One of the most pernicious parts of the COVID-19 crisis is how uncertain everything is. Researchers and officials cite statistical models that estimate infection rates, death counts and when things will go back to normal, but those estimates are changing rapidly. And as the forecasts bounce around, so do the rest of us living through the crisis. How can one feel settled when the future feels so volatile?
Still, there’s a way to at least get a sense for what the experts are thinking. For the past five weeks, infectious disease researchers from institutions around the United States have been taking a survey that gathers their thoughts on the trajectory of the COVID-19 virus. The researchers come from academia, government and industry and are experts in modeling the spread of viruses like this one. The survey asks about things like how many people will eventually get COVID-19 and how many Americans will die.
The top-line numbers are sobering. The most recent survey, taken on March 16 and 17, found that, as a group, the experts think that as of March 15, only 12 percent of infections in the U.S. had been reported by the Centers for Disease Control and Prevention. They think there’s a 73 percent chance of a second wave of hospitalizations this fall. And they expect approximately 200,000 deaths in the U.S. by the end of the year.
But averages can only tell you so much. When forecasting the future, it also matters what a person (or model) thinks the range of possibilities could be — how uncertain the forecast is, in other words. In this survey, the experts gave three answers to most questions, representing the most likely future scenario and the best-case and worst-case scenarios.
Collecting responses in this form captures both the best-guess estimate from each respondent and the uncertainty surrounding it. It also lets the people in charge of the survey — Thomas McAndrew and Nicholas Reich, both biostatisticians at the University of Massachusetts Amherst — convert the responses to a probabilistic consensus forecast,1 something that can answer questions like, “According to these researchers, what is the probability that we will have 50,000 reported cases by March 29?”
Expert consensus forecasts give you what a model does — a forecast that gives a measure of its uncertainty — without being overly reliant on just one way of thinking about a problem. In this instance, each expert has their own assumptions about how likely the virus is to spread or to be fatal, as well as assumptions about the ways humans might try to mitigate its damage.
Here’s what the researchers collectively had to say in the March 16-17 survey.
How many total COVID-19 cases in the U.S. will the CDC report on March 29?
At the time the survey was in the field, about 3,500 cases had been reported. But the experts estimated that by Sunday, March 29 — a little under two weeks after they took the survey — the country would have seen anywhere from 10,000 to 75,000 cases. (The current count is 15,219.) The experts’ confidence in those estimates, however, varied greatly:
Andrew Lover, an epidemiologist from the University of Massachusetts who took the survey, said his estimates were “semi-quantitative” and based on the virus’s progression in other countries. “The doubling times have been 5-8 days most places, so it’s a matter of applying that with some sliding-scale adjustments (testing rates, population density, etc.) based on the ‘feel’ of the epidemic curves.”
The consensus forecast generated by the individual responses indicates that we should expect roughly 19,000 reported cases by March 29, with an 80 percent chance of seeing between 10,500 and 81,500 cases.
What percentage of all COVID-19 cases in the U.S. had been reported as of March 15?
Experts estimated that, on March 15, when the CDC had publicly identified about 3,500 cases, only between 5 percent and 40 percent of actual COVID-19 cases had been reported. But experts’ confidence in those numbers was shaky.
The consensus model indicates that only 12 percent of cases had been reported at that time. In other words, the researchers think there were actually about 29,000 infections in the U.S. as of March 15, more than eight times the known tally.
How likely is it that there will be a second wave of hospitalizations later this year?
Just as flu season can have two peaks, the surveyed experts think there’s a good chance there will be a second wave of coronavirus-related hospitalizations sometime between August and December. Individual estimates for the likelihood of that second round of cases ranged from 40 percent to 96 percent, with an expert consensus of a 73 percent chance.
How many people will die in the U.S. due to COVID-19 this year?
Experts’ estimates of the number of COVID-19 deaths in the U.S. in 2020 ranged from 4,000 all the way to 1 million, a huge range that highlights how much we still don’t know about this disease.
The expert consensus is to expect about 200,000 deaths in the U.S. from COVID-19 this year, but the uncertainty around that number is also huge: There’s an 80 percent chance the final number will be between 19,000 and 1.2 million, according to these estimates.
The researchers plan to continue conducting these weekly surveys. As some elements of the pandemic become clearer — such as the virus’s incubation period and fatality rate, and how far the U.S. is willing to go to slow the spread of the virus — these ranges will presumably narrow.
But for now, there’s a lot the experts still aren’t certain of. Just like the rest of us.