COVID-19 Megathread 5: The Trumps catch COVID-19

[Fmr. Gov. NickG]

Today I heard news about a British study that found blood type is apparently connected to heavily course of Covid19 (here is a report). According to the study, blood type A has a 50% higher risk for heavy symptoms, while blood type 0 has the lowest risk.

It's kinda interesting because I remember a conversation at lunch with my boss from February and March and I said to him that I believe more unknown factors determine how mild or how bad the disease turns out. One suggestion I raised was blood type. I didn't read anything like that at the time and I have no deeper knowledge of medical issues in general, I was just guessing.

Yeah, I saw these studies a couple weeks back.  It actually prompted me to donate blood for the first time in a decade because I couldn’t remember my blood type and I knew they’d tell me if I did! (Turns out I’m O+, which is the most common type.)

[Fmr. Gov. NickG]

It doesn’t seem like many people are harshly criticizing California’s response.  And despite locking down very early, they are following the same path as Texas and Florida.

This just seems to me like more evidence that lockdowns alone are not very effective unless they are also bolstered by partial herd immunity.  The reason that places like Italy and Spain and New York were effective last month (and it seems like the UK has been effective a month later) is that they didn’t lock down until a significant portion of their population had already been infected.  But of course, this also resulted in horrible death tolls in all these places.  

So you could definitely make an argument that California locked down too early.  Ultimately, it was impossible for them to stop the spread of cases before enough people had been infected that they had an immunity buffer.  But at least now we have much better testing and treatment that we can identify and treat cases where there is a new surge, resulting in a lot less death than we saw in Spain or NYC.

Except New York and Italy only have 5%-15% infected, and you need much higher than that (70%) for herd immunity. There is no meaningful benefit below the threshold. Besides, other states like Ohio that locked down very early are seeing low cases. California is an exception to the general trend that the numbers tend to be worse in states that locked down late and opened early, and good in states with very strict rules.

I’m not talking about strict herd immunity.  But it is absolutely not true that “there is no meaningful benefit below the threshold”.  If you institute a lockdown that reduced R to 1.1 when a very small percentage of the population has been infected, the virus will continue to grow.  But use the same measures when 15% of the population is immune, and the R will be .93, and the virus will slowly die out.   That’s why the same measures that lead to the decline of the virus in NY might lead to its continued growth in CA.

[Fmr. Gov. NickG]

It doesn’t seem like many people are harshly criticizing California’s response.  And despite locking down very early, they are following the same path as Texas and Florida.

This just seems to me like more evidence that lockdowns alone are not very effective unless they are also bolstered by partial herd immunity.  The reason that places like Italy and Spain and New York were effective last month (and it seems like the UK has been effective a month later) is that they didn’t lock down until a significant portion of their population had already been infected.  But of course, this also resulted in horrible death tolls in all these places.  

So you could definitely make an argument that California locked down too early.  Ultimately, it was impossible for them to stop the spread of cases before enough people had been infected that they had an immunity buffer.  But at least now we have much better testing and treatment that we can identify and treat cases where there is a new surge, resulting in a lot less death than we saw in Spain or NYC.

There is no herd immunity buffer in California--they had a strict lockdown that was enforced and kept it under control, keeping hospitals from getting overwhelmed, and then reopened in phases. Look at New York for example which waited too long to go into lockdown, and ended up with a far higher number of cases and deaths despite having less people than CA.

Right now, cases are exploding in CA, while they have been in decline in NY for almost two months.  If CA had waited until they had an immunity buffer, their cases might be declining now too.  But their death toll would be much higher right now.  My post was not meant to come down definitively in favor of one strategy, just to point out why different states might get different results from the same measures.

[Fmr. Gov. NickG]

If we're going to head for herd immunity anyway (or close to it), I am truly starting to think a strategy of "intentionally infect all the young and healthy people" would have been what was best for two reasons.  First (and most importantly), we would be choosing who gets it.  Based on the CDC's estimate of a 0.26% fatality rate and assuming 50% as the herd immunity threshold, a non-targeted herd immunity strategy would result in about 430K deaths.  However, their estimate is a 0.025% fatality rate in those under 50.  Hitting herd immunity with engineering the infections to be in the young would have only resulted in about 41K deaths.  Now, it probably would have been a little more because we wouldn't have been perfect in stopping older individuals from getting sick.

The second reason is that we could have gotten this whole crisis over with in a matter of a few weeks to maybe a month.  Things truly would have likely been 100% normal by Easter Sunday if we had gone down that path.

This is more or less what I’ve been saying for at least two months.
 
I don’t think we should be infecting anyone without their consent.  But we should be providing immunity certificates to those who have recovered, and offering an option for young and healthy people to get a certificate through a low-dose, medically-monitored, quarantined, deliberate infection.  And we should only be fully “opening up” the country for those who already have immunity certificates until we have reached herd immunity levels.

[Fmr. Gov. NickG]

If we're going to head for herd immunity anyway (or close to it), I am truly starting to think a strategy of "intentionally infect all the young and healthy people" would have been what was best for two reasons.  First (and most importantly), we would be choosing who gets it.  Based on the CDC's estimate of a 0.26% fatality rate and assuming 50% as the herd immunity threshold, a non-targeted herd immunity strategy would result in about 430K deaths.  However, their estimate is a 0.025% fatality rate in those under 50.  Hitting herd immunity with engineering the infections to be in the young would have only resulted in about 41K deaths.  Now, it probably would have been a little more because we wouldn't have been perfect in stopping older individuals from getting sick.

The second reason is that we could have gotten this whole crisis over with in a matter of a few weeks to maybe a month.  Things truly would have likely been 100% normal by Easter Sunday if we had gone down that path.

Half of this is blatant disinformation, you ought to be ashamed of yourself.

But seriously, since Covid is so tame, please deliberately go out and get infected, and once you experience symptoms, lock yourself at home and don’t ever think of going to a hospital, no matter how sick you get. (remember it’s just like the flu!)
Honestly go out, get sick, and just don’t ever go to a hospital so you don’t waste resources.

There is no doubt that not getting infected with covid is safer for all individuals than getting infected with covid. 

However, at an aggregate level, we don’t have much choice.  60-70% of the population is going to get infected eventually.  The only choice is who is in that 60-70% and who is in the 30-40% who avoids it.  And it is much safer for that 60-70% to be composed of young and healthy people who have excellent medical care than for it to be composed of an essentially random sample of individuals, many of whom are extremely vulnerable, and most of whom will never know exaclty when they became infected.

[Fmr. Gov. NickG]

eventually the deaths are also going to catch up and start going up as well

Maybe if people keep repeating this mantra enough times without evidence it will turn out to be true.

Cases in California have been steadily increasing almost every week since the middle of March.  But deaths peaked on April 23, and it is now almost two months later.  How much of a lag do you actually expect?

[Fmr. Gov. NickG]

Also, what New York (among other places) shows is that 60%-70% don't have to be infected. The virus can be contained, at minuscule levels, until a vaccine or treatment is available.

This is a good point, and one that I have never really diagreed with. 

California (and to a slightly lesser extent Florida and Texas) almost certainly radically reduced their death toll by locking down before they saw a surge, and waiting until testing and treatment were more widely available before the inevtiable surge occurred.  This is why we see deaths still declining as we open up more and more, and why we should also be much less concerned with rising case numbers than we might have been two months ago.

[Fmr. Gov. NickG]

Also, what New York (among other places) shows is that 60%-70% don't have to be infected. The virus can be contained, at minuscule levels, until a vaccine or treatment is available.

We don’t actually know that.  At this point, it’s probably a question of whether a vaccine is available before a second wave comes.  But at least we do have treatments now!

[Fmr. Gov. NickG]

Also, what New York (among other places) shows is that 60%-70% don't have to be infected. The virus can be contained, at minuscule levels, until a vaccine or treatment is available.

We don’t actually know that.  At this point, it’s probably a question of whether a vaccine is available before a second wave comes.  But at least we do have treatments now!

Second wave?  We're still in the first wave.

I do think it’s reasonable to say that the “first wave” is essentially over in New York.

[Fmr. Gov. NickG]

Also, what New York (among other places) shows is that 60%-70% don't have to be infected. The virus can be contained, at minuscule levels, until a vaccine or treatment is available.

We don’t actually know that.  At this point, it’s probably a question of whether a vaccine is available before a second wave comes.  But at least we do have treatments now!

Second wave?  We're still in the first wave.

I do think it’s reasonable to say that the “first wave” is essentially over in New York.

In New York, sure, but look at what the abrupt re-openings have done to FL, TX, etc.

The discussion you are quoting was specifically about NY.

[Fmr. Gov. NickG]

Also, what New York (among other places) shows is that 60%-70% don't have to be infected. The virus can be contained, at minuscule levels, until a vaccine or treatment is available.

We don’t actually know that.  At this point, it’s probably a question of whether a vaccine is available before a second wave comes.  But at least we do have treatments now!

Second wave?  We're still in the first wave.

I do think it’s reasonable to say that the “first wave” is essentially over in New York.

In New York, sure, but look at what the abrupt re-openings have done to FL, TX, etc.

The discussion you are quoting was specifically about NY.

And my comment is directly related to the discussion on the effects of the first wave vs. the second wave.

No one is questioning that many states are still in the “first wave”.  The question was whether New York is an example of successfully containing the virus at an infection level well below herd immunity.  And I think the answer to that will depend on whether we see a second wave in NY, not whether other states are still in the first wave.

[Fmr. Gov. NickG]

https://twitter.com/ProjectLincoln/status/1274096165470994438?s=20

I would certainly agree that the pandemic has been astoundingly mismanaged by Trump and the whole administration.  But that graph is pretty misleading.  The CFR in most major European counties is much higher than in the US, and the death rate has fallen at a mostly similar rate in the US as in Europe, just about three weeks later.

[Fmr. Gov. NickG]

Latest U.S. vs. European case and death graphs

The next few weeks will be a pretty definite test for whether deaths are indeed a lagging indicator of confirmed infections.  Cases in the US have been exploding the last few days, while deaths continue their remarkably steady decline of around 15% per week.

[Fmr. Gov. NickG]

Meanwhile, Brazil reported 55k new cases today.

That's appalling.

If you think that’s bad, look at Chile.

[Fmr. Gov. NickG]

Are people still seriously citing case numbers?

Florida’s death rate has shown no signs of increasing.  It is still below the national average:the US has had about 2 daily deaths per million over the last, compared to about 1.5 in Florida.  New Jersey still gets 6 daily deaths per million after three months of shut-down, and I don’t see many people criticizing them.

Clearly, Florida’s policies have lead to an increase in cases.  But more new cases are -good- if they aren’t accompanied by more deaths.  It might mean testing has increased.  It might mean treatment has improved.  It might mean the virus has mutated and become less lethal (and likely also more contagious).  It might mean a younger population is getting infected, providing more of an immunity buffer against infections among vulnerable people.  If might mean all of these things.  It’s hard to break down these factors, but they are all positives. 

I know people are going to come in here and say that increased deaths will inevitably follow.  But there’s just been no evidence for that in the US or anywhere in the world that has had a major outbreak.

[Fmr. Gov. NickG]

Here’s an interesting article about the spike in new cases in Florida among children:
https://www.naplesnews.com/story/news/2020/06/15/young-floridians-test-positive-coronavirus-almost-twice-rate/3190098001/

Some notable stats:
- Children are almost twice as likely to test positive for covid as adults (9.2% vs. 5%)
- Children who test positive are more than five times less likely to be hospitalized than adults
- More than 3,000 children have tested positive in FL so far, but zero have died

[Fmr. Gov. NickG]

Are people still seriously citing case numbers?

Florida’s death rate has shown no signs of increasing.  It is still below the national average:the US has had about 2 daily deaths per million over the last, compared to about 1.5 in Florida.  New Jersey still gets 6 daily deaths per million after three months of shut-down, and I don’t see many people criticizing them.

Clearly, Florida’s policies have lead to an increase in cases.  But more new cases are -good- if they aren’t accompanied by more deaths.  It might mean testing has increased.  It might mean treatment has improved.  It might mean the virus has mutated and become less lethal (and likely also more contagious).  It might mean a younger population is getting infected, providing more of an immunity buffer against infections among vulnerable people.  If might mean all of these things.  It’s hard to break down these factors, but they are all positives. 

I know people are going to come in here and say that increased deaths will inevitably follow.  But there’s just been no evidence for that in the US or anywhere in the world that has had a major outbreak.

"There is no evidence that more cases results in more deaths" is some incredible galaxy-brain logic, and it somehow isn't even the most ridiculous claim in here (that honor goes to "more new cases is good").

When I said there is no evidence that “increased deaths will inevitably follow“, I might have been a little unclear.

I meant there is no evidence that if a spike in deaths occurs, it will lag by some significant time the spike in cases.  There many are locations where we saw spikes in cases and deaths more of less simultanously.  But very few where we’ve seen a spike in cases followed by a spike in deaths several weeks later.

[Fmr. Gov. NickG]

Are people still seriously citing case numbers?

Yes.

https://talkelections.org/FORUM/index.php?topic=370528.msg7405122#msg7405122

I don’t know what you are linking to.  Whenever someone posts a link to another message in the same thread, that link always seems to send me to the bottom of a message, and I’m not sure whether I’m supposed to be looking at that one or the one after it; both seems only vaguely relevant in this case.

[Fmr. Gov. NickG]

Ok well yes we're still seriously citing case numbers. Today's looking pretty bad at about 32K, even more than Brazil, and it's Saturday.

I’ll gladly take a 30% rise in cases (with a 0.5% bump in positivity rate) in exchange for a 20% drop in deaths, which is what today’s numbers look like week-over-week.

[Fmr. Gov. NickG]

Another day, another 33k (new cases).

The positivity rate also ticked very slightly up again today.

I’ll gladly take a 30% rise in cases (with a 0.5% bump in positivity rate) in exchange for a 20% drop in deaths, which is what today’s numbers look like week-over-week.

That's a bad trade (if we could hypothetically make such a trade) and a bad way to look at it. Hypothetically, we would be better off and it would likely mean fewer deaths in the long run if we instead had a 20% drop in cases and a 30% rise in deaths than vice versa.

Increasing cases (unless the increase is purely attributable to testing) imply that the virus is still spreading and R is > 1. As long as R > 1, more and more people will be infected, and the number of true new cases per day will rise exponentially at some rate that is higher the more R is above 1, but which is exponential even if R is only slightly above 1. As long as R remains > 1, that implies the virus will keep spreading amongst the population until herd immunity is achieved, which means as long as the death rate is > 0, you will get more deaths over time as a greater and greater share of the population becomes infected.

By contrast, if new cases were decreasing by 20%, then as long as we were not simply testing less, that would imply R < 1. Which would mean that the virus was disappearing, and there would be fewer deaths in the long run because the virus would wither away.

If we continue to improve treatment or alter the profile of the infected population such that a 15-20% week-over-week drop in deaths is sustainable, I don’t see why anyone wouldn’t take this trade.   It doesn’t matter much what the R of the disease is if almost no one is dying from it.   At that point, everyone will simply get infected and recover until we have achieved herd immunity.  (We’d also want to monitor hospitalizations, but I believe those are still decreasing on aggregate.)

One the other hand, if new cases kept dropping but deaths kept increasing, that would either mean the virus has gotten much more lethal or that we had gotten worse at identifying infections.  I can’t see how either of those would be a good thing.

[Fmr. Gov. NickG]

If we continue to improve treatment or alter the profile of the infected population such that a 15-20% week-over-week drop in deaths is sustainable, I don’t see why anyone wouldn’t take this trade.   It doesn’t matter much what the R of the disease is if almost no one is dying from it.   At that point, everyone will simply get infected and recover until we have achieved herd immunity.  (We’d also want to monitor hospitalizations, but I believe those are still decreasing on aggregate.)

One the other hand, if new cases kept dropping but deaths kept increasing, that would either mean the virus has gotten much more lethal or that we had gotten worse at identifying infections.  I can’t see how either of those would be a good thing.

I had assumed that when you were talking about a 15-20% drop in the fatality rate, that you were talking about a one-time effect. A one time drop such as that (or perhaps more of a drop, perhaps a 30%, 40%, or even 50%) drop would be plausible. However, if what you actually are proposing is that you think there is/will be/will continue a persistent trend of the fatality rate dropping by 20% or so each week, then on the one hand your argument at least makes more sense mathematically. On the other hand though, I would very much question what justifies the notion that such a thing is remotely plausible.

Suppose, just as an example, that the base fatality rate is 0.7%. What happens if that decreases by 20% every week?

The fatality rate each subsequent week will be given by F = .007 * (1 - .2) ^ (t - 1), where F stands for the fatality rate and t stands for the week (starting at t = 1 for the first week with the full fatality rate). If you calculate that out, you get the following fatality rate trend week over week:

Week   Fatality rate
1   0.70%
2   0.56%
3   0.45%
4   0.36%
5   0.29%
6   0.23%
7   0.18%
8   0.15%
9   0.12%
10   0.09%
11   0.08%
12   0.06%
13   0.05%
14   0.04%
15   0.03%
16   0.02%
17   0.02%
18   0.02%
19   0.01%
20   0.01%

So over the course of just 10 weeks, the fatality rate drops to 0.1%, and after another 10 weeks it drops to 0.01%. And as you go through additional weeks, the fatality rate asymptotically approaches zero.

You don't honestly believe that that is a sustainable trend, do you? The only possible way that something like that could be plausible is if we suddenly get some magical miracle treatment that is pretty much 100% effective and drives the fatality rate to 0. But there is no such treatment now, and while it is theoretically possible that we might at some point get one, there is no particular reason I am aware of to think that we necessarily will, much less that we will soon. I would hope that is not an accurate representation of what you are trying to say.

Ask yourself the question, what is actually medically plausible? What seems plausible to me is that the base fatality rate for a representative slice of the population (whether it is about 0.7% or some other number) has had a pretty much one-time reduction by some amount due to: a) semi-effective treatments such as plasma therapy/remdesivir and b) due to doctors gaining basic experience with the virus and it not being entirely new, so they know better things like when to use ventilators and how. Let's be pretty generous and say that reduced the fatality rate by 50% from 0.7% to 0.35% (I would doubt it is really that much, but let's just say it is). Barring more effective treatments arising, there is no reason to suppose that the base fatality rate for a representative population will drop by any more after that.

However, what has been happening at the same time is that the median age of people who are being newly infected has been going down. This implies that the overall realized fatality rate would drop for the moment for the moment because the proportion of the new cases that are more vulnerable to the virus will have declined. But unless you believe that a lower and lower share of new infections will continue to be old/have pre-existing conditions that make them vulnerable over time (i.e. unless you believe the median age of new cases will continue to drop forever), then eventually if the overall number of new cases continue to rise, then that effect will start to be reversed and the fatality rate will start to rise again. As a matter of math and basic logic, this will have to be the case to some degree, if cases rise indefinitely towards a herd immunity threshold. To illustrate that point, consider that at the limit, if 100% of the population were hypothetically eventually infected, then the median age of people who had been infected would have to be exactly the same as the median age of the population as a whole. And if it were 70% of the population, then there is room for the median age of the population to differ somewhat from the median age of infected people, but there is only so much room for deviation. As you go down to lower and lower percentages of the population who are infected, there is more freedom for deviation in the medium age (that is where we are now). But the more people get infected, the more the general tendency of the median age will be to approach the median age of the total population. And that means that if currently the median age of infections is lower and that is temporarily causing a lower fatality rate, then eventually the median age will start to go back up and the fatality rate will rise. And of course since the overall # of cases will be going up, not only will the fatality rate rise, but also the absolute # of fatalities.

I never said a 20% drop in “fatality rate”.  I said a 20% “drop in deaths”, i.e. total number of daily deaths.  I don’t know why this would be an implausible long-term trend, given that this is exactly what we have been seeing very consistently for almost two months now.  (It’s probably more like 15% per week over that time frame.)

[Fmr. Gov. NickG]

I never said a 20% drop in “fatality rate”.  I said a 20% “drop in deaths”, i.e. total number of daily deaths.  I don’t know why this would be an implausible long-term trend, given that this is exactly what we have been seeing very consistently for almost two months now.  (It’s probably more like 15% per week over that time frame.)

I was assuming, to be charitable to you, that when you said a 20% drop in deaths, you meant fatality rate, not merely in the raw number of cases.

What we were discussing is what would happen if cases are rising, which they have been recently (along with hospitalizations and the positivity rate, where we have data in the problem states). If you don't think that the fatality rate will continue to decrease indefinitely, it is more difficult to get a persistent long term drop in deaths, not less.

https://twitter.com/COVID19Tracking/status/1274472380941791233

Deaths apparently lag by ~28 days or so:

https://twitter.com/COVID19Tracking/status/1274473854849609728

What has been happening up until now is that we have had cases declining, and medical treatment improving somewhat (should be pretty much a one-time effect of starting to use Remdesivir/Plasma treatment/doctors learning the basics of what to do), and the median age of infection dropping. Those things have combined to lower the # of deaths we have observed over time.

Now, however, cases are starting to go up. If they continue to go up, the raw number of fatalities are going to start also going up with a lag, unless something happens to cause the the fatality rate to decline further such that the effect of the fatality rate on the total # of deaths exceeds the effect . Logically, there are two possible ways that the fatality rate could hypothetically go down further - due to new/more effective medical treatments on top of plasma etc, or due to the median age of people who are infected going down even further. As far as I can see, while it is possible that new/better medical treatments could appear tomorrow, there is no reason to expect that. And while the median age of new cases could continue to trend down for a short time, it won't continue to do so indefinitely if cases are rising, simply because at some point there will be fewer young people left to infect, and the remaining new infections will be older simply because the "low hanging fruit" of easily-infected youngs will already have been infected.

I actually do think the fatality rate will continue to drop.  Our treatments are getting better over time, not worse.  There also seem to be other factors beside treatments that are leading to a lower fatality rate, and it is difficult to break down which factor is contributing how much, or how these other factors will change over time.  But it seems very plausible to me that within a few months, we may have a combination of treatments that cures 90% of cases that would have resulted in deaths back in March.

And while you can say that at an anecdotal, individual level, death tends to lag 28 days from symptom onset, there is almost no evidence for this sort of lag in the aggregate data.

[Fmr. Gov. NickG]

In early April, New York was reporting almost 10,000 cases per day and almost 1000 deaths per day at the same time.
Over the last few days, Florida is reporting about 4,000 cases per day and 40 deaths.

That is a tenfold decrease in CFR.  

There are likely many factors that have led to this trend.  But it is real and it has been progressing steadily for two months.  People have been expecting it to suddenly reverse itself for at least six weeks, but it has just kept going.   Testing keeps increasing, treatments keep getting better, it seems the infected population keeps getting younger.   Why would any of these trends reverse now?

(I suppose there is a possibility of the virus mutating to a less lethal and more contagious strain, though this factor seems impossible to predict.)  

[Fmr. Gov. NickG]

https://twitter.com/JohnJHarwood/status/1275061777911463936

This is an interesting page that I’m surprised I hadn’t looked into before.
In particular they seem to show a strong negative relationshipbetween Rt today and Rt three months ago.  

But a couple of the Rt number of three months ago struck me as strange.  In particular, two of the states with the lowest Rt three months ago (March 22) were New York and Michigan.  But New York went from having about 2000 cases a day on March 22 (7 day average), to 9000 cases a day two weeks later on April 5.  Michigan similarly went from around 250 cases a day on March to 1500 cases a day on April 5.  The states were right at the beginning of some of the fastest growth the virus has experienced in this country three months ago.  How could they have the lowest Rts?

[Fmr. Gov. NickG]

Amazing how 3 months into lockdown, California is clearly at its worst point with the number of new cases vastly increasing and hospitalization hitting a record high too.  Time for Newsom to lock things back down.

Cailfornia’s problem is that lockdowns provide very little benefit in places where the virus doesn’t already have a strong presence.  They just delay the inevitable outbreak.  Newsome locked the entire state down based on an outbreak in the Bay Area but basically no significant infections in the southern half of the state, and now LA and the inland empire are seeing the consequences.

Edit: I probably shouldn't imply that the lockdown had -no- benefit in Southern California.  By delaying the surge in infections, they held off the majority of cases until we had developed much better testing and treatment, likely reducing the death toll by many thousands.  But the infections were still inevitable.  They could have stayed locked downs even longer, and likely treatment would continue to make improvements.  But the infection spike would still come.