Coronavirus

[jhm]

I found this webpage today, seems pretty interesting. I guess we'll see how well the model holds up. I haven't read through all of their assumptions. Obviously, depending on what is done to restrict travel between states, it will dramatically impact the results even at this point. (I don't know whether their upper and lower bounds account for some of this uncertainty - which is huge.)

Looks like peak deaths on ~April 14th, with integrated deaths around ~81k. Lots of nice plots you can scroll around, etc. Also has state by state breakdown so you can see the delays on peaks in different states. I should note that I've seen models showing a much more optimistic model for Oregon, for example - so I hope that this is all way too pessimistic.

IHME | COVID-19 Projections

All pretty horrifying. I hope they are wrong, though these numbers were within my expectations. Aligns fairly well with my wild guess of a few days ago of ~10,000 deaths by the end of next week. I'd say peak fear will probably show up next week!

Seems very unlikely that the lockdown will be able to stop until the end of May - and then only in some states. Note that practically, the upper and lower bounds don't make much difference (maybe 1-2 weeks) to when we can "stop." (Whatever "stop" means.) I guess the good news is we should have plenty of time to plan the exit strategy! "Plenty" when it comes to this administration may not be enough time though.

View attachment 526599

I really like that there is a predictve envelop in this chart. It is so important to get a feel for the range of possibility and degree of uncertainty.

 

[HG Wells]

How do you think the market will respond when we reach 2000 deaths per day ?
Is it already baked in ?

 

[Norbert]

I would strongly suggest that we not get hung up on debating or trying to estimate IFR or CFR. None of us in this forum have the data or experience to do this properly. We just wind up arguing about things we can't see.

I have found in my own experience as a statistician that it is very important to stay focused on things you can actually measure and get data for. Latent variables, censored observations and missing data can easily send one down imaginative paths that lead nowhere. What we often find is the we must make many more assumptions than we can test. We ultimately impose our own subjective biases as we select which results are worth sharing with others.

The when we share this with others we can run into conflict. But we wind up arguing not about observable facts, but over all the imaginative stuff we have no data. If we had the data, we wouldn't need to waste much time arguing.

So this is a clue, if you can't resolve an argument with someone, chances are you are not arguing about somethings you can actually quantify or gather good data for, or else someone in the argument is unwilling to yield to the data which is on hand. Past a certain point you are no longer engaged in science. You are probably just wasting time.

Among the tactics of "Merchants of Doubt," propagandists who try to undermine science, is the tactic the red herring. Get people to debate non-essential things for which available data cannot resolve. Often this is used to destroy the credibility of data that would answer critical questions. It is always possible to bash data sources. You can ask millions of questions about a given source for which you can't easily resolve with other sensible data. For example, climate gas lighters love to engage people in questions about statistical adjustments to raw data. This gets most people way out of their depth in a hurry. What is sensible data cleaning and structuring to disciplined scientist doing the adjustments becomes and endless litany about "what about this issue, what about that," which just confuses and bewilders the propagandist's target. All this is wastes time, raises needless doubt, and distracts from the more important issues. It can also get the target to disbelieve the data and adopt a cynical attitude towards the scientists who produce or use the data.

But here's the thing. All data is crap. It's messy, incomplete, not fully representative. We don't make progress by dismissing data because it is messy or wishing we had better data. How we make progress is to look for consistency across lots of crappy data, especially from different sources. That is, we look for the reliable trends that can be corroborated by different lines of inquiry and different sources of data.

So please, don't get hung up on IFR or CFR. Let academics publish papers on that years from now. That should not be our focus in the present moment. What are the big trends we should be paying attention to? Stay focused on that. You'll nice that kind of charts I like to post. These usually represent what I think are the big, reliable trends that the data are actually telling us. It important to actually look at the observable data. If you are curious about death rates, I'd recommend focusing on death growth rates. This is a fast-moving, dynamic process. IFR is a non-observable end state, who cares? Actual deaths are happing right now, let's see how this is unfolding so we can make sensible choices along the way. Also if you are worried about data issues in counting Covid19 related deaths, know that these data are coming form different countries and local jurisdictions. So the methodologies, issues and biases are all over the place. That's not the issue. When we see similar dynamics emerge from different data sources, we know that we are getting a robust picture of what Covid19 can do. So don't badger the data; compare results from different data sources instead.

We need to keep our heads up and eyes open. Avoid getting lost in things that cannot be seen at this point in time.

For me it is for example about learning to understand the data we are given. The value of the data, as well as its limitations. We might stop discussing IFR and CFR on this thread, but that doesn't mean the rest of the world will, and this thread seems to be a great place for a non-scientists to participate in that greater discussion which also informs political decisions. We also want to understand how much of a point there is in the arguments we are reading about from elsewhere.

Analyzing the death rate and trajectory is surely one of the more reliable tools available, but unfortunately it lags 2 weeks or so behind, and in the current situation that's an eternity. For me, that is not satisfying enough for developing a larger picture of the situation.

 

[jhm]

IFR can be quantified and it is a function of the intrinsic properties of COVID-19 and medical care. R0 changes with our behavior, there is no evidence that I'm aware of to support IFR changing over time for COVID-19. My understanding is that they don't think it's mutating fast enough for that to happen.

I think you are oversimplifying. Lot's of other factors are in play with any medical outcome.

 

[Daniel in SD]

What you didn't do is get hung up on some narrow range of issues that can't be nailed down with exacting certainty. Stay adaptive!

I guess I should clarify my thoughts on IFR since I feel like that might be directed at me. Obviously we can't know exactly what it is, that's why every professional analysis of the data includes an estimate of the uncertainty. I am however very certain that it is somewhere between 0.5% and 2%.

 

[Dr. J]

How do you think the market will respond when we reach 2000 deaths per day ?
Is it already baked in ?

I think it will get baked in over the next two weeks as reality starts to dawn. But what do I know? (Not that much.)

Probably a more important question: does that reality change policy? I'm not sure that it does, but I'm willing to bet that the country doesn't reverse the lockdown by Easter.

 

[wdolson]

Just for the record, loss of smell and taste is not a unique symptom of SARS-CoV-2 infection. It doesn't come with influenza, but it is pretty well-known to happen with other viruses. The MSM has latched on to this and overblown it as a symptom.

Until you provide an antibody status that proves otherwise, as a physician I strongly question your assertion that you had a SARS-CoV-2 infection. I hear people near me stating the same thing, and every one of them keeps forgetting, or ignoring, that there is NOT a unique specific symptom or constellation of symptoms for infection of this virus.

I do want the antibody test to be sure. I hate not having the objective data.

I'm not saying ignore the symptom of loss of smell and taste, but it needs to be taken with a grain of salt and put into perspective. This is NOT a binary indicator.

I have temporarily lost my sense of smell when my sinuses were clogged, which is common with colds and flu. In this case my sinuses were completely clear. Other than damage to the olfactory membranes from a drug taken nasally (and that tends to be permanent) I haven't seen anything on anyone losing sense of smell without sinus congestion before this. I can't find anything online about flu symptoms and this, but few were talking about this specific symptom before a few days ago.

Flu and colds both usually involve the sinuses and mine were clear throughout this. Which is unusual for me. My sinuses were trashed from what turned out to be a fairly severe allergy to smog when I was growing up in Los Angeles. I was congested all the time for almost 18 years and cleared up after I moved away.

Sadly the problem with this is that we don't know for certain that having antibodies confers immunity!!! Right now that is the hope but the actual data aren't clear or really available now - and if the virus mutates????
Of course we need MASSIVE testing for antibodies and to get this question answered ASAP (clearly if immunity is clear than allowing those people back to work would be a great first step)

I saw an interview with Anthony Fauci last night where he was asked about whether antibodies confer immunity. He said nobody knows for sure, but with most viruses there is at least some period of immunity after the infection clears. With SARS and MERS it appears to be long term immunity, and the closest relative to COVID-19 is SARS.

As @bkp_duke has pointed out, for an RNA virus family coronaviruses tend to mutate fairly slowly because of an internal checking mechanism. It does appear this virus mutated from SARS which was loose 10 years ago.

There is also speculation that people who had SARS may be immune to this, but there is no data to back it up. There are other viruses where being exposed to one virus gives you immunity to other related viruses. The most famous being small pox and cow pox. Cow pox is a related virus to small pox which is not fatal to humans, but it was found that having had cow pox gives one immunity to small pox. The small pox vaccine was made from cow pox.

Ultimately we don't know one way of the other for sure, though the odds are relatively high that having had COVID-19 and recovered gives one immunity for at least a period of time. And that will vary from one individual to the other.

This is the next phase of the conversation. As we're seeing in NYC and I'm seeing in Pennsylvania, locking down is slowing but certainly not stopping spread. People are still walking around in supermarkets, most epidemic specialists agree this is going to spread through half the population in relatively short order.

The UK study was flat out wrong. Under no scenario, short of leaving patients in the street, were we going to see 2.2M deaths in the US. Could we still get to 400k is we really really tried to screw this up? Perhaps. But that doomsday scenario of 120M infections and 2% death rate has been completely disproven. We're now looking at 120M infections and something south of .2% death rate.

If you don't believe me, the author of the original report said so in front of Parliament 2 days ago.

The original study by Neil Ferguson described the outcome if we did not do social distancing at all. This death toll includes what happens if the spread is so fast that hospitals become overwhelmed and can't give every patient the care they need to survive, ensuring that a lot of severely ill, but savable patients will die.

Every model has assumptions. Some of those are based on data, while others are based on educated guesses. The death rate from COVID-19 is open to a lot of debate because we have absolutely no idea how many people have had symptomless cases, or such mild cases the medical profession never knew about it. If we were able to do widespread antibody testing, that would get us a much better picture, but while there are a few antibody tests now, we're in the very early days of distribution.

What data we do have is about those who got sick enough to pop up on the medical radar. I came across this today. It appears about 10% of hospital cases in the US result in death
Coronavirus Trend: One in 10 of Those Hospitalized Die

That's with a hospital system that may be badly stressed, but it still keeping up with demand. If hospitals reach a point where they can't give everyone who needs critical care the care they need, that percentage will go up.

The Diamond Princess gives us a laboratory of cases and spread, but it has factors that may skew the results. Most notable was the population was much older than the general population.
80% of Diamond Princess coronavirus patients had mild or no symptoms | The Japan Times

About 80% don't appear to have gotten sick, though I have not heard anybody getting an antibody test. It's possible most of that 80% did get it, but their immune systems reacted so quickly they never tested positive for the virus. There were cases of people who tested negative for the virus, had symptoms, and CT scans were consistent with COVID-19. About 20% had symptoms severe enough to require hospitalization, but this could be skewed by the older population on the ship. Data we don't know.

I would like to see data on what the infection rate was vs where someone's room was on the ship. On those cruise ships they recycle the air and pump it into the internals of the ship, but it's possible their normal filtering system was enough to stop droplets in the air from recirculating. It's also possible that passengers with sea facing cabins that could open a window or door to get fresh air may have had a lower infection rate. More speculation with no data one way or the other.

I can't find it now, but a German contact tracing study with one known introduction and all the people that person was in contact with who went into isolation found that there were quite a few people who were aysmptomatic, but those who tested positive for the virus, but we asymptomatic had the highest viral loads. Those with symptoms tended to have lower viral loads than those without. Some people never tested positive, which could mean they have some kind of innate immunity, or their immune systems reacted fast enough that they never tested positive for the virus, or while in the same space with the first patient, they didn't get infected at all. Again, a lot of unknowns.

It is possible that the death rate with good medical care for COVID-19 is close to that of the flu. However, it is obvious the hospitalization rate for COVID-19 is much higher than the seasonal flu. The US flu season in the US peaked around Christmas and hospitals had more patients than other times of the year, none were stressed to the breaking point. In several parts of the US, hospitals are stretched to the breaking point now. That's hard data showing that this is more dangerous than the flu.

Hospitals are not taking people who are very sick, but not critical right now. I mentioned the other day someone from my SO's office who was sick enough that under normal times he would have spent a night or two in the hospital, but they sent him home with an inhaler. Those who are being admitted to the hospital now are those who are unlikely to survive without it. That contributes to the 10% hospital death rate.

If the case load stays below the critical level in US hospitals, the death rate might be on par with the seasonal flu. But because so many people need hospitalization with this to survive, a lot of those people will die because they can't get the care they need if the hospitals go over capacity. That's where you get into the 2.2 million dead scenarios. It all hinges on whether the critical supply of hospital resources is breached or not.

It's like a slow moving flood. You're watching the river rise and hoping it doesn't go over the levies. If it stays below the levies, the damage will be minimal and it makes for some dramatic pictures. If the levies breach, it becomes an epic disaster with large portions of cities destroyed. You want to do everything to you can to prevent breaching the levies.

 

[wdolson]

I think it will get baked in over the next two weeks as reality starts to dawn. But what do I know? (Not that much.)

Probably a more important question: does that reality change policy? I'm not sure that it does, but I'm willing to bet that the country doesn't reverse the lockdown by Easter.

Trump doesn't have the power to reverse the lockdowns. Those are in the hands of individual governors and in some cases mayors. Trump can make noise about it and some governors will comply, but most won't. He will then go on Twitter and throw tantrums about the big bad governors who won't obey him and he may retaliate by restricting aid, but that's the limits of his powers.

 

[SMAlset]

I wonder if Tesla Could add these to what they are doing to help out with CV.

Face Shields — open source plans
Open Source Face Shield V1.1 - NYU COVID-19 Task Force

“The Open Source Face Shield was designed byUriel Eisen, Andrew Eisen, Vadim Gordin, andCallil Capuozzo as part of the Open Face PPE Project in collaboration with the NYU COVID-19 Task Force and many others.

Instruction and Face Shield illustrations byRachel Ciavarella

About the Open Face PPE Project
We’re a group of volunteer engineers and designers in NYC and abroad. We operate twelve 3D printers and a full machine shop that we’re using to design and ship COVID-19 supplies.“

 

[EVNow]

How do you think the market will respond when we reach 2000 deaths per day ?
Is it already baked in ?

The exponential thing slowly creeps upon us - I don't think any of these numbers will shake the market much.

Governmental actions and economic data are likely to be the big movers.

 

[Norbert]

I think you may have misread the chart -- worldometers reports 890 total cases in Iceland with 2 deaths (there were 88 new cases today, so that may be the cause for the mix-up)

Yes, slow start today on my side. However. The report's March 27th update still ignores delay of death vs test date. Unfortunately, worldmeters has no historic data about Iceland so far. But I found this article from March 15th:

Large scale testing of general population in Iceland underway

It mentions 425 positives "to date", which I assume refers to March 15th, not the "updated on March 21". Still that is only 12 days ago, less than the usual delay for death date I would use, as in the references by Daniel and Alan.

If you relate the current 2 deaths to 425, you get 0.47%. If it were just one more death, it would already be 0.7%.

There is the big possibility that all this means is just that they started testing very early and extensively.

Based on worldometers data, it looks like the German CFR has increased to 0.67% from 0.59%. Perhaps CEBM will adjust their IFR estimate for Germany, or there may be other factors that come into play.

It's obviously an evolving story, as they note: "Estimating CFR and IFR in the early stage of outbreaks is subject to considerable uncertainties, the estimates are likely to change as more data emerges."

Yes, and all indications point to CFR values going further up.

 

[SageBrush]

It's possible most of that 80% did get it, but their immune systems reacted so quickly they never tested positive for the virus.

No

 

[JRP3]

you will see rationing of care because of #1 - Canada and the UK already have this and it's not called "rationing" it's called "extended wait times to see a specialist".

You cannot have the nearly on-demand services of the US system with the price controls of a socialized medicine system.

When I had shoulder problems I had to wait almost a month to see a Dr. because I was a new patient. A friend of my mother's has water on the brain causing him increasing functional problems and the soonest he could see a specialist is June. They are trying to find some way to get him in sooner somewhere. Do not continue to pretend that care isn't rationed in the US.

 

[jhm]

I guess I should clarify my thoughts on IFR since I feel like that might be directed at me. Obviously we can't know exactly what it is, that's why every professional analysis of the data includes an estimate of the uncertainty. I am however very certain that it is somewhere between 0.5% and 2%.

Please don't take my comments personally. They were not directed at any individual. It's the sort of thing I see all over, not just here.

I just want to encourage us all to stay focused and adaptive. There is at least an order of magnitude uncertainty about how devastating this will be. So we've got to take each day at a time.

All the best.

 

[troym]

Germany is still in middle of things and doesn't really know its own state.

China and South Korea have stabilized more or less, and it would be nice if we could use them as an upper limit.
But the US already has more deaths per capita than either of them, so the US showed that those are not upper limits.

I think it’s fairer to use the Asian countries that have demonstrated some stabilization as lower bounds instead of upper...

 

[Webeevdrivers]

How do you think the market will respond when we reach 2000 deaths per day ?
Is it already baked in ?

That would be horrible. Our hearts our with you southern neighbours.

 

[HG Wells]

Florida will set up road checkpoints along the Panhandle border to direct motorists who have been to Louisiana to quarantine


Florida expands quarantine to visitors from Louisiana

 

[generalenthu]

5 minutes point of care FDA approved test from Abbott

Abbott receives approval for test that can detect coronavirus in 5 minutes

 

[TheTalkingMule]

If the case load stays below the critical level in US hospitals, the death rate might be on par with the seasonal flu. But because so many people need hospitalization with this to survive, a lot of those people will die because they can't get the care they need if the hospitals go over capacity. That's where you get into the 2.2 million dead scenarios. It all hinges on whether the critical supply of hospital resources is breached or not.

I agree with your entire post and thank you for taking the time to write out your thoughts. So much of this is unknown that it's almost futile to talk about, but this report is simply wrong in it's base assumptions.

Regardless of care or hospital status, plenty of the mortality rates used universally within the study are wrong. 40-49 year olds have absolutely nowhere near a 1.5% mortality rate in reality. I'm fairly confident we could leave all the US cases in this age bracket in the street and see a 0% to .5% mortality rate if the infection rate were as wide as Ferguson's estimates.

It may seem nitpicking and perhaps it is, but a LOT of unnecessary panic has been created by that report. Not just at the governmental level, but friends and family who've emailed/called me absolutely terrified. 2.2M was not a remotely viable possibility on the day this report was published(3/16). That's either a top epedemiologist being very wrong or acting very unethically. He should be retracting it, not deflecting. I was a consultant long enough to know when someone's layering more datasets and scenarios to try and cover a mistake.

 

[sparcs]

Florida will set up road checkpoints along the Panhandle border to direct motorists who have been to Louisiana to quarantine


Florida expands quarantine to visitors from Louisiana

the funny part is that the number of cases are escalating pretty fast in Florida and they still haven't locked down the state. Do they really think that there won't be any people traveling between different parts of Florida? Same playbook as the guy who wants to secure the canadian and mexican borders?

Anyone know if all the population centers and surrounding counties in Florida have strong safer at home rules? seems like that should be the bare minimum.