but the current CFR estimates for COVID-19 are based on hospitals not being overwhelmed
That seems a bit misleading because current CFR estimates are largely driven by cases from Hubei province, where the hospitals are already overwhelmed. You could argue that there’s room for things to get worse, and I probably agree with that. But I don’t see how this consideration warrants such a huge update to the CFR estimates. It seems to be already factored in to quite a large degree.
For specifics: The Imperial College estimate gave a 95% confidence interval for the “total case fatality rate” around February 10th. The interval ranged from 0.5% to 4%, with 1% being the median/EV (so much higher likelihood for lower values in that range). The way I interpreted this, their total CFR referred to their best guess about the prognosis of all people infected on February 10th. That included a large majority of patients in Hubei province. For patients who were diagnosed there, the estimated CFR was 18% (11-85% for the 95% confidence interval). Those scarily high 18% are already part of the Imperial College’s 1% expected death rate estimate. It’s still only 1% instead of anything more close to 18% both because disease runouts in other countries (with less overcrowded hospitals) were less severe, and because they estimated that even in Hubei province, there must be a large number of cases with only mild symptoms. The second factor must be responsible for the bigger update (because there simply hadn’t been many cases overseas or in other Chinese cities at that time).
Elsewhere in this comment thread, you write that mild or initially asymptomatic cases may later develop to become more severe. I haven’t heard that point before. If that was true, that would be a stronger reason for me to conclude that CFR estimates by the Imperial College will be way too low. (And just to be clear, I think it’s defensible to argue that hospital crowding and other conditions related to a worst-case pandemic can become a lot worse than even the situation in Hubei province, so I’m not saying I’m confident that current CFR estimates will remain accurate during worst-case pandemic conditions. I’m just pointing out that the consideration is already factored in to some degree, so arguing about how large the update should be exactly seems to require more detailed arguments.)
That seems a bit misleading because current CFR estimates are largely driven by cases from Hubei province, where the hospitals are already overwhelmed. You could argue that there’s room for things to get worse, and I probably agree with that. But I don’t see how this consideration warrants such a huge update to the CFR estimates. It seems to be already factored in to quite a large degree.
My understanding is that China sent a large number of doctors from other provinces to Hubei (can’t find the source now), and the outbreak was controlled via draconian means unlikely to be reproducible in most parts of the world. Hubei has a population of 58 million and only 66907 confirmed cases, so the hospitals were not nearly as overwhelmed as they could have been.
ETA: Looking at the Imperial paper you cited, the (uncorrected) estimated CFR being 18% was not the result of hospitals overwhelmed, but undercounting infections within Hubei (see page 10), which they then corrected to 1%. So I think this is consistent with the CFR estimate being based on hospitals not being overwhelmed, and it being much higher in the future when hospitals throughout the world will be overwhelmed.
Good point about importing doctors from other cities. This won’t be possible anymore if a country has the virus everywhere.
Room upward for more hospital crowding: Okay yeah, crowding could be a lot worse still. But self-isolation of the entire population seems like an option that can be implemented by any country to slow the peak of an outbreak. It doesn’t seem to require China-level control over a population. So maybe it’s unlikely that things would become worse than they did in Hubei? I guess you could argue that unless the government is extremely strict about people not being allowed to go outside, then too many people will still do it and get infected.
I think you’re giving convincing arguments to be a lot more pessimistic than the 1% expected CFR. The main counterconsiderations I see are that with more preparedness, countries can better slow down the peak of an epidemic. I don’t think cities are likely to quickly end up in a state that’s as bad as in Wuhan because by the time the hospitals start to crowd, the cities will already have been giving strongly worded advice about self-isolating at home for many weeks. This time window was missed in Wuhan. (But I was already aware of those considerations, so for me your points still provided a large update. I don’t expect to end up anywhere close to the high numbers you’re giving, but I’ll have to think about it more and I definitely think 2% total CFR seems possible or even likely now as we forecast scenarios approaching one billion cases.)
But self-isolation of the entire population seems like an option that can be implemented by any country to slow the peak of an outbreak.
0.1% of Hubei’s population have a confirmed infection, and its hospitals are already at the breaking point (even with national resources transferred into it). If this is the limit, then hospitals can treat at most 0.1% of the population per month, so it would take 50-100 months to treat the 5-10% who will require hospitalization, which is not a realistic amount of time for self-isolation. Even 10 months is not realistic (without totally ruining the economy) so it seems “slowing the peak” just won’t help much.
The WHO report said some 90% of people had fever. Maybe the tests have a significant false neg rate, or theyre not testing enough people, but I am worried that few mild cases are being missed.
I’m very interested in seeing how the Diamond Princess plays out. On feb 20, some half of the cases were asymptomatic. Wikipedia hasn’t been updated with any data since then. This bit feels cruxy to me.
FWIW, https://www.worldometers.info/coronavirus/ counts all confirmed cases and has a table by country, which lists the Diamond Princess separately (“international conveyance”). It doesn’t distinguish asymptomatic from mild, but does separate out “serious, critical” cases, which stand at 36/705 (plus 7/705 deaths and 100/705 recovered).
That seems a bit misleading because current CFR estimates are largely driven by cases from Hubei province, where the hospitals are already overwhelmed. You could argue that there’s room for things to get worse, and I probably agree with that. But I don’t see how this consideration warrants such a huge update to the CFR estimates. It seems to be already factored in to quite a large degree.
For specifics: The Imperial College estimate gave a 95% confidence interval for the “total case fatality rate” around February 10th. The interval ranged from 0.5% to 4%, with 1% being the median/EV (so much higher likelihood for lower values in that range). The way I interpreted this, their total CFR referred to their best guess about the prognosis of all people infected on February 10th. That included a large majority of patients in Hubei province. For patients who were diagnosed there, the estimated CFR was 18% (11-85% for the 95% confidence interval). Those scarily high 18% are already part of the Imperial College’s 1% expected death rate estimate. It’s still only 1% instead of anything more close to 18% both because disease runouts in other countries (with less overcrowded hospitals) were less severe, and because they estimated that even in Hubei province, there must be a large number of cases with only mild symptoms. The second factor must be responsible for the bigger update (because there simply hadn’t been many cases overseas or in other Chinese cities at that time).
Elsewhere in this comment thread, you write that mild or initially asymptomatic cases may later develop to become more severe. I haven’t heard that point before. If that was true, that would be a stronger reason for me to conclude that CFR estimates by the Imperial College will be way too low. (And just to be clear, I think it’s defensible to argue that hospital crowding and other conditions related to a worst-case pandemic can become a lot worse than even the situation in Hubei province, so I’m not saying I’m confident that current CFR estimates will remain accurate during worst-case pandemic conditions. I’m just pointing out that the consideration is already factored in to some degree, so arguing about how large the update should be exactly seems to require more detailed arguments.)
My understanding is that China sent a large number of doctors from other provinces to Hubei (can’t find the source now), and the outbreak was controlled via draconian means unlikely to be reproducible in most parts of the world. Hubei has a population of 58 million and only 66907 confirmed cases, so the hospitals were not nearly as overwhelmed as they could have been.
ETA: Looking at the Imperial paper you cited, the (uncorrected) estimated CFR being 18% was not the result of hospitals overwhelmed, but undercounting infections within Hubei (see page 10), which they then corrected to 1%. So I think this is consistent with the CFR estimate being based on hospitals not being overwhelmed, and it being much higher in the future when hospitals throughout the world will be overwhelmed.
Good point about importing doctors from other cities. This won’t be possible anymore if a country has the virus everywhere.
Room upward for more hospital crowding: Okay yeah, crowding could be a lot worse still. But self-isolation of the entire population seems like an option that can be implemented by any country to slow the peak of an outbreak. It doesn’t seem to require China-level control over a population. So maybe it’s unlikely that things would become worse than they did in Hubei? I guess you could argue that unless the government is extremely strict about people not being allowed to go outside, then too many people will still do it and get infected.
I think you’re giving convincing arguments to be a lot more pessimistic than the 1% expected CFR. The main counterconsiderations I see are that with more preparedness, countries can better slow down the peak of an epidemic. I don’t think cities are likely to quickly end up in a state that’s as bad as in Wuhan because by the time the hospitals start to crowd, the cities will already have been giving strongly worded advice about self-isolating at home for many weeks. This time window was missed in Wuhan. (But I was already aware of those considerations, so for me your points still provided a large update. I don’t expect to end up anywhere close to the high numbers you’re giving, but I’ll have to think about it more and I definitely think 2% total CFR seems possible or even likely now as we forecast scenarios approaching one billion cases.)
0.1% of Hubei’s population have a confirmed infection, and its hospitals are already at the breaking point (even with national resources transferred into it). If this is the limit, then hospitals can treat at most 0.1% of the population per month, so it would take 50-100 months to treat the 5-10% who will require hospitalization, which is not a realistic amount of time for self-isolation. Even 10 months is not realistic (without totally ruining the economy) so it seems “slowing the peak” just won’t help much.
The WHO report said some 90% of people had fever. Maybe the tests have a significant false neg rate, or theyre not testing enough people, but I am worried that few mild cases are being missed.
I’m very interested in seeing how the Diamond Princess plays out. On feb 20, some half of the cases were asymptomatic. Wikipedia hasn’t been updated with any data since then. This bit feels cruxy to me.
FWIW, https://www.worldometers.info/coronavirus/ counts all confirmed cases and has a table by country, which lists the Diamond Princess separately (“international conveyance”). It doesn’t distinguish asymptomatic from mild, but does separate out “serious, critical” cases, which stand at 36/705 (plus 7/705 deaths and 100/705 recovered).