'How Do You Quantify Panic?'

The public response to an outbreak often far outweighs the actual threat. In a new paper, researchers say they've created a formula to measure disease-induced hysteria.

One of the more memorable incidents of panicked Ebola news coverage came in early October, when CNN’s Ashleigh Banfield asked her guests on Legal View: Is Ebola the ISIS of biological agents?

Almost immediately, the wisecracks started rolling in.

“Without a doubt,” Stephen Colbert deadpanned on The Colbert Report. “Scientists have long compared diseases to murderous madmen. That’s why epidemiologists call tuberculosis ‘Lung Hitler.’”

“Is Ebola the Boko Haram of AIDS? Is Ebola the al-Shabab of dengue fever?” the New Yorker asked. A headline from New York magazine: “Ebola coverage goes extra dumb.” And one from The Independent: “CNN is asking the stupid Ebola questions.”

Others tried to channel public Ebola panic into something more productive, urging everyone concerned about an epidemic to stop worrying and get a flu shot instead.

From the numbers, it’s hard to say that it worked: Less than half of people in the U.S. last year were vaccinated against the flu, a disease that hospitalizes around 200,000 Americans annually. Meanwhile, a Washington Post poll published in October found that two-thirds of Americans were concerned about a large-scale outbreak of Ebola, which hospitalized 10.

The moral of the story: The threat of the disease and the size of the panic it causes are often very, very different things.

In a paper recently published in the journal Interface, researchers from the Massachusetts Institute of Technology, the Draper Laboratory, and the disease-forecasting company Ascel Bio say they’ve found a way to predict overreaction to outbreaks.

Scientists already have a fairly full toolbox when it comes to ways of tracking the spread of disease, from Google Flu Trends to monitoring the sale of over-the-counter medicines to analyzing patterns in Wikipedia searches. But predicting the social reaction to a disease is more difficult. As study co-author Marta Gonzalez, a professor of civil engineering at the Massachusetts Institute of Technology, said in a statement: “How do you quantify panic?”

To find out, Gonzalez and her colleagues analyzed a year’s worth of data from Ascel Bio on 300 diseases across 200 countries. Combing through around 12,000 separate sources—hospital records, news reports, social media posts—they compared the attention surrounding a given disease to its actual impact. Plugging in variables like biosafety level and local familiarity with a disease, they were able to create a model that could track the relationship between the panic incited by an outbreak and the threat it actually posed.

Judging the threat of an outbreak is a combination of local and global information, or incidents a person witnesses of a disease coupled with what they know about it more broadly.  But “most current models of disease and information spread assume that individuals in the population are rational actors, who assess the likelihood of infection and take protective action accordingly,” the study authors wrote. “While this assumption may hold for some health behaviors, it neglects the emotional component and the social component of health decision-making.” Among the emotional and social factors in play: A disease was more likely to incite panic, they found, if it was perceived as novel, if local experts were inexperienced in handling it, or if it was considered more severe, even if it had relatively few cases or was slow to spread.

Overreaction also bred more overreaction—“agents are biased towards adopting the opinions of their most concerned neighbors, rather than their calm ones,” the authors wrote—as did extensive media attention. (Past research has documented the same phenomenon; in one 2008 study, for example, participants tended to rate avian flu as more severe than yellow fever).

The team then tested its model on two case studies: one comparing the 2003 SARS outbreak and 2009 H1N1 outbreaks in Hong Kong, and other comparing the spring and fall 2009 outbreaks of H1N1 in Mexico City. In both cases, the first outbreak was both less severe and sparked a larger social reaction than the second. Again using health records and media sources, the researchers found that their model could accurately predict the relationship between the disease and its public response.

Down the road, the researchers wrote, they plan to use the model to help policymakers better prepare for disease-induced hysteria, which in more extreme cases can lead to riots, hoarding of supplies, economic damage, or migration that actually furthers the spread of an outbreak. “If we could predict that these bad social and economic consequences are going to happen that might cost a lot of money and might cost a lot of lives, [people] can take measures to counteract these effects,” Gonzalez said in the press release.

As their own model proved, though, fear is a hard thing to extinguish—even when the public is flooded with scientific facts meant to counteract it—and the line between vigilance and overreaction can be a tricky one to navigate. Quantifying panic is one challenge; preventing it before it starts is entirely another.

And that’s assuming that everyone with the power to prevent it will go along. As The New Yorker wrote, with what may very well be the best burn to come out of the Legal View incident: “Let me open the discussion up to our panel and ask whether Ebola is merely the Fox News of explosive incontinence, or whether the situation is much worse than that and Ebola is, in fact, the CNN of CNN.”