[jmnicolas]

If this thing was killing half of the infected persons there would already be much more deaths than there is now.

Unless most the infections occurred very recently which is pretty unlikely.

I think my 98% is probably a bit too optimistic, but your 53% is way beyond pessimistic.

[thebigspacefuck]

If you look at the case study of the first patient in the US from New England Journal of Medicine the patient didn’t get to a poor condition until day 11 of the illness https://www.nejm.org/doi/full/10.1056/NEJMoa2001191 at which point they tried an experimental antiviral and he recovered. 11 days ago, there were 527 cases. Patient in study didn’t come in until 2 days after having the sickness so if you look at -9 days infected was at 943. Currently at 305 deceased and 443 recovered, total of 748, so it seems like the infected between 9 and 11 days ago are the ones recovering or dying. The number of infected goes up rapidly from that point (now we are at over 16x number of infected from 9 days ago), so we’ll see the numbers get more accurate soon. Although number of cases rapidly increases, the antiviral treatments seem to help dramatically and they have just begun to start experimenting with them. Currently sitting at 40% mortality using D/D+R, but I imagine we’ll see more recoveries as antivirals are used.

[ericb]

Keep in mind, this is a self-selected group of people. There could be 10x more asymptomatic, or low impact cases who thought they had the flu, but didn't feel bad enough to go to the hospital. Imagine if I tried to capture the survival rate of the flu, but chose my population of patients from the ICU. Would the numbers be right?

I think this is born out by the really small number of children showing up in the mortality numbers. Proportionally, kids usually over-represent transmission, I think, from kid-level hygiene and being crowded into small classrooms. They seem under-represented in the study population, which makes me think there could be a lot of selection bias being captured.

Additionally, the standard for recovery is really high and is built in a way that it will severely lag deaths (which have no waiting period). I think it was something like total clearance of the virus, not "feels good, now surfs reddit all day."

[cstejerean]

There could also be many X more cases that haven’t been confirmed because they can’t get access to transportation to a hospital, or because the hospitals are overwhelmed, or because they are running low on test kits or behind the schedule on testing. I wouldn’t assume that the unreported numbers are strictly positive.

[colsmit]

I appreciate you linking this - I've read at least a books worth of panicked news coverage over the past week and had not seen a single story about potentially successful application of antivirals. I guess fear sells better.

[chongli]

The number of cases is growing exponentially. If it takes a few weeks for the disease to kill a person, then the low number of deaths relative to the number of infections is indicative of that lag time.

You might do better to compare the number of deaths to the number of infected from a few weeks ago, which was a much lower number than today.

[moralestapia]

>there would already be much more deaths than there is now

Why? People can be sick for quite some time (in this case, a month, maybe?) and then ultimately die from the disease, this is actually like ... the most common scenario ...

I have been unpleasantly surprised by how much people struggle to understand this simple concept. Recovered / (recovered+death) is how you do it. It's common sense, come on ...

[caconym_]

Since we are in early days and it probably takes longer to recover than to die of it, and the reporting of infections and recoveries and deaths coming out of China is likely undercounting all three due to different systematic factors, I think it's too early for the results of that formula to be taken at face value.

In the end, when everything is tallied up, sure. But in the meantime, in the presence of so much uncertainty, "common sense" is only going to lead you astray.

[moralestapia]

I agree with you that the correct answer is "we have to wait more".

[stale2002]

> I have been unpleasantly surprised by how much people struggle to understand this simple concept. Recovered / (recovered+death) is how you do it.

The mistake that you are making in your "simple math equation" is as follows. It is easier to count the number of deaths than it is to count the number of recovered. This messes up your math.

If you miscount a large amount on your recovered numbers, then this will make the death rate look worse than it actually is.

[marcosdumay]

> People can be sick for quite some time (in this case, a month, maybe?) and then ultimately die from the disease

Usually, not from a viral infection. The usual scenario from a viral infection is that if you are not dead in 2 weeks, you will almost certainly recover.

There are exceptions, but nothing I've read about this virus leads to the conclusion that it is one of those.

[cmendel]

Yes, but the incubation period on this one is 14 days. That changes the calculus.

[marcosdumay]

Well, that delays things from the infection point. But doesn't change anything from the point where people notice they are sick.

Anyway, this is a disease that spreads very widely very quickly. We are arguing about the known numbers when the most likely scenario is that it is already everywhere but not reported, so the known numbers are only noise.

[jmnicolas]

I think it's meaningless to make stats on unknown things. We have no idea how many of the infected people will die.

The only thing we can say is : as of now the survival rate is 98%. It will probably go down, nobody knows how much.

[allovernow]

How did you calculate 98% from the available figures when the vast majority of cases are less than a week old?

[allovernow]

This outbreak just started in earnest weeks ago. It has a 2 week incubation period and the illness may take 1-2 weeks before recovery or death. That's why you must look at dead/recovered. I agree that 53% for the general population is pessimistic because as usual there is a selection bias towards those with weakened immune systems, not by as much as you may think.

[ericb]

Are you sure?

As a thought experiment, if you infected an entire city with the flu, the ones to show up at the hospital are the worst cases, and if you just built your data from those people, you'd get a really really skewed perspective.

[DuskStar]

Infections are doubling every few days, so yes most infections occurred very recently. If it takes a month from initial infection to declare someone "disease-free", then there might only be ~300 people to that point.