"New York has twice the density of London but three times the deaths, and the differential is even higher [comparing NYC to] cities such as San Francisco and Los Angeles. Deaths have occurred disproportionately in poorer areas, where the incidence of long untreated morbidities such as heart disease and diabetes have contributed significantly. But the same is true in all other cities. The high dependence on mass transit also seems to be a factor. In other major cities, car commutes are much more common. As Joel Kotkin, a scholar of cities at Chapman University in California, says, it may be the lethal convergence of all three factors. 'If you put together density, levels of poverty and reliance on a mass-transit system, you have a hat trick,' he told me.… But even that may not explain the extent of New York’s unique catastrophe. Around the world, the highest death rates have occurred where hospital systems were overwhelmed in the early stages of the crisis. This is especially true in northern Italy. Anecdotally, at least, it seems that the same happened in New York: Large numbers of sick people never got to hospitals, arrived too late or, in the impossible circumstances that medical personnel were confronted with, were given ineffective treatment.… It will be a while before we get a proper understanding of what went so tragically wrong...."
From
"The Covid-19 Catastrophe Unfolding in New York Is Unique" (Wall Street Journal), quoted
at my son John's Facebook page.
John writes:
I'm not sure this is a logical argument:
"Density alone doesn’t seem to account for the scale of the differential between New York’s fatality rates and those of other cities. New York has twice the density of London but three times the deaths, and the differential is even higher for cities such as San Francisco and Los Angeles."
Doesn't that assume there's a linear relationship between density and infection rates, and isn't that not necessarily the case?
My question is about the comparison of New York to northern Italy, where hospitals were overrun. Were NY hospitals overrun? I thought they weren't. I think the 3 factors named — density, reliance on mass-transit, and the bad health conditions represented by the term "poverty" — are enough to explain what happened. These things are interactive. Shouldn't we talk about
Bayes theorem?
Posts
219 comments:
«Oldest ‹Older 201 – 219 of 219Density and COVID Deaths/100,000 in the 5 boroughs
NYC Boroughs Total Deaths Per 100,000 Population
Bronx 225
Brooklyn 180
Manhattan 120
Queens 200
Staten Island 140
NYC Boroughs Population per Sq. Mile
Bronx 34,653
Brooklyn 37,137
Manhattan 72,033
Queens 21,460
Staten Island 8,112
Correlation of Density and COVID death rate in the 5 boroughs of NYC: -.352
https://gothamist.com/news/coronavirus-statistics-tracking-epidemic-new-york
https://en.wikipedia.org/wiki/Demographics_of_New_York_City Borough density
I would say with humility that statistics is a huge field and statistics professionals are all very specialized and no one since Fisher can claim to be up to date in all the obscure methods. I have an MS in statistics and am in a part time PhD program, and the only thing I can be sure of is that all the young bio statistics assistant professors are out there doing all the research they can on the relationships for this disease, so they can get published and get tenure. I am quite content to wait a few months and read all their great stuff when it is reviewed.
As for Bayesian statistics, you can’t start there, you have to learn frequentist statistics first. When you learn to play baseball you don’t try to learn to throw a curve ball before you learn how to tie your cleats. So don’t worry about it, learn to do linear regression and t-tests in excel first. I say this in all humility.
After reading all the comments, I see one item is missing. Viral load. How strong is the virus attack. Someone exposed to a small amount may be able fight it off. The same person hit with a massive dose will not be able to fight it off, and have a worse case. Think subway.
This virus does not do well outdoors. It is killed by sun and heat and high humidity. So shelter in place can end up making the outbreak worse by having people breath in the same air indoors, sealed off from the outdoors.
I have been making a list of how bad our response to this was. We trusted China to tell us the truth. We depleted the supply of PPE 10 years ago, and never replenished it. The CDC said don't bother with masks, until they didn't. The damocrats played political games. I could go on...
There is one critical data item that everyone has missed. This is a virus with high infection rate, and low death rate. Maybe a little higher than flu, lower than bad flu. China had this for 3 months before we did anything in March. They had to know the true infection and death rate. They said nothing. This is an act of war.
This is the smoking gun. China knew. They let us think and act as though that this was a disease with 5% fatality. You had to shut down the economy to save the village. "Millions will die". In March all the damocrat govs issued the unconstitutional orders.
My question now: Did China let some know earlier? Some who acted to shut down the economy to win an election? New York Senator Chuck S. did it in 2008. He said a bank was shaky, that provided many business loans. The bank failed. Someone manipulated the price of oil in 2008. The economy crashed. They won. Is this the same tactic? What did they know, and when did they know it?
Blogger Howard said...
Graffiti has nothing to do with viral load or transmission. This is just a throwback to the racist theory of disease. Also it is very interesting to note that the deplorables depiction of how New Yorkers live like rats is very similar to the Nazi description of how Jews lived in Germany and Poland.
--
Good point.
And I was worried the discussion was going off track.
Oh no..track..trains. Bring on the Zyklon B showers. I can't help myself!
Re: RichardJohnson:
Take a look at Google's community mobility reports. People in Manhattan (New York County) seem to have started pulling back on recreational outings earlier than people in the other boroughs. Only by about a week, but a week makes a huge difference in a situation like this.
Also, as I argued above using my simplistic calculations, it's not per sqmi density as such that matters for spread, but the local density in enclosed spaces -- the number of people close together and hence the number of contacts each person engages in. That's why places like bars, nightclubs, religious services, etc. have spread the virus so effectively. And the NY subway too, perhaps.
Evaluating the relative density of the different boroughs is not sensible: the baseline of NYC is a high density, though there are parts of the Bronx, Queens and Staten Island that are positively suburban. Even Brooklyn, sorta. The density of the entire area is the matter, and the fact that few people travel in cars and that public transport covers all of it densely. The folks who got sick out in the hinterlands of the city all have really long subway and/or bus rides. There are also immigrant communities, very insular, slow to react the secular world around them. Also with different standards of hygiene and a communal lifestyle. That's all density.
Disclosure: this has little to do with the awfulness of NY and statistical analysis and is more just a tossing-out of another possible factor in world infection patterns generally. Back when HCQ/zinc was first advanced, before the media and the Left got their act in gear and trashed it for all intents and purposes, someone at a weather blog put together a world map showing the countries at that time with substantial COVID infections, then put up another map showing countries with endemic malaria, meaning that their populations had some means of acquiring antimalarials and motivation to take them. At the time, and this has likely changed since though to what degree I don't know, there was virtually no overlap in the coverage. In another interesting discussion a French oncologist who had been following a study population of cancer patients who had received chemo pointed out that of the 2400 patients involved none had developed COVID infections despite their obviously impaired immune systems. The common factor in his study population was that all, post chemo, had received an experimental mitigation therapy consisting in part of vitamin supplements and methylene blue, a substance that's had many applications as a medical intervention in its time -- most notably now as an antimalarial. It is related to HCQ, in fact.
I am convinced that the media and the Left have succeeded in destroying any hope for HCQ as a useful intervention, so this is not posited as an argument in its favor. I do think, though that in a global analysis of infection factors and what causes some populations to be less vulnerable needs to take common access to antimalarials into account.
Megaera,
Considering how it's being used in Italy, amazing how normally Euro sympathetic media here is smothering this.
https://www.trustnodes.com/2020/03/29/italy-finally-starts-mass-treatment-with-hydroxychloroquine
Althouse,
"People making decisions in real time and trying to shape policy have to come up with a good enough way to process the limited information they do have."
First, do no harm. The exact opposite of the Politicians' Credo: "We Must Do Something!"
Curious indeed. But where we seem to be heading, inexorably, to the anodyne assurance of "Oh, but it's for your own good" is life imprisonment for anyone guilty of being over 65. I note that no one seems to be endorsing virtually endless lockdowns for the obese (at least until they lose weight) or those with high blood pressure or diabetes. It's the one condition you stand no chance of improving or escaping -- except for HCQ/zinc timely administered, which seemed to vastly improve older patients' chances of getting away with minimal symptoms. Look at this suddenly widespread insistence that we have to keep older people locked up (to protect them, poor dears) which seems to involve a total removal of all civil rights until, well, come the millennium. Or a TOTALLY EFFECTIVE VACCINE. Or death. Meanwhile we'll just nail the oldsters up in their houses and see what happens.
On the ground here in SF, people were taking it seriously quite early on. Hand sanitizer and wipes were hard to find in stores at the end of February and people were getting all worked up over it.
My company switched to WFH mode as early as March 10, with downtown streets emptying out completely by St. Patrick's Day. Our public transportation system (I can't speak for BART, separate system) pared down swiftly, with only essential lines still running. Buses were subbed for streetcars and have not returned to use.
I do believe, as was stated in media reports recently, that San Francisco probably caught the leading edge of this back between Thanksgiving and Christmas. My colleagues were reporting bouts of the flu that left them wiped out for a good solid two weeks. What also helped was a warm, mild winter this year, with relatively little, wet crappy weather.
Bayes' Theorem is well past the theory stage. Insurance companies have at least a century of sickness data documented which shows that older folks get sick more often and frequently die from these illnesses.
Megaera,
Interesting to note media attention on Florida beache visuals instead of their very large retirement/elderly population and how they've managed.
Unknown -- well, yes, actuarial risk tables have been around for a lot longer than that ... weren't such risk factors part of the fun of tontines, after all?
walter: I had noticed, yes. All those predictions of massive wrinkly disaster, come to naught...and after all that effort and barely-disguised hope on the reporters' part too. Sad.
To Birx's credit, when baited by press, she acknowledged the professional abilities of Florida's county health folks.
Here in WI, it is very unclear what role county pros are playing.
It's striking how innumerate the average journalist is. It's a wonder that some of them manage to pay their monthly bills without getting evicted or having the electricity turned off. Anyway... I suspect it's all about elevators and subways. There probably isn't any other city with quite as many high rises, stuffing millions of people onto a small island. The average person who works in Manhattan probably spends more time in elevators than any other population in the world. Many cities have extensive subway systems, but Manhattan is the only place I know of where it is almost impossible to get around any other way. Taxis and bringing your own car both present high costs and inconvenience. On top of that the measure of reducing the subway schedule - thus increasing rider density - was absolutely brilliant, provided that your *intention* was to maximize the spread of the disease.
Thanks for the Monty Hall problem, Yancey Ward.
I followed the link, read the first two lines so that I understood the question, and then thought about it.
I realized this has everything to do with the game show host having no choice about what goat to reveal two-thirds of the time. Then I verified this more directly.
There are eighteen equally possible scenarios, some of which seem the same but are arrived at by different paths, and in two thirds of them the contestant would win if they switched their choice.
So it's a pretty cool problem, and it all hinges on the fact that this is a game and when the game show host makes his move he has no choice two-thirds of the time.
I have some doubt though that this is really analogous to Bayesian statistics. I get the sort of parallel in that a Bayesian statistical analysis starts with a "prior" or in a sense a guess of the person doing the analysis of what the answer is going to be.
Note that I know about priors, or rather I know something about priors, even though I don't really know Bayesian statistics.
I know for instance that a Bayesian statistical analysis will give a different result for a different prior. And I believe, although I need to verify this, that as you put more data into a Bayesian statistical analysis that, for a given problem, all the different priors will eventually converge on the same solution. But as long as there is a limited amount of data one can get very different answers depending on the assumed prior.
I gather that part of the advantage of Bayesian statistics is the ability to get better answers with less data if one knows something about the problem domain and one can reflect that knowledge in the prior.
I also suspect that there may be a hidden bias in "frequentist" methods even though I can't quite put my finger on what it is.
Had to laugh.
"People making decisions in real time and trying to shape policy have to come up with a good enough way to process the limited information they do have. The public health experts like those on the task force must have worked out something that they are able to do."
Assumes facts not in evidence.
A am a little bothered on 2 fronts:
1. The effort was designed to reduce the load on hospitals, not in any way designed to reduce the death count. Where there are no large demands on hospitals, it is reasonable to scale back; if the intent to to reduce the death toll, maybe not - but that's a HUGE mission creep.
2. The responsibility for individual health has always belonged to the individual; maybe that's not a good thing or we wouldn't have this big obesity problem, but that's been the way up til now. You get your advice from medical resources and then you choose to act on it or not. We backed into this corner to reduce hospitalization because that's a community resource that might have been challenged. That appears to be somewhat questionable.
Finally, I wonder if there are physiological differences in those of African descent, like the difference that leads to sickle cell anemia, which might be another unconsidered factor in the Bayesian realm.
Howard said...
Balfegor: that's a nice conceptual model of the density factor. If walls mitigate the affects from population density than one could say masks do as well probably not as well that's a wall but nearly as good
4/26/20, 10:09 AM
Also someone mentioned elevators before this... I'd be interested to know about architecture.
That is, the kind of buildings different people live in. Is a hundred-year-old five-floor walk-up more or less dangerous than one of the new 50-story towers on 11th Avenue where you can't open the windows? Sick building syndrome and that kind of thing. Stairways versus elevators. High vs. low floors. Any differences in sanitation, floor level incinerators / garbage chutes versus walking or trash outside. And in general, apartment living versus detached houses. Also, HVAC. Most buildings in New York at this time would be in heating rather than cooling mode.
As for subways and public transportation, regarding the greater incidence in the outer boroughs, people outside of Manhattan are typically spending more time on public transportation traveling into the city core, than Manhattan residents. Possibly making multiple transfers. Maybe the touch point is at fare boxes and everybody swiping their Metro pass is getting a hit of virus.
Not only should trains be maximized rather than minimized, buses likewise, but A, at each end of the line the vehicle, bus or train car, should be thoroughly decontaminated with whatever is the most effective disinfectant process, whether mopping with bleach or beaming with ultraviolet light or whatever, but B, first, each car should be sampled at many points for coronavirus, on floors, windows, seats, handholds, maps.
There should be collection points added to the trains to sample air quality and viral load and carriage density at each stop on the train, but grossly, it makes perfect sense that someone riding 20 or 30 stops has more exposure than someone riding two or three. It would also make sense that some stops might be more infectious than others, and that could lead to tracking down hotspots.
As for the ugly remarks about the characteristics of New Yorkers, they are scarcely worthy of response, but New Yorkers, past the necessary armor of daily existence, can be and often are exceedingly kind and considerate. As for loud spittle - sure, like Negroes eat fried chicken and watermelon. No prejudice there, no stereotyping.
As for dirty, I've worked in offices with French and other European people-they often smell, they do not wash properly, I don't know what is going on with that but when a Croatian supermodel type who somehow got on an audit team moves closer to you, your heart is going pitty-pat until she breathes at you or lifts her arms, then her name is Reek and you have business elsewhere.
NY Metro is having some undiscovered complicating factor/s. (It's not just because we suck, blah blah blah. Our pols stink on ice but I don't think that's all either.) Maybe a different, more communicable or more lethal strain. It would behoove us to figure it out, lest everybody else have the kind of time we're having next time, God forbid there is a next time.
Post a Comment