It seems like your arguments can be summed up as “if we slow the spread enough, hospitals won’t be overwhelmed” but the US only has 924,107 beds in total, and if each case takes 4 weeks to recover (“People with more severe cases generally recover in three to six weeks.”) we can treat 11 million people or 3.4% of the population over a year, but that would mean death rates from other diseases would rise a lot since those patients wouldn’t have beds. Many countries do have a lot more beds per capita than the US (which surprises and confuses me) but presumably they’re almost all being used already?
ETA: Actually the limiting factor probably isn’t hospital beds but equipment for treating respiratory disease. For example according to this paper:
The median number of full-feature mechanical ventilators per 100,000 population for individual states is 19.7 (interquartile ratio 17.2-23.1)
This works out to be about 65,000 in the whole country which is a small fraction of what’s needed to treat the number of people who will need them (.05 * .5 * 327e6 = 8,175,000) even spread out over a year or two.
I’m also arguing that we might just have many fewer severe cases than these right-tail estimates are indicating. So far, Hubei has only had .1% of their population get confirmed cases, for example, and I think that many scenarios in which >10% of people are infected globally are ones in which the actual number of cases in Hubei is much larger than .1%.
I also think there are more reasons for expecting fewer severe cases in many parts of the world than in China, like the increased prevalence of smoking in China, relative to places like the US.
in which the actual number of cases in Hubei is much larger than .1%.
I think Wei Dai’s position is compatible with there being 10x as many undiagnosed cases in Hubei as diagnosed ones. But maybe you’re suggesting that it could be more like 50x?
10x currently seems like the right ballpark to me.
Those predictions are based on 80% of cases being mild. My claim is that if 90% of cases are undiagnosed, then substantially less than 20% will be severe.
It seems like your arguments can be summed up as “if we slow the spread enough, hospitals won’t be overwhelmed” but the US only has 924,107 beds in total, and if each case takes 4 weeks to recover (“People with more severe cases generally recover in three to six weeks.”) we can treat 11 million people or 3.4% of the population over a year, but that would mean death rates from other diseases would rise a lot since those patients wouldn’t have beds. Many countries do have a lot more beds per capita than the US (which surprises and confuses me) but presumably they’re almost all being used already?
ETA: Actually the limiting factor probably isn’t hospital beds but equipment for treating respiratory disease. For example according to this paper:
This works out to be about 65,000 in the whole country which is a small fraction of what’s needed to treat the number of people who will need them (.05 * .5 * 327e6 = 8,175,000) even spread out over a year or two.
I’m also arguing that we might just have many fewer severe cases than these right-tail estimates are indicating. So far, Hubei has only had .1% of their population get confirmed cases, for example, and I think that many scenarios in which >10% of people are infected globally are ones in which the actual number of cases in Hubei is much larger than .1%.
I also think there are more reasons for expecting fewer severe cases in many parts of the world than in China, like the increased prevalence of smoking in China, relative to places like the US.
I think Wei Dai’s position is compatible with there being 10x as many undiagnosed cases in Hubei as diagnosed ones. But maybe you’re suggesting that it could be more like 50x?
10x currently seems like the right ballpark to me.
Those predictions are based on 80% of cases being mild. My claim is that if 90% of cases are undiagnosed, then substantially less than 20% will be severe.
Then you wouldn’t expect close contacts to both have severe disease frequently. I think this has occurred too frequently.