Fantasy Football - Footballguys Forums
Welcome to Our Forums. Once you've registered and logged in, you're primed to talk football, among other topics, with the sharpest and most experienced fantasy players on the internet.
*** OFFICIAL *** COVID-19 CoronaVirus Thread. Fresh epidemic fears as child pneumonia cases surge in Europe after China outbreak. NOW in USA (5 Viewers)
- Thread starter Big League Chew
- Start date
culdeus
Footballguy
Look at the CDC data, back it up your two weeks. Tell me how correlated the deaths are to cases. You don't even have to make it a regression, but if you would like I could probably arrange that for you.
Doug B
Footballguy
Grace Under Pressure
Footballguy
Discharges outpacing patients added is one of the first metrics that point to successfully beating the thing back. That's exactly the time to stay at home as much as possible, wear masks 100%, stay far away from people, almost to act like there is a shutdown, and act like you have it and that everyone has it.
3-4 weeks of that behavior, coupled with discharges exceeding new patients, is a good formula. Rooting for Miami-Dade.
culdeus
Footballguy
It's not correlated to the case count, the IFR is still dropping. Even in hot spots. I mean Florida is the best example of this.
Maybe they are cooking their books, but I just have to deal with the information we have.
COVID cases are absolutely correlated with COVID deaths. The former is a neccesary cause for the latter. That correlation may be changing as more testing gets you closer to an accurate death rate, but to say they are not correlated is patently false.
Last edited by a moderator:
culdeus
Footballguy
The median death is at 13-14 days from infection. If that's changed slightly fine, but it doesn't allow room for "Wait six weeks" in any situation. You still have to apply whatever the median Infection to fatality number there is.
culdeus
Footballguy
Correlations at some point lose a significance, or lose their significance below certain thresholds. That is what is happening. IFR is dropping so fast that the correlation is weaker, to the point that you could either suppose the virus is weaker or we over-estimated it's impact. That was poor word choice on my part.
parasaurolophus
Footballguy
Wait six weeks is the new wait two weeks.
Its almost like somebody has stressed this point. He shall remain anonymous.
shader
Footballguy
You're dealing with information overload, and are struggling to put it all together. It's because you're trying to listen to both sides of this thing. It's got your mind all clouded, I can tell. I talk to a lot of people about this.
We've known for 3 months that the IFR is around 1%. I'm not sure what that has to do with anything. The reported fatality rate will never come close to what we saw in New York, because they couldn't test back then.
Ultimately that's where many people get confused. Back in the day, New York was reporting 10k cases a day and 1k cases a day. Florida is reporting 10k cases a day and 50 deaths a day.
Certain people out there take that difference and use it to get likes/follows/retweets, but they never think about why that is.
The reality is that New York didn't have 10k cases a day, they had 50-100k cases a day, they just weren't reporting them.
You have to start at square one and think about what this virus does. It infects a host, and a certain percentage of people get hospitalized. Then when people totally give out, a certain % of those die. Those deaths sometime happen fast, and they often happen many weeks, or sometimes months later.
Right now the people dying in Florida, for the most part, caught this virus 2 or more weeks ago. 2 weeks ago Florida set a record of 5,511 cases in one day. Even if we assumed that their records were perfect (meaning they are testing everyone perfectly), that would only be about 55 deaths from those cases.
In the last two weeks, Florida has not increased their testing. The % positive has skrocketed, and the cases per day have doubled. (the average age has also gone up)
If there are more cases, more people will die. I don't know what to tell you, it's that's simple. If you deny that, you're literally denying the facts.
shader
Footballguy
If correct, then like clockwork deaths are starting to rise just when we'd expect them to. That's why it's a certainty that deaths will rise over the next week, because we had more cases last week than we did two weeks ago.
Last edited by a moderator:
CR69
Footballguy
Source?
With an incubation phase of up to 14 days there is literally 0% chance the median death is two weeks after infection.
Nathan R. Jessep
Footballguy
Yep, and the part in red is a big factor, IMO, especially in trying to predict fatalities as an outcome.
Overall hospitalizations going up (in a state) aren't going to necessarily lead to a large increase in deaths, simply because some (or most, even) of the hospitalizations may be in areas that were vastly under their capacity at the time. That context matters greatly.
parasaurolophus
Footballguy
shader
Footballguy
I'd love a source as well. I found a WHO document that says that death ranges from 2-8 weeks AFTER symptom onset (not infection), but I'm sure no one wants to hear what they have to say about it.
shader
Footballguy
I've read the same, the incubation period isn't as long as first thought. It can go to 14 days, but 97% of people develop symptoms in the first 11 days and the average is 5-6 days.
Brony
Footballguy
They put stickers on the floor in my local grocery store, but there is no other signage nor anyone enforcing it. I can't really tell if people are following it. I personally am not religiously following it. Seems like more of a risk for me to be in the store to walk 2 extra one way aisles for something that I could grab in 5 seconds.
I was raised Catholic so I have an appreciation for following arbitrary rules that don't seem to make sense but everyone does it mostly so that they can shame those that don't. This feels like that category. I don't put masks and social distancing in that category.
culdeus
Footballguy
Using Wuhan and Princess Cruise ship figures, which may be the most "pure" this was the estimate at the time. Willing to hear otherwise. 13-14 days as the median came out of the cruise ship. I've yet to see it refuted.
https://www.mdpi.com/2077-0383/9/2/538
To the extent that cases are well matched with deaths I would present this
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7286834/#bib0005
They propose that deaths x7 - some time scale = cases. It's also compelling and would like some review of this. It would also suggest a overall CFR of 0.17% which is less than the flu. Also, they would suggest that deaths lag to a constant CFR in a matter of days, not weeks. Each country has their own lag, at that time it was 4 days for the US.
TLEF316
Footballguy
This makes no sense to me.
By restricting traffic to one way, you are allowing for easier social distancing. If you allow people to go both ways, that becomes impossible. (based on the width of most grocery store aisles)
This isn't arbitrary at all.
Last edited by a moderator:
culdeus
Footballguy
The stores around here have gone to one way nearly exclusively. Seems to make sense. The issue is the checkout. Still letting way too many people into that area.
parasaurolophus
Footballguy
According to the CDC...
Mean number of days from symptom onset to death
0-49 years: 14.9
50-64 years: 15.3
≥65 years: 12.9
So add 5.4 days and that is time from infection to death. For the purposes of seeing positive test result though to death, 14 days seems very accurate and 6 weeks is absurd.
Mean number of days from symptom onset to death
0-49 years: 14.9
50-64 years: 15.3
≥65 years: 12.9
So add 5.4 days and that is time from infection to death. For the purposes of seeing positive test result though to death, 14 days seems very accurate and 6 weeks is absurd.
shader
Footballguy
So the average is 2 weeks. 6 weeks is certainly not the norm, but it does happen.
Doug B
Footballguy
So, roughly half of all who will die from a given round of infection will take longer than 13-14 days to expire?
Besides, the literature suggests that the figure is more chaotic than a simple median can account for. A 391-case Chinese study (The Lancet, 4/27/2020) reported the following on their first three deaths:
culdeus
Footballguy
CR69
Footballguy
In the abstract of your first link it says the median death is 13 days AFTER onset of illness.
Also, since most people aren't NBA players they get to wait 7-10 days for test results to come back. Deaths aren't also reported the day they happen. So today's cases that are being reported are infections from weeks ago and today's deaths are likely the same.
This has followed an EXTREMELY predictable path as far as more cases leads (still climbing) to more hospitalizations (already happening) and then more deaths. I don't get how you're trying to argue this time is going to be different.
Battersbox
Footballguy
Yes, deaths will rise with more and more infections. Nobody is arguing they won't. The magnitude of the rise is quite relevant, however, and thus far it is nothing like what we saw out of New York.
It's an unfair argument to simply say 'deaths will rise, it's a fact' and not include the fact that it's just a sliver of what was seen in New York.
peak NY: 51 deaths/day/mm 7 day MA TX: 1.43 FL: 2.25 CA: 1.69
https://twitter.com/boriquagato/status/1280871098028920833
culdeus
Footballguy
I mean I don't really see how this makes the wait six weeks argument valid, using 3 anecdotal cases out of China from March.
shader
Footballguy
parasaurolophus
Footballguy
I have come to like it quite a bit as well. I like that he is no longer very strong in any of his opinions. He obviously learned from some of his early assertions. You can tell by how often he brings up New Zealand that he regrets what he said about travel bans causing more harm than good.
He says we need to learn to live with this virus. Says he understands that as many people as possible need to be back to work, but we need to pick the highest risk businesses and shut them down. Then immediately stresses that this is inherently unfair and that in exchange for being shutdown those specific industries need to be compensated.
He seems to try and cover every angle of things and stresses repeatedly that he just doesnt know for sure.
shader
Footballguy
Some are arguing that. And Florida is currently nowhere near where NY was, I'm not arguing that they are. But they can get there (on proportional level) if they don't change what they are doing.
Last edited by a moderator:
culdeus
Footballguy
13 days after onset or symptoms is a minor concern. CDC is still using the roughly 2ish week figure. Nowhere close to six.
AAABatteries
Footballguy
I think there's a lot of talking past each other because 1. most of us aren't experts and 2. we aren't using precise language. Of course a COVID case relates to a COVID death - I agree with you. But we can take things to an extreme to show that the conversation is more complex. Say I went overboard with a certain persons suggestion and stopped testing living people completely and only test someone after they die. My new cases and deaths are now 100% correlated. As we introduce changes in that - testing some subset and treating those who test positively differently than previous cases we make the correlation murky. I'm thinking that what culdeus is saying is that we just don't know enough yet to say if we are at a point where those changes we are introducing will potentially impact the correlation to a point where we see positive results. I tend to think we haven't and deaths will rise but not like they did in other locations.
shader
Footballguy
parasaurolophus
Footballguy
shader
Footballguy
shader
Footballguy
This was the best post of the page, and probably where we should leave this side tangent.
culdeus
Footballguy
Doug quoted an article suggesting we should wait six weeks to evaluate the death counts. It started a separate trail. When Doug puts a ton of block quotes in a post it makes it nearly impossible to respond to directly so it splinters the conversation to some extent (board issue)
AAABatteries
Footballguy
Don't know about you but I'm of two minds on this. One, I like that somebody, even someone as knowledgable as him, doesn't come across as knowing everything - it makes me trust him more. YMMV. Howerver, it scares the #### out of me that someone that is supposed to be an expert can't really tell us more.
parasaurolophus
Footballguy
shader
Footballguy
.If 2 weeks is the average, you're not going to get all the deaths in 2 weeks. A certain percentage of them will happen AFTER two weeks. That's why most of us have typically said that we have to wait 2-4 weeks to order to see deaths rise. (because at 2 weeks, you only have half the deaths).
Waiting until 6 weeks is fine too, there's nothing wrong with waiting to see. New York is a great example, we still have deaths coming in from the initial surge (broadway star as an example). But not many deaths, and not enough to really move the needle.
This isn't difficult. If accurate testing is happening, 2-4 weeks after a significant rise in cases we should start to see deaths rise. Which is exactly what we're seeing.
culdeus
Footballguy
Basically yeah, I feel we have gotten close to the point where we can declare victory in treating this thing. It's trending towards, if not at, a IFR rate that is similar or weaker than the worst flus we have seen in the last 50 years. Yes, the long term damage to survivors are regrettable, and if you want to lump those into some other greater figure I can get on board.
The real question is really what public policy can be instituted or should be in light of the weak(end) IFR. And how do you manage schools and school age kids, and workplaces that are not WFH friendlyish. Is it worth even shutting down schools for cases? Workplaces?
shader
Footballguy
Ignoratio Elenchi
Footballguy
Haven't followed the discussion closely but I've seen the "6 week" thing mentioned in relation to the likely time lag between certain large gatherings and death spikes because it accounts for several iterations of transmission, e.g. a bunch of 20 year olds go party on the beach and COVID spreads there, most of them won't die but they'll also take it home with them and spread it to their parents, some of whom will die and many of whom will spread it to their elderly parents and then we see an uptick in deaths, etc. Basically the theory is that we may witness some correlation between infections and deaths over a longer time period than just 2 weeks not because it takes that long for an individual to die, but because it takes that long to spread through their network to people who are more likely to die. Not making any claims to the validity of the theory, just providing a different way to think about the 6 week thing and where the idea may have come from.
AAABatteries
Footballguy
I disagree with this for several reasons - the main one being we haven't even lived with it for a year yet. I'm concerned now - I'm downright scared of what happens this late Fall/Winter if we haven't made any more progress in treatments and stopping larger hotspots like we have now.culdeus said:
culdeus
Footballguy
That's not how statistics work, we can be fine if we say most of what we expect to happens has happened by a certain time. This is more or less what I do in my job, on occasion.shader said:
This is a situation where I imagine a Weibull distribution works well. There is a set time where you expect no defect detection, then they surge, then you have a bleed off of that over time (gamma). To know if you are getting better you only need to know where the magnitude of the peak, and you can just ignore the gamma. There's no reason to model the back end because you can already predict what that looks like (and whether you are ####ed or not just based on where the peak is running).
If this is getting more back end pressure (because people are staying alive and dying later) then that is a minor concern. They probably just shifted from dying in 21 days to 48 days. Post peak defects tend to stay there, and improvements can at time push the date of those out. It's not worth changing the current direction over.
culdeus
Footballguy
Glass half full would say that there seems no seasonal pause in the summer, so a seasonal peak in the winter is unexpected.
Doug B
Footballguy
I didn't post the Lancet article to support "six weeks!". I posted it to support "longer than two weeks".culdeus said:
...
However, I do think the "six weeks" things has something to it. IMHO, it's not that a lot of people will get infected today and then start dying in six weeks. It's more like Carrier A gets infected today, then infects Carrier B in five days who infects someone else ... and then Carrier G gets infected and passes within a few weeks of their own infection -- which took place two weeks or so after Carrier A's infection.
When case counts inflate ... it's not necessarily the cases you can confirm right this second that make up your entire cohort of infected persons. There's also the secondary people to whom the known carriers have already spread it -- and the people the secondary people spread it to. And so on.
That's one giant reason to keep track of confirmed case numbers and for communities to try and keep raw cases down -- confirmed case numbers are proxies for the number of infections you can't see or account for. But those infections can and do still spread all the same.
EDIT: Or what Ignoratio Elenchi posted above.
Last edited by a moderator:
