If you find these numbers suspiciously low for the global epicenter of the disease, and if you’re inclined to distrust any and all COVID data that comes from China: Well, me too.

But if I’m not mistaken, no antibody study of a regional or national population anywhere in the world has shown prevalence higher than around five percent so far. One would think that it must be much higher than that in nuclear hot spots like Lombardy, for instance, or in New York City, where a recent sample of pregnant women in one hospital found 15 percent infected at the time. But the sort of figures we’re hoping for — 30-40 percent or higher — to suggest we’re closer to herd immunity than we think are largely absent. Even that hyped Stanford study of Santa Clara County that made the rounds yesterday showed a local prevalence of only three percent or so.

If any place on Earth could produce big numbers in antibody testing, you would think it’d be a hospital in Wuhan. But no.

Wuhan’s Zhongnan Hospital found that 2.4% of its employees and 2% to 3% of recent patients and other visitors, including people tested before returning to work, had developed antibodies, according to senior doctors there.

“This is a long way from herd immunity,” said Wang Xinghuan, the head of Zhongnan hospital, one of the city’s largest. “So a vaccine may be our last hope.”…

The antibody tests at Zhongnan Hospital involved 3,600 healthy employees, including security guards and cleaners as well as doctors and nurses. They also included about 5,000 visitors, including people who were required to take the tests before being allowed to return to work or leave Wuhan…

It’s striking that antibody results were roughly the same for hospital visitors and staff, given that staff are more likely to have encountered the virus. One explanation could be that hospital staff had more exposure to the virus but also more protective gear, said Robert Garry, a virologist at Tulane University in New Orleans.

Right, and another explanation could be that the data has been doctored to support China’s lie that the virus was quickly contained in Wuhan thanks to old-fashioned communist managerial ingenuity. Chinese authorities know that western countries will soon begin reporting prevalence in their own hardest-hit areas. If New York City comes in at, say, 10 percent, having an official rate in Wuhan of 2.4 percent gives Beijing a new reason to say they did a much better job of controlling their outbreak than incompetent westerners did.

But I don’t know. A higher prevalence rate in Wuhan could also be exploited by China for propaganda. If the rate reported by the hospital were 35 percent, for instance, China could point to its bogus official death toll of 3,869 as remarkably low relative to the high prevalence of the disease locally. What an amazing triumph for communist medicine, saving so many lives and preventing so many severe cases despite the fact that so many people were infected!

We can’t count on the accuracy of Chinese data. But how about data from the Netherlands?

The Dutch have been hit hard by the disease, with the fifth-highest number of deaths per one million residents of any major country in the world and the seventh-highest number of confirmed cases per million. And yet only three percent have been infected per antibody testing. Recent testing in Denmark of health-care workers, among whom you might expect much higher prevalence, found just 4.1 percent infected.

Want something closer to home? Okay:

Less than one percent in King County, Washington, which saw one of the earliest outbreaks in the U.S. and could have been a disaster if not for early and aggressive social distancing.

This result from Massachusetts is interesting but one question hovering over many antibody studies is how they’re selecting people for their “random” samples. Is it truly random, which would give us a sense of how the total population is faring, or are the people being tested self-selecting somehow?

Nearly one third of 200 Chelsea residents who gave a drop of blood to researchers on the street this week tested positive for antibodies linked to COVID-19, a startling indication of how widespread infections have been in the densely populated city.

Sixty-four residents who had a finger pricked in Bellingham Square on Tuesday and Wednesday had antibodies that the immune system makes to fight off the coronavirus, according to Massachusetts General Hospital physicians who ran the pilot study.

The 200 participants generally appeared healthy, but about half told the doctors they had had at least one symptom of COVID-19 in the past four weeks.

Note the last bit. Inviting passersby to walk on up and get tested might logically attract more people who have reason to suspect they’ve had the disease recently (e.g., they had a cough for a week), especially if those people were unable to get tested when they were under the weather. There might be much higher prevalence in a group like that than across the entire city. Likewise, as NRO’s Robert VerBruggen points out, people who are willing to take the risk of being out and about right now might logically have been more likely to contract the virus over the past few weeks than the share of the population that’s holed up at home for fear of getting sick. We’d expect risk-averse people to be less likely to get infected, but risk-averse people are probably underrepresented in a group that’s strolling past the town square.

Selection bias was also a potential problem in the Stanford study I wrote about yesterday. Scientists and other statistically-minded people have begun to pick apart that study’s methodology in the past 24 hours, with one critique focused on the fact that participants were recruited via Facebook ads. That poses the same problem as the Chelsea study by potentially attracting people who are *eager* to be tested, as a person who’s eager may have more reason than the average person to believe that he or she has been infected. If that’s true then the prevalence in Santa Clara County would be even less than three percent.

But we’re getting into the weeds. Here’s a common-sense approach from investor Balaji Srinivasan to the grand question of how many people are infected. Looking around at the hard numbers we have in the U.S. now, with more than 32,000 dead in the official count and many more “excess deaths” on top of that, we would need a gigantic number to have been infected already to produce a death toll like that if the fatality rate from the disease was as low as the Stanford study (and the Oxford model) imagines it to be.

In order to generate these thousands of excess deaths in just a few weeks with the very low infection fatality rate of 0.12–2% claimed in the paper, the virus would have to be wildly contagious. It would mean all the deaths are coming in the last few weeks as the virus goes vertical, churns out millions of infections per week to get thousands of deaths, and then suddenly disappears as it runs out of bodies.

I guess this isn’t strictly impossible, but I was skeptical when this theory was first mooted because it’s different from the way past influenza-like pandemics have played out, including H1N1 in 2009. That took 12+ months to infect 11–21% of the world over multiple waves.

COVID-19 would need to be way, way more infectious than scientists have come to believe, especially relative to other flu-like illnesses. And even if it was, remember that herd immunity is a moving target. The more infectious a disease is, the more of us need to be infected in order to slow down the rate of transmission. For a disease that’s wildly contagious like the measles, herd immunity requires something like 93 percent of the population to be infected for herd immunity. Whether the current prevalence in the U.S. really is on the order of five percent and we need 50-60 percent to get sick or whether the prevalence is more like 20 percent and we need 80 percent or whatever to get sick, we’ve still got a long way to go, it seems. Even if we ignore the new Wuhan data.