This is very interesting - people have been saying that serosurveys will give us the only 'true' estimate of infection-fatality rate for #COVID19
I took some recent serosurvey results and added them to the meta-analysis of published estimates. Almost identical to modeled IFRs
I took some recent serosurvey results and added them to the meta-analysis of published estimates. Almost identical to modeled IFRs
Interestingly, adding these estimates also makes the publication bias look a lot less left-skewed, which to me indicates that perhaps there have been more low IFR studies published than high ones so far
This is from the SR/MA pre-printed here, that I'll update some time in the next day or two with some new info medrxiv.org/content/10.110…
Oh, and another interesting point is that when I use a modeled correction for publication bias, the IFR comes out as .78%, so potentially a touch higher if we include more estimates
