The Shape of the Covid19 Curve

[saveinla]

Heard that Washington is dismantling the field hospital set up at their stadium as they have not used it. Hope this is a good sign and will keep that way.


www.seattletimes.com/seattle-news/politics/inslee-sending-back-centurylink-field-hospital-to-federal-government-to-help-states-hit-harder-by-coronavirus/

[djAdvocate]

the shape of the COVID curve is breaking into three categories.

the upper tier is the US, France, Spain, Belgium, and the UK.

the second tier is the other European countries.

the third tier is Asian countries.

whether this is maintained is anyone's guess. but the top tier countries are doubling every 5 days now rather than every 2-4, which is, of course, great.

impossible to tell how it is all going to turn out, but it is getting more clear by the day.

[haapai]

There's a 4th tier too, just no data yet, and possibly never.  If there were reasonably accurate data being reported, the trend would show uncontrolled exponential growth. 

Sorry, I just had to point that out, having lived in the so-called third world as a child.  Some of those places developed, some of them developed at slow rates from a low base, and some of those states just failed.  The later two will follow a different pattern from the other three tiers described, but we are unlikely to get data showing that happening. 

[djAdvocate]

Apr 9, 2020 15:22:03 GMT -5 haapai said:

There's a 4th tier too, just no data yet, and possibly never.  If there were reasonably accurate data being reported, the trend would show uncontrolled exponential growth. 

Sorry, I just had to point that out, having lived in the so-called third world as a child.  Some of those places developed, some of them developed at slow rates from a low base, and some of those states just failed.  The later two will follow a different pattern from the other three tiers described, but we are unlikely to get data showing that happening. 

I think the DEATH curves are reasonably accurate.

if you are talking about the case curves, I agree. I already don't trust them.

including ours.

[haapai]

Sadly, I grew up in places where deaths from any cause are poorly counted.  Neither the new cases nor the Covid-19 deaths from these places should be trusted.  And don't even think about dropping back a step and looking at how the number of deaths reported recently compares to how many deaths were reported during the same period last year, or the year before.  I've lived in places where deaths don't get counted well.  These places are going to get pummeled if COVID-19 gets a foothold and we may never get a clear picture of how this thing from hell mowed over them. 

On a brighter note, once New York and New Jersey peak, the curves for the US should flatten out dramatically. 

And, yes, I haven't been too happy by the news that I have been receiving in the last few days.  As much as you tell yourself that the peaks will be hard to identify when they happen, you still hope for a dramatic drop off in deaths that I just haven't seen yet. 

[pulmonarymd]

Apr 9, 2020 17:51:43 GMT -5 haapai said:

Sadly, I grew up in places where deaths from any cause are poorly counted.  Neither the new cases nor the Covid-19 deaths from these places should be trusted.  And don't even think about dropping back a step and looking at how the number of deaths reported recently compares to how many deaths were reported during the same period last year, or the year before.  I've lived in places where deaths don't get counted well.  These places are going to get pummeled if COVID-19 gets a foothold and we may never get a clear picture of how this thing from hell mowed over them. 

On a brighter note, once New York and New Jersey peak, the curves for the US should flatten out dramatically. 

And, yes, I haven't been too happy by the news that I have been receiving in the last few days.  As much as you tell yourself that the peaks will be hard to identify when they happen, you still hope for a dramatic drop off in deaths that I just haven't seen yet. 

Deaths are a lagging statistics. The people on ventilators can take a while to die

[happyhoix]

Saw on the news tonight that the long lines of people crowding the sidewalks going into the hospitals in NYC two weeks ago are gone (at least at the hospital they were showing) indicating the number of new infections have dropped, which is a good indicator that the social distancing is working. 

Good news! 

[haapai]

I'm embarrassed to ask this question.  I think that I should already have seen or figured out the answer.  But here goes. 

Should I be preparing myself to see or expecting to see asymmetry in the graphs of new deaths per day?  What do Italy and math tell us to expect in NY?  That is, when all the data gets plotted, are we likely to see a rapid increase in the number of new deaths per day, followed by a long plateau, followed by a relatively slow decrease in the number of new deaths per day?  

Not having seen the curves of different reporting regions side by side or plotted on the same graph is really slowing down my understanding of what to expect. 

[pulmonarymd]

Apr 10, 2020 9:05:24 GMT -5 haapai said:

I'm embarrassed to ask this question.  I think that I should already have seen or figured out the answer.  But here goes. 

Should I be preparing myself to see or expecting to see asymmetry in the graphs of new deaths per day?  What do Italy and math tell us to expect in NY?  That is, when all the data gets plotted, are we likely to see a rapid increase in the number of new deaths per day, followed by a long plateau, followed by a relatively slow decrease in the number of new deaths per day?  

Not having seen the curves of different reporting regions side by side or plotted on the same graph is really slowing down my understanding of what to expect. 

Deaths will be the last thing to decline. Even when number of new infections slows, deaths will still go up. Think of it as pent up demand. These people are sick in the hospital, and their numbers go up as infections go up. Many people can be on the ventilator 10-14 days. So the number will lag due to that

[haapai]

What happens when a disease hits and the vast majority of the population uses social distancing to bring their R(0) below one but another, much smaller,  part of the population does not or cannot do this and has an R(0) of greater than one?  What kind of curve do you get when you overlay the two resulting curves? 

[djAdvocate]

Apr 11, 2020 14:49:56 GMT -5 haapai said:

What happens when a disease hits and the vast majority of the population uses social distancing to bring their R(0) below one but another, much smaller,  part of the population does not or cannot do this and has an R(0) of greater than one?  What kind of curve do you get when you overlay the two resulting curves? 

superposition, of course. the sum of the two curves.

[djAdvocate]

the growth rate is down significantly- in the single digits now.

so, I got (2) of my three predictions right from last week- we passed Italy and Spain on the number of deaths (though I was 1 day off on Italy- it happened in the wee hours Saturday, not Friday).

I was WAY off on the number of cases. and that's wonderful. so I am paring down my prediction for next Friday. I am thinking we will come in at under 750k cases by next Friday, which would be great.

the models out of Washington claim we are past the peak, but I am not convinced of that. having said that, today was clearly our most hopeful day in the last week- with only 1500 new deaths today. if that holds up, I would be a little surprised, however, as our case peak hit only a week ago, and was much LOWER 2 weeks ago than it is right now. if it scales up by cases, our peak will be closer to 3k/day, and will happen about a week from now- but I will admit that there may be something about the math I am missing. I hope I am wrong. AGAIN.

[oped]

With excess deaths numbers, particularly died at home in NY numbers that we know of, and potentially elsewhere with reasons to obfuscate, not being included in death numbers, I am not sure how accurate anything can be...

Italy hasn't had a ‘peak’ they’ve had a plateau... for weeks now. 

[haapai]

I refuse to get hopeful because of numbers reported over Easter weekend, particularly those of Italy and Spain, which have very predictable dips every weekend. 

[haapai]

Apr 13, 2020 1:02:25 GMT -5 djAdvocate said:

Apr 11, 2020 14:49:56 GMT -5 haapai said:

What happens when a disease hits and the vast majority of the population uses social distancing to bring their R(0) below one but another, much smaller,  part of the population does not or cannot do this and has an R(0) of greater than one?  What kind of curve do you get when you overlay the two resulting curves? 

superposition, of course. the sum of the two curves.

One curve is trending downward.  The other has exponential growth. 

[pulmonarymd]

Apr 13, 2020 6:57:38 GMT -5 haapai said:

Apr 13, 2020 1:02:25 GMT -5 djAdvocate said:

superposition, of course. the sum of the two curves.

One curve is trending downward.  The other has exponential growth. 

So it will be somewhere in between. These sort of curves are never smooth. Always a lot of noise

[thyme4change]

Apr 13, 2020 6:29:29 GMT -5 haapai said:

I refuse to get hopeful because of numbers reported over Easter weekend, particularly those of Italy and Spain, which have very predictable dips every weekend. 

People don't die over the weekend?

[pulmonarymd]

Apr 13, 2020 10:17:08 GMT -5 thyme4change said:

Apr 13, 2020 6:29:29 GMT -5 haapai said:

I refuse to get hopeful because of numbers reported over Easter weekend, particularly those of Italy and Spain, which have very predictable dips every weekend. 

People don't die over the weekend?

Data collection is not as complete over the weekend, so it is sometimes less accurate. Fewer people around to report it

[haapai]

Apr 13, 2020 10:17:08 GMT -5 thyme4change said:

Apr 13, 2020 6:29:29 GMT -5 haapai said:

I refuse to get hopeful because of numbers reported over Easter weekend, particularly those of Italy and Spain, which have very predictable dips every weekend. 

People don't die over the weekend?

Have you looked at the daily new deaths graphs for Italy and Spain?  Notice any pattern? 

[djAdvocate]

Apr 13, 2020 6:57:38 GMT -5 haapai said:

Apr 13, 2020 1:02:25 GMT -5 djAdvocate said:

superposition, of course. the sum of the two curves.

One curve is trending downward.  The other has exponential growth. 

if the attenuation in one equals the acceleration of the other, the trend will be linear up.

[djAdvocate]

Apr 13, 2020 10:31:13 GMT -5 haapai said:

Apr 13, 2020 10:17:08 GMT -5 thyme4change said:

People don't die over the weekend?

Have you looked at the daily new deaths graphs for Italy and Spain?  Notice any pattern? 

not really, no.

[haapai]

Apr 13, 2020 10:39:29 GMT -5 djAdvocate said:

Apr 13, 2020 6:57:38 GMT -5 haapai said:

One curve is trending downward.  The other has exponential growth. 

if the attenuation in one equals the acceleration of the other, the trend will be linear up.

Yeah, except I seem to have forgotten that it's not two discrete populations.  The portion of the population practicing social distancing strategies can be thought of as one population, but the portion of the population that is not practicing social distancing is actually made up of several separate groups.  The small populations of these non-distancing groups means that while exponential growth occurs, it does not go on forever.  Death, immunity and desertion kick in fairly quickly and exponential growth slows. 

Not exactly noise, but not exactly continued exponential growth either.  

It's a pretty awful thing for anyone who lives or works in a group living situation to be facing. 

I've got my doubts as to whether the IHME model attempts to take this into account. 

[haapai]

Today was not a good day for folks who have been following the trends in places that got hit hard and early by Covid-19.  Even if you have made some attempts to anticipate reporting delays, weekend delays and noise, this was not a good day. 

I think that it is time for us to admit that the curve goes up very fast and goes down much more slowly than any of our models predicted.  The curves in early-peaking areas have definitely slowed, possibly flattened, but they are not decreasing as quickly as the models that are widely available would predict. 

I hope to hell that I am an amateur alarmist. 

[djAdvocate]

when I first started watching this site, it said 60k dead.
I thought that was low.
they then revised to 90k dead, which seemed a little low, but in the ballpark.
they then went down to 60k again.
yesterday, they revised it to 69k.

incidentally, lots of people are quoting this work on this site.
but I still think they are low.

we'll, see. but the number today is 68841

covid19.healthdata.org/united-states-of-america

[teen persuasion]

Apr 15, 2020 0:47:17 GMT -5 djAdvocate said:

when I first started watching this site, it said 60k dead.
I thought that was low.
they then revised to 90k dead, which seemed a little low, but in the ballpark.
they then went down to 60k again.
yesterday, they revised it to 69k.

incidentally, lots of people are quoting this work on this site.
but I still think they are low.

we'll, see. but the number today is 68841

covid19.healthdata.org/united-states-of-america

Have they just stopped updating this site?  Last data added seems to be April 12.

[lynnerself]

The site said they were trying to update every 3 days. Data from the 12th was published on the 13th. So hopefully new data will be today.

[lynnerself]

IHME site has been updated today. With some new formatting as well as projections. There is now an area after the curves for the states called "containment strategy".

The projection for my state for peak utilization went from April 26th to today.

Projected deaths for the country is now 60,308

[oped]

I’m sorry but the shape of that graph seems absolutely ridiculous given what we know about the path this virus has taken elsewhere...

[djAdvocate]

Apr 17, 2020 18:09:12 GMT -5 oped said:

I’m sorry but the shape of that graph seems absolutely ridiculous given what we know about the path this virus has taken elsewhere...

I am inclined to agree.

they have also missed the peak twice already, including today.

I don't think their model is working out quite right.

again, I think we peak closer to 3000, and it takes a lot longer to come down.

my reasoning is that the 40% still think it is their God Given Right to infect people.

it only takes ONE to make a problem for many.

[djAdvocate]

i have been looking at the data again this afternoon.

so, we have approximate the same number of cases per million as France. France has lower number of tests per capita, which might account for the difference. but they also have nearly 3x the number of deaths per million as we do. so, from that, I am projecting the following range of possible deaths in the US;

currentUSdeaths*deathspermillion(france)/deathspermillion(US)*deaths = 105,000 max
105000*testspermillion(france)/testspermillion(US) = 60,000 min

of course, this assumes that France has no more deaths, which is obviously wrong. they have 5400 critical cases right now, and 100k active cases that are not critical. many of them will die. so those numbers will go higher (both min and max).

I don't know how the Washington site is coming up with 68k, but it does fall in the range, so maybe things will go far better for us than France, but I doubt it.

edit: that site has underestimated the peak death twice already.  we are clearly NOT at the peak yet.  but we might be getting near to it.  sometime in April, I would guess.....?