The CDC article describes five scenarios for planning purposes. Two assume 1% overall fatality, two have 0.2%, and one has 0.4%. They explicitly say they "[a]re not predictions of the expected...
The CDC article describes five scenarios for planning purposes. Two assume 1% overall fatality, two have 0.2%, and one has 0.4%. They explicitly say they "[a]re not predictions of the expected effects of COVID-19."
I don't know, but that's what's listed on the web page. The numerator is deaths and the denominator is infections with symptoms, so it seems like it couldn't be a lot higher without either missing...
I don't know, but that's what's listed on the web page. The numerator is deaths and the denominator is infections with symptoms, so it seems like it couldn't be a lot higher without either missing a lot of deaths or overcounting infections by a lot.
Though, this is specific to the US. In an overwhelmed healthcare system or a country with more elderly, the death rate could be higher.
Apparently the 0.4% scenario is based on their current best estimates. The other scenarios are supposed to be best and worst cases.
In New York things got pretty bad, but it's also the case that a lot of extra capacity got set up that ultimately wasn't used. I don't know what the results were when it comes to the fatality rate.
In New York things got pretty bad, but it's also the case that a lot of extra capacity got set up that ultimately wasn't used. I don't know what the results were when it comes to the fatality rate.
Looks like the worst of it has been in New York and Seattle. Most other cities aren’t quite so bleak. I think it’s a capacity thing. A cousin of mine in NYC ended up passing away of a mild heart...
Looks like the worst of it has been in New York and Seattle. Most other cities aren’t quite so bleak. I think it’s a capacity thing. A cousin of mine in NYC ended up passing away of a mild heart attack that would have been survivable in normal times. But the whole system was so slammed nobody could get to him.
Mortality rate isn't really something for the CDC to plan for, but rather is something for economic organisations to plan for (which is why that 'not a prediction' disclaimer that was pointed out...
Mortality rate isn't really something for the CDC to plan for, but rather is something for economic organisations to plan for (which is why that 'not a prediction' disclaimer that was pointed out is so important).
The relevant stats the CDC will be concerned with are transmission rates (R0), hospitalisation rates, and 'survivable only with intervention' rates.
Well, but it's also safe to assume that the outbreak in the US is mostly undiscovered as of yet. Their case fatality rate is at 6% or so, which is not anywhere close. Honestly, a infection...
Well, but it's also safe to assume that the outbreak in the US is mostly undiscovered as of yet. Their case fatality rate is at 6% or so, which is not anywhere close. Honestly, a infection fatality rate of somewhere around 0.4 or 0.5% is most likely right now, I think. Either the US has failed to save a whole bunch of people who could have been saved with better treatment or the US has only found 1 in 10 cases. Add that the deaths only occur much later, and you probably only found 1 in 10 cases of the cases that were here 2-3 weeks ago, nevermind the current ones.
Does anyone know why the US is reporting ~30000 cases every day for almost 2 months now? That doesn't look like the disease is under control (stronger decreasing trend) or like the disease is out of control, but observed. It looks like it's either deliberately controlled to be at that level with an ungodly amount of precision [1] or it's not being observed properly anymore. The basically same case number for weeks on end is not a result I'd expect from an unstable exponential process.
[1]: This is not to say they are deliberately spreading it or anything. But if someone were to say that 30k cases per day were acceptable, but to reopen as much as possible otherwise, this is the behaviour you could expect - except no one can control a disease that well because it's a lagging system with imperfect information.
I'd say the disease is both uncontrolled and unobserved in the USA. The number of cases being reported is limited by the amount of testing being done. I suspect there are a lot of infections not...
That doesn't look like the disease is under control (stronger decreasing trend) or like the disease is out of control, but observed. It looks like it's either deliberately controlled to be at that level with an ungodly amount of precision [1] or it's not being observed properly anymore.
I'd say the disease is both uncontrolled and unobserved in the USA. The number of cases being reported is limited by the amount of testing being done. I suspect there are a lot of infections not being detected because the tests aren't being done.
Also, I read something somewhere about cases picking up in some states while they're falling in other states. That would give the overall impression of a flatlining pandemic, when it just means the centre of infection has moved on.
I agree with your first assessment - I figured unobserved implied uncontrolled. The second part with multiple states imitating a flatline when in conjunction.... That'd be a slim chance of it...
I agree with your first assessment - I figured unobserved implied uncontrolled.
The second part with multiple states imitating a flatline when in conjunction.... That'd be a slim chance of it being such a flat line. I'd expect a lot more unexplained variance in that case. The weekly variance we see is easily explained away.
Sure the lower IFR means that the worst case is less bad. But it also means that the US is headed towards the worst case, because we know that there's a lot of cases going on that we don't know...
Sure the lower IFR means that the worst case is less bad. But it also means that the US is headed towards the worst case, because we know that there's a lot of cases going on that we don't know of.
Point is: It's very unlikely to be 0.2%. 1% is pessimistic. Studies elsewhere suggest 0.4-0.5%.
What we don't know is how infectious Covid-19 currently is in the US. A R of 1.3 vs 0.9 makes a world of difference. 1.3 is a ticking bomb and will give you a healthcare system overload. It's not flat enough. 0.9 is a problem solved, at least if the current level of the infections is acceptable. But since apparently only one in ten cases is discovered, it's not looking good.
The total case number also is the base for your infections. So if you have ten times the cases, you have ten times more new infections. That gets you a let closer to your healthcare system capacity.
We already know how full the hospitals have been and about how many people died in the hospital. Finding a lot more previously undiscovered infections can't change that history. We also have some...
We already know how full the hospitals have been and about how many people died in the hospital. Finding a lot more previously undiscovered infections can't change that history. We also have some idea of what the trends are.
Though, in both cases, we're looking in the rear-view mirror.
The CDC article makes no sense to me. If you divide the New York fatality rate of 23,488 by the fatality rate of 0.26%, you get an infection rate of 10.5 billion. That is over half the population...
The CDC article makes no sense to me.
If you divide the New York fatality rate of 23,488 by the fatality rate of 0.26%, you get an infection rate of 10.5 billion. That is over half the population of NY State. NYC has only 20% seroprevalence and had the bulk of the infections. That puts the IFR closer to 1, which is consistent with Princess Diamond
Two possible explanations: elderly people were more often affected in new York. Similar to the diamond princess. Other option is large scale healthcare failure. In Germany, which had plenty of...
Two possible explanations: elderly people were more often affected in new York. Similar to the diamond princess.
Other option is large scale healthcare failure. In Germany, which had plenty of capacity, a third of cases didn't leave the ICU alive. If we assume that no one died outside of ICUs and that everyone in ICUs is at risk of dying if they do not receive treatment, but everyone else can fend for themselves, the result is that healthcare failure results in triple the death count. That's a back of the envelope calculation; the truth is likely less than triple imo. And new York wasn't fucked to the point of providing almost no care, so that alone can't explain what you observed.
I've written elsewhere that my reading of the situation is an IFR of 0.4 to 0.5%. that would scale up to about 25% of NY infected.
It's looking like New York State screwed up big time when it comes to nursing homes. For a while they were requiring nursing homes to take people with COVID-19, which is the opposite of what you...
It's looking like New York State screwed up big time when it comes to nursing homes. For a while they were requiring nursing homes to take people with COVID-19, which is the opposite of what you want to do. Almost 5% of nursing home residents died.
The CDC article describes five scenarios for planning purposes. Two assume 1% overall fatality, two have 0.2%, and one has 0.4%. They explicitly say they "[a]re not predictions of the expected effects of COVID-19."
The article is a poor summary.
So 1% is the worst case they are planning for?
I don't know, but that's what's listed on the web page. The numerator is deaths and the denominator is infections with symptoms, so it seems like it couldn't be a lot higher without either missing a lot of deaths or overcounting infections by a lot.
Though, this is specific to the US. In an overwhelmed healthcare system or a country with more elderly, the death rate could be higher.
Apparently the 0.4% scenario is based on their current best estimates. The other scenarios are supposed to be best and worst cases.
I thought the US's healthcare system was overwhelmed. I've been seeing pictures of tents set up in parks to deal with all the sick people.
In New York things got pretty bad, but it's also the case that a lot of extra capacity got set up that ultimately wasn't used. I don't know what the results were when it comes to the fatality rate.
Looks like the worst of it has been in New York and Seattle. Most other cities aren’t quite so bleak. I think it’s a capacity thing. A cousin of mine in NYC ended up passing away of a mild heart attack that would have been survivable in normal times. But the whole system was so slammed nobody could get to him.
Mortality rate isn't really something for the CDC to plan for, but rather is something for economic organisations to plan for (which is why that 'not a prediction' disclaimer that was pointed out is so important).
The relevant stats the CDC will be concerned with are transmission rates (R0), hospitalisation rates, and 'survivable only with intervention' rates.
It's meaningless unless you know how infectious it is.
Don't we already have a pretty good picture of what SARS-CoV2's R0 is?
Well, but it's also safe to assume that the outbreak in the US is mostly undiscovered as of yet. Their case fatality rate is at 6% or so, which is not anywhere close. Honestly, a infection fatality rate of somewhere around 0.4 or 0.5% is most likely right now, I think. Either the US has failed to save a whole bunch of people who could have been saved with better treatment or the US has only found 1 in 10 cases. Add that the deaths only occur much later, and you probably only found 1 in 10 cases of the cases that were here 2-3 weeks ago, nevermind the current ones.
Does anyone know why the US is reporting ~30000 cases every day for almost 2 months now? That doesn't look like the disease is under control (stronger decreasing trend) or like the disease is out of control, but observed. It looks like it's either deliberately controlled to be at that level with an ungodly amount of precision [1] or it's not being observed properly anymore. The basically same case number for weeks on end is not a result I'd expect from an unstable exponential process.
[1]: This is not to say they are deliberately spreading it or anything. But if someone were to say that 30k cases per day were acceptable, but to reopen as much as possible otherwise, this is the behaviour you could expect - except no one can control a disease that well because it's a lagging system with imperfect information.
I'd say the disease is both uncontrolled and unobserved in the USA. The number of cases being reported is limited by the amount of testing being done. I suspect there are a lot of infections not being detected because the tests aren't being done.
Also, I read something somewhere about cases picking up in some states while they're falling in other states. That would give the overall impression of a flatlining pandemic, when it just means the centre of infection has moved on.
Of course, this is all speculation.
I agree with your first assessment - I figured unobserved implied uncontrolled.
The second part with multiple states imitating a flatline when in conjunction.... That'd be a slim chance of it being such a flat line. I'd expect a lot more unexplained variance in that case. The weekly variance we see is easily explained away.
If there are a lot more undiscovered infections then this implies a lower fatality rate, because that's what you're dividing by.
Sure the lower IFR means that the worst case is less bad. But it also means that the US is headed towards the worst case, because we know that there's a lot of cases going on that we don't know of.
Point is: It's very unlikely to be 0.2%. 1% is pessimistic. Studies elsewhere suggest 0.4-0.5%.
What we don't know is how infectious Covid-19 currently is in the US. A R of 1.3 vs 0.9 makes a world of difference. 1.3 is a ticking bomb and will give you a healthcare system overload. It's not flat enough. 0.9 is a problem solved, at least if the current level of the infections is acceptable. But since apparently only one in ten cases is discovered, it's not looking good.
The total case number also is the base for your infections. So if you have ten times the cases, you have ten times more new infections. That gets you a let closer to your healthcare system capacity.
We already know how full the hospitals have been and about how many people died in the hospital. Finding a lot more previously undiscovered infections can't change that history. We also have some idea of what the trends are.
Though, in both cases, we're looking in the rear-view mirror.
The CDC article makes no sense to me.
If you divide the New York fatality rate of 23,488 by the fatality rate of 0.26%, you get an infection rate of 10.5 billion. That is over half the population of NY State. NYC has only 20% seroprevalence and had the bulk of the infections. That puts the IFR closer to 1, which is consistent with Princess Diamond
I get 9,033,846. ( 23488 ÷ 0.0026 ) What did I miss?
Nope, you are right, I messed up my spreadsheet.
Two possible explanations: elderly people were more often affected in new York. Similar to the diamond princess.
Other option is large scale healthcare failure. In Germany, which had plenty of capacity, a third of cases didn't leave the ICU alive. If we assume that no one died outside of ICUs and that everyone in ICUs is at risk of dying if they do not receive treatment, but everyone else can fend for themselves, the result is that healthcare failure results in triple the death count. That's a back of the envelope calculation; the truth is likely less than triple imo. And new York wasn't fucked to the point of providing almost no care, so that alone can't explain what you observed.
I've written elsewhere that my reading of the situation is an IFR of 0.4 to 0.5%. that would scale up to about 25% of NY infected.
It's looking like New York State screwed up big time when it comes to nursing homes. For a while they were requiring nursing homes to take people with COVID-19, which is the opposite of what you want to do. Almost 5% of nursing home residents died.
https://newyork.cbslocal.com/2020/05/21/new-york-state-coronavirus-nursing-home-mandate/
Yeah, that'll do. So not much excess mortality through lack of resources required to explain the numbers.