A brief morning thread on the emerging results from antibody testing. There is a lot of interest in these as being able to nail down how many people might have been infected already and have some immunity, and the true range of severity 1/n
First we need to know that the tests can tell the difference between someone who has had covid-19 and someone who has had one of the other betacoronaviruses. Getting this wrong is bad for understanding immunity in the population, and disastrous for an individual 2/n
If you think you have a good test, like @dylanhmorris has said the way to use it to ask about population immunity is to go somewhere there has been a *lot* of disease & you might think a large proportion has been infected. Think Lombardy, or NYC (Wuhan is now small potatoes) 3/n
That assures you of something important, that there will have been enough exposure for a random sample to capture people who were asymptomatic, mildly ill, severely ill (and let’s not forget the dead) in numbers such that we can accurately estimate the proportion of each 4/n
In contrast smaller studies in places currently with little covid activity, like CA, will struggle because the signal in the first place is expected to be weaker. Smaller numbers are hard to estimate accurately. Biased sampling through social media won't help either 5/n
Also, if your sample is non-random, then you’re probably not doing useful science directed at understanding this question. You’re keeping the people who do the real work busy at the pipettes https://www.forbes.com/sites/davidnikel/2020/04/22/sweden-health-agency-withdraws-controversial-coronavirus-report/#7c1d4b994349 6/n
There’s a simple smell test here – we know that outbreaks of this result in severe strain to healthcare. If a study reports large amounts of population immunity without much disease, we’ve got to ask why the pandemic is magically different there from other places 7/n
Finally we cannot escape the fact that the poorly designed studies and the way they are reported all skew in one direction – the one which says there is more immunity in the population for less cost in lives. This is dangerous 8/n
I don’t know what the reasons for that are. Maybe wishful thinking. But the virus will carry on regardless. It doesn't care about your study, whether it is well designed or not. Remember nature cannot be fooled 9/n
I believe that there are quite huge numbers of uncounted cases, and much mild disease. I just don’t think it’s as much as these studies claim. There is a real answer out there, and we are still waiting for it #serosurveynow, but make it a good one 10/fin
(coda – apologies to the various journalists who have contacted me the last few days – esp at the @latimes for comments that would have been similar to this thread. Also kudos to @nataliexdean for being relentless on this and much smarter than me in her comments on it)