Sunday, March 22, 2020

We Are About to Lose New York City to Covid


The above data shows the Covid-19 case count in New York (from here).  The data are for New York State, but the case count is almost completely dominated by New York City and its suburbs.  I plotted the data only from the day before it crossed 100, and the scale is logarithmic.  The bottom scale is the date in March, and the graph runs to 35 (that is, April 4th).  Thus it extends two weeks from the last data point, which was yesterday (March 21st).

The blue straight line is an exponential fit to all the data here, and that translates to a 40% daily increase.  The red line is an eyeball fit only to the last six days, and translates to a daily increase of about 47%.  I have extrapolated those two lines by nine days.

You can see that these rates result in crossing 1m cases around the end of March, or in the first week of April.  The population of the New York Metropolitan area is around 20m people, and that is denoted as the very top of the graph.  You can see that if these spread rates continue, the epidemic will saturate some time in April - most people in the NYC metro area will have caught it, and then the herd immunity will be high enough that further spread will slow.

Probably this is driven by the extremely high densities of people in New York, especially Manhattan.  Probably the subway system is a key vector since in normal operation it requires New Yorkers to stand shoulder-to-shoulder for half an hour at at time.  In those circumstances, anyone infectious on the subway could easily infect dozens of other people.  But just walking around on the street in Manhattan under normal conditions is probably somewhat conducive to spreading the virus.

It seems it would be difficult to stop this process now  The subway cannot be shut down, and the population of essential workers may be large enough that transmission amongst them is high enough that the epidemic will not be stopped and then they will infect many others (by definition of being essential, they must interact with others).

It is probable the lockdown taking effect tonight will have some slowing effect.  However, consider the time factors in the disease.  The data in the graph show positive test results.  According to the WHO joint mission to China, mean time from infection to symptoms is 5 days.  I have marked that as a black double-ended arrow above.  However, there's also the time from starting to get any symptoms to actually having a known test result. That is is unknown in NY, but has probably also been several days.  I have shown a four day bar on the graph.  So there is a nine(ish) or so day lag from infections that have already been seeded now to possible test results in the future.  So the next nine(ish) days of data would be driven by infections that have already occurred under non-locked-down conditions.  So there's probably already 100k to 1m cases seeded in NYC today, and these will drive a lot of transmission within the enclaves separated by the lockdown.

However, the process will not be well documented going forward - the volume of spread is likely to far exceed the testing capacity, and indeed it already is.

It will not be possible to get relevant quantities of PPE, ventilators and extra hospital capacity into New York this fast.  Most people will not get much if any medical attention.  Stats elsewhere suggest 15-20% of people with this disease need hospitalization, and they will not get it.  About a third to half of those will die to judge by the experience in Hubei and Italy.  So that suggests we'll see 1m to 2m fatal cases in the New York area next month.  New York will become the global icon of the disease (taking over from Italy, which in turn had taken over from China).

As this unfolds, terrified people will stream out of New York City to anywhere else they can get to.  They will bring the virus with them.  Even under normal conditions, New York is a very highly connected city.  The right thing to do will be for authorities not to let people leave, but instead to close down the roads, airports, etc out of the NYC area to all but HGV and emergency vehicle traffic.  However, what we've consistently seen in this epidemic is that a virus growing at 30% to 40% a day is moving too fast for western governments to think it through in time.  The US federal government has been particularly useless so far, so an effective and timely federal response seems far from assured.  My fear is that what may happen instead is that New York City will then put much of the rest of the US into a state of uncontainable epidemic.

Just a note on my expertise, since I'm making strong claims here.  I do not have a human epidemiology background, but I am a legit expert in computer worm epidemiology, and the considerations are somewhat similar.

7 comments:

Benjamin Cole said...

You were wrong about "Peak Oil." I hope you are as wrong with this prediction.

Poul Andreasen said...

From a credible source, I learn that aerosol-borne Coronavirus survive up to 3 hours. This means, for example, that a sneeze in the subway can infect many people. Only drastic measures such as curfew, closure of public transport, etc. will be able to stop the virus in a city like New York. So I fear your analysis is correct. Pray that you are not.

Unknown said...

Thanks Stuart for the remarkable, thorough analysis (as good as ever)--my daughter lives in Brooklyn Heights -she has felt for a month what your analysis could possibly lead to. I hope the intangibles lessen the spread and severity. My best to you and your family, Dan

drchuck said...

Herd immunity is not a given:

1. What R are you using to establish the HI percentage?
2. HI depends on a non-changing virus, but several accounts note the mutations of the strain already taking place.
3. HI is not even proven given 1 and 2.

Unknown said...

What we can do is wear home-made face shields made from a report cover. There are many youtube instructions, for example https://youtu.be/wBJIRolOGZQ

CDesmond said...

Have they decided it is aerosol-borne? The last stuff I read said it was still considered large droplet, but that was a few days ago. (Still an airborne risk, of course, but for a little less time - not sure it makes much difference in a city that is so densely populated.)

Sam Charles Norton said...

Although it is for the worst of reasons, I am glad you are blogging again.