A Project of The Annenberg Public Policy Center

The Facts on Coronavirus Testing


In the wake of an increasing number of new coronavirus infections in the U.S., federal officials have provided confusing and sometimes contradictory statements about the number and availability of tests to diagnose the new respiratory illness.

President Donald Trump claimed on March 6 that anyone who “wants a test can get a test,” while Secretary of Health and Human Services Alex Azar said in a television interview the same day that there “is no testing kit shortage, nor has there ever been.”

Contrary to both men, we were told by the Association of Public Health Laboratories on March 9 that although things had improved since a week ago, “demand for testing is greater than the tests available.”

At the same time, at least one Democrat has exaggerated the problems.

Former Vice President Joe Biden said on March 1 that the Trump administration didn’t “even have a test kit” for the new coronavirus. That’s not true. While the Centers for Disease Control and Prevention is widely viewed to have bungled the rollout of its viral test kits, there was a test kit at the time, and it had been successfully used by the agency and some states.

The novel coronavirus, now known as SARS-CoV-2, is responsible for the pneumonia-like illness called COVID-19, which first appeared in an outbreak in Wuhan, China, at the end of 2019.

We’ll explain how testing works, what happened with the CDC’s coronavirus test and what’s known about how many tests are available in the U.S.

How does a coronavirus test work?

Most of the available diagnostics are molecular tests that use a standard molecular biology technique known as the polymerase chain reaction, or PCR, to detect even a very tiny amount of the virus in a specimen.

The test takes advantage of some natural features of biology to essentially scan through all of the RNA present in a sample — such as a nose or throat swab, or some phlegm — and search for the presence of coronavirus RNA. The scanning happens with test components known as primers and probes, which are sequence-based, and therefore can be made specific to the virus. 

In those reactions, if the novel coronavirus is present in the sample, then a short snippet of DNA is copied many times over, which can be read out on a machine. If there is no virus — or if there’s a different virus, such as influenza — then no DNA fragment is made and there’s no signal to detect.

The CDC’s test was designed to use three main sets of primers and probes — two that match just the novel coronavirus, and one that matches a variety of highly similar viruses. Someone infected with COVID-19 should test positive for all three, while someone who’s not infected should be negative for all three. A person who has SARS would only test positive using the third primer and probe set, thereby offering a way to theoretically discriminate between SARS-CoV-2 and other similar viruses. That concern, however, is hypothetical in this case, since there isn’t a known outbreak of SARS, or severe acute respiratory syndrome.

Eventually, the CDC will have a different so-called serology test that looks for whether a person developed antibodies in response to infection with SARS-CoV-2. The test, which does not need to be done when someone is actively infected, can yield a positive result whether or not someone developed symptoms. It therefore may provide answers about how many COVID-19 infections there were and how deadly the virus is in the larger population.

What went wrong with the CDC’s test kits?

Early on, all COVID-19 testing was done at the CDC with its own kit, and it performed well. The kit was successfully used to diagnose the first COVID-19 case in the U.S. on Jan. 20 in a patient returning from Wuhan to Washington state. A CDC report also notes that between Jan. 18 and Feb. 23, the agency tested 2,620 specimens from just over 1,000 people.

Problems arose after Feb. 5, when the CDC began shipping its kits out to qualified state and local public health labs in an effort to expand testing. Initially, the agency said it was releasing 200 kits — each capable of testing around 700 to 800 specimens — giving labs just one kit each. But by Feb. 12, the CDC told reporters that some labs had been experiencing issues when doing verification tests to make sure the tests worked.

The CDC has not clarified exactly what the issue was, but has said that there was a manufacturing problem related to the third test component, which led to “inconclusive” results. The Association of Public Health Laboratories told us more specifically that a negative control, which should not have reacted to any of the primer and probe sets, was coming back positive for the third set, which was the test for all SARS-like coronaviruses.

For a while, the agency promised to re-manufacture the faulty third component and distribute that to the public health labs and, in the meantime, asked any lab that had difficulties to send samples into the CDC for testing. As a result, as of Feb. 25, only 12 state or local labs out of more than 100 nationwide could do their own testing.

More than two weeks after first announcing the testing issue, the CDC said it had come up with a fix — labs could proceed without the third test, and just use the two other primer and probe sets to make accurate diagnoses. The agency also said that it had manufactured new tests with just the two components, omitting the third.

What other factors limited testing?

Other than the lack of CDC test kits, several other circumstances prevented wider scale testing early in the outbreak.

One was the criteria the CDC used to determine who would be tested for the virus. At first, the CDC recommended testing only for those who had a fever and/or lower respiratory symptoms, and had traveled to Wuhan or had contact with a suspected or confirmed coronavirus case.

In late February, after a patient in California was found to be infected despite a lack of known exposure, the CDC guidelines were relaxed to include anyone with a fever who was hospitalized with a respiratory illness. Then, on March 3, Vice President Pence announced that the CDC would lift all restrictions, as long as a doctor ordered the test.

The CDC updated its clinical criteria web page the next day, stating that clinicians “should use their judgment” in deciding who should be tested, keeping in mind the local epidemiology and COVID-19 symptoms — and strongly encouraging testing for other respiratory infections, such as influenza.

Another issue had to do with who could do COVID-19 testing.

Academic labs and other similar facilities that had the expertise to develop tests were advised not to use their own tests for diagnosing patients until they sought permission from the Food and Drug Administration. As we’ve explained before, this is not a law but has been a de facto FDA policy. On Feb. 29, the FDA announced that validated in-house tests could be used for diagnosis immediately, as long as the agency received applications for the emergency use of the tests within 15 business days.

How problematic was the delay?

It’s not yet clear exactly what the delays meant for the trajectory of the COVID-19 outbreak in the U.S. But some experts say the lack of testing in many of the nation’s public health labs hampered early detection. 

The CDC reported the first case of possible community spread on Feb. 26, and the first reported death, in a Seattle suburb, followed three days later. It would later turn out that two COVID-19-related deaths happened as early as Feb. 26. A genetic analysis of some of the first cases in Washington state also suggested that the virus had been circulating undetected since mid-January. The state did not begin its own testing until Feb. 28.

Scott Lindquist, an epidemiologist with the Washington Department of Health, told the Seattle Times that problems with the CDC’s test kit “really did set us back, there’s no denying that.”

In a CDC telebriefing announcing the first U.S. death, Jeffrey Duchin, a University of Washington epidemiologist and health officer for Public Health Seattle and King County, noted that the patient who died had been tested locally, but only because the lab had just received kits and because of “very recent” changes in CDC criteria for testing. “If we had the ability to test earlier,” he said, “I’m sure we would have identified patients earlier in the community.”

With more available tests and less stringent clinical criteria for testing, the U.S. could have set up a more proactive system for detecting coronavirus cases, although it’s possible the outcome would still be the same. 

The CDC, in fact, wanted to set up a five- or six-site surveillance network in cities such as Los Angeles, San Francisco, Seattle, Chicago and New York City, which would begin testing for COVID-19 in people with flu-like symptoms who test negative for influenza. The plan was to then scale up the network to each state. But with the test kit difficulties, that schedule was set back. On Feb. 28, the CDC said that it expected to have the first of six sites up and running by the following week. When we asked on March 10, the agency did not have an immediate update on the program.

How many tests are available?

Throughout last week, officials gave confusing and conflicting answers about how many tests would be available — and when — in various appearances.

Secretary of Health and Human Services Azar said on March 1 that “we now have the ability to test 75,000 people in the field at this moment,” but later said that those tests would be shipped by the end of the week.

Vice President Mike Pence also at times referred to test kits, when he meant the number of people who could be tested, and said on March 4, “We’ll have over a million tests in the field today,” a statement that conflicted with other administration estimates of 1 million tests shipping out by the end of the week.

Part of the confusion has to do with the fact that testing is happening at several different levels — by the CDC, by state and local public health labs, and at hospitals and other facilities — and that the words used to describe the testing aren’t always consistent.

A test “kit” means an entire box of supplies to run tests to diagnose many people. A CDC test kit, for example, is able to test 700 to 800 specimens. Other test kits come with enough reagents for 500 “reactions.” Because controls need to be run, fewer than 500 samples can be tested with such a kit. And in both cases, the actual number of people who could be diagnosed with each will be lower, since patients may have multiple samples or samples may be run in duplicate. The same issue applies to the word “test.”

FDA Commissioner Stephen Hahn provided the clearest answer about testing capacity on March 7, when he said that the CDC had shipped enough kits to state and local public health labs to test 75,000 people for COVID-19. 

In addition, more than 1.1 million tests manufactured by Integrated DNA Technologies had been shipped as of March 6 for non-public health labs to use, with another 400,000 possibly shipping by March 9. A separate manufacturer had also produced 640,000 tests that could have shipped as early as March 9, once quality control testing was complete.

“The actual number of tests that have shipped is larger than the number of patients that can be tested,” Hahn explained, noting that the approximately 2.1 million tests from IDT and another company would allow roughly 850,000 Americans to be tested.

On top of those tests, he added, would be any testing by commercial or academic labs, plus an expected additional 4 million tests from IDT and the other company by the end of this week. The CDC itself also can test approximately 350 specimens per day.

In a March 9 telebriefing, the CDC’s Nancy Messonnier clarified that the 75,000 people that can be tested using the CDC’s test kit in public health labs is a cumulative figure. She also said that in the coming weeks, as more manufacturers join the effort, the “majority” of testing will be from the commercial sector.

Several commercial companies have stepped forward to increase production. Quest Diagnostics said it could begin testing on March 9, and LabCorp said its test was available for ordering on March 5, with results taking three to four days once a specimen is received in the mail. 

Can “anybody that wants a test” get one, as the president said?

Not quite. On a March 6 visit to the CDC, Trump said, “Anybody that wants a test can get a test,” adding, “Anybody right now and yesterday — anybody that needs a test gets a test. We — they’re there. They have the tests. And the tests are beautiful.”

When we asked the White House about this, the vice president’s office told us that tests were available through state labs, but that doctors needed to contact the state lab to get one. In other words, if someone wants a test, and a physician does not agree, a person cannot be tested.

This point was also made by Anthony Fauci, the director of the director of the National Institute of Allergy and Infectious Diseases, on “Fox News Sunday,” when he said tests “are out there now,” but “it’s up to the doctor to order the test.”

It’s worth mentioning that even if a test has been shipped, it hasn’t necessarily been received or is ready to use right away; tests also take some time to complete, and so results may not be available immediately.

The CDC maintains that no person was ever denied a coronavirus test, nor was there ever a backlog at the agency. “Our guidance for who can be tested has always allowed room for clinical discretion,” said CDC press officer Richard Quartarone in an email. “That is, if a doctor had a patient who they wanted tested, CDC has not refused any of these requests.” The only exception to that, Quartarone told us, might be a damaged sample.

Some groups, however, dispute the idea that tests are fully available or that there has never been a shortage.

“We are now in the phase of the response where demand for testing is greater than the tests available, even when the private/commercial sector is doing testing,” Eric Blank, the chief program officer of the Association of Public Health Laboratories, told us in a March 9 email. “So there is a test shortage but it is probably not as great as it was a week or so ago.”