy of their test was questionable. They used a test made in China. A study in Denmark rated that particular test the least accurate of the tests they tried, with an accuracy of <90%. That is huge. Consider that if you test 1000 people, and 10 of them are infected. That would be a 1% infection rate. Now, if you are wrong on the 10% of the positives, the test might show 9 or 11. But what about the 990 that should be negative? It might show only 900 negatives, meaning you could show 100 positives (10 real ones, and 90 false positives), when the real number was only 10. I'm not sure that you can draw any reasonable conclusions without a more accurate test.
Next, consider the way they got people to test. They solicited them through Facebook. Might someone who wonders if they had Covid19 opt for the test? Very likely. So you end up with a self selected sample, most likely a group with a higher than typical chance of having been exposed to Covid19.
I'm sure that more people have had Covid than we know about. We can conclude that from data from places like Iceland where they did a lot of testing, or from the city of Vo, Italy. It would be nice if the real number is 50-85 times higher, but the Stanford study doesn't appear to sufficient to reach much of any conclusions.
As regards deaths, sometime this week we should hit two times the number of deaths from the flu this year (which killed 24,000). The IMHE model shows deaths vanishing after May 1. I hope that happens, but I am skeptical. Even then it doesn't address a second wave in the Fall. Flu plandemics normally kill a few in the Spring, then come back and kill more on the second wave, in the Fall. Will that also be true of Covid19? We'll have to wait and see, I guess.
Edit - Apparently for larger samples, the expected error of the test used by Stanford is more like +/- 2%. Still, if you are finding 2% positive, and the error is +/- 2%, the real infection rate is most likely in the 0-4% range. That error, combined with a non-random sample of people means that the Stanford "study" is a nice start, but not very convincing.
The good news is that if the real infection rate is 50x higher than previously know, the death rate is a lot lower, and that we are 1/20th of the way to herd immunity already. The bad news is that the R0 must be far higher than we knew, and it is going to be nearly impossible to avoid. Either way, buckle up for a year we won't forget (but won't want to remember).
Edit2 - There is another study out today with similar results. LA County found 4.1% of the people had antibodies, 28-55x higher than would be expected. The more studies that show this, the better. On the other hand, it is important to remember that when you are dealing with a low percent positive, the incidence of false positives is important. If a test gets 2% false positives, and you test 1000 people, none of whom are actually positive, you will get 2% positive. I don't know if they adjusted for that, but hopefully they did.
Last Edited: 4/21/2020 1:31:38 PM by L.C.