Hacker News Reader

Comment by marvin | Hacker News Reader

Here's my wild-ass guess: The percentage of lethal cases is actually close to 0.1% when the serious cases get the best care possible (my claim). If this is true, then Italy had 200*1000 = 20.000 early-stage cases 8 days ago. This yields on the order of 80.000 cases today, hopefully less as the quarantines should have an effect by now.

This assumes a median of 8 days from onset to death in the lethal cases, and a doubling time of # infections of 4 days, which seems to be around the current level for the world outside of China. The number is very sensitive to average lethality rate, and hence also to the age & health distribution of the population.

It's completely implausible that Italy has only 5000 cases, as people who have visited Italy for their winter holidays have very frequently tested positive after returning home. They haven't all randomly met the same 5000 Italians.

View on HN

Uh... data from this Chinese study suggests that death & recovery take longer than that... I read it as about 30 - 40 days from exposure.

Also suggests you've got a critical / potentially terminal patient occupying an ICU bed for 2 - 3 weeks in the process, which becomes a significant load on your available hospital capacity...

https://www.medrxiv.org/content/10.1101/2020.03.03.20028423v...

Can't you just send out test kits to 10000 random Italians and then use that to calculate the number of infected people? If its really common then you'd find 100-1000 infected people and then you can calculate the real mortality rate simply by monitoring them and recording how many of them have died. Of course it is possible that proper medical care results in a much lower than usual mortality rate compared to untreated people but that sounds more like good news than bad news.

> Can't you just send out test kits to 10000 random Italians and then use that to calculate the number of infected people?

I think at this point, it's still very early in the spread of infection and the distribution of infection might not be random and might be clustered around certain locations, social circles etc.

I would LOVE to see a full population test sample... Would be expensive (probably need 5 K - 10 K tests for a significant read, most of which will likely be wasted on "likely healthy" members of the community... but would decisively settle this outstanding question of asymptomatic infection frequency.

The closest data point we have is likely that South Korean church, where I think they hunted down more or less every member and tested them... Doesn't give you a true view of asymptomatic presence in the general population but does provide a cue about relative size vs. known infected...

The scary part is the WHO mission were adamant that they failed to find a large number of asymptomatic/mild cases in the population at large, and they tested 300.000 people. I can't get this to match up with my other hunches, so I might be very wrong on this.

Reply on news.ycombinator.com