Instead of a single peak moment, we want to think about “the time period during which medical supplies and services are overwhelmed with demand”. And that starts, in my rough estimation, the moment all the hospital beds are full.
In the US, we have 3 hospital beds for every 1000 people, and 2 of them are occupied on average. So we’re going to start having problems once 1 in 1000 people want to go to the hospital for coronavirus, which corresponds to an infection rate around 1%.
So that pushes the moment of great worry forward by quite a bit!
On the other side, it’s hard to predict when supply will again overtake demand. Maybe governmental intervention comes through on a massive scale, maybe mass quarantine works, maybe the weather warms up and transmission declines. But I’m worried it will take months for any of those to happen after the crisis times begin.
I would modify your advice to “2 weeks before all the hospital beds are full (in your local region)”, because 2 weeks is roughly the lag time between exposure and needing hospitalization (I think?). With exponential growth @ 5day doubling time, you really want to not catch it when 0.01%ish of the local population is hospitalized [assuming per-capita hospital beds in your region is typical of the USA]. My region has ~7M people, so I would be thinking about upping my social-isolation game when 700 people in my region are in the hospital, or something vaguely like that. Probably adjust that down quite a bit for uncertainty in the input parameters, and for not all cases being diagnosed (even in the hospital). Adjust down even more if lots of hospital staff are likely to get sick or quarantined because they’re not taking appropriate precautions, which seems probable at the moment.
Instead of a single peak moment, we want to think about “the time period during which medical supplies and services are overwhelmed with demand”. And that starts, in my rough estimation, the moment all the hospital beds are full.
In the US, we have 3 hospital beds for every 1000 people, and 2 of them are occupied on average. So we’re going to start having problems once 1 in 1000 people want to go to the hospital for coronavirus, which corresponds to an infection rate around 1%.
So that pushes the moment of great worry forward by quite a bit!
On the other side, it’s hard to predict when supply will again overtake demand. Maybe governmental intervention comes through on a massive scale, maybe mass quarantine works, maybe the weather warms up and transmission declines. But I’m worried it will take months for any of those to happen after the crisis times begin.
I would modify your advice to “2 weeks before all the hospital beds are full (in your local region)”, because 2 weeks is roughly the lag time between exposure and needing hospitalization (I think?). With exponential growth @ 5day doubling time, you really want to not catch it when 0.01%ish of the local population is hospitalized [assuming per-capita hospital beds in your region is typical of the USA]. My region has ~7M people, so I would be thinking about upping my social-isolation game when 700 people in my region are in the hospital, or something vaguely like that. Probably adjust that down quite a bit for uncertainty in the input parameters, and for not all cases being diagnosed (even in the hospital). Adjust down even more if lots of hospital staff are likely to get sick or quarantined because they’re not taking appropriate precautions, which seems probable at the moment.
Why do you believe that about the number of hospital beds? This seems like an important number to have, I wasn’t sure of it myself.
Hospital beds per 1000 people - some world data
As leggi said, the USA has about 3 hospital beds per 1000 people, and utilization is about 67%.
As for why it’s a good proxy… I couldn’t think of a better one that’s simple and objective. Can you?