Nobody knows how many cases of Coronavirus there actually are - healthy people don’t get tested. But this interesting approach in The Economist suggests 20% of Americans might have had Covid. An interesting read.
A brief quote from the article:
Since there are multiple problems with testing, The Economist “obtained estimates of covid-19’s infection-fatality rate (the likelihood that infection will lead to death) by age and gender from a paper recently published in Nature, which combines death tallies and seroprevalence data (the proportion of blood samples with antibodies) from 22 studies. We then combined these with the total number of covid-19 deaths by age and sex per week in America, provided by the CDC… Finally, because the CDC’s breakdown of deaths is incomplete for recent weeks, and infections tend to pre-date deaths by about 14 days, we used official case data to estimate new infections in the past six weeks… [assuming] that the relationship between reported cases and actual infections remained similar for the past two months.
Our analysis suggests that between 18% and 24% of Americans have been infected so far, or between 60m and 82m people… This means that over the course of the pandemic, less than a quarter of infections have been recorded.”
It’s behind a paywall for me, I thought The Economist were making all their COVID stuff freely available?
We’ve always known that the infection rate is some significant multiple of the number of confirmed cases, right? Because as you say most people are asymptomatic and don’t get tested. Here’s a recent paper, the second figure shows the estimated detection rate for each country, i.e. confirmed cases as a percentage of total infections. They have the U.S. detection rate at around 20%, so multiplying the confirmed cases (17 million) by 5 gives a similar number to the estimate from The Economist.
My understanding differs considerably from this.
You think people with no symptoms and no specific reason to suspect exposure are getting tested in large enough numbers to detect any significant proportion of asymptomatic infections? Where do you get your understanding?
There have been a few efforts at random sampling studies, but that’s not the same thing.
Some healthy people get tested. Many don’t. In some countries, many can’t. The article goes into more detail about other problems with testing - how symptomatic one might be, how long antibodies stay positive, false results, the initial tests, etc. I was succinct because it’s not the main point.
On my campus alone we tested thousands of asymptomatic people last semester. We’re a tiny campus compared to the huge state schools.
My mother’s home care workers were getting tested twice a week, along with nurses, other healthcare professionals, nursing home and rehab center workers, across the nation and so on. All asymptomatic.
Yes, and I thought that was overwhelmingly common knowledge.
A significant proportion of asymptomatic people? Cite, please.
Yup. Our university system (6 schools total) just spent an entire semester testing 32,000 students twice a week, and close to 5,000 faculty and staff weekly.
Indiana’s first official case of COVID wasn’t until early March, but I wonder if I didn’t have it about a month earlier. I was in bed for 24 hours with something that I thought was a bad cold, except for the fact that it hit really hard, and then let up just as quickly. Colds don’t usually work like that in me, but that fact is, that since I had my tonsils out 20 years ago, I’ve gotten very few colds, so my points of reference were the stick-like-napalm colds I used to get before I had my tonsils out, when I’d get three or four bad ones a year that lingered forever.
This thing I had in February had me exhausted, and because I have a sleep disorder that is usually insomnia, but about every 18 months results in a period or 3-6 days of hypersomnia, where I’ll sleep from 9pm to 2pm the next day, I thought maybe I was starting one of those, until I started to get a scratchy throat and mild cough. I spent the day in bed, thinking if it didn’t let up, I’d get DH to drive me to the doctor to see if maybe somehow I’d acquired mono.
Basically, on Friday, I had a case of the sniffles, and was a little tired, but I went to work. Saturday, I was in bed all day with a temperature, but not a high one; Sunday, I was up and around a little, but taking Tylenol for a sore throat, and cold medicine. I couldn’t taste much, but I get that a lot when my nose is stuffy, so I didn’t think about it a lot. Monday, I was almost completely well. I was still taking some cold medicine, and my sense of taste was still off-- I could taste things, but sweet things tasted odd; my energy was back. I wasn’t scheduled to work, so I took it easy-- all I did all day was walk the dog. I didn’t even make dinner, so I wouldn’t pass on whatever I had. We had take-out that day. By Tuesday, I was perfectly fine.
At the beginning of January, my son and I had flown back from LA through LAX, on a completely booked plane. If there were any place to be exposed in the US in January, 2020, it would be LAX. I realize that the incubation period is about 2 weeks, and I was sick a month later, but I thought that maybe my son caught it (he was 13 at the time), was completely asymptomatic, and passed it on to me at the tail end of having it.
Assuming that I had it, there could be lots of stories like mine, especially considering how much travel there is at the end of December and the beginning of January.
It’s possible that you can look at the first confirmed case in every state; assume that the actual first case was really about a month earlier; that between the theoretical case, and the confirmed case, there were escalating numbers of cases for which the first confirmed case represented n cases; and where n is not only greater than one, but probably some pretty significant factor of the population.
If that is true, than yeah, there are, in fact, a heck of a lot of people who have been infected and never knew it. They were infected before there was a test, and couldn’t have known it, and probably even so far before a test for a previous infection, that it might even be unknowable, if it’s true that immunity recedes after a while.
I’d happily volunteer for a test for a previous infection, but my illness wasn’t documented, because I never saw a doctor, so a negative test is meaningless.
The thread was meant to be about the article. But since it is saying Coronavirus levels may be almost four times measured levels, discussing the flaws of testing seems fair.
The problem is not just that asymptomatic people are less likely to be tested, though this is often true. Certainly, some businesses, universities, researchers etc. have tested more widely.
The Economist article, above, also says:
“THIS WEEK America’s official tally of covid-19 cases surpassed 15m, around 5% of the population. Nearly 1.3m of those were recorded in the past seven days. But the true scale of the outbreak is almost certainly many times larger. By combining new research with official death tallies, The Economist estimates that 60m-82m people have been infected in America so far, around 3.2m of them in the past week.
It is not easy to get a handle on the spread of covid-19. Many people who are infected have no symptoms and never get tested. When they do, tests regularly give false positives and results may not even enter official tallies. Deaths are more difficult to ignore, and have no false positives. But not all deaths related to covid-19 are picked up. Our excess-deaths tracker (article has link) shows that nearly a quarter of deaths in America attributable to covid-19 may go unrecorded as such. And the number of deaths associated with covid-19 offers incomplete information on the spread of infections. A small outbreak among the old may be as deadly as a larger one among the young.
One way to track infections is by using mass random tests for covid-19 antibodies, which remain in a person’s blood after they have been infected. But these “serosurveys” are limited too. They can only reveal past infections and are not 100% reliable, and antibodies can become undetectable over time.”
After discussing there conclusion that between 18% and 24% of Americans have been infected so far… “over the course of the pandemic, less than a quarter of infections have been recorded. This proportion has improved from just one in 12 cases being detected in March, to more than one in three now, consistent with the huge increase in testing. The overall number of infections is also consistent with national estimates of seroprevalence conducted this summer. Another study in the Lancet (article has link) by researchers at Stanford University put America’s past infection rate at 9.3% in July. Since then, covid-19 deaths in the country have roughly doubled.”
I’m not sure anyone knows - if you had Covid in January 2020, what proportion of people with few symptoms (no symptoms, or ones they dismissed as minor at that time without seeking care) would now test positive if tested once. When they were tested is also relevant if 10-20% of cases (known and hypothesized) in the US were in the past week alone. Not all tests are the same, and immunity can perhaps show up in T or B cells instead of, or in addition to antibodies.
The article certainly deserves scrutiny and I am quite interested in reasonable criticism of their methods. But experts would still not be surprised to learn the actual number of cases is several times higher than the measured number.
Is it possible to be tested as having anti-bodies to CV19 but not have enough to effectively fight off a full infection?
I am thinking here or people who might have had extremely low level exposure that they have fought it off without even noticing and yet will test positive.
I think it is likely that many folk will have had very low levels - perhaps only one or two infective virii, I know that the initial loading in some infectious agents can make a huge difference to the full progress and extent of the illness
I don’t think these details are known in much detail - how much virus is required, how this may vary between different people, nor why some healthy young people get very ill.
All the data we do have indicate exactly what you’d expect given that most people are asymptomatic - that confirmed cases are only a fraction of the number of infections. With increased testing of asymptomatic people, that fraction is increasing somewhat - but increased testing is offset by increasing prevalence, and if you want to pick up asymptomatic infections you must either repeat PCR tests periodically or test for antibodies.
As you’d expect, places like South Korea are an exception. Low prevalence, massive amounts of testing, so they pick up a much larger fraction of infections.
Well, South Korea is starting to put up some bigger numbers. They might be missing more than we think. We might soon see how even a rather top notch contact tracing system handles a real outbreak.
I completely agree that we aren’t testing the majority of infected asymptomatic people though. People working in a school or nursing might have a very skewed view. People working in Walmart or a lightbulb factory aren’t getting tested twice a semester.
If the number who have been infected is higher than previously thought, isn’t that actually good news?
No.
The argument that this is good because it implies a lower infection fatality rate only applies if the only possible endpoint is herd immunity through eventual infection. If everyone is eventually going to catch the virus anyway, then a lower infection fatality rate than we thought is good.
But when we’re on the brink of immunizing the population through vaccination, the fact that there are more asymptomatic carriers is bad news, unless the number of infections is already so high that we’re close to herd immunity through infection. Barring that (and we’re not close to those numbers) it just means that the virus has a greater probability of reaching everyone it can kill before they are vaccinated.