Washington Health Metrics Forecasts 40,000 to 160,000 COVID-19 Deaths for USA

The Institute for Health Metrics (IHME) at the University of Washington has modeled forecasts for each state and the USA.

The forecasts and models were run two days ago. They forecasted deaths, ventilator needs, ICU beds and hospital beds and shortages.

The US will need between 130,000 and 480,000 hospital beds. The most likely is a US peak need for 232,000 beds.

The number of cases will be about 400,000 to 1.5 million.

They forecast
USA – 40,000 to 160,000 Death range. 82,000 deaths main projection.
New York – 5,000 to 27,000 Death range. 10,240 deaths main projection
California – 800 to 17,000 death range. 6,100 deaths main projection.
Texas – 4,200 to 7,600 death range. 5,800 deaths main projection.
New Jersey – 1,700 to 11,000 Death range. 4,100 deaths main projection
Michigan – 1,400 to 11,000 death range. 4,060 deaths main projection.
Florida – 600 to 9000 death range. 3,340 deaths main projection.
Massachusetts – 400 to 7,200 death range. 2,230 deaths main projection.
Illinois – 500 to 5,900 death range. 2,450 deaths main projection.

They project the US will need about 19,000 invasive ventilators.
There will be a need for 35,000 ICU beds with about 15,000 new ICU beds needed.

The USA is predicted to peak April 14, 2020.
New York is predicted to peak on April 6, 2020.

This study used data on confirmed COVID-19 deaths by day from WHO websites and local and national governments; data on hospital capacity and utilization for US states; and observed COVID-19 utilization data from select locations to develop a statistical model forecasting deaths and hospital utilization against capacity by state for the US over the next 4 months.

SOURCES : IHME COVID-19 health service utilization forecasting team. Forecasting COVID-19 impact on hospital bed-days, ICU-days, ventilator days and deaths by US state in the next 4 months. MedRxiv. 26 March 2020.
Summary Written by Brian Wang, Nextbigfuture.com

86 thoughts on “Washington Health Metrics Forecasts 40,000 to 160,000 COVID-19 Deaths for USA”

  1. what I find unnecessarily rude is someone claiming this isn’t a big deal and taking that misinformation online, risking people’s lives.

  2. *we* should teach our kids to eat healthy, but then *they* (the government) show up in the next line.

    Which is exactly what I meant in my original post. Grow up and take responsibility for your own life!

  3. I’ll grant you swine flu
    AIDS was 40 years ago. Hardly “the last 2 decades”.

    And your other examples depends on what you mean by “we”.

  4. The whole world is aligned behind the (approx.) spherical Earth and the existence of gravity.

    It isn’t ALWAYS flawed.

  5. Is there literally ANY person in the western world who hasn’t been told, time and time again to:
    — Eat more fruit and vegetables
    — Eat less sugar
    –Drink less softdrink
    — Don’t drink too much alcohol
    — Don’t eat to excess
    — Lay off the deep fried junk food
    — Do regular exercise
    — Don’t smoke
    –Don’t do drugs
    — Maintain a healthy weight
    — Etc. etc

    But when they ignore all this and end up out of condition, they come up with “Nobody told us. We though cigarettes were full of vitamins and stuff. I thought deep fried Mars bars were the basis of a healthy diet.”

    Well, we just don’t believe them. Stop coming up with this rubbish because nobody is buying it.

  6. And they had it under control in china because they locked stuff down. That is the point.

    And what utter nonsense. The economy being hurt is noty going to kill more us citizens then died in ww2. Stop believing insane nonsense like that. Do you REALLY think the economy in the us crashing would kill millions of us citizens? Because if so I have a bridge to sell you.

  7. Yeah because the better choice is to let it go wild, killing millions instead of hundreds of thousands. Overwhelming our medical facilities will cause immensely more deaths. That is why we slow it. Sigh.

  8. See that 2 million dead figure? That is arguably on the low end, but more importantly it would cause immense damage to our economy.

    Heart attacks kill 1 out of a thousand of us per year. This would kill about 10X as many-at a minimum. and dragging it out IS THE POINT. It prevents our hospitals from being as overwhelmed and saves a ton of lives.

  9. Might be pointing out that over a dozen states arent locked down, and florida still has it beaches open. The forecasts are garbage because theyre based on us doing everything perfectly.

  10. Gosh if only we taught kids how to eat healthier they might grow up knowing how to eat better….you know like the stuff Trump got rid of that were put in under Obama’s administration.

    Or if the poor could actually afford to eat healthier. Maybe they should stop being poor right?

    Theres a LOT more to this then people choosing to be unhealthy.

  11. Looks like a simple plug in the excel as a cvs baysian regression R based model running a sim for maybe 1 million.

  12. yes.

    But it helps that I wrote … “words” a few years ago. It was winter, and I was bored… so I figured out how to (literally) cut-and-paste a bunch of fairly well regarded online dictionaries. Stuffed all the hideously unformatted entries into a giant PERL <DATA> structure, parsed that (only takes about 5 seconds) and built a search list. Deduplicated, and case-insensitive.  

    So, searching for /podes$/ in RegEX ‘language’ is no more difficult than typing ‘words’ on a command line and ‘podes$’ on the prompt. Gives them all as a list, as shown. 

    So, while bored, … i have bored-genius tools. 

    LOL … leaving enough time to bold and italic things!

    ⋅-=≡ GoatGuy ✓ ≡=-⋅

  13. We were being told by the WHO and China they had it under control and it was just a flu then.

    Also crashing the worlds economy may kill more people than the virus.

  14. No it isn’t.

    Cultures are dying everywhere while the people in big cities around the world are becoming clones.

    We need people who think and see things differently and act and believe differently.

    Globalism has brought us ruin time after time except for those who hold the reigns of power and wealth around the world.

    Let Globalism die.

  15. Jello you realize this is if we do what we are doing and keep doing it right?

    Meaning if we hadnt there could have been 2 million dead by the end of the year and the country has fallen apart.


    Why is this so hard to understand.

    We have less hospital beds per person than italy or Iran.Z

    BTW. Some of us have never stopped working.

  16. One aspect of doubling time seems to be not getting enough attention. When the new case rate is at a rate where the doubling time is longer than hospital treatment, hospital loading will go down. You won’t be battling desperately against a rising tide, you’ll be managing the patients that you have, although there are always more patients. For COVID-19, hospitalization time seems to be around ten days on average.

    That is a daily new case rate of 7%. Italy has been hovering agonizingly around this level for several days. I hope they can sustain it, and gain some relief.

    When the rate is at this level, with a strong community testing program in place, some of the lockdown restrictions might be loosened.

  17. Nope, I didn’t say anything about that. I’m actually fairly old myself.

    However, if all those people that messed up their own health hadn’t, the hospitals would have far more resource available to take care of those old people.

  18. If we could convince them to only eat virtual bats, pangolins and racoons and other wildlife, that could be a real winner.

  19. So the old peoples mistake was that they grew old?
    Obviously if they had snorted their lines, stacked on the pounds and got drunk everyday, they would have offed themselves 10 years ago and they wouldn’t have a worry now.

  20. If all the ICUs get saturated even the strong and young that get involved in accidents and will need surgery might die. Not to mention the fact that we don’t know if the virus causes long term issue: what happens if everyone gets infected 1% dies for the infection and 50% of the recovered have permanent damage to the immune system or develop cancer in the next few years?

  21. No, it will not. For that to happen 50-60% of people will have to contract the disease and survived it. Looks like Trump agrees now, since he extended social distancing until end of April.

  22. The majority of people in New York (I live in Albany) will not have been exposed in that time frame. There will not be a herd immunity by then, and elderly people like my parents and grandmother will still be at risk (her nursing home is in lockdown). As for the rest of the country that has not been hit hard by this, if they allow people to start socializing again they will spread the disease. The two week time frame is not when the disease will have run its course, it will be simply to flatten the curve so our medical services do not get overwhelmed. By all means though, kill off the people over 70 (sarcasm). At least it will make it easier to fix Social Security without the political drama and price.

  23. Why would the time for containment be over ?. And it would take a year to create herd immunity, do you realize how many would die ?, millions ?.

    By the way, do we even know of a drug that works with this virus ?. It certainly isn’t hydroxychloroquine.

  24. Just remember, in the Southern Hemisphere it is now Fall. Fall, Winter and Spring are the seasons where the virus should spread and it’s only Summer that is predicted to put a brake on it.

    Note, they also seem to be trying to ignore it which will only make it much worse.

  25. Breaking news: Researchers in Southern New Jersey discover that liberal, external application of Vagisil halts Covid19.

  26. Just heard from the news on France that patients in Italy are developing secondary aggravation of syndromes after 2-3 weeks of moderate/indolent disease. This could be due (I hope not) to the rise in anti-coronavirus antibody titers, mainly IgG. Deleterious role of some antibodies was reported for SARS, either via cross-reactivity with bronchial epithelial cells, or (less probably) via facilitated infection linked to anti-spike (viral protein) specificities. This makes it potentially risky to administer Immunoglobulins from patients who have recovered from coronaV.

    YS LIN et al. Antibody to severe acute respiratory syndrome (SARS)-associated coronavirus spike protein domain 2 cross-reacts with lung epithelial cells and causes cytotoxicity
    Clin Exp Immunol. 2005 Sep; 141(3): 500–508.
    doi: 10.1111/j.1365-2249.2005.02864.x
    PMID: 16045740

  27. Yes, we would all be dead from social insurrection or impoverished by bankruptcies. Good thinking.

  28. If in 2 weeks the parts of the US that are under control (not the East Coast Cities) begin a measured re-start, adjusting for impact over time, the world will not collapse. That is exactly what has been proposed.

  29. If President Trump gets his way and we start going back to work in the next week or two weeks that death toll will be closer to 160k than 40k. Until the population gets a large percentage of people that have had COVID and developed an immunity, or a vaccine, this disease could continue to blow up into exponential numbers at any time. Possibly he needs to get it himself so it effects him personally before he will start adjusting his priorities.

  30. I don’t think there will be any win there.
    1.The medical care in this illness costs far more than we would be paying.
    2.Influenza will kill far less than it usually kills. We could even end up with much less dead. I suspect Influenza will kill about 1/2 of what it would have otherwise killed. Coronavirus will probably fill that in. Possibly more but maybe no more than the bad flu season a couple years ago. Of course things could turn ugly.
    3.The elderly are in hiding. The smart ones that could also afford to stock up. New York lost 147 people between 18-64 and 369 older people. So you have to consider the loss of those working age people. That means the loss of breadwinners/tax payers and likely government assistance for the spouse and possibly children. And in the rest of the nation, people can mostly better isolate and store more food and supplies. That suggests this could be even more skewed to younger people getting this much more. So even with the lower probability of death, the high fraction of cases of the younger people means deaths could be almost 50-50 under/over 65 at New York infection levels.
    4.If this drags on many old people will cut their calorie intake to make their food last longer rather than risking their lives going to a grocery. This near fasting is very good for removing senescent cells…which can very easily lengthen their lives.

  31. Should have listened to the experts during the security briefs and locked down in late January. But hey, ‘we have it under control, we have beautiful control over it’.

  32. How much will this lower social security payments, medicare payments, spending by the now dead? Let put some dollar amount on this big cost reduction.

  33. I suppose if one is profoundly pessimistic, it is possible death rates could jump to 10% or worse in some countries because of infections people already have in some poor African countries, India and other poor countries around the world…on top of everyone being exposed. Or there could be a deadly mutation…perhaps even different enough to cause one to be infected twice. But that is bordering on the absurd. And I think this a 1 in 10,000 possibility or lower.

  34. Anything is possible between 1/4 million (pie in the sky, best case…effective vaccine in a month and ability to quickly make 7B+ doses) and 280 million (worst case scenario…nearly everyone exposed on the planet fairly quickly). And there are so many factors, it is near impossible to predict. Policy choices of governments, and treatments can radically shift the outcome.
    I think the 280m result is highly improbable. It looks like there might be a treatment: https://www.theguardian.com/world/2020/mar/18/japanese-flu-drug-clearly-effective-in-treating-coronavirus-says-china
    The trick is getting people this treatment very early in their bout with the disease. Plasma from people who fought it off can help save the very sick:
    This one…two punch could reduce the fatality rate to less than 0.5% from 3.5-4.5%. If we still have 100% exposure that could still be 7m. Fast tests and aggressive testing might result in quicker treatment and slower spread leaving room for vaccines to prevent many infections. Maybe 2/3 of the World will never get it, if enough people are vaccinated. So perhaps it could be stopped at 2.3m. And it could be lower if everyone on every airline was tested using 5-15 minute tests and assuming some countries are successful at stopping the spread in their counties.

  35. What I hope comes out of this (worldwide), is that people will start taking responsibility for _their own_ health, instead of seeking medical help when the body they messed up over decades starts failing.
    From all statistics it shows clearly that unless you’re old, obese or COPD patient, you have little to worry about. Statistically your healthy body’s immune system will be able to fight off this virus with relative ease.

    But, if we take for instance the U.S.A, where almost half the population is obese, chronically stressed and under-rested, it becomes clear where the high mortality rates are going to be.

  36. Hot and dry probably does help a little. Australia and parts of Africa seem to have a low trajectory. Though it is probably still too soon to say. Hot and humid probably does not work. I think Australia and Canada had a similar number of cases early on. And both had their first cases Jan 24 and a very similar story of early cases. But Canada has 61 dead and Australia has 16.

  37. spare me the noble medical professionals crap.
    of course i am talking about automating the process more like remote surgeries. I am sure that they have made great strides in minimizing the amount of supervision and human contact
    btw – this is an entertainment blog, not a serious place of intellectual discourse

  38. plah-tih-PODE-s, or plah-TIP-oh-dees?

    I believe the “from Greek” is supposed to follow Greek pronunciation penchants. Therefore the second one is ‘right’?

    Like hyperbole … rhapsodes … or maybe not. Here is my 365,000 word dictionary (dataset) list… looking for /podes$/

    podes$    : actinopodes   1
    podes$    : antipodes   2
    podes$    : apodes    3
    podes$    : bladderpodes   4
    podes$    : epodes    5
    podes$    : euronithopodes  6
    podes$    : fringepodes   7
    podes$    : hemipodes   8
    podes$    : megapodes   9
    podes$    : monopodes   10
    podes$    : octopodes   11
    podes$    : sicklepodes   12
    podes$    : spodes    13

    Harrumph. I dunno.
    Back in my hole.

    ⋅-=≡ GoatGuy ✓ ≡=-⋅

  39. If it hits a major city in Texas, they are screwed. Literally no support when you look at the staffed hospital beds per capita. Though my guess is that Texas will be OK since they have more dispersed living conditions with open air entry to their residences, lots of sunshine. With stay at home orders in place, its the enclosed entry residences that will get hit the hardest (ie. apartment buildings in colder places). Since people will go outside to avoid complete cabin fever and do laundry in apartment environments w/o in unit laundry.

  40. That’s like pointing out that traffic accidents kill less people than heart disease so you will just walk blindly into traffic from now on.

  41. WHO has 3.4% projected lethality for the Virus. That is the aggregate from different countries. Some have worse helathcare system, some have a better one. Lets say that U.S. can keep lethality at 1% as long as its healthcare system is not saturated that would mean 100 deaths for every 10,000 people.
    The society of critical care medicine indicates that more than 5 millions patients are put in intensive care annually but if no places in ICU are available is reasonable to assume that they will die too (obviously if the saturation of the healthcare system lasts 6 months and not a year the deaths will be 2.5 millions out of the 5 millons).

  42. Whats your method? Did you go Bayesian? Lets see the curve. I recommend a R package called Rethinking. Maybe a Markov Chain would work best if you want to look at spread between cities. Take a CVS file of the numbers (index in the code), run a sim off a regression model (maybe sim 200-300 mil), apply your priors (a 0-100 scale for a percentile range should work given the type of data) and then use the actual data with the sim data to get a simulated data set applying the actual numbers. You can even add in a multiplier to look at different scenarios for inadequate testing by altering the priors.

  43. Seeing the outbreak in Brasil, it doesn’t seem to be seasonal.

    We will have to rely in massive testing (as Korea successfully does) to prevent second waves without needing further nation-wide quarantines.

  44. Dawg… LOL… I’m working from home right now (and that might be permanent, YAY) but… that’s a little harsh. xD Unless you mean that with comical cynicism, which I COMPLETELY appreciate. Because I adore joking about dire situations. However, putting a halt to cultural mingling after this, is more dangerous than the virus.

  45. Agreed. In northern Harris County. However, even without the number of tested infections ballooning, it doesn’t mean we couldn’t seen a sudden related death uptick. That would suck HARD. I’m working from home (well, not currently; CURRENTLY I’m obliterating my immune system with sake), so I’m not leaving my home, but it isn’t impossible for a death rate explosion. Though, considering how I see people practicing social distance in my area, I see that as unlikely unless folks suddenly get too comfortable.

  46. The USA will peak in two weeks?? You will spend even more time figuring out why you so fail to make proper forecasts than figuring out how you it was ok to not prepare in any way at all…

  47. You are no doctor. You probably don’t even own a dictionary.

    A pandemic is an epidemic that goes beyond a single location.

    Swine Flu
    Viral hemorrhagic fevers

  48. The study assumes a steady state with no treatment developments. No hydroxychloroquine + antibiotic, for example. None of the other 70 drugs currently being trialed. Just social isolation dragging things out to flatline to…infinity or at least herd immunity sometime off the chart in August. That’s pretty pessimistic…unless they know something about cure suppression we don’t: https://www.opednews.com/articles/Why-France-is-hiding-a-che-by-Pepe-Escobar-Chloroquine_France_Frances-Fox-Piven_Tsunami-200328-903.html

  49. No such word as virii. The plural of virus is viruses. I didn’t know this, had to look it up. Sorry if you were just trying to be funny.

  50. We’ve had a few in the last two decades.

    Errr… remind me. I can’t think of any.

    We’ve had a few epidemics. But that’s not the same as a pandemic.

    The very “pan” part is the critical difference. That’s the difference between “a lot of people are sick in hospital” and “a lot of people are sick but not in hospital because the hospital is full”.

  51. An ICU room isn’t a room. We can whip up new rooms fairly easily. We can take over empty hotels. Put up prefab buildings. Erect tent cities in the warm areas.
    That’s not the problem.

    An ICU is a room, filled with expensive, uncommon equipment, and staffed with ICU trained medical staff. Those last two points are the problem.

  52. It’s only twice as bad *because* of the precautions we’re taking. If everybody goes back to work today, you see ~60% of the US getting this and millions dead.

  53. If true, wouldn’t that make it roughly twice as bad as the flu…not 10 times. In other words, get back to work everyone.

  54. yet another mathematically challenged person who doesn’t understand an exponential. here’s a clue – if you do nothing, it spreads everywhere and you get 2 million dead, plus or minus a million.

  55. Which equates to two persons dying out of every 10,000 population, correct? I hope we don’t completely destroy our country for that number. Texas currently has 30 deaths FYI.

  56. more likely that the tech to do a downloadable brain scan and then just upload you into a new model before we cure all potential for disease

  57. can’t we just get more of those ICU rooms in a bag? – like those preemie-containment-units on a gurney. Then we could rack’m and stack’m.
    Also, wondering about the whole – residence to ER to ICU logistics thing. Do we have the ambulances, parameds, and attendant staff (in right place and proximity from the actual cases to the actual facilities) to undertake this idealized pandemic-managing (or not quite) projection.
    Also, is there anywhere with a ‘complete’ stacked super graph including future projections – that is #tested, #presumptive/confirmed, #hospitalized, #ICU, #ventilator, #dead, #released – all stacked from US-Day 0 to +120days predict – since it would be good to see the trends – when all practicable # are tested, when #presum flattens, when #released/ recovered accelerates up, etc. I know that we have the statistical tech to do it (and then add a variable slider so that the graph changes with milestone/ policy/ behaviour changes (back to work, all vulnerable tested, etc.) Not that I am looking to play CoronaSimCity, just sayin.

  58. what happens when the next pandemic hits? We’ve had a few in the last two decades. This is undoubtedly the worst what if the next one is worse? You can’t keep quarantining people. It’s not going to work. We need cures.

  59. I think from now on the rest of the world should only travel to China using telepresence robots to avoid future Chinese pandemics …

  60. I think this will be something like a bad flu season, maybe a little worse. It sucks we will see people die, but…life. Could be like the Asian Flu from the way back machine when 100k died.

  61. we are going to be dealing with this virus for years if they keep trying to slow it down every time and don’t have a vaccine…

  62. I think France has 200,000 cases based on the number of critically ill…. they are just under reporting…

  63. I’m disappointed that there isn’t more precision in those projections.  How clumsy! Why are we denied 6 significant figure! Every dead person certainly is vitally important.  


    I’m tired.
    Its Saturday. 
    I’ve had 4 harrowing weeks of statistics and stupid math from Mathematical Idiots.

    When your variance is +100% / –50% for all numbers, it hardly is competent to include 4 significant digits for the average.  10,240 … is 4 sigfigs.  Yet, its range is said to be 5,000 to 20,000. (or with even more useless precision).

    I’ll fix it:

    America – – 40,000 to 160,000 Death range. 80 k deaths main projection.
    New York – — 5,000 to 27,000 Death range. 10 k deaths 
    California – – – – 800 to 17,000 death range. 4 k deaths 
    Texas – – – – – – 4,200 to 7,600 death range. 5 k deaths 
    New Jerse- – – 1,700 to 11,000 Death range. 4 k deaths 
    Michiga- – – – – 1,400 to 11,000 death range. 4 k deaths 
    Florida – – – – – – – 600 to 9,000 death range. 2½ k deaths 
    Massachusetts – 400 to 7,200 death range. 2 k deaths 
    Illinois – – – – – – – – 500 to 5,900 death range. 2½ k deaths 

    See? Less precision, better demonstration of principle.

    ⋅-=≡ GoatGuy ✓ ≡=-⋅

  64. That’s the thing about exponential calculations.
    Change your assumptions by only a tiny amount and you can get any number you like/fear.

    It’s the same as stock price calculations for technology startups.

    Once you’ve got an exponentially growing value, you can plug in numbers to a calculator and get 100 thousand, or 100 million with equally “reasonable” boundary conditions. Be slightly “unreasonable” and you get the values plunging to near zero or shooting up to billions.

  65. It would be useful to give the numbers for how many beds, ventilators etc. the USA actually has as well as how many they need.
    Otherwise you are just giving a number without context.

    Fortunately the fine print in some of the graphics gives you the shortfall, which is not really THAT bad. Well, except in New York.

  66. This is wave one we are experiencing now. If we loosen the shelter in place too soon we’ll see a second wave at that time and if we don’t, there probably be one next fall. This assumes the virus is seasonal but if it isn’t we’ll see a second wave regardless when we ease up on the shelter in place unless we do such a good job that we can get to zero cases (which I doubt.)

  67. Expect 100 fold more for the rest of the world. I think we could be easily looking at 20 million dead.

  68. I think we will have more cases and less deaths than they project in that time window. Though they obviously are not going out on a limb with their very wide range projection.
    If nothing else, blood plasma injections from survivors will save many lives. I am not a huge fan of this though, as we each carry many latent infections which could easily spread that way. Better if we can come up with treatments from more sterile sources and vaccines.
    I also see no reason to think August 4th is near the end of it. This thing could drag on for several months past that date still at a high level. I think we need to get serious about vaccine testing. Accelerate it.
    Even if we have to accept healthy volunteers to attempt infection after being given the vaccine.

Comments are closed.