Contribute Try STAT+ Today

The true number of coronavirus cases in the U.S. could be anywhere from six to 24 times higher than the confirmed number of cases, depending on location, according to a large federal study that relied on data from 10 U.S. cities and states.

The vast majority of Americans, however, are still vulnerable to Covid-19.

The study, published Tuesday in JAMA Internal Medicine, relied on serological tests — blood screens that search for antibodies to the virus and that determine whether someone was previously infected. They are different from diagnostic tests, which only detect people who currently have the virus, called SARS-CoV-2.


Overall, an estimated 1% of people in the San Francisco Bay Area have had Covid-19, while 6.9% of people in New York City have, according to the paper’s authors, who included researchers at the Centers for Disease Control and Prevention and state health departments. In seven of the 10 sites, the estimated number of cases was 10 times the number of reported cases.

The study was based on tests from more than 16,000 people across the 10 sites, but one limitation is that it relies on old data. The San Francisco samples were collected from April 23 to 27, while the New York tests were on blood from March 23 to April 1. The latest tests were conducted in May, and a lot can change during two months in the course of an outbreak. In South Florida, for example, researchers estimated that 1.9% of the population had antibodies to the virus. But that figure is based on samples collected from April 6 to 10, and given the spread of the virus since then in the state, the number now would certainly be some amount higher.


Still, the data reflect what CDC Director Robert Redfield recently said — that true case numbers are 10 times higher than confirmed diagnoses. Confirmed cases in the U.S. stand at more than 3.8 million.

The data underscore two other points: that testing in the U.S. is not capturing the full scope of the outbreak, and that even hard-hit communities are not close to reaching a herd immunity threshold — where enough people are immune from the virus (which scientists expect will happen for some amount of time after an initial infection) to slow down its spread to the point that unprotected people have a natural buffer.

“The study rebukes the idea that current population-wide levels of acquired immunity (so-called herd immunity) will pose any substantial impediment to the propagation of SARS-CoV-2 in the U.S., at least for now,” infectious disease experts Tyler Brown and Rochelle Walensky of Massachusetts General Hospital wrote in an editorial accompanying the study. Experts estimate that 60% to 70% of people in a given area would need to be protected from the virus — either through recovering from an infection or vaccination — to reach herd immunity.

Other locations included in the study and the estimated levels of antibodies in their residents:

  • Western Washington: 1.1%
  • Louisiana: 5.8%
  • The Philadelphia area: 3.2%
  • Missouri: 2.7%
  • Utah: 2.2%
  • Connecticut: 4.9%
  • The Minneapolis-St. Paul area: 2.4%

Overall, the researchers found that there was no association between infection rates and age or sex.

The results fit with other serosurveys that have found just a few percent of people in a given place have been infected with the SARS-CoV-2 virus, which causes the disease Covid-19. There have been a few outliers: One study in the hard-hit Boston suburb of Chelsea estimated that 30% of people had been exposed to the virus, while another survey in a German town where a carnival drove an outbreak found 14% of residents had antibodies.

Still, the new study landed on different estimates for New York City than a state-run survey released in April, which found that 1 in 5 people in the city had antibodies. The disparate results highlight how study design — such as how participants are recruited or what blood samples are included — can influence findings.

The new study relied on leftover blood samples collected from patients who sought medical care for any reason from March through May. Because so many appointments and procedures were canceled then, and because so many people were avoiding medical care during stay-at-home periods, the samples “are likely not representative of a typical prepandemic cohort,” Brown and Walensky wrote.

Experts note that the inability of diagnostic testing to keep up with cases is not just limited to problems with the tests, which have included a botched rollout, overwhelmed labs, and supply shortages. It’s also that some 20% to 40% of Covid-19 infections are asymptomatic. Those people can still spread the virus, as can people who eventually develop symptoms but don’t feel sick yet — which has complicated efforts to rein in the spread.

Researchers also stress that it’s still not confirmed if people who recover from Covid-19 are protected or for how long, or what levels (or titer) of antibody would be required to confer immunity. Some people with Covid-19 may not generate a robust antibody response, perhaps depending on how sick they get, though that remains an open question as well.

“At present, the relationship between detectable antibodies to SARS-CoV-2 and protective immunity against future infection is not known,” the study’s authors wrote. “Extrapolating these estimates to make assumptions about population immunity should not be done until more is known about the correlations between the presence, titer, and duration of antibodies and protection against this novel, emerging disease.”

  • 60%-70% Not really up on the herd immunity threshold statistics, are we, “Stat News”?

  • This article, like many, falsely equates not yet having a good explanation of HOW COVID-19 immunity works with not knowing THAT the vast majority of those who recover have immunity. The infection numbers would be completely different if recovery didn’t come with immunity. The falsehood is part of the attempt to keep people in line. There is no evidence at all of re-infection, which is extremely strong evidence of acquired immunity. The virus hasn’t been active long enough to know how long the immunity lasts, but there is no evidence of re-infection going back as far as we have infections.

    • mmm actually there are a lot of cases of patients who had the virus and recovered only to catch it again few weeks later because the immunity only last a few weeks for some people.

  • EXCELLENT ARTICLE!! As a physician and researcher I love statistics, and showing the stats from different areas and prognostic ideas really puts things in focus and shows that although we had several months of “” home study””and this is summer ,as my teacher used to say “”Mike, half of being smart is being careful “”.

  • All the stats are just flat out wrong and misleading. Start with the number of deaths. The USA does not currently have over 140,000 who died OF covid? We MIGHT have that many who died WITH covid, but not because of the virus. You can track the hundreds of news stories of people clearly dying from accidents, falls, injuries, drug overdoses, etc. who also test positive for covid (why are they testing them for covid upon death exactly?) who are in fact being marked as a covid death. My friend’s mother died of cancer (CLEARLY died of cancer) in April. He just received her death certificate from the county. She’s marked as a covid death and they won’t change it. So if the world can’t even agree that this is occuring, then we will get no where in examining and further numbers and stats. Multiple state health depts have admitted this is exactly what is happening. In the UK they are marking a death as covid if the person died say today from a fall, but they HAD covid in March. Look it up. Their health minister admitted to it last week and it was clealry spelled out on their web site. There isn’t one stat that is anywhere near accurate. Period. So trying to come up with a death rate is meaningless.

    • You are 100% correct. In the state of LA someone can test positive for Covid 8 times before testing negative and instead of it being reported as 1 new case it is reported as 8 new cases. This is so sad.

  • If infection rates are 24 times higher than reported cases (about 3.7 million as of last weekend), then the death rate (deaths — about 140,000 — divided by cases (3.7 million x 24= 89 million) is approaching the death rate of the seasonal flu, which is about .001 according to the CDC (maximum 62,000 deaths divided by 56 million cases).

  • I understand they are learning about this virus, but still have no settling answers. They need to make all people feel safe with some definite answers. Numbers keep changing from up to back down after real calculations. Vaccines aren’t making all people feel safe either that’s more scary than the virus. Let the politicians and CDC be the first to test it.

  • NO Dean W your math is wrong how you calculate a percentage is 140,630 /38,191,390=0.0036822436 and according to this website
    The 2018-2019 Flu season Estimates were 34,1557/35,520,883=0.0009616033
    So as of the information I used it looks you are about 4 times more likely to die from Covid 19 than last flu seasons Flu but still only less than 4 in 1000 and much higher as age goes up. That took me about 10 minutes of thinking and a little bit of internet research and some 5th grade math. Now if everyone would do this instead of just believing what ever agenda is being placed in front of them, then everything would be a lot better. With your math Dean W people would be dropping like flies.

    • Wrong Jesse. You state the ratio, but the original writer stated in PERCENTAGE and that is always ratio times 100. Hate to say so for the horrible numbers, but my math is right.

  • One more stab at the math if there are 140,630 fatalities from 38,191,390 that would be a death rate of .0037. If the USA has 327,000,000 people only .119 have been infected. The average likelihood death from COVID is .00044 to date. Of course 78 percent are people aged over 65. If you are under 65 the likelihood drops to .000097. I will take those odds. And, by the way, COVID-19 has been flying under the radar for almost a year.

Comments are closed.