When Gov. Jay Inslee announced a ban on large gatherings last week and warned schools to prepare for extended closures, he was standing next to a chart that showed the consequences of inaction: a relentless, upward trajectory in new coronavirus infections with 25,000 cases by early April and 400 people likely to die in King and Snohomish counties alone.
Those sobering projections were generated by a little-known team of local researchers who until recently focused mainly on tuberculosis, malaria and other scourges of the developing world. But as they watched the novel coronavirus spread around the globe and land in their own backyard, the epidemiologists and computer whizzes at the Institute for Disease Modeling (IDM) in Bellevue quickly shifted gears to help health officials and policymakers track and forecast the pathogen’s progression.
Their efforts over the past few months have influenced difficult decisions across the state, from closing schools to turning visitors away from nursing homes. Seattle Mayor Jenny Durkan quoted the institute’s estimates in a CNN interview. Many of the team’s findings have been freely shared with the public and other scientists, and their phones — and computer conference lines — have barely stopped ringing.
It’s an unfamiliar situation for scientists accustomed to working behind the scenes and publishing their results in academic journals where peer review can take months.
“As modelers … we try to do our work in the background,” said Mike Famulare, the group’s lead coronavirus modeler. Now he and his colleagues are being called on to advise state and local leaders about a fast-moving threat with lives at stake. “It feels like an enormous responsibility, and also a privilege,” Famulare said, speaking via Zoom from his home, where he’s been working remotely — like all of IDM’s staff.
Driven by a sense of urgency, Famulare and his colleagues are incorporating new data into their models as fast as they can. They’re also rushing to evaluate the impact of public health interventions and changes in behavior to determine what works best — and what doesn’t. A typical day is 11 to 13 hours of furious multitasking, with a short break to walk the dog, he said.
“To use an analogy I heard from someone else, we’re building the plane as we’re flying it.”
Facing the country’s most severe outbreak of the new virus, Seattle is fortunate to have a rich array of scientific expertise — as well as public health officials who work closely with scientists and value their advice, said Judith Wasserheit, chair of the UW Department of Global Health. “This means that the science actually gets translated into action here more quickly than in many other parts of the country,” she said in an email.
Modeling like IDM’s is particularly helpful, she added. “If you are trying to model a local or regional outbreak, having local modeling capacity is tremendously valuable because it increases the likelihood that local parameters will be used and interpreted correctly.”
IDM is one of several local research organizations that quickly pivoted to fighting the coronavirus, which had killed 40 and resulted in 642 confirmed positive cases in the state as of Saturday. UW Medicine’s Virology lab was one of the first university sites approved to test for the virus and will soon be processing 4,000 samples a day. Geneticists at the Fred Hutchinson Cancer Research Center are sequencing viral genomes to uncover the pathways the disease is following around the state.
The Bill & Melinda Gates Foundation pledged $5 million for the local response effort and is working with Amazon on a plan to deliver nasal swabs to people’s doorsteps, then collect them for speedy analysis. A project called the Seattle Flu Study, which was using similar home-test kits to track the spread of influenza, swiftly retooled to analyze for the new coronavirus.
The Institute for Disease Modeling was already oriented toward public health. It’s part of the Global Good Fund, a collaboration between Bill and Melinda Gates and Intellectual Ventures — a tech incubator founded by ex-Microsoft billionaire Nathan Myhrvold. IDM was established to provide computational firepower for disease control and eradication in the world’s poorest countries, and much of Famulare’s earlier work was on strategies to wipe out polio. Now, about a third of the institute’s 100 employees — from biostatisticians to experts in social mixing patterns — are devoted to the new coronavirus.
Famulare, who was also collaborating with the Seattle Flu Study, published an analysis in late January cautioning that the new pathogen appeared to have the potential to cause a pandemic worse than the 1957 flu pandemic that killed more than one million people worldwide — and possibly comparable to the 1918 influenza outbreak that killed at least 50 times that many. It was while doing those calculations that the true scope of the peril really struck him.
While the number of cases starts out small in each locality, experiences from China and other hard-hit countries show that diagnosed cases are just the tip of the iceberg. Because it can take two weeks for infected people to get sick enough to seek medical care and get tested, early in the outbreak the actual number of cases is probably eight times the number of confirmed infections, Famulare said.
“Whatever picture the raw data paint today, there is more that is just starting to happen, but you won’t have seen it yet.”
Washington is at the point where infections are poised to explode, or increase exponentially, as an ever-growing number of hidden infections continues to double every six days or so. As a modeler, exponential growth is a concept Famulare is very familiar with. But being in the thick of Washington’s outbreak has brought it home to him in a visceral way he never experienced before.
“I’ve been modeling exponential growth for years,” he said. “It’s only in the last two weeks that I’ve felt it in my bones.”
Every day is significantly different from the day before, and delays in action can have real, and terrible, consequences. Of all the problems facing society today, COVID-19 is the only one that will get twice as bad every six days if we don’t act to slow it, Famulare said. “That creates this enormous pressure to try to understand things as soon as possible, because a week’s delay means a much bigger problem.”
But there’s also a hopeful side to the modeling, and in the way those exponential curves have been flattened in China, Japan and other countries that took decisive action and are now seeing substantial drops in infection rates.
Famulare and his colleagues estimate that social distancing measures aimed at cutting the transmission rate in half could reduce the number of future infections in the Seattle area by a factor of five — from the roughly 25,000 predicted by early April if the virus remains unchecked to 4,800. Deaths would be reduced from a predicted 400 down to 100.
Now, they’re trying to quantify the impact from specific changes such as school closures and the shift to remote working. They’re also modeling the impact on health care facilities from the increasing number of sick people and a lack of equipment, at the same time many nurses and other workers might have to stay home to care for kids who are out of school.
Famulare characterizes much of his team’s works as “educated guesses,” and he scrupulously points out their assumptions and margins of error — which can be wide, given the high degree of uncertainty and incomplete information. They’re doing their best to fill in those gaps as they go, he said.
“We’re asking ourselves: ‘What did we get wrong last week that we’re going to do better this week?’ ”