[burger]

The Los Angeles seroprevalence study was finally published as a short note: https://jamanetwork.com/journals/jama/fu...le/2766367

Stanford med school faculty member Jay Bhattacharya was a co-author (hopefully future ex-faculty member).

Raw results: 4.06% positive (35/850)

Estimated prevalence for the county: 4.65%

Estimated IFR based on above as of early April: 0.065% [this isn't in the paper--it's using deaths as of that date]

The study took place April 10-11.

I think this is way overestimated.  Like pretty much every other antibody study, this one seems to have a selection bias.  Around 10% of their study population had symptoms similar to covid-19 in the preceding two months (6% had loss of smell/taste).  Assuming a 50% symptomatic rate, they should've had more like 2% at most.  People with symptoms had much higher rates of positive tests than others.



My current opinion is that these antibody studies are all but useless for estimating the number of people who've been infected.  Someone needs to make these people consult with survey statisticians before doing any future studies.

[burger]

The Los Angeles seroprevalence study was finally published as a short note: https://jamanetwork.com/journals/jama/fu...le/2766367

Stanford med school faculty member Jay Bhattacharya was a co-author (hopefully future ex-faculty member).

Raw results: 4.06% positive (35/850)

Estimated prevalence for the county: 4.65%

Estimated IFR based on above as of early April: 0.065% [this isn't in the paper--it's using deaths as of that date]

The study took place April 10-11.

I think this is way overestimated.  Like pretty much every other antibody study, this one seems to have a selection bias.  Around 10% of their study population had symptoms similar to covid-19 in the preceding two months (6% had loss of smell/taste).  Assuming a 50% symptomatic rate, they should've had more like 2% at most.  People with symptoms had much higher rates of positive tests than others.



My current opinion is that these antibody studies are all but useless for estimating the number of people who've been infected.  Someone needs to make these people consult with survey statisticians before doing any future studies.

[dabigv13]

My wife got an antibody test through a Cedars-Sinai study, which she was eligible for as an MD associated with them. Negative IgG. Bummer, guess that bad cold we both had in mid February wasn't it. I don't know the specificity and sensitivity of the Cedars test but I don't think its one of the Chinese imports used in this study, which have a high false positive rate that I believe is leading to an overestimate of the prevalence here in LA, along with some selection issues that burger points out.

[dabigv13]

My wife got an antibody test through a Cedars-Sinai study, which she was eligible for as an MD associated with them. Negative IgG. Bummer, guess that bad cold we both had in mid February wasn't it. I don't know the specificity and sensitivity of the Cedars test but I don't think its one of the Chinese imports used in this study, which have a high false positive rate that I believe is leading to an overestimate of the prevalence here in LA, along with some selection issues that burger points out.