Editor's note: This story was updated at 12:00 a.m. E.D.T. on Sunday, April 19
Way more people may have gotten coronavirus than we are detecting.
That's the takeaway from a small study of coronavirus antibodies in more than 3,000 people in Santa Clara County, California. The results suggested that between 2.5% and 4.2% of people in the county have contracted COVID-19, which is 50 to 85 times greater than the number of cases being reported at the time. Not everyone is convinced the true prevalence is that high, however, with some saying the antibody test the researchers used was not reliable.
However, this type of antibody testing, or serologic study, should be rolled out more broadly, epidemiologists told Live Science.
"I think this is a great start to beginning a serologic survey in the U.S., and I agree that we should expand this testing as much as possible so that hopefully we can figure out what level of antibodies, if any, is necessary to maintain immunity," said Krys Johnson, an epidemiologist at Temple University in Philadelphia.
So what does this mean for how deadly the virus is, how widely it has spread, and when we can ease social distancing? The answers aren't straightforward, epidemiologists told Live Science.
Related: Coronavirus in the U.S.: Latest COVID-19 news and updates
First, the study: Stanford University researchers used Facebook ads to find volunteers to be tested for antibodies to the novel coronavirus, or proteins produced by a person’s immune system to fight off a specific virus that has invaded the body. Roughly 3,300 of those volunteers came to a drive-through testing site on April 3 and April 4. One in every 66 tested positive for antibodies to the novel coronavirus. White women and affluent people were overrepresented in the population, while Latinos and Asians were underrepresented compared with Santa Clara's overall population.
A total of 50 tests came back positive. After adjusting for differences in zip code, race and sex between the sample population and Santa Clara as a whole, the researchers estimated that between 48,000 and 81,000 people in the 2-million-strong county had contracted coronavirus at some point. At the time, the health department was reporting about 1,000 positive cases.
The findings were posted Friday (April 17) to the preprint database medrXiv; they have not gone through peer review.
—Coronavirus in the US: Map & cases (opens in new tab)
—What are coronavirus symptoms? (opens in new tab)
—How deadly is the new coronavirus? (opens in new tab)
—How long does coronavirus last on surfaces? (opens in new tab)
—Is there a cure for COVID-19? (opens in new tab)
—How does coronavirus compare with seasonal flu? (opens in new tab)
—How does the coronavirus spread? (opens in new tab)
—Can people spread the coronavirus after they recover? (opens in new tab)
Less deadly than thought?
Using their data, the team estimated that the true "infection fatality rate" of coronavirus — or the number of infected people who die from the disease — is between 0.12% and 0.2%, or between 20% and two times more deadly than seasonal influenza (which kills about 0.1% of people it infects, on average). Other studies have estimated infection fatality rates between 0.5% and 0.9%, Nature news reported (opens in new tab).
Some experts have questioned the results, saying that when few people in a population have the virus, even a few false positives on the test could create the impression that there are many more coronavirus cases than actually exist, according to Nature.
The test used in this study has not been approved by the Food and Drug Administration (FDA) yet.
"They are constrained by the fact that the antibody tests they used were not very good, which they had to try and adjust for" who were infected, said George Rutherford, professor of epidemiology and biostatistics at the University of California, San Francisco (UCSF).
"The market's been flooded with these tests." Rutherford told Live Science. "But the FDA has relaxed its rules so there's not the same degree of quality control."
The crude rate of positives they found before making adjustments — about 1.5% — is probably about right, Rutherford said. However, using statistical adjustments to arrive at the range of 2.5% to 4.2%, and then to infer fatality rates, was likely a stretch, he added.
"The interpretation of the ratio of cases to death is an over-interpretation," Rutherford told Live Science. "
What's more, because they didn't take a random sample, the study is subject to what's called selection bias, Rutherford said.
"They may have picked off a piece of the population that was more likely to be infected or less likely to be infected, we just don't know," Rutherford said. (An example of potential selection bias: if someone suspected they had been infected earlier, but couldn't be tested when symptomatic, they might be more motivated to pursue antibody testing.)
Johnson, meanwhile, thinks the true prevalence in Santa Clara could be even higher.
"I think if they'd had an ethnically representative sample in this study as they'd hoped, they may have found an even higher proportion of people with antibodies, based on current reports that minorities are disproportionately affected by COVID-19," Johnson told Live Science in an email. "This would mean that even the informative conclusions here are still a conservative estimate of the likely number of infected people in Santa Clara County and throughout the U.S."
But on the other hand, the infection fatality rate in Santa Clara can't be directly translated to other spots in the U.S., which face higher rates of obesity and other chronic conditions known to worsen the outcomes of COVID-19. So infection fatality rates in other American cities may be higher than the Santa Clara County estimate, Johnson said.
Ultimately, it's just one sample in a single locale, said Dr. William Schaffner, an infectious diseases specialist at Vanderbilt University in Tennessee.
Schaffner suspects the 50 to 85 times higher prevalence "is on the high side" — meaning the true infection fatality rate could potentially be higher. But without doing antibody testing in several other places and populations, there is no way to know for sure, Schaffner told Live Science.
Mild disease and catastrophic impacts
If the numbers are in fact representative, though, how can this relatively low fatality rate be reconciled with the catastrophes that have unfolded around the world? How can a disease that's only slightly more deadly than the flu have caused China to shut down its economy for two months, brought the country's largest city to the brink of collapse, and kept 1.5 billion children out of school?
It turns out, that's definitely possible, because before late last year, no one on Earth had been exposed to this virus, so everyone could catch it. By contrast, many people will be immune to viruses that have circulated before, and only a fraction of the population is susceptible to catching those. Even if the novel coronavirus virus is not that deadly, it could kill many more people than a known, but similarly deadly bug simply because it has the potential to infect a greater proportion of the population. That can easily overwhelm the health care system, Schaffner said.
The flip side of this data is that nowhere in the U.S. is likely to have most of its population exposed to SARS-CoV-2 at this point, Schaffner told Live Science. So the idea of us being close to "herd immunity" — when enough people have gotten the virus and are immune that the disease can no longer spread — is wishful thinking.
In Santa Clara, at least 95% of the population is still susceptible to the virus, Schaffner said. "So we can't depend on any kind of herd immunity to slow down this virus yet."
Extrapolating data from one locale to another is always statistically dicey, but even in New York City — where reported deaths from COVID-19 already exceed 0.1% of the city's population — some other numbers suggest that about 15% of the population has been infected. That's well below what's needed to naturally slow the spread of coronavirus, Johnson said.
That said, the numbers do suggest caution before mandating social distancing too far out into the future based on epidemiological models, especially without taking into account practical factors, such as the societal costs of social distancing, Schaffner said. (Some health experts have suggested some form of social distancing may linger into 2022, unless a vaccine becomes available sooner.)
"Social distancing, into the fall and winter, I think is reasonable, and then let's see," Schaffner said.
Editor's Note: This story was updated to include comment from George Rutherford.
- The 9 Deadliest Viruses on Earth
- 28 Devastating Infectious Diseases
- 11 Surprising Facts About the Respiratory System
Originally published on Live Science.
OFFER: Save 45% on 'How It Works' 'All About Space' and 'All About History'!
For a limited time, you can take out a digital subscription to any of our best-selling science magazines for just $2.38 per month, or 45% off the standard price for the first three months.
One is the respiratory version where the virus containing droplets are inhaled, it infects the lungs, symptoms are severe often putting victim in in intensive care, or even causing death.
The other is the gastrointestinal version where virus is ingested (oral transmission) where victim has relatively mild nausea or diarrhea or stomach pain and recovers without serious consequences.
Obviously, until the virus has completely run its course (12 to 18 months) we are going to have to do everything we can to avoid inhaling it. This means practice of social/physical distancing, any wearing of masks in situations where distancing is not practical.
Good hygiene (washing of hands before preparing or eating food), and cleaning and sanitation of food handling surfaces and utensils to prevent spread of any food borne illnesses and cross contamination of course continues to be important as always.
Some easing of the more onerous business shutdowns, travel restrictions, and stay-at-home precautions if everybody got serious about distancing and/or wearing of masks should be possible, but so far large percentage of our societies seems to be incapable of even practicing these simple measures. SAD!
But if we can't fix stupid, maybe its time to ease restrictions a bit and let Darwinian selection take care of some of the more stupid aspects of our gene pools.
1. The false positive rate for this antibody test is too high.
2. The participants were not randomly selected, they were allowed to put themselves in the study. The critics pointed out there is evidence from social media that many participated in the study to get a free test, because they suspected they had been infected. '
If they suspected correctly, then a much higher percentage of these people were actually infected than the population at large. How much higher we do not know, but it basically invalidates the study.
This is very important - a person looking at the reported numbers might think "1,000 cases in a county with 1 million people means any one person I encounter has only a 1 in 1,000 chance of being infected why should I worry, the odds are way on my side" - but the true number was so, so, so much higher.
This very misleading way of reporting probably led many to be careless and get the virus.
James - your claims above are not well based in the known facts. Please read up some more, but also, check my Comments on this article, which will explain there are a lot of problems with this study - and also, PLEASE READ THE STUDY OR IT"S DATA AT LEAST - the numbers found in the study were something like 5% of the participants had antibodies - that is nowhere near "herd immunity" - And MOST of the country has lower known rates of virus than Santa Clara County has - so, they would have LESS than 5% of people with antibodies- this is ASSUMING the study is valid at all - we are nowhere near herd immunity and to get there, if we were at 5% immnity, we would need to have 20 times the infections - which at the current time would also mean 20 times the deaths- not 35,000, but 700,000 or more - to get to it - but in reality, since this study is wrong, it would be millions. Please do some research.
A simple, though error-filled computation of one infected person infecting one additional person per day, over a month, while the other infected person(s) likewise infect one each additional person per day, you wind up with a rapidly expanding population of infected people, reaching over a billion infected people within 30 days. Of course that is unrealistic as you quickly run out of non-infected people to infect, long before reaching a billion. But consider we had a quarter-million people arriving into the USA from China (as new arrivals or returns from family visits) BEFORE Trump shut the border. How many of them were infected? Nobody knows, but if only 10% were infected, that's 25,000 infected people, spreading across the land. By the one-person-daily preliminary calculation above, we could expect a tremendous number of people infected over the period from November through January. And aside from the problems you note in the study under question, the numbers of infected-asymptomatic should be even larger still, maybe 10, 100 or 1000 times as many, percent-adjusted for the full population.
Now, I'm retired from university teaching of such as population dynamics and environmental issues, including a bit of epidemiology, but such as I describe above could be organized into yet another computer model, and with some careful presumptions, my intuition suggests both high and low numbers of infected-asymptomatic-immune people to be even greater than what that study concluded. Which means,
1) the death rates of the Wu-Flu are an even smaller percentage of those who got infected.
2) we closed the border after "the horses left the barn" (or the quarter million ran into the barn) so to speak.
3) all the isolations, distancing, masking and shut-downs happened only AFTER the major sweep and spread of infections took place.
An incredible number of people were already infected by the end of January, but the vast majority recovered or never showed symptoms.
This suggests, strongly, that herd immunity has been upon us probably since early February, that no vaccine is necessary (and who in their right mind would take it, after all the despotic revelations about Fauci and Bill Gates, their wish for "vaccine certificates" and such), and that the numbers from the CDC cannot be trusted anymore (they began including deaths from all different causes into WuFlu category, without any testing, with money offered to hospitals for doing so). From this it is very clear, the nation should emerge from lock-downs immediately, to face the very real threat and danger now posed by 23 million new unemployed, with massive small business bankruptcies, home forclosures, a new group of impovershed people, and all the morbidity and mortality that attends to economic devastation and depression.
Also the unmasking of so many tin-pot dictators at our state and local levels, wishing to punish people for ordinary social behavior, is disgusting, and yet another reason to remove this issue from the hands of the medical and virological professions, who have done a grave disservice to the nation, and the world.
I predict, a year from now, the medical and academic professions, and the Communist Chinese, will have HELL to pay for this disaster they foisted upon the nation. If this had been treated more rationally like prior corona viruses from China, in a manner similar to how Sweden did (no lockdowns or ubiquious anti-social distancing, masking, etc.), we would be at about the same situation in terms of deaths as we are today. There are no overwhelmed hospitals, with plenty of available beds, the ventillator hysteria calmed down when thousands of them remain unused, and so forth. And there is NO evidence that this reduction is due to any of the lockdowns, distancing or masking. The nation got a huge dose of infections spreading across the land by the end of January, and somewhat similar to other flu episodes. It killed a lot of older people with serious pre-existing conditions, but not in numbers significantly larger than last year's flu season, which claimed ~80,000. What remains unusual is only that young people and children remained largely unscathed, which is not so typical of infectious epidemics.
Also, if the epidemic had almost silently run it's course just a few months back, so that herd immunity is in effect near Stanford, then their study should be indicating much higher numbers of people with antibodies, but they claim only 5%.. Even if that number is 100% correct and their estimates of case fatality rate are correct, 1/50th of previously reported - that might be a reason to reopen the country, but it is not anywhere near hear immunity.
https://www.nytimes.com/2020/04/10/nyregion/coronavirus-new-york-update.htmlNYC had plenty of "non-infected" by which the numbers could continually build for a longer time than elsewhere, thereby yielding worst-case symptoms later than other places. In California, which got the brunt of the travellers from China, reached its peak much sooner. Interesting conversation, but I got to go. Like all others, I look at the data and make my best evaluation of it. Let's hope I'm correct, as it would be better for all. Cheers, JD
The sample used for the COVID-19 data in this case is not representative, the test is not approved, and furthermore, unlike your mortality calculation based on this "study", the mortality of influenza is not assessed based on the number of possibly seropositive people in a population, but on the number of clinical cases, which are often not even confirmed by a PCR test. It means that many of the fatalities attributed to the seasonal influenza have not been confirmed by a PCR-test, just as little as a portion of the registered deaths attributed to coronavirus
Stop comparing apples and oranges, and jumping to conclusions based on this.