[jhpriestley]

1) I'm not a researcher and I can't claim to have a comprehensive view of the evidence, but I think there have been enough high-quality studies based on seroprevalence to give a good estimate. This one https://www.medrxiv.org/content/10.1101/2020.08.06.20169722v... found an IFR of 0.83% in spain, or 1.07% if counting excess deaths. This one https://www.medrxiv.org/content/10.1101/2020.06.27.20141689v... found an estimate of 1.39% in NYC. The WHO has estimated 0.6%. These studies are based on representative antibody surveys with good statistical power, they should be accurate. There are factors like demographics and comorbidities that can push the number much higher or lower in a given community, but the range does not seem that wide to me.

2) I haven't been tracking hospitalization data very closely, you may be right that this has changed ... I've seen little discussion of it.

I agree that mask use in particular was controversial for a rather long time, with bodies like the CDC seemingly dragging their feet on recommending the use of masks. Still, I think that it's been a pretty settled question since April (CDC recommended masks from April 3 https://www.livescience.com/cdc-recommends-face-masks-corona...).

[PaulDavisThe1st]

I would agree with you that it does look as if the overall IFR is in the range of 0.8-1.6%, and that the Spanish study was well conducted. But it is also horrifying that a pandemic that has killed more than 1M people worldwide still has so few similar studies to help answer these questions.

2. I don't want to make strong claims about this, but I do think it has moved quite a bit.

3. Yes, mask utility seems really well established now and has been for "many" months.

I still wish we would hear more public speakers using the numbers from 1. above combined with "flattening the curve" to make it clear what we're trying to do. 1% of the US is a million people. That could still be the outcome, but at the very least, we'd like that not happen all in the same month!

[nradov]

We now know that antibody seroprevalence studies missed many of the mild cases. This is naturally difficult to quantify.

https://www.bmj.com/content/370/bmj.m3364

So the actual IFR is probably lower than the numbers you cited.

[jhpriestley]

If you look at the responses to that editorial, they note that the 11% positivity of igA tests, cited as suggestive of a high level of undetected infection, is actually the false positive rate of the test, among other embarrassing errors.

The first citation on your linked editorial is Ioannidis, which is the same Stanford researcher in the OP article here, and the same group responsible for the utterly flawed Santa Clara study. It seems like it's literally this one group, and a few other weirdos and contrarians, singlehandedly raising spurious doubts about the science, and then being amplified beyond all reason.

The cited Ioannidis study also sucks, he takes a number of seroprevalence studies, including some very flawed or underpowered ones, and then takes the unweighted median for some reason? More details here, including an illustration of how the high-quality, randomly-selected samples do not vary so much https://twitter.com/GidMK/status/1283232023402868737

[nradov]

You're missing a big part of the picture. It is well established that some patients infected with SARS-CoV-2 either never produce detectible levels of antibodies, or have them fade away quickly. In order to measure actual infection numbers we need to look beyond antibody seroprevalence, and factor in CD4+ and CD8+ T cell assays as well.

https://doi.org/10.1016/j.cell.2020.08.017

[jhpriestley]

90% of PCR-positive subjects in the ENE-COVID test, for example, did also have a positive antibody test. So actual infection numbers might be ~10% higher than the raw numbers from the antibody tests (and I believe that this is factored into the results reported by such studies). It doesn't change our big-picture understanding of the disease.