Model Revisions Have Rare Good News About Coronavirus Pandemic

Apr 9, 2020
Originally published on April 9, 2020 9:51 am
Copyright 2020 NPR. To see more, visit https://www.npr.org.

DAVID GREENE, HOST:

We have some rare good news about the coronavirus epidemic. A major model has lowered its prediction for the death toll in the United States. This model predicts some states will start to see fewer deaths from COVID-19 each day. It actually says some states may have even passed their peak. This was Dr. Anthony Fauci, the government's top infectious disease expert, yesterday.

(SOUNDBITE OF ARCHIVED RECORDING)

ANTHONY FAUCI: What you do with data will always outstrip a model. You redo your models depending upon your data, and our data is telling us that mitigation is working.

GREENE: OK. We have heard from the White House about this particular model often. It was created by researchers at the Institute for Health Metrics and Evaluation at the University of Washington, which is why it's called the IHME model. So can we feel optimistic based on this? Well, I want to bring in NPR data editor Sean McMinn. Hi there, Sean.

SEAN MCMINN, BYLINE: Good morning.

GREENE: So as you look at this model and these revisions, what changes stand out to you?

MCMINN: Well, first of all, this model looks at each state and projects when it (inaudible) start seeing a consistent downward shift in new deaths. That's what they call the peak. For the U.S. as a whole, the model places the peak daily deaths on this upcoming Easter Sunday. In total, it projects between 31,000 and 127,000 people dying in this country. Now, that's lower than some numbers that they had previously reported.

What we're seeing with these new projections are more intense spikes but generally shorter ones. So what that means is that at the height of the disease in each state, there could be a fairly large number of people who die on a single day. But the model expects there to be fewer of those days overall, bringing the total death count down. They also looked at new data on how many people with the disease need to be hospitalized, and they said that they think that number is lower than what they originally expected.

GREENE: OK. So Fauci said - I mean, this is something you do; you redo your models. So what exactly is behind the changes in these projections?

MCMINN: So the team at the University of Washington was originally just using Wuhan as a case study in how long it takes for a region the top off after their first deaths are reported. Well, since then, they're now getting data in from places like Spain and Italy. And when they baked those into the model, they saw it was actually a shorter amount of time between the first cases and the peak than what they originally thought, which is good news - right? - because it means that this thing will slow down faster than what they were thinking before as long as people continue to maintain social distance.

GREENE: Well - and that's a key question, what people should do. I mean, when a state has reached its peak, does that mean people can relax a bit and start to resume normal lives or no?

MCMINN: No. Unfortunately, it does not. The team behind this model has really stressed that for the numbers to go down and stay down, people need to continue social distancing at least through the end of May. That's their assumption. Chris Murray, who's the lead researcher, made that point at a press conference on Monday.

(SOUNDBITE OF PRESS CONFERENCE)

CHRIS MURRAY: If you ease up prematurely, the epidemic can rebound to right back to the level we are at now in a matter of weeks. So the potential for rebound is enormous if we let up on social distancing.

MCMINN: So even though the model has had some increasingly optimistic projections, it's all based on the premise that social distancing continues. It also assumes that states that haven't mandated stay-at-home orders yet will do so in the next week, which of course is not guaranteed to happen. So I think a helpful way to think about it is this - that the model doesn't give us a hard date when life can go back to normal; it helps state officials and hospitals have an idea for when their health systems need to be ready for a surge of patients.

GREENE: Let me just - I mean, this is one model, right? Is there a reason to trust this over the others that are out there?

MCMINN: Yeah. So the IHME researchers have told us that they're getting requests from all sorts of government agencies, including the White House, for information on how this model sees the situation on the ground and asking them to run different scenarios. The thing is, the White House's own guidelines only advise social distancing through the end of April, which is shorter than what this model assumes. So any kind of hopeful conclusions that this model might show about what death counts are expected to do, that's predicated on something that the White House, so far, has not said it is willing to do.

Now there's other models out there, including one at Columbia University, which projects later peaks than what IHME does (ph) (inaudible). The two models look at social distancing differently, though. IHME assumes people will follow stay-at-home and other government orders, whereas Colombia expects people will do a better job social distancing once things gets bad where they live. So neither of these are really definite forecasts; they're more the guideposts for what states should be preparing for.

GREENE: All right. That was NPR's data editor Sean McMinn. And we should say, online we have data for each state's projection as well as a list of all the states in order of when they are expected to hit their peaks. You can see all of that at npr.org. Transcript provided by NPR, Copyright NPR.