Author Topic: How long can we wait while flattening the curve?  (Read 248166 times)

HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2200 on: May 26, 2020, 08:22:43 PM »

What he actually said was:
Quote
Extending the analysis to consider absolute risk, we estimate the 95% confidence interval for future vCJD mortality to be 50 to 50,000 human deaths considering exposure to bovine BSE alone, with the upper bound increasing to 150,000 once we include exposure from the worst-case ovine BSE scenario examined.

https://pubmed.ncbi.nlm.nih.gov/11786878/

Correct, I will amend.  Ferguson predicted that up to 150,000 people would likely die from exposure to BSE (mad cow disease) in beef. In the U.K., there were only 177 deaths from BSE.
« Last Edit: May 26, 2020, 08:40:12 PM by HBFIRE »

Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2201 on: May 26, 2020, 08:40:32 PM »

What he actually said was:
Quote
Extending the analysis to consider absolute risk, we estimate the 95% confidence interval for future vCJD mortality to be 50 to 50,000 human deaths considering exposure to bovine BSE alone, with the upper bound increasing to 150,000 once we include exposure from the worst-case ovine BSE scenario examined.

https://pubmed.ncbi.nlm.nih.gov/11786878/

Correct, I will amend.  Ferguson predicted that up to 50,000 (150 K?) people would likely die from exposure to BSE (mad cow disease) in beef. In the U.K., there were only 177 deaths from BSE.

Taken in the context of the full abstract I don't think it's accurate to say they "predicted" anything. Projections are not predictions, especially when the projections require certain conditions be met. Here's the full abstract for context:

Quote
Following the controversial failure of a recent study and the small numbers of animals yet screened for infection, it remains uncertain whether bovine spongiform encephalopathy (BSE) was transmitted to sheep in the past via feed supplements and whether it is still present. Well grounded mathematical and statistical models are therefore essential to integrate the limited and disparate data, to explore uncertainty, and to define data-collection priorities. We analysed the implications of different scenarios of BSE spread in sheep for relative human exposure levels and variant Creutzfeldt-Jakob disease (vCJD) incidence. Here we show that, if BSE entered the sheep population and a degree of transmission occurred, then ongoing public health risks from bovine BSE are likely to be greater than those from cattle, but that any such risk could be reduced by up to 90% through additional restrictions on sheep products entering the food supply. Extending the analysis to consider absolute risk, we estimate the 95% confidence interval for future vCJD mortality to be 50 to 50,000 human deaths considering exposure to bovine BSE alone, with the upper bound increasing to 150,000 once we include exposure from the worst-case bovine BSE scenario examined.

HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2202 on: May 26, 2020, 08:43:50 PM »

Taken in the context of the full abstract I don't think it's accurate to say they "predicted" anything. Projections are not predictions, especially when the projections require certain conditions be met. Here's the full abstract for context:



Feel free to swap "prediction" for "projection".  Either way, my points stand.

waltworks

  • Magnum Stache
  • ******
  • Posts: 4265
Re: How long can we wait while flattening the curve?
« Reply #2203 on: May 26, 2020, 09:19:22 PM »
Is it fair to say, simultaneously:

-2-3 months ago when we didn't know much about this, it was worth freaking out.

-Now, when we know more, the freakout was probably an overreaction. Very elderly people in nursing homes dying is horrible, but it's probably not worth ruining the economy, if we're going to be rational about it.

Unfortunately this is now a partisan issue, so half of everyone won't admit the latter, and half won't admit the former.

-W

Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2204 on: May 26, 2020, 09:26:16 PM »

Taken in the context of the full abstract I don't think it's accurate to say they "predicted" anything. Projections are not predictions, especially when the projections require certain conditions be met. Here's the full abstract for context:



Feel free to swap "prediction" for "projection".  Either way, my points stand.

If you don't know how important the distinction between those two words is, you shouldn't be critiquing Neil Ferguson.

Prediction - a probabilistic statement that something will happen in the future based on what is known today

Projection - a probabilistic statement that it is possible that something will happen in the future if certain conditions develop

If we're going to take the upper bounds of a projection and call it a prediction, we could make anyone who's ever modeled a model look like an idiot.

HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2205 on: May 26, 2020, 09:30:00 PM »

-Now, when we know more, the freakout was probably an overreaction. Very elderly people in nursing homes dying is horrible, but it's probably not worth ruining the economy, if we're going to be rational about it.

Unfortunately this is now a partisan issue, so half of everyone won't admit the latter, and half won't admit the former.

-W

I agree with this.  We had bad data but needed to act quickly.  Now we should act on accurate data.  The projection models have been embarrassingly bad.
« Last Edit: May 26, 2020, 10:11:31 PM by HBFIRE »

HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2206 on: May 26, 2020, 10:17:45 PM »

If we're going to take the upper bounds of a projection and call it a prediction, we could make anyone who's ever modeled a model look like an idiot.

No need to do that.  The so called projections speaks volumes.

According to Ferguson's projections, the current Swedish government’s response – if permitted to continue – would pass 40,000 deaths shortly after May 1, 2020 and continue to rise to almost 100,000 deaths by June.


Again, Sweden is at 4,125 deaths today and peaked weeks ago.

Any model that is off by order of magnitude needs to be pointed out, particularly when the world is using it to lead its major policy decisions.  This wasn't the first time.

Ferguson admitted in March that his model was based on an undocumented 13 yr old computer code originally designed for an influenza pandemic rather than a novel coronavirus outbreak. He refused to release his original code so other scientists could check his results.
« Last Edit: May 26, 2020, 10:25:22 PM by HBFIRE »

Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2207 on: May 26, 2020, 10:26:20 PM »

-Now, when we know more, the freakout was probably an overreaction. Very elderly people in nursing homes dying is horrible, but it's probably not worth ruining the economy, if we're going to be rational about it.

Unfortunately this is now a partisan issue, so half of everyone won't admit the latter, and half won't admit the former.

-W

I agree with this.  We had bad data but needed to act quickly.  Now we should act on accurate data.

What bad data? Can you give specific examples?

Bad data comes from things like bad tests giving too many false positives or under/over-attributing deaths to coronavirus or intentional lying.

Perhaps you meant bad conclusions drawn from good data? Can you give specific examples?

And I mean examples from experts, not the media. Bad conclusions drawn from good data (and research) are rampant on both sides of the argument in the media which is a problem if decisions are made based on bad reporting... which they probably have been. But can you give specific examples?

HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2208 on: May 26, 2020, 10:31:22 PM »


What bad data? Can you give specific examples?


Bad data = we thought the fatality rate was 3.5% initially, now it is estimated to be ~ 0.3% according to the CDC's estimate based on over 50 serological/PCR studies performed around the world.

We also initially thought the R0 was as high as 3+, and data now indicates it is 2.5, much less contagious than initially thought.

This is why back in March, Dr Ionnadis of Stanford, one of the top epidemiologists in the world, questioned our policy decisions based on incomplete data.  Dr Ionnadis's latest study is what the CDC is using for its most up to date IFR projections.




« Last Edit: May 26, 2020, 10:39:40 PM by HBFIRE »

Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2209 on: May 26, 2020, 11:00:25 PM »

If we're going to take the upper bounds of a projection and call it a prediction, we could make anyone who's ever modeled a model look like an idiot.

No need to do that.  The so called projections speaks volumes.

According to Ferguson's projections, the current Swedish government’s response – if permitted to continue – would pass 40,000 deaths shortly after May 1, 2020 and continue to rise to almost 100,000 deaths by June.

Ferguson wasn't even an author on that paper. The model type was based on his research, he was only a reference.

https://www.medrxiv.org/content/10.1101/2020.04.11.20062133v1.full.pdf

Quote
Again, Sweden is at 4,125 deaths today and peaked weeks ago.

Any model that is off by order of magnitude needs to be pointed out, particularly when the world is using it to lead its major policy decisions.  This wasn't the first time.

Ferguson admitted in March that his model was based on an undocumented 13 yr old computer code originally designed for an influenza pandemic rather than a novel coronavirus outbreak. He refused to release his original code so other scientists could check his results.

I don't have time to read the paper right now, but lets assume that this model was as wrong as you say. Was this model used to decide public policy? Has it even been peer-reviewed?


HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2210 on: May 26, 2020, 11:18:05 PM »
Was this model used to decide public policy? Has it even been peer-reviewed?

Yes, for both the US and UK.

Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2211 on: May 26, 2020, 11:19:10 PM »


What bad data? Can you give specific examples?


Bad data = we thought the fatality rate was 3.5% initially, now it is estimated to be ~ 0.3% according to the CDC's estimate based on over 50 serological/PCR studies performed around the world.

Who thought that? Do you have citations? The WHO said that the case fatality rate was 3.4% based on current data in early March, which was correct. Did they actually estimate a 3.4% infection fatality rate? That seems unlikely but I'll wait for the citation.

Quote
We also initially thought the R0 was as high as 3+, and data now indicates it is 2.5, much less contagious than initially thought.

R0 values aren't inherent to the virus, they change based on human behavior. To analyze any further we would need citations for your figures.

HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2212 on: May 26, 2020, 11:24:29 PM »


Who thought that? Do you have citations? The WHO said that the case fatality rate was 3.4% based on current data in early March, which was correct.


https://khn.org/morning-breakout/mortality-rate-placed-at-3-4-but-some-experts-say-thats-a-crudely-calculated-snapshot-that-will-change/

It was recognized that it was a crude estimate, nonetheless, it was based on incomplete data.

HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2213 on: May 26, 2020, 11:25:07 PM »

R0 values aren't inherent to the virus, they change based on human behavior. To analyze any further we would need citations for your figures.

R0 doesn't change, it's a base value. 

Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2214 on: May 26, 2020, 11:27:09 PM »
Was this model used to decide public policy? Has it even been peer-reviewed?

Yes, for both the US and UK.

The one titled "Intervention strategies against COVID-19 and their estimated impact on Swedish healthcare capacity" which was not peer reviewed and posted as a preprint on 4/15/20? That seems unlikely.

I think you're mixing up this study with "Impact  of  non-pharmaceutical  interventions  (NPIs)  to reduce COVID-19 mortality and healthcare demand" which Neil Ferguson was an author on.

HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2215 on: May 26, 2020, 11:29:01 PM »


The one titled "Intervention strategies against COVID-19 and their estimated impact on Swedish healthcare capacity" which was not peer reviewed and posted as a preprint on 4/15/20? That seems unlikely.


No, the Imperial College Model was used to shape US and UK government policies.

Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2216 on: May 26, 2020, 11:29:15 PM »


Who thought that? Do you have citations? The WHO said that the case fatality rate was 3.4% based on current data in early March, which was correct.


https://khn.org/morning-breakout/mortality-rate-placed-at-3-4-but-some-experts-say-thats-a-crudely-calculated-snapshot-that-will-change/

It was recognized that it was a crude estimate, nonetheless, it was based on incomplete data.

Once again, citations? It was never given as an estimate by the WHO, it was given as the current CFR and was correct. Who used this as an estimate?

ETA: The headline from your citation is enough to disprove your interpretation. You turned "crudely calculated snapshot" into crude estimate". You made up the estimate part entirely.
« Last Edit: May 26, 2020, 11:39:52 PM by Davnasty »

Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2217 on: May 26, 2020, 11:31:08 PM »

R0 values aren't inherent to the virus, they change based on human behavior. To analyze any further we would need citations for your figures.

R0 doesn't change, it's a base value.

"R0 is not a biological constant for a pathogen as it is also affected by other factors such as environmental conditions and the behaviour of the infected population."

https://en.wikipedia.org/wiki/Basic_reproduction_number

marty998

  • Walrus Stache
  • *******
  • Posts: 7039
  • Location: Sydney, Oz
Re: How long can we wait while flattening the curve?
« Reply #2218 on: May 26, 2020, 11:31:49 PM »
All this talk about Ferguson is irrelevant.

Especially the references to him being left wing or his mistress being a left wing activist. It’s a nonsense for a number of reasons, not least being teams of people and students would be involved in the research. His name just goes on the paper (the figurehead so to speak).

If it’s all a left wing socialist conspiracy and if it has been used to inform public policy then please explain to me why you think the right wing government in the UK swallowed it? They are free to reject it but they did not.

It never ceases to amaze me how when conservatives are in power their supporters still act like they are not in power. Ultimately governments have the choice to accept or reject research from a variety of sources.  They would not rely on a single model to shut down an entire economy. That is patently stupid. They would be assessing all sources and evidently they all said “shut down is the best option you have” so they went with it.


Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2219 on: May 26, 2020, 11:35:19 PM »


The one titled "Intervention strategies against COVID-19 and their estimated impact on Swedish healthcare capacity" which was not peer reviewed and posted as a preprint on 4/15/20? That seems unlikely.


No, the Imperial College Model was used to shape US and UK government policies.

It was, but this conversation began with your critique of the other study. That's the one that you cited to show Neil Ferguson was wrong even though he did not author that paper.
« Last Edit: May 26, 2020, 11:46:36 PM by Davnasty »

HBFIRE

  • Handlebar Stache
  • *****
  • Posts: 1249
  • Age: 42
  • Location: Huntington Beach, CA
Re: How long can we wait while flattening the curve?
« Reply #2220 on: May 27, 2020, 12:01:52 AM »

Once again, citations? It was never given as an estimate by the WHO, it was given as the current CFR and was correct. Who used this as an estimate?

ETA: The headline from your citation is enough to disprove your interpretation. You turned "crudely calculated snapshot" into crude estimate". You made up the estimate part entirely.

https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(20)30244-9/fulltext



"R0 is not a biological constant for a pathogen as it is also affected by other factors such as environmental conditions and the behaviour of the infected population."

https://en.wikipedia.org/wiki/Basic_reproduction_number

Ah yes, I stand corrected.  I was thinking of how R changes as more become immune and how that is different than R0.  Thanks, R0 is quite complex.



It was, but this conversation began with your critique of the other study. That's the one that you cited to show Neil Ferguson was wrong even though he did not author that paper.

My point is that we used a very flawed model (Ferguson's Imperial College) to base our major policy decisions on.  The entire model assumed an IFR of 0.9% and this is why it predicted as many as 2.2 million Americans would die.  This goes back to my point that we need to base our policy decisions on better data now that we have it.  We needed a good estimation of IFR, and we finally have that.  A model is only as good as the inputs, and we had the incorrect inputs.



Ferguson wasn't even an author on that paper. The model type was based on his research, he was only a reference.


"Although ICL only released scenarios and associated forecasts for the United Kingdom and United States, its model is adaptable to any country by changing the inputs to reflect its population, demographics, and the date its specific policies took effect. In early April around the peak of the academic community’s backlash against the Swedish government’s strategy, a group of researchers at Uppsala University attempted to do just that. They released an epidemiological model for Sweden that adapted the ICL COVID-19 model from Ferguson and his colleagues, and attempted to project the effects of Sweden’s unique response on both hospital capacity and total fatalities."

They applied Ferguson's model to Sweden.  The reason it was off by so much is that it uses an IFR input of 0.9%.
« Last Edit: May 27, 2020, 12:37:54 AM by HBFIRE »

mathlete

  • Handlebar Stache
  • *****
  • Posts: 1638
Re: How long can we wait while flattening the curve?
« Reply #2221 on: May 27, 2020, 12:37:56 AM »
The cheapest of second guessing.

mathlete

  • Handlebar Stache
  • *****
  • Posts: 1638
Re: How long can we wait while flattening the curve?
« Reply #2222 on: May 27, 2020, 01:32:10 AM »
We've talked a lot about emotional leverage in this thread in terms of people ostensibly "controlled by fear", but spent a relatively smaller amount of time talking about what is in my view, more abundant emotional leverage.

People (myself included) like being right on the internet. Some people (myself not included) do this by adopting and dying by contrarian takes to the point of silliness. Examples include;

-Deciding that a source they previously derided, like the CDC, now has it right when they come out with an estimate that we like better
-Making a majority of their posts on a forum about a single contrarian take
-Selectively deciding to feign concern for food insecure African grandmothers

You guys ever notice how much time we spend on the Imperial College study? That is so obviously because it's a big number that probably won't come to fruition. Never mind that it was caveated by twenty pages of work. Never mind that it was only one of now dozens of white papers put out on the virus. Let's go ahead and use it to discredit the notion of trying to quantify the impact of assumptions to support decision making.

You ever notice how little time we spend talking about the statement from late March by Dr. Fauci that we could see between 100K and 240K deaths? Unless something drastically changes, the US death toll will probably fit comfortably in that interval. At least in the first (hopefully only?) wave.

You ever notice how much time we spend on presumptive asymptomatics vs. the reporting that excess mortality may be greater than reported COVID deaths? It's because on balance, we're emotionally leveraged more towards one side than another.

Statements like, "We should make policy decisions based on better data" are truisms. No one would disagree with this. But where is this better data? Or more specifically, where was it in late Feb/Early March? The answer is that it didn't exist. We had deaths and we had limited testing data. We can study that to make educated guesses about IFR, R0, and presumptive asymptomatics. Of course the possibility was always there that the latter two metrics were higher, meaning the IFR could be lower and still support the number of deaths we're seeing. But no one effectively made an empirical case that this was more likely than not.

While I don't agree that IFRs in the neighborhood of 1.0% are completely out of the picture (see antibody studies), I'd be thrilled if the CDC's latest estimates based on scenario testing are accurate. Because it probably means fewer deaths and opening up sooner. Who could be against that? But what does that mean in the context of three months ago?

It means that the cost of inaction was merely hundreds of thousands of US lives instead of over 1 million.

You see similar, anti-intellectual attitudes in elections polling. People reduce a very complicated system to, "Polls said Clinton would win/Brexit would be defeated. Polls are useless." I would love to make wagers with these people. We put money on 100 election results. They can leverage media coverage and other heuristics, but must black out polling data. I get to use the polling. I would clean up.

Similarly, if you're offering me the practice of statistical inference and empirical modeling, I'll take it every damn time.

When scientists publish research or white papers or build models, they open themselves up to criticism. As they should. They are and should be held to a high (but not unreasonable) standard. When people make comments on the internet about how this is no deadlier than the flu, or talk about how Swine Flu killed more people, they're not opened up to very much critics because no one holds them to any standard or expectations as at all. I'm eternally grateful when public policy is driven by the former, rather than the latter.

Bloop Bloop

  • Handlebar Stache
  • *****
  • Posts: 2140
  • Location: Melbourne, Australia
Re: How long can we wait while flattening the curve?
« Reply #2223 on: May 27, 2020, 02:57:58 AM »
I think the mandated lockdown response feeds into left wing ideology.  The idea that we are all equal, the response must be uniform, because we are all equal and not different.  As more is acknowledged about the virus nothing can be further from the truth, reasonably healthy children have almost no risk from CoVID-19, while frail people in long-term care facilities have a fatality rate that could be close to 50%.  But, that acknowledgement comes with a lot of cognitive dissonance for people that think we are all equal and that the ends and result must also be equal. 

As humans we are able to fix problems and make things better and the government has the means to do it through mandated and uniform policy.  That's a great idea, but viruses don't work in the framework of an egalitarian mindset.

Spot on.

+1

Bloop Bloop

  • Handlebar Stache
  • *****
  • Posts: 2140
  • Location: Melbourne, Australia
Re: How long can we wait while flattening the curve?
« Reply #2224 on: May 27, 2020, 03:04:15 AM »
Cheap, emotionally leveraged second-guessing on COVID comprises almost the entirety of T-Money's contributions to this forum. I'm a pretty polite guy, but frankly, I've been working on COVID-19 for months and I'm losing patience for it. It's almost too delicious that the very first words he had on this topic were in regards to the death toll of H1N1, a take that has aged like milk.

For weeks, I've been encouraging those who think the reaction is overblown to get into the offices of decision makers to present competing analysis. They will never ever ever ever ever do this  though because their interest in this topic only goes as far as trying to be right on an Internet forum.

What do you want us to do, telephone the prime minister? Anyway, the points I've been making (1 - we need to adjust our lockdown timeframe (TICK) - 2 - we need to dial down the apocalyptic rhetoric (TICK) - 3 - we need to be a lot more attuned to the economic cost (TICK) - have all now been actioned in the last 3 weeks. Obviously had nothing to do with me saying it, but it seems like the Aussie authorities have been willing and able to change things on the fly; as soon as it was clear that this thing was a lot less of a problem than first thought, within about 2 weeks they changed tune.

And I think we can thank a lot of the commentators in the media and other "contrarians" who saw the flattening of the curve, saw the minimal death rate particularly in those under 60, and started asking the hard questions about what price we should really be paying.

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2225 on: May 27, 2020, 03:24:34 AM »

My point is that we used a very flawed model (Ferguson's Imperial College) to base our major policy decisions on.  The entire model assumed an IFR of 0.9% and this is why it predicted as many as 2.2 million Americans would die.  This goes back to my point that we need to base our policy decisions on better data now that we have it.  We needed a good estimation of IFR, and we finally have that.  A model is only as good as the inputs, and we had the incorrect inputs.
....

"Although ICL only released scenarios and associated forecasts for the United Kingdom and United States, its model is adaptable to any country by changing the inputs to reflect its population, demographics, and the date its specific policies took effect. In early April around the peak of the academic community’s backlash against the Swedish government’s strategy, a group of researchers at Uppsala University attempted to do just that. They released an epidemiological model for Sweden that adapted the ICL COVID-19 model from Ferguson and his colleagues, and attempted to project the effects of Sweden’s unique response on both hospital capacity and total fatalities."

They applied Ferguson's model to Sweden.  The reason it was off by so much is that it uses an IFR input of 0.9%.

Our modelleres have upped their assumption for the IFR from 0.3% to 0.7% based on local data and recent antibody studies. This is, as everywhere very sensitive to the assumption on how many have actually been infected, but that number (the denominator) was recently lowered.

Pooplips

  • Bristles
  • ***
  • Posts: 463
  • Age: 34
Re: How long can we wait while flattening the curve?
« Reply #2226 on: May 27, 2020, 04:32:16 AM »
What do you want us to do, telephone the prime minister? Anyway, the points I've been making (1 - we need to adjust our lockdown timeframe (TICK) - 2 - we need to dial down the apocalyptic rhetoric (TICK) - 3 - we need to be a lot more attuned to the economic cost (TICK) - have all now been actioned in the last 3 weeks. Obviously had nothing to do with me saying it, but it seems like the Aussie authorities have been willing and able to change things on the fly; as soon as it was clear that this thing was a lot less of a problem than first thought, within about 2 weeks they changed tune.
 

Those are 3 good points. I had not realized Australia had been adjusting so quickly.

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2227 on: May 27, 2020, 04:54:39 AM »
Our R number is still well below 1 according to the models after gradually reopening after a "lockdown" that was quite lax by international standards, albeit more strict than neighboring Sweden. More interesting, however, is that R was estimated to be just slightly above 1 the week before our version of a shutdown in mid-march. The morale from that is that pretty basic measures, like keeping distance and good hand hygiene goes a long way and it isn't really necessary to close everyhting and force everyone to stay at home.

Here the effect of social distancing and our partial shutdown was way more effective than anyone could have imagined. The talk was earlier to flatten the curve but now it looks more like keeping new infections around the current stupidly low rate - we now average around 15 new cases / day which adjusted for population would mean around 900 / day in the US and we currently have way more test capacity than we could possibly use so it's not like there are large numbers of sick people who does not get tested if they want to. We have the royal amount of 11 people in intensive care (672 scaled to US population) and death toll is 238 (15.700 US), and that includes all cases, also those outside hospitals in places like care homes - the latter group is the majority of deaths.

The experience might not transfer fully to other countries, Norway has a relatively healthy population on the younger side of thing (median age around same as US), and by US standards we only have 1 place qualifying as a city, and that would not make top 20 in the US populationwise so the population is fairly spread out. There are many parts of the country where number of infected people is now estimated to be zero.

Kyle Schuant

  • Handlebar Stache
  • *****
  • Posts: 1297
  • Location: Melbourne, Australia
Re: How long can we wait while flattening the curve?
« Reply #2228 on: May 27, 2020, 05:03:50 AM »
I went to the hospital tonight for a minor injury, I wanted my child to keep me company. My child was turned away and sent home, "Because it's too much risk, children can pass covid on." But schools have been reopened because, we are told, there's no risk. There seems to be some confusion here.

beltim

  • Magnum Stache
  • ******
  • Posts: 2964
Re: How long can we wait while flattening the curve?
« Reply #2229 on: May 27, 2020, 05:12:59 AM »
Is it fair to say, simultaneously:

-2-3 months ago when we didn't know much about this, it was worth freaking out.

-Now, when we know more, the freakout was probably an overreaction. Very elderly people in nursing homes dying is horrible, but it's probably not worth ruining the economy, if we're going to be rational about it.

Unfortunately this is now a partisan issue, so half of everyone won't admit the latter, and half won't admit the former.

-W

If your premise is true, that would be reasonable.  But the disease doesn't just kill the very elderly in nursing homes. Even among 40-49 year olds, the estimated fatality rate is still about 0.4%. 

Spud

  • Stubble
  • **
  • Posts: 161
  • Location: Southwest England, UK.
Re: How long can we wait while flattening the curve?
« Reply #2230 on: May 27, 2020, 05:20:53 AM »
@habaneroNorway I'd be really interested to learn more about the Norwegian lockdown. We hear a lot about how much better than Sweden the likes of Norway and Finland have done, but what we're never told, is what are the differences between the measures imposed by Norway and Sweden. I'm aware that I've fallen into the trap of thinking that Norway's numbers look really good, so therefore they must have locked down really early and really hard. They must have closed schools and nursing homes and their entire economy. I realise however, that may not be the case.

Please can you share your perspective as a citizen of Norway? Has lockdown been strict? Are there severe penalties if people do not comply? Or are they trusted to comply and left to their own devices without any penalties?

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2231 on: May 27, 2020, 05:20:58 AM »
Even among 40-49 year olds, the estimated fatality rate is still about 0.4%.

No, its not. Your graph lists the case fatality rate which has only confirmed cases in the denominator. For that age cohort our modelers use a hospitalization rate of 0.17% of all infected, meaning a death rate pretty close to zero as most who go to the hospital does not go to the ICU and for that age cohort the ICU survival rate is around 90%

beltim

  • Magnum Stache
  • ******
  • Posts: 2964
Re: How long can we wait while flattening the curve?
« Reply #2232 on: May 27, 2020, 05:26:52 AM »
Even among 40-49 year olds, the estimated fatality rate is still about 0.4%.

No, its not. Your graph lists the case fatality rate which has only confirmed cases in the denominator. For that age cohort our modelers use a hospitalization rate of 0.17% of all infected, meaning a death rate pretty close to zero as most who go to the hospital does not go to the ICU and for that age cohort the ICU survival rate is around 90%

It only has confirmed deaths from COVID-19 in the numerator also.  I would believe 0.2% in that age group as well.  I disagree that that's close to 0.  That would mean, if unchecked, more deaths in that age group than from all other causes combined in a given year.  Are you really going to try to argue that "pretty close to zero" 40-49 year olds die?

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2233 on: May 27, 2020, 05:36:54 AM »
Please can you share your perspective as a citizen of Norway? Has lockdown been strict? Are there severe penalties if people do not comply? Or are they trusted to comply and left to their own devices without any penalties?

It has been the strictest since WW2 (which doesnt really say much as its the only one since then).

Yes it was pretty strict. Schools closed. People worked from home in vast numbers (but there was not really any law stating that, it was also partly a function of schools / day care closed so parents didn't have much choice). Ski resorts closed, no large gatherings (down to max 5, now its 20, soon 50 I think), restaurants in the capital had to pretty much close as they couldn't sell alcohol (now lifted), pubs (i.e. non-food-places) will reopen on Monday. Kids activities, sports etc all closed, now partly reopened. And the list goes on. But you were never forced to stay in your own home, could go wherever you wanted (a short ban om cabin visits during the easter school holiday, however) and domestic travel is mildly discouraged but was never forbidden except for the cabin thing.

There were some fines, but think just a very small handful were handed out, think it was about 2000 US for breaking quarantine after being confirmed infected. No shops had to close, some did due to lack of customers. Malls stayed open.

The current situation is pretty similar between Norway and Sweden. People have this rather strange concept of Sweden being a Covid-19-anarchy, but the swedes work from home in vast numbers since March, keep their distance when out, lots of stuff has closed or had to close as no customers. The main part separating Norway and Sweden was really schools and chilldren's activiteis and for some time also bars/restaurants. Now we are getting closer to the swedish apporach, but we are on a completely different part of the epidemic curve so the numers are much smaller. For Sweden the main "action" is also in Stockholm, which is much larger than Norway's capital. If you take out Stockholm all swedish numbers drop to roughly half. And also in Sweden the numer of new patients admitted to ICU has been steadily declining for weeks which doesnt really come acrooss in the international press coverage.

There is also a cultural component. In general, Swedes have a lot of respect for authorities, so it's generally sufficient to encourage people to do this and that and not that and most will comply, in Norway it has been slightly more focus on rules and regulations, but in the end it is not terribly different, at lest less than the impression one can get from the press.

At the end of the day a very important factor is that people in the Scandinavian countries have a lot of faith in their governments and are not rebellious + the social welfare net is quite generous. Yes, there has been lots of complaints from people bearing the financial burden (lowly-paid private sectore employees in general, think tourism, bars/restaurants etcetc) and vast numbers have been and still are on government support as they are temporarily suspended from work (which legally is different than being fired), but the numbers have started going down as more and more return to work.

« Last Edit: May 27, 2020, 05:51:34 AM by habaneroNorway »

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2234 on: May 27, 2020, 05:45:22 AM »
It only has confirmed deaths from COVID-19 in the numerator also.  I would believe 0.2% in that age group as well.  I disagree that that's close to 0.  That would mean, if unchecked, more deaths in that age group than from all other causes combined in a given year.  Are you really going to try to argue that "pretty close to zero" 40-49 year olds die?

Our health folks have a hospitalization rate of 0.17% for that age cohort of which around 15% go to the ICU of which 90% survive. That should yield a probability of dying of 0.00255% if my math is correct, so yes, pretty much zero. Even if the number is off by an order of magnitude it's still much lower than the risk of dying from something else any given year (around 0.2% for that age group).

After the latest model update based on local numbers Covid-19 is more dangerous for the elderly than previously assumed and less so for the younger. For the US the numbers would probably be higer as public health is much worse and we have stuff like free, universal access to high-quality health care and our hospitals has never been stressed by a tsunami of patients. Norway's ICU survival numbers apparantly are among the highest in the world, probably for said reason, plus the fragile elderly in care homes never gets sent to the ICU in the first place.
« Last Edit: May 27, 2020, 05:49:26 AM by habaneroNorway »

beltim

  • Magnum Stache
  • ******
  • Posts: 2964
Re: How long can we wait while flattening the curve?
« Reply #2235 on: May 27, 2020, 05:55:30 AM »
It only has confirmed deaths from COVID-19 in the numerator also.  I would believe 0.2% in that age group as well.  I disagree that that's close to 0.  That would mean, if unchecked, more deaths in that age group than from all other causes combined in a given year.  Are you really going to try to argue that "pretty close to zero" 40-49 year olds die?

Our health folks have a hospitalization rate of 0.17% for that age cohort of which around 15% go to the ICU of which 90% survive. That should yield a probability of dying of 0.00255% if my math is correct, so yes, pretty much zero. Even if the number is off by an order of magnitude it's still much lower than the risk of dying from something else any given year (around 0.2% for that age group).

After the latest model update based on local numbers Covid-19 is more dangerous for the elderly than previously assumed and less so for the younger. For the US the numbers would probably be higer as public health is much worse and we have stuff like free, universal access to high-quality health care and our hospitals has never been stressed by a tsunami of patients. Norway's ICU survival numbers apparantly are among the highest in the world, probably for said reason, plus the fragile elderly in care homes never gets sent to the ICU in the first place.

Why don't you just get actual numbers of cases and deaths?  Your multiplication strategy leaves out of a lot of cases - people who die without being tested, before going to the hospital, or without going to the ICU.  Given that your estimate is two orders of magnitude lower than the data from many countries actual data, I hope you'll understand why I trust said actual data rather than your unsourced estimates.

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2236 on: May 27, 2020, 06:10:54 AM »
Your multiplication strategy leaves out of a lot of cases - people who die without being tested, before going to the hospital, or without going to the ICU.  Given that your estimate is two orders of magnitude lower than the data from many countries actual data, I hope you'll understand why I trust said actual data rather than your unsourced estimates.

No, as said, Norway has more test capacity than we could possibly put to any meaningful use now and we count every single death (which means also those outside hospitals) even if a small handful slip through the system it doesn't total to much. The multiplication might be off, but that's the estimate they use and is the best number we have.

If we look at the confirmed cases in Norway, for 40-49 we have 4 deaths and 1500 confirmed cases, yielding the more common 0.2%. We know this is an abseloute upper bound as the nominator is most likely fully accurate (noone dies under the radar here, especially from Covid-19 these days) while the denominator is higher as people with mild (or no) symptoms were not tested in the early stages due to test capacity (which is no much higher than needed). Given how low the nominator is, however, you can't really calculate much based on this number as the uncertainty is very large and random events might have pushed it higher or lower than the long-term equilibrium rate.

They estimate that there has been roughly 4 times as many cases as confirmed in Norway, so adjusting for that we get to 0.07% IFR for age 40-49. Which I still count as "close to zero" despite being siginficantly higher than the multiplication I did (about 25 times higher).

For the group below, 30-39 we have 0 deaths so hard to calculate a ratio then, but the IFR is likely to be very low for anyone below 40 at least. Ditto for 0-9,10-19 and 20-29, all zero deaths. But Norway's total death toll is only 240 people in total so they are not the most reliable numbers.

For comparison, the country with by far the best track on Covid-19 is Iceland and they have a case fatality rate of 0.5% across the entire population (albeit only 10 deaths so same caveat applies there).
« Last Edit: May 27, 2020, 06:26:51 AM by habaneroNorway »

beltim

  • Magnum Stache
  • ******
  • Posts: 2964
Re: How long can we wait while flattening the curve?
« Reply #2237 on: May 27, 2020, 06:25:07 AM »
So you’re doubling down on the death rate of 40-49 year olds from all causes  being close to zero. Got it

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2238 on: May 27, 2020, 06:30:28 AM »
So you’re doubling down on the death rate of 40-49 year olds from all causes  being close to zero. Got it

I see your point, but my point is that for me (age 40-49) Covid-19 seems to be about as dangerous as life itself, probably less so, especially for me personally (no underlying conditions). If you multiply a small percentage with a large number the output can be a large number, but for me personally, Covid-19 isn't anything I freak out about, nor should, in my opinion, anyone in my age group unless they have one or more underlying factors skewing the odds by a large amount.

When I checked the numbers for my probability of dying based on my age it was higher than i would have guessed, but still not something I freak out about as the number is most likely much lower for me personally as I am in good health and don't have any meaningful risk for a workplace accident etc. But still bad shit can happen to anyone, also from Covid-19 despite being young & healthy, but the probability of it is very low in my opintion. So guess at the end of the day we disagree on what's "low" then...  One of our deaths in the discussed age group (40-49) was one of those, he was fit and healthy but passed away in the ICU.
« Last Edit: May 27, 2020, 06:34:35 AM by habaneroNorway »

OtherJen

  • Magnum Stache
  • ******
  • Posts: 3602
  • Location: Metro Detroit
Re: How long can we wait while flattening the curve?
« Reply #2239 on: May 27, 2020, 06:37:48 AM »
So you’re doubling down on the death rate of 40-49 year olds from all causes  being close to zero. Got it

I see your point, but my point is that for me (age 40-49) Covid-19 seems to be about as dangerous as life itself, probably less so, especially for me personally (no underlying conditions). If you multiply a small percentage with a large number the output can be a large number, but for me personally, Covid-19 isn't anything I freak out about, nor should, in my opinion, anyone in my age group unless they have one or more underlying factors skewing the odds by a large amount.

When I checked the numbers my probability of dying based on my age was higher than i would have guessed, but still not something I freak out about as the number is most likely much lower for me personally as I am in good health and don't have any meaningful risk for a workplace accident etc. But still bad shit can happed to anyone, also from Covid-19 despite being young & healthy, but the probability of it is very low in my opintion. So guess at the end of the day we disagree on what's "low" then...

As someone in the same age bracket with no known risk factors, I’m less concerned about dying now from COVID-19 and more concerned about 1) losing my clients if I become too ill to work for weeks and 2) suffering long-lasting lung, heart, kidney, and/or CNS damage for the rest of my life.

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2240 on: May 27, 2020, 06:42:41 AM »
2) suffering long-lasting lung, heart, kidney, and/or CNS damage for the rest of my life.

That is also a point that gets a bit lost these days. There are several stages between "fit and healthy" and "dead" that are not desirable, and there are some signs of lasting problems among those who survive. One dude here lost 70% of his lung capacity and was now in a training program normally used for people suffering from severe COPD.

Kyle Schuant

  • Handlebar Stache
  • *****
  • Posts: 1297
  • Location: Melbourne, Australia
Re: How long can we wait while flattening the curve?
« Reply #2241 on: May 27, 2020, 06:48:51 AM »
Some years back I was involved in the peak oil scene. The interesting thing is that most of them were climate change denialists.

I went over to a climate change modelling forum, and asked if they'd modelled for the decline in the availability of fossil fuels, since having less to burn overall might mitigate things a bit, and at the same time, the declining energy return on energy invested might mean more emissions per unit of oil burned. They handwaved that away and said their model assumed fossil fuels were infinite, and that this was good enough on the timescales they were using.

Now come to this virus, and on the one hand we have "models" consisting of "if everyone in a population of X gets infected and the death rate is Y% then X x Y% will die," and on the other hand people like the psychologists here in Australia who claimed we'd get a 50% rise in the suicide rate, or the people here engaged in statistical shenanigans to try to handwave one hundred thousand dead.

And then of course there's probably someone who thinks it's all just toxic masculinity.

People seem to suffer from a Problem Exclusion Principle, so that once one problem is acknowledged as real, they have to deny or dismiss any other problems as either irrelevant, or at best simply a problem derived from the One True Problem. Unfortunately the truth is that the world has more than one problem, and we can never solve them all, we just balance one issue against another.

No more shenanigans.

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2242 on: May 27, 2020, 07:17:15 AM »
people who die without being tested, before going to the hospital, or without going to the ICU.

A small comment on this, for someone living in the US (don't know if you do...) it can probably be a bit hard to grasp just how much better control the government has on things in a small non-federalist country where approacahes, standards, routines, reporting systems etcetc are the same everywhere and national and the public sector is highly digitalized. And very, very few people live "under the radar" which is very hard in a country like Norway. This is also a part of the explanation for Sweden's high death toll relative to population - they count every single Covid-19-death they are aware of regardless of where it happens. There is a siginficant lag in some of the reporting, but after a week or so the swedish figures are about 90% up-to-date while some earlier deaths do trickle in a week or two later. This is partly due to never really having a need for instant reporting of deaths in elderly care homes (after all, it's pretty much the last stop before death anywaay) so there was no real reporting system in place for something like this.

You have probably seen the graphs of "excess deaths" from various places the last month which is used as an estimate of the actual number of Covid-19-deaths and to gauge the unreported deaths. The norwegian weekly death numbers for the last 2 months are lower than the mean, the explanation given is that this year's regular flu season was an unusually mild variety and it was effectively killed off completely due to Covid-19-related social distancing etc. And also with much lower activity there are less workplace accidents and so on. And Covid-19-deaths (so far) represent a very small number, 240 total compared to an underlying annual death rate of roughly 40.000 people / year for the country. If you took out a few elderly care homes the numbers would be much lower due to local outbreaks in such facilities, so it's bit bad luck as well, but that is doomed to happen in a situation like this I guess.

beltim

  • Magnum Stache
  • ******
  • Posts: 2964
Re: How long can we wait while flattening the curve?
« Reply #2243 on: May 27, 2020, 07:25:07 AM »
people who die without being tested, before going to the hospital, or without going to the ICU.

A small comment on this, for someone living in the US (don't know if you do...) it can probably be a bit hard to grasp just how much better control the government has on things in a small non-federalist country where approacahes, standards, routines, reporting systems etcetc are the same everywhere and national and the public sector is highly digitalized. And very, very few people live "under the radar" which is very hard in a country like Norway. This is also a part of the explanation for Sweden's high death toll relative to population - they count every single Covid-19-death they are aware of regardless of where it happens. There is a siginficant lag in some of the reporting, but after a week or so the swedish figures are about 90% up-to-date while some earlier deaths do trickle in a week or two later. This is partly due to never really having a need for instant reporting of deaths in elderly care homes (after all, it's pretty much the last stop before death anywaay) so there was no real reporting system in place for something like this.

You have probably seen the graphs of "excess deaths" from various places the last month which is used as an estimate of the actual number of Covid-19-deaths and to gauge the unreported deaths. The norwegian weekly death numbers for the last 2 months are lower than the mean, the explanation given is that this year's regular flu season was an unusually mild variety and it was effectively killed off completely due to Covid-19-related social distancing etc. And also with much lower activity there are less workplace accidents and so on. And Covid-19-deaths (so far) represent a very small number, 240 total compared to an underlying annual death rate of roughly 40.000 people / year for the country. If you took out a few elderly care homes the numbers would be much lower due to local outbreaks in such facilities, so it's bit bad luck as well, but that is doomed to happen in a situation like this I guess.

But your multiplication result is so far away from the actual number that there is an error there.  You gave a good response on why the first isn't likely, but there's still the other two.  Plenty of people die in hospitals who aren't in the ICU, I assure you.  And plenty of people die outside hospitals.  Nevertheless, I gave those as possibilities for where the error might be, but honestly the most likely source of error is the completely nonsensical hospitalization rate you used to start with.  Does Norway really think the hospitalization rate is lower than the death rate observed by many other countries?  If so, do you really have confidence in that estimate? 

Davnasty

  • Magnum Stache
  • ******
  • Posts: 2582
Re: How long can we wait while flattening the curve?
« Reply #2244 on: May 27, 2020, 07:43:08 AM »

Once again, citations? It was never given as an estimate by the WHO, it was given as the current CFR and was correct. Who used this as an estimate?

ETA: The headline from your citation is enough to disprove your interpretation. You turned "crudely calculated snapshot" into crude estimate". You made up the estimate part entirely.

https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(20)30244-9/fulltext

Could you tell me what you think this article says? Because I see nothing about using 3.4% as an estimated IFR.

Quote

It was, but this conversation began with your critique of the other study. That's the one that you cited to show Neil Ferguson was wrong even though he did not author that paper.

My point is that we used a very flawed model (Ferguson's Imperial College) to base our major policy decisions on.  The entire model assumed an IFR of 0.9% and this is why it predicted as many as 2.2 million Americans would die.  This goes back to my point that we need to base our policy decisions on better data now that we have it.  We needed a good estimation of IFR, and we finally have that.  A model is only as good as the inputs, and we had the incorrect inputs.

Honestly, I don't think that was you point. It was pretty clear from context that you mixed up the two studies and now you won't admit to it. I don't think you're arguing in good faith.

Also you're falling back on the assumption that it was a very flawed model, but none of the evidence you've cited for that conclusion has been correct. Even if it turns out an IFR of .9% was high, it wasn't off by an order of magnitude like you claimed. More importantly, in the do nothing scenario where the model predicted 2.2 million deaths, the IFR would have been much higher than our current estimates due to hospitals being severely overwhelmed.

Quote

Ferguson wasn't even an author on that paper. The model type was based on his research, he was only a reference.


"Although ICL only released scenarios and associated forecasts for the United Kingdom and United States, its model is adaptable to any country by changing the inputs to reflect its population, demographics, and the date its specific policies took effect. In early April around the peak of the academic community’s backlash against the Swedish government’s strategy, a group of researchers at Uppsala University attempted to do just that. They released an epidemiological model for Sweden that adapted the ICL COVID-19 model from Ferguson and his colleagues, and attempted to project the effects of Sweden’s unique response on both hospital capacity and total fatalities."

They applied Ferguson's model to Sweden. The reason it was off by so much is that it uses an IFR input of 0.9%.

Is that why it was off by so much*? How do you know those researchers used the models correctly? Their paper has not been peer reviewed. Have you read it and confirmed that they used the model correctly?

And if that is the reason and the real IFR is closer to .3%, shouldn't their prediction have been 3x high, not 10x high?

*I still haven't read the paper and given your track record of interpreting models, I suspect the conditions for that projection were not met, which again is a critical component in modeling.
« Last Edit: May 27, 2020, 07:48:03 AM by Davnasty »

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2245 on: May 27, 2020, 08:00:29 AM »

But your multiplication result is so far away from the actual number that there is an error there.  You gave a good response on why the first isn't likely, but there's still the other two.  Plenty of people die in hospitals who aren't in the ICU, I assure you.  And plenty of people die outside hospitals.  Nevertheless, I gave those as possibilities for where the error might be, but honestly the most likely source of error is the completely nonsensical hospitalization rate you used to start with.  Does Norway really think the hospitalization rate is lower than the death rate observed by many other countries?  If so, do you really have confidence in that estimate?

Well, as said, I'm pretty confident that we have very close to 100% control of everyone who died from Covid-19, mostly because there are so few of them so not that hard to keep track of. I don't see any reason why plenty of people would die in the wards in Norway. Firstly, the ICUs have never been even remotely close to capacity, and secondly the elderly in care homes etc are most often not sent to the hospital at all. There might be some old people who are judged not to be able to get any benefit from and/or endure ICU treatment, but again, our hospitalization numbers are so low that's not a big source either.

I don't know what the actual hospitalization rate is, but I know hat most, if not all countries, does not have a clue on how many people have actually been infected. As for the rate mentioned (0.17%) my hunch says its on the low side, but it's also hard to fact-check due to the very low number of deaths in that age group and the denominator is also a big unknown. I do, however think that for norwegian figures it's best to use norwegian data and not infer from other countries. Our public health authorities pointed this out when they recalibrated their models and finally could use domestic data. There are major differences in population mix, public health and other factors which might render data from other places less relevant for local use.

Edit: I dug a bit deeper into the data and hereby change my stance to it's most likely quite a bit too low, which I assume its because it's a data point where the curve radically shifts shape. Thx for challenging it!

And its a model input used to fit data to actual hospitalizations, so it's a estimate after all. It can probably be higher, and even a lot higher, but it still does not mean it's a very dangerous disease for the young and healthy among us.

Here are the parameters they use for hospitalization rate of infections to get best fit to actual hospitalizations. As you can see the 40-49y is right on where the curve really shifts upwards so its probably quite sensitive to a few cases more or less.

0-9y      0.02%
9-19y    0.02%
20-29y  0.06%
30-39y  0.13%
40-49y  0.17%
50-59y  3.5%
60-69y  7.1%
70-79y 11.3%
80y+    27%

Above 70 the rate is reduced as many in those age groups never gets sent to the hospital as pointed out above.
Currently they estimate that 15% of hospitalizations need ICU treatment.
The survival rate of 90% is not in the report, but Sweden has similar numbers for their ICU patients in that age group and thats based on many more cases.

In all, this model gives an overall IFR of 0.7% but it's extremely skewed towards the older population. 65% of actual deaths are >80y. If you include 70-79 you get 87% of all deaths.
« Last Edit: May 27, 2020, 08:07:06 AM by habaneroNorway »

mathlete

  • Handlebar Stache
  • *****
  • Posts: 1638
Re: How long can we wait while flattening the curve?
« Reply #2246 on: May 27, 2020, 08:06:28 AM »
I don't even feel the need to die on the hill of the Imperial study. The conclusions were what they were. Caveats, conditions and all. That was one paper of many. One model of many. It drives me up a wall that we're sitting here putting the word "model" in scare quotes.

It's crazy to me that we're here defending the practice of statistical inference, and using methods to quantify assumptions. What is the alternative? Make no assumptions? That can't happen. I suppose you could have policy makers, as a practice, ignore all statistical inference, but the policy makers would themselves be making assumptions anyway. And they'd probably be really bad assumptions. And since we're putting model in scare quotes, we'd have no mathematical method for quantify what these bad assumptions mean in order to support decision making.

We could wait for perfect data to materialize. But we'd be waiting forever because perfect data is a mythical concept. And in the mean-time, we're going to lose hundreds of thousands more lives.

We're dealing with a novel and emerging risk. Data is limited, and that sucks. But statistical inference and mathematical modeling identified this as a virus that is many times deadlier than the flu and capable of causing lots of excess mortality. That's what motivated policy decisions. And even with pretty extreme measures taken, COVID-19 still became the leading cause of death in the US for weeks. That sure sounds like a novel emerging risk capable of causing lots of excess mortality to me.

The research and the modeling did its job. So now it becomes about whether the lives saved are worth it. I'd be glad to entertain discussion on that, but I'd like to see quantification and assumption setting for quantifying the cost that is similar in rigor to the work done on researching the virus and making forecasts and projections.

the_fixer

  • Pencil Stache
  • ****
  • Posts: 899
  • Location: Colorado
  • mind on my money money on my mind
Re: How long can we wait while flattening the curve?
« Reply #2247 on: May 27, 2020, 08:18:52 AM »
So you’re doubling down on the death rate of 40-49 year olds from all causes  being close to zero. Got it

I see your point, but my point is that for me (age 40-49) Covid-19 seems to be about as dangerous as life itself, probably less so, especially for me personally (no underlying conditions). If you multiply a small percentage with a large number the output can be a large number, but for me personally, Covid-19 isn't anything I freak out about, nor should, in my opinion, anyone in my age group unless they have one or more underlying factors skewing the odds by a large amount.

When I checked the numbers my probability of dying based on my age was higher than i would have guessed, but still not something I freak out about as the number is most likely much lower for me personally as I am in good health and don't have any meaningful risk for a workplace accident etc. But still bad shit can happed to anyone, also from Covid-19 despite being young & healthy, but the probability of it is very low in my opintion. So guess at the end of the day we disagree on what's "low" then...

As someone in the same age bracket with no known risk factors, I’m less concerned about dying now from COVID-19 and more concerned about 1) losing my clients if I become too ill to work for weeks and 2) suffering long-lasting lung, heart, kidney, and/or CNS damage for the rest of my life.
Plus the possibility of outrageous bills from a hospital stay or god forbid in ICU that could destroy years of savings.

As someone in that 40 - 49 age group I fully understand my chances of dying are low, but the possibility of additional health issues for the rest of our lives and financial concerns are what keeps me from throwing caution to the wind.

In Colorado the age group of 40 - 49 accounts for 17.25% of positive cases the highest in the state.


Sent from my iPhone using Tapatalk

OtherJen

  • Magnum Stache
  • ******
  • Posts: 3602
  • Location: Metro Detroit
Re: How long can we wait while flattening the curve?
« Reply #2248 on: May 27, 2020, 08:29:43 AM »
So you’re doubling down on the death rate of 40-49 year olds from all causes  being close to zero. Got it

I see your point, but my point is that for me (age 40-49) Covid-19 seems to be about as dangerous as life itself, probably less so, especially for me personally (no underlying conditions). If you multiply a small percentage with a large number the output can be a large number, but for me personally, Covid-19 isn't anything I freak out about, nor should, in my opinion, anyone in my age group unless they have one or more underlying factors skewing the odds by a large amount.

When I checked the numbers my probability of dying based on my age was higher than i would have guessed, but still not something I freak out about as the number is most likely much lower for me personally as I am in good health and don't have any meaningful risk for a workplace accident etc. But still bad shit can happed to anyone, also from Covid-19 despite being young & healthy, but the probability of it is very low in my opintion. So guess at the end of the day we disagree on what's "low" then...

As someone in the same age bracket with no known risk factors, I’m less concerned about dying now from COVID-19 and more concerned about 1) losing my clients if I become too ill to work for weeks and 2) suffering long-lasting lung, heart, kidney, and/or CNS damage for the rest of my life.
Plus the possibility of outrageous bills from a hospital stay or god forbid in ICU that could destroy years of savings.

As someone in that 40 - 49 age group I fully understand my chances of dying are low, but the possibility of additional health issues for the rest of our lives and financial concerns are what keeps me from throwing caution to the wind.

In Colorado the age group of 40 - 49 accounts for 17.25% of positive cases the highest in the state.


Sent from my iPhone using Tapatalk

Michigan is similar. That age group accounts for 16% of all cases and is tied for second place with those aged 60-69 years. The most cases have been reported in those aged 50-59 years (18%).

From the utilitarian perspective that seems to be favored by many, even if a fraction of those survivors develop significant long-term medical issues as a consequence of COVID-19, that's still going to place significant burdens on the economy, society, and healthcare systems.

habanero

  • Pencil Stache
  • ****
  • Posts: 668
Re: How long can we wait while flattening the curve?
« Reply #2249 on: May 27, 2020, 08:31:46 AM »
So now it becomes about whether the lives saved are worth it. I'd be glad to entertain discussion on that, but I'd like to see quantification and assumption setting for quantifying the cost that is similar in rigor to the work done on researching the virus and making forecasts and projections.

The current wisdom in my part of the world is that the most economically sound thing to do is to hit the virus really hard and then try and maintain a low rate of new infections (as in not only flatten the curve but to really suppress it). This is, as everything else a conclusion reached with a ton of uncertain parameters. but the experience so far seems to be that it's possible to maintain a low infection rate without draconian measures or a total surveillance of the population in Asia-style. Part of the input for this conclusion is putting a monetary value on quality-adjusted life-years saved so it can come across as a bit unsentimental - but it is a big modelling job behind it. If it proves to be correct in the long run is another matter, of course. It is also, at least for the time being, the most acceptable approach in the electorate, but that might change if the economic pain becomes long-lasting and hard. We also have a large part of the workforce in the public sector where everyone has kept their job and pay, a factor which means for a lot of employed, someone else is suffering the actual pain from job loss and/or reduced pay.