Some Covid-19 Studies Are More Meaningful Than Others

(Bloomberg Opinion) -- New antibody tests are allowing researchers to begin to make more-informed estimates of how many people actually have Covid-19. They’re not perfect, though, and they’ve been saying greatly varying things about the spread and severity of the disease. Bloomberg Opinion columnists Justin Fox and Noah Smith — who, let’s be clear, are not epidemiologists, but have been reading a lot of epidemiology research lately — met online recently to discuss where things stand.

Justin Fox: Some early studies are showing that a lot more people have been infected with the new coronavirus than the confirmed-case numbers indicate. Testing of pregnant women in two Manhattan hospitals found that 15% had Covid-19, which if it were true today for the entire city's population would mean about 10 times more infections than confirmed cases. Testing for coronavirus antibodies in residents of Chelsea, Massachusetts, found that 30% were infected, which if true for the entire population would be about 15 times the confirmed cases.

Two brand-new new studies of California counties by an overlapping group of researchers have come to even more dramatic conclusions. They found that only a small percentage of the people tested in Santa Clara County (aka Silicon Valley) and Los Angeles County had antibodies that indicated they had had the disease (1.5% in the first, 4.1% in the second), but after a few statistical adjustments the estimates of infection was 50 times to 85 times the number of confirmed cases in Santa Clara County and 28 to 55 times that number in Los Angeles County. This would also imply that the infection fatality rate from the disease was close to the 0.1% of the seasonal flu, possibly even less.

This would be ... great news, an indication that the coronavirus pandemic might be over much sooner than most experts have been predicting. But it also seems too good to be true. And before I go into why I think it seems too good to be true, how about you, Noah? Do you think it's too good to be true?

Noah Smith: I think we need to be extremely cautious when interpreting these studies. The first reason is the possibility of false positives. As experts have been quick to point out, if most people in an area haven’t been infected, then even a small rate of false positives can make the number look much larger than it is. This is a problem for both of the studies you mention because California has low infection rates.

To see how this works, imagine a place where only 0.01% of people have actually been infected — essentially nobody. Now imagine giving everyone in the area a test that has a 1% false positive rate — a very accurate test! That test will nevertheless show an infection rate 100 times higher than it actually is.

Experts have identified other problems with the way these studies selected the people to test. Together, these issues suggest that we don’t know much yet about how many people have really been infected with this virus, and should continue to err on the side of caution. Much more data will be coming in soon.  

JF: And when it does, I take this to mean that we should pay a lot more attention to studies conducted in places that we already know have been hard-hit by the disease. It’s useful to know that 50 of 3,300 people in Santa Clara County and 35 of 863 in Los Angeles County tested positive (and that 13 of 2,283 randomly selected Icelanders did), but it seems like we shouldn’t be drawing any grand conclusions about the severity of the global spread of the disease from those results. A similar study in New York City or Northern Italy, though, could presumably tell us things with much broader application. If 20% of the population is infected, that false-positive rate of 1% only bumps the result up to 21%. Right?

NS: Exactly right. Note that there have been some large-scale tests of hard-hit regions such as the Italian town of Vo. That study found that 43% of infected people were asymptomatic at the time they were tested (some of those might have developed symptoms later). That’s very close to the asymptomatic percentage recorded on the cruise ship Diamond Princess, where everyone was tested, and in Iceland, where 5% of the entire population was tested. These are PCR (polymerase chain reaction) tests that check for the virus itself; in contrast, the studies you mentioned earlier are antibody tests, which are newer and perhaps less reliable.

It’s also worth noting that coronavirus has already killed about 14,000 people in New York City, which is 0.16% of the population. That should conclusively reject the idea that this is a disease with a 0.1% infection fatality rate. Add that to the fact that the virus is still killing many New Yorkers each day, that many others have probably not yet been infected, and it seems clear that this is a much deadlier disease than the seasonal flu.

JF: But let’s say 25% of New Yorkers have been infected. That implies an infection fatality rate of 0.67%, which happens to be almost exactly what the epidemiologists at Imperial College London have estimated for China. I’m a New York City resident, and I’ve been struggling to get my head around whether this will be helpful going forward because so many of us might have at least temporary immunity, or a disaster because there are probably still so many infectious people here that we’ll never get it under control.

NS: Yes, 0.67% is just about at the lower end of the reasonable estimates. If that’s the true fatality rate, and coronavirus eventually infects 60% of the U.S. population of 330 million, that’s 1.33 million dead Americans. Plus, remember that survivors of this disease often appear to have long-term damage to their lungs or other organs.

So allowing the virus to rampage unchecked would be an enormous loss in human terms, but it’s not as if we’d be getting a restored economy out of the deal. It’s highly likely that reopening the economy without a system in place to suppress the virus would cause mass panic and continued efforts at social distancing, thus extending and deepening economic losses. So let’s err on the side of caution and keep lockdowns going for now, until we have better data and a suppression system in place.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Justin Fox is a Bloomberg Opinion columnist covering business. He was the editorial director of Harvard Business Review and wrote for Time, Fortune and American Banker. He is the author of “The Myth of the Rational Market.”

Noah Smith is a Bloomberg Opinion columnist. He was an assistant professor of finance at Stony Brook University, and he blogs at Noahpinion.

©2020 Bloomberg L.P.