Daily deaths appear to have stopped growing roughly 15 days after the quarantine was enacted. Notably, 15 days is the lag time for death’s used by Annan and Hargreaves. No I haven’t compared to their model. This graphs is just doing the “back of the envelope” time comparison to see if the quarantine seems to have an effect in the time scale we would expect to see it have an effect.
Open Thread: Of course, we know the discussion will be All Covid… all the time.
The problem with this argument is that the curve in Illinois is basically the same as in states that locked down earlier or later and in states with more or less strict lock downs. That would not be the case if the lock downs were responsible.
I think it is important to separate two questions:
(1) Did individuals changing their behavior make a difference?
(2) Did government imposed lock downs make a difference?
Public awareness and concern changed dramatically between Monday, March 9 and Monday, March 16. Individual behavior also changed dramatically all across the country. So there is a stronger case for answering (1) in the affirmative than (2).
But if (1) is the right question, the lag time is closer to 3 weeks than 2 weeks. That is longer than it should be. Also, it is now five weeks and we are still not seeing a drop.
Without a proper control to know what would have happened if we had done nothing, the claim that the lock downs worked is at best premature, at worst unfounded.
I think social distancing is working, it’s possible the downturn in cases is a coincidence of timing with new rules, but I find it very unlikely. I think the burden of proof is on those who claim otherwise to put up evidence to refute the argument. Differences in timing can be from multiple sources (testing, reporting lags, etc.). The downturn also very likely has multiple parameters. It would seem obvious that being close to other humans would be a major forcing.
.
An argument can be made that social distancing without government imposed lock downs may be equally effective or effective “enough”.
Well, again as a fan of lower math, I just look at Worldometer and if a state (or country) shows that 5% or less of their total cases occurred in the past 24 hours I think they’re on the right path for the moment.
Here’s an analysis of comparative state performance by Kevin Drum:
https://www.motherjones.com/kevin-drum/2020/04/which-states-are-fighting-covid-19-most-effectively/
Cities and sates that were already suffering budget woes because of long term retirement liabilities are going to have a major problem with this covid budget setback. The pleas for fed bailouts are already beginning. NYC projects a $7.4B loss in revenue.
.
“The mayor announced earlier in the week that New York City would have to slash more than $2 billion in municipal services over the next year.”
“There was that famous Daily News cover that said ‘Ford to City: Drop Dread,’†Mr. de Blasio said. “So my question is, Mr. Trump, Mr. President, are you going to save New York City or are you telling New York City to drop dead? Which one is it?â€
.
Trump wouldn’t be saving NYC, it would be taxpayers everywhere else. As for de Blasio from this taxpayer: Drop dead. This is the very last guy I want to send a check too.
Yeah, Tom–we’re not a country! We’re 50 states. Their mayor is an outspoken member of the wrong political party–let the people die.
Riddle me this, Tom: How do you feel about each of the following?
1. The Payroll Protection Plan
2. Boosting unemployment payments by $600 / weekly
3. The stimulus checks of $1,200 per capita
4. The tax cut of 2017 of $1 trillion, 80% of which went to the wealthiest members of our polity
5. Tariffs on China which increased consumer prices by between $500 and $1,000 per family
Drum: “FWIW, I take it as yet another piece of data that we’re locking down without knowing which measures really work and which are window dressing. There’s just too much evidence accumulating that the value of (some) countermeasures is simply not as great as we’ve been assuming.”
.
This is what I’ve been thinking as well. Too much binary thinking without evidence, and nobody collecting evidence and reporting on it. Can we hardcore social distance an office or factory floor and get 95% of the same effective as closing the business?
.
There is almost zero discussion on this across the media and politics. I don’t think testing infrastructure is going to be wished into existence in the next month, hopefully I’m wrong, but we need a plan B.
Fuller,
Blasio is using extreme rhetoric against Orange Man to deflect his own failures and to give an alternate target for the coming blow back as city services need to be cut. Decades of financial mismanagement are the problem here, and he is transparently trying to use the crisis as cover for a fed bailout. NYC is already getting disproportionate fed help for the pandemic which is fine, but that’s not what he is asking for. Cities like Chicago are in worse shape.
.
Florida will be seeing huge budget shortfalls as the tourist industry disappears for half a year or more. Is Trump telling us to Drop Dead?
Thomas,
Do you believe that people who do not pay federal income taxes should benefit from a federal income tax cut?
Hi Mark,
Yes. Next question please?
In case you want clarification, I believe that the richest country on earth should feed the hungry, house the homeless, treat the ill, etc.
The biggest difference between a state and a group of people occupying the same piece of dirt is how it helps the needy.
Thomas Fuller (Comment #183198): “Here’s an analysis of comparative state performance by Kevin Drum:
https://www.motherjones.com/kevin-drum/2020/04/which-states-are-fighting-covid-19-most-effectively/”
.
Interesting. All states are pretty much the same. Too bad it is a week old; an update would be nice.
———
Addition: It looks like this is the source:
http://systrom.com/topic/coronavirus/
Tom, after what has spewed out of Trump the Chump’s mouth and Twitter feed you say Blasio is using extreme rhetoric?
Thanks Thomas.
.
I work my ass off to provide for my family and my extended family. I am happy to tell you I’m taxed more than enough already, and you can take your self righteous screed and stick it as far as I am concerned. The funny thing is, you spout crap like this and wonder why Trump is President.
Have a nice day.
Mark, when the unhappy day arrives, as it surely will, when you or someone you love needs help, to whom will you turn if your attitude spreads across the country?
To myself Thomas. The same as I do every day.
Do you imagine I got a stimulus check? Do you believe I benefited from the 2017 tax cuts? You’d be wrong on both counts. [Edit: I’m somewhat too successful for such things to be aimed at me, in case that wasn’t clear.]
.
The government doesn’t much trouble to take care of those who take care of themselves. As it should be, fine. But all of your grandiose ideas come at the expense of people just like me, who are working hard to take care of themselves instead of sitting on their asses and waiting for the government to solve their problems.
.
So, thanks for your response. I’m not impressed.
My mistake above. I got to keep a percent or two more of my income. It was lost in the noise.
Somebody took an overdose of virtue signaling pills this morning.
I’m not trying to impress you, Mark, I’m trying to understand you.
I’m in the same category as you apparently, at least in terms of being above the threshold for stimulus payments. I am helping support my father in a retirement facility, a mother in law living alone in France and still manage to save money and plan for my upcoming retirement, as well as earmarking modest amounts for helping others.
I do not feel the same way as you do about many things. Your responses don’t impress me much.
Yes, Tom, Mark is doing his best to signal his virtue as a hard working family man concerned about his own situation, come what may to the rest of the world.
Oh–you weren’t referring to him? 🙂
Good for you Thomas.
See, but I’m not the guy who showed up bitching about the tax cut; that’d be you. So, if you don’t like my response, how bout you take your observations someplace where the replies will impress you more. Do all of us a favor.
No sir. Here in the bible belt we’re all sinners, and I’m no exception. I’d like you progressive bloodsuckers to leave me the hell alone, that’s why I’m speaking.
Thanks for the opportunity to clarify that Thomas.
What state do you live in, Mark?
Alabama.
Ah, the state that receives $3.35 back from Washington for every $1 its citizens pay in federal tax.
So I’m the state of Alabama now.
Maybe it’s hard for collectivists to grasp, but no I’m not the state of Alabama. I’m an individual. Go think about the difference Thomas.
I’d venture to guess you aren’t trying to understand anybody here.
Well, it’s been fun, but — surprise!– I’ve got work to do.
Ciao Thomas.
Tom, you’re mistaken. Back in the day when there was still a climate conversation I would repeatedly tell the consensus people (Tobis, ATTP, etc.) that their lack of curiosity about their opponents was one very significant reason why they got their asses handed to them in every conversation.
I think understanding those on the other side of the fence is key not only to victory but to reconciliation.
WSJ offers up some safe at work advice.
.
How to Keep Workers Healthy on the Job
Employers can stagger breaks, deep-clean surfaces, limit meetings and offer on-site testing.
https://www.wsj.com/articles/how-to-keep-workers-healthy-on-the-job-11587311057
.
The first step is reducing the risks of commuting.
Businesses can encourage carpooling or deploy corporate vans or other forms of transport where hygiene and social distancing are easier to control.
Companies can also continue to encourage telework, and Zoom calls aren’t only for employees stuck at home.
Employers can, say, split workers into groups and alternate who shows up in person and who works from home.
Companies can allow staggered schedules, especially to avoid rush hours on crowded trains and buses.
Another consideration is meals. Companies should encourage workers to bring food to the office or have lunch delivered in prepackaged containers.
Then there’s cleaning. Employees will need reassurance that surfaces are disinfected regularly. Give disinfecting wipes to all employees and ask them to clean surfaces they often touch, such as keyboards and telephones, regularly.
employers should keep settings such as break rooms or kitchen areas closed.
Finally, it’s essential to provide access to rapid diagnostic testing if an employee becomes ill. Testing can be brought to work to facilitate screening for those who might have mild symptoms.
Tom Scharf,
Based on reports of what’s happening where I used to work, limiting meetings would drastically improve the efficiency of the technical workforce. They might actually get some useful work done.
Mark, I think your admirable attempt to communicate differing ideas about government here with some participants is doomed from the start. There are people who are giving of family, friends and the needy who also feel that the government should see to it that others do the same or more and through the auspices of the state. I think they have this view of government involvement through the eyes of a Civics 101 course where government gets a pass from the less than admirable traits of the politicians who run the show. The needy are needed by the politicians whereby they have a dependent class that is inclined to feel that they need to vote for those politicians upon whom they feel they depend for the necessities of life. A prime example of that is the plantation politics practiced in most large cities in the US. A sincerely motivated politician would certainly want the needy to become less needy without becoming permanently dependent on the government and politicians.
Hey, wait a minute! That graph is on a log scale. Plot it on a linear scale and it does not look at all like it is leveling off.
Thomas, what is the evidence that tariffs on China increased costs $600-$1000 per family, when inflation has been low?
Dr Birx showed a chart of top metros without NYC, and the scale dropped by a factor of 10.
#1 was Chicago, however it was Chicago/Naperville/Elgin, which is not comparable to #3 Boston/Cambridge/Newton.
Chart has 9 metros over 10000 cases. #9 is DC/Alexandria/Arlington. CDC website has 10529 cases for Virginia and DC combined. Either that metro is bigger than I thought, or Maryland is responsible for a substantial portion of those cases.
Chicago 25k, Boston 25k, Detroit 23k, Philly 20k, Miami 13k, New Orleans 13k, LA 11k, DC 11k
Mike M
The time of the change points strongly to the dramatic change in behavior as a result of the lockdown is what caused the change. And besides that from a mechanistic point of view, “awareness” and “concern” don’t change the basic reproduction rate. I live here. Individual behavior did not change “drastically” in Illinois on March 9 nor on March 16. I and many others were still taking dance lessons, working out in gyms, shopping at the mall and so on. I and many others still went to dance parties with many participants. Many people were going to work. Lots of people were still going to restaurants.
.
Maybe people’s behavior changed drastically somewhere else. But this is Illinois data. In Illinois. Our behavior changed drastically between March 19 when our gorvenor announced the lock down and implemented it on March 20. The timing of the flattening is just what you’d expect as a consequence of a change around March 19-20. It’s late if you think it was caused by some mystery event that no one in Illinois could even notice at the time that supposedly happened on March 9.
Thanks Kenneth.
I wasn’t really trying to communicate. My aspiration was more along the lines of taking out the garbage. I agree with you though that it’s largely a doomed effort, and I know it clogs up otherwise interesting threads. I’ll try to cut back.
>I and many others were still taking dance lessons, working out in gyms, shopping at the mall and so on. I and many others still went to dance parties with many participants. Many people were going to work. Lots of people were still going to restaurants.
Did you notice a dropoff at the time of how busy things were?
MikeN,
Yes. There was a drop off at one of the dance parties. Probably about 20% fewer than normal– so a “low” turnout, but still decent. It had also snowed. So the cause is hard to say. At another dance, the room was packed.
.
My dance teacher said about similar for his students during the week before the shut down. We discussed it. Some of his students were organizing their lessons to have no physical contact, so he watched them do drills and stuff.
.
(BTW: his watching them do drills and other solo work was good practice for running later online lessons. That’s what we are doing.)
Zumba attendance was unaffected. Everyone who signed up in Early March showed up on the last day. After the session, we talked about what might happen soon.
But 10-20% reduced attendance at some things (but not others) pales in comparison to the 100% shut down that happened when Pritzker did the shut down. FWIW: if Pritzker announces an end of stay at home, I think some people are going to stay at home anyway. I’m not going to go to large dance parties unless the daily death rates have declined noticeably and consistently from their level just before March 20. I will take lessons sooner than I’ll go to dance parties. The number of people I would contact is drastically different at the two things.
Interesting. Restaurant attendance was way down even before they announced a shutdown. In fact, I recently learned that restaurants are not officially shutdown, as I observed a Popeyes with dining room open. Many places have the door locked and they will bring your order to the door.
In Illinois restaurants were down but did not appear to drop off as much as what I noticed in Virginia. However, schools had closed locally a few days before the state order, and this likely influenced behavior.
MikeN
https://coronavirus.illinois.gov/s/stay-at-home-faqs#a3About%20the%20Order
Our stay at home order allows restaurants to open only for “Restaurants for: Delivery, take-out, curbside delivery”.
Dine in restaurants are ordered closed in Illinois. I can’t speak to whether they are all complying. (Is the Popeye’s you saw open in Illinois?)
I do know of at least one Illinois bar that allowed at least on “sneak” happy hour. Heard of it from someone who went. I and his two brothers gave him grief. The next week, his group met for a Zoom happy hour.
Were are you getting your restaurant numbers?
.
I do think some restaurants were starting to ramp up curbside pickup where possible. It’s very possible for places like Starbucks, Dunkin Donuts, McDonalds etc. Not so easy for the “white table cloth” places.
.
Customers did want that and took advantage of it.
.
That’s going to continue even after the “stay at home” ends. Some behavior changes will persist because people are cautious for themselves, loved ones, and yes, the population as a whole.
.
Part of the reason the stay at home is working is that aspects of it are accepted as reasonable. When that is no longer the case, people will begin violating and protesting. (See Michigan where some of the rules are nutty. There’s no reason to not allow landscape businesses to put down mulch, plant trees and so on!)
I would think by now that just about everybody would agree that mitigation interventions can reduce the Ro value to something like Rt<1. The only way that this would reasonably not be the case that Rt<Ro would be for the virus to travel and infect over large distances and not be impeded by individual protective measures. It could also be reasonably assumed that reducing the speed limit to 10 mph and strictly enforcing it would reduce auto traffic deaths dramatically. The questions that remain are related to what mitigation measures have been the most effective and with what part of the population and are those measures realistic given that mitigation will have to be eased up before a vaccine or prophylactic is available. Even the issue of when the mitigation started has long term issues to which the link below points.
Unfortunately with any uncertainty at all with an issue such as we are experiencing with Covid-19, governments tend to overkill whereby politicians can cover their posteriors and the immediate consequences of their actions can be put into the unseen future.
If we are ever to resolve or even partially resolve these government tendencies we need to have discussions and analysis whether that be with foresight or hindsight.
https://personalpages.manchester.ac.uk/staff/thomas.house/blog/modelling-herd-immunity.html
lucia,
Interesting. Were any public health measures taken in Illinois prior to the stay-at-home order?
Here in New Mexico, a public health emergency was declared on March 11, schools were closed and all large gatherings cancelled within a day or two of that. On the 16th, bars and restaurants were ordered to limit occupancy; my impression is that by that time people were substantially modifying their behavior. But it is hard to tell since I was modifying my behavior. On the 19th, a whole lot of things were ordered closed. For me, the stay-at-home order on the 24th had no observable effect, but I suppose it reduced the number of people going to work.
Kenneth Fritsch (Comment #183286): “I would think by now that just about everybody would agree that mitigation interventions can reduce the Ro value to something like Rt<1."
.
I don't agree, at least not as to the measures that have been taken. I don't care how many times something is asserted to be true, I believe it only when I see evidence.
MikeM
Can you be more specific about what you consider a public health measure? People were being advised to wash hands.
MikeM,
Our schools were closed Wed March 17 meaning they were open on March 17 and closed on Thursday and Friday the 18th and 19th. That’s insignificantly different from Saturday, March 20. Most people don’t work on Saturday, so in terms of traffic, commuting and so on, kids stopped going to school 1 day before most people stopped going to work.
.
Other states my have had gradual closings over the course of a wee: We didn’t. With respect to time scales of contagion, ours was a pretty discrete event. The whole state closed.
.
We can argue whether it was a mistake or not. But our Illinois close down was a pretty clean break.
FWIW: Jim and I went to a dance party hosted by my former dance teacher at Ballroom city Saturday March 7. It was packed Everyone was snarfing down snacks, dancing and so on. I did not go to their party March 14 because I only go to the ones Devin hosts. I have no idea what attendance was like.
I went to a different dance party Wed March 11. There were plenty of dancers there. It was a new venue for me so I can’t compare if it was fewer or more than usual. One lady I talked to had driven in from Indiana.
Edit above: I have the day of weeks wrong: Students were off school schoof Wed March 18, Thursday March 19. Everyone else was locked down March 20. So schools were closed 2 days before people stopped going to work. Our lockdown was fast.
I can still golf in FL in my county. One person to a cart and they wipe down the carts between people. The clubhouse is a weak link and you still have to sign the credit card machine, yuck. I hear in some places you must pay online and no getting into the clubhouse. No option to pay online here.
.
They took away all the rakes and put Styrofoam rings in the holes so you don’t have to touch the flag stick. I walk so the cart is not an issue. However some of my fellow golfers are idiots and are just asking to get the course shutdown. A bunch of 50+ age groups hanging out outside at a table drinking (the inside is closed).
.
The 6 foot rule is fine, unless there is wind which nobody seems to understand. A bunch of people lined up on the driving range with wind going right down the line doesn’t seem like an ideal situation. My wife pointed out that if you can smell somebody’s perfume when they walk by …
.
It’s very unclear how dangerous any of that even is. There still seems to be almost no information to gauge the relative risk.
lucia (Comment #183289): “Can you be more specific about what you consider a public health measure? People were being advised to wash hands.”
.
Advising people as to protective measures (hand washing, face touching, etc.). Closing schools. Advising people to avoid crowds. Advising people to work from home if possible and to make only necessary trips outside the home. Limiting occupancy in places where people might be in close contact. Closing such places, such as bars, restaurants, theaters, malls.
.
lucia (Comment #183290): “Other states my have had gradual closings over the course of a wee: We didn’t. With respect to time scales of contagion, ours was a pretty discrete event. The whole state closed.”
.
So from one extreme to another. Wow. Is your stay-at-home order enforced? Ours is not, as least not as regards individuals. Maybe they are enforcing on businesses.
.
I think that reasonable public health measures are justifiable and probably have at least some effect. One can argue as to what is or is not reasonable and as to how large the effect might or might not be. My main objection is to the fascist stuff going on in places like NYC and California.
.
I have a second problem with hard and fast orders from the state: They do not seem to have an exit strategy. If you modify behavior via recommendations and education, then restrictions can ease off and on as evolving conditions warrant. But a hard order is either on or off and, as we are seeing, going from on to off is problematic.
The effective use of recommendations and education requires a lot of hard work from the authorities. Flat orders are much easier for them. But inconvenience for the authorities is not what should drive decision making.
Tom Scharf (Comment #183227): “Businesses can encourage carpooling or deploy corporate vans or other forms of transport where hygiene and social distancing are easier to control.”
That is a horrible idea since it will produce extended contact in a confined space. If one member of the carpool gets infected, then all are sure to get infected, unless already immune. The van is even worse, since different people will likely be exposed on different days. The only way social distancing can be done in a motor vehicle is one occupant per vehicle.
———-
Tom Scharf (Comment #183294): “The 6 foot rule is fine, unless there is wind which nobody seems to understand.”
.
If there is wind, there is no problem since there is also rapid dilution. Windy conditions produce large eddy diffusion coefficients.
.
Tom Scharf: “A bunch of people lined up on the driving range with wind going right down the line doesn’t seem like an ideal situation.”
.
That would be more of an issue since any exposure would be sustained. The rule is actually no more than six seconds within six feet.
.
Tom Scharf: “My wife pointed out that if you can smell somebody’s perfume when they walk by”
.
Not a good analogy. A person who has taken a bath in perfume is emitting odor at a much higher level, relative to the detection limit, than an infected person is emitting virus particles.
It is a myth that a single virus particle will cause infection.
———–
Hmm. That perfume argument gives me an idea – maybe one can protect oneself by not bathing. 🙂
In Albuquerque, as the various restrictions took hold, bus ridership dropped. So the city cut service, apparently to increase bus occupancy levels.
I was about to say “it does not get much stupider than that”. But then I remembered that stupidity has no upper bound.
The restaurants I observed were in Virginia, I think the week of Mar 14-20, when schools were closed statewide, but no shutdown order for everything else. Popeyes is still open, and the state order says no dining over 10 people. Even prior to this order, but after the school closing which locally was on Mar 12, I saw Cheesecake Factory nearly empty when they normally have lines out the door Friday evening. I think they are doing more business now with to go than then.
Anecdotally, I have heard from owners in Illinois of a dropoff but still getting hundreds of customers a day prior to the restaurant shutdown order, more like a 50% drop.
NYC took much longer to close. In mid-March a nursing home was discussing banning visitors https://www.fox5ny.com/news/brooklyn-care-home-tolls-55-dead
A county where I grew up had no cases thru end of March, but had shutoff visitation to nursing homes and prisons by Mar 10.
Notice the 55 dead at the nursing home involves people who they don’t know if they had COVID.
Mike M.,
Oh? Not according to this.
However:
I heard a long time ago that if you swim or SCUBA dive in even slightly polluted water with an open wound, that only one virion in the wound can give you hepatitis.
First “advising†is not the mechanism for changing the infection rate. People taking actions is. But if you think “advising†does is an importat thig: People were advising we wash hands from the time of the very first reports in January. People were already washing hands in early Feb before we had any infections. It can’t explain a “change†in the rate of increase of deaths observed in April.
(I did start stocking the pantry in Feb in anticipation that there would, eventually, be a lock down. I didn’t know what the parameters might be.)
Our schools were closed 2 days before the state was locked down. The St. Pats parade was cancelled at the very last minute. That was 3 days before the lockdown. I knew people preparing to go. (I learned not to go after freezing my buns off, not getting a good view, and being crowded out of pubs years ago.)
Once again: “Advising†isn’t a mechanism for reducing the rate of infection. Actually avoiding crowds does. Those organizing big events were still organizing them. The organizers for Windy City Open were still signing people up. Dancers were still signing up.
Once again “advising†isn’t the relevant thing. It’s people actually starting to work from home. My sister started to work from home the Monday after the lock down. That’s typical.
None to speak of. This happened on the lockdown. Our lockdown was a major distinct change over a short time scale. Yours might not have been.
MikeM,
I don’t have an issue with your objecting to hard and fast lockdown which are mostly economic and social. But we had one. The apparent rate of change in deaths in our state happens the appropriate amount of time after the lock down to be consistent with the lockdown being the cause.
One thing a fast hard lock down does do is create cleaner data. That may not be a good reason to have one, but that we had one happened. It’s in the past. I’m not going to not look at the data just because someone thinks the lockdown shouldn’t have happened. I’m not going to pretend we had some sort of slow diffuse lock down when we didn’t. I’m not going to pretend that behavior that was happening through the entire period explains a change in the last two weeks.
I’m also not going to deny other factors have an effect. But people were already washing hands in February, so that effect is in the “pre-lockdown” data. We locked down fast– over 3 days in march.
MikeM
Large eddies doesn’t really result in rapid dilution. If you are looking at a “batch release” it results in unpredictability of where that “batch” goes. Smaller and mid scales will mix that batch with air around it.
.
On the “on viron” infection theory: We don’t know what is or is not a myth about infection processes with this virus. We barely know with flu!
DeWitt Payne (Comment #183301),
One virion causing infection is theoretically possible, but it is not significant in practice. It typically takes a small but significant initial infectious dose to produce an immune response, a larger dose to produce symptoms, and an even larger dose to produce serious illness. Vaccination is based, in part, on that fact. Of course “small” and “large” vary from person to person; an individual with a compromised immune system often can not safely receive a vaccine.
Here is an excellent article on the subject:
https://www.newyorker.com/magazine/2020/04/06/how-does-the-coronavirus-behave-inside-a-patient
lucia (Comment #183306): “Large eddies doesn’t really result in rapid dilution.”
.
Large eddies produce smaller eddies which produce smaller eddies and so on, all the way down to the Kolmogorov length, where molecular diffusion takes over.
Were Illinois schools closed the same time as everything else?
I ran my amended Annan Bayesian code in R instead of R Studio in case it needed debugging. It did not and as a result I obtained a graph and table in R using the US daily deaths in the program. These results are in the linked Dropbox. I would appreciate those here who are more familiar with this than I am in looking at the results and letting me know if I have at least got the variables assigned correctly.
The Bayesian posteriors were very close to the priors which means I should redo the analyses with different priors.
https://www.dropbox.com/s/ie9mot6c4elnx6q/Annan_Bayesian_Covid_US.pdf?dl=0
Several more small studies have come back with pretty high asymptomatic carrier counts. 88% of pregnant women who tested positive in a NY hospital had no symptoms. I don’t think any of these anecdotal studies I have heard about came to a different conclusion.
I think the people protesting stay at home laws are pretty much cranks and clowns at the moment, but this is not the correct response:
.
https://arstechnica.com/tech-policy/2020/04/facebook-nixes-anti-quarantine-event-listings-for-violating-state-rules/
.
“Facebook has nixed event pages for planned protests against stay-at-home measures in California, New Jersey, and Nebraska. The actions come at the request of state officials, who say the gatherings would violate state orders.”
It is probably just me but I found this list in the WSJ on suggestions for having sex during the Covid-19 crisis both silly and funny. Now I must preface this by admitting that those hormones connected to this activity are no longer at the raging pace they once were, but if you have to follow suggestions about it from the WSJ or another publication I would say why bother.
It appears as patronizing as the lists of favored political candidates that the Chicago Tribune (and probably other newspapers) came up with that could be conveniently taken into the voting booth. If you have to have a paper decide for you why bother or at least perhaps consider why should I be voting.
kenneth Fritsch,
“The Bayesian posteriors were very close to the priors which means I should redo the analyses with different priors.”
.
That raises real doubt, because some of those priors (only two days infective?!?) seem unlikely to be anywhere near accurate. I don’t think the Annan/Hargreves paper showed Baysian posteriors, so it is very good you do. Have you communicated with James about what you are doing?
Tom Scharf,
“Facebook has nixed event pages for planned protests against stay-at-home measures in California, New Jersey, and Nebraska. The actions come at the request of state officials, who say the gatherings would violate state orders.”
.
You gotta hand it to politicians. First they implement what appear on their face to be unconstitutional restrictions on personal actions, then claim any communication about protesting those unconstitutional restrictions no longer has first amendment protections.
So, no religious worship allowed: check. No freedom of speech allowed: check. No freedom of the press: check. No right to assemble and petition the government: check. Yup, the numskulls have declared the entire first amendment is null and void. The SC will have to get involved at some point. Sooner would be better than later.
Article I section 6 of the NJ State Constitution
NJ Supreme Court should block most of Murphy’s restrictions. But they are mostly progressives, so they won’t.
Kenneth,
Days to death/recover is negative??? Oh, wait, that makes sense since you only know when death happened, more or less, and have to calculate when the infection happened. But two days for the infectious period seems really short and -14 days seems a bit short too. Try looking at just New York instead of the entire US. The Wikipedia page has the data.
mark bofill (Comment #183216)
April 19th, 2020 at 11:14 am
Mark is doing his best to signal his virtue as a hard working family man concerned about his own situation, come what may to the rest of the world.
No sir. Here in the bible belt we’re all sinners, and I’m no exception.
________
mark, I hope you aren’t too hard on yourself. Being a sinner used to really weigh on me, but not so much anymore. Sinning doesn’t mean you are evil.
If I understand you, like me you put family first. I have 16 close relatives, and stand ready to help any who need financial assistance because of the pandemic. Fortunately, none are in need so far, but that could change.
While my wife and I take measures to protect others and ourselves from being infected with Covid-19, what we can do right now to help the economy is limited. We have increased our purchases, mostly to stock up on essentials, but of course our greater inventory will result in fewer purchases of these items later on. We have switched from margarine to butter to help dairy farmers, since demand for butter dropped when restaurants were locked down. Other changes in diet that could help farmers most in need are being considered.
When it’s safe, we do plan to eat out more than usual, take more trips by air than usual, go to movies, and do other things to help the industries that are hurting most. My wife says I could help my barber recover lost business by getting haircuts more frequently. Nothing new there, she always thought I should get more haircuts.
Giving to charities is another way we could help, but we are wary of charitable organizations, suspecting much of what we give never gets to those who need help, but goes to administrative expenses instead. There may be some we could trust.
If I do become infected, after recovering I would consider doing volunteer work since at that stage I obviously would not be putting myself or others at risk.
Sweden claims that their approach is working and that cases are flattening out without overwhelming the health care system. Just like here.
https://www.msn.com/en-ca/news/world/sweden-says-controversial-virus-strategy-proving-effective/ar-BB12TP18?li=AAggFp4
DeWitt Payne (Comment #183317)
DeWitt: New York state or NYC?
Kenneth,
NYC would probably be better as it would be more nearly homogeneous. However, since most of the cases are in the city and the surrounding counties, it might not make much difference.
Kenneth,
I think you are now the official local expert on running Annan/Hargreaves code. 🙂
Kenneth,
I find that WSJ article funny too. I mean… Imagine having a heart to heart discussion with your sweetie and saying: “Oh love of my life. I need to manage my stress. Let’s have at it! “
SteveF,
In the text of the paper, I think the priors are 2 days ± something infective, 4day ± some latent. But 15 days as the amount of time to die. I either didn’t see or don’t remember the justification of the 2 and 4 days are. I did find the justification for the 15 days.
Markbofill and TomScharf, you make good points about family. The family I was born into was big on family, but my wife’s family was even bigger on family and had actions that demonstrated it over 4 generations. My wife instilled in me very early in our marriage that it was family first and foremost and with no apologies given or needed.
I think the family unit can be and should be the most powerful unit for maintaining the well being of individuals and doing it in a voluntary manner. The more people depend on the government for well being the less important family can become. If politicians depend on government dependent people the family can become a secondary unit for them.
Real question:
Ever wonder why you can give to other people’s family members indirectly through a charity and obtain a potential tax deduction, but if you give directly to a family member in need no such luck.
Mike M
Big whorls have little whorls that feed on their velocity.
Little whorls have littlerly whorls and so on to viscosity.
Neverthless, Individual large scale bursts don’t instantly have fine grain structure present. That happens during a cascadeprocess. And, equally importantly, the fine grain structure eats itself up. So you can end up with only large scale structures with very little fine scale structure.
A blob of something dropped in the large scale structure will not diffuse to dilute quickly unless there is fine scale structure. It will be difficult to predict where it will go, but it will not quickly dilute.
Max, thanks for your concern, but don’t let it trouble you.
.
Kenneth, I strongly agree with you.
.
My apologies for the terse responses, I’ve been (and continue to be) short of free time.
lucia (Comment #183328)
Your comment was the funniest.
“Lets have at it” is very much to the point but never would have worked in my bedroom.
Annan”s priors have fairly wide distributions and I think were intended to be uninformative – which usually means the observed evidence will have more influence on the posterior results. I have used Stan in some fairly complicated Bayesian analysis and it is probably more robust than the MCMCpack program that Annan used. I know Annan has published Bayesian analyses on climate science topics a few times and I judge he is more knowledgeable than I am (and probably much more). Before I would communicate with him I would want to be more familiar with what the code is doing and would probably bother Nic Lewis about it first.
Kenneth,
Somewhere (perhaps his blog) he said he picked wide distributions to make the posterior take on information from the data.
This paper has some information on lag times that would them be put in SEIR.
https://www.nature.com/articles/s41591-020-0869-5
The lag times are short …. because people seem to be infectious before symptoms and not very infectious once symptoms show.
An indication of survival rates after more than ten days on a ventilator: A patient who went home after 14 days on a ventilator made the NBC nightly news tonight. I don’t remember them giving his age, but my eyeballing estimate would be late 30’s-early 40’s
I am linking a Dropbox pdf file with the R studio report from the Annan Bayesian analysis for NYC. My initial attempt failed because the program refused to run a statement that had mixed negative and zero dimensions. I was at an impasse since I do not yet posses a complete understanding of what all the code is doing. I figured it must be related to how far apart the start and intervention dates were. I added some zeros to the start and changed the start date accordingly. The start date that I obtained from the Wikipedia data that Dewitt linked was later than that for most nations listed at Worldometer. The program ran after that change, but even though I thought the start date compensated for the added zeros, it appears that the final data point in the results graph is 4 days behind the actual date which should be April 18, 2020. The final data point is also an outlier with only 24 daily deaths.
The posterior distributions look lumpy but appear to have reasonable confidence intervals (a term the Bayesians do not use and instead they talk about credible limits or some such terminology). Compared to the US as a whole from my previous analysis NYC would appear to be currently in a better place. When I have time today I plan to obtain the New York state data from Worldometer and run it.
https://www.dropbox.com/s/j6yxeveymfxcs1q/Covid_NYC.pdf?dl=0
DeWitt Payne (Comment #183344): “A patient who went home after 14 days on a ventilator made the NBC nightly news tonight. I don’t remember them giving his age, but my eyeballing estimate would be late 30’s-early 40’s”
.
Was that the 25 year old star athlete in New Jersey? He was the subject of a segment on Tucker Carlson’s show yesterday. He was in the news (at least on Carlson’s show) previously as an example of how the Wuhan virus can be a threat even to people in perfect health.
lucia has described how Illinois went from little concern to lock down in a matter of a few days. It occurs to me that this site has other denizens from Illinois. I am wondering if they concur with lucia’s description.
I am not doubting lucia’s observations. I am merely recognizing that different people have differing perceptions of events.
MikeM
I didn’t say little concern. Concern is a state of mind. You keep wanting to use words that focus on peoples state of mind and away from their actual activities. Of those two, only activities affect the transmission of this virus.
I said: very, very limited closures and reduction in people circulating to big drop in circulation. Many people were worried. But our library (and Kenneth’s in Wheaton) were open, our park district was open, most of our restaurants were open, people were going to work… and so on.
If they hadn’t been doing this, then
(a) no stay at home order would have been required to make them stop and
(b) you ought to have no complaint about the stay at home order because it would have zero economic consequence in a situation where everyone was somehow not going to work and staying at home even though no stay at home was in place.
That you gripe about the economic impact of the stay at home suggests you know quite well that it did result in a drastic change in the number of people going to work, going to resturants and so on.
“an example of how the Wuhan virus can be a threat even to people in perfect healthâ€
.
With 789K cases in the US this data point represents zero information to me. I assume this is an anecdotal story to try to get young people to be more vigilant. You can mine all those cases and make any point you want “disabled lesbian Jews with Hispanic fathers at risk!â€.
.
The Nature paper linked above says 44% of transmission occurs * before * symptoms show. Not exactly good news. An Ohio prison tested their entire population and 73% tested positive, 1800+ cases, no deaths.
https://www.npr.org/sections/coronavirus-live-updates/2020/04/20/838943211/73-of-inmates-at-an-ohio-prison-test-positive-for-coronavirus
.
This thing looks quite virulent, and probably really difficult to stop at this point. The good news is that the mortality rate is probably much lower than previous estimates given the large number of asymptomatic and low symptom cases.
.
When the economy reopens there is very likely going to be an increase in cases (or increasing R), if I was a governor I would be doing some expectations management.
.
A change in R after social distancing rules are relaxed will also provide further evidence that these rules work.
Off-topic, but has anyone heard if the US govt has been buying oil for the Strategic Petroleum Reserve? Perhaps the storage facilities are already full. Or perhaps they’ve disbanded that, now that the US seems to be energy self-sufficient.
And on-topic, has anyone tried to figure out what R would be with “acceptable” measures such as mask wearing, physical distancing in public where possible (say, beaches & parks; not likely on buses or in subways), etc.? “Acceptable” in this context means without lockdown of “non-essential” workers, although one would expect that where work from home is reasonably efficient, it would continue. It strikes me that less formidable measures might be effective to reduce Rt below 1, with less economic damage.
The link below is the R Studio report for the New York State daily death Annan Bayesian analysis. I obtained the data from Wikipedia through 4/20/2020. I added zeros at the beginning and adjusted the start time. The New York state posterior results differ dramatically from those for NYC and are moved away from the priors more than those for the US.
https://www.dropbox.com/s/4h83fh1uco2rw44/NewYork_Covid_Bayesian.pdf?dl=0
Harold,
We don’t know the effect of any individual mitigation measure on “R”. Annan’s type model could only tease out an estimate of the value before a mitigation and after. You also need to have a not to variable “before” policy and a not to variable “after” one and they need to be quite distinct policies.
.
No one has done and experiment of having no one wear masks and then suddenly change a law requiring wearing masks with no other measure changing. It’s also not going to happen.
.
To the extent anyone suggests the effect of “R”, it’s a SWAG (scientific wild ass guess) of something that is subsequently put in some other model. It appears they help. The effect on R may well be unknowable.
https://www.wsj.com/articles/why-oil-is-11-a-barrel-now-but-three-times-that-in-autumn-11587392745
“At a press briefing Monday, President Trump said his administration again was exploring taking oil off the market by pumping barrels into the Strategic Petroleum Reserve. “This is a great time to buy oil,†he said.”
Can we expect this from the orange man who sings Every thing is wonderful and Every thing is beautiful?
I think this goes towards answering Harold’s comment. We also might have known going in what parts of the mitigation works (best/better) had it been applied piecemeal.
HaroldW, I am currently too lazy to look but I recall Trump talking about the oil reserves and hearing somewhere that it was at 3/4 filled. Yesterday oil futures went negative which means you have to pay somebody to take the oil from you. Ships at sea are full of oil waiting for the prices to go up. Crazy times.
Harold,
I also think the only political expedient path to get out of this will be staged unwinding of the stay at home. It will go something like
Stage I:
Allow elective surgery.
Allow hairdressers, barbers, services to open, with some limit on number of customers a week for each service provider and another limit on number of people in a particular space. (For some business, the number customers/week will be the important limitation; for others occupancy thresholds.)
k-12 schools still closed; the occupancy is too high. (It’s summer anyway.)
libraries open with occupancy limits.
Allow some outdoor sports provided people are required to space out on benches. Perhaps leaving at least 3 feet laterally between each person watching and only filling every other row. (Minor league baseball might not make enough money, but some summer high school sports camps might fly in some circumstances. Little league might work fine.)
.
Monitor new cases and death rates vigilantly. If we pass, move on to stage II.
Get contract tracing operating and working.
Get test kits out there so people can self test.
.
Stage II: ?? Depends on apparent R at the end of stage 1, and how well contact tracing is working.
.
Like it or not, I don’t think the public at large is going to be willing to open more than stage 1 initially. Yes, people want the economy to work. But people want to feel safe.
The goalposts are being moved. Endless lectures on flattening the curve to not overwhelm the medical system are now being changed to eliminating the virus entirely and keeping it on a downward trajectory.
.
A flattish trajectory with the economy open is now verboten from the the tut-tutters, which was previously the stated goal, or an acceptable outcome.
Tom Scharf (Comment #183384): “The goalposts are being moved. … A flattish trajectory with the economy open is now verboten from the the tut-tutters, which was previously the stated goal, or an acceptable outcome.”
.
Indeed. And the new goalposts have been strategically located at infinity.
Sigh. NPR answers the question nobody is actually asking.
.
What Happens If U.S. Reopens Too Fast? Documents Show Federal Coronavirus Projections
https://www.npr.org/sections/health-shots/2020/04/21/839456638/what-happens-if-u-s-reopens-too-fast-federal-documents-show-coronavirus-projecti
“the “best guess” for how things will play out without further mitigation says that coronavirus cases and deaths would double about every five and a half days”
.
This is estimating what happens when all social distancing restrictions are removed, which is not anybody’s plan. Even if some crazy governor removed them all, the people wouldn’t follow that plan anyway. This is now the media narrative, binary thinking.
Promising results from anecdotal testing of coronavirus drug touted by Trump.
https://www.statnews.com/2020/04/16/early-peek-at-data-on-gilead-coronavirus-drug-suggests-patients-are-responding-to-treatment/
Tom
I guess I’m no one then!
I think that’s a very good question.
I think Ro is likely below 0.7 in Illinois right now. I also think it needs to stay below 1. That’s why I think reopening needs to be staged.
.
As for those who want it blown full open (if any exist), I think politically, you’re just going to have to settle for partial reopening with monitoring. And I think if you want to do the most to accelerate re-opening, make the case for what restrictions to relax, and what other programs to fund.
.
My understanding is that programs to carryout contact tracing have expanded: that’s great
I hope testing is expanding. We need it now and we need it even more if we are going to re-open.
.
And guess what? For the people who think the flattening of deaths in Illinois is a “natural” process: Then the partial reopening will just proceed more quickly because monitoring will reveal that R is staying below zero and we can gradually reopen. (In case you think the idea someone thinks the flattening of deaths is due to hitting the natural “top”… well, I’ve talked to at least one.)
Tom Sharf
I don’t think it’s any governors plan. But I have met at least one person whose proposed plan is nothing more than recommending masks (with no enforcement), and then letting everyone do whatever they feel like. (He does think some people will stay home voluntarily. But pretty much no social distancing restrictions. Just whatever the now frightened public takes on their own.
.
Don’t kid yourself: there is always someone who takes a very extreme position. But no: I don’t think any governor who put a stay-home in place is proposing that.
The model they are referring to is the pre-shutdown model, so they are just going down the “their own models tell them not to reopen the economy” story line, and not adding any useful information from my point of view. They are inferring that this model predicts partial reopening outcomes which I think is misleading.
.
There are definitely some blow it wide open people, but I don’t think that will happen. I think we should wait a couple more weeks, reopen the economy and throttle it to keep R = ~0.9 or so. The Northeast may need to wait another 4+ weeks.
.
It will be hard to gauge where R really is with the lag and a media that is very incentivized to see new infection spikes everywhere. It would be useful to decide and announce a metric ahead of time for clamping back down so everyone’s bias doesn’t make up the metrics as they go along. We know how that works.
.
Mass gathering are gone for at least the next 3 months, probably until a vaccine is found or we reach herd immunity in 12 months.
.
If R > 1 with just a partial reopening and testing expansion delayed then: “Damn! We’re in a tight spot!”. Decisions get pretty hard.
Tom–
I don’t think they are referring to the “preshut down” model. I didn’t don’t remember the exact times but I’m pretty sure they were doubling faster about once every 2 or 3 days. Doubling every 5 days is a lot less. So they seem to be assuming some sort of self-cloistering would continue.
.
I don’t know whether reopening would raise the current Ro above one or not. But I think it’s an important question to ask.
That’s what I think. I gave a suggestion about what to open and do in “stage I” above.
.
I don’t think it’s all that useful to try to announce a metric. Ideally, it would be R<0.9. But R is a model parameter and everyone would just argue what the value currently. So announcing that as the metric is about as useful as not announcing anything.
.
If you were to have a "hard" metric, it would need to be something like
"Current week new cases and current week deaths no greater than past weeks and down at least 5% from average over past two weeks. Then perhaps when we reach a suitably low number we allow it to fluctuate around that low value (10 deaths a day? Something like that?)
.
Yes. Mass gatherings need to be out.
Kenneth,
The fit for NYC looks terrible. The line is quite far away from the data points. NY state looks somewhat better, but still doesn’t look much like the plots in the Annan’s paper.
lucia (Comment #183388): “I think Ro is likely below 0.7 in Illinois right now. I also think it needs to stay below 1.”
.
I don’t think that is feasible. We are not going to drive this virus into extinction. You can do that in models with Reff<1, but models are not the real world.
So the Wuhan virus will likely be with us forever and will be a problem until a reasonable degree of herd immunity is achieved. With Reff < 1, that will take forever or until a vaccine is available, whichever come first. It is not at all obvious that a vaccine wins that race since the history of coronavirus vaccines is not encouraging.
lucia,
Any state where health care workers are being furloughed, hospitals losing lots of money and capacity for treating COVID-19 patients looks adequate should immediately allow ‘elective’ surgery at hospitals and same day or ambulatory surgery centers. Btw, in many cases the term ‘elective’ means not immediately life threatening and can be scheduled as opposed to cosmetic plastic surgery.
ER’s are also seeing a lot fewer heart attack and stroke cases. Apparently people are so frightened of catching COVID-19 that they may be risking dying from other causes.
https://www.washingtonpost.com/health/patients-with-heart-attacks-strokes-and-even-appendicitis-vanish-from-hospitals/2020/04/19/9ca3ef24-7eb4-11ea-9040-68981f488eed_story.html
lucia (#183380) and Tom Scharf (#183381) —
Thanks for your responses.
Below I have a link to the Annan Bayesian analysis of the IL daily death rates. The analysis posterior results look reasonable with symmetrical distributions. The Rt value is centered at 0.65.
lucia (Comment #183388): “I think Ro is likely below 0.7 in Illinois right now. I also think it needs to stay below 1.â€
Lucia, did you peek or do you do Bayesian in your head. Should I call the governor with these results?
https://www.dropbox.com/s/kk929xmdzzqzw9x/IL_Covid_Bayesian.pdf?dl=0
DeWitt,
” Apparently people are so frightened of catching COVID-19 that they may be risking dying from other causes.”
.
Sure. But at least they die politically correct. Maybe that is an important consideration at Heaven’s gate. Maybe not.
MikeM
If it’s below 1 now, it’s feasible to keep it below 1. Whether we drive it to extinction is a different issue.
Kenneth,
Very interesting. Unfortunately, the governor doesn’t give a sh!t about the Rt or about realistic projections of future deaths in Illinois. I think he wants to be re-elected, and everything else is window dressing. Am I cynical? No, just experienced.
Kenneth,
No. I’ve been running SEIR models and eyeballing shapes!! 🙂
Kenneth,
I posted your graph as a new open thread. Glad someone is running Annan’s model. It’s good to see!!!
Today’s data is out…. sadly, it is high… 🙁
R0 I think is a number that doesn’t change for a particular virus. This is what I was asking as to whether Lubos used ‘herd immunity’ correctly.
Mike N,
As used in an SEIR model, the base reproduction range, Ro doesn’t change unless the social/behavioral situation changes for the population. Then the instantaneous Rt is the product of the susceptible fraction of the population, Csus and Ro. That is Rt =Csus Ro. The susceptible fraction typically declines as people gain immunity after being infected and recovering.
But Ro can change. I don’t know what Lubos said about herd immunity. But typically, heard immunity is reached when the Susceptible fraction of the population, Csus reaches 1/Ro. This is because at that point Rt = (1/Ro) *Ro.
Usually, I think we’d really only say we have “herd immunity” if the susceptible fraction was below (1/Ro) using Ro without distancing measures. So that would be the pre-lockdown Ro. (And it has to be the value that applies for a population without immunity. ) Assuming before lockdown Ro was about three, we wouldn’t reach that until only 1/3rd of the population was susceptible, which is after about 2/3rd had been exposed. There is nothing to suggest Illinois was there. (And really nothing to suggest the country was there.)
Mike N,
Ro depends on multiple behavioral and cultural factors, not just the virus. Ro in NYC may not be Ro in Montana or Chile.
MikeN,
It’s also important to know that the epidemic doesn’t stop when herd immunity is reached. It stops growing. But a lot of people still get sick after herd immunity is reached. You just see the number of cases declining with time instead of growing with time.
.
However, if a population already had herd immunity and one or two infected people arrive, then the contagion dies out.
.
The other thing to be aware of is that all the published estimates of Ro assume no one was immune before this hit. In an SEIR model under that assumption, an estimate of the infectious time period, Ï„, and the rate of rise of cases or deaths, the value of Ro can be estimated. Basically you use 1/N * dN/dt = (Ro -1 ) Ï„ when Csus = 1 . Here N is the number of infected or exposed people.
.
You get 1/N * dN/dt from data from cases over time. You get Ï„ from attempts to monitor sick people (it’s estimated near 2 days.) Then you back out (Ro -1 ). Most outbreaks seem to be getting 2 <& Ro<4.
.
MikeN,
My time scale estimate above isn’t quite right because I forgot about the time scale from the intermediate incubation step…. but you do basically back out R0 from (1/N dN/dt) and the time scales. The time scales are often estimated from studying people from the time of infection (which is not easy to do.) James is using those studies as priors and doing fancier fitting.