1/ OK, the results of this Stanford serosurvey are back (in preprint)- and they show...
that we need to do a rigorous serosurvey with neighborhood sampling in a high prevalence area.
that we need to do a rigorous serosurvey with neighborhood sampling in a high prevalence area.
2/ What percent had past infection? 50/3330 (1.5%).
The test kit maker says that among 371 pre-COVID samples, 2 tested false positive, so 18 of them could be false positives
Our twitter poll results over-estimated prevalence, and no, @VDHanson there is no herd immunity
The test kit maker says that among 371 pre-COVID samples, 2 tested false positive, so 18 of them could be false positives
Our twitter poll results over-estimated prevalence, and no, @VDHanson there is no herd immunity
3/ That's a lot of infections
In fact, if you weight the sample to match county residents (eg increase weight given to hispanic and asian positives) you arrive at 2.81% infection rate- that translates to between 48,000-81,000 infections vs 956 diagnosed cases!
is that right?
In fact, if you weight the sample to match county residents (eg increase weight given to hispanic and asian positives) you arrive at 2.81% infection rate- that translates to between 48,000-81,000 infections vs 956 diagnosed cases!
is that right?
4/ This has huge implications
if there are 50x as many infections as diagnosed cases (and Santa Clara test availability has been as good or better than anywhere in US) then the infection fatality rate is tiny.
If inflate 50 deaths there x 2, CFR ~ 0.12-0.2%
flu territory
if there are 50x as many infections as diagnosed cases (and Santa Clara test availability has been as good or better than anywhere in US) then the infection fatality rate is tiny.
If inflate 50 deaths there x 2, CFR ~ 0.12-0.2%
flu territory
5/ Does this make sense?
That would mean that in a place like NYC, with an estimated 10k extra deaths (so far) then 3 weeks ago there were already 5 million infections??
Might the discussion section discuss potential sources of bias in the estimates?
People responded to an ad
That would mean that in a place like NYC, with an estimated 10k extra deaths (so far) then 3 weeks ago there were already 5 million infections??
Might the discussion section discuss potential sources of bias in the estimates?
People responded to an ad
6/ Is it possible that people who thought they might have had it (or even those already tested positive by PCR) would be more motivated to get tested?
seems highly likely. I'd be cautious
They did collect survey data on prior clinical symptoms but it's not reported here ☹️
seems highly likely. I'd be cautious
They did collect survey data on prior clinical symptoms but it's not reported here ☹️
7/ How could we possibly address selection bias?
You draw a grid of all households in a defined census area, randomly select your sample, knock on their doors, and politely ask them to roll up their sleeves for the public good.
When I did one (WNV) 700/1220 eligible gave blood
You draw a grid of all households in a defined census area, randomly select your sample, knock on their doors, and politely ask them to roll up their sleeves for the public good.
When I did one (WNV) 700/1220 eligible gave blood
8/ So, in conclusion,
Can we please do a household serosurvey to answer one of the major unanswered freaking questions with this outbreak??
I never imagined in Feb that we still wouldn't know the answer to these in April
Can we please do a household serosurvey to answer one of the major unanswered freaking questions with this outbreak??
I never imagined in Feb that we still wouldn't know the answer to these in April
9/ lest I wasn't clear enough up above.
I don't believe that the CFR for COVID is flu territory at all. O think the study is overestimating prevalence because of selection bias
More people died in NYC in March than has ever been seen with influenza, at least since 1918
I don't believe that the CFR for COVID is flu territory at all. O think the study is overestimating prevalence because of selection bias
More people died in NYC in March than has ever been seen with influenza, at least since 1918
