The Horror of the Coronavirus Data Lag

Some reopening states are already claiming victory over the coronavirus. But the real consequences won’t be clear for weeks.

The boardwalk at Ocean City, Maryland, crowded with people
Eric Thayer / Getty

The most universal experience of the coronavirus pandemic in America might not be a sense of fear or anxiety, but a profound confusion over what exactly is going on. Novel pathogens are confounding by definition, and since the first COVID-19 case was confirmed in the United States, in January, information about severity, spread, and a seemingly ever-expanding list of symptoms has trickled slowly and inconsistently out of emergency rooms and local health departments.

Conditions are improving in some of the country’s major cities, but outbreaks continue to grow in others, as well as in prisons and rural areas—especially those home to large meatpacking plants. Every state tests at a different rate, makes those tests available to different types of people, and counts its results differently in official statistics. As some states have begun to partially lift shelter-in-place orders and allow businesses such as restaurants and hair salons to reopen, one particularly high-stakes point of confusion has emerged: When can you tell if a state’s reopening guidelines are keeping infection numbers down, and how long do you have to wait before you feel sure?

Humans tend to think of illness as a binary. You are sick or you aren’t, which feels simple and knowable. In reality, most types of communicable disease are far less black and white: People are infected before they become aware that they’re sick. For much of the outbreak in the United States, local officials have asked people who suspect that they have COVID-19 but aren’t having serious trouble breathing to avoid seeking medical attention. That means that by the time most infections become official cases, people have spent days or weeks trying to heal on their own before seeking care. The periods of information lag with COVID-19 are longer than those of the colds and flus to which it’s often compared, which has been a central problem in containing the virus since the beginning. In order to make predictions and policy from any particular day’s case numbers, health experts have to synthesize information about testing rates, positive rates, local guidelines, and anything known about how people are responding to instructions to stay home or get back to business. And in order to make everyday decisions and assess their own risk, people have to live simultaneously in the past, present, and future.

Asking the general public to think in such broad and uncertain timescales is a tall order. “The concept of an incubation period or the onset of severe disease—those aren’t things that the public is well versed in, nor should we expect them to be,” says Tom Hipper, who manages the Center for Public Health Readiness and Communication at Drexel University. As a result, keeping the slow progression of a disease in mind can be difficult for regular people trying to make sense of official statistics. “As humans, I think we like instant gratification and we like instant feedback on things,” Hipper told me. The lull between new behavior and its measurable results “can make it a little more difficult to see the connection.”

The snail’s pace at which COVID-19 infections seem to run their course makes public-health communications about the state of the current outbreak unusually complicated. The available evidence suggests that it usually takes about five days for an infected person to go from transmission to symptoms, but it can take as long as 14 days, and the person infected will be contagious for much of that time. The flu, by comparison, goes from transmission to symptoms in an average of two days and a maximum of four, according to the Centers for Disease Control and Prevention, and most people are infectious for only about a day before becoming ill.

Public-health experts can’t change the pace at which COVID-19 moves once a person is infected, but information doesn’t have to be delayed as much as it currently is in the U.S. The speed at which cases of the disease become known to the medical system, government monitors, and the general public depends on how hard a state or country is trying to find them. So far, the best-case scenario seems to be what’s happened in South Korea. In January and February, as the situation in Wuhan, China, deteriorated, South Korea quickly began identifying and isolating infectious travelers from the country, seeking the contacts of known cases, and testing those people before they became symptomatic. This program of testing, tracing, and swift isolation meant South Korea had something closer to real-time information about how the disease was spreading within its borders, and it was able to control the outbreak quickly. Such proactive approaches allowed South Korea, Germany, and Hong Kong to relax some of their restrictions on business and travel, and to quickly identify any new outbreaks that resulted. But even with those measures, all have experienced an uptick in new cases after reopening.

Although no particular outcome is guaranteed, public-health experts believe that something similar will happen in the United States in the weeks to come, as states begin to reopen. But here, we won’t have the benefit of such up-to-date information to judge the country’s progress. Andy Slavitt, a health-care official in the Obama administration, theorized at the beginning of May that the U.S. won’t see the cumulative effect of any reopening spikes until June, because of the weeks it takes for one case to go from transmission to death. Crystal Watson, a professor and risk-assessment expert at Johns Hopkins University’s Center for Health Security, recently told the Associated Press that she expects the lag in the United States to be even longer: five to six weeks from when businesses reopen.

Those assessments have not stopped proponents of reopening from declaring victory in states that have already eased restrictions on businesses, such as Georgia, Texas, and Colorado. Two weeks after reopening, none of them has experienced dramatic spikes. Georgia, whose governor received the most intense blowback—including from me—seems to have kept its rate of transmission largely stable in the earliest days after lifting lockdown, according to estimates commissioned by The Atlanta Journal-Constitution, with about one new transmission resulting from every newly discovered infection. But because the state backdates many of its cases to when patients first had symptoms, it takes two weeks for case counts for any particular day to be completed. Public-health experts are largely in agreement that any changes observed right now, in Georgia or elsewhere, are the result of behavioral patterns from weeks before states began to reopen. In other words, any positive effects are the result of older restrictions, not recent leniency.

The gap between how quickly some countries find new spikes and how long experts believe it will take in the United States can be attributed to a fundamental difference in case-finding strategies. Instead of proactive testing and contact tracing, the U.S. largely relies on identifying cases only when people become ill enough to seek treatment, which can take weeks after exposure. The high cost of medical care in the U.S. can extend the delay; in New York City, reports indicate that many people in the hardest-hit neighborhoods waited until they were near death to seek treatment, at least partly out of fear of the expense. People white-knuckling it through severe cases of COVID-19 at home doesn’t just harm their health, but also means that the country doesn’t know exactly how the disease is spreading or who it’s affecting until significantly later than countries that have more comprehensive test-and-trace systems or make medical care easier to access.

The results of these differences are cumulative, and the delays compound one another. “If you do good things now, you see good results three to six weeks from now,” says Tara Kirk Sell, a professor at the Johns Hopkins Center for Health Security. “If you don’t do the right things now, then it takes a little while to see those opposite results.”

Even in states without particularly ambitious reopening plans, evidence shows that tests and cases are being counted in ways that can mislead casual observers trying to understand how an outbreak is evolving. Virginia, for example, briefly combined its statistics for tests of active infections and antibody tests, which indicate that a person might have been infected at any point in the past. This makes tying daily testing numbers to particular restrictions or behaviors useless.

The lag in COVID-19’s progression isn’t particularly long compared with the lag of chronic illnesses such as emphysema and heart disease, for which public-health experts ask people to change their behavior to reap a payoff decades in the future. But because COVID-19 is a crisis-inciting communicable disease, the lag causes more problems. Society doesn’t have to shut down to mitigate the long-term impacts of smoking or eating too many fried foods, and late-in-life chronic health conditions don’t make attending a single birthday party or funeral an existential threat. When people are asked to change their whole life in onerous ways for an eventual benefit, as is the case with the coronavirus, they can get restless waiting for information about the results. “This long delay leads to a bit of an information void,” Kirk Sell told me. “We’re not able to tie facts together as efficiently, and there’s an opportunity for misinformation to grow.”

The effect of this misinformation can be magnified by the decentralized approach that the United States has taken to its pandemic response. When officials in different parts and levels of government publicly spar and contradict one another, it violates what Hipper calls the cardinal rule of good risk communication. “Mixed messaging from sources who are supposed to be on the same side is one of the best ways to get people to lose trust and question recommendations,” he said.

For now, daily cases are roughly stable in most states—though whether that’s a sign of fewer infections or a side effect of testing rates remains to be seen. More information is now emerging about what might be the most important factor in avoiding an immediate second spike in cases if the country returns to some degree of normalcy: how people in each state have been behaving in the past two months, including the weeks after lockdowns began to ease in some areas.

The available data paint a picture of state leadership whose wishes are secondary to what their constituents believe about acceptable risk. According to an analysis from the data company Opportunity Insights, people and businesses in states with late lockdowns began shutting down, staying at home, and taking the precautions recommended elsewhere, such as closing nonessential businesses and limiting visits to restaurants and bars, far before their own state leadership required it. Since lockdowns were lifted in some areas, the company’s data indicate that people have been extremely slow to return to work, dining out, and shopping, in spite of anecdotal reports of the occasional packed restaurant or store. In Georgia, in-restaurant dining is still down 92 percent from pre-pandemic levels, according to data from OpenTable. Reopening itself is far less dangerous for a populace when most people decline to play along.

Kirk Sell calls signs that individuals have continued taking their own precautions in reopened states “very encouraging.” Those behaviors indicate “people having an understanding of their own risks and the situations that lead to increased risks,” she said, which is no easy feat in a situation as confusing as America’s present reality.

But doing that math, and living in three different time frames, for every trip to the grocery store and every afternoon in the park or at the beach becomes exhausting, even for people who believe that their vigilance is a matter of life and death. When staying at home, wearing masks, and avoiding prolonged contact with others starts to work and the immediate danger appears to recede, people will naturally want to stop abiding by them. “The more cases go down, and the longer this goes on, the more it opens people up to the thought that perhaps this isn’t as severe as we once thought,” Hipper said.

To ensure that the country stays relatively united, public-health officials will need to use accumulating information about spread and risk to help people decide which activities they can add back into their life. “We know the age ranges and the risk factors. We know that it requires close contact—having a close, face-to-face conversation or being in the same household with someone,” Kirk Sell explained. “You need to be around people for quite some time, and a short interaction is much lower risk.” That decision making, she said, is what the next phase of the country’s communication should be about. The best way to avoid an information vacuum is to fill it with things people can actually use.