The Black Hole in America’s COVID-19 Data

States know very little about the results from rapid coronavirus tests.

an illustration of coronavirus tests in the shape of a question mark.
Getty / The Atlantic

For months, public-health experts have been calling for a new line of defense against COVID-19. Antigen testing, also known as rapid testing, is cheaper and faster than standard testing, and it can happen at points of care—long-term-care facilities, prisons, schools, and doctors’ offices. The technology has been around since the spring, but the real investment came in August, when the federal government ordered 150 million Abbott BinaxNOW tests, which show the result on a card in just 15 minutes. At the time, the federal government hoped to use the tests to pinpoint outbreaks in long-term-care facilities and protect the elderly, while states looked to deploy them in schools and universities to get students back in the classroom. Faster, more frequent testing in these places could help mitigate the pandemic’s worst effects.

But the process of distributing those tests, and then reporting their results, has been slow and uneven. The results from most of the antigen tests that have been sent to states have been unreported, according to separate data from the COVID Tracking Project at The Atlantic and the U.S. Department of Health and Human Services, leaving gaps in our knowledge of the pandemic. In some cases, the tests are not getting used at all. Almost a year into this crisis, the U.S. still does not have a clear picture of how many people are being tested for COVID-19.

The value of antigen tests is based on a simple tradeoff. They’re not as sensitive as polymerase chain reaction (PCR) tests, which are the gold standard for diagnosing COVID-19, but they’re most effective when people with the virus are most infectious. And since they’re cheaper, faster, and easier to perform, the ability to do more tests can balance out the lower sensitivity. If their convenience means you take eight antigen tests a month, versus, say, two PCR tests a month, the antigen regime might catch cases that the PCR regime misses. One recent analysis suggests that using antigen tests on a population three days a week would be virtually as effective in catching infections as using the superior PCR test the same number of times.

But for an effective antigen-testing plan to work, the tests have to be given frequently and the results quickly reported. While PCR tests are done at labs, which are practiced in reporting results to health agencies, antigen tests are intended for places such as schools and nursing homes, which have to develop their own reporting systems. And that’s been a problem since the tests started going out in May. By August, the test manufacturers Quidel and BD combined to produce 3 million antigen tests a week. But by mid-September, states had reported only 215,000 antigen-test results. There was, as The Atlantic’s Alexis Madrigal and Robinson Meyer wrote then, “a hole where data about antigen testing should be.” At the time, only six states made antigen-testing numbers available (now 20 states do), and extrapolating those results suggested that only 1.4 million antigen tests had been conducted nationwide.

The federal government pressed on with antigen testing, though, and ordered 150 million tests from Abbott when the company’s more advanced technology came online. In October, the Associated Press reported that while about half of the country’s testing capacity consisted of antigen tests, the overall reported test numbers didn’t reflect the flood of antigen tests that had made their way to states. In November, The New York Times found the same hole in the data, reporting that state and local public-health officials were on the hunt for rapid-testing facilities that were not disclosing their data.

And that hole doesn’t seem to have gone anywhere. On December 14, during a biweekly CDC call with laboratory representatives to address their issues in dealing with COVID-19, the Department of Health and Human Services said that a majority of the more than 50 million antigen tests that have been distributed to states “do not have test results flowing to HHS.” According to the CDC data, that number is now up to 94.4 million. Data collected by the COVID Tracking Project similarly suggest that states are not reporting antigen-test results. Out of more than 235 million test results reported since June, only about 9 million are clearly reported as antigen-test results. Admittedly, many states still don’t split out separate figures for antigen tests and PCR tests. But even among the states that do, antigen tests make up only about 10 percent of all tests they have reported. Extrapolated nationwide, that would be about 27 million tests, or a bit more than a quarter of the tests the federal government has distributed to date.

Collecting antigen-test results has been a struggle, says Richard Danila, the deputy state epidemiologist in Minnesota, which does report antigen tests separately. The state needed four people working full time for three weeks to set up a reporting system with its 300 nursing homes after they received the antigen tests—not just getting results from frequent testing, but ensuring that those results are always connected to the patient’s name and demographic information. In the best-case scenario, the department helped them create flat electronic files like Excel sheets, but some of the data came in as faxes or emails. “We certainly don’t have, probably, full reporting of the Abbott BinaxNOW positive tests,” Danila says. “You’re relying on the good nature of people that remember to report to us.” All in all, Minnesota has reported nearly 273,000 antigen tests during the pandemic, compared with nearly 5.7 million PCR tests. (Some of this gap, Danila says, is due to Minnesota’s investment in a saliva-based PCR test, making the state less dependent on antigen tests.)

If tests are being given but not reported, that doesn’t mean they’re completely useless; people and individual institutions can obviously take action based on the results. But it means that crucial information isn’t available to health officials. “This lack of visibility can lead to a lot of challenges, especially as the administrators are trying to conduct contact tracing or decide where to distribute additional testing supplies, and identify hot spots,” Kristen Honey, a senior adviser to the assistant secretary of HHS, said on the CDC call.

Perhaps more worryingly, officials on the same call said that in some cases, the antigen tests the government ordered are not being used at all. Marcus Plescia of the Association of State and Territorial Health Officials described a nationwide survey his organization had conducted in November about the use of Abbott’s BinaxNOW rapid tests. One of the highest priorities for these tests was using them in schools, to catch infections early and help keep facilities open. But “use in K–12 has not been substantial so far,” Plescia said. An antigen-testing program’s logistics can be daunting, and as an article in The New England Journal of Medicine put it, schools have received “strikingly little substantive guidance on testing” from federal and state officials. When ASTHO did its November survey, 60 percent of respondents were still in the planning stage. That may be changing: Massachusetts, for example, had been using rapid antigen testing for symptomatic students and staff in 134 of its 403 school districts since November, but is now rolling out pooled testing across the state for all students and staff, regardless of symptoms, which will greatly boost the number of tests being used. Other school systems, such as Chicago Public Schools and New Orleans Public Schools, had instituted remote learning, but are going back to in-person education with antigen-testing plans.

New technologies may ease and speed antigen-test reporting, filling in the data hole that’s been lingering for months. At-home tests that don’t require a prescription are going on sale this month, and the kit includes a device that transmits the results via Bluetooth to an app on the user’s phone. And the CDC recently launched a new interface, SimpleReport, to make it easier for institutions such as K–12 schools and universities to submit testing results. But these systems still require people and institutions to take the tests and choose to report their results. Better tests alone won’t solve that problem.