Sunday, April 5, 2020

The data I'm not seeing reported on the Chinese Flu

There are two data points that are being reported: number of total cases and number of deaths.  This is unsatisfactory, because these data do not (directly) measure the most critical concern, the reason that we've shut down our whole economy and thrown 10M+ people out of work: the capacity of the health care system.  This (stolen from Gorges) sums up the inadequacy of the data as reported:


We're missing something.  Clearly if we overload the system then all sorts of Bad Things happen, but the data don't show this.  So the question is: are we overloading the health care system?  We need different data to understand this.  I would propose the following:

  • Number of admissions to the hospital
  • Number of ICU beds in use
  • Number of ventilators in use
  • Number of discharges from hospital
  • Number of discharges from ICU

I quite frankly have no idea what the numbers are for this, and haven't been able to find them.  I'm pretty convinced that these are the figures to look at, which will be key in deciding when (and how much) to reopen the economy.

If anyone has a link to where some of this can be found, I'd be very appreciative.

UPDATE 5 April 13:18: Lots and lots of really good, thoughtful comments.  I highly recommend everyone read them as well as this post.

30 comments:

ASM826 said...

There were over 1500 deaths in the U.S. yesterday, so we have crossed a death a minute, whatever treatments are being used.

JustPeachy said...

ASM826: where are you getting your numbers from? I've been keeping track on the Worldometers site, and their numbers don't match yours. I am looking for a more reliable/accurate tally.

Tacitus said...

Florida seems to have some numbers...

https://www.clickorlando.com/news/local/2020/04/03/coronavirus-heres-how-many-hospital-beds-are-available-in-florida/

And less specific, older data from New York

https://thecity.nyc/2020/03/new-york-hospital-icus-nearing-limit-as-covid-19-surges.html

I figure ICU bed occupancy is the best measure right now. Anybody there in theory is ill enough that a vent is possible. And it might be a better measure of projected mortality, as people who die of other things due to scarce resources are equally, and sometimes preventably, dead.

TW

FredLewers said...

I too use Worldometers. Not all the data but more than John Hopkins site. It looks like it's shifting from logarithmic to linear growth now. But it's still craptastic.

JustPeachy said...

Borepatch: the lack of data availability is distressing. I watched something happen last night that bothers me a lot, and suggests some data curating is going on. I've been tracking the US deaths-per-day using this site:

https://www.worldometers.info/coronavirus/country/us/

It has a big spreadsheet listing today's deaths, which resets at 8pm every evening (midnight GMT). Below that are some graphs, one of which is a bar graph listing deaths by day from the beginning of the epidemic up through the previous day. It also updates at 8pm (not precisely, but they've got the new graph up by 8:30pm generally).

So, yesterday, I had the page up right before 8pm, had refreshed it, so it was current, and when I looked at it, it listed around 1049 deaths for the day (4/4), and the bar graph had the previous day(4/3) as 1300+ deaths (consistent with the previous day's spreadsheet number at 8pm). But then, when everything updated around 8:20 or so, the bar graph listed 4/4's deaths as 1331, and 4/3's deaths as 1045. What happened? 300 people who died 4/3 suddenly traveled through time and died on 4/4 instead? It should have read as a spike in deaths on 4/3, and then a significant decrease 4/4, but it's flipped now, to show a constant upward trend. WTF?

I tried following their data source link, but the link is broken for 4/4, and for 4/3 no links are listed at all.

JustPeachy said...

Florida also keeps pretty fair stats here:

https://experience.arcgis.com/experience/96dd742462124fa0b38ddedb9b25e429

It lists cases, deaths, and hospitalizations by county.

Borepatch said...

Thanks for the links - I'll go dig around.

Tim, it seems like the discharge rate is what will tell us when we have effective treatments, but I am not by any means a specialist.

Beans said...

Cases at the moment vs beds at the moment.

Turn-and-burn cleaning and processing of beds time.

Graphs showing by the hour or by the day, comparing #people in the USA or the Area being shown, #tested, #hospitalized, #beds used vs # beds available and #treated and released. And do not talk about NYC or New Orleans stats when talking about Podunk USA. They are very different places.

The data will show, as the meme at the beginning shows, a very big disconnect between what we are hearing (Death STALKS THE STREEETSSSS) and what is actually happening (death comes to a small but very sick segment of the overall affected, and there are a lot more hospital beds out there available and a lot more med supplies available now than ever before.)

I read a couple machining boards, one dealing with 3D printing. At the beginning of this whole 'crisis,' local agencies were begging small 3D printing people (small manufacturers and gifted amateurs) for masks, face shields, other printable parts of vent systems. Now that the 3D printing parts are rolling out of people's garages and basements, the agencies that were screaming for the parts are stepping back and saying they have enough stuff already and are reluctantly accepting the stuff they contracted or begged for just a week ago. Why? Because supplies are more available than what are being talked about by the media. MASK SHORTAGE OMG!!! True, if you are an asshat like Cuomo... But the rest of the nation, not so much.

Tacitus said...

Discharges, be it from ICU or from hospital generally, will have validity over time. But in the short run...

Lots of people will be discharged from ICU if/when a wave of deathly ill folks hits town. Many of these are less ill folks being kept for monitoring and who did not "need' to be there. I think empty ICU beds is the only real measure of remaining capacity.

Effective treatment is very slippery. Things that work for a 60 year old won't work for a 99 year old. How ill people are before they hit the door, how many co-morbid conditions your community has, etc.

Hospital discharges generally could be GIGO. Most elective stuff has been cancelled. A hospital with a big orthopedic unit will be admitting and discharging fewer people these days. Said people are just limping around on their bad joints and hoping to get rescheduled for their elective joint replacement at some point.

I'd also look to regional numbers not just facility specific. To an extent patients can and should be moved about. Less ill people to community hospitals, really sick ones to tertiary care centers. This also will mess with your admit/discharge numbers unless you factor in transfers somehow.

My sense is we are nearing peak badness in the US. Gonna go soak me up a bunch of them virus killing photons now...

TW

Mike said...

50000 flu deaths per year is roughly 1000 per week. We are currently running around 8000 deaths per week from Covid, so figure that's a minimum factor of 8x. Anyone who doesn't see how that would put a strain on the system is lacking in basic common sense.

I'd like more info, but I have two family members that are medical care providers and the numbers they were giving me on case load were much higher than the official numbers. Same for a co-worker whose wife is a PA. I suspect that some of that is due to the immediate concern of the providers being to treat the patients, not make sure accurate data is being promulgated to public health figures, but it gives me little to no confidence in the official numbers.

There is some reason for optimism. Do a search for Ivermectin Covid treatment. It's showing very good results so far.

Old NFO said...

Nope, the numbers seem to 'fluctuate' in some odd patterns... And they don't 'match' the panic that is surrounding this.

SiGraybeard said...

I like what Mike @1322 has to say, but I look at it slightly differently. No offense intended, it's just how I think.

The total annual death rate is around 7800 peeps/day (2018 data from CDC) - which includes the flu. Not all of those are in hospitals because some are fatalities or suicides cleaned up by first responders, but most are hospitalized. Suicides and injuries account for about 680 deaths per day, putting the total closer to 7100 On top of whatever number that is, we've added around 1200/day, about 17% increase. Is our system so fragile that we can't handle a 17% overload?

That said, it's not the same everywhere and some cities and hospitals are worse off than others. Some are getting more than 17% overload and some may have fewer deaths than that year.

I'm getting these numbers from
https://www.cdc.gov/nchs/nvss/deaths.htm and a graph on
https://www.cdc.gov/nchs/products/databriefs/db355.htm
and my (t)rusty TI-86 calculator.

Skinnedknuckles said...

Sean at https://www.ncgunblog.com/ is doing a great job compiling this type of important data for NC. I think state-by-state effort will be the only way to get it. CT has been posting some good data but it doesn't cover as much as Sean has been able to find.

Divemedic said...

Part of it is the lag of testing results. My son admitted someone for viral pneumonia on 3/15. On admission, they tested him for COVID. He died on 3/20. The results of his COVID test came back positive on 3/23.

I know what you are thinking: Why did it take eight days to get the results? The lab is backed up, and the tests that are giving fast results are not accurate enough for hospital use, sort of like how there is a pregnancy test you can do at home by peeing on a stick that isn't accurate enough for health care use.

It doesn't matter at this point how fast the test is because the treatment for viral pneumonia is the same as COVID at this point.

Nick Flandrey said...

There are also posts from Drs that they are sending people home that they would have hospitalized in normal times. They know the people will be back when their condition worsens, but it keeps a bed free for another night or two.

We are used to transparency and instant access to data, but the only way we are getting CV info out of the facilities, is if someone tallies, codes, and sends it somewhere. That someone might be tired, overworked, incompetent, or just late getting numbers from subordinates. They might be getting direction to report only CDC approved test results, or they might be told to report all the 'looks like a duck, quacks like a duck' cases.

I don't think we can count on any numbers to be "accurate" any more than we can know the temperature of a city on a given day. If the methods of collection and reporting remain constant, we CAN use them for trends.

We can also see the second order effects and draw conclusions from them, forex- trailers full of dead people=not normal levels of dying.

nick

Ed Bonderenka said...

Well, that didn't clear it up for me.

JustPeachy said...

https://covidtracking.com/data/us-daily/

seems to be keeping track of the daily number history, and is keeping track of a lot of specific bits and bobs that you can't find on Worldometers.

Glen Filthie said...

Via The Rott:

https://theconservativetreehouse.com/2020/04/04/ground-reports-healthcare-focus-whats-going-on-in-your-city-town-neighborhood/

It’s just the flu, bro.

alo 89 said...

What I really want to know is total deaths for the first 3 months of the year. We can get monthly deaths by states for 1999 through 2017 from the CDC. For the 1918 Spanish Flu they used a calculation called excess deaths. If we could get monthly total numbers we could see for ourselves what part of total deaths equal excess deaths and see what coronavirus is actually doing without having to debate which deaths are which.

Aesop said...

Anybody - ANYBODY - anywhere, that thinks there's some mythical glut of PPE of any type is either delusionally psychotic, or smoking crack. There is no third option.

We have been critically short since February, in the backwater of this outbreak, and it's getting worse, not better, and our case load is still small, but trending upward, just doing so more slowly in Califrutopia than it is in cities like NYFC where their idiot mayor still has people riding the frickin' subway everyday.

STxAR said...

I made an ER visit this weekend to the local portal of doom. No visitors allowed, standard questions, took my own mask. It's a small town hospital, part of a big chain.

No wait, no issues noted with drugs or PPE. I was seated near the ADC and the PPE storage. Everyone was masked and gloved, with boxes of spares available. Spent some time hanging around the entrance waiting for my ride home, and visited with the nurse screening entrants. She said NYC is paying 6000/wk for agency nurses. No real issues down here in the south country that I was able to find out about.

I mean no offense, but one of the commenters here rings the panic bell on everything that comes our way it seems. I remember his Ebola predictions. I consider his predictions absolutely worst case.

I don't know of anyone that has contracted it inside my circle of friends and acquaintances. My daughter had a bad cold with crud hanging on for several weeks in early january, as did a couple co-workers. Their symptoms track pretty well with what I've read about the Red Peril Lung Malf. With Chairman Mao Jr's honesty and forthrightness, I'd expect CV-19 has been around the world twice before we even found out about it.

Richard said...

https://covid19.healthdata.org/projections

This is the UW model that the Feds appear to be using. It has actual data to date and then the projections out there. Read the methodological section. They are clearly using actual data to modify the model in something approximating real time. The uncertainty ranges are large but that is reflecting data limitations and lags. They are saying that there will be a revised version tomorrow.

The evil climate people have poisoned the well about models but my assessment of this is much more positive. It is certainly more transparent and is definitely changing as more and better data become available.

Glen Filthie said...

The problem I see, Richard, is that all the data coming in is inconsistent. If BP is turning in numbers 5 times mine, and 10 times ASM’s, and somebody else is coming 100 times mine... then that suggests that our data is compromised to the point of uselessness. You can shovel as much garbage in as you want... and all you will get is garbage out. There is no scientific basis for panic here at all.

We’ve seen politics run away with science before with the warble gloaming crowd. That seems largely to be a power and money grab with any number of fake scientists and paid actors cashing in. I’d like to see a similar grift going on here, but nobody seems to be in control of this particular narrative.

Yet.

Aesop said...

@SiG,
A virtual overnight 17% increase in patient load would crush healthcare.
You've leaned it out, like airlines, to the point that if it's not running at >90% most days, it goes broke.
There's no surge capacity, and they can't survive financially with 80% occupancy and 80% staff.
(Who do you know that's willing to sit around idle 20% of their working year?)
So it's high 90s, most days and times.
Thus little stretch if catastrophe strikes.
Like a pandemic.
When it backs up, because normal flu season/holidays combined, the full hospital forces everything in the ER to wait, and nobody outside can get in.

Think of your toilet if someone tried to flush 50 pounds of quilts down the drain, and you've got it.

And STxAR,
Just curious, using your logic:
How many people do you know personally who died in Iraq?
Or the WTC?
Or Vietnam?
Or WWII?

Does your separation from any of those mean they didn't happen?
Or weren't a big deal?

Personally, I've never, to my knowledge, met a murderer nor a child molester, but I still think it's a good idea to build prisons to keep them in, despite no firsthand acquaintance with the results of their predations. Nobody in my family, not even my extended family, has ever had a heart attack either, but I still think paramedics are a great idea.

Put another way, how many heat tiles on the shuttle Columbia had to fail during re-entry before it became a problem? How many O-rings on Challenger? How many sparks on how many switches had to happen in the cockpit of Apollo I?

Those are just three examples for what happens when you ignore the worst case scenario.

That something isn't affecting you, personally, or seems small, far away, and insignificant, and therefore isn't important, is really a rather juvenile level of reasoning, despite being 100% true and accurate factually, as far as it goes. You're basically planning on only learning from your own mistakes. No one has to think outside their own small circle of experience, but as a rule, doing so is generally a pretty good idea.

This pandemic isn't the Chicken Heart That Ate New York.
https://www.youtube.com/watch?v=fE0hHEtkkQA&t=1s
But it has the potential to do disproportional levels of damage, much like a squirrel gnawing through an electrical line can take out power to the entire Northeast U.S.

Nick Flandrey said...

"Personally, I've never, to my knowledge, met a murderer nor a child molester, but I still think it's a good idea to build prisons to keep them in, despite no firsthand acquaintance with the results of their predations. Nobody in my family, not even my extended family, has ever had a heart attack either, but I still think paramedics are a great idea."

--I've met and even worked for murderers. They're more common than you might think. Two of them were working in the same union local in TX for a while... and one of my old bosses beat a worker to death after hours on a job site.

--I've got family and acquaintances who have been accused credibly of child molestation. Also common.

--and I've got family and friends who've had heart attacks too.

--lastly to Star's point, I also know someone with actual test results saying she's got COVID. She's one of 1395 confirmed cases in our county, IF they've included her in the count as of today.

Aesop, you've been incredibly lucky, or maybe I'm just incredibly unlucky :-)

STxAR- your point of view is incredibly shallow, and all of Aesop's response applies in spades.

Too many people are still looking for reasons to discount this. Stop twisting yourselves into pretzels and use Occam's razor. Or Sherlock Holmes' reasoning. The actual evidence says it's real, it's bad, and it's going to get worse.

If you aren't personally affected, count your blessings and realize that people outside your own skull ARE, and are suffering devastating consequences.

Start positioning yourself for bad things happening.

Because if things do get bad, and you start whining about how no one told you, blah blah, you're not getting anything but scorn from me.

nick

Richard said...

@GlenFilthie
Yep the data is crap and therefore everybody's analysis is crap. But it doesn't stop anyone from having an opinion. Thing I like about the UW model is they are updating and disclosing. As we learn more the model will grow closer to reality. The disease is new and there hasn't been a lockdown before. Beats burning witches, though.

TechieDude said...

The term I heard for when someone thinks something isn't happening or won't happen to them because they don't know anyone it's happened to is "Infantile Omnipotence".

Usually it's used for dumbshit things. "I don't know anyone who's board broke riding down a stair rail. They probably screwed up. Wont happen to me." Then snap, bang, ouch...your off to the hospital with a pair of busted plums.

That said, there's a drastic difference in stats between cities. DFW is either behind, or doing something right, because cases are low, as are deaths compared to others. And with those low numbers, I know of probably 1/2 dozen or more that have/had it.

I must know 2-3 people, including my wife, that had what appeared to be godawful flu, but the test was negative. In the wife's case, a month later, she got the flu for real. So went through misery twice. She got the flu from the grandkids, who had it. I was with them when she picked it up. To this day, I don't have it. I usually don't get it (flu), for whatever reason.

I just got an email from an friend who's mom was in the hospital for some repairs after a fall, got the KungFlu there, Which they noticed the day she was to go to rehab. Never left the place. She lives in a state where cases are in the single digits.

Devin Page said...

The data is not comprehensive for many reasons, including lack of kits early on, untested and asymptomatic people, etc. It's inconsistent because it's being collected and reported by different entities all across the world, various states in the U.S., using whatever methods have been established from city to city, or from hospital to hospital. When it's all over and the data has been aggregated, disaggregated, and vetted, this event and the American response will be interpreted very differently by the left and right. For example, if the total U.S. deaths tops out at under 100,000, the left will say that (in spite of a late response by the government) mamy people were smart enough to follow the stay-at-home orders and that otherwise the totals would have climbed to the predictions of 100,000 - 150,000. The right will say that the original predictions of 100,000 - 150,000 deaths were too high, the media was fear-mongering, and that the lock-downs unnecessarily crippled the economy. Even if the worldometers site and every other source had the same data, people are going to interpret the numbers to fit their narrative and align with their pre-existing political leanings. That's fine, as long as it's just healthy skepticism and people stay at home. Thoughts: When it's all said and done, will we, as a country, learn from this? How long will the slogan "we're all in this together" last? Or, will COVID-19 divide the United States politically? Are there lessons (good and bad) to be learned from how we reacted to 911? How did Germamy rebuild after WWII? Japan after Hiroshima? How will the scientific and medical communities rebound? What are the parallels? Can we learn from the 1918 flu? Did we not learn enough the first time? What about global interdependence? Good or bad? Is American exceptionalism dangerous? Is self-reliance an asset or liability in a situation like this? When under attack by an invisible enemy, can individuals maintain a spirit of independence while also obediently unifying in our efforts to shelter on place? Are these mindsets mutually exclusive? Skepticism is healthy and expected in he scientific community, but is there a cost? All I ask is that we please interpret whatever numbers we read thoughtfully... comparing one month's worth of coronavirus numbers to a year's worth of flu numbers doesn't make sense no matter how cool it looks on a meme. The whole country changed after 9/11 and that took 3,000 lives. COVID-19 will take 20 - 30 times as many lives as 9/11 did, probably 30,000 more than the seadonal flu, maybe more. We had better learn how to rebuild from this, and become a tighter, more streamlined version of America. There will be lots of data... How were urban communities more vulnerable? Some more so than others? Why were urban and rural blacks dying at higher rates (please, do not respond with racism)? How did the virus affect different regions of the country? Did seasons or weather impact the virus? People smarter than me will hopefully find answers by examining our response, the factors that led to the virus' spread, and the same data for other countries, and we will have make changes as a country--at the national and local levels. To succeed, making necessary changes will require humility, courage, talent, and buy-in from millions of Americans.

Jen Miller, RN DON said...

CDC website has accurate numbers updated daily

Borepatch said...

Jen. I have to disagree. The CDC posts numbers every day but not the ones I talked about here. The gaps in the CDC data are so wide that we have a very imperfect view of what's happening.