Over the past week, the number of deaths thought to be caused by COVID-19 has nearly doubled, from about 16,000 to 30,000. New York City changed the way it counts its death toll on Tuesday to include people who did not receive a positive COVID-19 test but who were likely to have died from the virus. That increased the number of the city’s coronavirus death estimate by 3,700 to more than 10,000.
The number of cases and fatalities continue to change rapidly, so to get a better understanding of where they’re headed, we’ve been following a weekly survey of infectious-disease researchers from institutions around the U.S.
This week’s survey, conducted April 13 and 14, shows that the expert consensus is that reported deaths will increase to around 47,000 by May 1, although they think there could be as many as 82,000 by that time. The experts also think fatalities are most likely to peak in May, but they still see about a 1 in 3 chance that deaths won’t peak until June or later. They also expect that between eight and 11 states will report more than 1,000 deaths by May 1.
The survey — organized by Thomas McAndrew and Nicholas Reich, both biostatisticians at the University of Massachusetts, Amherst — asks experts to either assign a probability to an outcome or to give a most-likely, best-case and worst-case estimate. Structuring the survey in this way lets the organizers generate probabilistic consensus forecasts, a tool that can answer questions about how likely various scenarios are.
Here’s what the experts had to say this week.
How many deaths will be reported on May 1?
Experts were asked to give the smallest, largest and most likely number of fatalities they thought will be reported by The COVID Tracking Project as of May 1, and all but two thought it most likely that there would be 50,000 or fewer deaths, although some of the worst-case ranges extended far beyond that.
The expert consensus is that the U.S. will have reported around 47,000 COVID-19 deaths by May 1, with a 90 percent chance of having between 32,000 and 82,000.
This is very close to the latest forecast from the Institute for Health Metrics and Evaluation at the University of Washington, which projects roughly 51,000 deaths by May 1, with an uncertainty range from about 28,000 to 112,000.
Which month will see the most COVID-19 deaths?
For five weeks, surveyed experts were asked what factors led to the difference between their best-case and worst-case estimates in longer-term death forecasts. The answers have had a common refrain: They can’t perfectly predict what stay-at-home orders will be enacted, how long they will last, and how well people will follow those orders. That last issue became particularly salient this week as protesters in Michigan, Ohio and North Carolina organized in opposition to their state’s stay-at-home orders.
Experts think deaths are most likely to peak relatively soon, with a 65 percent chance that the highest number of fatalities will be reported in either April or May. But a later peak — perhaps because of relaxed social distancing measures — is still possible, with experts estimating a 35 percent chance that the peak will be in June or later.
How many states will have more than 1,000 COVID-19 deaths by May 1?
At the time the survey was conducted, three states — New York, New Jersey and Michigan — had reported more than 1,000 COVID-19 deaths. As of Thursday, Louisiana, Massachusetts and Illinois have also passed that threshold.
The experts were asked how many states will report more than 1,000 deaths by May 1. They estimated that the most likely number was between eight and 11 states, with a nearly 1 in 3 chance that at least 12 states would report that many deaths.
In this week’s survey, the experts were asked to report the proportion of their responses that were based on forecasting models they were directly involved with building, as opposed to “general knowledge about disease dynamics.” They said 44 percent of their responses were based on models they’ve helped build, up from 32 percent in a similar question two weeks ago.
The Institute for Health Metrics and Evaluation model has revised its projected death counts downward, as it appears that social distancing measures have helped flatten the curve of the coronavirus. But given the uncertainty that remains in these experts’ forecasts, it seems like they aren’t convinced the curve won’t start to rise again.