COVID19 update, April 26, 2020: NYC vs. rest of USA; super-spreader events analyzed; reopening non-emergency care; cyclic lockdowns

(1) Bret Stephens in the NYT: lockdowns are good for NYC, but why should the rest of the USA have to play by the same rules as congested NYC? (archive version

As of Friday, there have been more Covid-19 fatalities on Long Island’s Nassau County (population 1.4 million) than in all of California (population 40 million). There have been more fatalities in Westchester County (989) than in Texas (611). The number of Covid deaths per 100,000 residents in New York City (132) is more than 16 times what it is in America’s next largest city, Los Angeles (8). If New York City proper were a state, it would have suffered more fatalities than 41 other states combined.

It isn’t hard to guess why. New York has, by far, the highest population density in the U.S. among cities of 100,000 or more. Commuters crowd trains, office workers crowd elevators, diners crowd restaurants. No other American city has the same kind of jammed pedestrian life as New York — Times Square alone gets 40 million visitors a year — or as many residents packed into high-rises. The city even has a neighborhood called Corona, which, it turns out, has among the highest rates of coronavirus infections.

Consider a thought experiment in which metropolitan New York weren’t just its own state, but its own country. What would the crisis for what remained of America look like, then? In this slightly smaller nation of a little more than 300 million people, the death toll would amount to about 7.5 per 100,000, slightly above Germany’s levels.

I also suspect that, if NYC were at the same latitude as Miami or Houston, it would have seen much lower mortality even with the same population density. Although the NYC subway still strikes me as “the mother of all super-spreaders”, and a study from MIT agrees.

(2) Speaking of super-spreader events: Quillette’s Canadian editor Jonathan Kay did some research of his own into 38 such events in 28 countries. (He restricted himself to documented events in languages he could read: the Purim super-spreader events in Israel he therefore dismissed, despite the extensive documentation I have seen in the Hebrew-language press and social media.) At any rate, let me (over?)summarize some of the patterns seen over and over:

  • mass events (at high density), not people going about normal day-to-day business (unless said business involves catering at crowded mass events, of course)
  • long duration
  • extremely close physical contact — kissing, hugging, or shouting into people’s faces or ears (e.g., because the music at the event was so loud).

Buying seeds or “nonessential” household items is not on this list, needless to say. (If of course said purchase requires queueing in tight quarters for hours, that’s another matter.)

(3) Rick Jackson, chairman and CEO of Jackson Healthcare, the 3rd largest health care staffing company in the USA, points out in an op-ed in Newsweek [!] that many hospitals are standing empty and face closure unless allowed to treat “non-urgent” non-COVID19 patients again — and urges authorities to permit this. He points out that no fewer than 1 in 8 [working] Americans work in healthcare…

(4) The group of systems biologist Uri Alon at the Weizmann Institute developed this “Adaptive cyclic exit strategies” simulator

Their idea in a nutshell: work x days on, y days off (their recommendation is x=4 and y=10), for example by alternating shifts. I think the easiest to implement in practice would be x=5 and y=9, two shifts (i.e., alternating working weeks). The 9-10 days at home would mean anybody who got infected at work would either be showing symptoms or asymptomatically test positive by the time they’d have to go back to work.

Their simulations show that this is a way to achieve most of the benefit of a full lockdown, while still permitting about 50% economic activity in non-telecommutable sectors (so probably 70%+ overall).  The active case load, rather than a monotonic decay as in a full lockdown, would get a damped oscillation superimposed on it. Below is an example:

Anyway, have a look at the simulator. The key is to keep the effective reproductive number Rt (in their notation) below 1 so the epidemic will die out eventually. A number of larger Israeli companies have adopted this strategy, with two staggered shifts.

(5) As a final reflection: the more I think about COVID19, the more it sounds that, if we had a reliable way to prevent cytokine storm, or nip it in the bud if it appears, we would be following a drastically different strategy. I talked to a source in the drug design community, and it sounds like more and more immunosuppressant/immunomodulator drugs are being repurposed for clinical trials (or compassionate use) in severe COVID19.