From the statistics I've been seeing, of the people who get it, the mortality rate is closer to 10%. And the virus is so damned contagious that with no counter measures taken everyone would get it with say 18 months. And that would literally decimate the population. Not to mention all the devastating follow on covid-19 side effects some of which, it appears, never go away.
Do you have good insight into the effect on the economy of randomly killing a high percentage of type-2 diabetes? I know we like to ideate the poor into this state, but have you considered the effect on the economy, if the supersize-me generation at large in the economy at all levels suddenly took a king-hit?
We can roughly estimate this from the effects of the 1918 flu, which killed way more people, and killed healthy, working age people. The overall effect on GDP was tiny compared to the effect we're seeing from the lockdowns.
Two totally different economies. People will self-lockdown much more extensively today. Economies were much more local 100 years ago. GDP is also a meaningless metric when measured in currency, and is anyway antithesis of lockdown/war times. Sweden didn't lockdown but is still hit hard economically, and is more isolated today than surrounding countries.
Rule of thumb: Catching COVID doubles your chance of dying this year.
This rule lines up with age group and preexisting condition groups surprisingly well. It underestimates the death rate of older people, but is within a factor of two. It drastically over-estimates the death rate for children.
Corollary 1: If we do nothing, then hospitals and morgues will have to absorb a year worth of cases in about 2 months. Obviously, that’s a big problem.
Corollary 2: COVID will reduce life expectancies by much less than 6 months (actually, less than one month).
I suspect we could increase life expectancies in the US by much more than 6 months if we spent $3T on something other than the shutdown.
I think your rule of thumb is correct, but I don't see how this implies your corollaries.
I think Corollary 1 assumes 100% infection at T_0, which is clearly not true. It's debatable what the correct curve would look like, but even in the worst do-nothing case I don't think it could possibly be this steep.
I don't understand the logic for Corollary 2. Even if we assume Corollary 1 (which I'd argue we shouldn't) why would uniformly doubling the average death rate reduce life expectancy by less than 1 month? I can't say for sure that it wouldn't, but I don't think the math is obvious. Could you flesh this out?
Your final sentence seems possible, depending on what "much more" means, but I'd like to see more proof. Adding $3T would be equivalent to doubling current US health care spending.
It's nowhere near that. It's estimated that most cases aren't detected, and even of the ones who are, it varies between 15% and less than 1%. The actual fatality rate is estimated to be between 0.3% and 1% - varying wildly by age of course. That's still many many people and would be a huge tragedy which we should definitely avoid, but no one is considering numbers in that order or magnitude.
If the IFR was 10%, everybody would have been hiding in their homes and no one would be complaining about wearing masks.
10% sounds like a case fatality rate somewhere. That’s the percentage of people that see a doctor for it that die, and it is all over the place depending on which city, etc.
The WHO estimate is 0.5% mortality among people that get it. The CDC reports something strange, but if you multiply it out, they’re estimating 0.3% in the US.
> CDC and the Office of the Assistant Secretary for Preparedness and Responseexternal icon (ASPR) have developed five COVID-19 Pandemic Planning Scenarios that are designed to help inform decisions by public health officials who use mathematical modeling and by mathematical modelers throughout the federal government. Models developed using the data provided in the planning scenario tables can help evaluate the potential effects of different community mitigation strategies (e.g., social distancing). The planning scenarios may also be useful to hospital administrators in assessing resource needs and can be used in conjunction with the COVID-19Surge Tool.
They have five scenarios. These cover a range of cases. The CDC says this about the parameters in the scenarios:
> The parameters in the scenarios:
> Are estimates intended to support public health preparedness and planning.
> Are not predictions of the expected effects of COVID-19.
> Do not reflect the impact of any behavioral changes, social distancing, or other interventions.
They say this about the scenario that people are quoting:
> Scenario 5 represents a current best estimate about viral transmission and disease severity in the United States, with the same caveat: that the parameter values will change as more data become available.
How does anything you said contradict the claim that "[The CDC are]... estimating 0.3% [actually 0.26%] in the US"?
The commentator never claimed that it was an ironclad truth that could never change, just an estimate, which you confirm ("Scenario 5 represents a current best estimate...").
A well written overview. I think what is sometimes not read well by 'why don't we..' type comments is that the non-deterministic quality of this is quite strong. We can say your predicted risk as an 80 yo is higher than an 8 yo but we can't say much else. So, the 1% mortality and how it distributes over a cohort by age, sex, (ok more men) co-morbidities (ok so overweight, type-2 and pre-diabetic) .. but then we can't really say "oh, you're the high risk for subsequent heart attack" or "yep: you'll face life crippling mental acuity hits" or "we have to amputate now"
Ask anyone if they will take 1 in 100 risk and they will casually say yes. Ask again, if you present them with real risks of life changing outcome at shorter odds, they may reflect.
I wouldn't stand in front of a 1-in-100 russian-roulette machine personally.
Agreed. The penalty associated with the risk is important too. Probability theory and statistics can get us the risk fairly easily. But putting a number on the penalty associated with failure is a different problem altogether. If getting sick meant a 1:10 chance of getting a really bad headache for a week, I would take that chance. But if it was a 1:1000 chance of death I would probably not take the chance.
Sort of. It's important to point out that there are effects other than death. However, he's exaggerating like crazy, possibly by accident. Issue is that most of these complications happen to the same people. Young healthy people are nearly always fine. The virus is likely widespread in the military, and deaths/lasting impacts are minimal. Older, sicker, fatter people suffer an array of adverse consequences. In other words, the machine is highly biased depending on your circumstances. 1 in 100 is the worst case. Policy does not reflect this.
The problem is not the virus itself but instead it’s impact on the availability of medical care.
If you wish to talk about statistics the death rate associated with the virus is declining in the US even though the numbers of infection are surging. That is because medical services are better learning what treatments are effective. The death rate will continue to decline inversely to the growth of infection rate only so long as medical treatment is available.
The real problems occur when medicine is over whelmed by the case volume, particularly when hospital beds fill up. When hospitals lack the space to provide appropriate care and triage is when the associated death rate shoots up from declining below 0.1% to over 10% (greater than 100x). Quality and availability of care impacts all medicine. For example appendicitis is an easily treatable and relatively common medical problem in the US, but without available care the patient will die of a blood infection. If that example is not properly triaged due to lost medical capacity then it shifts from a minor concern to a potentially fatal problem.
>That is because medical services are better learning what treatments are effective.
Not sure that's true. The average age of infected is much lower than it was a couple months ago. It's likely that more people who aren't at risk of serious complications are being infected now and we're testing much more than we had been -- including people w/o symptoms. A couple months ago most people who weren't considered at-risk couldn't get a test even if they had symptoms.
Before you panic, note that this post is full of incorrectly computed conditional probabilities.
It claims 19% of the whole population will be hospitalized at 100% infection rate. This seems high, but let’s go with it.
It then cites a study saying 18% of hospitalized patients’ hearts showed evidence of scarring.
Later, it computes:
> 62,358,000 hospitalized.
59,076,000 people with permanent heart damage.
Multiplying the hospitalized count by 18% gives 11,224,440, or 3.7% of the population. I’m assuming that no people that avoid the hospital will have heart damage. The post assumes there is no correlation between disease severity and hospitalization.
The true rate is probably somewhere in between, and probably closer to 4% than 18%.
4% of patients having heart permanent damage is still 13,128,000 people. Qualitatively, those numbers are equally hard to grasp, so why would it be any less terrifying?
Heart scarring (what the study looked for) is extremely common. For instance, flossing your teeth commonly leads to an opportunistic infection that leads to heart tissue scarring.
If most of the 13M are expected to be impacted in some way by the heart damage, then the number is terrifying, but the evidence doesn’t back that up.
Also, there’s a huge difference between 1/5 people need heart surgery, and the more realistic 60% * 18% * 19%, which is closer to 1/50.
To help put that in perspective: ~1/2 of Americans suffer from heart disease.
Can you provide a link for the claim that “flossing your teeth commonly leads to [...] heart tissue scarring”?
After some googling, it appears that the opposite is true. If you don’t brush and floss your teeth, you increase your chances of gingivitis, which can become a systemic infection.
Just for information (i've said this multiple time on HN, i'm rambling), ive heard from my cousin (urgentist at Colmar) that he had a 26 yo young woman with lung fibrosis who wasn't hospitalized while she was infected.
That said, you're totally right, tissue scarring seems to be closer to 5% than 20%. And the IFR in France was predicted around april to be close to .7%, .8% for males. So far the prediction were right, so you should table for something around that in the US with a younger population, but more diabetes and obesity. Now that new treatments are found for the worst cases, this could be lowered further imo.
Hearth scarring is not really a problem if it does not cause arythmia. Lung scarring, however, can shove ten year of your life expectancy (and not the worst ten years, it works a bit like a thyroid ablation were you just age faster).
> Sweden hasn't even been worse than the 2000 Flu season yet.
Can you post the link to the Swedish statistics? I think you're counting flu and covid death differently.
For flu I think you're using a mix of all cause mortality and statistical modelling. For covid I think you're using either "died after being tested positive for covid" or "covid was mentioned as a cause on the death certificate".
What people are trying hard to ignore is that we are seeing these numbers _despite_ shutting down the country and everyone using masks and washing hands and avoiding contact.
With a business as usual attitude we would see death rates 100 times higher.
I just realized that this has similarities to the September 11th 2001 (9/11) terrorist attack.
A relatively small investment resulted in the USA invading Iraq, which had absolutely nothing to do with it, killing hundreds of thousands of civilians, indebting the US population for a few trillion dollars, and generally causing a ridiculous over-reaction in just about every way possible.
Has anyone seen any comparisons yet? If not I can't imagine it won't last. Both are astonishing failures of civil leadership and governance.
34 comments
[ 201 ms ] story [ 894 ms ] threadThis rule lines up with age group and preexisting condition groups surprisingly well. It underestimates the death rate of older people, but is within a factor of two. It drastically over-estimates the death rate for children.
Corollary 1: If we do nothing, then hospitals and morgues will have to absorb a year worth of cases in about 2 months. Obviously, that’s a big problem.
Corollary 2: COVID will reduce life expectancies by much less than 6 months (actually, less than one month).
I suspect we could increase life expectancies in the US by much more than 6 months if we spent $3T on something other than the shutdown.
I think Corollary 1 assumes 100% infection at T_0, which is clearly not true. It's debatable what the correct curve would look like, but even in the worst do-nothing case I don't think it could possibly be this steep.
I don't understand the logic for Corollary 2. Even if we assume Corollary 1 (which I'd argue we shouldn't) why would uniformly doubling the average death rate reduce life expectancy by less than 1 month? I can't say for sure that it wouldn't, but I don't think the math is obvious. Could you flesh this out?
Your final sentence seems possible, depending on what "much more" means, but I'd like to see more proof. Adding $3T would be equivalent to doubling current US health care spending.
If the IFR was 10%, everybody would have been hiding in their homes and no one would be complaining about wearing masks.
The WHO estimate is 0.5% mortality among people that get it. The CDC reports something strange, but if you multiply it out, they’re estimating 0.3% in the US.
This is being repeated a lot. I think people have misunderstood what CDC are saying.
https://www.cdc.gov/coronavirus/2019-ncov/hcp/planning-scena...
> CDC and the Office of the Assistant Secretary for Preparedness and Responseexternal icon (ASPR) have developed five COVID-19 Pandemic Planning Scenarios that are designed to help inform decisions by public health officials who use mathematical modeling and by mathematical modelers throughout the federal government. Models developed using the data provided in the planning scenario tables can help evaluate the potential effects of different community mitigation strategies (e.g., social distancing). The planning scenarios may also be useful to hospital administrators in assessing resource needs and can be used in conjunction with the COVID-19Surge Tool.
They have five scenarios. These cover a range of cases. The CDC says this about the parameters in the scenarios:
> The parameters in the scenarios:
> Are estimates intended to support public health preparedness and planning.
> Are not predictions of the expected effects of COVID-19.
> Do not reflect the impact of any behavioral changes, social distancing, or other interventions.
They say this about the scenario that people are quoting:
> Scenario 5 represents a current best estimate about viral transmission and disease severity in the United States, with the same caveat: that the parameter values will change as more data become available.
The commentator never claimed that it was an ironclad truth that could never change, just an estimate, which you confirm ("Scenario 5 represents a current best estimate...").
Ask anyone if they will take 1 in 100 risk and they will casually say yes. Ask again, if you present them with real risks of life changing outcome at shorter odds, they may reflect.
I wouldn't stand in front of a 1-in-100 russian-roulette machine personally.
If you wish to talk about statistics the death rate associated with the virus is declining in the US even though the numbers of infection are surging. That is because medical services are better learning what treatments are effective. The death rate will continue to decline inversely to the growth of infection rate only so long as medical treatment is available.
The real problems occur when medicine is over whelmed by the case volume, particularly when hospital beds fill up. When hospitals lack the space to provide appropriate care and triage is when the associated death rate shoots up from declining below 0.1% to over 10% (greater than 100x). Quality and availability of care impacts all medicine. For example appendicitis is an easily treatable and relatively common medical problem in the US, but without available care the patient will die of a blood infection. If that example is not properly triaged due to lost medical capacity then it shifts from a minor concern to a potentially fatal problem.
Not sure that's true. The average age of infected is much lower than it was a couple months ago. It's likely that more people who aren't at risk of serious complications are being infected now and we're testing much more than we had been -- including people w/o symptoms. A couple months ago most people who weren't considered at-risk couldn't get a test even if they had symptoms.
I’ve heard it hypothesized but never supported. I agree it’s plausible but we need evidence before confidence.
>In Florida, the median age of new COVID-19 cases fell from 65 in March to 35 in June.
It claims 19% of the whole population will be hospitalized at 100% infection rate. This seems high, but let’s go with it.
It then cites a study saying 18% of hospitalized patients’ hearts showed evidence of scarring.
Later, it computes:
> 62,358,000 hospitalized. 59,076,000 people with permanent heart damage.
Multiplying the hospitalized count by 18% gives 11,224,440, or 3.7% of the population. I’m assuming that no people that avoid the hospital will have heart damage. The post assumes there is no correlation between disease severity and hospitalization.
The true rate is probably somewhere in between, and probably closer to 4% than 18%.
If most of the 13M are expected to be impacted in some way by the heart damage, then the number is terrifying, but the evidence doesn’t back that up.
Also, there’s a huge difference between 1/5 people need heart surgery, and the more realistic 60% * 18% * 19%, which is closer to 1/50.
To help put that in perspective: ~1/2 of Americans suffer from heart disease.
After some googling, it appears that the opposite is true. If you don’t brush and floss your teeth, you increase your chances of gingivitis, which can become a systemic infection.
That said, you're totally right, tissue scarring seems to be closer to 5% than 20%. And the IFR in France was predicted around april to be close to .7%, .8% for males. So far the prediction were right, so you should table for something around that in the US with a younger population, but more diabetes and obesity. Now that new treatments are found for the worst cases, this could be lowered further imo.
Hearth scarring is not really a problem if it does not cause arythmia. Lung scarring, however, can shove ten year of your life expectancy (and not the worst ten years, it works a bit like a thyroid ablation were you just age faster).
https://www.reuters.com/article/us-health-coronavirus-sweden...
And I don't recall the 2000 Flu season shutting down Sweden/USA/World
Quora might be a good jumping point when doing research, but like 4chan and other hive minds you still have to think.
Can you post the link to the Swedish statistics? I think you're counting flu and covid death differently.
For flu I think you're using a mix of all cause mortality and statistical modelling. For covid I think you're using either "died after being tested positive for covid" or "covid was mentioned as a cause on the death certificate".
This also talks about May -
https://emanuelkarlsten.se/swedens-two-corona-months-are-not...
With a business as usual attitude we would see death rates 100 times higher.
A relatively small investment resulted in the USA invading Iraq, which had absolutely nothing to do with it, killing hundreds of thousands of civilians, indebting the US population for a few trillion dollars, and generally causing a ridiculous over-reaction in just about every way possible.
Has anyone seen any comparisons yet? If not I can't imagine it won't last. Both are astonishing failures of civil leadership and governance.