Lots of excitement about this result on social media today, although there are two ways to look at it. One is the optimistic way, which I’ve highlighted in the headline. If the number of Americans who’ve had COVID is 50 to 85 times the number of confirmed cases then we’re further along towards herd immunity than we thought. Suddenly the Oxford model of the outbreak, which imagines many millions of undetected infections, is back in play.

The pessimistic way to look at it is the fact that, in Santa Clara County, a prevalence of 50-85 times the number of confirmed cases would mean that … three percent or so of the population there has had the disease. That’s right in line with experts’ current estimations, including Scott Gottlieb’s, that something like one to five percent of Americans have had the virus so far. The scientists who are encouraging lockdowns would expect a result like this.

As more antibody studies are done, my guess is that we’ll have that same reaction to many of them, if not all. Lots more people than we thought have had coronavirus, annnnnnd … we’re still nowhere near the numbers we’d need across the population to meaningfully reduce transmission rates in the near term via herd immunity.

Background Addressing COVID-19 is a pressing health and social concern. To date, many epidemic projections and policies addressing COVID-19 have been designed without seroprevalence data to inform epidemic parameters. We measured the seroprevalence of antibodies to SARS-CoV-2 in Santa Clara County. Methods On 4/3-4/4, 2020, we tested county residents for antibodies to SARS-CoV-2 using a lateral flow immunoassay. Participants were recruited using Facebook ads targeting a representative sample of the county by demographic and geographic characteristics. We report the prevalence of antibodies to SARS-CoV-2 in a sample of 3,330 people, adjusting for zip code, sex, and race/ethnicity. We also adjust for test performance characteristics using 3 different estimates: (i) the test manufacturer’s data, (ii) a sample of 37 positive and 30 negative controls tested at Stanford, and (iii) a combination of both. Results The unadjusted prevalence of antibodies to SARS-CoV-2 in Santa Clara County was 1.5% (exact binomial 95CI 1.11-1.97%), and the population-weighted prevalence was 2.81% (95CI 2.24-3.37%). Under the three scenarios for test performance characteristics, the population prevalence of COVID-19 in Santa Clara ranged from 2.49% (95CI 1.80-3.17%) to 4.16% (2.58-5.70%). These prevalence estimates represent a range between 48,000 and 81,000 people infected in Santa Clara County by early April, 50-85-fold more than the number of confirmed cases. Conclusions The population prevalence of SARS-CoV-2 antibodies in Santa Clara County implies that the infection is much more widespread than indicated by the number of confirmed cases. Population prevalence estimates can now be used to calibrate epidemic and mortality projections.

The unvarnished good news in every report of millions of hidden infections is that it means the fatality rate is much less than feared. The case fatality rate right now in the U.S. is 4.6 percent, a number no one takes seriously. If 50 times as many Americans are infected than we know right now, the true number of cases would swell to 33 million and the fatality rate would drop to flu levels, around 0.1 percent. That’s the good news.

The bad news is that multiplying the known cases by 50 would mean that just 10 percent or so of the total population has been infected, and that 10 percent has cost us 30,000 deaths so far. If we need to get to, say, 60 percent being infected in order to enjoy herd immunity, that would mean we should expect 180,000 dead before transmission rates decline and the epidemic tails off. Which happens to be right in the range of deaths predicted by the White House model.

The Santa Clara study isn’t the only news today about a huge number of hidden infections in a population. Down in Guam, tests are being run on the crew of the USS Theodore Roosevelt:

Roughly 60 percent of the over 600 sailors who tested positive so far have not shown symptoms of COVID-19, the potentially lethal respiratory disease caused by the coronavirus, the Navy says. The service did not speculate about how many might later develop symptoms or remain asymptomatic.

“With regard to COVID-19, we’re learning that stealth in the form of asymptomatic transmission is this adversary’s secret power,” said Rear Admiral Bruce Gillingham, surgeon general of the Navy.

The figure is higher than the 25% to 50% range offered on April 5 by Dr. Anthony Fauci, director of the National Institute of Allergy and Infectious Diseases and a member of President Donald Trump’s coronavirus task force.

I don’t want to confuse apples and oranges here. The Santa Clara test was an antibody test, showing everyone who’s had the disease — even those who’ve recovered. The Roosevelt test is, I believe, a PCR test with the nasal and throat swab, showing who has the disease *right now.* It’s possible that some sailors who had the disease and beat it are testing negative on the PCR but would test positive on the antibody test, meaning that the share of the crew that’s been infected is even larger than we think. (On the other hand, there’s no way to know right now how many sailors without symptoms will go on to develop some and how many won’t ever have any, the true asymptomatics.) The key point in both studies is the same, though: There’s obviously a *large* number in any population where the virus has circulated that’s been infected but has remain hidden to scientists, whether because they haven’t had symptoms or because they couldn’t be tested for some reason. The question is figuring out what that number is nationally.

And regionally. Maybe the regional number is more important, in fact. If the number of true infections in New York City exceeded the number of known infections by the same magnitude that it did in Santa Clara County — 50 to 85 times — then right now in NYC there would be, uh, 5.8 million to 9.9 million infected people. Note: NYC only has 8.4 million people. It’s a cinch that the share of the population in New York that’s infected is higher than it is in most/all other parts of the country, but it can’t be that they’ve already reached herd immunity. (I think.) Even if they had, it would mean that a huge chunk of the national population that’s survived COVID is located in one small geographic area. For most of the country, it’ll still be a long road to herd immunity.

There are other reasons to doubt the Santa Clara study. Note the bit in the excerpt about how people were recruited — through a Facebook ad, not through some sort of purely random sampling. It may be that there were people in Santa Clara who had symptoms last month, desperately wanted a test at the time but couldn’t get it, and now raced to sign up for the antibody test to see if they’re immune. If that’s so, then we might expect the Santa Clara sample to have more infected people in it than a true random slice of the population would have. Likewise, the reason the number with antibodies was 50-85 times the number of confirmed cases may have less to do with prevalence in the population than with the scarcity of testing. Look at it this way: If you had a population of 100,000 people, and 1,000 of them were infected, but only 10 were *known* to be infected via PCR testing, an antibody test later might reveal those extra 990 “hidden” infections to you and give you a mind-boggling rate of 99 times as many unknown infections as confirmed ones.

But the rate doesn’t mean much there. In reality, just one percent of the overall population is infected. The reason you got that gaudy “99 times!” number is simply because you tested so few people via PCR that you only found 10 to begin with. Do a crappy enough job of diagnostic testing, as we have in the U.S., and sure, your rates of true infections will seem sky high.

There’s another reason to doubt the study. Namely, if the accuracy of the tests they used is even a leeeetle bit worse than expected, their results could go up in smoke:

Here we go. They acknowledge this possibility, like good researchers. If test specificity <98%, the significance of the results goes away. pic.twitter.com/6Rr5IC9Zne

— Non-essential Fred (@LesserFrederick) April 17, 2020

Not to be too much of a downer about it but Scott Gottlieb says he would have expected a *higher* prevalence than three percent in Santa Clara County because it was an early hot spot. The bottom line is encouraging — many, many people out there have had COVID and beaten it, or are en route — but it’s not some vindication of the Oxford model.