A Project of The Annenberg Public Policy Center

How Many COVID-19 Tests Are ‘Needed’ to Reopen?


The Trump administration has repeatedly claimed that there are enough COVID-19 tests for states to begin reopening their economies. While that may be true for select locations, experts say more tests are needed, even if they don’t agree on a particular number.

In the past couple of weeks, President Donald Trump and other members of the coronavirus task force have insisted that there are adequate levels of diagnostic testing for the coronavirus to allow states to lift stay-at-home orders and restart some businesses.

“We are continuing to rapidly expand our capacity and confident that we have enough testing to begin reopening,” Trump said in an April 27 briefing, adding, “And the testing is not going to be a problem at all. In fact, it’s going to be one of the great assets that we have.”

In the same briefing, Vice President Mike Pence said, “[W]e have a sufficient amount of testing today for every state that qualifies to enter phase one to begin to reopen their economies.” Phase one refers to the first part of the White House’s plan for a gradual return to quasi-normal life, which allows gyms, restaurants and movie theaters, for example, to open under continued social distancing, but for schools to remain closed and workplaces to encourage telework where possible.

The vice president’s office did not explain which metrics Pence and others were using to make that determination, so it’s hard to fully evaluate his statement. But while some places may have enough tests, multiple scientists told us that as a whole, the U.S. still lags behind what’s needed to execute an effective strategy to rein in outbreaks.

“Testing is still limited,” Michael Mina, an epidemiologist at the Harvard T.H. Chan School of Public Health, told us in response to Pence’s statement. 

“If you’re not including testing as a large part of phase one opening, then sure, we have plenty of tests,” he said. “But I think most of us who are working on thinking about what are the safest and best ways to open up the economy again place pretty significant amounts of testing front and center to prevent outbreaks and monitor for them.”

The question of how many tests it will take is hotly debated, with some groups recommending at least around half a million tests a day, and others suggesting several million or upwards of 20 million or even 35 million a day.

According to the COVID Tracking Project, testing for the coronavirus in the U.S. has steadily improved — to around 264,000 a day, on average over the past week, up from about 217,000 tests per day on average for the week ending on April 26. But the nation is still below any of the proposed benchmarks from experts.

Those figures also mean there aren’t enough tests for every American who wants to return to work to get one, as Trump suggested in a May 5 interview with ABC News. When asked whether “any worker who’s nervous about going back” would have access to both diagnostic tests for the virus and antibody, or serological tests, “right now,” Trump replied, “There should be no problem.”

But Yonatan Grad, a professor of immunology and infectious diseases at Harvard, said the answer “is quite clearly no,” adding, “we don’t yet have the capacity to offer broadly either virologic or serologic testing.”

National Institute of Allergy and Infectious Diseases Director Anthony Fauci told CNN’s Jake Tapper on April 28 that everyone who needs a test, “according to the way we’re approaching the identification, isolation, contact tracing, keeping the country safe and healthy,” should be able to get one by the end of May or the beginning of June.

“That’s what I’m being told by the people who are responsible for the testing,” he added. “I take them for their word.”

We’ll review several of the estimates and explain why testing is so important — but also why it’s not the only consideration for reopening businesses, schools and other places that have been largely closed for a month or more.

Wide Range of Estimates

Several groups have attempted to put a figure on how many tests are needed for the U.S. to reopen. On the lower end, researchers at the Harvard Global Health Institute calculated that the U.S. would need to do approximately 500,000 tests per day, as a bare minimum, by May 1 — a figure they have since revised to 900,000 for May 15.

On the other extreme, there’s Nobel laureate and New York University economist Paul Romer’s plan to test the entire U.S. population once every two weeks, or around 25 million tests a day. He has since raised that to 35 million tests per day, or more, to include daily testing of front-line workers.

In between, Harvard University’s Edmond J. Safra Center for Ethics suggests 2 million tests per day to start, then ramping up to at least 5 million tests per day by early June to get the full workforce back in action — with that growing to as many as 20 million tests per day.

The 5-million-a-day figure has received perhaps the most attention, in part because when asked about the number on April 28, the president claimed the U.S. would hit that goal “very soon.”

That morning, however, Adm. Brett Giroir, who is in charge of the government’s COVID-19 testing efforts, told Time magazine that “there is absolutely no way on Earth, on this planet or any other planet, that we can do 20 million tests a day, or even five million tests a day.” 

A day later, Trump insisted that he had never made such a claim. “I think we will,” he said, referring to hitting the 5-million-per-day benchmark, “but I never said it.”

There isn’t a consensus on the number of tests, in part because the groups are making different assumptions and taking different approaches to tackling the pandemic on different timelines. Key to all of the strategies, however, is using testing to identify people who are infected so that they can be isolated and no longer spread the virus to others. 

Many public health experts endorse testing combined with contact tracing, which involves tracking down others with whom the infected person had contact, so that those people can be warned, and ideally be put into quarantine or be tested as well.

“Without ubiquitous testing — testing everyone with symptoms, and those they have come in contact with — we simply cannot be confident that a reported decline in cases represents a true decline in infections – and that it’s safe to open,” explained Ashish Jha, faculty director of Harvard Global Health Institute, or HGHI, and a co-author of the group’s estimates, in a Time editorial. “If a state misjudges its true underlying infection trajectory, it may suffer large flare-ups of the disease, necessitating a long and painful lockdown again.”

The HGHI estimates were calculated in two main ways. In the first, the team followed a strategy of testing everyone with flu-like symptoms and the contacts of those who test positive. Assuming that 1% of cases prove fatal, the group worked backward from a projected number of deaths to estimate the number of new infections in the U.S. on a given day, which for May 15, is about 100,000. To then approximate the number of tests, the group assumed that three-fourths of those people would have symptoms and be tested, and each person would have 10 contacts to trace.

Thomas Tsai, a surgeon and health policy researcher on the HGHI team, told us the team looked through the literature to determine the number of contacts, and ended up choosing a “middle-of-the-road estimate.”

Then the group added in the typical number of influenza-like illness cases for the season, reaching a total of around 900,000 tests per day. A similar calculation, based on trying to get the percentage of positive tests below 10% — in line with recommendations from the World Health Organization — produced a similar figure.

The researchers, however, caution that their approach in every way veers toward an undercount, and therefore should be viewed as a “floor, not a goal.”

“If we don’t ramp up our testing,” said Tsai, “we’re just condemned to repeating the mistakes of both February, March and April.”

The Harvard Safra figures, which were based on modeling efforts described in two white papers, assume an intensive testing, isolation and contact tracing approach.

Danielle Allen, the director of the center and co-author of the group’s April 20 report, told us the numbers were generated by aiming to have each infected person spread the disease to less than one other person (specifically 0.75 people), which would allow the outbreak to eventually fizzle out.

With just essential workers, or about 40% of the workforce in action, the group estimated the U.S. would need 2 million tests per day, rising to 5 million for 100% of the workforce.

“Both of these numbers depend on effective contact tracing,” Allen said. “If contact tracing is moderately effective the amount of testing needed rises to 20 million a day.”

Allen, however, said that with subsequent modeling, the group now thinks 5 million tests per day may be sufficient, and is planning to revise the numbers in a forthcoming report.

Two other groups, also adopting a contact tracing strategy, offer other testing benchmarks. The Rockefeller Foundation suggests an initial 3 million tests per week, or about 430,000 tests per day, by mid-June, then growing that number 10-fold over another six months.

Resolve to Save Lives, a public health initiative headed by former Centers for Disease Control and Prevention Director Tom Frieden, tallied up the number of highest-priority people who need tests, and came up with a lower bound number of 350,000 to 700,000 tests per day. This includes high-risk patients with COVID-19 symptoms and any sick hospital workers, public safety officers, prisoners, or nursing home residents and their symptomatic contacts.

“The minimum number of tests that need to be done per day,” the group’s report notes, “will depend on the stage of the epidemic and number of outbreaks, hospitalization rates, number of cases and contacts identified, and decisions about the frequency and extent of testing in nursing homes, hospitals, and essential services.”

If contact tracing is not part of the equation, even more tests are needed because the approach becomes an untargeted national surveillance method. Because diagnostic testing only reveals whether someone has detectable virus in their body at one point in time, tests must be given to pretty much everyone repeatedly to ensure those who are infected are aware and can be isolated. 

That’s how one arrives at the even higher figures, such as Romer’s 25-million-per-day goal. Instead of prioritizing people who are sick, as is being done now — and is recommended in the vast majority of proposals — Romer’s plan calls for tests to go first to people without symptoms to pick out those who are spreading the virus unwittingly.

State-Specific Thresholds

While most of the test estimates are for the entire country, the reality is that for any targeted testing strategy — which is the only strategy that’s currently possible — locations with relatively few infections won’t need as much testing as places with huge outbreaks.

“You can’t just take the national number and scale it to states by their population,” HGHI’s Jha told the health news website STAT, in a story sharing an updated state-by-state breakdown of the group’s original estimate. “You have to base it on the size of the outbreak in a state.”

The latest figures from HGHI show that just nine states, the majority of which are clustered in the sparsely-populated West, are close to or have surpassed their testing targets. The analysis also reveals that even states that have been performing a large number of tests per capita, such as Rhode Island and Massachusetts, are still short of their estimated minimums. Conversely, some states that have relatively low per-capita testing rates, such as Hawaii and Alaska, are exceeding their targets.

Because the estimates were made using a model that assumed states would stay locked down until May 15, the researchers caution that states could very well need more tests than their calculations suggest if social distancing measures are loosened.

“The moment you relax, the number of cases will start climbing,” Jha noted in the group’s write-up of the project. “And therefore, the number of tests you need to keep your society, your state from having large outbreaks will also start climbing.”

None of the other estimates provide state-level figures, although Allen said the Harvard Safra group plans to release another report breaking down its national numbers “to make visible what is necessary in high prevalence, low prevalence, and no prevalence locales.”

Test Positivity

Short of a specific test threshold, one metric experts say can get at whether a country or state is doing something close to enough testing is the test positivity rate.

Caitlin Rivers, an epidemiologist at Johns Hopkins University, told us that it’s “very difficult to say what the ‘right’ number” of tests is, but because a high fraction of tests are still positive, “it is clear we need more capacity.”

As a Johns Hopkins website explains, if a large proportion of tests are positive, that suggests a community is only able to test the sickest patients, and is likely missing people with milder infections who can still spread the disease.

The WHO has said that countries doing extensive testing see around 3-12% of tests come back positive, and has recommended that nations aim for or a test positivity of roughly 10% to make sure there’s enough testing happening to detect all of the cases.

As of May 8, about 15% of all tests in the U.S. have been positive, according to data from the COVID Tracking Project. In the last few days, the current test positivity rate has begun to dip below 10%, although many states have a significantly higher percentage of positive tests.

“I prefer positivity to numbers of tests or tests per capita because positivity is tied to the burden of disease in a location,” said Jennifer Nuzzo, a senior scholar at the Johns Hopkins Center for Health Security, in an email.

Nuzzo noted, for example, that the U.S. has conducted around 10 times as many tests per capita as Taiwan, but the American epidemic is some 200 times bigger.

And she has concerns about many states that are thinking of or have already reopened. “[W]e’ve found that a number of these [states] aren’t doing enough testing and have too high positivities,” she said. “This makes me worried that they are missing the bottom of the iceberg in terms of infections and not seeing the full picture of illness in their communities.”

Jeffrey Shaman, an epidemiologist at Columbia University’s Mailman School of Public Health, agreed that per-capita test numbers are “not as informative as looking at testing per positive infection identified.” 

The U.S., he said, has done only about six tests per confirmed case, whereas South Korea has done 60. “They’re testing 10 times as many per infection found,” he said in an interview.

This is especially relevant because Trump has frequently boasted of the total number of tests the U.S. has performed, and has suggested that the U.S. is outpacing South Korea and other nations. 

Speaking in Arizona on May 5, Trump said, “We’re over 7 million tests. Germany is at two and a half. Italy is less than that. Japan is down here and South Korea, which we talk about, and again, I’m very friendly with South Korea, and with the president of South Korea, and he calls to congratulate us on our great testing. South Korea is over here.”

Trump is correct that the U.S. has done more total testing, but all of the other countries he mentioned have still performed more tests per positive case, meaning they’ve been able to do more testing relative to the size of their outbreaks.

The next day, White House Press Secretary Kayleigh McEnany again referenced South Korea. “We always hear about South Korea and their tests, there are 11 tests per thousand,” she said. “Here in the United States, that’s 17 tests per thousand.”

McEnany’s statistics are slightly off — for May 4, Our World in Data shows South Korea with 12 tests per thousand people and the U.S. with 22. 

But again, the comparison is misleading because it ignores the fact, as we have previously written, that South Korea rapidly expanded testing early on in the outbreak. Ultimately, the country was able to successfully halt its outbreak — something the U.S. has failed to do.

On May 8, the country posted just 12 new cases, while the U.S. is still seeing around 25,000 new infections per day, or approximately 325 times more cases than South Korea when adjusted for the population. It took months — until April 17 — for the U.S. to catch up to South Korea’s per-capita testing level, and the U.S. has yet to match the nation’s less than 2% test positivity rate. South Korea doesn’t need to compete with the U.S. on per-capita tests because the country is already doing a massive amount of testing compared to its caseload.

Contact Tracing

While experts say increased testing is a requirement for reopening, they also emphasize that by itself, it’s not enough.

“What’s more important now is not so much the testing number,” said HGHI’s Tsai. “It’s really the testing strategy and transparency around the goalposts as states come up with their specific plans for reopening.”

A key question is whether states will be able to snuff out new chains of viral transmission through contact tracing, as many experts advise. While a staple of public health, contact tracing is a resource-intensive process, and involves not just piecing together an infected person’s contacts, but also providing support to those who are quarantined or isolated.

“They may need food. They may need medicines,” said Emily Gurley, an epidemiologist at Johns Hopkins, on a call with journalists. “They may need help with simple things like doing their laundry.”

Gurley contributed to a report, written in collaboration with the Association of State and Territorial Health Officials, that estimated the U.S. would need to add around 100,000 contact tracers to combat the spread of the coronavirus.

While Shaman thinks contact tracing is absolutely critical, he noted that South Korea was exceptionally well set-up to make the technique more effective.

Following an outbreak of MERS in 2015, he said, the country passed laws that allowed the government to monitor people’s credit cards and phones — and also to arrest anyone refusing to participate in a quarantine.

Those levers aren’t available in the U.S., and privacy concerns may keep at bay proposals to deploy more scalable technology-based contact tracing tools.

Other scientists are pessimistic that contact tracing will work in the U.S., especially if case numbers remain high.

“It’s clear that you can do a lot of control if you do contact tracing really well. Singapore managed with mainly contact tracing for several months,” said Marc Lipsitch, an epidemiologist at Harvard University, in a call with reporters. “But then eventually even Singapore lost its control of the epidemic and had to resort to social distancing types of measures.”

“In conjunction with really aggressive measures to get case numbers down — and significant resources — contact tracing can be a useful piece of the control approach,” he added. “But I think it’s very challenging with an ongoing epidemic.”

According to a New York Times analysis, 21 states are planning to or have already begun to reopen, even though they have not cleared all of the criteria set forth by the White House guidelines. Most of the states do not meet the gating criteria for COVID-19 cases, which can be met by showing either a “downward trajectory of documented cases within a 14-day period” or a “downward trajectory of positive tests as a percent of total tests within a 14-day period.” 

Opening before states have solid testing infrastructures and strategies that they know are working, Tsai said, would be premature. “It’s basically just turning back the hands of time to February,” he said.

FactCheck.org does not accept advertising. We rely on grants and individual donations from people like you. Please consider a donation. Credit card donations may be made through our “Donate” page. If you prefer to give by check, send to: FactCheck.org, Annenberg Public Policy Center, 202 S. 36th St., Philadelphia, PA 19104.