The predictive value of COVID positivity

[Roadrunner]

CA has a real Covid fatigue I think. We had an early lockdown in March along with NY. Only we weren’t as hard hit as NY. However, NY at least somewhat enjoyed the summer, but we went from lockdown to surge and just as the numbers were starting to get batter, a surge again. We have barely left our home since March. My kids are starting to have mental health issues. I need to get them out for several days and away from here. 
This vaccine can’t come soon enough. 

[Not_a_Number]

8 minutes ago, Roadrunner said:

CA has a real Covid fatigue I think. We had an early lockdown in March along with NY. Only we weren’t as hard hit as NY. However, NY at least somewhat enjoyed the summer, but we went from lockdown to surge and just as the numbers were starting to get batter, a surge again. We have barely left our home since March. My kids are starting to have mental health issues. I need to get them out for several days and away from here. 
This vaccine can’t come soon enough. 

Yeah, our friends in California are really tired. 

I would have thought that there are more outdoor activities in CA than in NY nowadays, though? I mean, when it's not shelter in place, it seems like a better place to be in the winter... 

[Roadrunner]

13 minutes ago, Not_a_Number said:

Yeah, our friends in California are really tired. 

I would have thought that there are more outdoor activities in CA than in NY nowadays, though? I mean, when it's not shelter in place, it seems like a better place to be in the winter... 

true, but trails and beach were absolutely mobbed by tourists. We stayed home mostly.
 

[Bootsie]

13 hours ago, kand said:

Sure. I agree with all of this. My point was that people *say* things like “You have a 96%  chance of surviving Covid, why are we shutting everything down for a virus that isn’t even dangerous?!“  this is a thing I have heard and read from many, many people, using numbers ranging from 96% to 99%.  My point was for them to think that a 1 to 4% chance of dying of this was no big deal  shows they don’t have a good handle on what that would mean. 
 

eta:  because I’m still not sure it’s clear that it’s not me who is saying someone has a 1 to 4% chance of dying, I want to reiterate my point is that there are people saying that this is the chance but still thinking that that’s no big deal. That’s the disconnect.  )There certainly are people in demographics where their chance is that high if they do catch it, though.)

It’s particularly worrisome in California, given that they have the third lowest hospital beds per capita 🙁. 

The chance of not dying if you are in a plane crash is only 90-96%.  So, if you ARE in a plane crash you are more likely to die than if you DO get COVID.   But, the likelihood of being in a plane crash is much lower than the likelihood of contracting COVID.  The disconnect is people do not think in terms of conditional probability.  

[Not_a_Number]

11 minutes ago, kand said:

I think we’re talking past each other. It wasn’t that important of a point though, merely an observation. Hopefully most people understood what I meant.

I knew what you meant. 

For what it's worth, I think that if you act like those people are acting, their chance of contracting COVID this year is probably way over 1/2, so you may as well not do much conditioning. 

[Bootsie]

34 minutes ago, Not_a_Number said:

I knew what you meant. 

For what it's worth, I think that if you act like those people are acting, their chance of contracting COVID this year is probably way over 1/2, so you may as well not do much conditioning. 

My point is that people's experiences and frames of reference are based upon conditional probabilities.  They don't realize that they are comparing death rates with things that have very different conditional probabilities.  Working with a lot of young people, 96% chance of survival of COVID sounds very high when they hear that they chance of surviving a plane crash is 96% or less.  They mistakenly conclude that COVID is just as risky as flying.  They fail to take into consideration that if there is a 50% chance of contracting COVID and a 4% chance of dying--that is a 2% chance of contracting COVID and dying.  Whereas if the chance of getting into a plane crash is .001% (I don't know what the statistic actually is--probably much smaller) the chance of dying in a plane crash is only 0.004%.  

What I am pointing out is not that those individuals who say 96% survival rate--that is pretty good--are correct.  I am arguing that they are failing to take into consideration the higher probability of contracting COVID than some of their other activities for which they are comparing similar death/survival rates.  I am not arguing that they should greatly reduce their assessment of risk of COVID by conditioning the probability, I am arguing that when they make the comparison to other things that they are failing to condition those events with low occurrence. 

 

[The Governess]

Here are current numbers of deaths per day and 7-day average. I will add a new line and repost each day so we have an easy reference. 🙂

Mon, Nov 23 +991, 7-day 1581

Tues, Nov 24 +2210, 7-day 1661

Wed, Nov 25 +2332, 7-day 1712

Thurs, Nov 26 +1443, 7-day 1623

Fri, Nov 27 +1356, 7-day 1532

Sat, Nov 28 +1224, 7-day 1491

Sun, Nov 29 +820, 7-day 1483

Mon, Nov 30 +1275, 7-day 1522

Tues, Dec 1 +2667, 7-day 1589

Wed, Dec 2 +2874, 7-day 1665

Thurs, Dec 3 +2926, 7-day 1878

Fri, Dec 4 +2738, 7-day 2074

Sat, Dec 5 +2266, 7-day 2224

Sun, Dec 6 +1089, 7-day 2262

Mon, Dec 7 +1508, 7-day 2296

The 7-day average is still climbing, although the rate has slowed the last couple days.

[Not_a_Number]

Just now, lovelearnandlive said:

Here are current numbers of deaths per day and 7-day average. I will add a new line and repost each day so we have an easy reference. 🙂

Mon, Nov 23 +991, 7-day 1581

Tues, Nov 24 +2210, 7-day 1661

Wed, Nov 25 +2332, 7-day 1712

Thurs, Nov 26 +1443, 7-day 1623

Fri, Nov 27 +1356, 7-day 1532

Sat, Nov 28 +1224, 7-day 1491

Sun, Nov 29 +820, 7-day 1483

Mon, Nov 30 +1275, 7-day 1522

Tues, Dec 1 +2667, 7-day 1589

Wed, Dec 2 +2874, 7-day 1665

Thurs, Dec 3 +2926, 7-day 1878

Fri, Dec 4 +2738, 7-day 2074

Sat, Dec 5 +2266, 7-day 2224

Sun, Dec 6 +1089, 7-day 2262

Mon, Dec 7 +1508, 7-day 2296

The 7-day average is still climbing, although the rate has slowed the last couple days.

Thanks! 

What it looks like to me is that we've caught up. If you redistribute the last two weeks, the slope would be right around the slope of the weekend increases. Which (to me) suggests that we're still going up, but slowly... but we're going up from the actual average of 1900 or 1950 or so, not from what we are now. So I tentatively expect a decrease tomorrow. But I can't predict things perfectly day by day... next week is definitely the moment of truth, though. We'll see if positivity is still being predictive or not. 

 

[Ausmumof3]

290,000 deaths now.  If the 2000 plus deaths per day continues it will pass 300,000 this week.  Rolling average is still going up though there’s a weekend dip happening for the last two days. 

[The Governess]

10 minutes ago, Not_a_Number said:

Thanks! 

What it looks like to me is that we've caught up. If you redistribute the last two weeks, the slope would be right around the slope of the weekend increases. Which (to me) suggests that we're still going up, but slowly... but we're going up from the actual average of 1900 or 1950 or so, not from what we are now. So I tentatively expect a decrease tomorrow. But I can't predict things perfectly day by day... next week is definitely the moment of truth, though. We'll see if positivity is still being predictive or not. 

 

Yeah. I’m think that if we keep seeing increases this week that something may have changed. Like you I’m hoping for a decrease tomorrow. We will need to have fewer than 2667 deaths. 

[Not_a_Number]

13 minutes ago, lovelearnandlive said:

Yeah. I’m think that if we keep seeing increases this week that something may have changed. Like you I’m hoping for a decrease tomorrow. We will need to have fewer than 2667 deaths. 

One more day of a small increase... maybe. Anything after that, that’s weird.

[Corraleno]

FWIW, this is the IHME prediction for US deaths up to April 1st. And they often underestimate. ☹️

[Ausmumof3]

8 minutes ago, Not_a_Number said:

One more day of a small increase... maybe. Anything after that, that’s weird.

I’m assuming any positivity reduction due to increased testing for people looking to travel pre thanksgiving would have gone through reporting now?  

[Not_a_Number]

15 minutes ago, Corraleno said:

FWIW, this is the IHME prediction for US deaths up to April 1st. And they often underestimate. ☹️

 

IHME is junk, though. Their model isn't really using anything useful last I checked. 

... not that they are going to be that far off, whatever happens 😕 . Here's hoping the vaccine makes a real difference by taking care of the superspreaders and the vulnerable people. 

Edited December 8, 2020 by Not_a_Number

[Not_a_Number]

15 minutes ago, Ausmumof3 said:

I’m assuming any positivity reduction due to increased testing for people looking to travel pre thanksgiving would have gone through reporting now?  

I would think so. The positivity has been hovering around the same place for a while. We'll see what happens next. 

[Ausmumof3]

9 minutes ago, Not_a_Number said:

I would think so. The positivity has been hovering around the same place for a while. We'll see what happens next. 

Bno has it dropping 1pc today.  No idea where they got that from though.  It sounds positive.  Would be nice if you are peaking and turning around now.

[Laura Corin]

On 12/6/2020 at 9:03 PM, Roadrunner said:

It’s crazy. They are mostly living on food they packed and doing dinner takeout, but yes, she is telling me everywhere you to go pick up some food, it’s all as if there was no covid. 

Wow. Different here. I  just came back from picking up Hobbes (Scotland to England and back). Empty motorways and when I  picked up coffee in the morning, I  was the only customer.

[Not_a_Number]

OK, we've clicked over for the day. There was a mild increase in deaths, although most of it seems to be Iowa's new bookkeeping, which added something like 168 deaths to the total. If you get rid of those, things have stayed almost even from last week, inching up slightly. 

Taking those out of the equation, the 2-week average (which still seems much more meaningful than any other statistic due to Thanksgiving, unfortunately) is 1879. 

I'm very curious what happens for the rest of the week. My mild expectation is that the numbers stay relatively stable, like I've said, or perhaps go down to balance out the fact that last week was inflated. But I don't have strong opinions on this. 

Edited December 9, 2020 by Not_a_Number

[The Governess]

 

Tues, Nov 24 +2210, 7-day 1661

Wed, Nov 25 +2332, 7-day 1712

Thurs, Nov 26 +1443, 7-day 1623

Fri, Nov 27 +1356, 7-day 1532

Sat, Nov 28 +1224, 7-day 1491

Sun, Nov 29 +820, 7-day 1483

Mon, Nov 30 +1275, 7-day 1522

Tues, Dec 1 +2667, 7-day 1589

Wed, Dec 2 +2874, 7-day 1665

Thurs, Dec 3 +2926, 7-day 1878

Fri, Dec 4 +2738, 7-day 2074

Sat, Dec 5 +2266, 7-day 2224

Sun, Dec 6 +1089, 7-day 2262

Mon, Dec 7 +1508, 7-day 2296

Tues, Dec 8 +2913, 7-day 2337

Added today’s numbers. The 7-day average is still increasing - looks like it’s been a pretty consistent increase of around 40 per day for the last 3 days.

Feels like we’ll be seeing some 3000+ days soon. 😞

[Not_a_Number]

20 minutes ago, lovelearnandlive said:

 

Tues, Nov 24 +2210, 7-day 1661

Wed, Nov 25 +2332, 7-day 1712

Thurs, Nov 26 +1443, 7-day 1623

Fri, Nov 27 +1356, 7-day 1532

Sat, Nov 28 +1224, 7-day 1491

Sun, Nov 29 +820, 7-day 1483

Mon, Nov 30 +1275, 7-day 1522

Tues, Dec 1 +2667, 7-day 1589

Wed, Dec 2 +2874, 7-day 1665

Thurs, Dec 3 +2926, 7-day 1878

Fri, Dec 4 +2738, 7-day 2074

Sat, Dec 5 +2266, 7-day 2224

Sun, Dec 6 +1089, 7-day 2262

Mon, Dec 7 +1508, 7-day 2296

Tues, Dec 8 +2913, 7-day 2337

Added today’s numbers. The 7-day average is still increasing - looks like it’s been a pretty consistent increase of around 40 per day for the last 3 days.

Feels like we’ll be seeing some 3000+ days soon. 😞

It's possible. Last week was pretty close, so it wouldn't be a huge surprise. 

On the other hand, I still maintain that last week was inflated, so we'll see. 

[Not_a_Number]

The reason I'm cautiously optimistic is that I think the true graph looks like this, with the true 7-day averages in red: 

That's very approximately what you get if you rebalance things between the last two weeks. So, I do think it's going up, but it's not going up a ton. And since last week was inflated, staying on this curve would LOOK like going down. 

Edited December 9, 2020 by Not_a_Number

[Ausmumof3]

Worldometer has 2,960 today, 7 day rolling average at 2,344 (a little over the the April peak). Cases seem to be flattening?  If positivity is going down as well should be a slow down?

Edited December 9, 2020 by Ausmumof3

[Not_a_Number]

Just now, Ausmumof3 said:

Worldometer has 2,969 today, 7 day rolling average at 2,344 (a little over the the April peak). Cases seem to be flattening?  If positivity is going down as well should be a slow down?

And some of the 2969 is Iowa changing how it counts cases, at that -- it had an extra 170 or so from that, I think. 

My personal belief is that we'll go back down to meet the curve in red I drew above. But I'm not very sure about that -- just my prior. We'll see. 

[Not_a_Number]

Yikes. The 7-day average for positivity just jumped in JHU's database. I don't know if there's a statistical anomaly there... any ideas? It's because the positivity yesterday was very high. 

If this is correct, then we aren't going to have even a short reprieve. The best we can hope for is to come back to the curve in red I drew above (which would look like a decrease but wouldn't really be one) and then to inch up and up... 😞 

[Danae]

Ignoring the reporting blip around Thanksgiving, MN state positivity appears to have peaked on Dec 3.  That would be 2 1/2 weeks after the Governor tightened restrictions on indoor dining, gyms, etc and most schools moved to all online.  

Unfortunately that means deaths stay at peak levels for another 6 days.

[Not_a_Number]

46 minutes ago, Danae said:

Ignoring the reporting blip around Thanksgiving, MN state positivity appears to have peaked on Dec 3.  That would be 2 1/2 weeks after the Governor tightened restrictions on indoor dining, gyms, etc and most schools moved to all online.  

Unfortunately that means deaths stay at peak levels for another 6 days.

I'm seeing an earlier peak here: 

https://coronavirus.jhu.edu/testing/individual-states/minnesota

Is this graph wrong? 

[The Governess]

1 hour ago, Not_a_Number said:

Yikes. The 7-day average for positivity just jumped in JHU's database. I don't know if there's a statistical anomaly there... any ideas? It's because the positivity yesterday was very high. 

If this is correct, then we aren't going to have even a short reprieve. The best we can hope for is to come back to the curve in red I drew above (which would look like a decrease but wouldn't really be one) and then to inch up and up... 😞 

I wonder how much effect CA is having on the totals. We have only recently gone over 10% average positivity and have been reporting a lot of cases - sometimes as many as 30,000 positive cases per day.  As recently as 10 days ago our average positivity was 6%. 

[Roadrunner]

4 minutes ago, lovelearnandlive said:

I wonder how much effect CA is having on the totals. We have only recently gone over 10% average positivity and have been reporting a lot of cases - sometimes as many as 30,000 positive cases per day.  As recently as 10 days ago our average positivity was 6%. 

Do you know where we can get a summary look of all new case breakdowns by county? I would love to know say daily out of 25k new cases which counties  make up what percentage. I know LA Times has it behind a paywall, but curious if you know any place I can find it for free?

[Not_a_Number]

21 minutes ago, lovelearnandlive said:

I wonder how much effect CA is having on the totals. We have only recently gone over 10% average positivity and have been reporting a lot of cases - sometimes as many as 30,000 positive cases per day.  As recently as 10 days ago our average positivity was 6%. 

It probably doesn't make a huge difference. We had 213,000 positive tests yesterday. 

[Danae]

1 hour ago, Not_a_Number said:

I'm seeing an earlier peak here: 

https://coronavirus.jhu.edu/testing/individual-states/minnesota

Is this graph wrong? 

It's only slightly different than mine.  Their pre-Thanksgiving peak is a little higher, and post-no-cases-reported-on-Thanksgiving-dip peak is a bit lower.  I think considering it a noisy plateau from November 12 to December 3 is the best way to deal with Thanksgiving reporting effects.

 

Edited to add:  My graphs are the MN Department of Health numbers with cases counted on the day they are reported, not backdated to the day the sample was taken.  That probably accounts for small differences.

Edited December 9, 2020 by Danae

[Danae]

Quoted instead of edited.  Gah.

 

Edited December 9, 2020 by Danae

[The Governess]

48 minutes ago, Roadrunner said:

Do you know where we can get a summary look of all new case breakdowns by county? I would love to know say daily out of 25k new cases which counties  make up what percentage. I know LA Times has it behind a paywall, but curious if you know any place I can find it for free?

Yes, worldometers breaks it down by county - case numbers and deaths.

https://www.worldometers.info/coronavirus/usa/california/

Todays numbers aren’t posted yet but you can click to see yesterday’s numbers.

 

[Not_a_Number]

41 minutes ago, lovelearnandlive said:

Yes, worldometers breaks it down by county - case numbers and deaths.

https://www.worldometers.info/coronavirus/usa/california/

Todays numbers aren’t posted yet but you can click to see yesterday’s numbers.

 

There's probably also a CA dashboard? Or no? 

[Roadrunner]

1 hour ago, lovelearnandlive said:

Yes, worldometers breaks it down by county - case numbers and deaths.

https://www.worldometers.info/coronavirus/usa/california/

Todays numbers aren’t posted yet but you can click to see yesterday’s numbers.

 

I see totals. Their daily new count is ongoing though. I was hoping to see something like - 35k new cases in CA, of which 10k in Los Angeles, 5k in SF..... daily break down all together by county. I know I can compile them from all available data,  but I am too lazy. Hoping somebody did it for me. 

[Not_a_Number]

1 minute ago, Roadrunner said:

I see totals. Their daily new count is ongoing though. I was hoping to see something like - 35k new cases in CA, of which 10k in Los Angeles, 5k in SF..... daily break down all together by county. I know I can compile them from all available data,  but I am too lazy. Hoping somebody did it for me. 

They post daily totals by county, no? 

[Roadrunner]

43 minutes ago, Not_a_Number said:

They post daily totals by county, no? 

Total cases by county. Yes. But I would have to load them daily and then do net new. I would love to see historicals on net new. It’s too much work for me to recreate. Now if this lives somewhere in a spreadsheet to download, let me know! 

Edited December 9, 2020 by Roadrunner

[The Governess]

44 minutes ago, Roadrunner said:

I see totals. Their daily new count is ongoing though. I was hoping to see something like - 35k new cases in CA, of which 10k in Los Angeles, 5k in SF..... daily break down all together by county. I know I can compile them from all available data,  but I am too lazy. Hoping somebody did it for me. 

If you click on the “yesterday” link above the totals, it is a complete breakdown of yesterday’s numbers. Or maybe I’m not understanding what you are looking for? Totally possible today... 🤪

[The Governess]

Just now, lovelearnandlive said:

If you click on the “yesterday” link above the totals, it is a complete breakdown of yesterday’s numbers. Or maybe I’m not understanding what you are looking for? Totally possible today... 🤪

 

1 minute ago, Roadrunner said:

Total cases by county. Yes. But I would have to load them daily and then do net new. I would love to see historicals on net new. It’s too much work for me to recreate. Now if this lives somewhere in a spreadsheet to download, let me know! 

They don’t keep it historically though, beyond one day. I’d like that, too. 

[The Governess]

1 hour ago, Not_a_Number said:

There's probably also a CA dashboard? Or no? 

There is covid.ca.gov but I find worldometers to be the easiest way to get a quick breakdown of county-level info. Maybe I just haven’t dug deep enough on the state website...

[Danae]

50 minutes ago, Roadrunner said:

Total cases by county. Yes. But I would have to load them daily and then do net new. I would love to see historicals on net new. It’s too much work for me to recreate. Now if this lives somewhere in a spreadsheet to download, let me know! 

CSV file of historical data by county.

[Not_a_Number]

Yikes. We're over last Wednesday's numbers already, and it's only 6:22 Eastern 😕 . We're clearly about to have a "over 3000" day, and I'm worried positivity will stop being the right statistic.

[Not_a_Number]

Ugh. Seems like there's ANOTHER data dump today -- Colorado dumped 267 deaths, which is way more than it ever had before, and these are certainly not recent. 

Thanksgiving has also apparently shoved a whole bunch of data dumps, which is also messing with the numbers. Grrrrrrr. 

[The Governess]

40 minutes ago, Not_a_Number said:

Ugh. Seems like there's ANOTHER data dump today -- Colorado dumped 267 deaths, which is way more than it ever had before, and these are certainly not recent. 

Thanksgiving has also apparently shoved a whole bunch of data dumps, which is also messing with the numbers. Grrrrrrr. 

I noticed that and tried to do a quick search to see if there was any sort of explanation - didn’t see anything. And Worldometers doesn’t have a note (they have one for yesterday’s Ohio numbers).

[Not_a_Number]

Just now, lovelearnandlive said:

I noticed that and tried to do a quick search to see if there was any sort of explanation - didn’t see anything. And Worldometers doesn’t have a note (they have one for yesterday’s Ohio numbers).

It has to be a data dump, though, lol. They don't even have particularly high case numbers recently, and they've never had more than 100 before! 

I'm kind of annoyed that the week after Thanksgiving is polluted by data dumps by random states. Ugh. 

[The Governess]

9 minutes ago, Not_a_Number said:

It has to be a data dump, though, lol. They don't even have particularly high case numbers recently, and they've never had more than 100 before! 

I'm kind of annoyed that the week after Thanksgiving is polluted by data dumps by random states. Ugh. 

I agree that it must be a dump, but is it a dump that should be spread out over a week or 2 weeks or 2 months...? Someone should let us know!! 🤪

 

[Not_a_Number]

Just now, lovelearnandlive said:

I agree that it must be a dump, but is it a dump that should be spread out over a week or 2 weeks or 2 months...? Someone should let us know!! 🤪

 

Lol. Yeah, I wish they'd let us know, too!! Probably not recent data, either way, though. Data dumps never are... 

[The Governess]

Just now, Not_a_Number said:

Lol. Yeah, I wish they'd let us know, too!! Probably not recent data, either way, though. Data dumps never are... 

If you look at their graph of new cases, they were averaging 837 in mid-October and then increased to an average of almost 5k by mid-November, then leveled off around there. So they did have a big surge a few weeks ago. It would make sense if most of those deaths were recent, but definitely not one day’s worth. 🤔

[Not_a_Number]

1 minute ago, lovelearnandlive said:

If you look at their graph of new cases, they were averaging 837 in mid-October and then increased to an average of almost 5k by mid-November, then leveled off around there. So they did have a big surge a few weeks ago. It would make sense if most of those deaths were recent, but definitely not one day’s worth. 🤔

Yeah, I dunno. Maybe Thanksgiving.. 

[Ausmumof3]

2 hours ago, Not_a_Number said:

Yikes. We're over last Wednesday's numbers already, and it's only 6:22 Eastern 😕 . We're clearly about to have a "over 3000" day, and I'm worried positivity will stop being the right statistic.

Just went over 3000.

[The Governess]

Thurs, Nov 26 +1443, 7-day 1623

Fri, Nov 27 +1356, 7-day 1532

Sat, Nov 28 +1224, 7-day 1491

Sun, Nov 29 +820, 7-day 1483

Mon, Nov 30 +1275, 7-day 1522

Tues, Dec 1 +2667, 7-day 1589

Wed, Dec 2 +2874, 7-day 1665

Thurs, Dec 3 +2926, 7-day 1878

Fri, Dec 4 +2738, 7-day 2074

Sat, Dec 5 +2266, 7-day 2224

Sun, Dec 6 +1089, 7-day 2262

Mon, Dec 7 +1508, 7-day 2296

Tues, Dec 8 +2913, 7-day 2337

Wed, Dec 9 +3243, 7-day 2398

 

It looks like we probably crossed 3000 even accounting for the fact that CO’s numbers are high. Not a great day either way. Besides CO, 4 other states posted over 200 deaths (almost 300 for TX). 😞