The case fatality rate for COVID-19 is not "about 2%" - it's 3.4%

[The_jackalope]

Based on tonight's numbers from https://www.worldometers.info/coronavirus/ : Deaths 2804, Cases 82,183, the case fatality ratio is 3.41%

Not only that, but the fatality rate has been climbing steadily for the last two weeks:

12-Feb 2.13%
13-Feb 2.15%
14-Feb 2.27%
15-Feb 2.41%
16-Feb 2.49%
17-Feb 2.55%
18-Feb 2.67%
19-Feb 2.81%
20-Feb 2.93%
21-Feb 3.04%
22-Feb 3.13%
23-Feb 3.31%
24-Feb 3.37%
25-Feb 3.41%
26-Feb 3.41%

That climb could just be an artifact of slowing case growth, but maybe not. I'm no pandemiologist , but this behaviour of the data worries me. At the very least, the white coats and empty suits should be using more accurate numbers, in the interests of transparency.

[PoindexterOglethorpe]

1. It's also helpful to realize that many cases are not diagnosed

because people are not very sick or asymptomatic altogether.

Also, in China they've changed the definition of who has the coronavirus at least twice, which makes statistics a bit tricky. But the overall death rate for those who get the virus is significantly less, given the many undiagnosed cases.

[The_jackalope]

3. "May be" significantly less.

Maybe there are 30x as many hidden as reported cases (and nobody is dying undiagnosed), and the eventual fatality rate will turn out to be just the same as the flu?

Or not.

[PoindexterOglethorpe]

6. If you honestly think that the published number so those who've contracted

this coronavirus is anywhere near the actual number, I beg to differ.

Especially in China. Or possibly in places like Iran.

[The_jackalope]

11. A more effective way to make things look better than they are

Would be to minimize the death rates, not the case rates. If the Chinese are jiggering any of the numbers, my money would be on death reporting.

[wnylib]

25. Wouldn't there be an incentive

to under report cases in order to minimize the extent of its spread?

[The_jackalope]

35. Yes, but there's an at least equal incentive to under report deaths

Both to obscure the lethality, and also to make the reporting country look more effective at combating the disease itself, as opposed to just looking more effective at slowing its spread.

Under-reporting deaths tends to make the reporting country look better at managing cases, under-reporting cases makes the reporting country look better at managing the spread of the infection.

I have no idea if such intentional, politically motivated under-reporting is going on. We may get some better indications as the global case load rises. At the moment, the most likely thing is that asymptomatic cases are simply being missed.

[wnylib]

37. So far, US sources say there are 15

confirmed cases inside the US. But I just read on the BBC site that there are 60 cases in the US. Wonder why the discrepancy and where the BBC got its figure from.

[The_jackalope]

38. Johns Hopkins web site reports 60.

[mr_lebowski]

5. That percentage is probably still helpful as a benchmark vs other similar diseases though ...

Since you could say the same thing about most diseases I would think.

[Sunsky]

30. Exactly

For example, the flu. Many people sick with the flu don't seek help. Most people nurse the flu at home. I would believe with the fanfare surrounding coronavirus more people will be inclined to seek help.

[PoindexterOglethorpe]

43. That's a very good point.

[greenjar_01]

39. Much more likely

I want to see a parallel case of a 3-4% mortality rate for a pulmonary virus. It would be unusual, at the very least.

[Amishman]

41. SARS and MERS are both 10%+ and are in the same family

We are fortunate that this thing is actually much less deadly than its closest cousins

[applegrove]

2. It is probably going up because many people are just carriers or

mild cases and, have not been counted as having caught COVID-19. So the ratio is not as spot on as it was when this started.

[The_jackalope]

4. Early data is always noisier than later data

We still don't know how reliable the reports are, but I'm quite willing to entertain the idea that we're seeing something that has one foot in the H1N1 camp and one foot in Spanish Flu territory.

We'll see.

[Lucky Luciano]

7. Spanish flu was really bad in that it was most lethal for young adults.

Not the case here.

[The_jackalope]

10. So maybe it's turns out to be half as lethal as Spanish Flu, and affects older people

but has a somewhat higher R-nought than SF? That's pretty cold comfort.

[applegrove]

8. Well the early data was more accurate because they could always almost

work it back to the first carriers. So they knew geographically where to look for victims and carriers. So it was cleaner data. Now with it on every continent, those mild cases are much harder to track. So the stats are off. I hope.

[The_jackalope]

9. But Chinese politicians were (and are) heavily involved in the data management.

But of course I'm sure they wouldn't have played around with the death counts. Right? I mean, Xi is a totally transparent guy when the Chinese reputation is at stake. Right?

The point is that given all the uncertainty, our politicians are choosing to spin the numbers for public consumption by using dodgy data from a couple of weeks ago. Not surprising, but something to factor into one's thinking.

[applegrove]

14. But if you don't count all the carriers, and the virus has gotten out of containment,

then the fraction is off. If 20 people are carriers/mild/sick coronavirus and ten die of it, 10/20 is 50% you have a mortality rate of 50 percent.

But if you undercount the number of carriers/mild/sick coronavirus and say it is 15 instead of 20, and the number who died stays the same (they are easier to count) 10/15 the mortality rate goes up to 66%. That is 66.666%.

So if you undercount all the people who carry/mild/sick the coronavirus the mortality rate goes up.

[The_jackalope]

15. Overcounting the cases or undercounting the deaths both make the situation look better.

If I had money to bet, I'd say the Chinese authorities been under-reporting the deaths so as to keep the fatality rate low and not spook the horses. That may be getting harder to do as the case load grows, so the case fatality rate may be climbing as a consequence.

The situation easily could be worse than it appears. Only time and more transparent reporting from Western democracies will tell us what's really going on.

[applegrove]

16. I think the Chinese are scientists. They dislike bad data as much as

the next scientist. I admit they were slow to react to the Dec 1 first cases of novel coronavirus. And did not react until january. But i doubt they are jigging the numbers now. I still think it is that the carriers/mildcases/sick are now undercounted as it is harder to do as the virus has spread. The people who die are easier to count.

[The_jackalope]

17. The problem might not be bad scientists

But "normal" politicians...

eta: the issue is not so much counting the dead, as reporting the count.

[applegrove]

19. But it is spreading. Other countries will be able to figure out the mortality rate

for themselves. For instance canadian scientists looked at Iran's numbers and were able to say they are way undercounting how much the virus has spread there. By about 20,000.

[Demonaut]

12. the powers that be recognize the significance

[The_jackalope]

13. It looks like they do.

Luckily for them the American public tends toward innumeracy.

[lunasun]

20. Rec Always appreciate your posts

[Laffy Kat]

21. The good news: it's not killing kids.

So far, the known death rate for 0 to 9-year-olds is zero.

[3catwoman3]

22. Unlike the flu. So far this season, there have been...

...105 pediatric deaths attributed to influenza, compared to 41 last season.

[Laffy Kat]

23. Yes, thanks; I was just checking the CDC for peds.

It appears there's been a steady increase in pediatric influenza deaths over the last few seasons. It wasn't clear by the data I saw how many of those deaths were from non-vaccinated children. Regardless, I wonder if that trend will continue.

[cstanleytech]

24. The majority of those are in China though and alot of those I suspect are people that were more

advanced with the virus and thus weakened greatly before they were able to come up with a stronger treatment plan.
The real % should become more evident as we see what happens outside of China in countries like S Korea, Italy and others.

[JCMach1]

26. Correct answer is we don't know yet, but it's a number 2-9%

Really, really bad, but not Captain Trips

[wnylib]

27. No one here has mentioned another

possibility for an increase in death rates. Long incubation period, long lasting illness before an infected person recovers or dies. The death numbers lag behind the infection numbers, but increase as time goes on.

[Wounded Bear]

28. It's actually a bit early. Stats will move around...

Does look concerning, especially with the transmissability being so high.

[Ms. Toad]

29. It has been around that since early on -

if you calculate today's deaths against the number of cases 2 weeks ago. (Using the number of cases 2 weeks ago gives you a better estimate, since it is about 2 weeks from symptoms to death. That means some of the brand new cases in your denominator will die in the next two weeks, so shouldn't be treated as resolved yet.)

[Blues Heron]

31. In Korea it's less than 1 percent

11 fatalities 1766 cases

That's a developed country with a big heads up on this outbreak, with the most cases outside China.

I'd pull back on the tin foil re. China hiding deaths - that's not really helpful.

[Coventina]

34. This virus starts off very mild and gradually gets worse.

People are sick with it for a very long time.

The outbreak in S. Korea is very recent. Their death rate will climb.

[beachbumbob]

32. we have little dependable data as China held back for so long, and I am more concern about

who are dying and it looks like the "low risk" ones are

[uponit7771]

40. +1,

[Squinch]

33. OK, you talking about behavior of the data makes me think you know more about this stuff than

I do. So let me ask you a question.

There was that big jump in new cases in one day a while back. 15000 new cases in a day. The explanation was that they had been using a test that gave 50 to 70 percent false negatives, and they had changed to a new, more accurate test.

Prior to that, the new case line in the graph had a pretty steady slope.

Then, there's that one day where the slope gets very steep.

But then, in subsequent days, the slope of the new cases line goes back to pretty much the same as it was before they changed the test.

If they are using a new test that DOESN'T give the 50 to 70% false negatives, shouldn't that line have become much steeper and stayed steeper as the new test produced many fewer false negatives?

[The_jackalope]

36. Intuitively I agree with you.

Better tests should result in a steeper curve.

Something else may have been going on behind the scenes to cause that spike. If one were to be cynical, one might speculate that there was an accidental release of "true" data that couldn't be retracted. The data minders then made sure the leak was plugged. But that's just me being cynical. We don't know nearly enough yet.

[onenote]

42. A controlled population might be a better measure.

If, as I understand to be the case, all 3700 people on board the Diamond Princess was tested, and around 700 were determined to have contracted the virus and five have died, that's a fatality rate of around 0.7 percent. That may be low, but it may be a more accurate indicator that a rate based on comparing fatalities to tested/confirmed cases when there is almost certainly a significant number of people who have contracted the virus but haven't been tested and recovered.