IHME now estimating 600,000 US COVID-19 deaths by May 1 (was 350K by Jan)

[Nithavela]

It's been more than two weeks since the last update. They musta gone on staycation.

Probably got orders to stay quiet.

 

[essentialsaltes]

Ha, my obsessive refreshing is good for something after all!

August 6 update
215,338 by October 1
250,258 by November 1
295,011 by December 1 [newly added estimate]

August 21 update
218,290 by Oct 1
254,717 by Nov 1
309,918 by Dec 1

 

[essentialsaltes]

August 21 update
218,290 by Oct 1
254,717 by Nov 1
309,918 by Dec 1

August 28 update
216,645 by Oct 1 <-- interesting that this one decreased.
259,483 by Nov 1
317,312 by Dec 1

 

[essentialsaltes]

August 28 update
216,645 by Oct 1 <-- interesting that this one decreased.
259,483 by Nov 1
317,312 by Dec 1

Sep 3 update, now with Jan 1 forecast
211,660 by Oct 1 <-- again a slight decrease
254,936 by Nov 1 <-- slight decrease
325,907 by Dec 1 <-- no more good news
410,451 by Jan 1

 

[GreatLakes4Ever]

Sep 3 update, now with Jan 1 forecast
211,660 by Oct 1 <-- again a slight decrease
254,936 by Nov 1 <-- slight decrease
325,907 by Dec 1 <-- no more good news
410,451 by Jan 1

The last line is worse. That’s a projection of over 3,000 deaths a day for the month of December.

 

[essentialsaltes]

The last line is worse. That’s a projection of over 3,000 deaths a day for the month of December.

It does look terrible, but I wonder about the model. It factors in seasonality ("We expect the daily death rate in the United States, because of seasonality and declining vigilance of the public, to reach nearly 3,000 a day in December.") but I wonder how much we know about that. Obviously there was the idea that the heat of summer would stop things.

So the question is whether that effect is real or not.

If it's real, then our not-very-good summer numbers are actually much better than they would have been, and that's what's reflected in the December forecast.
If the effect is not real, or not as strong, then the model may be way too grim for the colder parts of the year.

Time will tell. Other tidbits from the update notes:

• If a herd immunity strategy is pursued, namely no further government intervention is taken from now to January 1, then the death toll could increase to 620,000 by January 1. Compared to the reference scenario, this would be 210,000 more deaths from now to the end of the year.
  
• Increasing mask use remains an extraordinary opportunity for the US. Increasing mask use to the levels seen in Singapore would decrease the cumulative death toll to 288,000, or 122,000 lives saved compared to the reference scenario. This would be a 30% reduction in the deaths expected from now until the end of the year.
  

 

[essentialsaltes]

Sep 3 update, now with Jan 1 forecast
211,660 by Oct 1 <-- again a slight decrease
254,936 by Nov 1 <-- slight decrease
325,907 by Dec 1 <-- no more good news
410,451 by Jan 1

Sep 11 update

211,292 by Oct 1 <-- again a slight decrease
253,618 by Nov 1 <--again a slight decrease
327,368 by Dec 1 <--no more good news
415,090 by Jan 1 <--worse

I'm no Pollyanna, but again I wonder about the model's dependence on a seasonal variation. We'll see.

 

[Ken-1122]

The IHME COVID-19 estimates aren't necessarily the best, but they were one of the first and still most accessible models. As of the June 15th update, it predicts just over 200,000 deaths in the US by October 1st.

Other models exist, of course, but it's hard (for me) to see what they project out to October. but they generally agree that by the end of July we may be around 150K.

Back in March, Dr. Fauci was saying 100,000 to 200,000 might die.

"I would say between 100,000 and 200,000 cases," he said, correcting himself to say he meant deaths. "We're going to have millions of cases." But he added "I don't want to be held to that" because the pandemic is "such a moving target."

At that time, Trump said that a final US coronavirus death toll somewhere in the range of 100,000 to 200,000 people would indicate that his administration has “done a very good job.”

“So if we can hold that down, as we’re saying, to 100,000, it’s a horrible number, maybe even less, but to 100,000, so we have between 100 [thousand] and 200,000, we altogether have done a very good job.”

Trump seemed to be trying to draw a contrast between the scenario where the government literally did nothing and the path America is now on. With his reelection on the line, he set the stage to eventually proclaim that even 200,000 Americans deaths from the coronavirus would be proof that he saved lives.

It seems very likely to me that we will reach that 200,000 deaths by October. And sadly, it won't be over yet.

At the time of posting, the US is #7 in terms of both cases and deaths per capita (for countries over 1 million in population).

Is there data on how many people currently have the virus?

 

[essentialsaltes]

Is there data on how many people currently have the virus?

Worldometers purports to show active cases, but I honestly don't know what it's based on.

 

[essentialsaltes]

Sep 11 update

211,292 by Oct 1 <-- again a slight decrease
253,618 by Nov 1 <--again a slight decrease
327,368 by Dec 1 <--no more good news
415,090 by Jan 1 <--worse

Sep 18 update

204,296 by Oct 1 <-- again a decrease
229,587 by Nov 1 <-- quite a decrease
279,424 by Dec 1 <-- quite a decrease
378,321 by Jan 1 <-- quite a decrease

Keeping my fingers crossed that the seasonal effect was being exaggerated, and now the model is figuring that out.

ETA: Looks like the model is more delaying than cancelling the seasonal effect:

• This decline is driven by steeper than expected declines seen in deaths in several states. In our model, these declines push out further into January and February the expected seasonal surge, reducing the number of deaths expected by January 1. We expect over 3,000 deaths per day by the end of December.
  

 

[essentialsaltes]

Sep 18 update

204,296 by Oct 1 <-- again a decrease
229,587 by Nov 1 <-- quite a decrease
279,424 by Dec 1 <-- quite a decrease
378,321 by Jan 1 <-- quite a decrease

Sep 24 update

206,856 by Oct 1 <--a slight increase
236,077 by Nov 1 <--a slight increase
285,614 by Dec 1 <--a slight increase
371,509 by Jan 1 <-- a slight decrease

Looks like the model has settled down... not that that makes it any more accurate.

 

[essentialsaltes]

Sep 24 update

206,856 by Oct 1 <--a slight increase
236,077 by Nov 1 <--a slight increase
285,614 by Dec 1 <--a slight increase
371,509 by Jan 1 <-- a slight decrease

Oct 2 Update
206,491 by Oct 1 <--a tiny decrease [though in the past, this is still a projection as their accepted data lags about 10 days]
234,981 by Nov 1 <--a slight decrease
281,653 by Dec 1 <--a decrease
363,269 by Jan 1 <--a decrease

Current situation

• Daily cases remain steady around 40 thousand per day since late August (Figure 1).
  
  
• Daily deaths increased slightly to around 750 per day (Figure 2).
  
  
• Effective R calculated using data on cases, hospitalizations and deaths remains above 1 in North Dakota, Arizona, New Mexico, Colorado, Oklahoma, Arkansas, Missouri, Illinois, Michigan, New York, New Jersey, Pennsylvania, Virginia, Mississippi, Alabama, and Georgia (Figure 3).
  
  
• The daily death rate remains over 4 per million in six states: Arkansas, Mississippi, Florida, Georgia, South Carolina, and North Dakota (Figure 6).
  
  1. Projections
     
     
     • Daily deaths are expected to rise to over 2,900 a day by the end of December (Figure 13). We expect 360,000 cumulative deaths by January 1, a further 156,000 by the end of the year.
       
       
     • Expanding mask use to the level observed in Singapore can save 82,000 lives more than half of the deaths expected by the end of the year (Figure 12).
       

 

[hedrick]

Their model used to show deaths falling to zero soon. Now it shows then growing a lot. 2900 a day is worse than the first peak. I’m skeptical

 

[essentialsaltes]

2900 a day is worse than the first peak. I’m skeptical

I am as well. The model assumes COVID-19 is seasonal, but I find it hard to believe we know *how* seasonal it is. Already the model has been shifted to 'delay' the seasonal effect. I'm hoping that the seasonal effect will be smaller than the model shows.

A side-effect of the efforts in New Zealand was that COVID was suppressed, but so was the winter flu.

 

[hedrick]

Even if we have another peak like the spring peak, I think there has been enough learned that there will be fewer deaths. To get more you’d need a kind of peak that seems highly unlikely. Some kind of growth does seem likely.

 

[essentialsaltes]

Oct 2 Update
206,491 by Oct 1 <--a tiny decrease [though in the past, this is still a projection as their accepted data lags about 10 days]
234,981 by Nov 1 <--a slight decrease
281,653 by Dec 1 <--a decrease
363,269 by Jan 1 <--a decrease

Oct 9 Update

206,216 'observed' for Oct 1 <-- a tiny decrease
229,722 for Nov 1 <-- a slight decrease
265,238 for Dec 1 <-- a decent decrease
323,338 for Jan 1 <-- a whopping decrease
394,693 for Feb 1 <-- new estimate

The main forecast has daily deaths peaking out at 2,300 a day in January.
If we ease all mandates, the forecast has us rising toward 6,000 deaths a day at the end of January.
If everyone wore their durn masks, the forecast has us at about 1,400 deaths a day at the end of January.

 

[essentialsaltes]

206,216 'observed' for Oct 1 <-- a tiny decrease
229,722 for Nov 1 <-- a slight decrease
265,238 for Dec 1 <-- a decent decrease
323,338 for Jan 1 <-- a whopping decrease
394,693 for Feb 1 <-- new estimate

Oct 15 update
228,490 for Nov 1 <-- a tiny decrease
260,956 for Dec 1 <-- a decrease
316,935 for Jan 1 <-- a decrease
389,087 for Feb 1 <-- a decrease

While these numbers are coming down, they're not coming down much. My hope for a significantly smaller seasonal effect is getting dashed as case numbers are headed back up.

 

[essentialsaltes]

At that time, [March] Trump said that a final US coronavirus death toll somewhere in the range of 100,000 to 200,000 people would indicate that his administration has “done a very good job.”

“So if we can hold that down, as we’re saying, to 100,000, it’s a horrible number, maybe even less, but to 100,000, so we have between 100 [thousand] and 200,000, we altogether have done a very good job.”

Trump has not done a very good job on COVID-19.

 

[Mayzoo]

Trump has not done a very good job on COVID-19.

I gather it won't matter the final number is, he will still say he did an outstanding job. Better than anyone else, by far, could have done. Make room for the cheerleaders (who won't have on masks of course). Autographs to the left please, handshakes to the right. Donators, front and center.

 

[essentialsaltes]

Oct 15 update
228,490 for Nov 1 <-- a tiny decrease
260,956 for Dec 1 <-- a decrease
316,935 for Jan 1 <-- a decrease
389,087 for Feb 1 <-- a decrease

Oct 23 update
229,020 for Nov 1 <-- a tiny increase
262,918 for Dec 1 <-- a small increase
318,735 for Jan 1 <-- a small increase
385,611 for Feb 1 <-- a decrease

It's still absurd that the model assumes that North Dakota (for instance) is going to suddenly spring into action and lock things down before the end of October. Glad to see California is finally getting some credit for keeping things steadily bad instead of terrible and worsening. I think previous models have had CA with 40,000+ deaths by the end of the forecast and now it's at 30K, comparable to FL and TX.