A Virus Study You’ve Never Heard of Helped Us Understand COVID-19

What Columbia University researchers learned when they tried to get a complete picture of how respiratory viruses spread across Manhattan

Brooklyn Bridge
A woman wearing a mask walks the Brooklyn Bridge in the midst of the coronavirus (COVID-19) outbreak on March 20, 2020 in New York City. Victor J. Blue/Getty Images

Virus studies tend to be passive, not proactive. People get tested when they show up for treatment. But that paints only a partial picture of infections, one that misses those who are infected and spreading the disease but don’t go to the doctor.

Jeffrey Shaman of Columbia University's Mailman School of Public Health wanted to fill in the rest with a radical kind of study, one that tested and tracked seemingly healthy people to see who was unknowingly spreading disease. Beginning in March 2016, Shaman and his team at Columbia began the Virome of Manhattan, an ambitious project to build that picture of respiratory viral infections throughout the borough.

The U.S. Department of Defense funded Shaman’s work in the hopes of finding new ways to keep soldiers safe by predicting the spread of disease. But when COVID-19 pandemic hit earlier this year, the research became something more: a road map to the unfolding mysteries of the virus. The virome project asked questions that later would become crucial for beginning to understand SARS-CoV-2. Do people who come in for treatment represent the majority of infections or a minority? What allows viruses to persist in a community and move around? Can people be infected again?

A man covers his face as the sun rises behind in Manhattan on April 6, 2020 as seen from Weehawken, New Jersey. Kena Betancur/Getty Images

Past outbreaks, current trends

The Virome of Manhattan study surveyed three populations. Researchers tested patients in pediatric emergency departments and the people who accompanied them. They swabbed visitors to a tourist attraction, taking a medical history and asking participants if they'd felt cold or flu systems over the past 48 hours. They also recruited 214 volunteers who, every day from October 2016 through April 2018, entered into a phone app their symptoms including cough, fever, sore throat and muscle pain, and whether they stayed home, took medication or saw a doctor.

Researchers swabbed the volunteers weekly to track 18 different viruses, including the four endemic coronaviruses that were known at the time to infect humans. Those viruses cause the common cold and occasionally more serious complications like pneumonia or bronchitis. Their more dangerous cousins are responsible for Severe Acute Respiratory Syndrome (SARS) and Middle East Respiratory Syndrome (MERS), which together killed more than 1,600 people during outbreaks from 2002 to 2004 and 2012 to the present.

Unlike the flu, coronaviruses had no global surveillance system before the COVID-19 outbreak, which is why Shaman's project created a valuable glimpse into their behavior. "Endemic coronaviruses are still pretty ignored because they're pretty wimpy, which is not a smart thing considering we've already had SARS and MERS," Shaman says.

Amesh Adalja, a researcher at the Johns Hopkins University Center for Health Security focused on emerging infectious disease and pandemic preparedness, described Shaman’s study tracking specific viruses as long overdue. Major problems arise when there are undocumented chains of transmission, as with COVID-19, that end up infecting vulnerable people.
“I think this is something that we really need to think about doing as part of routine care, trying to understand what viruses people have,” Adalja says. “One of the things that I'd like to see continue after the acute phase of this pandemic is that respiratory virus testing becomes something that is done very, very normally.”

The Manhattan project found that one in nine people were infected with a respiratory virus during February, the peak of the flu season. Yet relatively few saw a doctor—just 22 percent of people with the flu and only 4 percent with endemic coronaviruses. That problem became startlingly clear with the arrival of COVID-19 and the widespread fear of silent transmission by asymptomatic carriers. “There's a large percentage of infections that are undocumented," Shaman adds. "They are contagious. Not as contagious as the confirmed cases. But because there's so many more of them, they're the ones who are setting up these silent chains of transmission, which we're unaware of until somebody gets sick enough that they go to see a doctor."

Until the Manhattan Virome project, Shaman says, researchers hadn't seen just how much these undocumented cases contributed to the viral spread. When SARS-CoV-2 emerged in January, his team realized its spread was probably fueled by the same phenomenon. Working with a team of researchers in China and elsewhere, they published a paper explaining how the novel coronavirus had been spread by people without symptoms. They found that 86 percent of all infections were undocumented before the Wuhan travel shutdown in January. Per person, the undocumented infected were only half as contagious as documented infections, yet they were the source of two-thirds of those infections.

"The thing that allows this virus to get around—in addition to being a newly-emergent disease to which the population is susceptible—is that it has this stealth transmission," Shaman says. "People whose symptoms are mild or nonexistent are out and about going shopping, going to work or school, taking business and vacation trips and bringing the virus into new populations."

Manhattan and COVID-19
Art by Rick A. Diaz of the statue of Liberty wearing a mask is seen on May 10, 2020 in the Manhattan borough of New York City. Jeenah Moon/Getty Images

Open COVID questions

Work by Shaman and others prodded governments to shut down in March and slow the progress of the virus. Then, he turned his attention to mine another crucial question about the COVID-19 threat: whether people can be infected again.

By following people through flu seasons, Shaman found that 137 had been infected with one of four coronaviruses that cause the common cold, two closely related to COVID-19. Twelve people were reinfected sometime between four and 48 weeks later. The evidence suggested two possible explanations: that immunity wanes over time, or that the virus picks up mutations that allow it to escape the body's immunity.

It's not yet clear how this lesson applies to COVID-19. So far, no clear evidence exists of repeated infections, but the disease is new. SARS, caused by another coronavirus, was tamped down so effectively that the chances for reinfection were miniscule. "One of the big ponderables for this current pandemic is are we going to be one and done with it or are we going to be subject to repeat infections?" Shaman says. "If we are subject to repeat infections, are those infections going to be more likely to be milder, about the same severity, or worse? That's something we also do not know yet at this point."
Shaman is trying to answer those questions by diving back into the Manhattan study’s data. He also seeks to discover whether a person’s genetics play a part in the severity of a coronavirus infection and whether the virus is seasonal. The genetic question arises because families in the Manhattan study suffered clusters of infections. Was that because they lived together and shared germs, or because they share the same genes? People can be tested for mutations like breast cancer genes, but there hasn’t been much study about genes and acute infectious diseases.

"That's something that we set out to try to do,” he says. “We're working on that with the Virome Project to try to understand why some individuals become sicker than others. This is an important field of inquiry that hasn't been delved into to the extent that it needs to be."

As for the question of COVID seasonality: Shaman and his team won a Centers for Disease Control competition in 2014 to predict the timing, peak, and intensity of the flu season. They optimized their model by testing it against real-time flu activity in the recent past and then determined whether the forecasts converged on a similar outcome.

The other coronaviruses, notably OC43 and HKU1, relatives to the COVID-19 virus, are also seasonal. But because COVID-19 has different reporting rates across countries, Shaman says, it's hard to tease out whether the novel coronavirus behind the disease will be seasonal and slow transmission during the summer, as happened during the pandemics of 1918, 1957, and 2009. "I'm not sure it will," he says. "I don't think the (seasonal) modulation is going to be substantive enough to preclude transmission. And we are still cranking along. We have 22,000 cases a day in the United States, which is much too high."

Shaman is still mining the data and plans to issue new papers soon. His team recently modeled the spread of the virus, concluding that if government officials had shut down a week earlier, they could have saved 36,000 deaths, about 40 percent of the total. He's been examining pandemics his entire career, but wasn't sure he'd see one like this.

"This is, I hate to say, the big one," he says. "This is the most disruptive thing that we have seen since 1918. I sit back some days and think, ‘I can't believe we're dealing with this now.’"

Get the latest Science stories in your inbox.