Reopening states will cause 233,000 more people to die from coronavirus, according to Wharton model

[steve2470]

https://finance.yahoo.com/news/reopening-states-will-cause-233000-more-people-to-die-from-coronavirus-according-to-wharton-model-120049573.html

New data from the University of Pennsylvania suggests that relaxing lockdowns across U.S. cities and states could have serious consequences for the country’s battle to contain the coronavirus, which has infected over a million people while killing more than 66,000 people.

According to the Penn Wharton Budget Model (PWBM), reopening states will result in an additional 233,000 deaths from the virus — even if states don’t reopen at all and with social distancing rules in place. This means that if the states were to reopen, 350,000 people in total would die from coronavirus by the end of June, the study found.

Kent Smetters, the PWBM’s director, said the decision to reopen states is ultimately a “normative judgement that comes down to the statistical value of life.”

He explained: “That’s not a crude way of saying we put a dollar value on life, but it’s the idea that people will take risks all the time for economic reward.”

[regnaD kciN]

1. That is a very confusingly-written article...

...since it says "reopening states will result in an additional 233,000 deaths from the virus — even if states don’t reopen at all and with social distancing rules in place." Reopening states will do that even if they don't reopen?



I'm guessing, from the way the article is written, that they left off a bit of text that was supposed to follow the em-dash. As far as I can tell, it was supposed to say that, even if states stayed closed, we can expect another 51,000 deaths by the end of June, for a total of 117,000. If states re-open (and I don't know if they mean every state, or just those who are doing so right now or about to do so), there there will be an additional 223,000 deaths on top of that, for a toll of 350,000.

Of course, it should be noted that the PWBM is one of the most pessimistic models, as its 117,000 is considerably higher than the IHME model, which is currently forecasting 72,433 fatalities. However, each model has a high level of uncertainty, and IHME's current model (which, I believe, was from last Monday) predicted the number could go as high as 114,228 by August, assuming the maintaining of current social-distancing measures.

[jojog]

2. It will be 72,433 in a week nt

[genxlib]

3. The IHME

Is quickly losing credibility. They will be bypassed in a few days.

They got lots of attention early because they were at the first outbreak in the US. Then they rocketed to attention because the White House likes their rosey projections. At the time they had dropped all the way to 60k and you see how well that held up.

I suspect that they are modeling the theoretical impacts of the control measures rather than the imperfect way they are actually being enacted.

[regnaD kciN]

4. You obviously know very little of the IHME analysis...

..and didn't bother to read my post very carefully.

Their model gives a wide range of possible outcomes, and is updated weekly based on data received at that time. While they give a "best estimate" number (which is currently at 72,433, and likely to go up with the next model run, since the current one is a week old), that's only one figure in a wider range of possible outcomes. As I stated before, their model allows for the possibility the death toll will go as high as 114,000+, and that's even with current restrictions kept in place. And, for those who think the White House favors them, you may want to note that IHME has also been making recommendations on when each state should consider relaxing their restrictions, which "the White House" has been studiously disregarding.

Incidentally, here is a 538 article that discusses the different models and their underlying principles.

[genxlib]

5. I understood your post just fine

You can be dismissive of my opinion if you want to but I am not the only one. https://www.vox.com/future-perfect/2020/5/2/21241261/coronavirus-modeling-us-deaths-ihme-pandemic

Moreover, right wing media has been making hay out of this for weeks because they love every chance to undermine science based policy making. I have been defending the IHME for weeks but I am getting tired of it because they don't seem to be getting their issues fixed.

I understand why they dropped to the low 60's when the Country looked to be serious about a shutdown. But they have been very slow to adapt off of that assumption when it has been clear for weeks that the efficacy of the shutdown was not working as well as they projected.

Here are my thoughts on the matter. The disease (on average) takes about 14 days to resolve. So the people dying today were already sick two weeks ago. They are already in the pipeline. The net result is that the near term 2-week projections should be much more reliable. On the other hand, the long term projections have many more variables since they involve predicting the full range of activity, transmission, illness and outcomes. There are really two kinds of projections going on here. The level of uncertainty for the 4-16 week projections should have a much wider range of possible outcomes.

Considering that, take note of the last few weeks...

the IHME number on 4/22 was 67,641. We have already passed that number in 12 days.

On 4/27, they updated their number to 74,073. We will probably pass that again within about 12 days.

That by itself represents an error but they actually revised the number down to 72,437 on 4/29. We will probably pass that in a few days (or 7 days from their projection)

It isn't that they are missing that bothers me. It is that they are missing in very short time frames that should have much higher certainty. Those deaths are already baked into the system and only requires them to evaluate the death rate of active cases.

As they have made adjustments, they continue to assume that the death rate will be dropping quickly from a peak. But the deaths have plateaued and continue at a relatively high rate. That isn't a huge surprise considering that the new case totals show a similar plateau.

By the way, I didn't say that the White House favors them. They don't favor using any science to make decisions. They are simply willing to use them as a tool when it fits their narrative. All I am saying is that it raised the IHME profile while other similar models remained more obscure.

Edited to add...the 14 day to resolution number is actually related to after onset of illness. I left out the incubation period since those cases are not known based on current testing shortfalls.

[genxlib]

6. And there it is

Just now, the IHME updated to 134,475.

They appear to be taking the relaxation of rules into account so the increase is not surprising.

But please note this passage from their notes. This is one of their major modeling improvements.

Hospitalizations of COVID-19 patients as an additional leading indicator for estimating COVID-19 deaths in the next eight days. As of today’s release, we now include two leading indicators for locations where hospitalization data are available: the number of COVID-19 cases (added for our April 17 release) and hospitalizations of COVID-19 patients. Each indicator is used to inform the trend in the number of COVID-19 deaths in the coming eight days. In other words, if the number of hospitalizations (and/or cases) has increased over the last few days in a given location, we want our model to predict that deaths are also likely to increase eight days later.

This lines up with my major criticism on the short term failures of their projections.