Note: The graphic in this story is no longer being updated. For more recent data, visit our new post on this topic.
To safely phase out social distancing measures, the U.S. needs more diagnostic testing for the coronavirus, experts say. But how much more?
The Trump administration said on April 27 that the U.S. will soon have enough capacity to conduct double the current amount of testing for active infections. The country has done nearly 248,000 tests daily on average in the past seven days, according to the nonprofit COVID Tracking Project. Doubling that would mean doing about 496,000 a day.
Will that be enough? What benchmark should states try to hit?
One prominent research group, Harvard's Global Health Institute, proposes that the U.S. should be doing more than 900,000 tests per day as a country. This projection, released Thursday, is a big jump from its earlier projection of testing need, which had been between 500,000 and 600,000 daily.
Harvard's testing estimate increased, says Ashish Jha, director of the Global Health Institute, because the latest modeling shows that the outbreak in the United States is worse than projected earlier.
"Just in the last few weeks, all of the models have converged on many more people getting infected and many more people [dying]," he says.
But each state's specific need for testing varies depending on the size of its outbreak, explains Jha. The bigger the outbreak, the more testing is needed.
On Thursday, Jha's group at Harvard published a simulation that estimates the amount of testing needed in each state by May 15. In the graphic below, we compare these estimates with the average numbers of daily tests states are currently doing.
Two ways to assess whether testing is adequate
To make their state-by-state estimates, the Harvard Global Health Institute group started from a model of future case counts. It calculated how much testing would be needed for a state to test all infected people and any close contacts they may have exposed to the virus. (The simulation estimates testing 10 contacts on average.)
"Testing is outbreak control 101, because what testing lets you do is figure out who's infected and who's not," Jha says. "And that lets you separate out the infected people from the noninfected people and bring the disease under control."
This approach is how communities can prevent outbreaks from flaring up. First, test all symptomatic people, then reach out to their close contacts and test them, and finally ask those who are infected or exposed to isolate themselves.
Our chart also shows another testing benchmark for each state: the ratio of tests conducted that come back positive. Communities that see about 10% or fewer positives among their test results are probably testing enough, the World Health Organization advises. If the rate is higher, they're likely missing a lot of active infections.
What is apparent from the data we present below is that many states are far from both the Harvard estimates and the 10% positive benchmark.
Just nine states are near or have exceeded the testing minimums estimated by Harvard; they are mostly larger, less populous states: Alaska, Hawaii, Montana, North Dakota, Oregon, Tennessee, Utah, West Virginia and Wyoming.
Several states with large outbreaks — New York, Massachusetts and Connecticut, among others — are very far from the minimum testing target. Some states that are already relaxing their social distancing restrictions, such as Georgia, Texas and Colorado, are far from the target too.
Jha offers several caveats about his group's estimates.
Estimates are directional, not literal
Researchers at the Global Health Initiative at Harvard considered three different models of the U.S. coronavirus outbreak as a starting point for their testing estimates. They found that while there was significant variation in the projections of outbreak sizes, all of the models tend to point in the same direction, i.e., if one model showed that a state needed significantly more testing, the others generally did too.
The model they used to create these estimates is the Youyang Gu COVID-19 Forecasts, which they say has tracked closely with what's actually happened on the ground. Still, the researchers caution, these numbers are not meant to be taken literally but as a guide.
If social distancing is relaxed, testing needs may grow
The Harvard testing estimates are built on a model that assumes that states continue social distancing through May 15. And about half of states have already started lifting some of those.
Jha says that without the right measures in place to contain spread, easing up could quickly lead to new cases.
"The moment you relax, the number of cases will start climbing. And therefore, the number of tests you need to keep your society, your state from having large outbreaks will also start climbing," warns Jha.
Testing alone is not enough
A community can't base the decision that it's safe to open up on testing data alone. States should also see a consistent decline in the number of cases, of two weeks at least, according to White House guidance. If their cases are instead increasing, they should assume the number of tests they need will increase too.
And, Jha warns, testing is step one, but it won't contain an outbreak by itself. It needs to be part of "a much broader set of strategies and plans the states need to have in place" when they begin to reopen.
In fact, his group's model is built on the assumption that states are doing contact tracing and have plans to support isolation for infected or exposed people.
"I don't want anybody to just look at the number and say, we meet it and we're good to go," he says. "What this really is, is testing capacity in the context of having a really effective workforce of contact tracers."
The targets are floors, not goals
States that have reached the estimated target should think of that as a starting point.
"We've always built these as the floor, the bare minimum," Jha says. More testing would be even better, allowing states to more rapidly tamp down case surges.
In fact, other experts have proposed that the U.S. do even more testing. Paul Romer, a professor of economics at New York University, proposed in a recent white paper that if the U.S. tested every resident, every two weeks, isolating those who test positive, it could stop the pandemic in its tracks.
Jha warns that without sufficient testing, and the infrastructure in place to trace and isolate contacts, there's a real risk that states — even those with few cases now — will see new large outbreaks. "I think what people have to remember is that the virus isn't gone. The disease isn't gone. And it's going to be with us for a while," he says.
Daniel Wood contributed to this report.