We live in an era defined by data — and by promises about the power of data. For a while now, we’ve been persuaded that all that data was making life better, simpler, and more predictable.
Until this spring, that seemed true enough. Anyone could see the ways in which the world was becoming more quantifiable, and thus more certain. Your Uber driver is three minutes away. It’s going to start raining on your block in about 14 minutes. Your package is on the truck and will arrive before 6 p.m. It seemed obvious that these numbers reflected something real and true about what will soon happen to you.
Now, the national experience of COVID-19 floats on a sea of numbers. Newscasts and headlines begin with updates on cases and deaths and numbers of tests. Dozens of daily trackers and models offer up every kind of statistic, from a rating of each county’s preparedness level to the number of available intensive care beds nationwide.
It’s the quantified epidemic: more numbers about more variables than anyone can absorb. Meanwhile, the more numbers you read, the less certain you feel. The most important data points, like how many people have actually been infected by the virus, are hard to pin down. The stats aren’t being hidden or distorted. The data just aren’t there.
Meanwhile, the swirl of models and projections, the curves that skyrocket or flatten, offer big-picture predictions that seem to hold all the answers, yet they don’t tell any one individual anything meaningful about his or her own life. All those charts and spreadsheets cannot answer that most urgent and terrible question: What is going to happen to us?
TURNING TO NUMBERS in a crisis is human nature, says medical anthropologist Monica Schoch-Spana, senior scholar at the Johns Hopkins Center for Health Security, who has been involved in many national working groups on disaster preparedness. “There’s an uncertainty in an epidemic,” she says. “People’s worlds are turned upside down, and they start engaging in a process of sense-making.”
Years ago, as part of a project to anticipate potential consequences of a deliberate bioterrorism attack, Schoch-Spana began studying the 1918 flu pandemic. She read the entire output of two Baltimore newspapers between March 1918 and December 1919.
The Baltimore Sun, she found, relentlessly reported daily tallies of cases and deaths that afflicted the city, constantly examining the meaning of these numbers. During the worst of it, the city’s health department could no longer keep up with the deaths and was not generating reliable statistics, so the paper’s reporters calculated their own tally by examining forms that undertakers had to fill out.
Today, the most crucial numbers in this pandemic are also surprisingly difficult to establish. The official tallies of COVID-19 cases include only people who have tested positive. But there haven’t been nearly enough tests available, and some sick people wait so long for a test that the virus is no longer detectable by the swab method, so they test negative. Still others — and this may be a huge group — never feel sick enough to ask for a test. The upshot is that many people who probably were infected are not included in official totals.
Various lines of evidence point in that same direction. In New York City, for example, about 146,000 people had confirmed cases of COVID-19 as of Friday. But when two New York hospitals tested every pregnant woman who arrived to deliver a baby over two weeks, they found 33 of the 215 women had the virus; 29 of those 33 women had no symptoms at the time. If the 215 women in the sample are typical, more than 1 million people have been infected with this coronavirus in the city alone.
Even the death count is not straightforward. In general, that number also reflects only people who have tested positive for the virus. But hospital physicians across the country say that their emergency rooms are strangely empty of the patients who normally show up with heart attacks and strokes and serious wounds. The doctors worry that many people, afraid of catching the virus, are trying to ride out their illness at home, and die there instead.
Indeed, emergency responders in New York City report that hundreds of New Yorkers have died at home every day since late March. That’s roughly eight times more than usual. More deaths than usual also have been documented in Massachusetts and Detroit.
Some of those people probably died from the virus at home or in a nursing home, without getting a test. Others might have been killed by the disease indirectly, because they were scared to go to the hospital and died of something else — or they lost their job and health insurance, and feared they couldn’t pay for a hospital visit.
In mid-April, New York City updated its death count to include people whose deaths were presumed (but not proved) to be the result of COVID-19. With a lot more detective work, a full count of cases and deaths eventually can be established. For now, though, the numbers are a work in progress — the best available information in the midst of a crisis.
BUT THERE’S A second type of uncertainty we must face that will never be resolved. It is caused by the fluid nature of the pandemic itself, and the human urge to interpret public health models as private health guidance.
Every day brings new updates from the epidemiological models created by researchers at places like Imperial College London or the Institute for Health Metrics and Evaluation (IHME) at the University of Washington. The models are created to guide policy makers and government officials, helping them to anticipate how many hospital beds might be needed in the next month, for instance, or to decide when to impose social distancing rules. Models cannot capture the individual reality of the pandemic. But we look for ourselves in them, hoping for some hints about our own fate. “People are attributing a power to models, that they are one-to-one representations of the real world, and they’re not,” says medical anthropologist Schoch-Spana. “They’re meant to be a decision-making aid, not a photograph of reality.”
At the end of March, the IHME model, which is often cited by the White House, forecast that COVID-19 would kill more than 93,000 Americans by August. But as data continued to flow into the model — specifically, new information about how strictly people were observing social distancing — the projection dropped in early April to 60,414. It’s not that the initial projection was wrong and the second one right. The data — such as they were — changed. Many epidemiologists think the the virus will bounce back in the fall, so that number is still not a final answer — just a provisional picture of the damage the virus might do by mid-summer, subject to revision.
Unlike a one-and-done flood or tornado, a pandemic is an unfolding catastrophe, changing every day. “In natural disasters, yesterday was your bad day,” says Schoch-Spana. “It’s not that way in pandemics. It’s a biological and organic process.”
Adding another layer of uncertainty, the predictions can actually change our collective behavior, says Jonathan Fuller, a doctor and philosopher of medicine at the University of Pittsburgh. “People’s perception of risk influences their behavior, and that’s a rational response,” says Fuller. “Higher death projections could lead people to take things more seriously, which might lead people to change their behavior, which feeds back into the model.” It forms a feedback loop with unforeseeable effects.
The virus probably will surge and retreat and surge again in the coming months. Lockdown orders will likely give way to more dynamic and local restrictions. In different locales, parts of life may start up again, then need to be shut down, repeatedly, through summer and fall. What’s called “the pandemic” will feel more like a whole variety of local epidemics, each with its own peaks and valleys, says epidemiologist Caitlin Rivers, also at the Johns Hopkins Center for Health Security. It’ll be confusing.
So in the short term, what comes next isn’t going to get any easier to predict. Not having a normal will become the new normal. Don’t believe anybody who tells you exactly what’s going to happen, says Rivers. There are too many possible futures. We need to just roll with it. The lovely fantasy that data were going to tame the future must now be put aside.
In the face of so much illness and death, the idea of living with that level of uncertainty seems unbearable. But as Schoch-Spana’s 1918 research shows, we’ve done it before. The task before us now is to learn to do it again.
Kat McGowan is a journalist in California who covers health, medicine, and science. Follow her on Twitter: @mcgowankat.