6 ms·

1/500 needs a citation. Because the WHO and CDC as of a few days ago said 1/50. But even 1/500 is an incredibly high mortality rate for something as infectious

1/500 needs a citation. Because the WHO and CDC as of a few days ago said 1/50. But even 1/500 is an incredibly high mortality rate for something as infectious as the flu, and to make light of it is bizarre. Even your number would make catching it about 1000x more dangerous than skydiving.

throwawayhhakdl 7y ago
What bothers me is the difficulty in interpreting such numbers because they’re not normalized to health. E.g. dropping sperm counts in men. Is that dropping in all men? Or is it a side effect of rising obesity in enough men to dent the average? Is the mortality 1/500 here because a good health 30 year old has a 1/500 chance of dying? Or because the 1/100 30 year olds with poor health have a 1/5 chance of dying?
anonuser123456 7y ago
So 1/500 sounds like a correct ballpark. A few days ago there was some data released by China on some 40,000 cases. When you age adjust the mortality stats by: - Assuming uniform age distribution - Accounting for the age distribution of the population. You get something like .4% for the 30-40 age range. This was a crude estimate... but that's why I say ballpark. It also could go up/down as we get more data. I agree the risk here is considerably high. People just don't understand scale well. Assuming the R0 is high, it could eventually sweep through the majority of the population. IIRC, the 1918 flu pandemic is estimated to have infected 80% of the population. So in the US alone that's 6,000,000 deaths, or one Sept 11. like attack every day for 200 days.
- Retric 7y ago
  Deaths are a lagging indicator. 2119 deaths as of yesterday / 0.004 = 530,000 cases that are ~2+ weeks old. In other words 530,000 infections around February 5th. That seems extremely unlikely based on the data we have. For a reasonably accurate estimate you need to look at percentage of total infections detected, and the lag between detection and death. Using say 50% detection rate and 7 days between detection and death you get 2119 / (63,851 / 0.5) = 1.7% fatality rate. That’s about as optimistic as I think is reasonable. PS: A 50% detection rate may be high or low, but that’s the real unknown.
  - anonuser123456 7y ago
    I don't disagree that the data has very large error bars. I'm simply saying that, using the current best guess we can do the following: - Stratify relative risk based on population distribution. - Guage the impact of the expected bad case of ~2% mortality. Will mortality rate come down? Very likely. But it may not... And if not, what is the outcome at scale?

throwawayhhakdl 7y ago

What bothers me is the difficulty in interpreting such numbers because they’re not normalized to health. E.g. dropping sperm counts in men. Is that dropping in all men? Or is it a side effect of rising obesity in enough men to dent the average? Is the mortality 1/500 here because a good health 30 year old has a 1/500 chance of dying? Or because the 1/100 30 year olds with poor health have a 1/5 chance of dying?

anonuser123456 7y ago

So 1/500 sounds like a correct ballpark. A few days ago there was some data released by China on some 40,000 cases. When you age adjust the mortality stats by: - Assuming uniform age distribution - Accounting for the age distribution of the population. You get something like .4% for the 30-40 age range. This was a crude estimate... but that's why I say ballpark. It also could go up/down as we get more data. I agree the risk here is considerably high. People just don't understand scale well. Assuming the R0 is high, it could eventually sweep through the majority of the population. IIRC, the 1918 flu pandemic is estimated to have infected 80% of the population. So in the US alone that's 6,000,000 deaths, or one Sept 11. like attack every day for 200 days.

Retric 7y ago
Deaths are a lagging indicator. 2119 deaths as of yesterday / 0.004 = 530,000 cases that are ~2+ weeks old. In other words 530,000 infections around February 5th. That seems extremely unlikely based on the data we have. For a reasonably accurate estimate you need to look at percentage of total infections detected, and the lag between detection and death. Using say 50% detection rate and 7 days between detection and death you get 2119 / (63,851 / 0.5) = 1.7% fatality rate. That’s about as optimistic as I think is reasonable. PS: A 50% detection rate may be high or low, but that’s the real unknown.
- anonuser123456 7y ago
  I don't disagree that the data has very large error bars. I'm simply saying that, using the current best guess we can do the following: - Stratify relative risk based on population distribution. - Guage the impact of the expected bad case of ~2% mortality. Will mortality rate come down? Very likely. But it may not... And if not, what is the outcome at scale?