re: Some Coronavirus Spread Analysis

quote:

---

And when the final chapter of this is written, the mortality rate will be lower than .5%. Your basic point that there are way more positives out there than we know is correct. Actually, I think we will never know the true number, because we are now learning that many positives never show any symptoms at all.

---

Also, I'm trying to guess the "true number" which is the point of the post. If the mortality rate is lower, say 0.1%, that just means there are a lot more people that actually have it right now than 25,600. It would mean that on Feb 25th 1000 people had the disease. Doubling that 6 times yields 32000 people are currently infected.

quote:

---

So you think that only 479 people in Louisiana have the virus right now? Because that's the only "accurate data" we have. I'm making projections here.

---

continue making shite up...I'm sure it's really useful for you

quote:

---

What percentage of these have underlying medical issues already?

That's a lot harder to guess on. We won't really know that until after the dust settles.

---

Huge problem with this whole thing. 99% of the fatalities in Italy had underlying conditions.

quote:

---

99% of the fatalities in Italy had underlying conditions.

---

That will probably be true here as well. But if a lot of people with underlying conditions contract the virus, you get a lot of deaths. That's why we're seeing vastly different mortality rates in different parts of the world.

quote:

---

How are you getting the number of actual cases if you are not using the number confirmed? Seems like you're making a pretty large assumptions that would totally invalidate your data.

---

He’s using deaths as a baseline instead of using confirmed cases. Which makes a lot of sense.. it seems likely that “number of COVID-related deaths” is currently a much more accurate number than “number of COVID cases”.

The problem is that there’s a ton of error built in when you try to back into the number of cases from the number of deaths:
- uncertainty around mortality rate
- small sample size (especially starting from the first death)
- potential for COVID deaths to have not been properly attributed to the virus prior to widespread testing

To the OP:
Backing up from the first death is probably not the best way, since that’s the point where the error for mortality is at its highest due to the smallest sample size. I would suggest fitting a trend line to the running total of deaths, then applying metrics for incubation period and mortality rate.

quote:

---

Huge problem with this whole thing. 99% of the fatalities in Italy had underlying conditions.

---

I wouldn’t call it a huge problem from a statistical standpoint. Data is data. You can, theoretically, build things like age distribution, smoking prevalence, etc. into a model.

But most people who understand statistics don’t think that a statement like “0.5% mortality rate” means you have a 0.5% chance of dying if you catch the virus. To know that you would have to factor in the data across age groups, medical histories, etc.

The fact is that there are many more cases than the number of confirmed cases. The mortality rate will be lower when final studies are completed in 2 or 3 years.

The only number that matters is the number of cases that require hospitalization. Once new hospitalizations and bed use declines, we can get back to normal. I wish this was the number being published.

Once we figure out which therapies work best, hospitals will be able to manage this and we will get back to normal.