Experts consult models to predict if coronavirus cases will spike
With months of real data filling in the profile of COVID-19, modelers are gaining confidence in projecting future outcomes of the disease — but not the human behavior that could determine the scope and timing of future waves of the pandemic.
With the economy opening and unexpected mass gatherings taking place, including those following the death of George Floyd at the hands of Minnesota police, projections for the summer and beyond are anticipating continuing waves of infection.
“The dominant source of uncertainty is we don’t know what people are going to do and the decisions that are going to be made even a few weeks out” that could encourage or limit mobility, said Jeffrey Shaman, an epidemiologist and modeler at Columbia University. “The models are going to be imperfect … Those uncertainties are enormous and they swamp it.”
William C. Miller, senior associate dean of research and professor of epidemiology at Ohio State University who has been working with Ohio state government to review models as it reopens, said the main factor is the lifting of strict stay-at-home orders, along with "sporadic cases" that could arise from recent demonstrations.
Loosening the lockdowns "has resulted in increased mobility and interaction," and will "very likely result in much more social interaction," and potential infection, he said, perhaps more so than demonstrations.
Miller said the demonstrations "do pose some risk for increases in cases. The biggest mitigating factors are that many people were wearing masks and the demonstrations were outside."
Worse, he said, any outbreak resulting from demonstrations will likely be more difficult to trace "as a person will not know who they were next to or when they were exposed."
Another concern, Miller said, is when groups of demonstrators are herded into smaller spaces.
"I expect we’ll see some sporadic cases but not enough to outweigh what is going on in every other aspect of the country in terms of reopening the economy," Miller said.
Mark Jarrett, chief quality officer and deputy chief medical officer at Northwell Health, said that while the models don’t project infections reaching April levels because controls likely would be tightened in the face of surges, there remain big ifs for the fall.
“We may also be dealing with flu season, and anything else that may be going on,” he said.
Accordingly, the model trajectories have various potential outcomes, like those shown on hurricane path projections on a weather forecast. “It’s never static, and we’d have to change our projections based on variables as we go along,” he said, adding surveillance based on testing and contact tracing could limit the scope of any second wave.
Still, the prospect of even a summer surge looms weeks after reopenings and the Memorial Day weekend. Stony Brook University has a new model for Suffolk that shows three projected courses for COVID-19 infections from the current 39,000-plus cases, including one that could see a spike to more than 45,000 by late June with a 50% increase in mobility. Just a 20% to 30% increase in mobility and inconsistent mask wearing can bring the level to around 42,500 by late June, the most likely scenario, the model states. The third, baseline scenario projects a relatively smaller increase in infections.
Sean Clouston, an associate professor at Stony Brook University’s School of Public Health who has been modeling the virus in Suffolk and for Stony Brook’s hospitals, said projections of the model to date have been relatively accurate, starting with the major decision to lock down society in mid March.
Now, he said, much will depend on how well people and companies practice and enforce social distancing rules and keep contacts to a minimum, particularly at newly reopened workplaces.
“As people start working more, they’re going to start engaging more in using shared spaces,” he said. “Mobility is OK, if people are driving alone, and going to a little office somewhere. But a lot of office buildings you have shared lobbies, elevators. The shared spaces are the hard part to figure out what the risks will be,” and how they could contribute to future infection increases.
Clouston said the impacts of the first reopenings should become evident by early this month.
Early models of the novel coronavirus projected more than a million deaths from the disease as it reached U.S. shores and began infecting major population centers. To date, the U.S. toll from COVID-19 just surpassed 100,000, with around one fourth that number hitting New York.
The projections were used to marshal equipment and medical forces throughout the country and most notably at the epicenter of the U.S. infection: New York City. Some models went so far as to project the specific number of ventilators that were needed — projections that vastly inflated the actual need.
What the models didn’t fully capture, experts say, was just how seriously most citizens would take warnings to shelter in place and keep their distance from one another. The waves began to dissipate.
Youyang Gu, an independent modeler and MIT graduate whose uncannily accurate projections have been carried by the U.S. Centers for Disease Control, said one of the surprises about the COVID-19 pandemic and the ability to model it involves assumptions about a summer abatement, which is typical of the flu.
“That’s an assumption a lot of the models are making, but there’s no data [indicating] that temperature is affecting this virus,” he said.
Still, Gu’s model for Nassau and Suffolk show a relatively stable line in anticipated deaths through the summer, based on the staggered and controlled reopenings in New York state.