Quote:
Originally Posted by Camelback
You're not a virologist or epidemiologist right?, you're providing a link to something (the same thing I did, from which you specifically asked for)?
|
No, but it's not obvious to me that the guy you quoted is one either. If he has some more sophisticated model than A/B, I'd like to see it.
Anyway you have no reason to listen to me. I sent you a link to a good primary source of seroprevalence data, which you can use to form your own models and interpretation.
In my view, the highest number on there is highly incongruous with the claim in the video you posted. ere. This may also be because the data is biased in some way that is undercounting, but at least the seroprevalence is a direct observable and does not require too sophisticated an analysis (one can assume with such high values, the effects of false positives and false negatives are unlikely to be very important... you may remember how improper statistical accounting of false positive rates lead to wildly incorrect conclusions in the early Santa Clara County study).
There is also independent data from blood donations, which may also be biased, but is 20% as of March, broadly consistent with the NIH data above (it's cited, for example, here:
https://www.cnn.com/2021/03/15/healt...ies/index.html, I'm not sure where the Red Cross publishes their reports).
Of course this is all from March, more people have been infected by now, but given the trajectory of COVID I'd guess it the total seroprevalence is unlikely to have increased by such a large factor.
Another handle is to use the number of estimated deaths. The population IFR is on the order of 0.5-1% (depending on the demographics of the population, naturally). Conservatively using that lower number and the number of deaths so far (600k), one gets approximately 120 million cases in the US. Of course some of these cases may have happened to the same person multiple times, but at face value, this gives us a prevalence estimate as of now of around 36%. This is higher than (admittedly rather stale) serology estimates, but it is using a low-ball number for population IFR, so I would consider this an approximate upper bound.
So I'd believe anywhere between 20-35% of Americans may have had COVID-19 right now, but 50% sounds implausible given the data.