Consumer Beware: Coronavirus Antibody Tests Are Still A Work In Progress

Yves here. In addition to providing a helpful overview of the state of play in coronavirus antibody tests, this article underscores a key point: scientists still don’t know what if any immunity one gets from having had a case of coronavirus.

By JoNel Aleccia, Senior Correspondent at Kaiser Health News, who previously reported for The Seattle Times,, and Originally published at Kaiser Health News

After hearing for months about serious access issues involving tests that diagnose COVID-19 based on swabs from the nose or throat, Americans are being inundated with reports about promising new tests that look for signs of infection in the blood.

There are high hopes for these antibody tests, which detect proteins that form in blood as part of the body’s immune response to an invading virus. Communities across the U.S. have been rolling out the results of serological surveys that examine blood samples from people who haven’t been diagnosed with COVID-19 to see if they were, in fact, previously infected.

The thinking is, if there are blood markers that can detect when people have been infected, such tests should be able to tell us how widely the novel coronavirus has spread. And equally optimistic: those same antibodies could convey immunity to the disease, signaling someone is safe from reinfection and able to get back to work.

Such high hopes, however, are running smack into the roadblocks of reality.

Infectious disease experts are raising pointed questions about the reliability of the early tests and the studies that hinge on their results. And they warn that state and local governments — as well as individuals — should be wary of shaping policy or changing behavior based on any single report.

In the sharpest caution to date, officials with the World Health Organization on Saturday warned against plans for proposed “immunity passports,” which would allow people who have recovered from the coronavirus to resume unrestricted travel and work.

“There is currently no evidence that people who have recovered from COVID-19 and have antibodies are protected from a second infection,” the agency wrote in a scientific brief.

Even before the WHO weighed in, other experts were urging restraint in interpreting early results of antibody screening.

“The science is catching up,” said Dr. Liise-anne Pirofski, chief of the division of infectious diseases at the Albert Einstein College of Medicine and Montefiore Health System. “Our ability to make a test at the moment is much greater than our understanding of what those antibodies we are testing for mean.”

In the past few weeks, more than 180 academic centers, hospitals and private manufacturers have notified the federal Food and Drug Administration that they intend to create serology tests for COVID-19, spokesperson Stephanie Caccomo said in an email. They’ve been able to jump into the fray because the FDA in March relaxed regulations for developing tests as part of its emergency response to the pandemic.

But the FDA has not reviewed the vast majority of tests on the market, and their validity, particularly point-of-care blood tests that promise rapid results within minutes, isn’t clear, said Dr. Michael Busch, director of the Vitalant Research Institute and a professor of laboratory medicine at the University of California-San Francisco.

“Some of them have sensitivities that are quite poor,” he said. “You may even miss some infected people completely.”

Other tests may flag people as positive for COVID-19 when they’re not infected. That’s especially true in regions of the country with little spread of the novel virus. If the prevalence of a disease is low, less than 5%, even an accurate test would yield a high number of false positive results because of the way such screening tools operate.

So when people see advertisements for finger-prick antibody tests becoming widely available at urgent care centers and medispas, they should think twice.

For one, antibody tests can’t be used to diagnose the disease. Antibodies may not be present in high enough levels to be detected in the earliest days of an infection. And because there are several other known coronaviruses — including those that cause the common cold — people infected with those viruses could produce antibodies that cross-react with those produced in response to the new virus.

Scientists still know too little about whether antibodies to COVID-19 convey immunity that could allow people to put away masks and halt social distancing, said Dr. Mary Hayden, director of the division of clinical microbiology at Rush University Medical Center in Chicago.

Immunity to a virus is a complicated process that takes place over one to two weeks, the WHO noted. The immune system makes antibodies in response to an infection. But the body also makes T-cells that recognize and eliminate other cells infected with the virus, creating what’s known as cellular immunity. Those two processes together may help a person recover and prevent reinfection. But it is not yet clear whether cellular immunity is required to bolster recovery and prevent subsequent infection with COVID-19.

“We do not know whether or not the antibodies detected are protective,” Hayden told reporters last week on a call organized by the Infectious Diseases Society of America. “We recommend that people with antibodies not change their behavior in any way.”

Scientists are hoping, however, that future COVID-19 studies may demonstrate immunity that could last for one or two years.

Concerns about the validity of the tests have cast a shadow on several recent reports aiming to quantify the spread of the virus in specific regions. Last week, New York Gov. Andrew Cuomo revealed the results of a serological survey that suggested that 1 in 5 New York City residents had been infected with the coronavirus. Statewide, the figure was 13.9%, according to the study of 3,000 New Yorkers in 19 counties who were recruited at grocery stores.

But the results quickly drew criticism. Dr. Demetre Daskalakis, who directs the city’s disease control, warned that the tests could produce “false negative or false positive results.” Florian Krammer, a microbiology professor at the Icahn School of Medicine at Mount Sinai who designs such tests, tweeted — and later deleted — that the results were “BS.”

“I think this is too high,” he said in a later tweet. “It is possible. But a 20% plus infection rate seems too high for NYC due to a number of reasons. I would think 6-8%, maybe 10% are closer to the truth. It would be nice to know more about the test, its sensitivity and specificity and the test population.”

Similarly, two serology studies in California, one in Santa Clara County and one in Los Angeles County, drew wide criticism about the recruitment of subjects and the analyses used.

In the Santa Clara study, Stanford University researchers tested 3,330 volunteers for antibodies showing exposure to COVID-19; about 1.5% were positive. They concluded that meant from 48,000 to 81,000 people were infected with the virus in the county.

“It was completely inadequate to interpret the results that 50,000 to 80,000 people were infected,” Busch said.

The L.A. study, conducted by University of Southern California researchers, concluded that 2.8% to 5.6% of the county’s adult population had been exposed to the coronavirus. That translates to 221,000 to 422,000 adult residents who have been infected. Critics, however, argued that the study sample was too small and that details of the methodology weren’t immediately available.

Busch understands the drive to conduct such tests.

“People are asking the questions: What’s the real denominator to judge the case counts and the death counts against?” he said. “People are urgently trying to get data.”

Unfortunately, that data simply is not available yet, other experts said. This coronavirus has never been seen before, so the science that will inform efforts to help communities respond and recover is playing out in real time.

“The problem is that the science has not kept up with the tests,” Hayden said. “Now we need to do the research to tell what the results mean.”

On the positive side, most of the scientific community has pivoted to focus on finding solutions, said Pirofski, who was also on the IDSA call. “We just have to slow our roll.”

“This is our first dive in trying to understand what’s going on,” she said. “I would say it’s a start.”

Print Friendly, PDF & Email


  1. sd

    Question: would using open source help move this process along? It feels like there are too many competing interests, a consistent lack of detail, etc. Now just seems like a good time to set aside profit motive.

    1. Carla

      No kidding. Well, we can always dream. That is, if any of us are able to sleep, perhaps we can dream…

  2. Ignacio

    I have been asked more than once by friends and family, what test should I have?
    My answer: unless you are asked in a survey or you have symptoms compatible with Covid-19 (yes, wide range of symptoms) there is no need to waste your money on tests. Given that the window for accurate PCR detection is relatively narrow (best results between 6-10 days after symptom appearance) one should call the physician in advance and have samples taken preferably within this period. I would also recommend something I learnt here: test yourself with an oximeter to detect “silent pneumonia”, this might give you more relevant information regarding your condition than the test itself. If you find that oxygen saturation goes below your normal level (let’s say somewhere between 94-99% saturation) call your physician. Insist if it goes below 92%.

    I wonder if in serology studies like that in Santa Clara or NYC they could be detecting cross-reactions with people that had been before infected by other virus that share epitopes with SARS CoV 2 or any other weird possibility. When you do massive tests you will have lots of surprises.

    Too many are too obsessed with tests.

    1. Dwight

      Are you suggesting the monitoring with oximeter for those with fever or other symptoms? I think so, but want to confirm. Also, would it be good to measure a few times before one gets sick, to establish a baseline? Thank you for this and all your great comments.

      1. Ignacio

        Monitoring with the oximeter even without symptoms. Yes, with a baseline taken before and doing the monitoring in the same conditions used for the baseline. The same clean and not cold finger with hand resting and nail not painted. At home is 98-99% O2 saturation all times all subjects so far.

        1. marku52

          Might vary with machine. My results vary from 94% to 98%. It would be a good idea to get a baseline with the machine you use before you think you are sick.

          1. Information v, Noise

            A friend of mine who is a professor of integrative physiology told me that oximeter readings are 2-4 percentage points lower on average for people living at altitude (I’m at 5,400 feet above sea level). That seem right Ignacio?

    2. polecat

      “one should call ‘the’ physician and have samples taken.”

      “call ‘your’ physician”

      Nearly worthless advice if seeing a doctor is unobtainium via gatekeeping & penury, in a system that survives wholely on Churn!

  3. voislav

    I’ll chime in a bit since this coronavirus antibody testing is my area of research and I’ve had many discussions with people over this in the past couple of weeks. For those who are unfamiliar with how antibody testing works, there are 2 main types of tests. First is the assay test, where you look and identify all antibodies in patients blood and then single out the disease-specific one. Second, more common, is the targeted test, where you use a protein target to bait the antibody out of patients blood. If the test is rapid, typically, there is no effort to identify the antibody that was baited out, just the fact that an antibody stuck to that specific protein bait is used as evidence of the coronavirus antibody.

    There are 3 main issues with COVID-19 antibody testing:

    1. Test Selectivity – how selective is the test for COVID-19 vs. other coronaviruses. Coronavirus family has a degree of commonality, so a protein bait that works for COVID-19 may pull out antibodies associated with another virus, creating a false positive. Lab testing is being done so fast for COVID-19 that there is limited opportunity to verify protein bait selectivity when used on a large population.

    2. Stage of the infection – Immune system reacts differently in different people in different stages of the disease. The antibodies that are released are different (IgM vs. IgG, or even different types of IgG’s). Most antibody tests typically targets a single antibody (unless it’s an assay test), so antibody selection has a great effect on whether a test is positive or negative.

    3. Test Integrity – these tests are being rushed out. In my experience, test results can be very different when performed by properly trained personnel who have been doing that job for years, vs. hastily trained technicians with no experience. One of our antibody tests was affected by how fast the person was pipetting the reagent. So a false positive rate of X in a lab with highly trained personnel does not equate to the same rate in the field with mass trained technicians.

    Which brings me to my final point. Any antibody test that will be administered en masse will need to be backed up with lower volume quality control assay testing on same patients. That way we can evaluate the quality of the mass testing and make sure that the positive rate is as realistic as possible.

    I’ll give props to NIH who are aggressively moving behind the scenes to make this happen. You won’t see this widely publicised, but I believe that NIH will have a reliable antibody testing system within 3 months and it will not involve any of the clowns that are rushing these initial tests through. What’s happening right now is the test equivalent of hydroxychloroquine, unreliable tests being pushed by media hungry individuals.

    1. Information v. Noise

      Thank you for this great explanation. Can you summarize how accurate the best anti-body tests are that are currently in use for other known viruses, so we can get a sense of how accurate such a test might be for COVID? What are their false negative/positive rates? For example for the flu, for measles, for polio, etc…

      1. voislav

        Test accuracy really depends on the virus. Quick tests are usually some variation of ELISA test and the accuracy does not depend on the test itself but on the virus. For example, COVID-19 spike protein is a common target, but two other betacoronaviruses that are common in humans also use similar spike protein. So, is the body going to produce antibody that targets only COVID-19 spike protein or will it attack the other two as well? Typically it takes us 12 months to work out the protocol, including target/antibody selection, and sometimes there is no good target, we can’t make the test specific enough.

        Assay tests are much better, they are basically 100% accurate, but they take longer to run, a few hours per patient, and require more specialised equipment. Before all this we were heavily invested in increasing the throughput for assay tests so that they can be used for most clinical applications. This is now ramping up even more as NIH is looking for some way to validate ELISA tests that will be done on general population.

    2. Cuibono

      yes the tests are being rushed out. Yes there is lots to know about their performance. but to call them all into question like this is really scientifically dishonest. In fact the results are in line with expecations from a number of other observations.

    3. rd

      Useful info. I had assumed, based on my general background in science and engineering, that it would be mid-to-late summer at the earliest when enough duplicate testing and other test validation protocols executed to have reliable antibody testing available. So I had targeted September to look for a test that could be reasonably accurate.

      I suspect that the general poor state of testing globally due to the short time frame and rushed protocols could be behind soem of the questions about whether or not people become immune. There is probably a threshold amount of antibodies required for solid immunity and the testing is probably insufficicent accurate right now to either pick up the coronavirus proper vs some other surrogate or quantify the amount of antibody accuracately.

    4. Barry

      In fairness to Dr Ioannidis he stressed in an interview that the tests have some problems and they still were good enough for the time being to extrapolate the number of people infected were likely 35 to 85 fold higher than what we had thought. His history with being careful and polite but critical of inaccurate and biased research makes me feel much better about things. I don’t feel he is politically biased here. Unlike the administration and WHO and especially the CDC.

  4. k teh

    This virus is not an Act of God, despite the growing cacophony of disparate fascists looking for the contractual exit door. There is no arbitrary limit to the number of productive human beings this planet can carry. There is a definite limit to to the number of MMT consumers however. The idea that the economy is 70% consumption and therefore the consumers must be subsidized is insane.

    Issuing debt in exchange for natural resources, to be given to whatever corporatists happen to be in charge and hiding the bill in actuarial carry, is just a crappy system humans have been employing for thousands of years. Nancy, Mitch and the gold bugs just haven’t got the news yet; globalization has run its course and de-urbanization is already well on its way.

    As far as welfare goes, absent allowing the real estate market to fall, all the programs – UBI, food stamps, etc. – are a hollow gesture. Real estate doesn’t pay for itself; real estate inflation drives income inequality and falling living standards. Setting up banks to arbitrarily decide who gets to collect interest and who must pay doesn’t work, as this bailout attests.

    It’s all the welfare programs for favored real estate “investors” and government taxation of unfavored users that is killing the economy. It’s still feudalism, but everyone gets an iphone. I guess a direct immunity chip to participate in the social credit system is next.

Comments are closed.