This is a very important piece of data. Stanford University researchers took the lead a few weeks ago in organizing a study to ascertain the true level of coronavirus exposure in the United States. They used antibody testing in Santa Clara County among 3300 volunteers to begin making that determination. Results from the initial phase of the study have been released. (ABC Story) They found that between 2.5% and 4.2% of the people tested had antibodies and had been infected. (As an aside, I assume the range results from ambiguity about the strength of the antibody response in some of the tested persons.) This ratio equates to 48,000 to 81,000 infected persons in the county at a time when the official positive number was around 1000. If you apply this to the country as a whole, according to Worldometers, we have and have had around 700,000 cases in the US. In reality that might be as many as between 3o million and 50 million cases in the country, assuming relatively even levels of initial exposure and timing. While I don’t know that the number is that high, I do think it is much higher than the reported level. And as always, with any study, the most important thing is to understand the caveats or limitations. For this one, that means the possibility that the sample was not truly random in some manner.
The significance of the results, however, cannot be understated. These were people who were asymptomatic or had mild illness. It means that the vast majority of infections result in no or low illness burden. The rates of serious infection and deaths would be very, very low. If there have actually been 36 million cases in the country, just to make the math easy, the fatality rate is one-tenth of a percent. So it has obvious implications for what kind of mitigation of spread measures we should implement. The study should help policymakers justify re-opening the country and relaxing mitigation of spread measures much more quickly.