News:

Welcome to the new (and now only) Fora!

Main Menu

Coronavirus

Started by bacardiandlime, January 30, 2020, 03:20:28 PM

Previous topic - Next topic

Treehugger

Quote from: zyzzx on March 25, 2020, 05:34:24 AM
Quote from: Caracal on March 25, 2020, 04:48:55 AM
Quote from: Treehugger on March 25, 2020, 03:47:28 AM
Just popping in from the rest of the Internet to report on the innumeracy of  the general populace.

A good one: "They had 5,000 new cases yesterday! Just because the new case numbers are lower doesn't mean they are flattening the curve." Actually, yes, that is exactly what that means.

Or: "No one is reporting that there are 100,000 recoveries. We need to stop cowering in our basements!" Yeah, 110,00 recoveries actually, but over 19,000 death which makes for a case fatality rate of 15%. So, 100,000 recoveries = not such great news after all.

This is innumeracy too. If you divide fatality rates by total number of cases, you could be underestimating fatality rate early on if there are lots of people who have recently gotten sick, those people skew young and if testing has been extensive. However, if you only count resolved cases when numbers have been rapidly increasing, especially if lots of sick people aren't being tested, you are going to get some really distorted numbers. You start getting deaths before you get recoveries.

None of this back of the envelope math is actually how epidemiologists figure out what a real fatality rate is though. To do that you have to find ways of estimating the actual number of people infected. I'm not going to pretend to have any sort of expertise, but I saw various people who do on Twitter saying that they think a study that did this through various methods and estimated a death rate of a bit over one percent in Hubei is probably the most accurate measurement of the death rate there. It doesn't do any good to pretend this isn't a big deal, but it also isn't helpful to overestimate the bad effects either.

With South Korea testing like crazy, seems like they should have the best handle on numbers of infected. They did aggressive testing of contacts, etc. and if they had missed a lot of asymptomatic cases, it seems unlikely that the new cases would be so low now. They are also well past their peak, so a large proportion of their cases have been resolved. Their death/recovery is 3.3% and death/cases is 1.4%. The first number will continue to go down, and the second number will continue to go up (assuming they don't get a second wave of cases), putting the final rate somewhere in between.
Similarly, the cruise ship where everyone was tested now has a death/case rate of 1.4%, which also still has the potential to increase.

So yeah, while we don't know the real death rate yet, we're not totally clueless. Best case scenarios where everyone can still receive top-quality care, and it seems like 1-2%. As health care systems get overwhelmed, this will go up.

Actually, one thing I have been wondering (and have not seen discussed) is how the case fatality rate of typical seasonal flus as well as epidemic flus are calculated. I have seen .1% cited as the case fatality rate for the former and 2.5% for the Spanish flu. But do these rates include all (or many) of the asymptomatic infections? I am thinking not — particularly in the case of the Spanish flu. Were there people who had the Spanish flu and were asymptomatic? Probably no way of knowing. I would also think it highly unlikely they were included in the CFR, if they existed. But how about H1N1 cousins of the Spanish flu? Can people be infected with H1N1 flu and remain asymptomatic? Are these cases included in the flu's case fatality rate?

I am bringing this up because it seems like one of the key things people do with various possible COVID-19 CFRs is turn around and compare them with those of various flus (generally with the intention to minimize COVID). For example, the CFR for the Spanish flu was 2.5%, but when you really test everyone, as they did in South Korea, the CFR for COVID-19 is much lower — only .4%. But, really, apples and oranges are being compared here —- unless all the asymptomatic or mild "cases" of the Spanish Flu were counted too, which I doubt.

Hegemony

Some of those questions are probably unanswerable, since we can't go back in time and retest for Spanish flu.

I've heard various arguments that Covid-19 is no worse than the flu, and I'm in no position to judge about fatality rates. What I do see is that thousands are dying from Covid-19, at a nastily accelerating rate. And because there is no vaccine still hardly any appreciable rates of immunity (and indeed it hasn't even been firmly established that surviving it confers immunity, though we hope so) — a huge percentage of the population is likely to get it, which is not the case with the flu.  If the flu were as transmissible and we were as vulnerable to it as to Covid-19, even the flu wouldn't be "just the flu." It would be something to be feared and avoided much more than it is now.

namazu

Quote from: Treehugger on March 31, 2020, 06:31:48 PM
Actually, one thing I have been wondering (and have not seen discussed) is how the case fatality rate of typical seasonal flus as well as epidemic flus are calculated.

For seasonal flu, here's how the CDC arrives at its estimates.

For the 1918 flu pandemic, there are numerous published papers giving (a wide range of) estimates and detailing the methodology used to estimate overall mortality and case fatality rates, some of which you can find here.

bacardiandlime

Thanks for the link, namazu. I had been wondering the same thing.

For this current pandemic, I'm feeling the % death rate being put out for many countries must be at this stage meaningless (given different rates of testing). Surely we will only really know the impact in about a year when we can compare the overall national mortality rate for 2020 with 2019 and see the difference?

nebo113

Quote from: backatit on March 31, 2020, 08:42:40 AM
Quote from: pigou on March 31, 2020, 08:13:02 AM
Quote from: backatit on March 27, 2020, 03:09:10 PM
They seem to be sold out both locally and on amazon (I'm trying to find one in a more out of the way sort of place, now, and will post on Nextdoor to see if anyone has a spare). I had been trying to find one before this, but they've been gone for a while now, unfortunately.
Have you looked at eBay? Looks like there's a large supply there: https://www.ebay.com/sch/i.html?_from=R40&_nkw=pulse+oximeter&_sacat=0&rt=nc&LH_BIN=1

Thanks, I hadn't thought to look there (I haven't shopped ebay in yonks!). You're a lifesaver (perhaps literally :D). We're actually both doing a bit better, fortunately, but someone else might need it or we might relapse, which I've heard is unfortunately common.

Never heard "in yonks" but love the way it rolls off the tongue!

Caracal

Quote from: namazu on March 31, 2020, 07:17:12 PM
Quote from: Treehugger on March 31, 2020, 06:31:48 PM
Actually, one thing I have been wondering (and have not seen discussed) is how the case fatality rate of typical seasonal flus as well as epidemic flus are calculated.

For seasonal flu, here's how the CDC arrives at its estimates.

For the 1918 flu pandemic, there are numerous published papers giving (a wide range of) estimates and detailing the methodology used to estimate overall mortality and case fatality rates, some of which you can find here.

It does highlight the really basic problems with the whole question. Nobody thinks it would be a good idea to try to estimate flu mortality with the number of confirmed cases as the denominator, because they know they are only getting confirming a minority of the total cases. There's no really good reason for most people to go to the doctor when they have flu symptoms. I always go, but that's just because I'm a worrier and want to hear a medical professional tell me that I seem normal sick and not alarming sick.

This is obviously much worse than normal flu, but it also has a really wide range of symptoms. Even in places like South Korea where they are testing a ton of people, they are still probably missing people who have mildish symptoms are asymptomatic, or very lightly symptomatic. There's also apparently a pretty high false negative rates for tests, maybe as high as 25 percent. You're just never going to get a reliable number by trying to do simple math, you have to do modeling of various sorts, and there's not enough data yet for these models to be particularly reliable.

Caracal

Quote from: bacardiandlime on April 01, 2020, 05:18:48 AM

For this current pandemic, I'm feeling the % death rate being put out for many countries must be at this stage meaningless (given different rates of testing). Surely we will only really know the impact in about a year when we can compare the overall national mortality rate for 2020 with 2019 and see the difference?

Even that might not tell you about COVID deaths per se. People might die at higher rates of other things if they are more reluctant to seek care, or if the medical system is under strain.

HomunculusParty

Quote from: Caracal on April 01, 2020, 05:46:03 AM
Quote from: bacardiandlime on April 01, 2020, 05:18:48 AM

For this current pandemic, I'm feeling the % death rate being put out for many countries must be at this stage meaningless (given different rates of testing). Surely we will only really know the impact in about a year when we can compare the overall national mortality rate for 2020 with 2019 and see the difference?

Even that might not tell you about COVID deaths per se. People might die at higher rates of other things if they are more reluctant to seek care, or if the medical system is under strain.

Yes, that's quite right. A family friend recently contracted an unrelated infection, but they were advised not to bring him to the overloaded hospital. It got rapidly worse, and he died two days after they finally brought him in. I know anecdote != data, but I'm sure others have experienced similar tragedies.

bacardiandlime

Quote from: HomunculusParty on April 01, 2020, 05:49:54 AM
Quote from: Caracal on April 01, 2020, 05:46:03 AM
Quote from: bacardiandlime on April 01, 2020, 05:18:48 AM

For this current pandemic, I'm feeling the % death rate being put out for many countries must be at this stage meaningless (given different rates of testing). Surely we will only really know the impact in about a year when we can compare the overall national mortality rate for 2020 with 2019 and see the difference?

Even that might not tell you about COVID deaths per se. People might die at higher rates of other things if they are more reluctant to seek care, or if the medical system is under strain.

Yes, that's quite right. A family friend recently contracted an unrelated infection, but they were advised not to bring him to the overloaded hospital. It got rapidly worse, and he died two days after they finally brought him in. I know anecdote != data, but I'm sure others have experienced similar tragedies.

Conversely I'm assuming that during the various lockdowns we will see a reduction in deaths from motor vehicle accidents.

spork

It's terrible writing, used to obfuscate the fact that the authors actually have nothing to say.

secundem_artem

Quote from: spork on April 01, 2020, 08:20:25 AM
Great U.S. data projections here:

https://covid19.healthdata.org/projections.

Agreed, but those 95% CI's are enormous.    I did not read what their underlying assumptions were but there's a lot of wiggle room in those numbers.  A few dozen more fundamentalist churches packing the pews for a couple more Sundays could blow these models to bits.
Funeral by funeral, the academy advances

nebo113

Quote from: secundem_artem on April 01, 2020, 09:03:05 AM
Quote from: spork on April 01, 2020, 08:20:25 AM
Great U.S. data projections here:

https://covid19.healthdata.org/projections.

Agreed, but those 95% CI's are enormous.    I did not read what their underlying assumptions were but there's a lot of wiggle room in those numbers.  A few dozen more fundamentalist churches packing the pews for a couple more Sundays could blow these models to bits.

https://www.usnews.com/news/us/articles/2020-03-31/louisiana-pastor-says-hell-keep-violating-coronavirus-ban

And reduce the numbers of Trump voters, while further crashing our fragile health care system.

Anselm

I am Dr. Thunderdome and I run Bartertown.

alto_stratus

This is making its way around the interwebz. Fun flashback to some of our earlier fora days.

https://www.rollingstone.com/music/music-news/robbie-robertson-the-weight-coronavirus-974210/

secundem_artem

Quote from: alto_stratus on April 02, 2020, 10:41:17 AM
This is making its way around the interwebz. Fun flashback to some of our earlier fora days.

https://www.rollingstone.com/music/music-news/robbie-robertson-the-weight-coronavirus-974210/

Here's another one.  Maybe the sweetest song ever from the Good Ol' Grateful Dead.

https://www.youtube.com/watch?v=MHo1fNnXFVU
Funeral by funeral, the academy advances