Jul 21, 2020 7:00 AM

To Navigate Risk In a Pandemic, You Need a Color-Coded Chart

Is going to the dentist more dangerous than grocery shopping? Public health groups want to help us weigh everyday risks with to easy-to-read guides.

different colored graphical cutouts layered

Half a year since the emergence of Covid-19 and months since cities and states went into lockdown—and, to varying degrees, came out again—we’re more than familiar with the basics of what we should do to protect ourselves: Stay home if you can, social distance when you can’t, and wear a mask when you have to be less than arms-length from people.

That guidance is simple—too simple, perhaps. It doesn’t account for the complexities of how to live in a pandemic, or answer the questions that come up every day. Is it safe to return to an office? To undergo procedures at the dentist? To sit in a doctor’s waiting room? If you want to exercise, is shooting hoops smarter than tennis? If you need solace, is a library safer than church?

There’s no federal guidance for such questions, but there are beginning to be guides: color-coded charts that estimate the relative risks of everyday activities. They might look familiar, if you live in a place like Georgia where pollen count sliders tell you when it’s safe to go outside, or if you’ve visited US National Parks, where fire danger signs let you know it’s OK to light a campfire; they are mostly colored in the familiar hazard range of green-yellow-red. They come from individual scientists, from academic researchers, and from professional organizations, and they provide insight and support for the daily decisions of life in a pandemic: the equivalent of a traffic light flashing stop or go.

The first to be published, the Covid-19 Risk Index, is the creation of three public health experts. Ezekiel J. Emanuel, chair of the department of medical ethics and health policy at the University of Pennsylvania, and formerly an adviser in the Obama White House, launched the guide in a personal blog post on June 30, explaining that the team plotted possible activities on a matrix with four axes: whether an activity would occur in an enclosed space; whether it would attract a crowd; whether it would subject someone to “forceful exhalations” (sneezing, yelling, singing or coughing); and how long the activity would last.

The other authors are James P. Phillips, a physician and the chief of disaster medicine at George Washington University Hospital, and Saskia Popescu, an infectious disease epidemiologist at the University of Arizona. Popescu simultaneously released it on Twitter, where it has been liked more than 2,300 times. The goal, she says, is to move communication about Covid-19 risks away from a binary “do this/don’t do that” message. “We wanted people to understand, as life opens up, that there is a range of possible risks, and there are things you can do to stay safe, and things you might want to avoid,” she says.

Giving people permission to do things that seem likely to be safe was an important part of formulating the next guide to be published, which was drawn up by the Covid-19 Task Force of the Texas Medical Association and released July 8. Like the Emanuel group’s guide, the Texas guidance starts in deep green and cascades to bright red, passing through lighter green, yellow and amber. Unlike its immediate predecessor, it stacks activities vertically instead of horizontally—and it also subdivides each color group into more and less risky activities. Within the light greens, for instance, it ranks playing golf as less risky than walking through a busy downtown; among the ambers, it judges a hair salon visit as less risky than a plane ride. (Since it originates in Texas, home of some of the country’s largest mega-churches, it specifically lists “religious service with 500+ worshippers” as something to avoid.)

“It was important to us that there be things that were doable,” says Ogechika K. Alozie, a task force member and infectious-disease physician in El Paso. “If everything on our list had been orange and red, people might have reacted with: ‘This isn’t achievable for me, so why should I bother?’”

To come up with the rankings, the task force, along with the medical association’s Committee on Infectious Diseases, amassed a list of the questions they were getting from patients and colleagues, and rated each activity using whatever they could extract from the scientific literature. The problem is, there still isn’t that much. Though possibly tens of thousands of papers on the novel coronavirus have been published or uploaded to preprint servers since January, there hasn’t been sufficient time for experiments that definitively prove the risks of different exposures.

There are extant papers, though, on exposure risks in the earlier coronavirus epidemics: SARS in 2003 and MERS beginning in 2012. And there’s abundant literature on other respiratory and airborne pathogens. The task force borrowed whatever seemed useful, and layered their best clinical judgement on top. “We consider these to be consensus guidelines based on expert opinion,” Alozie says.

The guides are meant to be reassurance, but they’re also an indictment. Guidance like this ought arguably to be published by the Centers for Disease Control and Prevention, normally the source of public health advice on risks from seatbelt use to smoking to unsafe sex. It’s obvious at this point that the CDC has been muted for political reasons; witness the absence of director Robert Redfield from the White House podium, and the move last week to take data on Covid-19 hospitalizations away from the CDC’s control. If the public health apparatus of the United States was performing its core tasks, private citizens wouldn’t have to step into the gap. “I wish we didn’t have to do this, but here we are,” says Popescu. “As hard as we push this, it won’t reach as many people as it would if the CDC had published it. So the lack of leadership and resources at a national level has far-reaching implications.”

To guarantee a chart informs the public, it’s not enough to research what to say. It’s also critical to figure out the right visual language to help people understand it. Ordering risky activities top to bottom like a list, or left to right like a scientific chart, might seem a simple choice, as self-evident a communication strategy as borrowing the green-yellow-red symbolism of a traffic light. In fact, though, it involves assumptions: about an audience’s cultural background, about their graphic and data literacy, even about their ability to see color. (About 8 percent of men with European Caucasian ancestry and at least 4 percent of men of Asian heritage have red-green color blindness; the rates are much lower in women.) Meanwhile, it’s so predictable that it has already been parodied, in a version likely to reach more people than any of the originals will: an xkcd comic featuring a hilariously complicated chart in which the pinnacle of risk is “winning a test-tube-eating contest at a Covid testing lab.”

David McCandless, a journalist and data visualization artist who runs the design consultancy Information is Beautiful, has been thinking about those underlying problems. (His own version of the risks of different activities goes left to right but only from yellow through orange to red; it relies not on blocks or bars, but on bubbles of different sizes.) Data visualizations, he points out, are driving the news cycles of the pandemic: the charts of job losses, the fever lines of case counts, the progress bars of trackers monitoring drug discovery and vaccine research. “In a climate of anxiety, which I believe it's fair to say we're in, there is a hunger for guidance,” he says.

Yet it’s possible they can distort strategic thinking, by conveying more certainty than actually exists. “Risk is a tricky thing to communicate,” he says. “You take what is ultimately qualitative—it has to be qualitative, because you are measuring environmental and human pressures, and things you can't control—and you crystallize it into a chart. And suddenly everything is very distinct.”

McCandless’s graphics are based on data culled from a range of sources including the Johns Hopkins Coronavirus Resource Center national and global dashboard, statistics released by individual nations (quoted in some cases by news reports), and numbers from health departments and states within the US. Still, the choices a designer makes in visualizing data can be inadvertently misleading, he acknowledges. “The challenge has always been how to make visualization a consistent thing: How do you depict uncertainty, how do you depict confidence intervals, how do you depict hazard ratios, how do you depict the subtlety of statistical ranges?”

At Georgia Tech in Atlanta, a team led by Joshua Weitz, who directs the Interdisciplinary Graduate Program in Quantitative Biosciences, is trying to address that certainty problem by building a risk map that is grounded in data that addresses hazard and risk. Their Covid-19 Event Risk Assessment Tool uses data on new and recent cases, harvested from state health departments and collated by the volunteer Covid Tracking Project, and subjects it to an algorithm that accounts for a predictable level of under-reporting because of the scarcity of tests. It feeds that result into a mathematical model designed to answer this question: What is the probability that someone capable of spreading Covid-19 is attending any large event?

The answers are plotted onto a county-level heat map of the United States using the classic hazard colors, though slightly softened instead of glaring: a pale yellow down to a deep bruised red. And they can be disturbing. Last Friday, according to the map, the chance that an infected person would be present in a gathering of 100 people in Atlanta was 99 percent. In Chicago, it was 68 percent. In New York City, it was 42 percent.

Weitz says he had been tinkering with exposure models since concerns for a possible epidemic began to percolate in January. In March, Atlanta was supposed to host the March Madness tournament—and though cases in Georgia were still low, he began thinking of how to quantify the possible risks. (Georgia is now one of the US hotspots, and mired in a legal battle between city mayors who have imposed mask mandates and a governor who refuses to allow them.) “As cases have begun to rise here and schools and businesses are reopening, people are asking hard questions: Can I send my child into a classroom? Can I safely go into a bar or a restaurant?” Weitz says. “Answering those questions is the core of what we are trying to do.”

One of the site’s features is a slider which the user can use to choose the size of event, from as few as 10 people (“a dinner party,” Weitz estimates) to as many as 10,000 for sports events. (The crowd size for March Madness might have reached 100,000.) If you zoom out to a view of the entire United States, and ask about risk at a dinner party, things don’t look too bad: a risk probability of about 25 percent in northern and western states, and 25 to 50 percent in the Southeast and South. Move the slider to larger gatherings, though, and things get worse quickly. A gathering of 100 people, the size of a small wedding in the Before Times, carries an exposure risk of 75 percent or higher in all of the Southeast and most of the Southwest. At 1,000 people, the size of a concert audience, only a smattering of counties remain low-risk, all in upper New England and the Great Plains.

It is a daunting view. But Weitz points out that the map doesn’t only show us how much we are at risk. It also shows us that lowering risk is possible. “I hope this empowers people to ask political leaders: ‘Why are we in this situation when others are not?’” he asks. “I'd like to think that people could use the site as a means to talk to a city council member, a school board, their business, and say, ‘We can do better, because these other places are doing better.’”

That’s a hopeful interpretation. It turns the standard risk chart, the bars and blocks that confine us to where we can exist without danger, into something quite different—into a road map, laid out in the colors of safety, that could lead us to where we want to be instead.

Update 7-21-2020 11:08 am EST: This story was updated to correct the crowd-size estimate for March Madness.