John, they have done field epidemiology in New York, Los Angeles and Santa Clara county using blood tests for antibodies to COVID-19. In all cases the investigators showed many fold more infections than have been reported through the normal hospital based testing that was going on at the time. When the preprints came out all kinds of hate posts started appearing bitching about the sampling protocols and the poor reliability of the tests which was false in the two CA testing regimes as the researchers did their own in house validation and didn't simply rely on the manufacturers data. New York used a version of the Mt. Sinai developed test which is approved by the FDA and one of the best ones out there. IMO, the infection rate is a minimum of 10X the number of reported infections and this is similar to what the Italians are finding as well. Is it as high as 50-70X as the Santa Clara study showed? I don't know and maybe it is somewhere in between.
The bottom line is we have the tools to do this kind of testing right now even if we are not going to engage in a massive testing program. The failure of will at the National level is just stupifying to me. I have friends in the public health community that have documented what is needed to reopen things in a sane manner. We have millions of people unemployed and some of these folks won't see their jobs return. Retrain people to help out on the public health side doing contact tracing. They can earn money and contribute to the public good. This is not rocket science!!
My problem with the California and New York tests involved the sampling. When I was studying sampling, it became apparent (in the academic studies) that very minor changes in sampling technique could lead to large errors in reported outcomes. (There was even an entire book devoted to the problem, called "Unobtrusive Measures.") In California, the people sampled (as I understand it) were actually recruited on-line, which raised the question of whether self-selection might have skewed results. In New York, sampling only people who were outside also probably skewed results; people who were self-quarantining were not tested. That said, I completely agree that the number of infections is probably far higher that what we have officially recorded. The question is, what is it? If the infection rate is, say, 5x higher, then the death rate would be many times worse than the flu. If it were 10x higher, then the actual death rate might be, say, .006% still quite a bit higher than the flu. But what if it turns out to be 100x higher? Then the death rate would be flu-like, and with adequate protections provided to the most vulnerable groups, we might be able to open the economy entirely, accepting flu-like death rates. (I doubt that would be the case, as we've never had this many deaths in so short a time, from a respiratory virus, since the Spanish Flu.)
As far as the false negatives are concerned, the doctor featured on CNN as their medical specialist, Sanjay Gupta, keeps citing the 15% false negatives as if that were an established fact; but I'm not sure that it continues to be, even though he continues to talk about that specific number.
In any case, I think it would be relatively simple and cheap to use standard polling techniques to find out about how many people are carrying the virus; I don't know why it hasn't been done. I'm not making any political accusations here -- neither red nor blue states have really done this. It may be a matter of "penny wise, pound foolish," thinking that the cost of a poll would be money better spent on other things related to the crisis. But that's probably incorrect, IMHO, for the relatively small amounts involved. I believe it would be really helpful to know about what we're dealing with -- I make no claims that such a technique would be perfect.
