3 ms·

Agree completely that this is far more dangerous and deadly than the flu (especially as a New Yorker who knows multiple people who have died from this disease).

Agree completely that this is far more dangerous and deadly than the flu (especially as a New Yorker who knows multiple people who have died from this disease). That being said, there is some selection bias in these results as those who were tested were much more likely to have been infected. These tests were not done in randomly selected people. Most who get tested have reason to believe they were infected (or were at high risk for having been infected). The fact that the 27% of the group of high risk people who chose to get tested showed they had the disease doesn't mean that 27% of all New Yorkers had the disease.

nimish 6y ago
Mortality is also highly age dependent. It mostly reflects that the virus ripped through care facilities since nobody isolated them in time.
raducu 6y ago
You are right, I missed that the number of tests was ~1 million, no way that was randon testing. So, probably the mortality rate is higher. I wonder if there is any bayesian tricks to estimate the infection range given 1/8 people got tested, regardless of selection bias. edit: actually the overall deatheate could be lower, because we know the number of deaths, but not the exact number of infected -- if overall the infection rate is lower then 27%, the deathrate is lower than 1%.
- StanislavPetrov 6y ago
  Nobody is suggesting the testing was random. The overall death rate and infection right could be either higher or lower, my only point is that you can't draw conclusions about the total population on a testing sample that wasn't random (and didn't try to be).
- rumanator 6y ago
  > because we know the number of deaths We only know estimates of the number of deaths. The correct number lies somewhere between the official number of covid19 fatalities and the total number of deaths during the same period, because one hand you have excessive mortality and on the other hand you have changes to the baseline due to changes in behavioral patterns. Excessive mortality are higher than officially recognized covid19 deaths and serve as a more accurate proxy for real covid19 deaths, but at the same time the baseline mortality might vary as well.
  - raducu 6y ago
    Do you know any numbers for that excessive mortality in NYC? That would be pretty interesting.

nimish 6y ago

Mortality is also highly age dependent. It mostly reflects that the virus ripped through care facilities since nobody isolated them in time.

raducu 6y ago

You are right, I missed that the number of tests was ~1 million, no way that was randon testing. So, probably the mortality rate is higher. I wonder if there is any bayesian tricks to estimate the infection range given 1/8 people got tested, regardless of selection bias. edit: actually the overall deatheate could be lower, because we know the number of deaths, but not the exact number of infected -- if overall the infection rate is lower then 27%, the deathrate is lower than 1%.

StanislavPetrov 6y ago
Nobody is suggesting the testing was random. The overall death rate and infection right could be either higher or lower, my only point is that you can't draw conclusions about the total population on a testing sample that wasn't random (and didn't try to be).
rumanator 6y ago
> because we know the number of deaths We only know estimates of the number of deaths. The correct number lies somewhere between the official number of covid19 fatalities and the total number of deaths during the same period, because one hand you have excessive mortality and on the other hand you have changes to the baseline due to changes in behavioral patterns. Excessive mortality are higher than officially recognized covid19 deaths and serve as a more accurate proxy for real covid19 deaths, but at the same time the baseline mortality might vary as well.
- raducu 6y ago
  Do you know any numbers for that excessive mortality in NYC? That would be pretty interesting.